Measuring Health: A Guide to Rating Scales and Questionnaires

Measuring Health: A Guide to Rating Scales and Questionnaires, Third Edition Ian McDowell OXFORD UNIVERSITY PRESS Measuring Health This page intentionally left blank MEASURING HEALTH A Guide to Rating Scales and Questionnaires THIRD EDITION Ian McDowell 1 2006 3 Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Copyright © 2006 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data McDowell, Ian, 1947– Measuring health: a guide to rating scales and questionnaires / Ian McDowell.—3rd ed. p. ; cm. Includes bibliographical references and index. ISBN-13 978-0-19-516567-8 ISBN 0-19-516567-5 1. Health surveys. 2. Social surveys. 3. Health status indicators—Measurement. [DNLM: 1. Health Surveys. 2. Health Status Indicators. 3. Pain Measurement. 4. Psychiatric Status Rating Scales. 5. Psychological Tests. 6. Questionnaires. WA 900.1 M478m 2005] I. Title. RA408.5.M38 2005 614.4'2—dc22 2005010668 9 8 7 6 5 4 3 2 1 Printed in the United States of America on acid-free paper  Preface The text of the first edition of this book argued cians. The information provided is intended to in 1987 that no concise source of information be sufficiently detailed to explain the strengths on health measurement methods existed; at the and limitations of each measure, without neces- time, this appeared to be a significant problem sarily providing an exhaustive systematic re- for epidemiological and health care researchers. view. This third edition differs from the second The methods in question cover a variety of top- in three main ways. It adds a completely new ics, including physical disability, emotional and chapter on anxiety measures; it includes new social well-being, pain, life satisfaction, general measures in most of the existing chapters, so health status, and quality of life. Several books that the number of scales reviewed has risen reviewing health measurements later appeared from 88 to 104. In addition, all of the existing in the early 1990s, validating our original argu- reviews have been updated to cover new ver- ment (1–3). Nonetheless, it remained challeng- sions of the measures and new information on ing to keep pace with the rapid development of reliability and validity. the field, and descriptions of methods were The book is intended to serve two main pur- widely scattered in social science, medical, poses. First, it describes the current status of management, and methodological journals. The health measurement, reviewing its theoretical second edition of this book, in 1996, included a and methodological bases and indicating those broader range of instruments than the first; the areas in which further development is required. book grew from 340 to 520 pages. The goal This discussion is intended to be useful for stu- was still to provide reasonably detailed, al- dents and for those who develop measurement though not exhaustive, information on a selec- methods. Second, and more important, the book tion of leading instruments. During the second describes the leading health measurement meth- half of the 1990s, several Web sites were estab- ods, providing practical information for those lished to provide brief overviews of measures, who intend to use a measure for clinical and re- but these generally did not provide detailed re- search applications. An underlying goal is to try views. These sites are useful in raising aware- and help the field move forward by a combina- ness of the range of instruments available but tion of comments and suggestions for improve- they rarely provide sufficient information to ment. help readers choose among rival instruments. The methods all use subjective judgments ob- Furthermore, the quality of the sites varies, and tained from questionnaires or clinical rating some include information that is simply incor- scales; they do not include laboratory tests or rect. Hence, this third edition of Measuring physical measures of function. The reviews give Health seeks to fill a niche by providing accu- full descriptions, where possible including copies rate and reasonably detailed information on a of the actual instruments, of over 100 measure- selection of the most commonly used instru- ment methods. Each review summarizes the reli- ments. The scope is not exhaustive, because the ability and validity of the method and provides intention is to focus on those measures most information necessary to allow readers to select likely to be useful for researchers and clini- the most appropriate measurement for their pur- vi Preface poses and to apply and score it correctly. As an References introduction to health measurement methods, this book should be of value to clinicians who 1. Wilkin D, Hallam L, Doggett MA. Mea- sures of need and outcome for primary wish to select a measure to record the progress health care. Oxford: Oxford University of their patients. It should also serve as a refer- Press, 1992. ence work for social scientists, epidemiolo- 2. Bowling A. Measuring health: a review of gists and other health care researchers, and for quality of life measurement scales. Milton health planners and evaluators: in short, for all Keynes, U.K. Open University Press, 1991. who need to measure health status in research 3. Spilker B, ed. Quality of life assessment in studies. clinical trials. New York: Raven Press, 1990  Acknowledgments The first edition of this book was written in the production department at Oxford University 1987 with the assistance of Claire Newell, who Press. Her calm and good cheer transformed the provided invaluable research and editorial input proofreading stage into a pleasurable experi- throughout the project. While the 1996 second ence. edition was being prepared, Claire was working Most of all, I owe a huge debt of gratitude to overseas and was only able to offer limited edito- my family. In reverse order of age: to my daugh- rial input. Responsibility for preparing this third ter Karin, who generously bequeathed me her edition has fallen to me alone; sadly I can no outdated computer to use for word processing longer share blame for anything! the manuscript; to my sons Graeme, Wesley, and I am grateful to my colleagues at work who Kris for their good cheer and patient guidance in tolerantly handled my absences from the office; I managing said computer’s occasional cognitive thank the University of Ottawa for its Internet losses and digestive indiscretions. Above all, my and library support, especially the previously gratitude to my wife Carrol, who has patiently unimagined luxury of on-line periodicals. I ex- endured being a book widow for far, far too tend warm thanks to Sylvie Desrochers who pa- many months. I do recognize that an author’s tiently assisted me during 2004 by finding copies obsession is hard to live with and is a selfish of numerous, often obscure journal articles and pleasure that comes at considerable cost to those book chapters. I also thank Anne Enenbach of around him. This page intentionally left blank  Contents List of Exhibits xiii Group Differences and Sensitivity to Change 37 1. Introduction 3 Construct Validity: Conclusion 39 Assessing Reliability 39 Background 3 Internal Consistency 42 Selection of Instruments for Review 3 Interpreting Reliability Structure of the Book 4 Coefficients 45 Style and Content of the Reviews 5 Summary 46 Format for the Reviews 5 Evaluating a Health Measurement: The User’s Perspective 7 3. Physical Disability and Handicap 55 2. The Theoretical and Technical The Evolution of Physical Disability Foundations of Health Measurements 55 Measurement 10 Scope of the Chapter 58 The Evolution of Health The Pulses Profile 62 Indicators 11 The Barthel Index 66 Types of Health Measurements 12 The Index of Independence in Theoretical Bases for Measurement: Activities of Daily Living, or Psychophysics and ADL 74 Psychometrics 16 The Kenny Self-Care Numerical Estimates of Health 18 Evaluation 78 Scaling Methods 18 The Physical Self-Maintenance Psychometric Methods 20 Scale 84 Methods Derived from Economics The Disability Interview and Decision Analysis 23 Schedule 88 Identifying and Controlling Biases in The Lambeth Disability Screening Subjective Judgments 25 Questionnaire 89 Conceptual Bases for Health The OECD Long-Term Disability Measurements 27 Questionnaire 93 The Quality of a Measurement: The Functional Status Rating Validity and Reliability 30 System 95 Assessing Validity 30 The Rapid Disability Rating Scale 97 Criterion Validity 31 The Functional Status Index 100 Construct Validity 34 The Patient Evaluation Conference Correlational Evidence of System 103 Validity 34 The Functional Activities Factorial Validity 36 Questionnaire 108 ix x Contents The Health Assessment The Positive and Negative Affect Questionnaire 111 Scale 225 The MOS Physical Functioning The Life Satisfaction Index 231 Measure 119 The Philadelphia Geriatric Center The Functional Autonomy Morale Scale 236 Measurement System 122 The General Well-Being The Functional Independence Schedule 240 Measure 141 The RAND Mental Health Conclusion 149 Inventory 247 The Health Perceptions Questionnaire 253 4. Social Health 150 The General Health Social Adjustment and Social Questionnaire 259 Roles 151 Conclusion 271 Social Support 152 Scope of the Chapter 153 The Social Relationship Scale 155 6. Anxiety 273 The Social Support Theoretical Approaches to Anxiety 274 Questionnaire 158 Anxiety and Depression 276 The RAND Social Health Anxiety Measurements 277 Battery 161 The Taylor Manifest Anxiety The MOS Social Support Scale 279 Survey 165 The Hamilton Anxiety Rating The Duke-UNC Functional Social Scale 286 Support Questionnaire 168 The Hospital Anxiety and The Duke Social Support and Stress Depression Scale 294 Scale 170 The Zung Self-Rating Anxiety The Katz Adjustment Scales 173 Scale 302 The Social Functioning The Zung Anxiety Status Schedule 177 Inventory 302 The Interview Schedule for Social The Beck Anxiety Inventory 306 Interaction 180 The Depression Anxiety Stress The Social Adjustment Scale 184 Scales 313 The Social Maladjustment The State-Trait Anxiety Schedule 194 Inventory 319 The Social Dysfunction Rating Conclusion 327 Scale 196 The Structured and Scaled Interview to Assess Maladjustment 199 7. Depression 329 Conclusion 203 Classifications of Depression 330 Measurement of Depression 332 Scope of the Chapter 333 5. Psychological Well-being 206 The Beck Depression Inventory 335 Scope of the Chapter 208 The Self-Rating Depression Scale 344 The Health Opinion Survey 210 The Center for Epidemiologic The Twenty-Two Item Screening Studies Depression Scale 350 Score of Psychiatric The Geriatric Depression Scale 359 Symptoms 216 The Depression Adjective Check The Affect Balance Scale 221 Lists 364  Contents xi The Hamilton Rating Scale for Questionnaire Techniques 473 Depression 369 Behavioral Measurements of The Brief Assessment Schedule— Pain 473 Depression 378 Analogue Methods 474 The Montgomery-Åsberg Depression Scope of the Chapter 474 Rating Scale 382 Visual Analogue Pain Rating The Carrol Rating Scale for Scales 477 Depression 387 The McGill Pain Questionnaire 483 Conclusion 390 The Brief Pain Inventory 491 The Medical Outcomes Study Pain 8. Mental Status Testing 394 Measures 496 The Oswestry Low Back Pain Measurements of Cognition, Cognitive Disability Questionnaire 498 Impairment, and Dementia 395 The Back Pain Classification Scope of the Chapter 396 Scale 501 The Dementia Rating Scale 399 The Pain and Distress Scale 506 The Cognitive Capacity Screening The Illness Behavior Examination 404 Questionnaire 508 The Clock Drawing Test 407 The Pain Perception Profile 514 The Alzheimer’s Disease Assessment Conclusion 517 Scale 411 The Information-Memory- Concentration Test 417 10. General Health Status and Quality The Dementia Scale 420 of Life 520 The Mental Status Measuring Quality of Life 520 Questionnaire 423 Scope of the Chapter 522 The Short Portable Mental Status The Arthritis Impact Measurement Questionnaire 426 Scales 526 The Mini-Mental State The Physical and Mental Examination 429 Impairment-of-Function The Modified Mini-Mental State Evaluation 535 Test 441 The Functional Assessment The Informant Questionnaire on Inventory 538 Cognitive Decline in the The Functional Living Index— Elderly 449 Cancer 541 The Clifton Assessment Procedures The Functional Assessment of for the Elderly 456 Cancer Therapy 546 The Cambridge Mental Disorders of The EORTC Quality of Life the Elderly Examination 460 Questionnaire 551 Conclusion 464 The Quality-Adjusted Time Without Additional Instruments 465 Symptoms and Toxicity Screening Tests 465 Method 559 Instruments for Clinical The Quality of Life Index 564 Application 466 The COOP Charts for Primary Care Diagnostic Instruments 467 Practices 569 Single-Item Health Indicators 578 9. Pain Measurements 470 The Functional Status Theoretical Approaches to Pain 470 Questionnaire 587 Approaches to Pain Measurement 472 The Duke Health Profile 591 xii Contents The OARS Multidimensional The Quality of Well-Being Functional Assessment Scale 675 Questionnaire 596 The Health Utilities Index 683 The Comprehensive Assessment and The EuroQol EQ-5D Quality of Referral Evaluation 604 Life Scale 694 The Multilevel Assessment Conclusion 702 Instrument 609 The Self-Evaluation of Life Function Scale 611 11. Recommendations and McMaster Health Index Conclusions 704 Questionnaire 612 The Current Status of Health The World Health Organization Measurement 704 Quality of Life Scale 619 Guidelines for Developing Health The Sickness Impact Profile 630 Measurements 705 The Nottingham Health Profile 639 Final Remarks 709 The Short-Form-36 Health Survey 649 The Short-Form-12 Health Glossary of Technical Terms 711 Survey 666 The Disability and Distress Scale 671 Index 719  List of Exhibits Exhibit 3.1 Summary Chart from the Exhibit 3.20 The Health Assessment Original PULSES Profile 64 Questionnaire 113 Exhibit 3.2 The PULSES Profile: Revised Exhibit 3.21 The Medical Outcomes Study Version 65 Physical Functioning Measure 121 Exhibit 3.3 The Barthel Index 67 Exhibit 3.22 The Functional Autonomy Exhibit 3.4 Instructions for Scoring the Measurement System 124 Barthel Index 68 Exhibit 3.23 The Functional Exhibit 3.5 Scoring and Guidelines for the Independence Measure: Items and 10-Item Modified Barthel Index 69 Levels of Function 143 Exhibit 3.6 Scoring for the 15-Item Exhibit 3.24 Functional Independence Modified Barthel Index 71 Measure (FIM) 145 Exhibit 3.7 The Index of Independence in Exhibit 4.1 Format of the Social Activities of Daily Living: Evaluation Relationship Scale 156 Form 75 Exhibit 4.2 The Social Support Exhibit 3.8 The Index of Independence in Questionnaire 159 Activities of Daily Living: Scoring and Exhibit 4.3 The RAND Social Health Definitions 76 Battery 163 Exhibit 3.9 The Sister Kenny Institute Self- Exhibit 4.4 Scoring Method for the RAND Care Evaluation 79 Social Health Battery 164 Exhibit 3.10 The Physical Self-mainte- Exhibit 4.5 The Medical Outcomes Study nance Scale 85 Social Support Survey 167 Exhibit 3.11 The Lawton and Brody IADL Exhibit 4.6 The Duke-UNC Functional Scale 86 Social Support Questionnaire 169 Exhibit 3.12 The Disability Interview Exhibit 4.7 The Duke Social Support and Schedule 90 Stress Scale 171 Exhibit 3.13 The Lambeth Disability Exhibit 4.8 The Katz Adjustment Scale Screening Questionnaire (Version 3) 92 Form R2 (Level of Performance of Exhibit 3.14 The OECD Long-Term Socially Expected Activities) and Form Disability Questionnaire 93 R3 (Level of Expectations for Exhibit 3.15 The Functional Status Rating Performance of Social Activities) 175 System 96 Exhibit 4.9 Example of a Section from the Exhibit 3.16 The Rapid Disability Rating Social Functioning Schedule 178 Scale-2 98 Exhibit 4.10 Example of an Item from Exhibit 3.17 The Functional Status the Interview Schedule for Social Index 101 Interaction 181 Exhibit 3.18 The Patient Evaluation Exhibit 4.11 The Social Adjustment Conference System 105 Scale—Self-Report 185 Exhibit 3.19 The Functional Activities Exhibit 4.12 Structure and Content of the Questionnaire 109 Social Maladjustment Schedule 195 xiii xiv List of Exhibits Exhibit 4.13 The Social Dysfunction Exhibit 6.4 The Hospital Anxiety and Rating Scale 198 Depression Scale 295 Exhibit 4.14 Scope of the SSIAM Showing Exhibit 6.5 The Zung Anxiety Status Arrangement of Items within Each Inventory (ASI) 303 Section 201 Exhibit 6.6 The Zung Self-Rating Anxiety Exhibit 4.15 An Example of Two Items Scale (SAS) 305 Drawn from the SSIAM: Social and Exhibit 6.7 The Depression Anxiety Stress Leisure Life Section 202 Scales 315 Exhibit 5.1 An Example of a Two- Exhibit 6.8 Examples of items from the Dimensional Conceptual Model of State-Trait Anxiety Inventory 321 Mood 209 Exhibit 7.1 The Zung Self-Rating Exhibit 5.2 The Original Version of the Depression Scale 346 Health Opinion Survey 212 Exhibit 7.2 The Center for Epidemiologic Exhibit 5.3 Main Variants of the Health Studies Depression Scale Opinion Survey 213 (CES-D) 352 Exhibit 5.4 Langer’s Twenty-Two Item Exhibit 7.3 The Geriatric Depression Screening Score of Psychiatric Scale 360 Symptoms 217 Exhibit 7.4 Depression Adjective Check Exhibit 5.5 The Affect Balance Lists, Form E 365 Scale 222 Exhibit 7.5 The Hamilton Rating Scale for Exhibit 5.6 The Positive and Negative Depression 370 Affect Scale 227 Exhibit 7.6 The BAS-D Brief Assessment Exhibit 5.7 The Life Satisfaction Scale for Depression 379 Index A 232 Exhibit 7.7 The Even Briefer Assessment Exhibit 5.8 The Philadelphia Geriatric Scale for Depression (EBAS DEP) 380 Center Morale Scale 238 Exhibit 7.8 The Montgomery-Åsberg Exhibit 5.9 The General Well-Being Depression Rating Scale 383 Schedule 241 Exhibit 7.9 The Carroll Rating Scale for Exhibit 5.10 The General Well-Being Depression 389 Schedule: Subscore Labels and Question Exhibit 8.1 Content of the Dementia Topics 243 Rating Scale 400 Exhibit 5.11 The RAND Mental Health Exhibit 8.2 The Cognitive Capacity Inventory, Showing Response Scales Screening Examination 405 and Factor Placement of Each Item Exhibit 8.3 Alzheimer’s Disease 249 Assessment Scale: Summary Exhibit 5.12 The Health Perceptions of Items 412 Questionnaire, Showing the Items Exhibit 8.4 The Information-Memory- Included in Each Subscore 255 Concentration Test 418 Exhibit 5.13 The General Health Exhibit 8.5 The Dementia Scale Questionnaire (60-item Version) 260 (Blessed) 422 Exhibit 5.14 Abbreviated Versions of the Exhibit 8.6 The Mental Status General Health Questionnaire 263 Questionnaire 423 Exhibit 6.1 The Taylor Manifest Anxiety Exhibit 8.7 Short Portable Mental Status Scale 281 Questionnaire 426 Exhibit 6.2 The Hamilton Anxiety Rating Exhibit 8.8 The Mini-Mental State Scale 287 Examination 431 Exhibit 6.3 The Structured Interview Exhibit 8.9 The Modified Mini-Mental Guide for the Hamilton Anxiety Rating State Examination 442 Scale 289 Exhibit 8.10 The IQCODE 450  List of Exhibits xv Exhibit 8.11 Summary Content of the Clinical Course of Cancer Described in Clifton Assessment Procedures for the the Q-TWiST 560 Elderly 457 Exhibit 10.8 Utility Plot for a Q-TWiST Exhibit 8.12 Summary of the Content Comparison of Two Treatments 561 of the CAMCOG Section of the Exhibit 10.9 Q-TWiST Gain Function in CAMDEX 462 Comparing Two Treatments 562 Exhibit 9.1 Alternative Formats for Visual Exhibit 10.10 The Quality of Life Index: Analogue Rating Scales Tested by Clinician Rating Version 566 Huskisson 478 Exhibit 10.11 The Quality of Life Index: Exhibit 9.2 Formats of the Numerical Self-Assessment 567 Rating (NRS) and Visual Analogue Exhibit 10.12 The Dartmouth COOP Scales (VAS) as Used by Downie Charts 571 et al. 480 Exhibit 10.13 The Delighted-Terrible Exhibit 9.3 The McGill Pain Scale 579 Questionnaire 485 Exhibit 10.14 The Faces Scale 579 Exhibit 9.4 Scale Weights for Scoring the Exhibit 10.15 The Ladder Scale 580 McGill Pain Questionnaire 486 Exhibit 10.16 The Circles Scale 580 Exhibit 9.5 The Brief Pain Inventory 492 Exhibit 10.17 Examples of Single, Exhibit 9.6 The Medical Outcomes Study Summary Item Health Scales 581 Pain Measures 497 Exhibit 10.18 The Functional Status Exhibit 9.7 The Oswestry Low Back Pain Questionnaire 588 Disability Questionnaire 499 Exhibit 10.19 The Duke Health Exhibit 9.8 The Low Back Pain Symptom Profile 593 Checklist 503 Exhibit 10.20 Contents of the OARS Exhibit 9.9 The Back Pain Classification Multidimensional Functional Scale, Showing Discriminant Function Assessment Questionnaire 598 Coefficients 504 Exhibit 10.21 The OARS Exhibit 9.10 The Pain and Distress Multidimensional Functional Scale 507 Assessment Questionnaire: ADL and Exhibit 9.11 The Illness Behavior IADL Sections 600 Questionnaire 509 Exhibit 10.22 Indicator Scales of the Exhibit 9.12 The Pain Perception Profile: CORE-CARE 606 Lists of Pain Descriptors and Sample Exhibit 10.23 An Example of an Indicator Page from a Pain Diary, Including Lists Scale from the CORE-CARE 606 of Pain Descriptors 516 Exhibit 10.24 The Scope of the Multilevel Exhibit 10.1 The Arthritis Impact Assessment Instrument 610 Measurement Scales, Version 2 Exhibit 10.25 The McMaster Health (AIMS2) 528 Index Questionnaire 614 Exhibit 10.2 The Physical and Mental Exhibit 10.26 Domains, Facets, and Items Impairment-of-function Evaluation 536 Included in the WHOQOL-100 and Exhibit 10.3 The Functional Assessment WHOQOL-BREF 622 Inventory: Summary of Ratings 539 Exhibit 10.27 The Sickness Impact Profile: Exhibit 10.4 The Functional Living Categories and Selected Items 632 Index—Cancer 542 Exhibit 10.28 The Nottingham Health Exhibit 10.5 The FACT-G, Version 4 548 Profile 641 Exhibit 10.6 The EORTC Quality of Exhibit 10.29 The Short-Form-36 Health Life Questionnaire (QLQ-C30), Survey, Version 2 651 Version 3.0 552 Exhibit 10.30 The Short-Form-12 Health Exhibit 10.7 The Three Stages of the Survey, Version 2 667 xvi List of Exhibits Exhibit 10.31 The Disability and Exhibit 10.35 The HUI3 Classification Distress Scale: Descriptions of States of System (See Exhibit for Questions Illness 672 from Which the Classification Is Exhibit 10.32 The Disability and Distress Derived) 685 Scale: Valuation Matrix for Health Exhibit 10.36 HUI3 Single Attribute States 672 Utility Functions 687 Exhibit 10.33 Dimensions, Function Exhibit 10.37 HUI3 Multi-Attribute Levels, and Weights of the Quality of Utility Function 687 Well-Being Scale 676 Exhibit 10.38 The EuroQol Quality of Exhibit 10.34 Symptom and Problem Life Scale 696 Complexes (CPX) for the Quality of Exhibit 10.39 Dimensions and Weights Well-Being Scale 678 for the EuroQol 697 Measuring Health This page intentionally left blank  1 Introduction Background reviews and copies of the questionnaires are available to paid subscribers. The Buros Mental The first edition of this book was written because Health Measurement Yearbooks are now avail- clinicians and researchers often seemed unaware able online and can be searched for a fee of the wide variety of measurement techniques at http://buros.unl.edu/buros/jsp/search.jsp. A available in health services research. This was compendium of quality of life measures can unfortunate, because research funds are wasted be purchased from www.euromed.uk.com/ when studies do not use the best measurements qolcompendium.htm. Another large collection of available, and less scientific evidence is accumu- quality of life measures assembled by Salek is lated if different methods are used in different available on CD or on the Internet at http://www studies. In addition to serving as a guide to avail- .wiley.com/legacy/products/subject/reference/salek able measures, the book also included several _index.html criticisms of the current state of development of In succeeding chapters, we review a selection health measurement overall. of the leading health measurement methods. In the years since the first edition was pub- Each chapter includes a critical comparison be- lished, progress has been made in consolidating tween the methods examined to help readers the field of health measurement. It remains true choose the method most suitable for their pur- that the quality of health measurements is some- poses. Descriptions in this book are intended to what uneven, but several promising signs are vis- be sufficiently detailed to permit readers to apply ible. In place of the enthusiastic proliferation of and to score the instrument, although in many hastily constructed measures that typified the cases a manual is available from the original au- 1970s, attention is being paid to consolidating thors of the instrument that supplies additional information on a narrower range of quality in- information. struments. These are being used more consis- tently in growing numbers of studies, providing genuinely comparative information. Time may Selection of Instruments have come to remove our comment in the first for Review edition that bemoaned the tendency to reinvent Activities of Daily Living scales. Furthermore, The large and growing range of health measure- methodological studies that test the accuracy of ment instruments demands that any review book measurements now use more sophisticated ap- be selective. Making such a selection has proven proaches, and obvious statistical errors are rare. immensely difficult and ultimately includes a Finally, several books are available to help the large element of subjectivity. It is therefore desir- user locate the most suitable measurement scale able to explain some of the principles that guided (1–12). Web resources are increasingly available, the current selection. First, we have tried to focus including a data base (www.qolid.org) run by on measures of good quality: there seems little the MAPI Research Institute in Lyon that pro- point in reviewing an indifferent method when a vides brief summaries of over 450 instruments; clearly superior one is available. The focus has 3 4 Measuring Health therefore been on scales for which evidence for re- dence of reliability and validity. The occasional liability and validity has been published. Because exceptions to this are scales that hold particular it takes time to establish such evidence, new scales conceptual or methodological interest in the de- that appear promising, but for which we lack evi- velopment of the field. Inevitably, another author dence, have been set aside for subsequent edi- would have made a different selection and, tions. Only published methods are considered; equally inevitably, some readers will be irritated unpublished scales have been reviewed elsewhere, by the omission of a particular favorite scale. To for example by Ward and Lindeman (13) and by all these people, I apologize in advance, and wel- Bolton (14) and on many Web sites. Second, this come suggestions for improvement. book focuses primarily on generic instruments that can be applied to a range of populations and health conditions. We have not included, for ex- Structure of the Book ample, diabetes-specific quality of life measures, but Ann Bowling has published brief descriptions Because this book is intended for a broad audi- of a wide range of disease-specific instruments ence, including those not familiar with the (9). However, this book does describe many methodological bases for health measurement, generic quality of life instruments that could be Chapter 2 reviews the historical origins and de- used with patients with diabetes (and where ap- velopment of the field and outlines the theoretical propriate, this is indicated). Having said that the and technical foundations of health measurement focus lies on generic measures, some notable ex- methods. This discussion introduces the central ceptions may be found all the same. The chapters themes of validity and reliability and explains the on anxiety and depression were included because various approaches to assessing measurement it simply seemed impossible to omit these funda- quality. The explanations are intended to be ade- mental areas, and because such measures are quate for readers to understand the descriptions widely used in surveys, as outcome measures in of the measurement methods, but the chapter is clinical trials, and in general medical practice. not intended to serve as a text on how to develop The book does omit psychological measurements a measurement scale; Streiner and Norman is a with a clinical focus such as neuropsychological better choice for that (17). A glossary of techni- tests or diagnostic batteries such as the Compos- cal terms at the end of this book helps readers ite International Diagnostic Interview (CIDI) in find definitions of technical terms without need- part because they have been described elsewhere ing to hunt through Chapter 2. (4; 15; 16), but also because they are so numer- Chapters 3 through 10 are the heart of the ous that they would fill a volume of their own. Fi- book, presenting detailed reviews of the instru- nally, the book does not include complete survey ments. Each chapter opens with a brief historical questionnaires or questions on risk factors such overview of the measurement techniques in that as smoking or alcohol consumption. The scales branch of practice. This is intended to illustrate that we do discuss are commonly incorporated the common themes that link the measurement within a broader survey questionnaire. They methods, because these are seldom developed in cover the following topics: physical disability and isolation from each other. The overview is fol- handicap, psychological well-being, social health, lowed by a table that gives a summary compari- depression, anxiety, mental status, pain, quality son of the measures reviewed. These tables are of life, and overall health status. intended to assist the reader in selecting the most Within each of these fields we have been suitable scale for a particular application. The ac- highly selective, attempting to review the best tual reviews of the measurements are loosely or- methods available. The definition of “best” re- dered from the simpler to the more complex lied principally on the evidence for the validity within each chapter. This is intended to reflect and reliability of each measurement. We there- the general evolution of the methods and corre- fore considered only measurements for which spond to a roughly chronological ordering and to published information is available, including evi- aid the reader in selecting a method of appropri-  Introduction 5 ate length and scope. What level of detail to pre- provide a factually accurate overview of each sent in each review is insoluble. The aim has been method; subjective statements or opinions are re- to provide enough information to permit the stricted to the Commentary section of each re- reader an informed choice among instruments. view. By the same token, we have avoided Other review books provide briefer reviews that repeating the interpretations of authors concern- merely repeat the contents of the article that orig- ing the validity of their scales; virtually all au- inally described an instrument; the goal here is to thors claim their method to be superior to the provide a representative picture of studies that average, so it seems simplest to let the statistics have evaluated each measure, and consequently speak for themselves. It is also perennially true our reviews vary in length. The accuracy and that the original authors report better reliability completeness of each review have been ensured and validity results than subsequent users do. in almost every case by having been checked by the person who developed the method originally. The conclusion of each chapter gives a brief Format for the Reviews summary of current thinking in that area of measurement and suggests directions for further A standard format is followed in reviewing each developmental work. The concluding section measurement. It should be stressed, again, that also mentions other measurements thought to although we have written the reviews, each was have merit but for which we did not include a checked for accuracy and completeness by the formal review because of lack of space or insuf- person who originally developed the method, or ficient evidence concerning their quality. by an acknowledged expert, to ensure that we The experience of writing these reviews made are providing an authoritative description of it clear that for some measurements remarkably each measurement. The following information is basic information is lacking: for whom precisely given for each measurement. is the measurement intended? How is it to be in- terpreted? How valid is it? Exactly how is the Title measurement to be administered? Because of this, the book concludes in Chapter 11 by offer- The title of each method is that given by the orig- ing some guidelines that might be followed by inal author of the instrument. those who develop and publish descriptions of health measurement methods. Author The attribution of each method to an author is Style and Content of the Reviews primarily for convenience; we recognize that most methods are developed by a team. In certain The reviews do more than merely reproduce ex- cases, additional authors are cited where they isting descriptions of the various methods, be- have had a continuing involvement in the devel- cause it is remarkable how often there are errors opment of the method. in the published descriptions. Where inconsisten- cies were found, such as between different ver- Year sions of a scale, we have sought the guidance of the original author about which version is cor- This is the year the method was first published, rect. Where possible, unclear statements in the followed by that of any major revisions. original publications have been elucidated through discussion with their authors. We have Purpose tried to avoid technical terms and jargon, but because some technical terms have to be used, The purpose of the measurement is summarized we have defined these in the Glossary of Techni- in our own words, based as far as possible on the cal Terms at the end of the book. The reviews description given by the original authors. We 6 Measuring Health have indicated the types of patient or client the the Description section of the method. The Al- method is intended for (specifying age, diagnos- ternative Forms section covers other variants and tic group) where this was stated. All too fre- translations. Again we have been selective: for quently the precise purpose of a measure is not some methods there has been a proliferation of made clear by its author; occasionally, it is re- less frequently used, minor variants that should stated differently in different publications and we not be encouraged. These have been ignored. have tried to resolve such inconsistencies. Reference Standards Conceptual Basis Where available, these provide normative infor- Where specified by the original author, this indi- mation with which the results of the user’s study cates which theoretical approach was taken to may be compared. measure topics described in the Purpose section. Commentary Description Descriptions of each measurement are as objec- The description indicates the origins and develop- tive as possible; the Commentary section offers ment of the instrument and shows the question- some remarks on the strengths and weaknesses naire or rating scale where space permits. Details of the method and on how it compares with oth- of administration and scoring are given. Where ers with a similar focus. In conjunction with the several versions of the instrument exist, we have summary tables at the beginning of each chapter, sought the advice of the author of the method, this is intended to help the reader choose between and in general present the most recent version. alternative measurements and to suggest where further developmental work may be carried out. Exhibits Address Within the Description section, we reproduce (where permitted) a copy of the questionnaire or For some scales, there is a mailing or Web site rating scale. Occasionally, where space does not address of a contact person from whom further permit us to show an entire instrument, we in- information may be obtained. This has been clude one or two sections of it and indicate done in cases where permission is required to use where the complete version may be obtained. a scale, where the user’s manual is not published, and where we have not reproduced a copy of the instrument. Reliability and Validity For most instruments we have summarized all in- References formation available when our reviews were being prepared. For a few scales (e.g., psychological Rather than list all available references to each well-being and the depression scales) data on va- method, we cite only those thought to provide lidity are so extensive that we have been selective. useful information on the instrument or on its Most of our information was taken from published validity or reliability. We have not cited studies sources, at times supplemented or corrected fol- that merely used the method but did not report lowing correspondence with the original author. on its quality. If needed, such references can be identified through the Science Citation Index. Alternative Forms Summary Tables Different versions exist for many of these mea- surements. Where revised versions have come to The Summary Table at the end of the introduction be widely accepted and used, we include these in to each chapter compares relevant characteristics  Introduction 7 of the measurements reviewed in that chapter. validity testing reported in nu- This serves as a consumer’s guide to the various merous studies. methods and gives the following information: Because the thoroughness of testing may be independent of the results obtained, two 1. The numerical characteristics of the scale: other ratings summarize results of the reli- nominal, ordinal, interval, or ratio. ability and validity testing: 2. The length of the test, as indicated by the 0 = No numerical results reported number of items it contains. This refers to ? = Results were not stated or are un- the full version of each scale; for many, interpretable abbreviated forms exist, as described in * = The evidence suggests weak relia- the reviews. bility or validity 3. The applications of the method: clinical, * * = Adequate reliability or validity research, survey, or as a screening test. * * * = Excellent reliability or validity: 4. The method of administering the scale: higher coefficients than those nor- self-administered, by an interviewer or mally seen in other instruments. trained staff member, or requiring an ex- pert rater (e.g., physician, physiotherapist As health measurement develops over time, more or psychologist). Where the length of time and more scales undergo extensive validity and needed to administer the scale was re- reliability testing. Compared with earlier edi- ported, this is indicated. tions, therefore, ratings for a few scales have 5. We have made a rating that indicates how been reduced to reflect their status relative to the widely the method has been used be- current standards of testing. cause, other things being equal, it will be advantageous to select a commonly used measurement technique. The rating refers Evaluating a Health Measurement: to the number of separate studies in The User’s Perspective which the method has been used rather than to the number of publications that Finally, we recognize that everyone would like a describe the method, because one study guide book to make recommendations about the may give rise to a large number of re- “best buy.” This, of course, is a difficult judgment ports. Three categories indicate how to make without knowing about the study in widely the scale has been used: a few which the reader intends to use the method. We (one to four) published studies have used give several indications of the relative merits of the method, several (five to twelve) stud- the scales, but all methods we review have differ- ies by different groups, or many studies ent strengths, so we can only make suggestions to (more than a dozen studies). the reader as to how to use the information in this 6. Four ratings summarize evidence of relia- book to choose an appropriate measurement. bility and validity. The first and third sum- The user must decide exactly what is required marize the thoroughness of reliability and of the measurement. For example, will it be used validity testing: to evaluate a program of care or to study indi- 0 = No reported evidence of reliabil- vidual patients? What type of patient will be as- ity or validity sessed (e.g., diagnosis, age group, level of * = Basic information only; informa- disability)? What time frame must the assess- tion only by the original authors ment cover: acute or long-term conditions? How of the scale broad-ranging an assessment must be made, and * * = Several types of test, and several how detailed does the information need to be? studies by different authors have For example, would a single rating of pain level reported reliability or validity suffice, or is a more extensive description of the * * * = All major forms of reliability or type as well as the intensity of the pain needed? 8 Measuring Health Bear in mind that this may require 15 minutes of 6. What degree of change can be detected by the patient’s time and that a person in pain will the method, and is this adequate for the be unenthusiastic about answering a lengthy purpose? Does the method detect qualita- questionnaire. The user must decide, in sum, on tive changes only, or does it provide quan- the appropriate balance to strike between the de- titative data? Might it produce a false- tail and accuracy required, and the effort of col- negative result due to insensitivity to lecting it. This information can be gleaned from change (e.g., in a study comparing two the tables at the beginning of each chapter. types of therapy)? Is it suitable as a screen- Turning to how the user evaluates the pub- ing test only, or can it provide sufficiently lished information on a measurement method, detailed information to indicate diag- the following characteristics of a method should noses? be considered: 7. How strong is the available evidence for reliability and validity? How many differ- 1. Is the purpose of the method fully ex- ent forms of quality testing have been car- plained and is it appropriate for the in- ried out? How many other indices has it tended use? The method should have been been compared with? How many different tested on the type of person who will be users have tested the method, and did they taking it. obtain similar results? How do these com- 2. Is the method broad enough for the in- pare to the quality of other scales? tended application, asking neither too many nor too few questions? Is it capable One difficulty commonly encountered in com- of identifying levels of positive health paring two indices is where one shows excellent where this is relevant? validity results in one or two studies and the 3. What is the conceptual approach to the other is more widely tested but shows somewhat measurement topic? For example, which less adequate validity. The reader should pay at- theory of pain does it follow, and is its ap- tention to the size of the validation studies: fre- proach consonant with the orientation of quently, apparently excellent results obtained the study? Is the theory well-established from initial smaller samples are not repeated in (e.g., Maslow’s hierarchy of needs) or is it larger studies. Ultimately the selection of a mea- an idiosyncratic notion that may not cor- surement contains an element of skill and even respond to a broader body of knowledge? luck; it is often prudent to apply more than one 4. How feasible is the method to administer? measurement. This has the advantage of rein- How long does administration take? Can forcing the conclusions of the study when the re- it be self-administered? Is professional ex- sults from ostensibly similar methods are in pertise required to apply or interpret the agreement, and it also serves to increase our gen- instrument? Does it use readily available eral understanding of the comparability of the data (e.g., information already contained measurements. in medical records) and will the measure be readily acceptable to respondents? References What response rates have been achieved using the method? Is the questionnaire (1) Mitchell JV. Mental measurements readily available? Is at free? Above all, yearbook. Lincoln, Nebraska: Buros Institute of Mental Measurements, 1985. does the instruction manual clearly spec- (2) Keyser DJ, Sweetland RC. Test critiques. ify how the questions should be asked? Kansas City, Missouri: Test Corporation of 5. Is it clear how the method is scored? Is the America, 1984. numerical quality of the scores suited to (3) Corcoran K, Fischer J. Measures for the type of statistical analyses planned? If clinical practice: a sourcebook. New York: the method uses an overall score, how is Free Press, 1987. this to be interpreted? (4) Sweetland RC, Keyser DJ. Tests: a  Introduction 9 comprehensive reference for assessments in Functional status measurement in primary psychology, education and business. care. New York: Springer-Verlag, 1990. Kansas City, Missouri: Test Corporation of (12) Salek S. Compendium of quality of life America, 1983. instruments. Chichester, U.K.: J. Wiley & (5) Bellamy N. Musculoskeletal clinical Sons, 1998. metrology. Dordrecht, the Netherlands: (13) Ward MJ, Lindeman CA. Instruments for Kluwer Academic Publishers, 1993. measuring nursing practice and other (6) Spilker B, ed. Quality of life assessment in health care variables. Vol. 2. Washington, clinical trials. New York: Raven Press, 1990. DC: Department of Health, Education and (7) Wilkin D, Hallam L, Doggett MA. Mea- Welfare (DHEW Publication No. HRA 78- sures of need and outcome for primary 54), 1978. health care. Oxford: Oxford University (14) Bolton B. Measurement in rehabilitation. Press, 1992. In Pan EL, Newman SS, Backer TE, et al., (8) Bowling A. Measuring health: a review of eds. Annual review of rehabilitation. Vol. quality of life measurement scales. 2nd ed. 4. New York: Springer, 1985: 115–144. Buckingham, U.K.: Open University Press, (15) Mittler P, ed. The psychological assessment 1997. of mental and physical handicaps. London: (9) Bowling A. Measuring disease: a review of Tavistock, 1974. disease-specific quality of life measurement (16) Sajatovic M, Ramirez LF. Rating scales in scales. 2nd ed. Buckingham, U.K.: Open mental health. Hudson, Ohio: Lexi-Comp, University Press, 2001. 2001. (10) Jenkinson, C, ed. Measuring health and (17) Streiner DL, Norman GR. Health measure- medical outcomes. London: University ment scales: a practical guide to their College London Press, 1994. development and use. 3rd ed. New York: (11) WONCA Classification Committee. Oxford University Press, 2003.  2 The Theoretical and Technical Foundations of Health Measurement or more than 100 years, Western nations measures of population health is described by F have collected statistical data characterizing social conditions. These data (e.g., birth and Murray et al. (5–7). Debate will probably always continue about death rates, education, crime, housing, employ- how best to measure health, in part because of ment, economic output) reflect issues of public the complexity and abstract nature of health it- concern and have often become the focal point self. The fundamental problem is that, as with for movements of social reform. Measurements attitudes or motivation, health cannot be mea- of health have always formed a central compo- sured directly, like length or weight; instead, the nent in such public accounting; they are used process of its measurement is indirect and in- to indicate the major health problems con- volves several steps. The first requires agreement fronting society, to contribute to the process of on a definition of what is to be measured: what setting policy goals, and to monitor the effec- does the concept health include? Should the def- tiveness of medical and health care. Rosser inition be broad or a narrow? How does health offers some very early examples of such ac- relate to quality of life or to well-being? Some counting, drawn from ancient Egyptian and prefer to keep the concept of health somewhat Greek times (1; 2). imprecise, so that it can be reformulated to re- Social indicators of this kind are based on flect changing social circumstances, while others aggregated data expressed as regional or na- define it in operational terms, which often tional rates; they are intended to give a picture means losing subtle shades of meaning (8). of the status, including the health status, of Measurement next implies assembling a selec- populations rather than of individual people. tion of indicators to represent the conception of In addition, many indicators of the health or health. Health indicators may be of many types, well-being of individuals have been developed; ranging from a specimen analyzed in a labora- these are principally used to contrast differ- tory to the flexion of a limb observed by a phys- ences in health between people, to diagnose ill- iotherapist, from estimates of working capacity ness, to predict the need for care, and to to expressions of personal feelings. Next, fol- evaluate the outcomes of treatment (3). Even lowing the quantitative tradition of science, nu- though both population and individual health merical scores are assigned to the indicators. indicators are necessary and both will continue Reflecting this multistage process, an often- to be developed, this book deals only with indi- quoted definition of measurement is “the assign- vidual indicators. Population health statistics ment of numbers to objects or events to (e.g., rates of morbidity and mortality, sum- represent quantities of attributes, according to mary measures of population health) are not rules” (9). The “objects or events” refer to the considered here; a brief review of them is given indicators that have been used to measure by Rosser (4) and the World Health Organiz- health, and these form the main focus of this ation’s international collaboration on summary book. 10  The Theoretical and Technical Foundations of Health Management 11 or prematurity, problems rarely encountered The Evolution of Health with high infant mortality. In a similar way, in- Indicators creased life expectancy in industrial countries raises the prevalence of disability in the popula- Indicators are deliberately chosen to reflect prob- tion (13; 14). In each case, the resolution of one lems of social concern or core values. Just as lan- health problem casts new issues into prominence guage molds the way we think, our health and reduces the usefulness of the prevailing measurements influence (and are influenced by) health indicator, necessitating its replacement by the way we define and think about health. Social others. It may also increase pressure to modify reforms are based on the information available the prevailing definition of health. to us, so the selection and publication of indica- The rising expectations of the past 150 years tors of health are actions that both reflect and have led to a shift away from viewing health in guide social and political goals. Hence, the pro- terms of survival, through a phase of defining it cess of measurement tends to influence the in terms of freedom from disease, thence to an health of the population; publication of an indi- emphasis on the person’s ability to perform his cator focuses attention on that problem, such as daily activities, and more recently to an emphasis infant mortality, and the resulting interventions on positive themes of happiness, social and emo- (if successful) tend to reduce the prevalence of tional well-being, and quality of life. Only where the problem, in turn reducing the value of the in- problems of high premature mortality are no dicator as a marker of current health problems. longer a pressing social concern does it become The identification of new concerns tends to raise relevant to think of health in the World Health a demand for new indices of health to monitor Organization’s (WHO) terms of “physical, men- progress toward the new goals, and so the cycle tal, and social well-being, and not merely the ab- begins again (10). Health indicators are in a con- sence of disease and infirmity” (15, p459). But tinuous state of evolution, and Erickson has de- here, again, measurements interact with scribed the life cycle of a measurement tool (11). progress. When it was introduced in the 1950s, The earliest measures of population health the WHO definition of health was criticized as used readily available numerical indicators such unmeasurable, but the subsequent development as mortality rates. Mortality is unambiguous and, of techniques to measure well-being contributed because death must be recorded by law, the data to the current wide acceptance of the definition. are generally complete. But as societies evolve, This book is concerned with the adequacy of health problems alter in salience and new health health measures: do the indicators in use success- indicators must be chosen to reflect changing fully reflect an explicit and accepted definition of health issues. As an example, the infant mortality health? This is the central theme of validity, the rate (IMR) is often used as an indicator of health assessment of what a measurement measures, levels in pre-industrial societies, where high rates and hence of how it can be interpreted. A valid are an important concern, and where reductions health index provides information about health can be relatively easily achieved. As IMR de- status, not about some other variable such as clines, however, a law of diminishing returns be- personality (3). There has long been concern gins to apply, and further reductions require over the interpretation of many of the statistical increasingly large expenditures of resources. As indicators of population health, and this stimu- the numerator becomes smaller, it also becomes lated the development of individual health mea- less representative as an indicator of the health of sures. For example, consultation rates with the broader population. The resolution of one physicians can only be interpreted as indicators type of health problem reveals a new layer of of health if we can determine how the numbers concerns, a process that Morris has called “the of services provided are related to health status. onion principle” (12). For example, as the IMR Do more consultations imply better health in a declines, growing numbers of survivors exhibit population, or that there is more illness to treat? health problems associated with low birth weight Consultation rates may form more valid indica- 12 Measuring Health tors of health expenditures than of health. Simi- of data, asking questions of a patient seemed to lar problems arise in interpreting indicators such be abundantly susceptible to bias. There was as rates of bed-rest days or work-loss days: these also the issue of completeness: mortality indica- may reflect a blend of health and the provision of tors are collected routinely of the whole popula- care, because improved access to care may result tion rather than of selected individuals. in activity restrictions ordered by physicians. Applying individual health measurements to Studies have, indeed, shown increases in disabil- whole populations is prohibitively expensive, al- ity days as the availability of medical services though questions on health were asked in the grows (16); without such care there might have census in Ireland as far back as 1851 (22). Grad- been less activity restriction but a greater risk of ually, however, indices of personal health that long-term damage to the patient’s health (17). relied on subjective judgments came to be ac- One way out of the dilemma in interpreting ag- cepted. The reasons for this included several gregated indicators is to ask questions of individ- methodological advances in survey sampling and ual, patients, providing a more direct reflection data analysis made at the time of World War II. of health. The questions are commonly called That conflict brought with it the need to assess “items,” simply because they may not be the physical and mental fitness of large numbers phrased as questions: some use rating scales and of recruits for military service, and indicators of others use statements with which the respondent the health of individuals applicable on a large may agrees or disagree. scale were accordingly developed and standard- For reasons of simplicity and cost, most ized. Wartime screening tests of physical capac- health measures rely on verbal report rather than ity later influenced the design of post-war observation, although in gerontology some em- questionnaires (22; 23), while the psychological phasize the benefits of assessments based on ob- assessment techniques developed by Lazarsfeld, serving actual performance (18–20). Subjective Guttman, Suchman and others during the war health measurements hold several advantages. formed the basis for the first generation of psy- They extend the information obtainable from chological well-being measurements in the post- morbidity statistics or physical measures by de- war years (see Chapter 5). After the war, survey scribing the quality rather than merely the quan- sampling techniques were refined by political tity of function. Subjective measures give scientists concerned with predicting voting be- insights into matters of human concern such as havior. This provided the technical basis for us- pain, suffering, or depression that could not be ing individual measurements to provide data deduced solely from physical measurements or representative of the larger population. Coming laboratory test results. They give information somewhat later, technical advances in data pro- about people whether they seek care or not, they cessing had a profound effect on the range of can reflect the positive aspects of good health, statistical analyses that could be applied to data. and they do not require invasive procedures or Computers greatly simplified the application of expensive laboratory analyses. They may also of- principal components or factor analysis in refin- fer a systematic way to record and present “the ing questionnaires; they also simplified the small, frantic voice of the patient” (21). Subjec- analysis and presentation of the voluminous in- tive measurements are, of course, little different formation collected in health questionnaires. from the data collected for centuries by physi- cians when taking a medical history. The impor- tant difference lies in the recent standardization Types of Health Measurements of these approaches and the addition of numeri- cal scoring systems. There are several ways to classify health mea- Despite these potential advantages of subjec- surements. They my be classified by their func- tive indicators, several problems delayed accep- tion, or the purpose or application of the tance of these instruments. Compared with the method; descriptive classifications focus on their inherent accuracy of mortality rates as a source scope, whereas methodological classifications  The Theoretical and Technical Foundations of Health Management 13 consider technical aspects, such as the tech- may be specific to a disease (e.g., a quality of life niques used to record information. scale in cancer patients), or to a particular type An example of a functional classification is of person (e.g., women’s health measures, pa- Bombardier and Tugwell’s distinction between tient satisfaction scales) or to an age group (e.g., three purposes for measuring health: diagnostic, child health indicators). Specific instruments are prognostic, and evaluative (24; 25). Diagnostic generally designed for clinical applications and indices include measurements of blood pressure intended to be sensitive to change after treat- or erythrocyte sedimentation rates and are ment. Generic instruments are commonly ap- judged for their correspondence with a clinical plied in descriptive epidemiological studies or diagnosis. Prognostic measures include screen- health surveys. They permit comparisons across ing tests, scales such as the Apgar score (26) and disease categories. In addition to specific and measures such as those that predict the likeli- generic measures, preference-based health mea- hood that a patient will be able to live indepen- sures may be distinguished (29). Whereas health dently following rehabilitation. Finally, evaluative status measures, whether generic or specific, indexes measure change in a person over time. record the presence and severity of symptoms or Kirshner and Guyatt also gave a functional clas- disabilities, preference-based measures record sification (27). In this, discriminative indexes the preferences of individual patients for alterna- distinguish between people, especially when no tive outcomes; this is relevant in policy analysis external criterion exists, as with IQ tests. Predic- and in predicting demands for care. Preference- tive indexes classify people according to some based measures generally combine several as- criterion, which may exist in the present (hence pects of health into a common numerical index equivalent to Bombardier’s diagnostic measures) that allows comparisons between different types or in the future (equivalent to prognostic mea- of health programs. Drawing these categories to- sures). A simpler functional classification was gether, Garratt et al. divided measurements into proposed by Kind and Carr-Hill (28). Measure- dimension-specific measures (e.g., a depression ments monitor either health status or change in scale); disease or population specific measures health status, and they may do this for individu- (e.g., an asthma quality of life scale); generic als or for groups. Measuring the health status of measures that can be applied to different popula- individuals is the domain of the clinical inter- tions; individualised measures that allow respon- view; measuring change in the individual is the dents to make their own judgments of the purpose of a clinical evaluation. Measuring importance of the domains being assessed (as in health status in a group is the aim of a survey in- the Patient-Generated Index), and utility mea- strument, while measuring group change is the sures developed for economic evaluation that in- domain of a health index (28, Table 1). corporate health state preferences and provide Health measurements may also be classified a single index score (e.g., the EuroQol EQ5D) descriptively, according to their scope or the (30). range of topics they cover. A common distinc- Many methodological classifications of health tion is drawn according to the breadth of the measurements exist. There is the distinction, for concept being measured. These range from example, used in the subtitle of this book, con- narrow-focus measures that cover a particular trasting rating scales and questionnaires; there is organ system (e.g., vision, hearing); next are the distinction between health indexes and scales concerned with a diagnosis (e.g., anxiety health profiles. Cutting across these categories, or depression scales); then there are those that there is the more complex distinction between measure broader syndromes (e.g., emotional subjective and objective measures. In essence, the well-being); then come measurements of overall contrast between rating scales and question- health and, broadest of all, measurements of naires lies in the flexibility of the measurement overall quality of life. A common distinction is process. In a rating scale an expert, normally a that between broad-spectrum generic health clinician, assesses defined aspects of health, but measures and specific instruments. The latter sometimes the precise questions vary from rater 14 Measuring Health to rater and from subject to subject. An example point for emics is that language forms a unique is the Hamilton Rating Scale for Depression: aspect of culture, and that the goal of translation Hamilton gave only a general outline of the is to review the pertinence of an idea (here, a types of question to ask and the clinician uses questionnaire item) to the target culture, seeking personal judgment in making the rating. By con- a metaphor that is equivalent. Whereas the trast, in self-competed questionnaires and in in- nomothetic approach tackles “What?” ques- terview schedules the questions are preset, and tions, the idiographic considers the “Why?” As we carefully train interviewers not to alter the Millon and Davis point out, the two approaches wording in any way. The debates over which ap- need not be in conflict; the success of theoretical proach is better generate more heat than light; propositions is ultimately judged based on how they also reveal deeper contrasts in how we well they explain individual clinical observa- approach the measurement of subjective phe- tions, whereas idiographic assessments are nomena. Briefly, the argument in support of merely descriptive unless they proceed from structured questionnaires holds that standardi- some theoretical base (32). zation is essential if assessments are to be com- Applied to designing a measurement, the pared among individuals; this consistency is seen nomothetic approach assumes that a standard as a cornerstone of nomothetic science. This is set of measurement dimensions or scales is rele- concerned primarily with abstract constructs vant to each person being measured and that and the theoretical relations among them, such scoring procedures should remain constant for as the links between dementia and depression. each. Thus, for example, in measuring social The goal of nomothetic science is to generalize, support, it would not accept the idea that social and it is inherently taxonomic; based on deter- isolation might be perfectly acceptable, even ministic philosophy, it searches for underlying healthy, for certain people, although undesirable commonalities and downplays individual vari- for many. In reaction to this, the idiographic ap- ability. The use of factor analysis to create mea- proach is more flexible and allows differences in sures of the theoretical concept underlying a set measurement approach from person to person. of indicators would typify this approach. In the For example, we should not assume that word- fields of linguistics and translation, this corre- ing a question in the same way for every respon- sponds to the etic approach, in which translation dent provides standardized information: we is approached from outside and seeks to derive a cannot assume that the same phrase will be in- non-culture-specific presentation of the underly- terpreted identically by people of different cul- ing ideas, which are assumed to be universally tural backgrounds. What is important is to applicable. A good example of this approach to ensure that equivalent stimuli are given to each translating a questionnaire is given by Cella et al. person, and this is the forte of the skilled clini- who sought to ensure not only semantic, but cian who can control for differences in the use of also content, conceptual, criterion, and technical language in rating different patients. Not only equivalences of an instrument in English and may symptoms of depression vary from person Spanish versions (31). By contrast, the idio- to person, but the significance of a given symp- graphic approach to measurement focuses on tom may vary from one patient to another, so assessing individuals; it particularizes and em- they should not necessarily receive the same phasizes the complexity and uniqueness of each score. This type of approach has, of course, long person being assessed. It is inherently clinical been used clinically in psychiatry; and more for- and corresponds to qualitative research methods. mal approaches to developing equivalent mea- The idiographic philosophy argues that because surement approaches for different subjects each person is the unique product of a particular include the repertory grid technique. Briefly, this environment, we cannot fully understand people classifies people’s thoughts on two dimensions: through the application of universal principles. the elements or topics they think about, and the Idiographic approaches also mirror the emic ap- constructs, which define the qualities they use to proach to language and translation. The starting define and think about the elements. An inter-  The Theoretical and Technical Foundations of Health Management 15 view (e.g., rating a person’s subjective quality of The distinction between objective and subjec- life) would identify the constructs the respon- tive measures reflects that between mechanical dents identify in thinking about quality of life, methods based on laboratory tests and those in and then rate each of these in their current situa- which a person (e.g., clinician, patient, family tion. This permits a more fully subjective assess- member) makes a judgment that forms the indi- ment of quality of life than is possible using a cator of health. Ratings that involve judgments structured questionnaire. Methods of this type are generally termed “subjective” measurements, have been used in quality of life measurement, and we use the term in this sense here. By con- for example in the SmithKline Beecham Quality trast, objective measurements involve no human of Life scale (33), or by Thunedborg et al. (34). judgment in the collection and processing of in- The second methodological classification formation (although judgment may be required refers to two contrasting approaches to summa- in its interpretation). This distinction is often not rizing data collected by generic instruments. clear, however. Mortality statistics are com- Scores may be presented separately, to represent monly considered “objective,” although judg- the various aspects of health (e.g., physical emo- ment may be involved assigning a code to the tional), giving a health profile. Alternatively, the cause of death. Similarly, observing behaviors indicators may be combined into an overall score, only constitutes an objective measure if the ob- termed a health index. Supporters of the profile servations are recorded without subjective inter- approach argue that health or quality of life is in- pretation. Thus, climbing stairs may be herently multidimensional, and scores on the dif- considered an objective indicator of disability if ferent facets should be presented separately. it is observed and subjective if it is reported by When replies to several contrasting themes are the person. Note that the distinction between added together, there are many ways a respon- “subjective” and “objective” measurements does dent can attain an intermediate score, so these do not refer to who makes the rating: objectivity is not provide interpretable information. This re- not bestowed on a measurement merely because flects the philosophy of the Rasch measurement it is made by an expert (35). Nor should we as- model (see Glossary), which holds that items to sume that subjective measures are merely “soft”: be combined should cover one dimension only, in longitudinal studies, subjective self-ratings of with separate elements presented as a profile. health are consistently found to predict subse- Single scores may be of two kinds: a single quent mortality as well as, or better than, physi- indicator (e.g., serum cholesterol) or an index, cal measures (36). which is an aggregation of separate scores into a The questions that comprise many health single number like the Dow Jones Industrial Av- measures can be worded either in terms of per- erage or the consumer price index. Single indica- formance (“I do not walk at all”: Sickness Im- tors require no particular discussion; they pact Profile) or in terms of capacity (“I’m unable necessarily cover a limited part of the broader to walk at all”: Nottingham Health Profile). concept of health. A health index, however, con- This distinction reflects the contrast between ob- fronts head on the issue of combining different jective and subjective measurement, in that per- facets of health. Critics argue that this mixes ap- formance can be recorded objectively whereas ples and oranges, but proponents argue that assessments of capacity tend to be subjective. finding connections between dimensions is nec- Active debate continues between those who fa- essary in making real life decisions. A single vor performance wording and those who favor score is often needed to address dilemmas such capacity wording. In general, capacity wording as choosing between two treatments, one of gives an optimistic view of health, whereas per- which prolongs life but at the cost of significant formance is conservative. Proponents of perfor- adverse effects, while the other produces shorter, mance wording argue that it gives a truer picture but disability-free, survival. Index scores are of what the person actually does, and not what commonly used in economic analyses and in they think they might be able to do on a good policy decision-making. day if they try. Proponents of capacity wording 16 Measuring Health argue that performance may be restricted by ex- more than a crude approximation to measure- traneous factors such as opportunities or per- ment. Indeed, many health measurements are sonal choice, so that these questions confound crude and merely affix numbers to qualitative health status with environmental and other con- subjective judgments. However, this need not be straints and tend to give a falsely conservative so, as will be seen from some of the more sophis- impression of health problems (37, pp242–243). ticated instruments reviewed in this book. To Thus, old people with equal capacity who live in introduce the scientific basis for subjective mea- institutional care typically have less freedom surement, a brief introduction to psychophysics than those in the community, so they will tend to and psychometrics describes the procedures used be rated less healthy by performance wording to assign numerical scores to subjective judg- than capacity. To compensate for this, the intro- ments. The following sections presume a famil- duction to performance questions typically iarity with some basic statistical terms that are, stresses that responses should focus solely on however, defined in the Glossary. limitations that are due to health problems. This Arguments for considering subjective judg- is complex, however, because health problems ments as a valid approach to measurement de- commonly interact with other factors such as the rive ultimately from the field of psychophysics. weather, making it hard for the respondent to Psychophysical principles were later incorpo- figure out which factor influenced their perfor- rated into psychometrics, from which most mance. The general consensus is that both word- of the techniques used to develop subjective ings have merit in particular applications; measurements of health were derived. Psy- capacity wording more closely reflects underly- chophysics is concerned with the way in which ing impairments, whereas performance wording people perceive and make judgments about is close to a measure of handicap. The user must physical phenomena such as the length of a line, be aware of the potential distortions of each. the loudness of a sound, or the intensity of a A major contribution to enhancing the accep- pain: psychophysics investigates the characteris- tance of subjective measures came from the appli- tics of the human being as a measuring instru- cation of numerical scaling techniques to health ment. indices. Because subjective reports of health are The early search for a mathematical relation- not inherently quantitative, some form of rating ship between the intensity of a stimulus and its method was required to translate statements such perception was illustrated by the work of Gustav as “I feel severe pain” into a form suitable for Fechner, whose Elemente der Psychophysik was statistical analysis. The scaling techniques origi- published in 1860. Subjective judgments of any nally developed by social psychologists to assess stimulus, Fechner discovered, are not simple re- attitudes soon found application in health in- flections of the event. For example, it is easy for dexes. The use of these, and later of more sophis- us to feel the four-ounce difference between a ticated rating methods, permitted subjective one- and a five-ounce weight, but distinguishing health measurements to rival the quantitative a 40-pound weight from another four ounces strengths of the traditional indicators. heavier is much less certain. To discern the math- ematical form of the link between a physical stimulus and our perception of it, Fechner pro- Theoretical Bases for posed a method of scaling sensations based on Measurement: Psychophysics “just noticeable differences;” he recorded the and Psychometrics objective magnitude of just noticeable differ- ences at different levels of the stimulus. In many What evidence is there that subjective judgments cases, our perceptions are more attuned to de- may form a sound basis for measuring health? tecting small differences at lower levels of a stim- Set against the measurement tradition of the ex- ulus than they are at higher levels. A minor act sciences, it is by no means self-evident that difference between two weights may be notice- subjective reports can be considered as anything able when the weights are small, but with heav-  The Theoretical and Technical Foundations of Health Management 17 ier weights the just noticeable difference in- power law recognized that humans can make con- creases. In 1962, Stevens wrote: sistent, numerically accurate estimates of sensory stimuli. It agreed, also, that the relationship be- If you shine a faint light in your eye, you have tween stimulus and subjective response was not a sensation of brightness—a weak sensation, linear, but it differed from Fechner’s law in stating to be sure. If you turn on a stronger light, the that the exact form of the relationship varied from sensation becomes greater. Clearly, then, one sensation to another. This was described by an there is a relation between perceived bright- equation with a different exponent for each type ness and the amount of light you put in the of stimulus, of the general form: eye. . . . But how, precisely, does the output R = k × Sb, of the system (sensation) vary with the input (stimulus)? Suppose you double the stimulus, where R is the response, k is a constant, S the does it then look twice as bright? level of the stimulus, and b an exponent that typ- The answer to this question happens to be ically falls in the range 0.3 to 1.7 (39, p83; 40, no. It takes about nine times as much light to p25). When the exponent b is unity, the relation- double the apparent brightness, but this spe- ship between stimulus and response is linear, as cific question, interesting as it may be, is only proposed by Weber’s law. Conveniently, the ex- one instance of a wider problem: what are the ponent for judging short lengths is unity, so that input-output characteristics of sensory sys- a two-inch line is judged to be twice as long as a tems in general? Is there a single, simple, per- one-inch line. The varying exponents for other vasive psychophysical law? (38, p29) judgments imply that subjective perceptions of different types of stimulus grow at different, al- Over 100 years earlier, Fechner had con- though characteristic, rates. The exponent is an cluded that a geometric increase in the stimulus indicator of psychological sensitivity to the stim- as received by the senses produces an arithmetic ulus. The exponent for force of handgrip is 1.7, increase in conscious sensation. This relation- whereas that for sound pressure is 0.67. For the ship is conveniently expressed as a natural loga- latter, a doubling of decibels will typically be rithm, and details of the derivation of Fechner’s judged as only two thirds louder. Sensitivity to law are given, for example, by Baird and Noma electrical stimulation is much greater, with an (39). Fechner’s approach generally agreed with exponent of 3.5; this holds implications for de- empirical data, it was intuitively appealing, and scribing responses to pain. it also incorporated Weber’s earlier law of 1846, Considerable attention has been paid to vali- which proposed that the magnitude of just no- dating the power law; some of the most convinc- ticeable differences was proportional to the ab- ing evidence comes from a complex-sounding solute level of the stimulus. Fechner’s law technique called cross-modality matching. In re- became accepted, and psychophysics turned its search to establish the power law and to identify attention to other issues for more than 70 years. the characteristic exponents b, judgments of var- During this time, however, accumulated experi- ious stimuli were made by rating responses on mental investigations of how people judge the numerical scales (39, p82). Knowing the re- loudness of sounds, the intensity of an electric sponse exponents, in terms of numerical judg- shock, the saltiness of food, and other quantities ments, for different stimuli (e.g., loudness, showed that the logarithmic relationship be- brightness, pressure), arithmetical manipulation tween stimulus and response did not always ap- of these exponents can postulate how a person ply, but it proved hard to find a more adequate would rate one stimulus by analogy to another. mathematical formulation. Thus, in theory, a certain degree of loudness By 1962, Stevens proposed that psychophysics should match a predictable brightness or pressure had apparently succeeded, and the logarithmic ap- of handgrip—the cross-modality matching. Ex- proach came to be replaced by the more generally perimental testing of the predicted match could applicable power law (38). Like Fechner’s law, the then be used to test the internal consistency of 18 Measuring Health the power law. As it turned out, the experimental mation: “On a scale of 0 to 100, how severe is fit between observed and predicted values was re- your pain?” This magnitude estimation ap- markably close, often within only a 2% margin proach is illustrated by the visual analogue of error (40, pp27–31). This holds valuable im- scales reviewed in Chapters 9 and 10. However, plications for health measurement: people can this may be a difficult task; many people find ad- make numerical estimates of subjective phenom- jectives (e.g., mild, moderate, or severe) far more ena in a remarkably consistent manner, even natural. Measurement requires the assignment when the comparisons are abstract, indeed, more of numerical scores to such descriptions, and abstract than those involved in subjective health this is achieved by using one of many scaling measurements. The validation experiments also procedures. These assign a numerical score to confirmed that the exponent for line length was each answer category for each topic covered unity, which justifies the use of visual analogue (e.g., pain or difficulty climbing stairs); combin- scales to represent abstract themes such as inten- ing the scores for a given pattern of responses sity of pain or level of happiness. Finally, studies provides a numerical indicator of the degree of validating the power law suggested that people disability reported. Scaling methods can also be can make accurate judgments of stimuli on a ra- used in rating composite descriptions of a per- tio, rather than merely on an ordinal scale of son’s overall health status when they have been measurement; that is, people can accurately judge rated on several dimensions (able to walk freely, how many times stronger one stimulus is than but with vision difficulties, occasional inconti- another. Judgments of this type are termed nence, and mild anxiety). This implies weighting “magnitude estimation” and are used in creating both the severity of the individual elements and ratio-scaled measurements (see page 19). the relative importance of each in the overall Traditionally, psychophysics studied subjec- score, and the result produces a health index. tive judgments of stimuli that can be objectively Methods to establish numerical scores vary in measured on physical scales such as decibels or their complexity, but fundamental to all scaling millimeters of mercury. In the social or health methods is a distinction between four ways of sciences, by contrast, we often use subjective using numbers in measurement; these lie in a hi- judgments because no objective physical ways erarchy of mathematical adequacy. The lowest yet exist to measure the phenomena under con- level is not a measurement, but refers to a classi- sideration. Psychometrics concerns the applica- fication: nominal or categorical scales use num- tion of psychophysical methods to measuring bers simply as labels for categories (e.g., qualities for which there is no physical scale 1 = male, 2 = female). The assignment of numbers (41–43), and this forms a cornerstone in the de- is arbitrary, and no inferences can be drawn velopment of health measurement methods. from the relative size of the numbers used. The The following sections introduce two psycho- only acceptable mathematical expressions are metric issues in making and recording subjective A = B or A ≠ B. For the second type, ordinal estimates of health. How are numerical values as- scales, numbers are again used as labels for re- signed to statements describing levels of health? sponse categories and their assignment is again And, how far are subjective judgments influenced arbitrary, except that the numbers reflect the in- by personal bias, rather than giving an accurate creasing quantity of the characteristic being reflection of the actual level of health? measured. Responses are ordered in terms of magnitude and a sequential code is assigned to each. “Mild,” “moderate,” and “severe” disabil- Numerical Estimates of Health ity might be coded 1, 2, and 3, with the property A < B < C. There are many limitations to this ap- proach, and Bradburn and Miles gave a critical Scaling Methods review (44). It is not an absolute rating: because The simplest way to quantify estimates of people use adjectives in different ways we cannot healthiness is to ask directly for a numerical esti- assume that “mild” implies the same thing to  The Theoretical and Technical Foundations of Health Management 19 different people, nor that “often” implies the and to calculate averages. It is not, however, pos- same frequency when referring to common sible to state how many times greater one tem- health problems as when referring to rare ones*. perature is than another: 40° is not twice as hot Thus, the actual value of the numbers in an ordi- as 20°. This is, however, the distinguishing fea- nal scale and the distance between each hold no ture of the fourth type of scale, the ratio scale. intrinsic meaning: a change from scale point 3 to The key here is a meaningful zero point, making point 2 is not necessarily equivalent to a change it possible to state that one score is, for example, from 2 to 1. Because of this, it is not strictly ap- twice another; this may be expressed as A × B = C propriate to subtract the ordinal scores taken be- and C/B = A. This is straightforward when num- fore and after treatment to compare the progress bers are used in measuring physical characteris- made by different patients. Nor is it appropriate tics such as weight or time, but things get more to combine scores by addition: this might imply, complicated when numbers are used to represent for example, that a mild plus a moderate disabil- abstract concepts. Here, variables that satisfy the ity is equivalent to a severe disability. This is not numerical requirements of an interval or a ratio to say that adding or subtracting ordinal scales scale should not necessarily be considered as cannot be done—it is frequently done. Purists such: classifying a scale as ordinal or interval de- may criticize (46; 47) but pragmatists argue that pends more on the way it will be interpreted the errors produced are minor (43, pp12–33). than on its inherent numerical properties. For Adding ordinal answer scales may lead to incor- example, age is often treated as an interval scale, rect conclusions and this is the main motivation but if age is being used, for example, as an indi- for developing more accurate scale weights for cator of maturation, the changing rate of growth answer categories. As a crude approach, some around puberty challenges the interpretation of ordinal scales deliberately leave gaps between the age as an interval scale. Rasch analysis (see page numerical codes to better represent the pre- 21) offers an approach to assessing whether a sumed distance between categories: see the 6, 5, measurement can be interpreted as an interval 3, 0 scoring in the Barthel Index (Exhibit 3.6). scale; an example is given by Silverstein et al. Cox and Wermuth have outlined methods for as- (49). sessing the linearity of ordinal scales (48), and In constructing a health measurement, scal- item response analyses (described later in this ing procedures may be used to improve the nu- chapter) can be used to transform ordinal mea- merical characteristics of response scales, sures into interval scales. typically by converting ordinal response codes Adding and subtracting scores is permissible such as 1, 2, 3 to interval scales. The scaling pro- with the third type of numerical scale. An inter- cedures are of several types that fall into the val scale is one in which numbers are assigned to broad categories of psychometric and economet- the response categories in such a way that a unit ric methods. Both involve samples of people who change in scale values represents a constant make judgments in a scaling task. Psychometric change across the range of the scale. Tempera- scaling procedures were originally developed to ture in degrees Celsius is an example. Here 30° − rate feelings, opinions, or attitudes; they record 20° = 50° − 40°, so it is possible to interpret values and concern the present. Values refer to differences in scores, to add and subtract them, preferences expressed under conditions of cer- tainty: these are underlying perceptions and *A nice illustration comes from the WHO’s work on there are no probabilities or choices in the situa- developing the WHOQOL measure. Szabo described a tion to which the weights are applied. When ap- study to establish the equivalence between response plied to health measures, psychometric scaling categories in nine regions of the world (45, Table 6). methods are used mainly to provide scores for For example, “quite often” in England appeared questions that refer to current health status. By equivalent to “often” in India, “from time to time” in Zambia, “sometimes” in Melbourne and Seattle, “now contrast, the econometric tradition derived from and then” in the Netherlands, and “usually” in Za- studies in economics and decision analysis of greb. consumer choices between products, often under 20 Measuring Health conditions of risk or uncertainty. Scaling meth- provide numerical scores for the response cate- ods derived from this tradition record utilities, gories “strongly agree,” “agree,” “disagree,” which are numbers that describe the strength of and “strongly disagree.” A method called “sum- a person’s preference for a particular outcome mated ratings” was described by Likert for doing when faced with uncertainty such as the possi- this, but in practice, the correlation between the bility of a future gain (50). The crucial distinc- scaled version and arbitrarily assigned scores of tion is that utilities capture both the person’s 1, 2, 3, and 4 is high (52, pp149–152). preference and their attitude toward risk. This is Magnitude estimation procedures have been relevant in predictive health measures, in clinical proposed as an improvement on category scaling decision analyses, and in studies of patient tasks. Because psychophysical experiments choices between alternative therapies for which showed that people can make accurate judg- the outcome lies in the future and remains uncer- ments of the relative magnitude of stimuli, mag- tain. In theory, if a person is risk neutral, values nitude estimation asks them to judge the relative and utilities will agree. If a person is risk-averse severity implied by each statement on scales (e.g., very careful to preserve good health), their without limitations on values. This has been utilities for a given state will be higher than their used, for example, in rating how much more se- values; the converse holds for a person who is rious one type of crime is, compared with an- risk seeking (51, Figure 1). Because more people other. Proponents argue that this approach are risk-averse, utility scores for a health state produces a ratio scale estimate of the absolute are generally higher than value scores. value of the stimulus, although critics argue that the precise meaning of judging one stimulus as twice as desirable as another is not clear. Magni- Psychometric Methods tude estimation has seldom been used in its psy- The many psychometric scaling procedures may chometric form as a way to score categorical be grouped into comparative techniques and responses in health measures, although it is magnitude estimation methods. Among those in widely used in econometric scaling procedures. the former category, several measurements de- Psychometric scaling methods are described in scribed in this book have used Thurstone’s detail in many sources (39; 40; 42; 52–55). They “Equal-Appearing Intervals” scaling method to demand more work during the development of a produce what is argued to be an interval scale. measurement but provide measurements with There are variants of the approach, but in essence better numerical qualities, which may be espe- a sample of people is asked to judge the relative cially important in scales with few items. How- severity of each response category. The people ever, item weights do not necessarily alter the making the judgments may be patients or experts impression gained from unweighted scores. As or a combination. Items such as “Pain prevents shown in the chapters that follow, weighted and me from sleeping,” “I have aches and pains that unweighted scores have been compared for in- bother me,” and “I require prescription medi- struments such as the Physical and Mental cines to control my pain” are sorted by each Impairment-of-Function Evaluation, the Multi- judge into rank-ordered categories of severity. level Assessment Inventory, the McMaster There are typically ten to 15 numbered cate- Health Index Questionnaire, the Nottingham gories. For each item, a score is based on the me- Health Profile, and the Health Opinion Survey: dian of the category numbers into which it was the correlation is uniformly high, generally be- placed by the group of judges; this is used as a tween 0.95 and 0.98. Item weights are most numerical indicator of the judged severity of that likely to exert an effect when a scale includes a response. Where disagreement between raters is small number of items that cover different top- high, the item may be eliminated because this ics. Where there are more than 20 questions and suggests ambiguity. This scaling approach has where these all measure a common theme, been used in instruments such as the Sickness Im- weights are unlikely to have a strong effect on pact Profile. Similar techniques may be used to the way a health measure ranks people.  The Theoretical and Technical Foundations of Health Management 21 This introduces two fundamental challenges mension of health, such as functional disability in scoring health measures. Health, and more es- rather than pain due to a limb problem that af- pecially quality of life, implies several dimen- fects walking ability. Thus, applied during the sions that should perhaps be scored separately, item analysis phase of test development, the and the internal structure of the dimensions may Guttman approach helps to identify questions be complex. Psychometric scaling techniques that do not fit the set of questions and should be were originally intended for scales in which an discarded. Instruments that use Guttman scaling affirmative answer to an extreme statement include Meenan’s Arthritis Impact Measurement would imply affirmative answers to less severe Scale and Lawton and Brody’s Physical Self- statements on the same dimension. The dimen- Maintenance Scale. This approach is less well sion score would be the total of the scale weights suited to the measurement of psychological at- for items answered affirmatively. Items on a tributes, which seldom form cumulative scales health measurement may have more complex (43, p75). logical connections between them. A simple A generalization of this approach to analyz- structure occurs with physical disability, in ing response patterns is item response theory which answers to questions on mobility will (IRT) (56; 57), occasionally also called latent generally form a pattern in which an affirmative trait analysis (58). This is based on a measure- answer to a major problem will imply affirma- ment model developed by Georg Rasch for as- tive answers to milder problems. However, this sessing abilities that is now frequently applied may not hold true in other areas of health. De- to health measurements. The “latent trait” pression, for example, may be experienced dif- refers to a theoretical continuum that the test ferently by different people so that a positive items are designed to measure. Using physical response to a severe symptom (e.g., suicidal function as an example, the continuum running ideation) need not imply that a person will also from fitness to severe disability (the latent trait) respond positively to other, lesser symptoms. is plotted along the horizontal axis of a graph. And a greater diversity of symptoms does not Items that measure different severities of dis- necessarily imply greater severity. This intro- ability are spaced along this continuum; items duces the themes of the number of dimensions such as “I can run a mile” would be near the being covered in a measurement, and of analyz- left, and “I am confined to a wheel chair” near ing the scale properties of sets of items. the right. The Rasch model is based on several Instead of providing scores for a single ques- postulates: tion, some category scaling procedures analyze the properties of the measurement as a whole. (i) Scores on a measure depend on the abil- These are the response-centered approaches that ity of the person and the level of disabil- provide scale values for individual respondents, ity implied by the items. rather than for items. A method sometimes used (ii) A good scale should have items that in measuring functional abilities is Guttman’s range in difficulty, and with a rank order approach to what was originally called “scalo- of difficulty that does not vary from re- gram analysis” (see Glossary). This procedure spondent to respondent. identifies groups of questions that stand in a hi- (iii) Good measurement requires that a per- erarchy of severity. Where questions form a son’s ability be accurately reflected re- Guttman scale, an affirmative reply to a question gardless of the scale used. indicating severe disability will also imply an af- (iv) Where a measurement contains several firmative reply to each question lower on the scales, such as physical, mental, and so- scale, so a person’s health status can be summa- cial, these should each be unidimen- rized by noting the question at which the replies sional, because if questions on different switch from affirmative to negative. This pattern topics are combined, the resulting score of responses provides evidence that the items cannot be clearly interpreted (unless the measure varying levels of a single aspect or di- score is extreme). 22 Measuring Health Applied to a measurement instrument, a ility value into a linear continuum. Overall Rasch analysis provides several statistics. Ad- scores for respondents are not based on a sum- dressing postulate (i), it shows where each pa- mation of item responses as with conventional tient fits along the measurement scale and also approaches, because this assumes that people where the items fit. The item severity parameter with the same overall score will have equivalent is defined in terms of the threshold along the la- patterns of responses. Instead, an individual’s tent trait at which the probability of endorsing score is based on the pattern of responses given a given item is 0.5. The greater the threshold to items of differing severity. For postulate (iv), value, the greater a person’s disability must be an error term and a fit statistic indicate how before she or he will endorse the item. For pos- well each item matches the ideal pattern of the tulate (ii), IRT analysis gives a score to each cumulative scale. As with Guttman analyses, item that reflects its difficulty on a parametric these statistics can be used to select those items scale, equivalent to a scaling task that translates that best define the continuum being measured. ordinal responses into an interval scale (59). Unlike Guttman analyses, however, the Rasch The distribution of items across the scale indi- approach also indicates how each patient per- cates the density of coverage at each level of the forms on the scale and which patients are giv- trait by the measurement instrument; this may ing idiosyncratic responses. indicate gaps in coverage or redundancies. For The Rasch model for IRT only indicates the efficiency, a measurement instrument should position of each item on the latent trait. An ad- contain items that are spaced evenly along the ditional parameter is required to indicate how continuum being measured, and IRT illustrates sharply an item demarcates people above and be- this graphically; this is helpful in identifying re- low that level—a notion closely related to sensi- dundant items and in selecting subsets of items tivity and specificity (61). The two-parameter from a larger test. Indicators of test efficiency model adds information on item discrimination, can be produced. Analysis also indicates how plotted on a vertical axis that indicates the cu- consistently the relationships among the items mulative probability of endorsing each item as hold in different subgroups of respondents the level of the trait increases. This produces S- (e.g., classified by age, diagnosis, overall score shaped, normal ogive “item characteristic on the test). For postulate (iii), the severity of curves” for each item starting at the base of the disability for each respondent can also be graph and sloping up more or less steeply to placed along the same continuum; an advan- meet the top line; a curve is produced for each tage of item response approaches is that they item. As before, the distance of each curve from produce scale values both for items and sub- the left of the graph indicates the threshold or jects. Computations are based on conditional severity of the trait at which the item will be an- probabilities of response patterns, rather than swered positively; the slope of the curve indicates on correlations between items as in factor the discriminal ability or accuracy of the item. analysis. It has been argued that correlational The ideal is a steep slope, suggesting that the approaches produce results that vary according item sharply demarcates people along the trait. to the particular sample of respondents used in In practice, item slopes may vary across the an analysis. As an illustration, Delis et al. severity range, with severe symptoms often showed empirically how memory tests applied forming more reliable indicators than mild to a sample of cognitively normal people can symptoms: it is harder to write a good question- form a unitary factor, whereas the same tests naire item reflecting mild illness than severe (58, applied to memory-impaired subjects can form p401). distinct factors if disease disrupts aspects of IRT is seeing increasing application in the memory processing; hence, the factor structure development of new health measures (58; 62). is sample-dependent (60). In place of correla- It provides valuable insight into the structure tions, IRT uses the logit as the unit of measure- and quality of a measurement and is ideal for ment; a logit is a transformation of a probab- selecting a subset of items from an initial item  The Theoretical and Technical Foundations of Health Management 23 pool when constructing a questionnaire. It scale be used, for example, running from 1.0 helps to explain why a measure may work well representing perfect health to 0.0, representing for one group of respondents and not for an- death. Utilities were chosen as the basis for other or be good for one task and poor for an- defining quality because they offer a way to inte- other. General discussions of item response grate morbidity and mortality into a single scale. theory are provided by Hambleton et al. (63; Utilities refer to preferences for particular out- 64). comes when faced with uncertainty (whereas values refer to preferences under certainty). The 1944 theory of von Neumann and Morgenstern Methods Derived from Economics offered a prescriptive model of decision-making and Decision Analysis under uncertainty, suggesting how such deci- Whereas many health measures were developed sions ought rationally to be made; this theoreti- by clinicians whose interest lay in recording the cal model underlies scaling methods, such as the health of individual patients, economists have time trade-off and standard gamble, which are contributed mainly to measures intended to be used to estimate utility values (50). applied to groups of people. To allocate medical The most common unit of measurement of resources rationally, we need a way to compare health benefit in this equation is the quality- the health benefits achieved per unit cost for dif- adjusted life year, or QALY (66). QALYs offer a ferent medical interventions. A table listing such way to integrate changes in both length and benefits could guide policy in directing resources quality of life produced by an intervention; they away from procedures that are relatively expen- are calculated as the average number of addi- sive for a given level of health benefit toward tional years of life gained from the intervention, those that are cheaper. While economists debate multiplied by a utility judgment of the quality of details of assessing both the cost and benefit life in each of those years. The QALY concept components, we are concerned here only with does not indicate how these weights are to be de- the measurement of benefits. Because the econo- rived, but this is typically done using a scaling mists’ focus has been on the general rather than task such as the standard gamble, described later the particular, the econometric approach has in this chapter. After the utility for a given state tended to develop health indexes rather than pro- has been estimated, QALYs are calculated by files and has focused attention on details of scal- multiplying the number of years to be expected ing rather than question wording. in each state by the utility for that state. For ex- Early economic evaluations approached ben- ample, if a statement of “Choice of work or per- efits in terms of whether a treatment reduced the formance at work very severely limited; person is costs associated with disability and lost produc- moderately distressed by this” were rated 0.942, tion. Accounting gains in purely financial terms, and if a person remained in this state for 10 however, appears to ignore the inherent value of years, the QALY would be 9.42 years (66). Tor- life; it would also lead logically to a preference rance and Feeny gave the example of a person for a quick death if a patient cannot be cured. placed on antihypertensive pharmacotherapy for Accordingly, economists switched to recording 30 years, which prolongs his life by 10 years at a benefits in terms of readily interpretable units quality level of 0.9. The need for continued drug such as cases of disease prevented or life-years therapy reduces his quality of life by 0.03. gained, and these were subsequently balanced by Hence, the QALYs gained would be 10 × 0.09 − considering the quality of those life-years (65). 30 × 0.03 = 8.1 years (50). Many refinements The goal was to indicate, in a single number, the have been proposed to this basic approach. For total amount of health improvement; this should example, QALYs may be adjusted to reflect the consider both quantity and quality of life and be individual’s preference or aversion for risk- applicable to patients with any medical condi- taking. This characterizes the “healthy years tion (50). To make comparisons across condi- equivalent” (HYE) indicator, which permits the tions, it was also desirable that a universal metric rate of trade-off between length and quality of 24 Measuring Health life to depend on the expected remaining life risk of operative mortality they would tolerate to span (65). avoid remaining in the condition described in the The main application of QALYs lies in policy first option. In principle, the more severe their as- analysis, comparing different interventions sessment of the condition, the greater the risk 1− rather than in evaluating the health status of in- p of operative mortality (perhaps five to ten dividual patients. For example, QALYs offer a percent) they would accept to escape. The sever- way to compare an intervention that extends life ity of the condition is expressed by subtracting but with high levels of disability, to an interven- the tolerated risk of operative mortality from 1 to tion that does not prolong life as much but offers give the utility of the state described in the first higher levels of well-being. QALYs have been option. used to propose levels of hypertension below These judgments are difficult to make, so that which intervention is judged not to be worth- various simplifying procedures may be used in while (67). Extending this, comparisons have of- administering the scaling task, including the ten been made of the cost per year of healthy life time trade-off technique (50; 55, p36; 73). gained from alternative therapies (50, Table 2; Methods such as “willingness to pay” or the 68, Tables 2.2–2.4; 69, Table 5). “time trade-off” are alternative ways to present Calculating the utility weights (e.g., the factors the standard gamble, rather than different meth- of 0.03 and 0.9 for the hypertension example) re- ods. As before, the judges are asked to imagine quires a scaling task. This commonly uses a vari- that they are suffering from the condition whose ant of the “standard gamble” approach (70). As severity is to be rated. They are asked to choose in psychometric scaling, these approaches involve between remaining in that state for the rest of a sample of people in making judgments; occa- their natural lifespan (e.g., 30 years for a 40 sionally the patient whose health is being evalu- year-old person), or returning to perfect health ated may provide the weights. Utility weights for fewer years. The number of years of life ex- provided by experts, patients, and population pectancy they would sacrifice to regain full samples are empirically similar (71); most studies health indicates how severely the condition is now take a sample that includes all three. Unlike rated. The utility for the person with 30 years of psychometric approaches, however, utility scaling life expectancy would be given as methods involve making a choice between health Utility = (30 − Years traded)/30. states that involves a notion of sacrifice, rather than merely judging value on a scale (72). A third approach is to ask the person how much The standard gamble involves asking sub- of their income they would be willing to pay to jects to choose between living in the relevant obtain a hypothetical cure. Note that these two health state (which is less than ideal) for the rest alternatives to the standard gamble do not in- of one’s life, or taking a gamble on a treatment volve the element of risk, so that they measure that has the probability p of producing a cure, values rather than utilities. The values obtained or a less desirable outcome (typically, immediate can, however, be transformed into estimates of death) with risk 1− p (50). Applied to health utilities using a power curve relationship derived states, the scaling procedure asks the rater to from studies that have measured both values and imagine that, for the first option, they are to live utilities (29, p262). the rest of their life with the chronic condition Econometric scaling generally rates the utility that is being evaluated; the symptoms, functional of composite health states, so a practical issue is limitations, pain, and other drawbacks are de- the large number of such health states that may scribed. However, they could undergo an imagi- need to be judged. Here, a set of assumptions nary operation that, if successful, would result in from multiattribute utility theory allows the in- complete recovery. However, the operation incurs vestigator to estimate utility weights for compos- a specified risk of death, and the risk of death is ite states from the weights of individual varied until the rater is indifferent between the components. Multiattribute utility theory identi- first and second options. This shows how great a fies a mathematical formula for extrapolating  The Theoretical and Technical Foundations of Health Management 25 utility estimates to a wide range of states from and consistent perceptual tendency to exaggerate direct measurement of preferences for a subset or underestimate stimuli described by the expo- of those states (37, p245). The multiattribute nent b of the psychophysical experiments. This functions may be additive or multiplicative, with may also be applicable to health; we know little each requiring different assumptions of the data about this as yet, although several studies have (37, p246). In practice, raters use the standard compared subjective responses with physical or gamble to judge the individual elements (e.g., laboratory measurements of health status (75). A pain, disability) plus a selection of multiattribute tendency also exists to alter response to a stimu- combinations and from this, utility weights for lus across time or under different situations, and all possible permutations can be estimated (50). this is termed bias. One person may exaggerate This approach has been used in the Health Utili- symptoms to qualify for sick leave or a pension, ties Index. whereas another may show the opposite bias and By providing a universal unit for measuring minimize ailments in the hope of returning to health status, economic methods permit us to work. Subjective ratings of health blend an esti- compare the impact of different forms of disabil- mate of the severity of the health problem with a ity, and the cost-utility of different treatments; personal tendency to exaggerate or conceal the they are being increasingly applied in health pol- problem—a bias that varies among people and icy analysis and in discussions of resource over time. allocation. These methods also allow us to ad- Biases in subjective measurement can arise dress broader philosophical questions, such as from the respondents’ personalities, from the whether there is a social consensus over the valu- way they perceive questionnaires, or from partic- ation of life, whether it is equally valuable to ex- ular circumstances of their illnesses. Illustrative tend the life of a 20 year-old and a 50 year-old, examples will be given here, rather than an ex- or whether it is equally valuable to extend one haustive list, for the main question concerns how life for 1,000 days, or 1,000 lives for a single to reduce the extent of response bias. Personality day (4). LaPuma and Lawlor sketched the his- traits that may bias responses include stoicism, tory of QALYs and the philosophical and ethical defensiveness, hypochondriasis, or a need for at- bases for their use and offer a cautionary discus- tention. The drive to portray oneself in a good sion of their potential misuse (67). Despite wide- light by giving socially desirable responses illus- spread enthusiasm for approaches to combining trates a bias that reflects social influences. Gold- quality and length of life, critics have reviewed berg cites the example of a person, regarded by some of the logical perils this may entail (74). outside observers as fanatically tidy, but who judged himself untidy (76). These biases are un- conscious, rather than a deliberate deception, Identifying and Controlling and are typically more extreme where questions Biases in Subjective Judgments concern socially undesirable acts, such as sexual behavior or the illicit use of drugs. Several scales Psychophysical experiments have shown that have been proposed to measure a person’s ten- people can make accurate and internally consis- dency to give socially desirable responses (77; tent judgments of phenomena. This is the case, 78), but these scales appear to show rather low at least, with laboratory experiments concerned intercorrelations (55, p57). A correlation of 0.42 with lights or noises in which the person has no between the Crowne-Marlowe and Edwards particular stake. Judgments about health may scales was reported in one study, for example not be so dispassionate: in real life, people often (79, Table 1). Attitudes may also bias responses have a personal stake in the estimation of their and this has long been studied (80). Biases can health. Bias refers to ratings that depart system- also arise from the way people interpret ques- atically from true values. We should, however, be tionnaire response scales: some prefer to use the careful to discriminate between two influences in end-position on response scales, whereas others the judgment process. There is the underlying more cautiously prefer the middle (76, 26 Measuring Health pp26–34). Other biases may be particular to the so as to hide their intent. This is commonly done health field and reflect the anxiety that surrounds with psychological measurement scales. For ex- illness. One example is named the “hello- ample, the “Health Opinion Survey” has noth- goodbye” effect, in which the patient initially ing to do with opinions; it is designed to identify exaggerates symptoms to justify their request for psychoneurotic disorders. Several of its ques- treatment (55, p58). Subsequently, the person tions appear to refer to physical symptoms (e.g., minimizes any problems that remain, either to upset stomach, dizziness) but are intended as please the clinician or out of cognitive disso- markers of psychological problems. A second nance (81). Similarly, in the rebound effect, a pa- approach is to have the questionnaire completed tient recovering from a serious illness tends to by someone who is familiar with the patient. Ex- exaggerate reported well-being (82). A related amples may be found in Chapter 4, on ratings of bias is known as “response shift,” whereby pa- social adjustment, and in Chapter 8, on ratings tients with a chronic condition may shift percep- of mental abilities. A third way of handling re- tion as the disease progresses—typically they sponse bias is to make an explicit assessment of lower expectations and thereby score better on the patient’s emotional response to their condi- health measures despite physical deterioration tion. Examples may be found in Chapter 9, on (83; 84). Examples are given by Paterson (85, measurements of pain. p875). A fourth approach is a statistical method of Two general approaches are used to deal analyzing patterns of responses that provides two with bias in health measurement. The first by- scores. The first is concerned with perception and passes the problem and argues that health care indicates the patient’s ability to discriminate low should consider symptoms as presented by the levels of the stimulus, a notion akin to estimating patient, bias and all, given that this forms a part the size of “just noticeable” differences. The sec- of the overall complaint: consideration of “the ond score reflects the person’s decision whether to whole patient” is a hallmark of good care. From report a stimulus; under conditions of uncer- this viewpoint, it can be argued that the biases tainty, this reflects a personal response bias. This inherent in subjective judgments do not threaten field of analysis derived from the problem of dis- the validity of the measurement process: health, tinguishing signals from background noise in ra- or quality of life, is inherently subjective and is dio and radar, where it is called signal detection as the patient perceives it. The second viewpoint analysis. The same analysis may also be applied to argues that this is merely a convenient simplifica- other types of decision (e.g., the behavior of base- tion and that the interests of diagnosis and pa- ball players in deciding whether to swing at the tient management demand that health ball, or of drivers in deciding when it is safe to measurements should disentangle the objective merge into traffic) and is here called decision estimate from any personal response bias. As an analysis or sensory decision theory. Where it is example, different forms of treatment are appro- difficult to judge whether or not a stimulus is priate for a person who objectively reports pain present (e.g., whether a radiograph shows a small of an organic origin and for another whose pain fracture), two types of error are possible: falsely is exacerbated by psychological distress; several reporting a fracture, or missing one. Where the pain scales we review make this distinction. radiograph is unclear, the decision is influenced Most health indexes do not disentangle sub- by factors such as the frequency of seeing frac- jective and objective components in the measure- tures of this type, clinical conservatism, and the ment and thereby tacitly (or overtly) assume that relative importance of avoiding each type of er- the mixture of subjective and objective data is in- ror. The analytic technique uses the notions of evitable. Among the relatively few indexes that “hits” and “false alarms.” A hit occurs where a do try to separate these components, we discern stimulus is present and I rate it as present; a false several different tactics. The simplest is to try to alarm occurs where I report a signal that is in fact mask the intent of the questions, either by giving absent. When it is important to detect a signal, I them a misleading title, or by phrasing questions may set my decision criterion to raise the number  The Theoretical and Technical Foundations of Health Management 27 of hits, even at the expense of also increasing false of varying rewards or penalties for making cor- alarms. Thus, my performance is characterized by rect or incorrect decisions (90–92). SDT analysis my trade-off of hits against false alarms, which has also been applied in studying the effect of can be shown graphically by the receiver operat- age on test scores: may declines in memory ing characteristic curve (ROC), which is a plot of scores among old people reflect changes in ap- the probability of detection (hits) against the proach to taking a test (e.g., cautiousness), probability of false alarms. Guyatt et al. have ap- rather than real reductions in memory (87)? plied this type of thinking to health measure- Although this is the original application of ments; the signal represents true differences in ROCs, similar curves are often drawn to sum- health that one wishes to detect; the noise repre- marize the validity of screening tests; this is be- sents measurement error over which the signal cause hits and false alarms are equivalent to must be detected (86, p1343). They then link sensitivity and 1-specificity (see page 33). In this these ideas to the purpose of the measurement, application, the area under the ROC curve indi- noting that for evaluative instruments, the rele- cates the discriminal ability of the instrument, vant signal concerns change over time, so the ranging from 0.5 (indicating no discrimination) signal-to-noise ratio is represented by a measure to 1.0 (indicating perfect discrimination). of responsiveness. For a discriminative measure, signal represents the ability to distinguish be- tween people, so that a signal-to-noise ratio is rep- Conceptual Bases for resented by a reliability coefficient (86, Table 2). Health Measurements Signal detection theory (SDT) has been ap- plied to analyzing responses to health measures It may appear obvious that a health measure- (87–90). For example, detecting pain involves ment must be based on a specific conceptual ap- the patient’s ability to perceive the painful stimu- proach: if we are measuring health, what do we lus and the tendency to describe the feeling as mean by the term? The conceptual definition of “painful.” These can both be evaluated experi- an index justifies its content and relates it to a mentally: two types of stimulus are presented in broader body of theory, showing how the results random order—noise alone or noise plus low lev- obtained may be interpreted in light of that the- els of signal—and the ability of an individual to ory. Yet by no means do all of the methods we identify the presence of a signal against the noise review in this book offer a conceptual explana- is recorded. Applied to pain research, the stimu- tion of what they measure or of the value judg- lus is usually an electric shock and the “noise” is ments they incorporate. A basic issue in a low level of fluctuating current. For each trial, constructing a health index is how to choose the respondent judges whether the shock was among the virtually unlimited number of ques- present and from the resulting pattern of true- tions that could potentially be included. There and false-positive responses, two indexes are are two ways of confronting this problem that calculated: discriminal ability and response bias. correspond to the nomothetic and idiographic Using some basic assumptions, it is possible to approaches: questions may be chosen from a estimate these two parameters from a person’s theoretical or from an empirical standpoint. rate of hits and false alarms; this is well de- Both approaches are represented among the in- scribed by Hertzog (87). In pain research, this struments reviewed in this book, although analysis has been used to study whether anal- purely empirically based measures are becoming gesics influence pain by altering discriminability less common. (i.e., by making the stimulus feel less noxious), The empirical method to index development or by shifting the response bias (i.e., making the is typically used when the measurement has a respondent less willing to call the feeling practical purpose, for example to predict which “painful”). Studies of this type are further de- patients are most likely to be discharged after re- scribed in Chapter 9. Presented in the form of habilitation is complete. After testing a large ROC curves, the results may show the influence number of questions, statistical procedures 28 Measuring Health based on correlation methods are used to select analytically, rather than simply descriptively: those that best predict the eventual outcome. studies using these methods may be able to ex- These “item analysis” statistics are described in plain, rather than merely to describe, the pa- the next section. Empirical selection of items has tient’s condition in terms of that theory. a practical appeal. It does, however, suffer the Underlying theories also provide a guide to the weakness that the user cannot necessarily inter- appropriate procedures to be used in testing the pret why those who answer a certain question in validity of the method. The conceptual ap- a certain manner tend to have better outcomes: proaches used in many indexes share common the questions were not selected in relation to any elements that will be introduced here, whereas particular theory of rehabilitation. Many illus- more detail about the conceptual basis for each trations exist; the questions in the Health Opin- method are given in the reviews of individual ion Survey were selected because they indices. distinguished between mentally ill patients and Most indicators of physical health reviewed unafflicted people, and although they succeed in in this book and many of the psychological doing this, debates over what exactly they mea- scales build their operational definitions of sure have continued for 35 years. A more recent health on the concept of functioning: how far is example is Leavitt’s Back Pain Classification the individual able to function normally and to Scale, which was developed empirically to distin- carry out typical daily activities? These are effect guish between pain of an organic origin and pain indicators. In this view, someone is healthy if related to emotional disorders. It succeeds well, physically and mentally able to do the things she but Leavitt himself commented: “Why this par- or he wishes and needs to do. The phrase “activ- ticular set of questions works as discriminators ities of daily living” epitomizes this principle. and others do not is unclear from research to There are many discussions of the concept of date.” Accordingly, the back pain scale may functional disability; early examples include have clinical value, but it does not advance our those by Gallin and Given (94) and Slater et al. understanding of the phenomenon of pain as (95). As Katz et al. pointed out, functional level a response to emotional disorders and of how may be used as a marker of the existence, sever- psychological factors may modify the pain re- ity, and impact of a disease even though knowl- sponse. edge about its cause and progression is not The alternative strategy to developing a advanced enough to permit measurement in health measurement is to choose questions that these terms (22, p49). Measuring functional are considered relevant from the standpoint of a level offers a convenient way to compare the im- particular theory of health, reflecting a nomo- pact of different types of disease on different thetic approach. As deductive science develops populations at different times. A common ap- and tests theories, so some indices have been de- proach to measuring health is therefore through signed to represent particular theories of health, the impact of disease on various aspects of func- and their use in turn permits the theory to be tion. The notion of impact is contained in the ti- tested. Melzack’s McGill Pain Questionnaire tles of several measures, such as the Arthritis was based on his theory of pain; Bush’s Quality Impact Measurement Scales or the Sickness Im- of Well-Being Scale was based on an explicit pact Profile. Stating that an index of disability conceptual approach to disability. Bech pro- will assess functioning does not, however, indi- posed a conceptual approach to specifying the cate what questions should be included. At this content of a measure of well-being that reflects more detailed level, measures diverge in their both the diagnostic approach of the Diagnostic conceptual basis; approaches that have been and Statistical Manual, and also Maslow’s hier- used include psychological theories such as archical model of human needs (93). Basing a Maslow’s hierarchy of human needs, biological measurement on a particular concept has impor- models of human development (as in Katz’s ADL tant advantages. Linking the measurement with scale), or sociological theories such as Me- a body of theory means the method can be used chanic’s concept of illness behavior.  The Theoretical and Technical Foundations of Health Management 29 Alterations in function may be assessed at In its revised version first outlined in 2001, various stages, from the bodily lesion to its ulti- the ICIDH was renamed the International Clas- mate effect on the individual. A widely used clas- sification of Functioning, Disability and Health, sification of functional limitations is that of the initially called the ICIDH-2 (100; 101) and sub- WHO 1980 International Classification of Im- sequently the ICF (102). This has refocused at- pairment, Disability and Handicap (ICIDH) tention away from the consequences of disease (96; 97), although different terms for the same toward functioning as a component of health concepts have been used by other writers (98). (102, p81). The ICF provides codes for the com- In the WHO definitions, “impairment” refers to plete range of functional states; codes cover body a reduction in physical or mental capacities. Im- structures and functions, impairments, activities, pairments are generally disturbances at the or- and participation in society. It also considers gan level; they need not be visible and may not contextual factors that may influence activity have adverse consequences for the individual: levels, so that function is viewed as a dynamic impaired vision can normally be corrected by interaction between health conditions (e.g., dis- wearing glasses. Where the effects of an impair- ease, injury) and the context in which the person ment are not corrected, a “disability” may re- lives including physical environment and cul- sult. Disability refers to restriction in a person’s tural norms relevant to the disease. It establishes ability to perform a function in a manner consid- a common language for describing functional ered normal for a human being (e.g., to walk, to states that can be used in comparing diseases and have full use of one’s senses). In turn, disability statistics from reporting countries. Compared can (but need not) limit the individual’s ability with ICIDH, the language is positive, so that “ac- to fulfil a normal social role, depending on the tivity” and “participation” replace “disability” severity of disability and on what the person and “handicap.” The ICF is described on the wishes to do. “Handicap” refers to the social WHO web site at www.who.int/classification/icf, disadvantage (e.g., loss of earnings) that may and chapter headings are listed by Üstün et al. arise from disability. A minor injury can restrict (102, Table 2). an athlete but may not noticeably trouble some- The positive aspect of health is often men- one else; a condition producing mild vertigo tioned and is linked to resilience and resistance could prove handicapping to a high-steel con- to disease. Here, health implies not only current struction worker but not to a writer. Although well-being but also the likelihood of future dis- medical care generally concentrates on treating ease. This is relevant because planning health impairments, a patient’s complaint is usually ex- services requires an estimate of what the burden pressed in terms of disability or handicap, and of sickness is likely to be in the future. Nor is it the outcome of treatment may therefore best be appropriate to consider as equally healthy two assessed using disability or handicap indicators people at equal levels of functional capacity if rather than measures of impairment. The ICIDH one has a presymptomatic disease that will seri- provided a classification of types of disability in- ously compromise future functional levels. For tended for use in statistical reporting of health this reason, some indexes (e.g., Functional As- problems and thereby in estimating needs for sessment Inventory, Quality of Well-Being Scale) care. Building on this general approach, most assess prognosis as well as current health status. health measures operationally define disability in This prognostic element relates only to health, terms of limitations in normal activities; these because there are external factors, such as the are generally termed “functional limitations” or personal financial situation, which can affect fu- “functional disability.” “Disablement” has been ture health levels, but are not an inherent aspect proposed to refer to both disability and handi- of a person’s health. cap: the field normally covered by subjective The conceptual basis for a health index nar- health indexes. These terms and their variants rows the range of questions that could possibly were discussed by Duckworth (97); Patrick and be asked, but a relatively broad choice remains Bergner also gave a brief overview (99). among, for example, the many questions that 30 Measuring Health could be asked to measure “functional disabil- lidity describes the range of interpretations that ity.” Whether the index reflects a conceptual can be appropriately placed on a measurement approach to health or has been developed on a score: What do the results mean? What can we purely empirical basis, procedures of item se- conclude about a person who produced a partic- lection and subsequent item analysis are used ular score on the test? The shift in definition is to guide the final stages of selecting and vali- significant, in that validity is no longer a prop- dating the questions to be included in the erty of the measurement, but rather of the inter- measurement. pretation placed on the results. This approach holds advantages and disadvantages. It may stimulate a search for other interpretations of an indicator, as illustrated in the development of un- The Quality of a Measurement: obtrusive measures (104). It can lead to discov- Validity and Reliability ering links between constructs previously thought to be independent. An advantage of the Someone learning archery must first learn how broader definition is also seen in the dementia to hit the center of the target, and then to do it screening tests we review in Chapter 8. Most are consistently. This is analogous to the validity and valid in that they succeed in their purpose of reliability of a measurement (103). The consis- identifying cognitive impairments (i.e., they are tency (or reliability) of a measurement would be sensitive). However, some tests show low speci- represented by how close successive arrows fall ficity in that cognitively normal people with lim- to each other, wherever they land on the target. ited formal education also achieve positive Validity would be represented by the aim of the scores, thus suggesting that the scales provide a shooting—how close, on average, the shots come more general indicator of intelligence or cogni- to the center of the target. Ideally, a close group- tive functioning. A disadvantage of broadening ing of shots should lie at the center of the target the definition of validity is that it may foster (reliable and valid), but a close grouping of shots sloppiness in defining the precise purpose of a may strike away from the center, representing an measurement. For example, much time has been archer who is consistently off target, or a test wasted over arcane speculation as to what cer- that is reliable but not valid, perhaps due to mea- tain scales of psychological well-being are actu- surement bias. ally supposed to measure. This can best be The core idea in validity concerns the mean- avoided by closely linking the validation process ing, or interpretation, of scores on a measure- to a conceptual expression of the aims of the ment. Validity is often defined as the extent to measurement and also linking that concept with which a test measures that which it is intended other, related concepts to indicate alternative to measure. This conception of validity, which possible interpretations of scores. There are reflects the idea of agreement with a criterion, is many ways to assess validity. The choice depends commonly used in epidemiology and underlies on the purpose of the instrument (e.g., screening the notions of sensitivity and specificity. It is a test or outcome measurement) and on the level of limited conception, however, because valuable abstraction of the topic to be measured. The fol- inferences from a measurement can sometimes lowing section gives a brief introduction to the be drawn that exceed the original intent of the methods that are mentioned in the reviews. test. Like using this book for a door stop, a mea- More extensive discussions are given in many surement may have other uses than those origi- standard texts (41; 43; 105–107). nally intended for it. As an example, mortality rates were recorded to indicate levels of health or need for care, but the infant mortality rate may Assessing Validity also serve as a convenient indicator of the so- cioeconomic development of a region or country. Most validation studies begin by referring to Hence, a more general definition holds that va- content validity. Each health measurement rep-  The Theoretical and Technical Foundations of Health Management 31 resents a sampling of questions from a larger for which no criterion exists. The former include number that could have been included. Similarly, screening and diagnostic tests and predictive the selection of a particular instrument is a measures. The latter group include measures of choice among alternatives, and the score ob- abstract concepts such as quality of life, happi- tained at the end of this multistage sampling pro- ness, or disability. Validation procedures for the cess is of interest to the extent that it is first group are relatively straightforward and in- representative of the universe of relevant ques- clude variants of criterion validation. Validation tions that could have been asked. Content valid- procedures for measures of the second type in- ity refers to comprehensiveness or to how volve construct validation that is more extensive adequately the questions selected cover the than criterion validation. These two approaches themes that were specified in the conceptual def- will next be described in sequence. inition of its scope. For example, in a patient sat- isfaction scale, do all the items appear relevant Criterion Validity to the concept being measured, and are all as- pects of satisfaction covered? If not, invalid con- Criterion validity considers whether scores on clusions may be drawn. Feinstein has proposed the instrument agree with a definitive, “gold the notion of sensibility, which includes, but standard” measurement of the same theme. This slightly extends, the idea of content validity option for validating a measure typically occurs (108). Sensibility refers to the clinical appropri- when a new instrument is being developed as a ateness of the measure: are its design, content, simpler, more convenient alternative to an ac- and ease of use appropriate to the measurement cepted measurement: can a self-report of anxiety task? Feinstein offered a checklist of 21 attrib- replicate what a psychiatrist would have diag- utes to be used in judging sensibility. Indeed, nosed? The new and the established approaches content validity is seldom tested formally; in- are applied to a suitable sample of people and stead, the “face validity” or clinical credibility of the results are compared using an appropriate a measure is commonly inferred from the com- indicator of agreement. The comparison may be ments of experts who review its clarity and com- used in a summative manner to indicate the va- pleteness. A common procedure is to ask lidity of the measure as a whole, or it may be patients and experts in the field to critically re- used in a formative manner during the develop- view the content of the scale. Alternatively, more ment of the new measure to guide the selection formal focus groups and in-depth interviews of items by identifying those that correlate best may be arranged to explore whether the ques- with the criterion. The latter forms part of “item tionnaire is covering all aspects of the topic rele- analysis” and involves cumulating evidence on vant to patients. Cognitive interviews involve the performance of each item from a range of va- having respondents verbalize their reactions to lidity analyses outlined subsequently in this each question as they answer them to indicate chapter. how the questions are perceived by respondents Criterion validity may be divided into con- (85). Occasionally, tests of linguistic clarity are current and predictive validity, depending on used to indicate whether the phrasing of the whether the criterion refers to a current or future questions is clear (109). It is difficult, perhaps state. To illustrate the former, results from a even impossible, to prove formally that the items questionnaire on hearing difficulties might be chosen are representative of all relevant items compared with the results of audiometric test- (110). ing. In predictive validation, the new measure- Following content validation, more formal ment is applied in a prospective study and the statistical procedures are used to assess the va- results compared with subsequent patient out- lidity of a measurement. Here, a distinction may comes (e.g., mortality, discharge). Because pre- be drawn between measures of concepts for dictive validation may demand a long study it is which there exists some type of criterion with rarely used; there are also logical problems with which the measure can be compared, and those the method. It is likely that during the course of 32 Measuring Health a prospective study (and also possibly as a result only that disease and not another condition. of the prediction) interventions will be applied Specificity corresponds to “discriminal validity” to treat the individuals at highest risk selectively. in the language of psychometrics. Accordingly, If successful, the treatment will alter the pre- testing specificity may involve comparing scores dicted outcome, contaminating the criterion and for people with the disease with those of others thus falsely making the test appear invalid. To who have different diseases, rather than to peo- avoid this predictive validity paradox, predic- ple who are completely healthy; as is so often the tions are more commonly tested over a brief case in research, the choice of comparison group time-interval, bringing this approach closer to is subtle but critical. concurrent validation. Because the purpose of a screening test is To illustrate the typical procedure for testing usually to divide people into healthy and sick concurrent criterion validity, Chapter 8 on men- categories, it is necessary to select a score on the tal status testing describes several screening tests; test that best distinguishes the two. The thresh- their validation represents a major category of old score that divides these two categories is criterion validation studies. Here, the test is ap- known as the cutting-point or cutting-score. For plied to a varied sample respondents, some of clarity in this book, cutting-points will be ex- whom suffer the condition of interest (e.g., de- pressed as two numbers, such as 23/24, indicat- mentia) and some who do not. The criterion ing that the threshold lies between these; this is takes the form of a diagnosis made indepen- helpful because a single number (as in “a dently by a clinician who has not seen the test re- cutting-point of 16 was used”) does not indicate sult; it could also include information from whether people with that precise score are in the magnetic resonance imaging scans, neuropsy- healthy or sick group. Choosing a cutting-point chological assessments or other diagnostic test- is challenging. First, the whole idea of a division ing. Statistical analyses show how well the test into two categories fails to reflect the notion of agrees with the diagnosis and also identifies the disease as a continuum but is done because of threshold score on the test that most clearly dis- convention and because decisions over treatment tinguishes between healthy and sick respon- options often require it. Second, the optimal dents. Note that in this screening test paradigm, cutting-point may vary for different applica- the goal is usually to show how well the new test tions; this derives from the problem that chang- divides the sample into two groups, healthy and ing the cutting-point alters the proportions of sick, but criterion validation can also be used to healthy and sick people correctly classified by show agreement with a scaled score of severity. the test such that an increase in sensitivity is al- In the two-category paradigm, two potential er- most always associated with a decrease in speci- rors can occur: the test may fail to identify peo- ficity. Hence, if the goal is to rule out a ple who have the disease, or it may falsely diagnosis, a cutting-point will be chosen that en- classify people without the disease as being sick. hances sensitivity, whereas if the clinical goal is The “sensitivity” of a test refers to the propor- to rule in a disease the cutting-point will be cho- tion of people with the disease who are correctly sen to enhance specificity. And third, as will be classified as diseased by the test, while “speci- seen many times in this book, it is awkward to ficity” refers to the proportion of people without compare the validity of two tests in terms of the disease who are so classified by the test re- both sensitivity and specificity. One may need, sult. For those unfamiliar with these terms, the for example, to compare one test with sensitivity crucial element to recognize is that the denomi- of 85% and specificity of 67%, with a test with nators are the people who truly have, or do not a higher sensitivity but also a lower specificity. have, the disease according to the criterion stan- Because of these complications, a succession dard. The terms sensitivity and specificity are of methods have been proposed to describe the logical: the sensitivity of the test indicates sensitivity and specificity of a test over a range of whether the test can sense or detect the presence possible cutting-points, thereby offering an over- of the disease, whereas a specific test identifies all summary of test performance or criterion va-  The Theoretical and Technical Foundations of Health Management 33 lidity. Because shifting the cutting-point to im- tive (or of a negative) test result. This is clinically prove sensitivity (i.e., increasing true-positive re- useful, because it indicates the probability that a sults) almost inevitably reduces specificity (i.e., patient with that particular test result will have increases false-positive results), a common ap- the disease. Finally, a further extension of the proach is to plot true-positive (sensitivity) against likelihood ratio is the diagnostic odds ratio false-positive results (1− specificity). If this is (118). This is defined as the ratio of the odds of a done for each possible cutting-point on the test, a positive test result in people with the disease rela- curve is produced that forms another application tive to the odds of positive results in unaffected of the ROC analyses described earlier in this people. A value of 1 indicates a test that does not chapter. The curve illustrates the trade-off be- discriminate; higher values indicate superior test tween sensitivity and specificity. The area under performance. the ROC curve (AUC) indicates the amount of in- Some cautions should be considered in inter- formation provided by the test; a value of 0.5 preting sensitivity and specificity figures. Al- would imply that the test is no better than merely though sensitivity and specificity are often guessing whether the respondent has the disease. presented as constant properties of a test, there Put formally, the AUC indicates the probability are several instances in which this may not be the that a randomly chosen healthy person will be case. Sensitivity and specificity figures typically identified as healthy by the test rather than a ran- vary according to the severity of the condition: it domly chosen person with the disease (91; 111). is easier to detect more serious illness than less AUC values of 0.5 to 0.7 represent poor accu- serious. This “spectrum bias” may distort the racy, values between 0.7 and 0.9 indicate a test impression of validity where the sample used in “useful for some purposes,” whereas values over evaluating the test differs in severity from that on 0.9 indicate high accuracy (112). Although use- which the test will later be applied (119). For ful, the AUC must be interpreted with caution. It this reason, a test that works well in a hospital shows the overall performance of the test, rather setting may not work in general practice, so au- than its performance at the recommended thors should always report the characteristics of cutting-point; a scale with a high AUC may have patients included in validation studies. It may be lower sensitivity and specificity values at its opti- wise to consider validity in terms of a distribu- mal cutting-point than another scale (113; 114). tion of sensitivities, according to the clinical fea- A modification of the ROC curve called the tures of each patient; the result of validity testing Quality ROC, or QROC, adjusts for a chance re- represents an average of these figures (120). Sen- sult (115; 116). Here, sensitivity is plotted sitivity and optimal cutting-points may also vary against specificity, producing a loop figure; the by factors such as age or educational level; the area under the QROC is derived from the area of dementia screening scales described in Chapter 8 a rectangle drawn from the origin to the optimal illustrate this. The reader should also critically point in the loop (113, Figure 1). An alternative review the study design in which sensitivity and way to summarize sensitivity and specificity is the specificity of a screening test were estimated. likelihood ratio. This indicates how much more The ideal is a study in which the gold standard likely is a given patient with a positive test result criterion and the test are applied to all people in to have the disease in question. The likelihood ra- an unselected population. This is rarely feasible, tio for a particular value of a diagnostic test is es- however, because it implies obtaining gold stan- timated as sensitivity divided by (1− specificity), dard assessments from large numbers of people or the probability of that test result in people who do not have the condition; there are often with the disease, divided by the probability of the financial or ethical constraints to this. Thus, a same result in people without the disease (117). A common alternative is to take samples of those convenient graphical nomogram allows the clini- with and those without the condition, selected cian to incorporate the estimated prevalence of on the basis of the gold standard; typically, these the condition in that population, converting a are people in whom the condition was suspected likelihood ratio into a predictive value of a posi- and who went to hospital for assessment. Sensi- 34 Measuring Health tivity and specificity are calculated in the normal veloped to guide the validation of measures of manner. The problem is that the comparison complex constructs for which there is no single group is not representative of the general popula- criterion standard. Hence, there is no set distinc- tion; this may give misleading validity results if tion between criterion and construct validation, the test is to be used on an unselected popula- and the increasing trend is to view all validation tion. An alternative approach is often used in of health measures as falling under the general validating a screening test in a population sam- heading of construct validation, of which crite- ple; here, the screening test is applied, and those rion validation forms a subcategory. who score positively receive the gold standard assessment, along with a random subsample of Construct Validity those scoring negatively. This approach incurs a bias known as “verification,” or “diagnostic For variables such as pain, quality of life, or hap- work-up bias,” which inflates the estimate of piness, gold standards do not exist and thus va- sensitivity and reduces that of specificity (119; lidity testing is more challenging. For such 121). This bias must be corrected, although re- abstract constructs, validation of a measurement grettably often it is not. There are several alter- involves a series of steps known as “construct native formulas for this (122). The net results of validation.” This begins with a conceptual defi- these various biases is that there is almost always nition of the topic (or construct) to be measured, a range of sensitivity and specificity figures for a indicating the internal structure of its compo- given test; readers will have to select results ob- nents and the way it relates to other constructs. tained from samples as similar as possible to the These may be expressed as hypotheses indicat- sample on which they plan to use the instru- ing, for example, what correlations should be ment. obtained between a quality of life scale and a One final note is that the use of a cutting- measure of depression, or which respondents point to indicate the likelihood of disorder as- should score higher or lower on quality of life. sumes diagnostic equivalence of the many None of these challenges alone proves validity possible ways of obtaining that score. This may and each suffers logical and practical limita- be adequate where a measurement scale gen- tions, although when systematically applied, uinely measures only a single dimension. How- they build a composite picture of the adequacy ever, where (as with many mental status tests) the of the measurement. A well-developed theory is items measure different aspects of a disorder, dif- required to specify such a detailed pattern of ex- ferent clinical pictures may emerge from different pected results, a requirement not easily met. The combinations of items (123). In such situations, main types of evidence used to indicate construct it may be more appropriate to replace the single validity include correlational evidence, which is cutting-score by a pattern of scores in defining a often presented in the form of factor analyses, positive response. This observation leads to a and evidence for the ability of the measure to dis- deeper consideration of the apparent simplicity criminate among different groups. The logic of of criterion validity. Most conditions for which these is briefly described here; their practical ap- health measures are developed (e.g., disability, plication is illustrated in the reviews of individ- depression, dementia) are syndromes that com- ual measurements. prise multiple signs and symptoms grouped into different facets. Indicators of criterion validity Correlational Evidence of Validity such as sensitivity or AUC offer only summary scores and give no indication of where the Hypotheses are formulated that state that the strengths and limitations of the screening test measurement will correlate positively with other may lie. They are of limited value in test develop- methods that measure the same concept; the hy- ment or in contrasting the relative strengths of ri- potheses are tested in the normal way. This is val measures. For this more detailed insight, we known as a test of “convergent validity” and is must turn to construct validation, which was de- equivalent to assessing sensitivity. Because no  The Theoretical and Technical Foundations of Health Management 35 single criterion exists in construct validation, the which the measure should logically be related; measurement is typically compared with several among others, studies of the EuroQol illustrate other indexes; multivariate procedures may be this (126). The expected strength of correlation used. Hypotheses may also state that the mea- coefficients (or the variance to be explained) surement will not correlate with others that mea- should be stated before the empirical test of va- sure different themes. This is termed “divergent lidity. validity” and is equivalent to the concept of Several guidelines may assist the reader in in- specificity. For example, a test of “Type A be- terpreting reported validity correlations. Coeffi- havior patterns” may be expected to measure cients range from −1.0 (indicating an inverse something distinct from neurotic behavior. association) through 0.0 (indicating no associa- Accordingly, a low correlation would be hypoth- tion at all) to +1.0. First, because some random esized between the Type A scale and a neuroti- error of measurement always exists, the maxi- cism index and, if obtained, would lend re- mum correlation can never reach 1.0 but can rise assurance that the test was not simply measuring only to the square root of the reliability of the neurotic behavior. A useful item in a test should measurement. Where two measurements are vary according to the characteristic being mea- compared, the maximum correlation between sured (here, a specific aspect of health) and not them is the square root of the product of their re- by other, extraneous factors. This idea was for- liabilities (106, p84). Where the reliability coeffi- merly called “content saturation,” which is high cients are known, the observed correlation may when an item’s correlation with its own scale is be compared with the maximum theoretically higher than its correlation with an irrelevant obtainable; this helps in interpreting the conver- scale. Thus, depression questions should not, by gent validity coefficients between two scales. For their wording, measure anxiety, even though example, a raw correlation of 0.60 between two anxiety and depression may occur together in a scales seems modest; but if their reliabilities are patient. This has been measured by Jackson’s 0.70 and 0.75, the maximum correlation be- Differential Reliability Index (124). This is cal- tween them would only be 0.72, making the 0.60 culated by taking the square root of the residual seem quite high. By extension, we can estimate obtained by subtracting the square of the corre- what the concurrent validity correlation would lation between an item and the irrelevant scale be if both scales were perfectly reliable: from the square of the correlation with its own rxy scale (125, p532). r xy′ = ———— , Correlating one method with another would √(rxxryy) seem straightforward, but logical problems arise. Because a new measurement is often not in our example, this is 0.83. These corrections designed to replicate precisely the other scales are only appropriate in large samples, of 300 or with which it is being compared (indeed, it may more; with smaller samples they can lead to va- be intended to be superior), the expected correla- lidity coefficients that exceed 1.0, which is over- tion may not be perfect. But how high should it optimistic for most measurements we review. be, given that the two indexes are inexact mea- Correlations can also be translated into more surements of similar, but not identical, con- interpretable terms. Imagine that a measure is be- cepts? Here lies a common weakness in reports ing used to predict a criterion; the purpose is to of construct validation: few studies declare what assess how much the accuracy of prediction is in- levels of correlation are to be taken as demon- creased by knowing the score on the measure- strating adequate validity. The literature contains ment. The simplest approach is to square the many examples of authors who seem pleased to correlation coefficient, showing the reduction in interpret arbitrarily virtually any level of correla- error of prediction that would be achieved by us- tion as supporting the validity of their measure. ing the measurement compared to not using it. As Construct validation should begin with a rea- an example, if a health measurement correlates soned statement of the types of variable with 0.70 with a criterion, using the test will provide 36 Measuring Health 0.72, or almost a 50% reduction in error com- cepts of interest cannot be measured directly; pared with simply guessing. This formulation in- they are unobservable and hypothetical. They dicates that the value of a measurement declines can only be measured indirectly, by indicators, rapidly below a validity coefficient of roughly such as questions or clinical observations, which 0.50. A glance at a scatter plot that illustrates a are incomplete and capture only part of the con- correlation below 0.50 will confirm this impres- cept to be measured. Factor analysis is a central sion. The derivation of this approach is given by analytical tool in describing the correspondence Helmstadter (106, p119), who also describes of alternative indicators to the underlying con- other ways of interpreting validity coefficients. cepts they may record. Using the pattern of inter- The adequacy of validity (and reliability) co- correlations among replies to questions, the efficients should also be interpreted in light of analysis forms the questions into groups or fac- the values typically observed. Convergent valid- tors that appear to measure common themes, ity correlations between the tests reviewed in each factor being distinct from the others. As an this book are low, typically falling between 0.40 example, Bradburn selected questions to mea- and 0.60, with only occasional correlations sure two aspects of psychological well-being falling above 0.70 for instruments that are very that he termed positive affect and negative affect. similar, such as the Barthel Index and PULSES Factor analysis of the questions confirmed that Profile. However, the attenuation of validity co- they fell into two distinct groups, which were efficients due to unreliability implies that a cor- homogeneous and unrelated, and which by relation of 0.60 between two measures inspection appeared to represent positive and represents an extremely strong association. negative feelings. This analytical method is com- Pearson correlations are often used in report- monly used to study the internal structure of a ing reliability and validity findings, but they health index that contains separate components, should be interpreted with caution, because they each reflecting a different aspect of health. quantify the association between two measure- Factor analysis can be used to describe the ment scales and indicate how accurately one rat- underlying conceptual structure of an instru- ing can be predicted from another, but they do ment; it shows how far the items accord in mea- not indicate agreement. This is because a perfect suring one or more common themes. Applied to correlation requires only that the pairs of read- validation, factor analysis can be used in study- ings fall on a straight line but does not specify ing content validity: do the items fall into the which line. For example, if one sphygmo- postulated groupings? Factor analysis can also manometer systematically gives blood pressure be used in test construction to guide the selec- readings 10 mmHg higher than another, the cor- tion of items on the basis of their association relation between them would be 1.0, but the with the trait of interest. Typically, separate agreement would be zero. For most health mea- scores would be calculated for these components sures, this does not matter because the numerical of the measurement instrument. Factor analysis values are arbitrary, but for assessing reliability can also be used in construct validation by indi- it is crucial (see page 40). Correlations are also cating the association among subscale compo- influenced by the range of the scale: wider ranges nents of measurements or even complete tend to produce higher correlations, even though measures. Scales measuring the same topic the agreement remains the same. Bland and Alt- would be expected to be grouped by the analysis man discussed these limitations, and proposed onto the same factor (a test of convergent valid- alternative approaches to expressing the agree- ity), whereas scales measuring different topics ment between two ratings (127). would fall on different factors (divergent valid- ity). This, in effect, applies a factor analysis to the results of a previous factor analysis and so is Factorial Validity termed second-order factoring. Factor analysis The major challenge in psychological measure- includes two parts: a structural model and a ment (as in measuring health) is that the con- measurement model (131). The structural model  The Theoretical and Technical Foundations of Health Management 37 posits underlying constructs to be measured, are often contravened. In such cases, item re- such as disability and handicap, and may pro- sponse analyses can be used to provide a factor pose the relationships between these in a dia- analysis applicable to binary (i.e., yes-no) re- gram in which ellipses are joined by arrows sponses. showing mutual association or influence. The These approaches to assessing construct va- measurement model shows the relationship be- lidity focus chiefly on the content coverage of tween variables recorded in a study (e.g., an- the measure and how this compares with other swers to questions) and the underlying concepts; measures. The next stage involves validating the by convention, arrows connect the ellipses show- performance of the measure: can it distinguish ing the constructs with rectangles indicating the healthy from sick people; can it show change measured variables. Confirmatory factor analy- over time? sis begins with the structural model and is used to test how far empirical data support the pro- Group Differences and posed conceptual structure. Exploratory factor Sensitivity to Change analysis begins with the measured variables and shows how they cluster together to represent un- An index that is intended to distinguish cate- derlying constructs, even where these have not gories of respondent (e.g., healthy and sick, anx- been formally defined (as is commonly the case). ious and depressed) may be tested by applying it An extension of factor analysis, which can to samples of each group, and by analyzing the also be applied to individual items or to com- scores for significant differences. Significant dif- plete scales, is multidimensional scaling. This ferences in scores would disprove the null hy- provides a graphical plot of the clustering of pothesis that the method fails to differentiate items in two (or more) dimensional space (128). between them. This approach is frequently used The resulting maps represent the similarity or and contributes to the overall process of con- dissimilarity of items by the distance between struct validation. However, like all other valida- them on the plot, the axes of which are concep- tion procedures, it suffers logical limitations. tually equivalent to the theta latent trait in the First, it may be necessary to standardize scores item response analysis plots. The maps are use- to correct for differences in age or sex distribu- ful in immediately identifying items that do not tions in the populations being compared (132). fit their scales and, because the axes represent Second, screening tests for emotional disorder underlying dimensions, the maps also identify may compare psychiatric patients with the gen- gaps in the overall coverage of the scale. Like eral population but this will underestimate the factor analysis, multidimensional scaling can adequacy of the method if some members of the help to indicate the number of dimensions repre- general population have undiagnosed mental or sented by the scale, and hence which items emotional disorders. Third, measures are often should appropriately be combined to form sub- developed using highly selected cases in clinical scores. research centers. Just as it is risky to generalize Factor analysis is widely used, and in the past the results of clinical trials from tertiary care was frequently misused in the studies reported in centers to other settings, so referral patients may this book; several principles guide its appropriate score highly on an index, yet high scores in a use (129; 130). Items to be analyzed should be community survey may not be diagnostic (110). measured at the interval scale level; the response New measurements should be retested in a vari- distributions should be approximately normal, ety of settings. and there should be at least five (some authors say For evaluative measures, one crucial charac- 20) times more respondents in the sample than teristic is the ability to detect change that actu- there are variables to be analyzed. Because so ally occurs. When the results of several drug many health indices use categorical response trials are combined in a metaanalysis, the aver- scales (e.g., “frequently,” “sometimes,” “rarely,” age impact of the treatment may be indicated us- “never”) the first and second of these principles ing the effect size statistic to compare mean 38 Measuring Health scores in treatment and control groups or before siveness statistic that divides the change in raw and after treatment. Results are expressed in scores by the standard deviation among stable standard deviation units: (Mt − Mc)/SDc. The ef- subjects (139); as an alternative, the standard- fect size can be compared with a z score in ized response mean (SRM) divides the mean which, assuming a normal distribution of scores, change in score by the standard deviation of in- the effect size indicates how far along the per- dividuals’ changes in scores (140; 141). Again, centile range of scores a patient will be expected values above 0.8 are considered large, and values to move following treatment. With an effect size between 0.5 and 0.8 are moderate. SES values of +1.00, a patient whose initial score lies at the are generally higher than the corresponding mean of the pretreatment distribution will be ex- SRM values. Further alternatives include relative pected to rise to about the 84th percentile of that efficiency (142), measurement sensitivity, re- distribution after treatment. The effect size of- ceiver operating characteristic analyses (in terms fers an insight into the clinical importance of an of the ratio of signal to noise) (143; 144), and an intervention: although large samples may make a F-ratio comparison (145; 146). Refinements to comparison between drug and placebo groups these basic formulas correct for the level of test- statistically significant, the effect size can be retest reliability (147). Effect size statistics are used to indicate whether the difference is clini- not completely independent of the sample: the cally important. Metaanalyses are now summa- largest effect size that can be attained for a given rizing the results of trials in terms of the effect measure in a given sample is the baseline mean sizes achieved by selected treatments; an exam- score divided by its standard deviation. ple is that by Felson et al. (133). These statistical indicators of effect size offer The idea of effect size for a treatment can be the reader no indication of the clinical magni- turned on its side and applied to measurement tude or importance of change represented by a instruments to illustrate how sensitive they are given shift in scores. Occasionally, authors pro- to detecting change—an example was given by vide helpful illustrations of changes in score; for Liang et al. (134). For outcome measurements, the Medical Outcomes Study measures (which responsiveness or sensitivity to change is a cru- were later incorporated into the SF-36), for ex- cial characteristic, and disease-specific measures ample, a 10-point deficit on the physical func- will generally be more responsive than generic tioning scale was described as being equivalent instruments. Finer-grained response scales also to the effect of having chronic mild osteoarthri- generally enhance responsiveness. However, no tis; a three-point difference on the mental health clear consensus exists as to how responsiveness scale is equivalent to the impact of being fired or should be assessed (135). Many indicators have laid off from a job (68, Table 11.1). Patrick et al. been proposed, and careful reading of articles is proposed some anchors to indicate the impor- often needed to identify which statistic has been tance of a given change in scores. Changes in used. Terwee et al. provide a review, listing no scores on the health measure can be compared fewer than 31 statistics (136, Table 2). Most in- with change judged significant by clinician or dicators agree on the numerator, which is the patient. Alternatively, changes can be compared raw score change; less agreement exists on the with a global health rating to identify score appropriate denominator. The basic measure is changes that correspond to a change of one cate- the change in mean scores on the measure di- gory on the global question (148, p29). Based vided by the standard deviation of the measure on these types of comparison, effect sizes of 0.2 at baseline (137); this is sometimes called the to 0.49 are generally considered small; 0.5 to standardized effect size, or SES. An SES greater 0.79 are moderate, and 0.8 or above are large than 0.8 can be considered large; moderate val- (137; 149). ues range from 0.5 to 0.8, whereas values below Knowing the responsiveness of an instrument 0.5 are small (138). Alternative formulations in- is valuable in calculating the power of a study clude a t-test approach that divides the raw score and the sample size required; as with all sample change by its standard error (139), or a respon- size calculations, this requires a prior estimate of  The Theoretical and Technical Foundations of Health Management 39 the likely effect size for a particular comparison those interviewed with versions of a question- (e.g., drug and placebo), with a particular mea- naire in different languages) are matched in surement instrument. To illustrate, if the treat- terms of their overall score, and their responses ment effect is expected to be only 0.2, about 400 to each item compared; this forms a variant of subjects will be required per group to show the item-total correlations. Groups for which DIF is contrast as significant, setting alpha at 0.05 and typically tested during item analysis include gen- power at 0.8. To detect a larger effect of 0.5, one der, age, education, ethnicity, and language. would need about 64 patients per group (137, pS187). Effect sizes also help in adjusting com- Construct Validity: Conclusion parisons between studies that used different measurements: an instrument with a large effect Although great progress has been made in for- size will show a higher mean improvement fol- malizing construct validation, it retains an ele- lowing treatment than an instrument with a ment of an art form. Construct validity cannot smaller effect size. be proved definitively; it is a continuing process A measure may discriminate between groups, in which testing often contributes to our under- but it should discriminate only in terms of a rel- standing of the construct, after which new pre- evant aspect of health and not to some other dictions are made and tested. This is the ideal, characteristic. Accordingly, many validation re- but the literature contains examples of inade- ports reviewed in this book consider whether an quate and seemingly arbitrary presentations of item or a measure as a whole records the health construct validity. We still see statements of the characteristic consistently in different groups. type “Our instrument correlated 0.34 with the For example, questions in a depression scale are AA Scale, −0.21 with the BB Index, and 0.55 sometimes worded in an colloquial manner (“I with the CC Test and this pattern of associations feel downhearted and blue”) that may not have supports its construct validity.” In the absence of equivalent meaning in different cultures or age a priori hypotheses, the reader may well wonder groups and thus may be difficult to translate into what pattern of correlations would have been in- other languages. Ideally, responses to an item terpreted as refuting validity. Mercifully, there should not vary for different types of people are several examples of systematic approaches to with the same level of depression. The theme of construct validation, such as those by McHor- differential item functioning, or DIF, refers to ney et al. (153) or by Kantz et al. (154). Good situations in which sources of variation other validation studies state clear hypotheses, test than the trait being measured influence the re- them, and also explain why those hypotheses are sponse to the item; this therefore reduces the va- the most relevant. Good studies will also try to lidity of the item as a measure of health. Even disprove the hypothesis that the method mea- though people of different ages (for example) sures something other than its stated purpose, may be expected to respond differently to ques- rather than merely assembling information on tions on their health, this should only occur be- what it does measure. cause of actual differences in health that occur with age, and not because people of different ages use language differently. Similarly, men Assessing Reliability might tend to underreport mental distress if an item were phrased in a manner that implies that Reliability, or consistency, is concerned with er- agreeing to it implied weakness. DIF can be illus- ror in measurement. In the metaphor of the tar- trated graphically (150, Figure 1). The various get used earlier, reliability was symbolized by the ways to identify DIF fall under the heading of dispersion of shots. This referred to the consis- item analysis techniques; they may use variants tency or stability of the measurement process of a regression approach, a factor analytic ap- across time, patients, or observers. Feinstein proach (150), or item response theory models suggested the term “consistency,” averring that (151; 152). People in contrasting groups (e.g., “reliability” carries connotations of trustworthi- 40 Measuring Health ness that may not be appropriate when, for enough observations are made, so the average example, a measurement repeatedly yields erro- score a person obtains if tested repeatedly gives neous results (108). Nonetheless, reliability re- a good estimate of the true score. mains a more widely accepted term and so is In classical test theory, reliability refers to the used here. extent to which a score is free of random error. Intuitively, the reliability of a measure can be More formally, reliability of a measurement is assessed by applying it many times and compar- defined as the proportion of observed variation ing the results, expecting them to match. Unreli- in scores (e.g., across patients or across repeated ability can therefore be seen in terms of the measurements) that reflects actual variation in discrepancies that would arise if a measurement health levels. This is normally written as the ra- were repeated many times. Unfortunately, re- tio of true score variance to observed score vari- peating a health measurement to assess its stabil- ance, or σ2T / σ2O. Because the observed score is ity is often not as simple as repeating a assumed to be the sum of true and error scores, measurement in the physical sciences: repeating this formulation is equivalent to σ2T / (σ2T + σ2E). a patient’s blood-pressure measurement may be This provides a number with no units that unwelcome and may even cause blood pressure reaches unity when all variance in observed to rise. This has led to alternative techniques for scores reflects true variance and zero when all assessing the reliability of health measurements. observed variance is due to errors of measure- There are many sources of measurement error; ment. To illustrate this idea, imagine that two the various approaches to estimating reliability nurses measure the blood pressure of five people. focus on different sources of error, and as with For simplicity, imagine also that each patient’s validity, there is no single way to express relia- blood pressure remains stable while the nurses bility, although most share a common ancestry make their measurements. The true variation in classical test theory. refers to the range of blood pressure readings Classical test theory views the value obtained across patients; error refers to discrepancies be- from any measurement as a combination of two tween the nurses’ ratings for any of the patients components: an underlying true score and some assessed. Reliability increases when true varia- degree of error. The true score, of course, is what tion increases and when error variation is small. we are trying to establish; “error” refers to im- Within this traditional approach, two types of precision in the measurement that frustrates our reliability are distinguished: whether different aim of obtaining a true score. Errors are com- raters assessing a respondent obtain the same re- monly grouped into two types: random errors or sult (inter-rater agreement or observer variation) “noise,” and systematic errors or bias. Tradi- and, whether the same result is obtained when tional reliability theory considers only errors the same rater makes a second assessment of the that occur randomly; systematic errors, or bi- patient (variously termed intra-rater reliability, ases, were generally considered under validity stability, test-retest reliability, or repeatability). testing. Random errors may arise due to inatten- Reliability can be translated into a convenient tion, tiredness, or mechanical inaccuracy that statistic in several ways; the choice depends may equally lead to an overestimation or under- mainly on which source of variation is to be con- estimation of the true quantity. The assumption sidered as “error.” We can introduce this by re- that such errors are random holds several corol- turning to the distinction between agreement laries. They are as likely to increase the observed and association that was introduced on page 36. score as to decrease it; the magnitude of the er- Agreement assesses whether our two nurses re- ror is not related to the magnitude of the true port identical blood pressure readings for the score (measurement error is no greater in ex- each patient, whereas association is less demand- treme scores), and the observed score is the ing and merely estimates whether the differences arithmetic sum of the error component and the in blood pressure readings that nurse A reports underlying true score that we are attempting to among the five patients are the same as the dif- measure. Random errors cancel each other out if ferences reported by nurse B (even though their  The Theoretical and Technical Foundations of Health Management 41 actual blood pressure readings may not agree). their relative standings on the two. Hence, if one Because measures of association were first used set of scores is systematically higher than the to describe reliability, we can begin with the cor- other, the ICC will not reach unity (144). Intra- relation coefficient. class correlations refer to a family of analysis of Association refers to a relationship (typically variance approaches that express reliability as linear) between two sets of readings and is com- the ratio of variance between subjects to total monly represented by a Pearson correlation co- variance in scores (157; 158); the procedure for efficient. The use of Pearson correlations to calculating the ICC is illustrated by Deyo et al. indicate inter-rater agreement or retest reliability (144). Because different sources of variance may was common in the past, but such practice has be considered in the numerator in different relia- passed from favor because it can seriously exag- bility studies, there is no single type of intraclass gerate the impression of reliability (155). As an correlation. Shrout and Fleiss described six illustration, Siegert et al. obtained a Pearson cor- forms of ICC, noting that research reports fre- relation of 0.95 and a Spearman coefficient of quently fail to specify the form used (159). 0.94 between self- and interviewer-administered intraclass correlations may also compare agree- questionnaires; despite these high correlations ment among more than two raters; for ordinal there was a precise agreement between interview measurement scales the equivalent statistic is and questionnaire in only 65% of cases (156, Kendall’s index of concordance (W). A statistical p307). The central point is that many types of relative of the intraclass correlation is the con- discrepancy are possible in pairs of ratings: as cordance correlation coefficient, which indicates with the example of blood pressure, the whole the agreement between the observed data and a distribution of scores may shift for one assess- 45° slope (the Pearson correlation indicates ment, or the relative position of certain individu- agreement between the data and the best-fitting als may change, or one rater may achieve greater line, wherever this may lie) (144, p151S). precision than the other, or one scale may be A graphical approach described by Bland and stretched compared with the other. Correlation Altman offers a useful way to conceptualize coefficients reflect some of these types of mis- intra-subject variation. This involves an exami- match between scores but ignore others. Thus, nation of the distribution of differences in the although a simple rule would advise against the pairs of scores for each person. The differences Pearson correlation for reporting reliability, a in scores can be presented graphically, plotting more sophisticated guide would begin by consid- them against the mean of the two scores. This ering which types of variation in scores are con- shows whether the error changes across the sidered erroneous. For example, if patients in a range of the scale and identifies outliers. In terms test-retest study are recovering and their average of the associated statistics, 95% of the differ- scores improve over time, the correlation coeffi- ences will lie within two standard deviations of cient will ignore this shift in the overall distri- the mean, and the standard deviation of the dif- bution of scores and will (perhaps quite appro- ferences can be calculated by squaring the differ- priately) focus on whether the relative position ences, summing them, dividing by N, and taking of each person was maintained. An index of the square root (127). An alternative way of ap- agreement, by contrast, would classify the gen- proaching variation of scores within subjects is eral improvement as unreliability in the measure called the standard error of measurement (this is and would show the test as being very unstable. unfortunately abbreviated to SEM, not to be As a measure of agreement, the intraclass confused with the standard error of the mean, correlation (ICC) is now normally used to indi- which shares the same abbreviation). The SEM cate reliability instead of Pearson or rank-order may be conceptualized as the standard deviation coefficients. Like the Pearson correlation, the of an individual score and is pertinent to ICC ranges from −1 to +1, but it measures the individual-level applications. A perfectly reliable average similarity of the subjects’ actual scores instrument would have a SEM of zero; all varia- on the two ratings, not merely the similarity of tion would be true variation (160). The standard 42 Measuring Health error of measurement is the square root of the index of agreement. When the ratings are di- within-subject variance (i.e., the overall variance chotomous (e.g., agreement over whether a chest less the variance between subjects) (161). Alter- radiograph shows pneumonia), a simple table natively, it may be estimated as the standard de- can indicate the proportion of agreement. How- viation × √(1−reliability) (160, p295). ever, a correction is necessary because chance Finally, it is intuitively clear that the repro- agreement will inflate the impression of reliabil- ducibility of a measurement lies in inherent ten- ity, especially if most cases fall in one category. sion with its responsiveness, or its ability to For example, if most chest radiographs do not detect change, which was discussed in the sec- show pneumonia, a second rater could spuri- tion on validity. To clarify the relationship, re- ously produce high agreement merely by guess- sponsiveness is defined in terms of the scale’s ing that all the films are normal. The kappa ability to detect reliable changes over time, omit- coefficient corrects for this by calculating the ex- ting random change. This leads to the notion of tent of agreement expected by chance alone and smallest real difference (SRD) or reliable change removing this from the estimation (see formula (RC). The reliable change index (RCI) is the in Glossary). Kappa coefficients can also be ap- smallest change in scale points on a given mea- plied to ordinal data in several categories, and a sure that represents a real change, as opposed to weighted kappa formula can be used to distin- chance variation. Establishing the RCI involves guish minor from major discrepancies between calculating an error margin around an individual raters (55, p95). measurement value that expresses the uncer- Although formerly a mainstay of test devel- tainty in a point measurement due to unreliabil- opment, traditional test theory has frequently ity. For a Type I error of 0.05, this width of the been criticized. More detailed approaches to re- margin is ±1.96 times the standard error of the liability testing, such as Rasch’s item response measurement. Next, in assessing the significance model and generalizability theory, are now be- of a change in scores, an error band can be cal- ing applied in testing health indexes (55; 63, culated around the difference between two mea- p98; 164; 165). The main shortcoming of con- surements, and when this band includes the ventional test theory is that it groups many value 0, it implies that the difference between to sources of error variance together, whereas we scores could be due to error alone; the formula may wish to record these separately to gain a proposed by Beckerman et al. is 1.96 × fuller understanding of the performance of a √(2×SEM) (161, p573). The denominator rep- measurement. Generalizability theory uses resents the variance in a person’s difference score analysis of variance to separate different sources and the derivation was explained (for example) of variation, distinguishing, for example, the ef- by Christensen and Mendoza (162). Thus, a fect of using an interview or a questionnaire, or graph can be drawn that plots pre- against post- the gender or age of the interviewer. The results scores. The diagonal rising from the origin of the indicate the likely performance of the measure- graph represents no change in scores, and a ment under different conditions of administra- shaded area either side of this, based on the RC tion. There will be a different reliability index, would include instances where change coefficient for each, which underscores the scores may be due to chance alone (163, Figure point that there is no single reliability result for 2). Note, however, that reliable change refers a measurement. only to the avoidance of chance errors and does not necessarily indicate clinically important Internal Consistency change. Clinical importance would imply that a change exceeded the RCI threshold and also rep- The notion of repeatability is central to reliabil- resented a return to normal function or was re- ity, but repeated assessments run the risk of a garded as valuable by the patient. false impression of instability in the measure if it In the case of nominal or ordinal rating correctly identifies minor changes that occur in scales, agreement is calculated using the kappa health between administrations. Hence, a sensi-  The Theoretical and Technical Foundations of Health Management 43 tive instrument may appear unreliable. To re- indicate how comparable the results would be if duce this risk, the delay between assessments the measurement had been split into two compo- should be brief. Unfortunately, this may mean nent versions. A simple approach is to correlate that recall by respondent or rater could influence two summary scores derived from the odd- and the second application so that the two assess- even-numbered questions (“split-half reliabil- ments may not be independent. Such is the the- ity”), but a more general approach is to estimate ory, but reassuringly, the interval between test the correlations between all possible pairs of and re-test may not be all that critical; in a study items, which introduces the theme of internal of knee function measures retest results at two consistency. The higher the intercorrelations days were similar to those at two weeks (166). among the items, the easier it would be to create Various tricks have been proposed to avoid two versions that are equivalent and therefore this dilemma and the underlying logic introduces reliable. Thus, in theory, the higher the internal the notions of equivalent forms and of internal consistency, the higher the test-retest reliability consistency. In theory, it is argued, if we could will be. Cronbach’s coefficient alpha is the most develop two equivalent versions of the test that frequently used indicator of internal consistency contain different questions but give the same re- (41, pp380–385). Alpha represents the average sults, this would overcome the problem of recall of all of the split-half correlations that could be biasing the second administration. This ap- calculated for an instrument and is used where proach has been used, for example, in the De- the items have more than two response options. pression Adjective Check Lists (see Chapter 7). It is also related to the saturation on the first fac- The lists (in the plural) are equivalent versions of tor in a factor analysis of a set of items (167). a measure that is designed to record changes in Alpha is intuitive: a value of zero indicates no depression before and after treatment, without correlation among the items, whereas a value of asking the same questions twice. The assessment 1.0 would indicate perfect correlation among of reliability then compares the two versions, them; a problem is that alpha varies with the which can be administered after a brief delay, or number of items in a scale (167, Table 2; 43). even at the same time. The concept of reliability There are several other formulas for internal has thereby shifted from the repeatability of the consistency, all of which estimate what the cor- same instrument over time to establishing the relation would be between different versions of equivalence of two sets of questions.* If the two the same measurement. Kuder and Richardson correlate highly, they are reliable; a score on one proposed their formula 20 as the equivalent of set could be predicted from a score on the other. alpha for dichotomous items. In a similar man- Note that here a correlation is appropriate, be- ner, the coefficient of reproducibility from cause we only wish to show that the two forms Guttman’s scalogram analysis can be used in as- give equivalent results, so that one could be sessing internal consistency, by indicating how translated into the other with a simple arithmetic perfectly the items fall in a single hierarchy of conversion. The next step in the logic holds that, intensity of the characteristic being measured because the forms are different, it will actually (52). Coefficient theta is a variant of alpha that be better to apply them at the same time to assess is suited to scales in which the items cover sev- their reliability, because this avoids the possible eral themes; coefficient omega is a variant de- problem of a real change in health occurring be- rived from factor analysis which estimates the tween administrations. Hence, reliability can be reliability of the common factors in a set of assessed by analyzing a longer instrument, ap- items (168). plied in a single session. Reliability is then as- Because many measures reviewed in this book sessed by using an appropriate statistic to cover several facets of health (for example, the various dimensions of quality of life), care must *Note that this also brings the conception of reliability close to validity. The more deeply one explores these be taken in deciding whether to estimate alpha concepts, the clear it becomes that they differ only in coefficients across the complete measure, or subtle perspective. within subcomponents. In part, this reflects the 44 Measuring Health earlier question of whether a measure provides graphs to illustrate this and also describe how an overall score, or a profile of separate scores for test length affects validity (172). each dimension. The discussion also returns us to But how high an internal consistency is ideal? the theme of item response theory, in which the As with most pleasurable things, moderation ideal is to construct a health measure using sepa- may be best. Where items intercorrelate highly rate, but internally consistent, subscales. Estimat- redundancy arises and the measurement and nar- ing coefficient alpha for each subscale does not rows in scope; it may thus be specific but at the indicate whether they are truly distinct, nor possible expense of sensitivity. If item intercorre- whether they could all have been combined into lations are kept moderate, each item will add a one score. Accordingly, Hays and Hayashi pro- new piece of information to the measurement posed an extension to internal consistency analy- (130). Nor is it reasonable to expect a high inter- sis called “multitrait scaling analysis” that nal consistency if the measurement covers several compares the consistency among items within a facets of a health syndrome—this is seen in cog- subscale to the agreement between items across nitive screening tests. A measurement that is subscales. A computer program identifies items broad in scope may also show lower repeatabil- that load more highly on other scales than those ity because there are more ways in which the to which they were initially assigned (169). For scores can vary from test to retest. A more de- example, an item whose correlation with another tailed analysis of these issues was given by Bollen scale exceeds its correlation with its own scale by and Lennox, who suggested that the optimal in- two or more standard errors may be considered ternal consistency will vary according to the de- to represent a clear scaling error (170). Discus- sign of a test (173). For example, where a sion about how to assess the internal consistency measure reflects the effects of the underlying of multidimensional scales brings us full circle to variable to be measured (as in an ADL scale that the themes of factor analysis and of item re- indicates disability through its effects), internal sponse theory. From this perspective, there is no consistency will be relatively high. However, inherent distinction between internal consistency where a measure records the inputs, or the cause reliability and convergent validity. of the variable to be measured (as in using life Assuming some unreliabilty is always pres- events to measure stress), no intercorrelation ent in individual measurements, the true score may be identifiable among the indicators and in- can be viewed as the mean score of repeated ternal consistency may be low. Measures based measurements. A clinician knows to take two or on symptoms (e.g., in depression scales) may three blood pressure measurements to get a provide intermediate internal consistency, be- truer reading. Hence, more observations give a cause different people typically present differing more accurate estimate of the mean, given the patterns of symptoms. Reliability typically also standard deviation narrows as the number of varies according to the topic being measured: observations increases. Thus, simply increasing Symptoms of depression probably vary more the number of items in an assessment increases than those of angina. Although the internal con- its internal consistency and its reliability. The sistency of a depression symptom checklist joint influence of the number of items and the might be improved by deleting questions that are reliability of each on the reliability of a scale not highly correlated with others, this might were described in formulas derived indepen- compromise content validity. Furthermore, the dently by Spearman and by Brown in 1910. For requirements for reliability differ according to example, a scale reliability of 0.8 can be the purpose of the measurement. A measure to achieved with two items with individual relia- be used for a single patient in a clinical setting bilities of 0.7, or with four of 0.5 or ten of 0.3 must have higher reliability than a survey instru- (171). With most health measurements, reliabil- ment intended to record a mean value for a thou- ity increases steeply up to about ten items, after sand respondents. For an evaluative instrument which the rate of improvement with additional that is sensitive to changes in health over time, items diminishes. Shrout and Yager provide stability is likely to appear low, but the impor-  The Theoretical and Technical Foundations of Health Management 45 tant quality is internal consistency so that the threshold for personality tests being 0.85, that score can be precisely interpreted. If a measure is for ability tests being 0.90, and for attitude tests, to be used to predict outcomes, it must be able to 0.79 (106, p85). A lower reliability, perhaps of predict itself accurately, so test-retest reliability >0.50 (106) or >0.70 (43), may be acceptable in is crucial. If it is intended mainly to measure cur- comparing groups. Empirical results help us to rent status, the internal structure is the most cru- interpret the level of agreement implied by a par- cial characteristic. This is especially true when ticular correlation coefficient; Andrews obtained the measurement is designed to reflect a specific a test-retest coefficient of 0.68 when 54% of re- concept of health, for greater reliability implies spondents gave identical answers on retest and a greater validity. Achieving these balances forms further 38% scored within one point of their the art of test development. previous answer on a seven-point scale (174, p192). Nonetheless, Williams cautioned that a reliability coefficient of 0.8 can mask significant Interpreting Reliability Coefficients variation. Consider a 10-item test, with each The reliability coefficient shows the ratio of true item measured on a 5-point response scale. In a score variance to observed score variance. Thus, simulation of test-retest reliability, responses to if reliability is calculated from an analysis of five of the items were held constant and the variance model as 0.85, this indicates that an es- other five were varied randomly; the resulting re- timated 15% of the observed variance is due to liability correlations centered around 0.80, so error in measurement. If a Pearson correlation is that a reliability of 0.8 is compatible with ran- used to express reliability, the equivalent infor- dom variation in 5 of ten items (175, p14). mation is obtained by squaring the coefficient Various guidelines for interpreting kappa val- and subtracting the result from 1.0. Reliability ues have been proposed; one example views val- also indicates the confidence we may have that a ues of less than 0.4 as indicating slight score for an individual represents his true score agreement, 0.41 to 0.6 as moderate, 0.61 to 0.8 or, put another way, our confidence that a as substantial, and over 0.8 as almost perfect change in scores represents a change in health (176). One guideline for interpreting intraclass status. As reliability increases, the confidence in- correlations is similar: values above 0.75 indicate terval around a score narrows, and so we be- excellent inter-rater agreement, 0.6 to 0.74 come more confident that the true score would shows good agreement; 0.4 to 0.59 indicates fair fall close to the observed score. To illustrate, to moderate, and below 0.4 is poor agreement with a standard deviation of 20 for an imaginary (177). Typical Pearson correlations for inter- 100-point measurement, a reliability of 0.5 rater reliability in the scales reviewed in this would give a 95% confidence interval of ±28.4; book fall in the range 0.65 to 0.95, and values a reliability of 0.7 would give a confidence inter- above 0.85 may be considered acceptable. Pear- val of ±22.0; 0.9 translates into ±12.8, and a re- son correlations for repeatability are often high, liability of 0.95 would give a confidence interval falling between 0.85 and 0.90. Intraclass corre- of ±8.8. lations tend to give slightly lower numerical val- Because our book is concerned with evaluat- ues than the Pearson equivalent. Articles by Rule ing measurement methods, we need to suggest et al. (178, Table 2) and by Yesavage et al. (179, what level of reliability coefficient is adequate. Table 2) illustrated alternative indicators of in- As with most interesting topics, no answer is ab- ternal consistency for depression scales. The solute; as stated above, the purpose of the mea- commonly used Cronbach alpha coefficients surement influences the standard of reliability consistently gave slightly higher values than required. Recommended values also vary from split-half reliability coefficients for the same statistic to statistic and are, at best, expressions scale, but the major contrast lies between these of opinion. Helmstadter, for example, quoted and the mean inter-item correlations and item- desirable reliability values for various types of total correlations. It is crucial to recognize that psychological tests intended for individuals, the alpha reflects not only intercorrelations between 46 Measuring Health items, but also the number of items. For a three- success of the treatment or, in a negative direc- item scale, an alpha of 0.80 corresponds to tion, the need for a change in the patient’s man- mean inter-item correlations of 0.57; for a ten- agement (183). In recent years, the term has item scale with the same alpha, the correlations been changed to “minimally important differ- will be 0.28 (167, p101). In a 20-item scale, an ence” (MID) (184, pp208–9). The MID gener- alpha of 0.80 corresponded to a mean inter-item ally considers both reliability and a subjective correlation of only 0.14 and an item-total corre- judgment of importance, perhaps made by the lation of 0.50. An alpha of 0.94 corresponded to patients being treated. Unfortunately, it is not a mean inter-item correlation of 0.36 and a me- easy to establish the MID for a measure; there dian item-total coefficient of 0.56 (179, Table are different approaches, and it is possible that 2). Similar results have been reported elsewhere: MIDs should vary according to the context, a mean inter-item correlation of 0.15 for 17 rather than representing a fixed number (185). items corresponded to an alpha of 0.76 for the MIDs are beginning to be reported for health Hamilton Rating Scale for Depression (180, measures, and where available, are included in pp34–35). So (and as explained earlier), high al- the reviews in this book. pha values are consistent with only modest agreement between individual items. McHorney et al. reported both inter-item correlations and alpha values for eight scales of the SF-36 instru- Summary ment; the alpha values were typically 0.15 to Recent years have seen continued technical ad- 0.30 higher than the inter-item correlations, vances in the methods used to develop and test whereas the correlation between the two statis- health measurement instruments. This has come tics was only 0.27 (181, Table 7). in part through the importation of techniques Finally, Wolf and Cornell complicated the from disciplines such as psychometry and educa- whole issue by warning against lightly dismiss- tional measurement, but we have also seen home- ing low correlations (182). They described a grown advances in procedures for assessing and technique that translates correlations derived expressing validity, reliability, responsiveness, from 2 × 2 tables into the metaphor of the esti- and clinically important change. In past years, mated difference in probability of success be- many widely used scales were developed by indi- tween treatment and control groups. They vidual clinicians, based mainly on their personal show, for example, that a correlation of 0.30 experience. These days seem to be numbered as (implying shared variance of only 9%) trans- we move toward greater technical and statistical lates into an increase in success rate from 0.35 sophistication. The process of developing a scale to 0.65: clinically speaking, a major improve- has become a long, complex, and expensive un- ment! Decisions on what may constitute an im- dertaking involving a team of experts, and in provement introduces the notion of the most cases, the quality of the resulting method is “minimal clinically important difference” better. We should be careful, however, not to for- (MCID). Because of the unreliability of individ- get the importance of sound clinical insight into ual measurements, we cannot interpret a small the nature of the condition being measured; the change in a health measure (e.g., pretreatment ideal is to use statistically correct procedures to and posttreatment) as necessarily representing refine an instrument whose content is based on an important contrast. Conversely, large sample clinical wisdom and common sense. sizes may make rather trivial differences in health scores between groups statistically signif- icant, but these may still not represent meaning- ful or important differences. The MCID refers References to a threshold for differences in scores on a (1) Rosser R. A history of the development of measure that represents a noteworthy change health indicators. In: Teeling-Smith G, ed. and could therefore be regarded as indicating Measuring the social benefits of medicine.  The Theoretical and Technical Foundations of Health Management 47 London: Office of Health Economics, years of the World Health Organization. 1983:50–62. Geneva: World Health Organization, 1958. (2) Rosser R. The history of health related (16) Colvez A, Blanchet M. Disability trends in quality of life in 101⁄ 2 paragraphs. J R Soc the United States population 1966-76: Med 1993;86:315–318. analysis of reported causes. Am J Public (3) Ware JE, Brook RH, Davies AR, et al. Health 1981;71:464–471. Choosing measures of health status for (17) Wilson RW. Do health indicators indicate individuals in general populations. Am J health? Am J Public Health 1981;71:461– Public Health 1981;71:620–625. 463. (4) Rosser R. Issues of measurement in the (18) Guralnik JM, Branch LG, Cummings SR, design of health indicators: a review. In: et al. Physical performance measures in Culyer AJ, ed. Health indicators. aging research. J Gerontol Amsterdam: North Holland Biomedical 1989;44:M141–M146. Press, 1983:36–81. (19) Podsiadlo D, Richardson S. The Timed (5) Murray CJL, Salomon JA, Mathers CD, et “Up & Go”: a test of basic functional al. Summary measures of population mobility for frail elderly persons. J Am health: concepts, ethics, measurement and Geriatr Soc 1991;39:142–148. applications. 1st ed. Geneva: World Health (20) Weiner DK, Duncan PW, Chandler J, et al. Organization, 2002. Functional reach: a marker of physical (6) Murray CJL, Lopez AD. Evidence-based frailty. J Am Geriatr Soc 1992;40:203–207. health policy–lessons from the Global (21) Elinson J. Introduction to the theme: Burden of Disease study. Science sociomedical health indicators. Int J Health 1996;274:740–743. Serv 1978;6:385–391. (7) Murray CJL, Salomon J, Mathers C. A (22) Katz S, Akpom CA, Papsidero JA, et al. critical examination of summary measures Measuring the health status of populations. of population health. Bull WHO In: Berg RL, ed. Health status indexes. 1999;78:981–994. Chicago: Hospital Research and (8) Rosser RM. Recent studies using a global Educational Trust, 1973:39–52. approach to measuring illness. Med Care (23) Moskowitz E, McCann CB. Classification 1976;14(suppl):138–147. of disability in the chronically ill and aging. (9) Chapman CR. Measurement of pain: J Chronic Dis 1957;5:342–346. problems and issues. In: Bonica JJ, Albe- (24) Bombardier C, Tugwell P. A Fessard DG, eds. Advances in pain research methodological framework to develop and and therapy. Vol. I. New York: Raven select indices for clinical trials: statistical Press, 1976:345–353. and judgmental approaches. J Rheumatol (10) Moriyama IM. Problems in the measure- 1982;9:753–757. ment of health status. In: Sheldon EB, (25) Tugwell P, Bombardier C. A methodologic Moore W, eds. Indicators of social change: framework for developing and selecting concepts and measurements. New York: endpoints in clinical trials. J Rheumatol Russell Sage, 1968:573–599. 1982;9:758–762. (11) Erickson P. Assessment of the evaluative (26) Apgar V. A proposal for a new method of properties of health status instruments. evaluation of the newborn infant. Anesth Med Care 2000; 38(suppl II):II-95– Analg 1953;32:260–267. II-99. (27) Kirshner B, Guyatt G. A methodologic (12) Morris JN. Uses of epidemiology. 3rd ed. framework for assessing health indices. J London: Churchill Livingstone, 1975. Chronic Dis 1985;38:27–36. (13) Gruenberg EM. The failures of success. (28) Kind P, Carr-Hill R. The Nottingham Milbank Q 1977;55:1–24. Health Profile: a useful tool for (14) Wilkins R, Adams OB. Health expectancy epidemiologists? Soc Sci Med in Canada, late 1970s: demographic, 1987;25:905–910. regional and social dimensions. Am J (29) Bennett KJ, Torrance GW. Measuring Public Health 1983;73:1073–1080. health state preferences and utilities: rating (15) World Health Organization. The first ten scale, time trade-off, and standard gamble 48 Measuring Health techniques. In: Spilker B, ed. Quality of life (42) Torgerson WS. Theory and methods of and pharmacoeconomics in clinical trials. scaling. New York: Wiley, 1958. Philadelphia: Lippincott-Raven, (43) Nunnally JC. Psychometric theory. 2nd ed. 1996:253–265. New York: McGraw-Hill, 1978. (30) Garratt A, Schmidt L, Mackintosh A, et al. (44) Bradburn NM, Miles C. Vague quantifiers. Quality of life measurement: bibliographic Public Opin Q 1979;43:92–101. study of patient assessed health outcome (45) Szabo S. The World Health Organization measures. Br Med J 2002;324:1417–1421. quality of life (WHOQOL) assessment (31) Cella D, Hernandez L, Bonomi AE, et al. instrument. In: Spilker B, ed. Quality of Spanish language translation and initial life and pharmacoeconomics in clinical validation of the Functional Assessment of trials. Philadelphia: Lippincott-Raven, Cancer Therapy quality-of-life instrument. 1996:355–362. Med Care 1998;36:1407–1418. (46) Merbitz C, Morris J, Grip JC. Ordinal (32) Millon T, Davis RD. The place of scales and foundations of misinference. assessment in clinical science. In: Millon T, Arch Phys Med Rehabil 1989;70:308–312. ed. The Millon inventories: clinical and (47) McClatchie G, Schuld W, Goodwin S. A personality assessment. New York: maximized-ADL index of functional status Guilford, 1997:3–20. for stroke patients. Scand J Rehabil Med (33) Stoker MJ, Dunbar GC, Beaumont G. The 1983;15:155–163. SmithKline Beecham “quality of life” scale: (48) Cox DR, Wermuth N. Tests of linearity, a validation and reliability study in patients multivariate normality and the adequacy of with affective disorder. Qual Life Res linear scores. Appl Stat 1994;43:347–355. 1992;1:385–395. (49) Silverstein B, Fisher WP, Kilgore KM, et al. (34) Thunedborg K, Allerup P, Bech P, et al. Applying pychometric criteria to functional Development of the repertory grid for mea- assessment in medical rehabilitation: II. surement of individual quality of life in Defining interval measures. Arch Phys Med clinical trials. Int J Methods Psychiatr Res Rehabil 1992;73:507–518. 1993;3:45–56. (50) Torrance GW, Feeny D. Utilities and (35) Mor V, Guadagnoli E. Quality of life mea- quality-adjusted life years. Int J Technol surement: a psychometric Tower of Babel. J Assess Health Care 1989;5:559–575. Clin Epidemiol 1988;41:1055–1058. (51) Torrance GW, Furlong W, Feeny D, et al. (36) Idler EL, Benyami Y. Self-rated health and Multi-attribute preference functions: mortality: a review of twenty-seven Health Utilities Index. Pharmacoeconomics community studies. J Health Soc Behav 1995;7:503–520. 1997;38:21–37. (52) Edwards AL. Techniques of attitude scale (37) Feeny D, Torrance GW, Furlong WJ. construction. New York: Appleton-Century- Health Utilities Index. In: Spilker B, ed. Crofts, 1975. Quality of life and pharmacoeconomics in (53) Young FW. Scaling. Annu Rev Psychol clinical trials. Philadelphia: Lippincott- 1984;35:55–81. Raven, 1996:239–252. (54) Stevens SS. Measurement, psychophysics (38) Stevens SS. The surprising simplicity of and utility. In: Churchman CW, Ratoosh P, sensory metrics. Am Psychol 1962;17:29– eds. Measurement, definitions and theories. 39. New York: Wiley, 1959:18–63. (39) Baird JC, Noma E. Fundamentals of (55) Streiner DL, Norman GR. Health measure- scaling and psychophysics. New York: ment scales: a practical guide to their Wiley, 1978. development and use. 3rd ed. New York: (40) Lodge M. Magnitude scaling: quantitative Oxford, 2003. measurement of opinions. Beverly Hills, (56) Lord FM. Applications of item response California: Sage Publications (Quantitative theory to practical testing problems. Applications in the Social Sciences No. 07- Hillsdale, New Jersey: Lawrence Erlbaum 001), 1981. Associates, 1980. (41) Guilford JP. Psychometric methods. 2nd (57) Hays RD, Morales LS, Reise SP. Item ed. New York: McGraw-Hill, 1954. response theory and health outcomes mea-  The Theoretical and Technical Foundations of Health Management 49 surement in the 21st century. Med Care (70) Froberg DG, Kane RL. Methodology for 2000;38(suppl II):II-28–II-42. measuring health-state preferences—II: (58) Duncan-Jones P, Grayson DA, Moran PAP. scaling methods. J Clin Epidemiol The utility of latent trait models in 1989;42:459–471. psychiatric epidemiology. Psychol Med (71) Kaplan RM. Application of a General 1986;16:391–405. Health Policy Model in the American (59) Wright BD, Masters G. Rating scale health care crisis. J R Soc Med analysis. Chicago: MESA Press, 1982. 1993;86:277–281. (60) Delis DC, Jacobson M, Bondi MW, et al. (72) Gudex C, Dolan P, Kind P, Williams A. The myth of testing construct validity Health state valuations from the general using factor analysis or correlations public using the visual analogue scale. Qual with normal or mixed clinical Life Res 1996;5:521–531. populations: lessons from memory (73) Feeny DH, Torrance GW. Incorporating assessment. Int J Neuropsychol Soc utility-based quality-of-life assessment 2003;9:936–946. measures in clinical trials. Med Care (61) Harris D. Comparison of the 1-, 2- and 3- 1989;27(suppl. 3):S190–S2004. parameter IRT models. Educ Meas Issues (74) Cox DR, Fitzpatrick R, Fletcher AE, et al. Pract 1989;8:35–41. Quality-of-life assessment: can we keep it (62) Teresi JA, Kleinman M, Ocepek-Welikson simple? J R Stat Soc [Ser A] 1992; K, et al. Applications of item response 155:353–393. theory to the examination of the (75) Kaplan SH. Patient reports of health status psychometric properties and differential as predictors of physiologic health mea- item functioning of the Comprehensive sures in chronic disease. J Chronic Dis Assessment and Referral Evaluation 1987;40(suppl):S27–S35. dementia diagnostic scale. Res Aging (76) Goldberg DP. The detection of psychiatric 2000;22:738–773. illness by questionnaire. London: Oxford (63) Hambleton RK, Jones RW. Comparison of University Press (Maudsley Monograph classical test theory and item response No. 21), 1972. theory and their applications to test (77) Crowne DP, Marlowe D. A new scale of development. Educ Meas Issues Pract social desirability independent of 1993;12:39–47. psychopathology. J Consult Psychol (64) Hambleton RK, Swaminathan H, Rogers 1960;24:349–354. HJ. Fundamentals of item response theory. (78) Edwards AL. The social desirability Newbury Park, California: Sage, 1991. variable in personality assessment and (65) Johannesson M, Pliskin JS, Weinstein MC. research. New York: Dryden, 1957. Are healthy-years equivalents an (79) Kozma A, Stones MJ. Social desirability in improvement over quality-adjusted life measures of subjective well-being: a years? Med Decis Making systematic evaluation. J Gerontol 1993;13:281–286. 1987;42:56–59. (66) Robinson R. Cost-utility analysis. Br Med (80) Bradburn NM, Sudman S, Blair E, J 1993;307:859–862. Stocking C. Question threat and response (67) LaPuma J, Lawlor EF. Quality-adjusted life bias. Public Opin Q 1978;42:221–234. years: ethical implications for physicians (81) Totman R. Social causes of illness. London: and policymakers. JAMA Souvenir Press, 1979. 1990;263:2917–2921. (82) Sechrest L, Pitz D. Commentary: measur- (68) Patrick DL, Erickson P. Health status and ing the effectiveness of heart transplant health policy: quality of life in health care programmes. J Chronic Dis evaluation and resource allocation. New 1987;40:S155–S158. York: Oxford University Press, 1993. (83) Schwartz CE, Sprangers MAG. Adaptation (69) Norum J, Angelsen V, Wist E, et al. to changing health: response shifts in Treatment costs in Hodgkin’s disease: a quality of life research. Washington, DC: cost-utility analysis. Eur J Cancer American Psychological Association, 2002. 1996;32A:1510–1517. (84) Sprangers MAG, Schwartz CE. Integrating 50 Measuring Health response shift into health-related quality of (96) World Health Organization. International life research: a theoretical model. Soc Sci classification of impairments, disabilities, Med 1999;48:1507–1515. and handicaps. A manual of classification (85) Paterson C. Seeking the patient’s relating to the consequences of disease. perspective: a qualitative assessment of Geneva: World Health Organization, 1980. EuroQol, COOP-WONCA charts and (97) Duckworth D. The need for a standard MYMOP. Qual Life Res 2004;13:871– terminology and classification of 881. disablement. In: Granger CV, Gresham (86) Guyatt GH, Kirshner B, Jaeschke R. Mea- GE, eds. Functional assessment in suring health status: what are the necessary rehabilitation medicine. Baltimore, Mary- measurement properties? J Clin Epidemiol land: Williams & Wilkins, 1984:1–13. 1992;45:1341–1345. (98) Nagi SZ. The concept and measurement (87) Hertzog C. Applications of signal detection of disability. In: Berkowitz ED, ed. theory to the study of psychological aging: Disability policies and government a theoretical review. In: Poon LW, ed. programs. New York: Praeger, 1979:1–15. Aging in the 1980s: psychological issues. (99) Patrick DL, Bergner M. Measurement of Washington, DC: American Psychological health status in the 1990s. Annu Rev Association, 1980:568–591. Public Health 1990;11:165–183. (88) Rollman GB. Signal detection theory (100) World Health Organization. International assessment of pain modulation: a critique. Classification of Functioning, Disability In: Bonica JJ, Albe-Fessard DG, eds. and Health. [Publ] A54/18. Geneva: Advances in pain research and therapy. Vol. World Health Organization, 2001. I. New York: Raven Press, 1976:355–362. (101) Simeonsson RJ, Lollar D, Hollowell J, et (89) Clark WC. Pain sensitivity and the report al. Revision of the International of pain: an introduction to sensory decision Classification of Impairments, Disabilities theory. In: Weisenberg M, Tursky B, eds. and Handicaps. Developmental issues. J Pain: new perspectives in therapy and Clin Epidemiol 2000;53:113–124. research. New York: Plenum Press, (102) Üstün TB, Chatterji S, Kostansjek N, et 1976:195–222. al. WHO’s ICF and functional status (90) Swets JA. The relative operating information in health records. Health characteristic in psychology. Science Care Financ Rev 2003;24(3):77–88. 1973;182:990–1000. (103) Ahlbom A, Norell S. Introduction to (91) McNeil BJ, Keeler E, Adelstein SJ. Primer modern epidemiology. Chestnut Hill, on certain elements of medical decisi Montana: Epidemiology Resources, 1984. on making. N Engl J Med (104) Webb EJ, Campbell DT, Schwartz RD, et 1975;293:211–215. al. Unobtrusive measures: nonreactive (92) Yaremko RM, Harari H, Harrison RC, research in the social sciences. Chicago: Lynn E. Reference handbook of research Rand McNally, 1966. and statistical methods in psychology. New (105) Anastasi A. Psychological testing. New York: Harper & Row, 1982. York: Macmillan, 1968. (93) Bech P. The PCASEE model: an approach (106) Helmstadter GC. Principles of to subjective well-being. In: Orley J, psychological measurement. London: Kuyken W, eds. Quality of life assessment: Methuen, 1966. international perspectives. Berlin: Springer- (107) American Psychological Association. Verlag, 1993:75–79. Standards for educational and (94) Gallin RS, Given CW. The concept and psychological testing. Washington, DC: classification of disability in health American Psychological Association, 1985. interview surveys. Inquiry (108) Feinstein AR. Clinimetrics. New Haven, 1976;13:395–407. Connecticut: Yale University Press, 1987. (95) Slater SB, Vukmanovic C, Macukanovic (109) Ley P, Florio T. The use of readability P, et al. The definition and measurement formulas in health care. Psychol Health of disability. Soc Sci Med Med 1996;1:7–28. 1974;8:305–308. (110) Seiler LH. The 22-item scale used in field  The Theoretical and Technical Foundations of Health Management 51 studies of mental illness: a question of (123) Clarke DM, McKenzie DP. A caution on method, a question of substance, and a the use of cut-points applied to screening question of theory. J Health Soc Behav instruments or diagnostic criteria. J 1973;14:252–264. Psychiatr Res 1994;28:185–188. (111) Murphy JM, Berwick DM, Weinstein (124) Jackson DN. A sequential system for MC, et al. Performance of screening and personality scale development. In: diagnostic tests. Arch Gen Psychiatry Spielberger CD, ed. Current topics in 1987;44:550–555. clinical and community psychology, (112) Swets JA. Measuring the accuracy of volume II. New York: Academic Press, diagnostic systems. Science 1970. 1988;240:1285–1293. (125) Ramanaiah NV, Franzen M, Schill T. A (113) Clarke DM, Smith GC, Herrman HE. A psychometric study of the State-Trait comparative study of screening Anxiety Inventory. J Pers Assess instruments for mental disorders in 1983;47:531–535. general hospital patients. Int J Psychiatry (126) Brazier J, Jones N, Kind P. Testing the Med 1993;23:323–337. validity of the EuroQol and comparing it (114) Lewis G, Wessely S. Comparison of the with the SF-36 health survey General Health Questionnaire and the questionnaire. Qual Life Res Hospital Anxiety and Depression Scale. 1993;2:169–180. Br J Psychiatry 1990;157:860–864. (127) Bland JM, Altman DG. Statistical (115) Kraemer HC. Assessment of 2 × 2 methods for assessing agreement between associations: generalization of signal- two methods of clinical measurement. detection methodology. Am Stat Lancet 1986;1:307–310. 1988;42:37–49. (128) Kemmler G, Holzner B, Kopp M, et al. (116) Clarke DM, McKenzie DP. Screening for Multidimensional scaling as a tool for psychiatric morbidity in the general analysing quality of life data. Qual Life hospital: methods for comparing the Res 2002;11:223–233. validity of different instruments. Int J (129) Comrey AL. Common methodological Methods Psychiatr Res 1991;1:79–87. problems in factor analytic studies. J (117) Fletcher RH, Fletcher SW, Wagner EH. Consult Clin Psychol 1978;46:648–659. Clinical epidemiology: the essentials. (130) Boyle GJ. Self-report measures of Baltimore, Williams & Wilkins, 1988. depression: some psychometric (118) Glas AS, Lijmer JG, Prins MH, et al. The considerations. Br J Clin Psychol diagnostic odds ratio: a single indicator of 1985;24:45–59. test performance. J Clin Epidemiol (131) Byrne BM, Baron P, Larsson B, et al. The 2003;56:1129–1135. Beck Depression Inventory: testing and (119) Ransohoff DF, Feinstein AR. Problems of cross-validating a second-order factorial spectrum and bias in evaluating the structure for Swedish nonclinical efficacy of diagnostic tests. N Engl J Med adolescents. Behav Res Ther 1978;299:926–929. 1995;33:345–356. (120) Lachs MS, Nachamkin IN, Edelstein PH, (132) Sun J. Adjusting distributions of the et al. Spectrum bias in the evaluation of Health Utilities Index Mark 3 utility diagnostic tests: lessons from the rapid scores of health-related quality of life. dipstick test for urinary tract infection. Qual Life Res 2003;12:11–20. Ann Intern Med 1992;117:135–140. (133) Felson DT, Anderson JJ, Meenan RF. The (121) Begg CB, McNeil BJ. Assessment of comparative efficacy and toxicity of radiologic tests: control of bias and other second-line drugs in rheumatoid arthritis. design considerations. Radiology Arthritis Rheum 1990;33:330–338. 1988;167:565–569. (134) Liang MH, Fossel AH, Larson MG. (122) Choi BCK. Sensitivity and specificity of a Comparisons of five health status single diagnostic test in the presence of instruments for orthopedic evaluation. workup bias. J Clin Epidemiol Med Care 1990;28:632–642. 1992;45:581–586. (135) Liang MH. Longitudinal construct 52 Measuring Health validity: establishment of clinical meaning (146) MacKenzie CR, Charlson ME, DiGioia in patient evaluative instruments. Med D, et al. A patient-specific measure of Care 2000;38(suppl II):84–90. change in maximal function. Arch Intern (136) Terwee CB, Dekker FW, Wiersinga WM, Med 1986;146:1325–1329. et al. On assessing responsiveness of (147) Lambert MJ, Hatch DR, Kingston MD, et health-related quality of life instruments: al. Zung, Beck, and Hamilton rating guidelines for instrument evaluation. Qual scales as measures of treatment outcome: Life Res 2003;12:349–362. a meta-analytic comparison. J Consult (137) Kazis LE, Anderson JJ, Meenan RF. Clin Psychol 1986;54:54–59. Effect sizes for interpreting changes in (148) Patrick DL, Wild DJ, Johnson ES, et al. health status. Med Care Cross-cultural validation of quality of life 1989;27(suppl):S178–S189. measures. In: Orley J, Kuyken W, eds. (138) Tidermark J, Bergström G, Svensson O, Quality of life assessment: international et al. Responsiveness of the EuroQol perspectives. Berlin: Springer-Verlag, (EQ 5-D) and the SF-36 in elderly 1993:19–32. patients with displaced femoral neck (149) Cohen J. Power analysis for the fractures. Qual Life Res behavioral sciences. New York: Academic 2003;12:1069–1079. Press, 1977. (139) Siu AL, Ouslander JG, Osterweil D, et al. (150) Fleishman JA, Lawrence WF. Demographic Change in self-reported functioning in variation in SF-12 scores: true differences older persons entering a residential care or differential item functioning? Med Care facility. J Clin Epidemiol 2003;41:III-75–III-86. 1993;46:1093–1101. (151) Morales LS, Reise SP, Hays RD. (140) Katz JN, Larson MG, Phillips CB, et al. Evaluating the equivalence of health care Comparative measurement sensitivity of ratings by whites and Hispanics. Med short and longer health status instruments. Care 2000;38:517–527. Med Care 1992;30:917–925. (152) Teresi JA, Kleinman M, Ocepek-Welikson (141) O’Carroll RE, Smith K, Couston M, et al. K. Modern psychometric methods for A comparison of the WHOQOL-100 and detection of differential item functioning: the WHOQOL-BREF in detecting change application to cognitive assessment mea- in quality of life following liver sures. Stat Med 2000;19:1651–1683. transplantation. Qual Life Res (153) McHorney CA, Ware JE, Jr., Raczek AE. 2000;9:121–124. The MOS 36-Item Short-Form Health (142) Liang MH, Larson MG, Cullen KE, et al. Survey (SF-36): II. Psychometric and Comparative measurement efficiency and clinical tests of validity in measuring sensitivity of five health status physical and mental health constructs. instruments for arthritis research. Med Care 1993;31:247–263. Arthritis Rheum 1985;28:542–547. (154) Kantz ME, Harris WJ, Levitsky K, et al. (143) Deyo RA, Centor RM. Assessing the Methods for assessing condition-specific responsiveness of functional scales to and generic functional status outcomes clinical change: an analogy to diagnostic after total knee replacement. Med Care test performance. J Chronic Dis 1992;30(suppl):MS240–MS252. 1986;39:897–906. (155) Bartko JJ. Measurement and reliability: (144) Deyo RA, Diehr P, Patrick DL. statistical thinking considerations. Reproducibility and responsiveness of Schizophr Bull 1991;17:483–489. health status measures. Statistics and (156) Siegert CEH, Vleming L-J, van den strategies for evaluation. Control Clin Broucke JP, et al. Measurement of Trials 1991;12:142S–158S. disability in Dutch rheumatoid arthritis (145) MacKenzie CR, Charlson ME, DiGioia patients. Clin Rheumatol D, et al. Can the Sickness Impact Profile 1984;3:305–309. measure change? An example of scale (157) Bartko JJ. The intraclass correlation as a assessment. J Chronic Dis measure of reliability. Psychol Rep 1986;39:429–438. 1966;19:3–11.  The Theoretical and Technical Foundations of Health Management 53 (158) Bartko JJ. On various intraclass (171) Bohrnstedt GW. Measurement. In: Rossi correlation reliability coefficients. Psychol PH, Wright JD, Anderson AB, eds. Bull 1976;83:762–765. Handbook of survey research. New York: (159) Shrout PE, Fleiss JL. Intraclass Academic Press, 1983:69–95. correlations: uses in assessing rater (172) Shrout PE, Yager TJ. Reliability and reliability. Psychol Bull 1979;86:420–428. validity of screening scales: effect of (160) McHorney CA, Tarlov AR. Individual- reducing scale length. J Clin Epidemiol patient monitoring in clinical practice: are 1989;42:69–78. available health status surveys adequate? (173) Bollen K, Lennox R. Conventional Qual Life Res 1995;4:293–307. wisdom on measurement: a structural (161) Beckerman H, Roebroeck ME, Lankhorst equation perspective. Psychol Bull GJ, et al. Smallest real difference, a link 1991;110:305–314. between reproducibility and (174) Andrews FM, Withey SB. Social responsiveness. Qual Life Res indicators of well-being: Americans’ 2001;10:571–578. perceptions of life quality. New York: (162) Christensen L, Mendoza JL. A method of Plenum, 1976. assessing change in a single subject: an (175) Williams JI. Ready, set, stop. Reflections alteration of the RC index. Behav Ther on assessing quality of life and the 1986;17:305–308. WHOQOL-100 (U.S. version). J Clin (163) Jacobson NS, Follette WC, Revenstorf D. Epidemiol 2000;53:13–17. Psychotherapy outcome research: methods (176) Landis JR, Koch GG. The measurement for reporting variability and evaluating of observer agreement for categorical clinical significance. Behav Ther data. Biometrics 1977;33:159–174. 1984;15:336–352. (177) Cicchetti DV, Sparrow SA. Developing (164) Evans WJ, Cayten CG, Green PA. criteria for establishing interrater Determining the generalizability of rating reliability of specific items: applications to scales in clinical settings. Med Care assessment of adaptive behavior. Am J 1981;19:1211–1220. Ment Defic 1981;86:127–137. (165) Brennan RL. Generalizability theory. Educ (178) Rule BG, Harvey HZ, Dobbs AR. Meas Issues Pract 1992;11(4):27–34. Reliability of the Geriatric Depression (166) Marx RG, Menezes A, Horovitz L, et al. Scale for younger adults. Clin Gerontol A comparison of two time intervals for 1989;9:37–43. test-retest reliability of health status (179) Yesavage JA, Brink TL, Rose TL, et al. instruments. J Clin Epidemiol Development and validation of a geriatric 2003;56:730–735. depression screening scale: a preliminary (167) Cortina JM. What is coefficient alpha? En report. J Psychiatr Res 1983;17:37–49. examination of theory and applications. J (180) Mehm LP, O’Hara MW. Item Appl Psychol 1993;78:98–104. characteristics of the Hamilton Rating (168) Ferketich F. Internal consistency estimates Scale for Depression. J Psychiatr Res of reliability. Res Nurs Health 1985;19:31–41. 1990;13:437–440. (181) McHorney CA, Ware JE, Jr., Lu JFR, et (169) Hays RD, Hayashi T. Beyond internal al. The MOS 36-item Short-Form Health consistency reliability: rationale and user’s Survey (SF-36): III. Tests of data quality, guide for the Multitrait Analysis Program scaling assumptions, and reliability across on the microcomputer. Behav Res diverse patient groups. Med Care Methods Instrum Comput 1994;32:40–66. 1990;22:167–175. (182) Wolf FM, Cornell RG. Interpreting (170) Kaasa S, Bjordal K, Aaronson N, et al. behavioral, biomedical, and The EORTC core quality of life psychological relationships in chronic questionnaire (QLQ-C30): validity and disease from 2 × 2 tables using reliability when analyzed with patients correlation. J Chronic Dis treated with palliative radiotherapy. Eur J 1986;39:605–608. Cancer 1995;31A:2260–2263. (183) Jaeschke R, Singer J, Guyatt GH. Mea- 54 Measuring Health surement of health status. Ascertaining and worsening. Qual Life Res the minimal clinically important 2002;11:207–221. difference. Control Clin Trials (185) Beaton DE, Boers M, Wells GA. Many 1989;10:407–415. faces of the minimal clinically important (184) Cella D, Hahn EA, Dineen K. Meaningful difference (MCID): a literature review and change in cancer-specific quality of life directions for future research. Curr Opin scores: differences between improvement Rheumatol 2002;14:109–114.  3 Physical Disability and Handicap ecause they cover an area of such fundamen- The Evolution of B tal concern in health care, there has been a proliferation of scales designed to measure Physical Disability Measurements physical disability and daily function. Available The concepts of impairment, disability, and measurement methods serve a variety of pur- handicap were introduced in Chapter 2 and this poses: some apply to particular diseases whereas conceptual framework is reflected in the evolu- others are broadly applicable; some assess only tion of functional assessments. Measurements in impairments, whereas others have a broader this field began with impairment scales (covering scope and cover disability, handicap, or prob- physical capacities such as balance, sensory abil- lems of the social environment; there are evalu- ities, and range of motion); attention then ative measures, screening tests, and clinical shifted toward measuring disability (gross body rating scales; some methods are designed for se- movements and self-care ability), and later verely ill inpatients, whereas others are for out- moved to assessments of handicap (fulfillment of patients with lower levels of disability. Well over social roles, working ability, and household ac- 100 activities of daily living (ADL) scales are tivities). described in the literature (1–11), but many Formal measurements of physical impairments fewer have achieved widespread use and we fo- began with diagnostic tests and standardized cus only on those here. We have omitted little- medical summaries of a patient’s condition; they used scales, those that have not stood the test of were typically used with older or chronically ill time, and those that lack published evidence for patients. The measurements were mostly rating validity and reliability. As it turns out, our selec- scales applied by a clinician; they are represented tion is comparable to that made quite indepen- in this chapter by the PULSES Profile. These mea- dently in other reviews (9; 12). The main surements were often used in assessing fitness for surprise, perhaps, will be the inclusion of sev- work or in reviewing claims for compensation for eral older scales in the selection. Many newer accidents and injuries; the emphasis was on stan- methods lack the information on reliability and dardized ratings that could withstand legal exam- validity that has been accumulated for the older ination. It was later recognized that although instruments, and we have not included methods impairment may be accurately assessed, it is by no simply because they are new. To introduce the means the only factor that predicts a patient’s basis for our selection, we begin with a brief need for care: environmental factors, the avail- historical overview of the development of the ability of social support, and the patient’s per- field. This identifies the main categories of phys- sonal will and determination all affect how far an ical and functional disability measurement, and impairment will cause disability or handicap. As the chapter reviews examples from each the scope of rehabilitation expanded to include category. the return of patients to an independent existence, the assessment of impairments was no longer suf- 55 56 Measuring Health ficient and it became important to measure dis- tients to meaningful social roles, and this has in- ability and handicap as well. Assessment methods spired measurement scales that cover social ad- were broadened to consider the activities a patient justment as well as physical abilities. To assess a could or did perform at his level of physical ca- patient’s ability to live in the community requires pacity. Assessments of this type are generally information on the level of disability, on the en- termed “functional disability” indicators. Most of vironment in which the patient has to live, on the scales we review are measures of functional the amount of social support that may be avail- ability and disability; the chapter title refers to able, and on some of the compensating factors “physical disability” to distinguish physical, that determine whether a disability becomes a rather than mental, problems as the source of the handicap. The IADL scales cover one part of this functional limitations. The ADL scales are typi- area, but other, more extensive scales have been cal; an early example was Katz’s index. This was developed to record factors that may explain dif- developed in 1957 to study the effects of treat- ferent levels of handicap for a given disability ment on the elderly and chronically ill. It summa- such as the type of work the patient does, his rizes the patient’s degree of independence in housing, his personality, and the social support bathing, dressing, using the toilet, moving around available. Such extensions to the original theme the house, and eating—topics that Katz selected to of functional disability produce measurements represent “primary biological functions.” Katz’s that are conceptually close to the indexes of so- scale is one of the few instruments to provide a cial functioning described in Chapter 4. IADL theoretical justification for the topics it includes. scales are also commonly used with less severely Unfortunately, most other ADL scales are not handicapped populations, often in general popu- built on any conceptual approach to disability, lation surveys, and cover activities needed for and little systematic effort has been made to spec- continued community residence. Forer’s Func- ify which topics should be covered in such scales. tional Status Rating System, the Kenny Self-Care In part because of this, progress in the field was Evaluation, and the SMAF are examples. They uncoordinated, and scales proliferated apparently improve on the sensitivity of ADL scales that at the whim of their creators. Furthermore, scant were found not to identify low levels of disabil- attention was paid to formal testing of the early ity, nor minor changes in level of disability. ADL methods and we know little about their Some general issues inherent in the design of comparative validity and reliability. ADL and IADL scales should be borne in mind ADL scales such as the Katz, Barthel Index, or by those who are choosing a scale for a particu- the Health Assessment Questionnaire are con- lar application. As noted earlier, most ADL cerned with more severe levels of disability, rele- questions reflect relatively severe levels of dis- vant mainly to institutionalized patients and to the ability and so are insensitive to variations at the elderly. In 1969, Lawton and Brody extended the upper levels of functioning, where most people ADL concept to consider problems more typically score. Care must be taken, therefore, in selecting experienced by those living in the community: a measure for use in surveys or with relatively mobility, difficulty in shopping, cooking, or man- healthy patients. Instruments such as the Med- aging money, a field that came to be termed “In- ical Outcomes Study Physical Functioning Mea- strumental Activities of Daily Living” (IADL) or sure include items on more strenuous physical “Performance Activities of Daily Living” (PADL) activities, while still retaining items on basic (13). IADLs are more complex and demanding abilities such as dressing or walking. The other than basic ADLs; they offer indicators of “ap- approach is to rely on IADL items to reflect plied” problems that include elements of the higher levels of function. Analyses using item re- handicap concept. sponse theory have, for example, shown that The development of IADL scales was stimu- ADL and IADL items fall on a single underlying lated in part by the movement toward commu- dimension (14). However, IADL scales are not nity care for the elderly. Rehabilitative medicine pure measures of physical function: activities has increasingly stressed the need to restore pa- such as cooking, shopping, and cleaning reflect  Physical Disability and Handicap 57 cognitive abilities and established social roles as such as “Do you have difficulty with . . . ?” as well as physical capacity. However much their used in the Lambeth Disability Screening Ques- wives may complain, there are reasons other than tionaire and the Organization for Economic Co- physical limitations why some men may not operation and Development (OECD) Disability cook or clean house. By comparison, ADL items Questionnaire. Jette used separate measures of on walking or bathing are more likely to offer pain, difficulty and dependency in his Functional pure measures of physical function. Although Status Index (see the reviews in this chapter). ADL scales tend to be universal in content, IADL The uncomfortable truth is that minor varia- scales may vary from culture to culture. For ex- tions in question wording may lead to large dif- ample, a British IADL scale included items on ferences in response patterns, which complicates making tea and carrying a tray, a Dutch scale in- comparisons between studies. Picavet and van cluded making a bed, and a New Zealand scale den Bos, for example, showed that if answer cat- covered gardening ability (15, p704). Although egories are phrased in terms simply of having the IADL scales are newer and have been some- difficulty in doing an activity, fewer people will what better tested than the ADL instruments, respond affirmatively than if the answer cate- many scales still offer little conceptual explana- gories distinguish between having minor or ma- tion or theory to justify their content. jor difficulty (17). Likewise, Jette showed that The distinction between a person’s physical response scales phrased in terms of experiencing capacity and his actual performance in managing difficulty with an activity may produce his life in the face of physical limitations has markedly higher estimates of disability than do been mentioned. Reflecting this contrast, there scales phrased in terms of requiring assistance, are two ways of phrasing questions on func- but the extent of the contrast varies according to tional disability. One can ask what a person can the question topic (18). An alternative is there- do (the “capacity” wording) or what he does do fore to use performance tests of function, in (“performance” wording). Both are common which the subject is assessed while actually per- and both hold advantages and disadvantages. forming the tasks. An early example is the 1980 Asking a patient what he can do may provide a Rivermead ADL scale, which includes observa- hypothetical answer that records what the pa- tions of ADL functions and of IADLs such as tient thinks he can do even though he does not counting change, light housework, and shopping normally attempt it. An index using such ques- (19; 20). A more recent example is the Timed Up tions may exaggerate the healthiness of the re- and Go measure that records the time taken for a spondent—perhaps by as much as 15% to 20% person to rise from a chair, walk 3 meters, turn (16, p70). Although the performance wording back and sit down again (21), and there are may overcome this, it can run the opposite risk. many other examples (22–31). The Functional Factors other than ill health may restrict behav- Independence Measure described in this chapter ior: for reasons of safety or lack of staff, hospital can be applied either as a performance assess- patients may be kept in bed. Performance ques- ment or a self-report questionnaire; the Rapid tions may therefore not be specific to health, so Disability Rating Scale and the PECS are other respondents are commonly asked to consider examples. only the health reasons why they did not do an Since the first edition of this book, there has activity. However, this may be difficult to judge been great progress in the development of func- because health interacts with factors such as the tional disability scales. Psychometric evidence for weather, making it difficult to determine virtually all measurements is accumulating; tech- whether an activity was not performed for niques of scale development such as item response health reasons. Hence, capacity wording is often theory are now routinely applied (32–36), and used in IADL questions that are more suscepti- reference norms for scales are being assembled. ble to such bias than ADL questions. Most ADL Nonetheless, there are inherent challenges in mea- indexes favor the performance approach, al- suring functioning in a culture- and gender-fair though an intermediate phrasing can be used, manner, and the introduction of techniques such 58 Measuring Health as analyses of differential item functioning are be- PULSES, Barthel, and Health Assessment Ques- ginning to be applied (37; 38). Because of the tionnaire are primarily designed for use in inpa- complexity of undertaking test development pro- tient settings. The Kenny, PSMS, Forer, Patient cedures such as these, we may have seen the end Evaluation Conference System, SMAF, and of the era in which an ADL scale could be person- Functional Independence Measure are intended ally developed, tested, and published by a clinician for rehabilitation patients, whereas the Katz or a graduate student. The rate at which new ADL scale, Linn’s Rapid Disability Rating Scale, general-purpose functional disability scales are and the Functional Status Index of Jette can be being produced has fallen, and as we already have used either in clinical or research settings. Five several good ones, this is probably appropriate. scales are designed for use as population survey Attention has shifted toward the development of instruments. These include Pfeffer’s Functional disease-specific measures, some of which are re- Activities Questionnaire, the questionnaire de- viewed in Chapter 10. veloped by the OECD, the Medical Outcomes Study scale, and two disability screening scales developed in England: the Lambeth scale and Scope of the Chapter Bennett and Garrad’s interview schedule. In ad- dition to the measures reviewed in this chapter, The uncertain quality of many available scales readers should bear in mind that many of the simplified the process of choosing which to re- general health measures reviewed in Chapter 10 view. The current review wrestled with finding a also include sections on physical functioning and balance between including large numbers of disability. Examples include the OARS Multidi- measurement methods that we could not really mensional Functional Assessment Questionnaire recommend because their quality remains un- and the Sickness Impact Profile, both of which known, and the opposite extreme of reviewing a contain ADL and IADL sections that may prove very small number of scales of proven quality. suitable for use as stand-alone scales. Some of The selection includes some scales of primarily the scales that were not included are described historical interest (e.g., the PULSES Profile), but briefly in the conclusion to the chapter. mostly the selection is intended to present scales Table 3.1 summarizes the format, length, and that merit serious consideration. We have in- use of each scale and published evidence on its cluded measurements for which the question- reliability and validity. naire is available, for which there is some evidence on reliability or validity, and those which have been used in published studies. We References have sought to keep the scope of the chapter (1) Bruett TL, Overs RP. A critical review of broad and have included methods whose pur- 12 ADL scales. Phys Ther pose is primarily clinical as well as those in- 1969;49:857–862. tended for survey research. (2) Berg RL. Health status indexes. Chicago: This chapter includes descriptions of seven Hospital Research and Educational Trust, traditional ADL scales and ten IADL or mixed 1973. instruments; several scales include both ADL (3) Katz S, Hedrick S, Henderson NS. The and IADL questions. The scales are presented in measurement of long-term care needs and chronological order, which generally corre- impact. Health Med Care Serv Rev 1979;2:1–21. sponds to the evolution from ADL toward IADL (4) Brown M, Gordon WA, Diller L. or mixed scales. It was not originally the inten- Functional assessment and outcome mea- tion to include so many older scales but in many surement: an integrative view. In: Pan EL, instances they have been more fully tested than Backer TE, Vasch CL, eds. Annual review the newer methods, and many are of historical of rehabilitation. Vol. 3. New York: importance in that they influenced the design of Springer, 1983:93–120. subsequent instruments. Of the 17 scales, the (5) Forer SK. Functional assessment  Table 3.1 Comparison of the Quality of Physical Disability Indexes* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results PULSES Profile (Moskowitz and ordinal 6 clinical staff many * ** * ** McCann, 1957) Barthel Index (Mahoney and ordinal 10 clinical staff many *** *** *** ** Barthel, 1955) Katz Index of ADL (Katz, 1959) ordinal 6 clinical staff many * * * ** Kenny Self-Care Evaluation ordinal 85 clinical staff several * * * ** 59 (Schoening et al., 1965) Physical Self-Maintenance Guttman 14 survey self, staff many * ** * ** Scale (Lawton and Brody, 1969) Disability Interview Schedule ordinal 17 survey interviewer few * * * * (Bennett and Garrad, 1970) Lambeth Disability Screening ordinal 25 survey self few * ? * ** (Patrick et al., 1981) OECD Disability Questionnaire ordinal 16 survey self several ** * ** * (OECD, 1981) Functional Status Rating ordinal 30 clinical staff few * ** * * System (Forer, 1981) (continued)  Table 3.1 (continued) Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Rapid Disability Rating Scale ordinal 18 research staff (2 min) several * ** * * (Linn, 1982) Functional Status Index ordinal 54 clinical interviewer several ** ** ** ** (Jette, 1980) Patient Evaluation Conference ordinal 79 clinical staff few * * ** * System (Harvey and Jellinek, 1981) Functional Activities ordinal 10 survey lay informant few * * ** ** 60 Questionnaire (Pfeffer, 1982) Health Assessment ordinal 20 clinical, self, staff many *** *** *** *** Questionnaire (Fries, 1980) research (5–8 min) Medical Outcomes Study ordinal 14 survey self few * ** * * Physical Functioning Measure (Stewart, 1992) Functional Autonomy ordinal 29 clinical staff several ** ** ** *** Measurement System (40 min) (SMAF) (Hébert, 1984) Functional Independence ordinal 18 clinical expert, many ** *** *** *** Measure (Granger and interviewer Hamilton, 1987) * For an explanation of the categories used, see Chapter 1, pages 6–7  Physical Disability and Handicap 61 instruments in medical rehabilitation. J J Epidemiol Community Health Organ Rehabil Evaluators 1982;2:29–41. 1996;50:86–93. (6) Liang MH, Jette AM. Measuring functional (18) Jette AM. How measurement techniques ability in chronic arthritis: a critical review. influence estimates of disability in older Arthritis Rheum 1981;24:80–86. populations. Soc Sci Med 1994;38:937–942. (7) Deyo RA. Measuring functional outcomes (19) Whiting SE, Lincoln NB. An ADL in therapeutic trials for chronic disease. assessment for stroke patients. Br J Occup Control Clin Trials 1984;5:223–240. Ther 1980;44–46. (8) Zimmer JG, Rothenberg BM, Andresen (20) Lincoln NB, Edmans JA. A re-evaluation of EM. Functional assessment. In: Andresen the Rivermead ADL scale for elderly patients EM, Rothenberg B, Zimmer JG, eds. with stroke. Age Ageing 1990;19:19–24. Assessing the health status of older adults. (21) Podsiadlo D, Richardson S. The Timed New York: Springer, 1997:1–40. “Up & Go”: a test of basic functional (9) Morris JN, Morris SA. ADL measures for mobility for frail elderly persons. J Am use with frail elders. In: Teresi JA, Lawton Geriatr Soc 1991;39:142–148. MP, Holmes D, et al., eds. Measurement in (22) Kuriansky J, Gurland B. The performance elderly chronic care populations. New test of activities of daily living. Int J Aging York: Springer, 1997:130–156. Hum Devel 1976;7:343–352. (10) Caulfield B, Garrett M, Torenbeek M, et al. (23) Guralnik JM, Branch LG, Cummings SR, Rehabilitation outcome measures in Europe et al. Physical performance measures in —the state of the art. Enschede: Roessingh aging research. J Gerontol Research and Development, 1999. 1989;44:M141–M146. (11) Lindeboom R, Vermeulen M, Holman R, et (24) Reuben DB, Siu AL. An objective measure al. Activities of daily living instruments: of physical function of elderly outpatients. optimizing scales for neurologic The Physical Performance Test. J Am assessments. Neurology 2003;60:738–742. Geriatr Soc 1990;38:1105–1112. (12) Gresham GE, Labi MLC. Functional (25) Cress ME, Buchner DM, Questad KA, et assessment instruments currently available al. Continuous-scale physical functional for documenting outcomes in rehabilitation performance in healthy older adults: a medicine. In: Granger CV, Gresham GE, validation study. Arch Phys Med Rehabil eds. Functional assessment in rehabilitation 1996;77:1243–1250. medicine. Baltimore: Williams & Wilkins, (26) Myers AM, Holliday PJ, Harvey KA, et al. 1984:65–85. Functional performance measures: are they (13) Lawton MP, Brody EM. Assessment of superior to self-assessments? J Gerontol older people: self-maintaining and 1993;48:M196–M206. instrumental activities of daily living. (27) Haley SM, Ludlow LH, Gans BM, et al. Gerontologist 1969;9:179–186. Tufts Assessment of Motor Performance: (14) Spector WD, Fleishman JA. Combining an empirical approach to identifying motor activities of daily living with instrumental performance categories. Arch Phys Med activities of daily living to measure Rehabil 1991;72:359–366. functional disability. J Gerontol Soc Sci (28) Gerety MB, Mulrow CD, Tuley MR, et al. 1998;53B:S46–S57. Development and validation of a physical (15) Fillenbaum GG. Screening the elderly: a performance instrument for the brief instrumental activities of daily living functionally impaired elderly: the Physical measure. J Am Geriatr Soc Disability Index (PDI). J Gerontol 1985;33:698–706. 1993;48:M33–M38. (16) Patrick DL, Darby SC, Green S, et al. (29) Judge JO, Schechtman K, Cress E. The Screening for disability in the inner city. J relationship between physical performance Epidemiol Community Health measures and independence in instrumental 1981;35:65–70. activities of daily living. The FICSIT (17) Picavet HS, van den Bos GA. Comparing Group. Frailty and Injury: Cooperative survey data on functional disability: the Studies of Intervention Trials. J Am Geriatr impact of some methodological differences. Soc 1996;44:1332–1341. 62 Measuring Health (30) Holm MB, Rogers JC. Functional dividual to perform routine physical activities assessment: the performance assessment of within the limitations imposed by various physi- self-care skills (PASS). In: Hemphill BJ, ed. cal disorders” (2, p2009). The profile is com- Assessments in occupational therapy monly used to predict rehabilitation potential, mental health: an integrative approach. to evaluate patient progress, and to assist in pro- Thorofare, New Jersey: Slack Publishing, gram planning (3). 1999:117–124. (31) Leidy NK. Psychometric properties of the Functional Performance Inventory in Conceptual Basis patients with chronic obstructive pulmonary As mentioned in Chapter 2, the need to assess the disease. Nurse Res 1999;48:20–28. physical fitness of recruits in World War II led to (32) Fisher AG. The assessment of IADL motor the development of a number of assessment skills: an application of many-faceted scales, mostly known by acronyms that summa- Rasch analysis. Am J Occup Ther rize their content. The PULSES Profile was devel- 1993;47:319–329. oped from the Canadian Army’s 1943 “Physical (33) Haley SM, McHorney CA, Ware JE, Jr. Standards and Instructions” for the medical ex- Evaluation of the MOS SF-36 physical amination of army recruits and soldiers, known functioning scale (PF-10): I. Uni- as the PULHEMS Profile. In this acronym, dimensionality and reproducibility of the P=physique, U=upper extremity, L=lower ex- Rasch item scale. J Clin Epidemiol 1994;47:671–684. tremity, H=hearing and ears, E=eyes and vision, (34) Revicki DA, Cella DF. Health status M=mental capacity, and S=emotional stability, assessment for the twenty-first century: with ratings in each category ranging from nor- item response theory, item banking and mal to totally unfit. The U.S. Army adapted the computer adaptive testing. Qual Life Res PULHEMS system and merged the mental and 1997;6:595–600. emotional categories under the acronym PUL- (35) Velozo CA, Kielhofner G, Lai JS. The use HES. Warren developed a modified version called of Rasch analysis to produce scale-free PULHEEMS to screen for disability in the gen- measurement of functional ability. Am J eral population (4). Moskowitz and McCann Occup Ther 1999;53:83–90. made further modifications to produce the (36) McHorney CA, Cohen AS. Equating health PULSES Profile described here (1). status measures with item response theory: illustrations with functional status items. Med Care 2000;38(suppl II):43–59. Description (37) Bjorner JB, Kreiner S, Ware JE, Jr., et al. The components of the PULSES acronym are: Differential item functioning in the Danish translation of the SF-36. J Clin Epidemiol P = physical condition 1998;51:1189–1202. U = upper limb functions (38) Fleishman JA, Lawrence WF. Demographic L = lower limb functions variation in SF-12 scores: true differences S = sensory components (speech, vision, hear- or differential item functioning? Med Care ing) 2003;41:III-75–III-86. E = excretory functions S = mental and emotional status The Pulses Profile The profile may be completed retrospectively (Eugene Moskowitz and Cairbre B. from medical records, or from interviews and McCann, 1957) observations of the patient (5, p146). In this Purpose vein, Moskowitz saw the profile “as a vehicle for The PULSES Profile was designed to evaluate consolidation of fragments of clinical informa- functional independence in ADLs of chronically tion gathered in a rehabilitation setting by vari- ill and elderly institutionalized populations (1; ous staff members involved in the patient’s daily 2). It expresses “the ability of an aged, infirm in- care” (6, p647).  Physical Disability and Handicap 63 Four levels of impairment were originally corresponded to the disposition of patients: specified for each component (1, Table 1) and those returning home were rated significantly the six scores were presented separately, as a higher than those sent to long-term institutions, profile. Thus, “L-3” describes a person who can who in turn scored significantly higher than walk under supervision and “E-3” indicates fre- those referred for acute care (3; 5; 8). Pearson quent incontinence (6). The original PULSES correlation coefficients between PULSES and Profile was reproduced in the first edition of Barthel scores ranged from −0.74 to −0.80 (the Measuring Health, page 46; the summary sheet negative correlations reflect the inverse scoring is shown in Exhibit 3.1. Moskowitz argued of the scales) (5, pp146–147). against calculating an overall score, which may In the study of 197 stroke patients, the obscure changes in one category that may be nu- PULSES Profile at admission and discharge cor- merically (but not conceptually) balanced by related −0.82 and −0.88 with the Functional In- opposite changes in another. For clinical applica- dependence Measure (FIM); the areas under the tions, Moskowitz later summarized the cate- receiver operating characteristic curve were vir- gories shown in Exhibit 3.1 and presented them tually identical for both instruments in predict- in a color-coded chart, which is reproduced in ing discharge to the community versus long term reference (6). care. In a logistic regression prediction of dis- In 1979, Granger proposed a revised version charge destination, the FIM accounted for no of the PULSES Profile with slight modifications further variance once the PULSES had been in- to the classification levels and an expanded scope cluded in the analysis (7, p763). In the same for three categories. This is now considered the study, a multitrait-multimethod analysis sup- standard version. As shown in Exhibit 3.2, the ported the construct validity of the PULSES. upper limb category was extended to include self-care activities, the lower limb category was Alternative Forms extended to include mobility, and the social and The Incapacity Status Scale is a 16-item disabil- mental category was extended to include emo- ity index based on the PULSES Profile and the tional adaptability, family support, and finances Barthel Index (9). (5, p153). This version provides an overall score, with equal weighting for each category to give Commentary a scale from 6, indicating unimpaired inde- The PULSES scale is the last of the physical im- pendence, to 24, indicating full dependence. pairment scales developed during World War II Granger suggested that a score of 12 distin- that still continues to be used, often in conjunc- guishes lesser from more marked disability and tion with other ADL scales such as the Barthel. It that 16 or above indicates very severe disability is also used occasionally as a criterion scale in (5, p152). validation studies. The PULSES and the Barthel Index both influenced the design of subsequent Reliability scales. Although the PULSES is often compared For the revised version, Granger et al. reported a with the Barthel Index, the two are not strictly test-retest reliability of 0.87 and an inter-rater re- equivalent. Granger et al. noted that the Barthel liability exceeding 0.95, comparable with their re- measures discrete functions (e.g., eating, ambu- sults for the Barthel Index (5, p150). In a sample lation), which may be relevant to clinical staff; of 197 stroke patients, coefficient alpha was 0.74 PULSES cannot do this. However, the PULSES at admission and 0.78 at discharge (7, p762). Profile is broader than the Barthel, tapping com- munication as well as social and mental factors Validity (5, p146). In a study of 307 severely disabled adults in ten As reflected in Table 3.1 in the introduction rehabilitation centers across the United States, to this chapter, the reliability and validity of the the PULSES Profile reflected changes between PULSES have been far less well-tested than those admission and discharge. Scores at discharge of alternative scales such as the Barthel Index.  Exhibit 3.1 Summary Chart from the Original PULSES Profile P U L S E S Physical Condition Upper Extremities Lower Extremities Sensory Function Excretory Social and Mental Functions Status cardiovascular shoulder girdles, pelvis, lower dorsal vision, hearing, bowel and emotional and pulmonary and cervical and upper and lumbosacral speech bladder psychiatric other visceral dorsal spine spine disorders disorders 1 1 1 1 1 1 NORMAL Health maintenance Complete function Complete function Complete function Continent Compatible with age 2 2 2 2 2 2 Occasional medical No assistance Fully ambulatory No appreciable Occasional stress No supervision MILD supervision required despite some loss of functional incontinence or required 64 function impairment nocturia 3 3 3 3 3 3 Frequent medical Some assistance Limited ambulation Appreciable Periodic incontinence Some supervision MODERATELY supervision necessary bilateral loss or or retention necessary SEVERE complete unilateral loss of vision or hearing. Incomplete aphasia 4 4 4 4 4 4 Total care bed or Nursing care Confined to Total blindness Total incontinence Complete care in chair confined wheelchair or bed Total deafness or retention psychiatric facility SEVERE Global aphasia or (including catheter aphonia and colostomy) Adapted from Moskowitz E. PULSES Profile in retrospect. Arch Phys Med Rehabil 1985;66:648. Exhibit 3.2 The PULSES Profile: Revised Version P - Physical condition: Includes diseases of the viscera (cardiovascular, gastrointestinal, urologic, and endocrine) and neurologic disorders: 1. Medical problems sufficiently stable that medical or nursing monitoring is not required more often than 3-month intervals. 2. Medical or nurse monitoring is needed more often than 3-month intervals but not each week. 3. Medical problems are sufficiently unstable as to require regular medical and/or nursing attention at least weekly. 4. Medical problems require intensive medical and/or nursing attention at least daily (excluding personal care assistance only). U - Upper limb functions: Self-care activities (drink/feed, dress upper/lower, brace/prosthesis, groom, wash, perineal care) dependent mainly upon upper limb function: 1. Independent in self-care without impairment of upper limbs. 2. Independent in self-care with some impairment of upper limbs. 3. Dependent upon assistance or supervision in self-care with or without impairment of upper limbs. 4. Dependent totally in self-care with marked impairment of upper limbs. L - Lower limb functions: Mobility (transfer chair/toilet/tub or shower, walk, stairs, wheelchair) dependent mainly upon lower limb function: 1. Independent in mobility without impairment of lower limbs. 2. Independent in mobility with some impairment in lower limbs; such as needing ambulatory aids, a brace or prosthesis, or else fully independent in a wheelchair without significant architectural or environmental barriers. 3. Dependent upon assistance or supervision in mobility with or without impairment of lower limbs, or partly independent in a wheelchair, or there are significant architectural or environmental barriers. 4. Dependent totally in mobility with marked impairment of lower limbs. S - Sensory components: Relating to communication (speech and hearing) and vision: 1. Independent in communication and vision without impairment. 2. Independent in communication and vision with some impairment such as mild dysarthria, mild aphasia, or need for eyeglasses or hearing aid, or needing regular eye medication. 3. Dependent upon assistance, an interpreter, or supervision in communication or vision. 4. Dependent totally in communication or vision. E - Excretory functions: (bladder and bowel): 1. Complete voluntary control of bladder and bowel sphincters. 2. Control of sphincters allows normal social activities despite urgency or need for catheter, appliance, suppositories, etc. Able to care for needs without assistance. 3. Dependent upon assistance in sphincter management or else has accidents occasionally. 4. Frequent wetting or soiling from incontinence of bladder or bowel sphincters. S - Support factors: Consider intellectual and emotional adaptability, support from family unit, and financial ability: 1. Able to fulfill usual roles and perform customary tasks. 2. Must make some modification in usual roles and performance of customary tasks. 3. Dependent upon assistance, supervision, encouragement or assistance from a public or private agency due to any of the above considerations. 4. Dependent upon long-term institutional care (chronic hospitalization, nursing home, etc.) excluding time- limited hospital for specific evaluation, treatment, or active rehabilitation. Reproduced from Granger CV, Albrecht GL, Hamilton BB. Outcome of comprehensive medical rehabilitation: measurement by PULSES Profile and the Barthel Index. Arch Phys Med Rehabil 1979;60:153. With permission. 65 66 Measuring Health References (1). It was intended for patients with conditions causing paralysis and has been used with re- (1) Moskowitz E, McCann CB. Classification habilitation patients to predict length of stay, of disability in the chronically ill and aging. J Chronic Dis 1957;5:342–346. estimate prognosis, and anticipate discharge out- (2) Moskowitz E, Fuhn ER, Peters ME, et al. comes, as well as being used as an evaluative in- Aged infirm residents in a custodial strument. institution: two-year medical and social study. JAMA 1959;169:2009–2012. Conceptual Basis (3) Granger CV, Greer DS. Functional status Items were chosen to indicate the level of nursing measurement and medical rehabilitation care required by a patient. A weighting system outcomes. Arch Phys Med Rehabil for the items reflects their relative importance in 1976;57:103–109. terms of the level of social acceptability and the (4) Warren MD. The use of the PULHEEMS nursing care required (1, p606). Granger has system of medical classification in civilian placed the Barthel Index conceptually within the practice. Br J Ind Med 1956;13:202–209. (5) Granger CV, Albrecht GL, Hamilton BB. World Health Organization (WHO) impair- Outcome of comprehensive medical ment, disability, and handicap framework (2). rehabilitation: measurement by PULSES Profile and the Barthel Index. Arch Phys Description Med Rehabil 1979;60:145–154. The Barthel Index is a rating scale completed by (6) Moskowitz E. PULSES Profile in a health professional from medical records or retrospect. Arch Phys Med Rehabil from direct observation (3). It takes two to five 1985;66:647–648. minutes to complete (4, p62), or it can be self- (7) Marshall SC, Heisel B, Grinnell D. Validity administered in about ten minutes (5, p125). of the PULSES Profile compared with the Two main versions exist: the original ten-item Functional Independence Measure for mea- form and an expanded 15-item version proposed suring disability in a stroke rehabilitation setting. Arch Phys Med Rehabil by Granger (3; 6; 7). The original ten activities 1999;80:760–765. cover personal care and mobility. Each item is (8) Granger CV, Sherwood CC, Greer DS. rated in terms of whether the patient can per- Functional status measures in a form the task independently, with some assis- comprehensive stroke care program. Arch tance, or is dependent on help. The ratings are Phys Med Rehabil 1977;58:555–561. intended to suggest the amount of assistance a (9) La Rocca NG. Analyzing outcomes in the patient needs and the time this will entail (8, care of persons with multiple sclerosis. In: p61). Item scores are added to form an overall Fuhrer MJ, ed. Rehabilitation outcomes: score that ranges from 0 to 100, in steps of five, analysis and measurement. Baltimore: Paul with higher scores indicating greater indepen- H. Brookes, 1987:151–162. dence. The items and scoring system are shown in Exhibit 3.3; rating guidelines are shown in The Barthel Index (Formerly the Exhibit 3.4. The “with help” category is used if any degree of supervision or assistance is re- Maryland Disability Index) quired. Wylie and White also published detailed (Florence I. Mahoney and Dorothea scoring instructions (9, Appendix). W. Barthel; in use since 1955, first Various modifications have been made to the published by originators in 1958) 10-item Barthel scale, including the version Purpose shown in Exhibit 3.5 which was proposed by The Barthel Index measures functional indepen- Collin and Wade in England (4). This reordered dence in personal care and mobility; it was devel- the original items, clarified the rating instruc- oped to monitor performance in long-term tions and modified the scores for each item. To- hospital patients before and after treatment and tal scores range from 0 to 20. This version also to indicate the amount of nursing care needed moves from the capacity orientation of the origi-  Physical Disability and Handicap 67 Exhibit 3.3 The Barthel Index Note: A score of zero is given when the patient cannot meet the defined criterion. With help Independent 1. Feeding (if food needs to be cut up = help) 5 10 2. Moving from wheelchair to bed and return (includes sitting up in bed) 5–10 15 3. Personal toilet (wash face, comb hair, shave, clean teeth) 0 5 4. Getting on and off toilet (handling clothes, wipe, flush) 5 10 5. Bathing self 0 5 6. Walking on level surface (or if unable to walk, propel wheelchair) *score 10 15 only if unable to walk 0* 5* 7. Ascend and descend stairs 5 10 8. Dressing (includes tying shoes, fastening fasteners) 5 10 9. Controlling bowels 5 10 10. Controlling bladder 5 10 Reproduced from Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Maryland State Med J 1965;14:62. With per- mission. nal to a performance rating, indicating what a slight dependency (10, p704). Lazar et al. pro- patient actually does, rather than what she could posed the following interpretation for 15-item do. In another variant, Shah et al. retained the scores: 0–19: dependent; 20–59: self-care as- original items, but rated each on a five-point sisted; 60–79: wheelchair assisted; 80–89: scale to improve sensitivity in detecting change wheelchair independent; 90–99: ambulatory as- (10). At present, little consensus exists over sisted, whereas 100 indicates independence (11, which should be viewed as the definitive version p820). For the 15-item version, Granger et al. if the ten-item scale is chosen. considered a score of 60 or lower as the thresh- Granger extended the Barthel Index to cover old for marked dependence (12). Scores of 40 or 15 topics, in an instrument sometimes called the lower indicate severe dependence, with markedly Modified Barthel Index. Two versions exist: a diminished likelihood of living in the community 1979 variant that includes eating and drinking as (2, p48). Twenty or lower reflects total depend- separate items (3) and a 1981 form that merges ence in self-care and mobility (3, p152). Later eating and drinking and adds an item on dress- studies continue to apply the 60/61 cutting ing after using the toilet (7, Table 1). The latter point, with the recognition that the Barthel In- version is recommended; it uses four-point re- dex should not be used alone for predicting out- sponse scales for most items, with overall scores comes (13, p102; 14, p508). ranging from 0 to 100 (6, Table 12-2; 7, Table 1). This version is outlined in Exhibit 3.6. Vari- Reliability ous guides to scoring are available on the internet ten-item version. Shah et al. reported alpha (e.g., www.neuro.mcg.edu/mcgstrok/Indices/Mod_ internal consistency coefficients of 0.87 to 0.92 Barthel_Ind.htm) but it is not clear how fully (at admission and discharge) for the original these have been tested. scoring system, and 0.90 to 0.93 for her revised Several authors have proposed guidelines for scoring method (10, p706). Wartski and Green interpreting Barthel scores. For the ten- or 15- retested 41 patients after a three-week delay. For item versions that use a 100-point scale, Shah et 35 of the 41, scores fell within 10 points; of the al. suggested that scores of 0–20 indicate total 6 cases with the most discrepant scores, 2 could dependency, 21–60 indicate severe dependency, be explained (15, pp357–358). 61–90 moderate dependency, and 91–99 indicate Collin et al. studied agreement among four Exhibit 3.4 Instructions for Scoring the Barthel Index Note: A score of zero is given when the patient cannot meet the defined criterion. 1. Feeding 10 = Independent. The patient can feed himself a meal from a tray or table when someone puts the food within his reach. He must put on an assistive device if this is needed, cut up the food, use salt and pepper, spread butter, etc. He must accomplish this in a reasonable time. 5 = Some help is necessary (when cutting up food, etc., as listed above). 2. Moving from wheelchair to bed and return 15 = Independent in all phases of this activity. Patient can safely approach the bed in his wheelchair, lock brakes, lift footrests, move safely to bed, lie down, come to a sitting position on the side of the bed, change the position of the wheelchair, if necessary, to transfer back into it safely, and return to the wheelchair. 10 = Either some minimal help is needed in some step of this activity or the patient needs to be reminded or supervised for safety of one or more parts of this activity. 5 = Patient can come to sitting position without the help of a second person but needs to be lifted out of bed, or if he transfers, with a great deal of help. 3. Doing personal toilet 5 = Patient can wash hands and face, comb hair, clean teeth, and shave. He may use any kind of razor but must put in blade or plug in razor without help as well as get it from drawer or cabinet. Female patients must put on own make-up, if used, but need not braid or style hair. 4. Getting on and off toilet 10 = Patient is able to get on and off toilet, fasten and unfasten clothes, prevent soiling of clothes, and use toilet paper without help. He may use a wall bar or other stable object of support if needed. If it is necessary to use a bed pan instead of a toilet, he must be able to place it on a chair, empty it, and clean it. 5 = Patient needs help because of imbalance or in handling clothes or in using toilet paper. 5. Bathing self 5 = Patient may use a bathtub, a shower, or take a complete sponge bath. He must be able to do all the steps involved in whichever method is employed without another person being present. 6. Walking on a level surface 15 = Patient can walk at least 50 yards without help or supervision. He may wear braces or prostheses and use crutches, canes, or a walkerette but not a rolling walker. He must be able to lock and unlock braces if used, assume the standing position and sit down, get the necessary mechanical aids into position for use, and dispose of them when he sits. (Putting on and taking off braces is scored under dressing.) 10 = Patient needs help or supervision in any of the above but can walk at least 50 yards with a little help. 6a. Propelling a wheelchair 5 = Patient cannot ambulate but can propel a wheelchair independently. He must be able to go around corners, turn around, maneuver the chair to a table, bed, toilet, etc. He must be able to push a chair a least 50 yards. Do not score this item if the patient gets score for walking. 7. Ascending and descending stairs 10 = Patient is able to go up and down a flight of stairs safely without help or supervision. He may and should use handrails, canes, or crutches when needed. He must be able to carry canes or crutches as he ascends or descends stairs. 5 = Patient needs help with or supervision of any one of the above items. 8. Dressing and undressing 10 = Patient is able to put on and remove and fasten all clothing, and tie shoe laces (unless it is necessary to use adaptations for this). The activity includes putting on and removing and fastening corset or braces when these are prescribed. Such special clothing as suspenders, loafer shoes, dresses that open down the front may be used when necessary. 5 = Patient needs help in putting on and removing or fastening any clothing. He must do at least half the work himself. He must accomplish this in a reasonable time. Women need not be scored on use of a brassiere or girdle unless these are prescribed garments. 9. Continence of bowels 10 = Patient is able to control his bowels and have no accidents. He can use a suppository or take an enema when necessary (as for spinal cord injury patients who have had bowel training). 5 = Patient needs help in using a suppository or taking an enema or has occasional accidents. 10. Controlling bladder 10 = Patient is able to control his bladder day and night. Spinal cord injury patients who wear an external device and leg bag must put them on independently, clean and empty bag, and stay dry day and night. 5 = Patient has occasional accidents or cannot wait for the bed pan or get to the toilet in time or needs help with an external device. Reproduced form Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Maryland State Med J 1965;14:62–65. With permission. 68 Exhibit 3.5 Collin and Wade Scoring and Guidelines for the 10-Item Modified Barthel Index General The Index should be used as a record of what a patient does, NOT as a record of what a patient could do. The main aim is to establish degree of independence from any help, physical or verbal, however minor and for whatever reason. The need for supervision renders the patient NOT independent. A patient’s performance should be established using the best available evidence. Asking the patient, friends/relatives and nurses will be the usual source, but direct observation and common sense are also important. However, direct testing is not needed. Usually the performance over the preceding 24–48 hours is important, but occasionally longer periods will be relevant. Unconscious patients should score “0” throughout, even if not yet incontinent. Middle categories imply that patient supplies over 50% of the effort. Use of aids to be independent is allowed. Bowels (preceding week) 0 = incontinent (or needs to be given enemata) 1 = occasional accident (once/week) 2 = continent If needs enema from nurse, then ‘incontinent.’ Occasional = once a week. Bladder (preceding week) 0 = incontinent, or catheterized and unable to manage 1 = occasional accident (max. once per 24 hours) 2 = continent (for over 7 days) Occasional = less than once a day. A catheterized patient who can completely manage the catheter alone is registered as ‘continent.’ Grooming (preceding 24–48 hours) 0 = needs help with personal care 1 = independent face/hair/teeth/shaving (implements provided) Refers to personal hygiene: doing teeth, fitting false teeth, doing hair, shaving, washing face. Implements can be provided by helper. Toilet use 0 = dependent 1 = needs some help, but can do something alone 2 = independent (on and off, dressing, wiping). Should be able to reach toilet/commode, undress sufficiently, clean self, dress and leave With help = can wipe self, and do some other of above. Feeding 0 = unable 1 = needs help cutting, spreading butter etc. 2 = independent (food provided in reach). Able to eat any normal food (not only soft food). Food cooked and served by others. But not cut up. Help = food cut up, patient feeds self. Transfer (from bed to chair and back) 0 = unable—no sitting balance 1 = major help (one or two people, physical), can sit 2 = minor help (verbal or physical) 3 = independent Dependent = no sitting balance (unable to sit); two people to lift. Major help = one strong/skilled, or two normal people. Can sit up. Minor help = one person easily, OR needs any supervision for safety. (continued) 69 70 Measuring Health Exhibit 3.5 (continued) Mobility 0 = immobile 1 = wheelchair independent including corners etc. 2 = walks with help of one person (verbal or physical) 3 = independent (but may use any aid, e.g., stick) Refers to mobility about the house or ward, indoors. May use aid. If in wheelchair, must negotiate corners/ doors unaided. Help = by one, untrained person, including supervision/moral support. Dressing 0 = dependent 1 = needs help, but can do about half unaided 2 = independent (including buttons, zips, laces, etc.) Should be able to select and put on all clothes, which may be adapted. Half = help with buttons, zips, etc. (check!), but can put on some garments alone. Stairs 0 = unable 1 = needs help (verbal, physical, carrying aid) 2 = independent up and down Must carry any walking aid used to be independent. Bathing 0 = dependent 1 = independent (or in shower) Usually the most difficult activity. Must get in and out unsupervised, and wash self. Independent in shower = “independent” if unsupervised/unaided. Total (0–20) Adapted from Collin C, Wade DT, Davies S, Horne V. The Barthel ADL Index: a reliability study. Int Disabil Stud 1988;10:63. ways of administering the scale: self-report, as- fifteen-item version. Granger et al. reported a sessment by a nurse based on clinical impres- test-retest reliability of 0.89 with severely dis- sions, testing by a nurse, and testing by a abled adults; inter-rater agreement exceeded physiotherapist (4). Kendall’s coefficient of 0.95 (3, p150). Shinar et al. obtained an inter- concordance among the four rating methods rater agreement of 0.99, and a Cronbach’s alpha was 0.93 (4, p61). This figure somewhat ob- of 0.98 for 18 patients (20, pp724, 726). They scures the extent of disagreement, however. also compared administration by telephone in- There was agreement for 60% of patients and terview and by observation for 72 outpatients. disagreement on one rating for 28%; 12% had Total scores correlated 0.97, and Spearman cor- more than one discrepancy (16, p357). Self- relations exceeded 0.85 for all but one item (20, report accorded least well with the other meth- Table 3). ods; agreement was lowest for items on transfers, feeding, dressing, grooming, and toi- Validity leting. Roy et al. found an inter-rater correla- ten-item version. Wade and Hewer reported tion of 0.99, whereas the correlation between validity information for the revised ten-item ver- ratings and patient self-report was 0.88 (17, sion shown in Exhibit 3.5. Correlations between Table 2). Hachisuka et al. reported agreement 0.73 and 0.77 were obtained with an index of among three versions of the self-report Barthel; motor ability for 976 stroke patients (21, p178). the self-report version correlated 0.99 with the A factor analysis identified two factors, which rating version, and coefficient alpha was 0.84 approximate the mobility and personal care (18; 19). groupings of the items. Wade and Hewer also  Physical Disability and Handicap 71 Exhibit 3.6 Scoring for the 15-Item Modified Barthel Index Independent Dependent I II III IV Intact Limited Helper Null 10 5 0 0 Drink from cup/feed from dish 5 5 3 0 Dress upper body 5 5 2 0 Dress lower body 0 0 −2 Don brace or prosthesis 5 5 0 0 Grooming 4 4 0 0 Wash or bathe 10 10 5 0 Bladder continence 10 10 5 0 Bowel continence 4 4 2 0 Care of perineum/clothing at toilet 15 15 7 0 Transfer, chair 6 5 3 0 Transfer, toilet 1 1 0 0 Transfer, tub or shower 15 15 10 0 Walk on level 50 yards or more 10 10 5 0 Up and down stairs for one flight or more 15 5 0 0 Wheelchair/50 yards—only if not walking Reproduced from Fortinsky RH, Granger CV, Seltzer GB. The use of functional assessment in understanding home care needs. Med Care 1981;19:489, Table 1. With permission. provided evidence for a hierarchical structure in plained 54 to 59% of the variance in EQ scores the scale in terms of the order of recovery of at different stages in the disease course; observed functions (21, Table 4). EQ scores correlated (intraclass correlation= Several studies have assessed predictive valid- 0.70) with those predicted from the Barthel (25, ity. In studies of stroke patients, the percentages Tables 3 and 4). Using the 0 to 20 scoring sys- of those who died within six months of admis- tem, Barthel scores of zero corresponded to a sion fell significantly (p< 0.001) as Barthel scores score of −0.25 on the EQ-5D (i.e., considered a at admission rose (9, p836; 12, p557; 22, p799; state worse than death); Barthel scores of 5 were 23, Table 4). Among survivors, admission scores equivalent to utility scores of zero, whereas also predicted the length of stay and subsequent scores of 20 corresponded to EQ-5D scores of progress as rated by a physician. Seventy-seven 0.75, which is close to population norms (25, percent of those scoring 60 to 100 points at ad- pp429–430). mission were later judged to have improved, compared with 36% of those scoring 0 to 15 fifteen-item version. Fortinsky et al. reported (22, p800; 24, p894). Most discrepancies be- correlations between Barthel scores and actual tween the change scores and the physician’s im- performance of 72 tasks. The overall correlation pression of improvement occurred because of was 0.91; the closest agreement was for personal the omission of speech and mental functioning care tasks (7, p492). Barthel scores also corre- from the index (9, p836). lated with age, psychological problems, and role An interesting cross-walk between the ten- performance (7, p495). item Barthel Index and the EuroQol EQ-5D Correlations between the Barthel and the health utilities index was based on a Dutch study PULSES Profile range from −0.74 to −0.80 (3, of 598 stroke patients (25). Barthel scores ex- p146), to −0.83 (2, p48), and −0.90 (12, p556) 72 Measuring Health (p< 0.001; the negative sign results from the in- place in the development of this field and has verse scoring of the scales). In a study of 45 elder- been incorporated into subsequent evaluation in- ly patients, Barthel scores correlated 0.84 with the struments such as the Long-Range Evaluation Functional Independence Measure (FIM), 0.78 System developed by Granger et al. (6; 30; 33). with the Katz Index of ADL, and 0.52 with More recently, the latter has been superseded by Spitzer’s Quality of Life Index (26, Table 2). the Uniform Data System for Medical Rehabili- Granger et al. found that four items (i.e., tation, which also incorporates the Barthel items bowel and bladder control, grooming, eating) of- along with the FIM (described later in this chap- fered predictions of return to independent com- ter). Validity data on the Barthel are more exten- munity living after six months that were sive than for most other ADL scales, and the comparable to predictions based on the entire results appear superior to others we review. scale (13, p103). Several criticisms have been made of the In a study of recovering stroke patients, the Barthel Index, mainly concerning its scoring ap- Barthel and the motor component of the FIM proach. Indeed, criticisms of the scoring stimu- proved equally responsive to change (27, Tables lated the development of many of newer 4 and 5). versions. Collin and Wade commented on the difficulty of interpreting the middle categories of Alternative Forms the scale and so proposed their more detailed There are many variants of the Barthel Index. guidelines (4). Shah et al. noted the insensitivity Among the less widely used are a 12-item ver- of the original rating scheme and also developed sion by Granger et al. (13; 14), a 14-item version more detailed intermediate categories (10). The (28), a 16-item version (29), and a 17-item ver- Collin and Wade or Shah approaches improve sion (5). Chino et al. proposed a scoring system on the original, but a definitive scoring approach for the 15-item version that is slightly different is still needed. Further development should be from that presented in Exhibit 3.6 (30). Because coordinated and norms by age, sex, and medical of these differing versions, caution is required condition are desirable. The Barthel Index is when comparing results across studies. Coordi- usually applied as a rating scale; self-reports may nation is needed in the further development of give results that differ from therapist ratings (4; the Barthel scale, and we recommend that users 5; 17; 20). The direction of the difference, how- select either the Collin version of the ten-item ever, is not consistent and in most cases is small. scale or the 15-item version proposed by Fortin- Reflecting its origins as a measure for severely sky et al. in 1981 (7). Other variants seem to ill patients, the Barthel Index is narrow in scope hold little advantage. and may not detect low levels of disability. Thus, The Barthel Index is available in numerous although a score of 100 indicates independence languages; an internet search can locate many of in all ten areas, assistance may still be required, these, although there is no coordination in the for example with cooking or house cleaning. development of translations. The Japanese ver- IADL scales address this issue, and scales such as sion has been relatively widely tested (18; 19; the PULSES Profile have achieved broader scope 30). An extended version of the Barthel Index in- of coverage through including topics such as cludes cognitive function (31). A study of the communication, psychosocial, and situational Barthel in Pakistan showed how differences in factors. A range of measurements with broader customs and even architecture meant that re- scope are described in Chapter 10. sponses were not comparable between rural and urban settings (32). References Commentary (1) Mahoney FI, Wood OH, Barthel DW. In use for decades, the Barthel Index continues Rehabilitation of chronically ill patients: to be widely applied and evaluated; it is a re- the influence of complications on the final spected ADL scale. It also occupies an important goal. South Med J 1958;51:605–609.  Physical Disability and Handicap 73 (2) Granger CV. Outcome of comprehensive (14) Granger CV, Hamilton BB, Gresham GE. medical rehabilitation: an analysis based The stroke rehabilitation outcome study— upon the impairment, disability, and Part I: general description. Arch Phys Med handicap model. Int Rehabil Med Rehabil 1988;69:506–509. 1985;7:45–50. (15) Wartski SA, Green DS. Evaluation in a (3) Granger CV, Albrecht GL, Hamilton BB. home-care program. Med Care Outcome of comprehensive medical 1971;9:352–364. rehabilitation: measurement by PULSES (16) Collin C, Davis S, Horne V, et al. Profile and the Barthel Index. Arch Phys Reliability of the Barthel ADL Index. Int J Med Rehabil 1979;60:145–154. Rehabil Res 1987;10:356–357. (4) Collin C, Wade DT, Davies S, et al. The (17) Roy CW, Togneri J, Hay E, et al. An inter- Barthel ADL Index: a reliability study. Int rater reliability study of the Barthel Index. Disabil Stud 1988;10:61–63. Int J Rehabil Res 1988;11:67–70. (5) McGinnis GE, Seward ML, DeJong G, et (18) Hachisuka K, Okazaki T, Ogata H. Self- al. Program evaluation of physical rating Barthel index compatible with the medicine and rehabilitation departments original Barthel index and the Functional using self-report Barthel. Arch Phys Med Independence Measure motor score. J Rehabil 1986;67:123–125. UOEH 1997;19:107–121. (6) Granger CV. Health accounting— (19) Hachisuka K, Ogata H, Ohkuma H, et al. functional assessment of the long-term Test-retest and inter-method reliability of patient. In: Kottke FJ, Stillwell GK, the self-rating Barthel Index. Clin Rehabil Lehmann JF, eds. Krusen’s handbook of 1997;11:28–35. physical medicine and rehabilitation. 3rd (20) Shinar D, Gross CR, Bronstein KS, et al. ed. Philadelphia: WB Saunders, Reliability of the activities of daily living 1982:253–274. scale and its use in telephone interview. (7) Fortinsky RH, Granger CV, Seltzer GB. Arch Phys Med Rehabil 1987;68:723–728. The use of functional assessment in (21) Wade DT, Hewer RL. Functional abilities understanding home care needs. Med Care after stroke: measurement, natural history 1981;19:489–497. and prognosis. J Neurol Neurosurg (8) Mahoney FI, Barthel DW. Functional Psychiatry 1987;50:177–182. evaluation: the Barthel Index. Md State (22) Wylie CM. Gauging the response of stroke Med J 1965;14:61–65. patients to rehabilitation. J Am Geriatr Soc (9) Wylie CM, White BK. A measure of 1967;15:797–805. disability. Arch Environ Health (23) Granger CV, Greer DS, Liset E, et al. Mea- 1964;8:834–839. surement of outcomes of care for stroke (10) Shah S, Vanclay F, Cooper B. Improving patients. Stroke 1975;6:34–41. the sensitivity of the Barthel Index for (24) Wylie CM. Measuring end results of stroke rehabilitation. J Clin Epidemiol rehabilitation of patients with stroke. 1989;42:703–709. Public Health Rep 1967;82:893–898. (11) Lazar RB, Yarkony GM, Ortolano D, et al. (25) van Exel NJA, Scholte op Reimer WJM, Prediction of functional outcome by motor Koopmanschap MA. Assessment of post- capability after spinal cord injury. Arch stroke quality of life in cost-effectiveness Phys Med Rehabil 1989;70:819–822. studies: the usefulness of the Barthel Index (12) Granger CV, Sherwood CC, Greer DS. and the EuroQol-5D. Qual Life Res Functional status measures in a 2004;13:427–433. comprehensive stroke care program. Arch (26) Rockwood K, Stolee P, Fox RA. Use of goal Phys Med Rehabil 1977;58:555–561. attainment scaling in measuring clinically (13) Granger CV, Hamilton BB, Gresham GE, important change in the frail elderly. J Clin et al. The stroke rehabilitation outcome Epidemiol 1993;46:1113–1118. study: Part II. Relative merits of the total (27) Wallace D, Duncan PW, Lai SM. Barthel Index score and a four-item Comparison of the responsiveness of the subscore in predicting patient outcomes. Barthel Index and the motor component of Arch Phys Med Rehabil 1989;70:100–103. the Functional Independence Measure in 74 Measuring Health stroke: the impact of using different ticular order, the most complex functions being methods for measuring responsiveness. J lost first. Empirically, the six activities included Clin Epidemiol 2002;55:922–928. in the index were found to lie in a hierarchical (28) Yarkony GM, Roth EJ, Heinemann AW, et order of this type whereas other items (e.g., mo- al. Functional skills after spinal cord injury bility, walking, stair climbing) did not fit the rehabilitation: three-year longitudinal pattern and were excluded (4). Katz et al. fur- follow-up. Arch Phys Med Rehabil ther suggested that, during rehabilitation, skills 1988;69:111–114. (29) Nosek MA, Parker RM, Larsen S. are regained in order of ascending complexity, Psychosocial independence and functional in the same order that they are initially acquired abilities: their relationship in adults with by infants (1, pp917–918). They concluded that severe musculoskeletal impairments. Arch the Index of ADL appears to be reflect “pri- Phys Med Rehabil 1987;68:840–845. mary biological and psychosocial function” (1; (30) Chino N, Anderson TP, Granger CV. 4–6). Stroke rehabilitation outcome studies: comparison of a Japanese facility with 17 Description U.S. facilities. Int Disabil Stud The Index of ADL was originally developed for 1988;10:150–154. elderly and chronically ill patients who had suf- (31) Jansa J, Pogacnik T, Gompertz P. An fered strokes or fractured hips. It assesses inde- evaluation of the extended Barthel Index with acute ischemic stroke patients. pendence in six activities: bathing, dressing, Neurorehabil Neural Repair using the toilet, transferring from bed to chair, 2004;18:37–41. continence, and feeding. Through observation (32) Ali SM, Mulley GP. Is the Barthel Scale and interview, the therapist rates each activity on appropriate in non-industrialized a three-point scale of independence, shown in countries? A view of rural Pakistan. Disabil Exhibit 3.7. The most dependent degree of per- Rehabil 1998;20:195–199. formance during a two-week period is recorded. (33) Granger CV, McNamara MA. Functional In applying the index: assessment utilization: the Long-Range Evaluation System (LRES). In: Granger CV, The observer asks the subject to show him Gresham GE, eds. Functional assessment in (1) the bathroom, and (2) medications in an- rehabilitation medicine. Baltimore: Williams & Wilkins, 1984:99–121. other room (or a meaningful substitute ob- ject). These requests create test situations for direct observation of transfer, locomotion, The Index of Independence and communication and serve as checks on in Activities of Daily Living, the reliability of information about bathing, or Index of ADL dressing, going to toilet, and transfer. (1, p915). (Sidney Katz, 1959, revised 1976) Purpose Full definitions of the six items are given by Katz The Index of ADL was developed to measure the et al. (5, pp22–24). physical functioning of elderly and chronically The first stage in scoring involves translating ill patients. Frequently, it has been used to indi- the three-point scales into a dependent or inde- cate the severity of chronic illness and to evaluate pendent dichotomy, using the guidelines shown the effectiveness of treatment; it has also been in the lower half of Exhibit 3.8. The middle cat- used to provide predictive information on the egories in Exhibit 3.7 are rated as “independent” course of specific illnesses (1–3). for bathing, dressing, and feeding, but as “de- pendent” for the others. The patient’s overall Conceptual Basis performance is then summarized on an eight- In empirical studies of aging, Katz et al. noted point scale that considers the numbers of areas that the loss of functional skills occurs in a par- of dependency and their relative importance Exhibit 3.7 The Index of Independence in Activities of Daily Living: Evaluation Form For each area of functioning listed below, check description that applies. (The word “assistance” means supervision, direction, or personal assistance.) Bathing—either sponge bath, tub bath, or shower Receives no assistance (gets in and out Receives assistance in bathing only one Receives assistance in of tub by self if tub is usual means of part of the body (such as back or a leg) bathing more than one part bathing) of the body (or not bathed) Dressing—gets clothes from closets and drawers—including underclothes, outer garments and using fasteners (including braces if worn) Get clothes and gets completely Gets clothes and gets dressed without Receives assistance in getting dressed without assistance assistance except for assistance in tying clothes or in getting dressed, shoes or stays partly or completely undressed Toileting—going to the “toilet room” for bowel and urine elimination; cleaning self after elimination, and arranging clothes Goes to “toilet room,” cleans self, Receives assistance in going to “toilet Doesn’t go to room termed and arranges clothes without room” or in cleansing self or in “toilet” for the elimination assistance (may use object for support arranging clothes after elimination or process such as cane, walker, or wheelchair in use of night bedpan or commode and may manage night bedpan or commode, emptying same in morning) Transfer— Moves in and out of bed as well as in Moves in and out of bed or chair with Doesn’t get out of bed and out of chair without assistance assistance (may be using object for support such as cane or walker) Continence— Controls urination and bowel Has occasional “accidents” Supervision helps keep urine movement completely by self or bowel control; catheter is used, or is incontinent Feeding— Feeds self without assistance Feeds self except for getting assistance Receives assistance in feeding in cutting meat or buttering bread or is fed partly or completely by using tubes or intravenous fluids Reproduced from Katz S, Downs TD, Cash HR, Grotz RC. Progress in development of the Index of ADL. Gerontologist 1970;10:21. Copyright © the Gerontological Society of America. Reproduced by permission of the publisher. 75 76 Measuring Health Exhibit 3.8 The Index of Independence in Activities of Daily Living: Scoring and Definitions The Index of Independence in Activities of Daily Living is based on an evaluation of the functional independence or dependence of patients in bathing, dressing, going to toilet, transferring, continence, and feeding. Specific definitions of functional independence and dependence appear below the index. A—Independent in feeding, continence, transferring, going to toilet, dressing and bathing. B—Independent in all but one of these functions. C—Independent in all but bathing and one additional function. D—Independent in all but bathing, dressing, and one additional function. E—Independent in all but bathing, dressing, going to toilet, and one additional function. F—Independent in all but bathing, dressing, going to toilet, transferring, and one additional function. G—Dependent in all six functions. Other—Dependent in at least two functions, but not classifiable as C, D, E or F. Independence means without supervision, direction, or active personal assistance, except as specifically noted below. This is based on actual status and not on ability. A patient who refuses to perform a function is considered as not performing the function, even though he is deemed able. Bathing (sponge, shower or tub) Independent: assistance only in bathing a single part (as back or disabled extremity) or bathes self completely Dependent: assistance in bathing more than one part of body; assistance in getting in or out of tub or does not bathe self Dressing Independent: gets clothes from closets and drawers; puts on clothes, outer garments, braces; manages fasteners; act of tying shoes is excluded Dependent: does not dress self or remains partly undressed Going to toilet Independent: gets to toilet; gets on and off toilet; arranges clothes; cleans organs of excretion; (may manage own bedpan used at night only and may or may not be using mechanical supports) Dependent: uses bedpan or commode or receives assistance in getting to and using toilet Transfer Independent: moves in and out of bed independently and moves in and out of chair independently (may or may not be using mechanical supports) Dependent: assistance in moving in or out of bed and/or chair; does not perform one or more transfers Continence Independent: urination and defecation entirely self-controlled Dependent: partial or total incontinence in urination or defecation; partial or total control by enemas, catheters, or regulated use of urinals and/or bedpans Feeding Independent: gets food from plate or its equivalent into mouth; (precutting of meat and preparation of food, as buttering bread, are excluded from evaluation) Dependent: assistance in act of feeding (see above): does not eat at all or parenteral feeding Reproduced from Katz S, Downs TD, Cash HR, Grotz RC. Progress in development of the Index of ADL. Gerontologist 1970;10:23. Copyright © The Gerontological Society of America. Reproduced by permission of the publisher. (shown in the upper half of Exhibit 3.8). Alter- Reliability natively, a simplified scoring system counts the Remarkably little formal reliability (or validity) number of activities in which the individual is testing has been reported. Katz et al. assessed dependent, on a scale from 0 through 6, where 0 inter-rater reliability, reporting that differences = independent in all six functions and 6 = de- between observers occurred once in 20 evalua- pendent in all functions (4, p497). This method tions or less frequently (1, p915). Guttman removes the need for the miscellaneous scoring analyses on 100 patients in Sweden yielded coef- category, “other.” ficients of scalability ranging from 0.74 to 0.88,  Physical Disability and Handicap 77 suggesting that the index forms a successful cu- Commentary mulative scale (7, p128). The Index of ADL has been very widely used: with children and with adults, with people with Validity mental retardation and people with disabilities, in Katz et al. applied the Index of ADL and other the community and in institutions (4; 6). It has measures to 270 patients at discharge from a been used in studies of many conditions, includ- hospital for the chronically ill. ADL scores were ing cerebral palsy, strokes, multiple sclerosis, found to correlate 0.50 with a mobility scale and paraplegia, quadriplegia, and rheumatoid arthritis 0.39 with a house confinement scale (5, Table 3). (2–4; 10–14). As with all ADL scales, the Katz In- At a two-year follow-up, Katz concluded that dex is only appropriate with severely sick respon- the Index of ADL predicted long-term outcomes dents; minor illness or disability frequently does as well as or better than selected measures of not translate into the limitations in basic ADLs physical or mental function (5, p29). Other covered in this scale. It is therefore unlikely to be studies of predictive validity are summarized by suitable for health surveys or in general practice. Katz and Akpom (4); typical of these findings Katz’s Index of ADL rose to prominence are the results presented by Brorsson and Ås- largely because it was the first such scale pub- berg. Thirty-two of 44 patients rated as indepen- lished. Illustrations exist in several areas of dent at admission to hospital were living at health measurement of acceptance of certain home one year later whereas eight had died. By scales by acclaim rather than following clear contrast, 23 of 42 patients initially rated as de- demonstration of validity and reliability; indeed, pendent had died and only eight were living in it is surprising that so little evidence has been their homes (7, p130). published on its reliability and validity. Consid- Åsberg examined the ability of the scale to erably more evidence has been accumulated, for predict length of hospital stay, likelihood of dis- example, on the Barthel Index. The work of charge home, and death (N=129). In predicting Brorsson and Åsberg partly filled this need, al- mortality, sensitivity was 73% and specificity though more evidence for validity and reliability 80%; in predicting discharge, sensitivity was is needed before it can be fully recommended. 90%, and specificity, 63%. Similar predictive Among the various critiques of the scale, poten- validity was obtained from ratings made by inde- tial users should be aware of the criticisms of the pendent physicians (8, Table IV). scoring system made by Chen and Bryant (15, Like all other ADL scales, the Index of ADL p261). Other scales should be reviewed closely suffers a floor effect whereby it is insensitive to before the Katz Index is selected. variations in low levels of disability. This has been reported many times; one example may suf- fice. Compared with the Functional Status Ques- References tionnaire (FSQ) in a study of 89 polio survivors, (1) Katz S, Ford AB, Moskowitz RW, et al. the Index of ADL rated 32 patients fully inde- Studies of illness in the aged. The Index of pendent, six partly dependent, and one depend- ADL: a standardized measure of biological ent. Using the instrumental ADL questions from and psychosocial function. JAMA the FSQ, only four of the same patients had no 1963;185:914–919. difficulty with walking several blocks, six had no (2) Katz S, Ford AB, Chinn AB, et al. Prognosis after strokes: II. Long-term difficulty with light housework, and only one course of 159 patients with stroke. patient had no difficulty with more vigorous ac- Medicine 1966;45:236–246. tivities (9, Table II). Even more indicative of the (3) Katz S, Heiple KG, Downs TD, et al. Long- limited sensitivity of the ADL questions, 15 of term course of 147 patients with fracture the patients had difficulty or required assistance of the hip. Surg Gynecol Obstet to stand up, and seven were unable to stand 1967;124:1219–1230. alone; ten could not go outdoors (9, Table IX). (4) Katz S, Akpom CA. A measure of primary 78 Measuring Health sociobiological functions. Int J Health Serv estimate a patient’s ability to live independently 1976;6:493–507. at home or in a protected environment. Intended (5) Katz S, Downs TD, Cash HR, et al. for use in setting treatment goals and evaluating Progress in development of the Index of progress, the method is limited to physical activ- ADL. Gerontologist 1970;10:20–30. ities and was designed to offer a “more precise (6) Katz S, Akpom CA. Index of ADL. Med measuring device than the traditional ADL Care 1976;14:116–118. form” (1, p690). (7) Brorsson B, Åsberg KH. Katz Index of In- dependence in ADL: reliability and validity in short-term care. Scand J Rehabil Med Conceptual Basis 1984;16:125–132. The topics included in the Kenny were selected to (8) Åsberg KH. Disability as a predictor of represent the minimum requirements for indepen- outcome for the elderly in a department of dent living (2, p2). The rating system considers all internal medicine. Scand J Soc Med of these self-care abilities to be equally important 1987;15:261–265. and assigns equal weight to them (3, p222). (9) Einarsson G, Grimby G. Disability and handicap in late poliomyelitis. Scand J Description Rehabil Med 1990;22:1–9. The Kenny evaluation is hierarchical. The re- (10) Steinberg FU, Frost M. Rehabilitation of vised version covers seven aspects of mobility geriatric patients in a general hospital: a follow–up study of 43 cases. Geriatrics and self-care: moving in bed, transfers, locomo- 1963;18:158–164. tion, dressing, personal hygiene, bowel and blad- (11) Katz S, Vignos PJ, Moskowitz RW, et al. der, and feeding. Within each category there are Comprehensive outpatient care in between one and four general activities, each of rheumatoid arthritis: a controlled study. which is in turn divided into component tasks. JAMA 1968;206:1249–1254. These comprise the steps involved in performing (12) Grotz RT, Henderson ND, Katz S. A the activity, for example, “legs over side of bed” comparison of the functional and is one of the steps in “rising and sitting.” In all, intellectual performance of phenylketonuric, there are 17 activities and 85 tasks (see Exhibit anoxic, and Down’s Syndrome individuals. 3.9). The questionnaire and a 24-page user’s Am J Ment Defic 1972;76:710–717. manual have been produced by the Publications (13) Katz S, Ford AB, Downs TD, et al. Effects of continued care: a study of chronic Office of the Sister Kenny Institute (2). illness in the home. (DHEW Publication Clinical staff observe the performance of No. (HSM) 73–3010) Washington, DC: US each task and rate it on a three-point scale: “to- Government Printing Office, 1972. tally independent,” “requiring assistance or su- (14) Katz S, Hedrick S, Henderson NS. The pervision” (regardless of the amount), or measurement of long-term care needs and “totally dependent.” Every task must be ob- impact. Health Med Care Serv Rev served; self-report is not accepted. If the rater be- 1979;2:1–21. lieves that the performance did not reflect the (15) Chen MK, Bryant BE. The measurement of patient’s true ability, special circumstances that health—a critical and selective overview. may have affected the score (e.g., an acute ill- Int J Epidemiol 1975;4:257–264. ness) can be noted in the “progress rounds” space on the score sheet. Rather than calculating a total score, the rat- The Kenny Self-Care Evaluation ings for the tasks within each activity are com- (Herbert A. Schoening and Staff of bined as follows: the Sister Kenny Institute, 1965, Revised 1973) Four: All tasks rated independent. Purpose Three: One or two tasks required assistance The Kenny Self-Care Evaluation is a clinical rat- or supervision; all others are done indepen- ing scale that records functional performance to dently. Exhibit 3.9 The Sister Kenny Institute Self-Care Evaluation Activities Tasks Evaluation Date: Progress Rounds: Progress Rounds: BED ACTIVITIES Shift position Turn to left side Moving Turn to right side in Bed Turn to prone Turn to supine Come to sitting position Maintain sitting balance Rising and Legs over side of bed Sitting Move to edge of bed Legs back onto bed TRANSFERS Position wheelchair Brakes on/off Arm rests on/off Sitting Foot rests on/off Transfer Position legs Position sliding board Maintain balance Shift to bed/chair Position wheelchair Brakes on/off Move feet and pedals Slide forward Standing Transfer Position feet Stand Pivot Sit (continued) 79 Exhibit 3.9 (continued) Activities Tasks Evaluation Date: Progress Rounds: Progress Rounds: Position equipment Manage equipment Toilet Manage undressing Transfer Transfer to commode/toilet Manage dressing Transfer back Tub/shower approach Use of grab bars Bathing Transfer Tub/shower entry Tub/shower exit LOCOMOTION Walking Locomotion Stairs Wheelchair DRESSING Hearing aid and eyeglasses Front opening on/off Upper Trunk Pullover on/off and Arms Brassiere on/off Corset/brace on/off Equipment/prostheses on/off Sweater/shawl on/off Slack/skirt on/off Underclothing on/off Lower Trunk Belt on/off and Legs Braces/prostheses on/off Girdle/garter belt on/off 80 Exhibit 3.9 Activities Tasks Evaluation Date: Progress Rounds: Progress Rounds: Stockings on/off Shoes/slippers on/off Feet Braces/prostheses on/off Wraps/support hose on/off PERSONAL HYGIENE Wash face Wash hands and arms Face, Hair, and Brush teeth/dentures Arms Brush/comb hair Shaving/make-up Wash back Trunk Wash buttocks and Wash chest Perineum Wash abdomen Wash groin Wash upper legs Lower Wash lower legs Extremities Wash feet BOWEL AND BLADDER Suppository insertion Bowel Digital stimulation Program Equipment care Cleansing self Manage equipment Bladder Stimulation Program Cleansing self (continued) 81 82 Measuring Health Exhibit 3.9 (continued) Activities Tasks Evaluation Date: Progress Rounds: Progress Rounds: Assemble equipment Fill syringe Catheter Inject liquid Care Connect/disconnect Sterile technique FEEDING* Adaptive equipment Finger feeding Feeding Use of utensils Pour from container Drink (cup/glass/straw) *If patient cannot swallow, he is to be scored 0 in feeding Two: Other configurations not covered in Reliability classes 4, 3, 1, or 0. The inter-rater agreement among 43 raters for One: One or two tasks required assistance or the Kenny total score was 0.67 or 0.74, accord- supervision, or one was carried out indepen- ing to whether it was applied before or after an- dently, but in all others the patient was de- other rating scale. The reliability of the pendent. locomotion score (0.46 or 0.42) was markedly Zero: All tasks were rated dependent. lower than that of the other scores, which (adapted from 2, p7) ranged from 0.71 to 0.94 (4, Table 2). Iversen et al. commented that the locomotion category is These 0 to 4 scales are entered under “Activity the most difficult to score (2, p14). Gordon et al. Scores” on the scoring sheet at the end of the ex- achieved higher inter-rater reliabilities: errors oc- hibit. Category scores are the average of the ac- curred in 2.5% of ratings (5, p400). tivity scores within a category (as shown under “Category Score”). The category scores may be Validity summed to provide a total score in which the Gresham et al. compared Kenny and Barthel In- seven categories receive equal weights. Equal dex ratings of stroke patients, giving a kappa co- weights were justified on the basis of empirical efficient of 0.42 and a Spearman correlation of observations suggesting that roughly equal nurs- 0.73 (p< 0.001) (6, Table 3). They found that the ing time was required for helping the dependent Kenny form tends to rate slightly more patients patient with each group of activities (1, as independent than other measures. Complete pp690–693). No guidelines are given on how to independence was designated in 35.1% of 148 interpret the scores. stroke patients by the Barthel Index, in 39.2%  Exhibit 3.9 (continued) Self-Care Score Activity Category Category Activity Category Category Activity Category Category Category Activities Scores Total Score Scores Total Score Score Total Score BED Moving in Bed ACTIVITIES Rising and Sitting ÷2 = . ÷2 = . ÷2 = . Sitting Transfer Standing Transfer TRANSFERS Toilet Transfer Bathing Transfer ÷4 = . ÷4 = . ÷4 = . Walking LOCOMOTION Stairs Wheelchair ÷3 = . ÷3 = . ÷3 = . 83 Upper Trunk and Arms DRESSING Lower Trunk and Legs Feet ÷3 = . ÷3 = . ÷3 = . Face, Hair and Arms PERSONAL Trunk and Perineum HYGIENE Lower Extremities ÷3 = . ÷3 = . ÷3 = . Bowel Program BOWEL AND Bladder Program BLADDER Catheter Care ÷2 = . ÷2 = . ÷2 = . FEEDING Feeding = .0 = .0 = .0 TOTAL SELF-CARE SCORE . . . Reproduced from Iversen IA, Silberberg NE, Stever RC, Schoening HA. The revised Kenny Self-Care Evaluation: a numerical measure of independence in activities of daily living. Minneapolis, Minnesota: Sister Kenny Institute, 1973. With permission. 84 Measuring Health by the Katz Index of ADL, and in 41.9% using specific ratings. Arch Phys Med Rehabil the Kenny instrument (differences not statisti- 1981;62:161–166. cally significant) (6, p355). (5) Gordon EE, Drenth V, Jarvis L, et al. Neurophysiologic syndromes in stroke as predictors of outcome. Arch Phys Med Commentary Rehabil 1978;59:399–409. Although its scope is limited, the Kenny Self- (6) Gresham GE, Phillips TF, Labi MLC. ADL Care Evaluation is distinctive in its detailed cov- status in stroke: relative merits of three erage and its requirement of direct observation standard indexes. Arch Phys Med Rehabil of the patient. In addition to this self-care scale, 1980;61:355–358. the Kenny Institute developed separate rating (7) Ellwood PM Jr. Quantitative measurement scales for behavior and for speech (7, p60). of patient care quality. Part 2—A system Available evidence suggests good inter-rater for identifying meaningful factors. agreement for the Kenny; as it breaks activities Hospitals 1966;40:59–63. down into their component parts, raters achieved high agreement because the narrower scope of The Physical Self-Maintenance Scale each evaluations reduced the number of behav- (M. Powell Lawton and Elaine M. ioral components that could be subjectively Brody, 1969) weighted (4, pp164–165). However, comparisons of the Kenny with sim- Purpose pler scales suggest that the additional detail may Lawton and Brody developed the Physical Self- not provide superior discriminative ability (4, Maintenance Scale (PSMS) as a disability mea- p164). The correlation of 0.73 with the Barthel sure for use in planning and evaluating treatment Index is high, and if this were corrected for atten- for elderly people living in the community or in uation due to the imperfect reliability of the two institutions. scales, it would imply that the simpler Barthel In- dex provides results that are virtually identical in Conceptual Basis statistical terms. Its detailed ratings may, however, This scale is based on the theory that human be- be advantageous for clinical applications. havior can be ordered in a hierarchy of complex- ity, an approach similar to that used by Katz for Address the Index of ADL. The hierarchy runs from Sister Kenny Institute, 800 East 28th Street at physical health through self-maintenance ADL Chicago Avenue, Minneapolis, MN USA 55407 and IADL, to cognition, time use (e.g., participa- tion in hobbies or community activities), and fi- nally to social interaction (1; 2). Within each References category a further hierarchy of complexity runs (1) Schoening HA, Anderegg L, Bergstrom D, from basic to complex activities (2, Figure 1; 3). et al. Numerical scoring of self-care status of patients. Arch Phys Med Rehabil Description 1965;46:689–697. The PSMS is a modification of a scale developed (2) Iversen IA, Silberberg NE, Stever RC, et al. at the Langley-Porter Neuropsychiatric Institute The revised Kenny Self-Care Evaluation: a by Lowenthal et al., which is discussed, but not numerical measure of independence in presented, in Lowenthal’s book (4). In turn, activities of daily living. Minneapolis. items from the PSMS have been incorporated Sister Kenny Institute, 1983 (reprint). into subsequent instruments. Brody and Lawton (3) Schoening HA, Iversen IA. Numerical scoring of self-care status: a study of the developed two scales: the PSMS, which includes Kenny Self-Care Evaluation. Arch Phys six ADL items (Exhibit 3.10), and an eight-item Med Rehabil 1968;49:221–229. IADL scale (Exhibit 3.11). They can be adminis- (4) Kerner JF, Alexander J. Activities of daily tered separately or together. Both are designed living: reliability and validity of gross vs for people over 60 years of age (1; 5, Appendix  Physical Disability and Handicap 85 Exhibit 3.10 The Physical Self-Maintenance Scale Circle one statement in each category A–F that applies to subject. A. Toilet 1. Cares for self at toilet completely, no incontinence. 2. Needs to be reminded, or needs help in cleaning self, or has rare (weekly at most) accidents. 3. Soiling or wetting while asleep more than once a week. 4. Soiling or wetting while awake more than once a week. 5. No control of bowels or bladder. B. Feeding 1. Eats without assistance. 2. Eats with minor assistance at meal times and/or with special preparation of food, or help in cleaning up after meals. 3. Feeds self with moderate assistance and is untidy. 4. Requires extensive assistance for all meals. 5. Does not feed self at all and resists efforts of others to feed him. C. Dressing 1. Dresses, undresses and selects clothes from own wardrobe. 2. Dresses and undresses self, with minor assistance. 3. Needs moderate assistance in dressing or selection of clothes. 4. Needs major assistance in dressing, but cooperates with efforts of others to help. 5. Completely unable to dress self and resists efforts of others to help. D. Grooming (neatness, hair, nails, hands, face, clothing) 1. Always neatly dressed, well-groomed, without assistance. 2. Grooms self adequately with occasional minor assistance, e.g., shaving. 3. Needs moderate and regular assistance or supervision in grooming. 4. Needs total grooming care, but can remain well-groomed after help from others. 5. Actively negates all efforts of others to maintain grooming. E. Physical ambulation 1. Goes about grounds or city. 2. Ambulates within residence or about one block distant. 3. Ambulates with assistance of (check one) a ( ) another person, b ( ) railing, c ( ) cane, d ( ) walker, e ( ) wheelchair 1 _____ Gets in and out without help. 2 _____ Needs help in getting in and out. 4. Sits unsupported in chair or wheelchair, but cannot propel self without help. 5. Bedridden more than half the time. F. Bathing 1. Bathes self (tub, shower, sponge bath) without help. 2. Bathes self with help in getting in and out of tub. 3. Washes face and hands only, but cannot bathe rest of body. 4. Does not wash self but is cooperative with those who bathe him. 5. Does not try to wash self and resists efforts to keep him clean. Reproduced from Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:180, Table 1. Copyright © The Gerontological Society of America. Reproduced by permission of the pub- lisher. B). They were originally developed as rating publication (1). The self-administered version of scales, but self-administered versions have been the PSMS uses simpler response scales of two, proposed; items focus on observable behaviors. three, or four categories (see 2, pp796–797). In There is some variation in the response cate- the rating version shown here, the ADL items use gories shown in Lawton’s various reports; the five-point rating scales ranging from total inde- ones in the Exhibits are taken from the original pendence to total dependence, whereas the IADL 86 Measuring Health Exhibit 3.11 The Lawton and Brody IADL Scale Circle one statement in each category A-H that applies to subject A. Ability to use telephone 1. Operates telephone on own initiative–looks up and dials numbers, etc. 2. Dials a few well-known numbers. 3. Answers telephone but does not dial. 4. Does not use telephone at all. B. Shopping 1. Takes care of all shopping needs independently. 2. Shops independently for small purchases. 3. Needs to be accompanied on any shopping trip. 4. Completely unable to shop. C. Food preparation 1. Plans, prepares and serves adequate meals independently. 2. Prepares adequate meals if supplied with ingredients. 3. Heats and serves prepared meals, or prepares meals but does not maintain adequate diet. 4. Needs to have meals prepared and served. D. Housekeeping 1. Maintains house alone or with occasional assistance (e.g., “heavy work-domestic help”). 2. Performs light daily tasks such as dish-washing, bed-making. 3. Performs light daily tasks but cannot maintain acceptable level of cleanliness. 4. Needs help with all home maintenance tasks. 5. Does not participate in any housekeeping tasks. E. Laundry 1. Does personal laundry completely. 2. Launders small items–rinses socks, stockings, etc. 3. All laundry must be done by others. F. Mode of transportation 1. Travels independently on public transportation or drives own car. 2. Arranges own travel via taxi, but does not otherwise use public transportation. 3. Travels on public transportation when assisted or accompanied by another. 4. Travel limited to a taxi or automobile with assistance of another. 5. Does not travel at all. G. Responsibility for own medications 1. Is responsible for taking medications in correct dosages at correct time. 2. Takes responsibility if medication is prepared in advance in separate dosages. 3. Is not capable of dispensing own medication. H. Ability to handle finances 1. Manages financial matters independently (budgets, writes checks, pays rent, bills, goes to bank), collects and keeps track of income. 2. Manages day-to-day purchases but needs help with banking, major purchases, etc. 3. Incapable of handling money. Reproduced from Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969; 9:181, Table 2. Copyright © The Gerontological Society of America. Reproduced by permission of the publisher. scale uses three- to five-point response scales. recorded for any answer other than the first in Both scales can be scored either by counting the each question. For the IADL items, the last an- number of items on which a specified degree of swer category in each question indicates a dis- disability is identified, or by summing the re- ability, with the following exceptions: for sponse codes for each item. In the first scoring questions B and C, answers 2 through 4 indicate approach, for the ADL items a disability is disability; for question F, answers 4 and 5 indi-  Physical Disability and Handicap 87 cate disability, and for question G, answers 2 lated 0.38 with a behavioral rating of social ad- and 3 indicate disability (1, Table 2). The scaled justment (1, Table 6). PSMS scores correlated scoring option produces an overall severity score 0.43 with an estimate of the time required for ranging from 6 to 30 for the PSMS and 8 to 31 caregivers to assist Alzheimer’s patients with for the IADL scale. Green et al. proposed a for- their daily activities (9). mula for transforming IADL scores to compen- The PSMS has been used quite frequently in sate for missing ratings (typically men who were clinical trials for treatment for Alzheimer’s dis- not rated on housework items) (6, p655). ease. The ADL items correlated 0.78 with scores Lawton and Brody’s original article focused on the Blessed test for a sample of Alzheimer’s on the PSMS and reported little information on patients; the correlation for the IADL items was the validity and reliability of the IADL items. 0.83 (6, p656). Sensitivity to change appears lower than that for the Mini-Mental State Exam Reliability (i.e., treatment altered cognition but did not af- The six PSMS items fell on a Guttman scale fect function) (10). Rockwood et al. found the when cutting-points were set between indepen- PSMS ADL questions to be less responsive (stan- dent (code 1 in each item) and all levels of de- dardized response mean 0.10) than the Barthel pendency. The order of the items was feeding Index (SRM 1.13) in evaluating the impact of a (77% independent), toilet (66%), dressing comprehensive geriatric intervention program. (56%), bathing (43%), grooming (42%), and The IADL questions (SRM 0.23) were slightly ambulation (27% independent). A Guttman re- better than the ADL but still not useful in detect- producibility coefficient of 0.96 was reported ing change (11). (N = 265) (1, Table 1). The IADL items formed a Guttman scale for women but not men, owing to Alternative Forms gender bias in the housekeeping, cooking, and The PSMS has been translated into several lan- laundry items; the reproducibility coefficient guages, although formal reports on the psycho- was 0.93 (1, Table 2). A Pearson correlation of metric properties of the translations are rare. 0.87 was obtained between pairs of nurses who Reliability results have, however, been reported rated 36 patients; the agreement between two re- for a Japanese version (7). search assistants who independently rated 14 pa- tients was 0.91 (1, p182). Hokoishi et al. Commentary compared ratings by a variety of personnel and The PSMS is being used quite frequently in stud- obtained ICCs ranging from 0.86 to 0.96 for the ies of treatment for Alzheimer’s disease, but it is PSMS items, and 0.90 to 0.94 for the IADL chiefly known through the incorporation of items (7, Tables 1 and 2). Rubenstein et al., how- PSMS items into other scales. Most notably, an ever, found that ratings made by nurses, rela- expanded self-rating version of the ADL scale tives, and the patients themselves may not agree was included in the 1975 OARS Multidimen- closely (8). Very high six-month retest reliability sional Functional Assessment Questionnaire has been reported: 0.94 for the ADL scale (item (MFAQ) and later in Lawton’s 1982 Multilevel range, 0.84–0.96); the value for the IADL items Assessment Instrument (both are reviewed in was 0.88 (range, 0.80–0.99) (6, Tables 3 and 4). Chapter 10) (2). The self-rating version of the PSMS is shown in Lawton’s 1988 article (2, Validity pp795–797), and the items are virtually identi- The PSMS was tested on elderly people, some in cal to the physical ADL items in the OARS an institution and others living at home. It corre- MFAQ shown in Exhibit 10.21. The IADL scale lated 0.62 with a physician’s rating of functional described by Lawton and Brody was also modi- health (N = 130) and 0.61 with the IADL scale fied for inclusion in the OARS MFAQ; the items (N = 77) (1, Table 6). As would be expected, it were then further adapted for the Multilevel As- correlated less highly (r= 0.38) with the Kahn sessment Instrument. Mental Status Questionnaire, and it also corre- The other noteworthy feature of Lawton’s 88 Measuring Health work is his carefully developed conceptual defini- (8) Rubenstein LZ, Schairer C, Wieland GD, et tion of competence in everyday activities. This hi- al. Systematic biases in functional status erarchical model of disability extends the scope of assessment of elderly adults: effects of Katz’s approach in his Index of ADL. It is some- different data sources. J Gerontol what curious that the PSMS is used in clinical tri- 1984;39:686–691. (9) Davis KL, Marin DB, Kane R, et al. The als, because it appears relatively insensitive to Caregiver Activity Survey (CAS): change. Green et al. noted that the ADL items development and validation of a new mea- “tended to change only in patients with moder- sure for caregivers of persons with ately severe dementia, while scores on the IADLs Alzheimer’s disease. Int J Geriatr changed over a broader range of mild-to- Psychiatry 1997;12:978–988. moderate dementia severity.” (6, p659). Corre- (10) Tariot PN, Cummings JL, Katz IR, et al. A sponding to this insensitivity to change, very high randomized, double-blind, placebo- test-retest results were reported. Within this limi- controlled study of the efficacy and safety tation (which applies to most ADL scales), the of donepezil in patients with Alzheimer’s PSMS appears to be a reliable and valid ADL disease in the nursing home setting. J Am scale for clinical and survey research applications. Geriatr Soc 2001;49:1590–1599. (11) Rockwood K, Howlett S, Stadnyk K, et al. Responsiveness of goal attainment scaling Address in a randomized controlled trial of Information on the scales originally developed at comprehensive geriatric assessment. J Clin the Philadelphia Geriatric Center (including a Epidemiol 2003;56:736–743. copy of the PSMS IADL scale) can be found at www.abramsoncenter.org/PRI/scales.htm. The Disability Interview Schedule References (A.E. Bennett and Jessie Garrad, 1970) (1) Lawton MP, Brody EM. Assessment of older people: self-maintaining and Purpose instrumental activities of daily living. This Disability Interview Schedule was designed Gerontologist 1969;9:179–186. to measure the prevalence and severity of disabil- (2) Lawton MP. Scales to measure competence ity in epidemiological surveys for planning in everyday activities. Psychopharmacol health and welfare services. Bull 1988;24:609–614. (3) Lawton MP. Environment and other Conceptual Basis determinants of well-being in older people. This interview schedule follows the standard dis- Gerontologist 1983;23:349–357. (4) Lowenthal MF. Lives in distress: the paths tinction between disability and impairment. Dis- of the elderly to the psychiatric ward. New ability was defined as limitation of performance York: Basic Books, 1964. in “essential” activities of daily living, severe (5) Brody EM. Long-term care of older people: enough to entail depending on another person. a practical guide. New York: Human Impairment was defined as an anatomical, patho- Sciences Press, 1977. logical, or psychological disorder that may cause (6) Green CR, Mohs RC, Schmeidler J, et al. or be associated with disability (1). Functional decline in Alzheimer’s disease: a longitudinal study. J Am Geriatr Soc Description 1993;41:654–661. Bennett and Garrad’s 1966 prevalence survey of (7) Hokoishi K, Ikeda M, Maki N, et al. Interrater reliability of the Physical Self- disability in London used a brief screening ques- Maintenance Scale and the Instrumental tionnaire, followed by a 20-page interview Activities of Daily Living Scale in a variety schedule. Bennett also described 18-item and 15- of health professional representatives. item disability screening questionnaires that are Aging Ment Health 2001;5:38–40. not reviewed here (2).  Physical Disability and Handicap 89 The present review covers only the disability format of the questionnaire is a notable feature. section from the survey; it was applied to a sam- The instrument is, however, old and lacks valid- ple of 571 respondents aged 35 to 74 years, ity testing; potential users should consider the drawn from those identified as disabled and/or OECD instrument as an alternative. The Disabil- impaired by the screening questionnaire. The ity Interview Schedule may serve as an example schedule shown in Exhibit 3.12 is administered for those designing new survey measurements of by interviewers trained to probe to identify ac- disability. tual levels of performance. The questions use performance rather than capacity wording, and References the highest level of performance is recorded. If an answer falls between two defined levels, the (1) Garrad J, Bennett AE. A validated less severe grade of limitation is recorded. Rec- interview schedule for use in population ognizing that there are reasons other than dis- surveys of chronic disease and disability. Br ability why people may not perform an activity, J Prev Soc Med 1971;25:97–104. (2) Bennett AE, Garrad J, Halil T. Chronic allowances are made in scoring the schedule, for disease and disability in the community: a example, men who do not perform domestic du- prevalence study. Br Med J ties (1, 2). Details of the scoring system are not 1970;3:762–764. given, although separate scores are provided for (3) Williams RGA, Johnston M, Willis LA, et each topic, rather than a single score, which al. Disability: a model and measurement “masks different levels of performance in differ- technique. Br J Prev Soc Med ent areas, results in loss of information, and can 1976;30:71–78. be misleading” (1, p101). (4) St. Thomas’s health survey in Lambeth: disability survey. London: St. Thomas’s Reliability Hospital Medical School, Department of Complete agreement was obtained on test-retest Clinical Epidemiology and Social Medicine, 1971. ratings for 80% of 153 subjects after a 12- month delay (1, p103). For 28 of the 31 res pondents exhibiting some change on the ques- The Lambeth Disability tionnaire, medical records corroborated that Screening Questionnaire there had been a change in impairment or dis- (Donald L. Patrick and ability status. others, 1981) Guttman analyses of the questions gave a co- efficient of reproducibility of 0.95 and a coeffi- Purpose cient of scalability of 0.69 for females and 0.71 This postal questionnaire was designed to screen for males (3, p73). There were slight differences for physical disability in adults living in the com- in the ordering of items in scales derived for munity. It provides estimates of the prevalence of males and for females (4, Tables I to III). disability for use in planning health and social services. Validity Data from medical and social work records of Conceptual Basis 52 outpatients were compared with information Based on the impairment, disability, and handi- obtained with the interview schedule. The clini- cap triad, questions on disability concern mobil- cal records listed disability in a total of 118 ar- ity and self-care; they are phrased in terms of eas, of which 108 (91.5%) were identified by the difficulty in performing various activities rather interview schedule (1, p102). than in terms of reduced capacity. A section on impairments records the nature of the illnesses Commentary causing disability, and questions on handicap This instrument is one of relatively few disability cover housework, employment, and social activi- measurements designed for survey use; the clear ties. Exhibit 3.12 The Disability Interview Schedule Note: Cross in any box marked with an asterisk indicates presence of disability. MOBILITY Walking Do you walk outdoors in the street (with crutch or stick if used)? If ‘Yes’: one mile or more If ‘No’: and: 1 ⁄4 mile Between rooms ∗ Unaccompanied 100 yds. ∗ Within room ∗ Accompanied ∗ 10 yds. ∗ Unable to walk ∗ Acc. & support ∗ Stairs Do you walk up stairs? Do you walk down stairs? To 1st floor or above Unacc. From 1 floor to another Unacc. 5–8 steps or stairs ∗ Acc. ∗ 5–8 steps or stairs ∗ Acc. ∗ 2–4 steps or stairs ∗ Acc. & Supp. ∗ 2–4 steps or stairs ∗ Acc. & Supp. ∗ 1 step ∗ No need to 1 step ∗ No need to mount stairs descend stairs Mounts stairs other Goes down stairs other than by walking ∗ than by walking ∗ Unable to mount stairs ∗ Unable to descend stairs ∗ Transfer Yes No Yes No Do you need help to get into Do you need help to sit down in a bed? ........................................ ∗ chair? ............................................ ∗ Do you need help to get out of Do you need help to stand up from a bed? ........................................ ∗ chair? ............................................ ∗ Bedfast........................................ ∗ Not applicable .................................. Travel Do you drive yourself in Do you travel by bus or a car? train? Normal (unadapt.) If ‘Yes’: If ‘No’: Adapted Whenever necessary Unable to use bus and train ∗ Invacar ∗ Only out of rush hour ∗ Unable to use bus, train and car ∗ and: Self-propelled vehicle (outdoors) ∗ Unaccompanied Does not travel by choice Does not drive Accompanied ∗ Uses private transport by choice SELF CARE Are you able to Are you able to Are you able to Are you able to use Are you able to wash feed yourself: dress yourself undress yourself the lavatory: yourself: completely: completely: Without any Without any Without any Without any Without any help help help help help With specially With assistance prepared food With help with With help with Receptacles with- for shaving, or containers ∗ fastenings ∗ fastenings ∗ out assistance ∗ combing ∗ hair, etc. With help other With help other Lavatory with With help for With assistance ∗ than fastenings ∗ than fastenings ∗ assstance ∗ bodily washing ∗ Not at all, Receptacles with must be fed ∗ Does not dress ∗ Not applicable ∗ assistance ∗ Not at all ∗ 90  Physical Disability and Handicap 91 Exhibit 3.12 DOMESTIC DUTIES Do you do your own: all part none preference unable Shopping ∗ Cooking ∗ Cleaning ∗ Clothes washing ∗ Men with no household duties OCCUPATION Do you have a paid job at present? If ‘Yes’: and: If ‘No’: Full-time Part-time Normal working Modified working Sheltered employment ∗ ∗ Males 65 and over Females 60 and over { Age retired Prem. retired Non-employed ∗ Males 64 and under Females 59 and under { Unemployed Unfit Non-employed ∗ Reproduced from an original obtained from Dr. AE Bennett. With permission. Description (a) the ambulation, mobility, body care, or This instrument was developed as a screen for movement items, except constipation or disability in a health survey in Lambeth, a district stress incontinence alone; (b) the sensory- in London. The original version of the Lambeth motor items except vertigo when no associ- Questionnaire contained 31 questions drawn ated illness condition was reported; and/or (c) from the questionnaires of Bennett and Garrad, the social activity items, except limitation in of Haber, and of Harris (1–3). Twenty-five ques- working at all or doing the job of choice tions covering impairment and disability were re- where the respondent was over retirement tained in the version used in the Lambeth age. (5, p66) survey.* Subsequently a third version was de- signed with 22 items, 13 of which were taken, Reliability unchanged, from the previous instrument; two Sixty-eight people identified as disabled on the were new, two were items reintroduced from the first version of the questionnaire were interviewed first version, and five were reworded from the three to six months later. All were still classified second version (4). The third version is shown in as disabled in the follow-up interview, although Exhibit 3.13. All three versions record difficulties there were discrepancies in replies to several items with body movement, ambulation and mobility, (1). No reliability information is available for ver- self-care, social activity, and sensory problems. sions 2 and 3 of the questionnaire. The third version is interviewer-administered and collects data on the respondent only. It uses Validity a yes/no response format and scoring weights Peach et al. reported low levels of agreement be- are given (4, p304). tween self-ratings and assessments made by fam- ily physicians. The low agreement was attributed Respondents were classified as “disabled” if primarily to the doctors’ ignorance of the pa- they reported difficulty with one or more of tients’ disabilities (1). *This version was shown on pages 96 to 100 in the In the Lambeth Health Survey, the screening second edition of Measuring Health. questionnaire was followed 6 to 12 months later 92 Measuring Health Exhibit 3.13 The Lambeth Disability Screening 65% of the respondents, a regression equation Questionnaire (Version 3) was derived to predict the FLP scores; the equa- tion was then applied to the replies of the second Because of illness, accident or anything related to your health, do you have difficulty with any of the group. For the physical subscale of the FLP, the following? Read out individually and code. actual scores correlated 0.79 with those predicted a. Walking without help from the screening instrument; for the psychoso- b. Getting outside the house without help cial scales the correlation was 0.50 (4, p302). c. Crossing the road without help d. Travelling on a bus or train without help e. Getting in and out of bed or chair without help Commentary f. Dressing or undressing without help The second version of the Lambeth question- g. Kneeling or bending without help naire is one of very few validated postal screen- h. Going up or down stairs without help ing instruments available. The instrument i. Having a bath or all over wash without help proved acceptable to respondents: a response j. Holding or gripping (for example a comb or pen) without help rate of 86.6% was obtained in the Lambeth sur- k. Getting to and using the toilet without help vey of 11,659 households. Of the remainder, 8% l. Eating or drinking without help could not be contacted, 0.2% provided informa- Because of your health, do you have . . . tion too inadequate to analyze, and only 5.2% m. Difficulty seeing newspaper print even with refused (5). Locker et al. discussed methods for glasses reducing the bias incurred in estimating preva- n. Difficulty recognizing people across the road lence due to nonresponse (3). The use of one even with glasses person to record details about other family mem- o. Difficulty in hearing a conversation even with a hearing aid bers was apparently successful. p. Difficulty speaking The Lambeth Questionnaire is based on an Because of your health, do you have difficulty . . . established conceptual approach to disability, al- q. Preparing or cooking a hot meal without help though the wording of the questions may not in- r. Doing housework without help dicate performance, as intended. Questions ask, s. Visiting family or friends without help “Do you have difficulty with . . . ?,” a wording t. Doing any of your hobbies or spare time that seems to lie between performance and ca- activities u. Doing paid work of any kind (if under 65) pacity: it does not tell us whether the person v. Doing paid work of your choice (if under 65) does or does not do the activity in question, or whether he cannot. Indeed, question phrasing is Reproduced from Charlton JRH, Patrick DL, Peach H. Use of crucial: Patrick et al. attributed lower disability multivariate measures of disability in health surveys. J Epi- demiol Community Health 1983;37:304. Reproduced with prevalence estimates obtained in previous sur- permission from the BMJ Publishing Group. veys to their use of capacity question phrasing (5). This questionnaire appears to be of good quality, but sadly there is very limited evidence by an interview survey of 892 respondents identi- for its psychometric quality. fied as disabled, and a comparison group of 346 non disabled (6). Compared with the Functional Limitations Profile (FLP), a British version of the References Sickness Impact Profile, the Lambeth question- naire showed a sensitivity of 87.7% and a speci- (1) Peach H, Green S, Locker D, et al. Evaluation of a postal screening ficity of 72.2% (6, pp31–35). Because a change questionnaire to identify the physically in health status may have occurred between the disabled. Int Rehabil Med 1980;2:189–193. two assessments, these figures provide low esti- (2) Patrick DL. Screening for disability in mates of the validity of the questionnaire. Lambeth: a progress report on health and Charlton et al. compared version 3 of the care of the physically handicapped. London: questionnaire with the FLP (4). The sample of St. Thomas’s Hospital Medical School, 839 was randomly divided into two groups. Using Department of Community Medicine, 1978.  Physical Disability and Handicap 93 (3) Locker D, Wiggins R, Sittampalam Y, et al. tional performance reflects the impact of long- Estimating the prevalence of disability in term disability, overlaid by short-term fluctua- the community: the influence of sample tions. Indicators of short-term disability already design and response bias. J Epidemiol exist in the form of restricted activity or disability Community Health 1981;35:208–212. days. The OECD group considered these adequate (4) Charlton JRH, Patrick DL, Peach H. Use of and so focused the questionnaire on measuring multivariate measures of disability in health long-term disability among adults (2). surveys. J Epidemiol Community Health 1983;37:296–304. (5) Patrick DL, Darby SC, Green S, et al. Description Screening for disability in the inner city. J Of the 16 questions, ten can be used as an ab- Epidemiol Community Health 1981;35: breviated instrument and represent a core set of 65–70. items for international comparisons. They are (6) Patrick DL, ed. Health and care of the shown in Exhibit 3.14. No time specification is physically disabled in Lambeth. Phase I attached to these questions to define long-term report of The Longitudinal Disability disability. Rather, the respondent is asked what Interview Survey. London: St. Thomas’s he can usually do on a normal day, excluding Hospital Medical School, Department of Community Medicine, 1981. Exhibit 3.14 The OECD Long-Term Disability Questionnaire The OECD Long-Term Note: The ten questions with an asterisk are included in the Disability Questionnaire abbreviated version. (Organization for Economic *1. Is your eyesight good enough to read ordinary Cooperation and Development, newspaper print? (with glasses if usually worn). 1981) 2. Is your eyesight good enough to see the face of someone from 4 metres? (with glasses if Purpose usually worn). The OECD questionnaire is a survey instrument 3. Can you hear what is said in a normal that summarizes the impact of ill health on es- conversation with 3 or 4 other persons? (with hearing aid if you usually wear one). sential daily activities. It was intended to facili- *4. Can you hear what is said in a normal tate international comparisons of disability and, conversation with one other person? (with through repeated surveys, to monitor changes in hearing aid if you usually wear one). disability over time (1). *5. Can you speak without difficulty? *6. Can you carry an object of 5 kilos for 10 metres? Conceptual Basis 7. Could you run 100 metres? In 1976, the OECD sponsored an international *8. Can you walk 400 metres without resting? effort to develop a range of social and health in- *9. Can you walk up and down one flight of stairs dicators. Participating countries included without resting? Canada, Finland, France, West Germany, the *10. Can you move between rooms? *11. Can you get in and out of bed? Netherlands, Switzerland, the United Kingdom, *12. Can you dress and undress? and the United States. The health survey ques- 13. Can you cut your toenails? tionnaire measured disability in terms of limita- 14. Can you (when standing), bend down and pick tions in activities essential to daily living: up a shoe from the floor? mobility, self-care, and communication. The dis- *15. Can you cut your own food? (such as meat, fruit, etc.). ruption of normal social activity was seen as the 16. Can you both bite and chew on hard foods? central theme (1). (for example, a firm apple or celery). Two aspects of disability were considered: temporary alterations in functional levels and Reproduced from McWhinnie JR. Disability assessment in population surveys: results of the OECD common develop- long-term restrictions such as those arising from ment effort. Rev Epidémiol Santé Publique 1981;29:417, Mas- congenital anomalies. A person’s current func- son SA, Paris. With permission. 94 Measuring Health any temporary difficulties. Four response cate- physical movements and a physicians’s rating of gories were proposed: Yes, without difficulty; mobility ranged between 0.21 and 0.61 (7, Table Yes, with minor difficulty; Yes, with major diffi- II). Item-total correlations ranged from 0.14 to culty and No, not able to. These were not strictly 0.54. adhered to in the field trials and sometimes cate- gories 2 and 3 were merged into “yes, with diffi- Alternative Forms culty.” A detailed presentation of the rationale for In 1992, the WHO coordinated the development question selection and administration is given in of a revised disability measure for use in health the OECD report (2). interview surveys (8). The WHO Disability Ques- tionnaire contains 13 items, including seven Reliability from the OECD instrument. A Dutch study re- Wilson and McNeil used 11 of the questions, ported a low level of agreement (kappa values slightly modified, in interviews that were re- ranging from 0.16 to 0.45 for the various sec- peated after a two-week delay (N = 223) (3). It tions) between self-report on this instrument and was not always possible to reinterview the origi- direct observation of abilities (8, Table 1). nal respondent, and in about half of the cases a proxy report was used. The agreement between Commentary first and second interviews was low, ranging The OECD questionnaire represents an early at- from about 30% to 70% for the 11 items. Con- tempt to develop an internationally applicable sidering the scale as a whole, fewer than two set of disability items; the WHOQOL and Euro- thirds of those who reported disabilities on ei- Qol instruments described in Chapter 10 offer ther interview reported them on both. Analyses more recent examples. As well as the studies showed that the inconsistencies were not due to cited here, the OECD scale was used in France using proxy respondents (3). (9), Japan, and West Germany. The questions A Dutch survey compared the responses to a continued to be used in Canadian national sur- self-administered version of the questionnaire veys in self-administered and interview formats (N = 940) with an interview version (N = 500). (10; 11). Many of the questions are similar to Although the two groups were very similar in those in the RAND Corporation scales and in age and sex, on average 3.1% more people de- the U.S. Social Security Administration disability clared some level of disability in the written ver- surveys. However, because none of the original sion (4, p466). contributing authors is still directly involved with this instrument, the questionnaire is un- Validity likely to see further improvement. Twelve of the OECD questions were included in Although the idea of an internationally stan- a Finnish national survey (N = 2,000). With the dardized scale is commendable, it was not fully exception of people over 65 years of age, most achieved. Most studies exhibited slight varia- expressed no difficulty with any of the activities tions in the questions or answer categories. covered (5, Table 1). Similar findings were ob- There are also certain illogicalities in the scale: tained in the United States and in the Nether- although it is intended to measure the behavioral lands (3; 4). The questions were applied to 1,600 consequences of disability, the questions are Swiss respondents aged 65 and over, and Ray- worded for capacity rather than performance. mond et al. reported sensitivity results (6). For The method is designed as a survey instrument, different medical conditions, sensitivity ranged but the questions cover relatively severe levels of from 61% to 85%, being highest for those with disability so that few people in the general adult vision, hearing, and speech problems. Specificity population answer affirmatively; the questions was 76% (6, p455). are most relevant to people over 65. The low In Canada, the questions were tested in inter- test-retest reliability reported in the United views with 104 rehabilitation outpatients. Cor- States is cause for concern; the distinction be- relations between the questions concerned with tween short- and long-term disability may not  Physical Disability and Handicap 95 have been adequately explained to the respon- personnes âgées en France, à l’aide de dents, who may have reported minor and tran- plusieurs indicateurs, dont les questions de sient difficulties rather than long-term problems l’OCDE. Rev Epidémiol Santé Publique (3). The distinction between acute and chronic 1981;29:451–459. disability is hard to draw, especially where a re- (10) McDowell I, Praught E. Report of the Canadian health and disability survey, spondent has problems of both types that may 1983–1984 (Catalogue No. 82-55E). interact. Linked to this, the instructions to the Ottawa, Ontario: Minister of Supply and respondents lack clarity. Services, 1986. Although it has been widely used, there are (11) Furrie A. A national database on disabled problems with this scale. Reliability and validity persons: making disability data available to results are poor. The instrument is narrow in users. Ottawa, Ontario: Statistics Canada, scope compared, for example, with the Lambeth 1987. questionnaire, which covers employment and so- cial activities as well as the ADL and IADL The Functional Status themes included in the OECD instrument. Rating System (Stephen K. Forer, 1981) References Purpose The Functional Status Rating System (FSRS) es- (1) McWhinnie JR. Disability assessment in timates the assistance required by rehabilitation population surveys: results of the OECD patients in their daily lives. It covers indepen- common development effort. Rev Epidémiol dence in ADL, ability to communicate, and so- Santé Publique 1981;29:413–419. (2) McWhinnie JR. Disability indicators for cial adjustment. measuring well-being. Paris: OECD Social Indicators Programme, 1981. Conceptual Basis (3) Wilson RW, McNeil JM. Preliminary No information is available. analysis of OECD disability on the pretest of the post census disability survey. Rev Description Epidémiol Santé Publique 1981;29:469–475. This rating scale was based on a method devel- (4) van Sonsbeek JLA. Applications aux Pays- oped by the Hospitalization Utilization Project Bas des questions de l’OCDE relatives à of Pennsylvania (HUP) initiated in 1974 to pro- l’incapacité. Rev Epidémiol Santé Publique vide national statistics on hospital utilization 1981;29:461–468. and treatment outcomes (1). A preliminary ver- (5) Klaukka T. Application of the OECD disability questions in Finland. Rev sion of the FSRS covered five ADL topics (2); the Epidémiol Santé Publique 1981;29:431–439. revised rating form described here is broader in (6) Raymond L, Christe E, Clemence A. Vers scope: 30 items cover five topics. The items are l’établissement d’un score global summarized in Exhibit 3.15; the scales on which d’incapacité fonctionnelle sur la base des the items are rated are shown at the foot of the questions de l’OCDE, d’après une enquête exhibit. An instruction manual gives more de- en Suisse. Rev Epidémiol Santé Publique tailed definitions of each item (3). Ratings are 1981;29:451–459. made by the treatment team member with pri- (7) McDowell I. Screening for disability. An mary responsibility for that aspect of care. Item examination of the OECD survey questions scores are averaged to form scores for each of in a Canadian study. Rev Epidémiol Santé the five sections. The scale can be completed in Publique 1981;29:421–429. (8) Wijlhuizen GJ, Ooijendijk W. Measuring 15 to 20 minutes. disability, the agreement between self evaluation and observation of performance. Reliability Disabil Rehabil 1999;21:61–67. Information is available for the preliminary ver- (9) Mizrahi A. Evaluation de l’état de santé de sion only. Inter-rater agreement was high, but Exhibit 3.15 The Functional Status Rating System Functional Status in Self-care A. Eating/feeding: Management of all aspects of setting up and eating food (including cutting of meat) with or without adaptive equipment. B. Personal hygiene: Includes set up, oral care, washing face and hands with a wash cloth, hair grooming, shaving, and makeup. C. Toileting: Includes management of clothing and cleanliness. Bathing: Includes entire body bathing (tub, shower, or bed bath). E. Bowel management: Able to insert suppository and/or perform manual evacuation, aware of need to defecate, has sphincter muscle control. F. Bladder management: Able to manage equipment necessary for bladder evacuation (may include intermittent catheterization). G. Skin management: Performance of skin care program, regular inspection, prevention of pressure sores, rashes, or irritations. H. Bed activities: Includes turning, coming to a sitting position, scooting, and maintenance of balance. I. Dressing: Includes performance of total body dressing except tying shoes, with or without adaptive equipment (also includes application of orthosis & prosthesis). Functional Status in Mobility A. Transfers: Includes the management of all aspects of transfers to and from bed, mat, toilet, tub/shower, wheelchair, with or without adaptive equipment. B. Wheelchair skills: Includes management of brakes, leg rests, maneuvering and propelling through and over doorway thresholds. C. Ambulation: Includes coming to a standing position and walking short to moderate distances on level surfaces with or without equipment. D. Stairs and environmental surfaces: Includes climbing stairs, curbs, ramps or environmental terrain. E. Community mobility: Ability to manage transportation. Functional Status in Communication A. Understanding spoken language B. Reading comprehension C. Language expression (non-speech/alternative methods): Includes pointing, gestures, manual communication boards, electronic systems. D. Language expression (verbal): Includes grammer, syntax, and appropriateness of language. E. Speech intelligibility F. Written communication (motor) G. Written language expression: Includes spelling, vocabulary, punctuation, syntax, grammar, and completeness of written response. Functional Status in Psychosocial Adjustment A. Emotional adjustment: Includes frequency and severity of depression, anxiety, frustration, lability, unresponsiveness, agitation, interference with progress in therapies, motivation, ability to cope with and take responsibility for emotional behavior. B. Family/significant others/environment: Includes frequency of chronic problems or conflicts in patient’s relationships, interference with progress in therapies, ability and willingness to provide for patient’s specific needs after discharge, and to promote patient’s recovery and independence. C. Adjustment to limitations: Includes denial/awareness, acceptance of limitations, willingness to learn new ways of functioning, compensating, taking appropriate safety precautions, and realistic expectations for long-term recovery. D. Social adjustment: Includes frequency and initiation of social contacts, responsiveness in one to one and group situations, appropriateness of behavior in relationships, and spontaneity of interactions. Functional Status in Cognitive Function A. Attention span: includes distractibility, level of alertness and responsiveness, ability to concentrate on a task, ability to follow directions, immediate recall as the structure, difficulty and length of the task vary. B. Orientation C. Judgment reasoning D. Memory: Includes short- and long-term. E. Problem-solving 96  Physical Disability and Handicap 97 Exhibit 3.15 Summary of Rating Scales Communication, psychosocial and cognitive function Self-care and mobility items items 1.0 = Unable—totally dependent 1.0 = Extremely severe 1.5 = Maximum assistance of 1 of 2 people 1.5 = Severe 2.0 = Moderate assistance 2.0 = Moderately severe 2.5 = Minimal assistance 2.5 = Moderate impairment 3.0 = Standby assistance 3.0 = Mild impairment 3.5 = Supervised 3.5 = Minimal impairment 4.0 = Independent 4.0 = No impairment Adapted from the Rating Form obtained from Stephen K Forer. varied according to the professional background instrument. Glendale, California: of the rater and the method of administration. Rehabilitation Institute, Glendale Adventist Correlations ranged from 0.81 to 0.92 (2, p362). Medical Center, December 1981. Validity Some predictive validity results were presented The Rapid Disability Rating Scale by Forer and Miller (2). Admission scores on (Margaret W. Linn, 1967, bladder management and cognition were found Revised 1982) to predict the eventual placement of the patient Purpose in home or institutional care. The instrument The Rapid Disability Rating Scale (RDRS) was de- was shown capable of reflecting improvement veloped as a research tool for summarizing the between admission and discharge for a number functional capacity and mental status of elderly of diagnostic groups (2, Table 2). long-stay patients. It may be used with hospitalized patients and with people living in the community. Commentary Despite its lack of validation, we have included Conceptual Basis this scale because of its broad scope and be- No information is available. cause, as a clinical instrument, the scale appears relevant in routine patient assessment and in set- Description ting rehabilitation goals. The lack of formal reli- The 1967 version of the RDRS contained 16 ability and validity testing makes it unsuitable as items covering physical and mental functioning a research instrument. and independence in self-care. A revised scale of A revised version of the scale presented in the 18 items was published by Linn and Linn as the Exhibit was incorporated into the Functional Inde- RDRS-2 in 1982 (1, 2). Changes included the pendence Measure described later in this chapter. addition of three items covering mobility, toilet- ing, and adaptive tasks (i.e., managing money, References telephoning, shopping); a question on safety su- pervision was dropped (1, p379). Four-point re- (1) Breckenridge K. Medical rehabilitation sponse scales replaced the earlier three-point program evaluation. Arch Phys Med scales. The RDRS-2 has eight questions on Rehabil 1978;59:419–423. (2) Forer SK, Miller LS. Rehabilitation ADLs, three on sensory abilities, three on mental outcome: comparative analysis of different capacities, and one question on each of dietary patient types. Arch Phys Med Rehabil changes, continence, medications, and confine- 1980;61:359–365. ment to bed (see Exhibit 3.16). The following (3) Forer SK. Revised functional status rating review refers mainly to the revised version. Exhibit 3.16 The Rapid Disability Rating Scale-2 Directions: Rate what the person does to reflect current behavior. Circle one of the four choices for each item. Consider rating with any aids or prostheses normally used. None = completely independent or normal behavior. Total = that person cannot, will not or may not (because of medical restriction) perform a behavior or has the most severe form of disability or problem. Assistance with activities of daily living Eating None A little A lot Spoon-feed; intravenous tube Walking (with cane or walker if used) None A little A lot Does not walk Mobility (going outside and getting None A little A lot Is housebound about with wheelchair, etc., if used) Bathing (include getting supplies, None A little A lot Must be bathed supervising) Dressing (include help in selecting None A little A lot Must be dressed clothes) Toileting (include help with clothes, None A little A lot Uses bedpan or unable cleaning, or help with ostomy, to care for ostomy/ catheter) catheter Grooming (shaving for men, hair- None A little A lot Must be groomed dressing for women, nails, teeth) Adaptive tasks (managing money/ None A little A lot Cannot manage possessions; telephoning, buying newspaper, toilet articles, snacks) Degree of disability Communication (expressing self ) None A little A lot Does not communicate Hearing (with aid if used) None A little A lot Does not seem to hear Sight (with glasses, if used) None A little A lot Does not see Diet (deviation from normal) None A little A lot Fed by intravenous tube In bed during day (ordered or self- None A little A lot Most/all of time initiated) (<3 hrs) Incontinence (urine/feces, with None Sometimes Frequently Does not control catheter or prosthesis, if used) (weekly +) Medication None Sometimes Daily, taken orally Daily; injection; (+ oral if used) Degree of special problems Mental confusion None A little A lot Extreme Uncooperativeness (combats efforts to None A little A lot Extreme help with care) Depression None A little A lot Extreme Reprinted with permission from the American Geriatrics Society. The Rapid Disability Rating Scale-2, by Linn MW and Linn BS (Journal of the American Geriatrics Society, Vol 30, p 380, 1982). 98  Physical Disability and Handicap 99 The rating scale is completed by a nurse or a tality was explained by the RDRS-2, which cor- person familiar with the patient. Because the rectly identified 72% of patients who would die scale describes performance, the rater must ob- (1, p382). serve the patient carrying out the various tasks Correlations of 0.27 were obtained between rather than rely on self-report. After the rater the RDRS-2 and a physician’s 13-item rating has made the observations, it takes about two scale of impairment of 172 elderly patients living minutes to complete the scale. in the community; a correlation of 0.43 was ob- Response categories are phrased in terms of tained with a six-point self-report scale of health the amount of assistance the patient requires and (1, p382). each item is weighted equally in calculating an overall score. Scores range from 18 to 72, with Alternative Forms higher values indicating greater disability. Three A French version has been used (5). subscores may be used, indicating the degree of assistance required with activities of daily living, Reference Standards physical disabilities, and psychosocial problems No formal reference standards are available, but (1, p380). Linn and Linn noted that for the RDRS-2 scores for elderly community residents with minimal Reliability disabilities average 21 to 22. For hospitalized el- Inter-rater reliability of the preliminary version derly patients, the average is about 32, and for was assessed by comparing ratings of 20 patients those transferred to long-term care facilities, it is made independently by three raters; a coefficient about 36 (1, p380). of 0.91 was obtained using Kendall’s W index of concordance (3, p213). The retest correlation af- Commentary ter a mean delay of three-and-a-half days was This is a broad-ranging scale that rates the 0.83, and the mean scores of the two sets of rat- amount of assistance required in 18 activities, ings were within one point of each other (3, broader in scope than the PULSES, Barthel, and p213). Linn et al. reported a one-week test-retest most ADL scales. It has been used in several correlation of 0.89 on 1,000 male patients for evaluative studies (4; 6; 7). Its research orienta- the original 16-item version of the RDRS (4, tion is reflected in the reliability and validity test- p340). Linn and Linn reported item reliability ing, which is more complete than other clinical results for the revised version: two nurses inde- scales. Nonetheless, the validity tests could be pendently rated 100 patients and item correla- improved. For example, correlations with physi- tions ranged from 0.62 to 0.98; the three lowest cians’ ratings commonly produce low coeffi- correlations were for the mental status items (1, cients because the physician is not aware of Table 2). Test-retest reliability on 50 patients af- details of the patient’s level of functioning. The ter an interval of three days produced correla- use of predictive validation is imaginative, but tions between 0.58 and 0.96 (1, pp380–381). because this is rarely attempted with such in- dexes it is hard to judge whether a 20% expla- Validity nation variance is high or low. It would be A factor analysis of ratings of 120 hospitalized helpful if studies of predictive validity reported patients provided a three-factor solution; the findings in a comparable manner: Granger ex- factors reflected ADLs, disability, and psycho- pressed the predictive validity of the Barthel In- logical problems (1, p381). The latter were la- dex in terms of percentages of patients with low beled “special problems” by the authors, as scores who died. shown in Exhibit 3.16. Criticisms have been made of the scoring sys- Ratings of 845 men (mean age, 68 years) tem. For example, the same weight is assigned to were used to predict subsequent mortality using different degrees of disability: “permanent con- multiple regression and discriminant function finement to bed” and “following a special diet” analyses. Twenty percent of the variance in mor- both rate three points (5, p345). This limits the 100 Measuring Health validity of the scale in giving absolute indica- “for their use of broad categories of activity tions of disability, although it may be less serious (e.g., dressing) which incorporated complex se- if the scale is used to monitor change over time. ries of activities involving many different joints and muscle groups” (4, p576). Jette also argued References that the outcomes of care should not be viewed solely in terms of independence, as is the case in (1) Linn MW, Linn BS. The Rapid Disability most ADL scales. Sometimes providing assis- Rating Scale-2. J Am Geriatr Soc tance to a patient, which increases dependence, 1982;30:378–382. may alleviate pain and reduce difficulty. He chal- (2) Linn MW. Rapid Disability Rating Scale-2 lenged “the exclusive emphasis on level of de- (RDRS-2). Psychopharmacol Bull 1988;24:799–800. pendence in previous work as well as the (3) Linn MW. A rapid disability rating scale. J assumption that assistance in ADL constitutes a Am Geriatr Soc 1967;15:211–214. loss of health” (4, p576). Accordingly, the FSI (4) Linn MW, Gurel L, Linn BS. Patient was designed to measure pain and difficulty, as outcome as a measure of quality of nursing well as level of dependence, in performing tasks. home care. Am J Public Health 1977;67:337–344. Description (5) Jenicek M, Cléroux R, Lamoureux M. Based on the Barthel, PULSES, and Katz instru- Principal component analysis of four health ments, the FSI was developed as a comprehensive indicators and construction of a global ADL assessment for adults living in the commu- health index in the aged. Am J Epidemiol 1979;110:343–349. nity (2; 3). The original FSI contained 45 ADL (6) Ogren EH, Linn MW. Male nursing home items (they are shown in reference 2, Table 1). patients: relocation and mortality. J Am Three questions were asked for each activity, Geriatr Soc 1971;19:229–239. yielding separate ratings for dependency, diffi- (7) Linn MW, Linn BS, Harris R. Effects of culty, and pain. The resulting 135 questions (45 counseling for late stage cancer patients. items × three dimensions) took between 60 and Cancer 1982;49:1048–1055. 90 minutes to administer (5). This proved unworkable, and factor analyses guided the abbreviation of the FSI to the current The Functional Status Index version, shown in Exhibit 3.17. The 18 items are (Pilot Geriatric Arthritis Program, grouped under five headings: mobility, hand ac- Alan M. Jette, 1978, Revised 1980) tivities, personal care, home chores, and social/ Purpose role activities. The FSI is administered by an in- The Functional Status Index (FSI) was designed terviewer and covers performance over the previ- to assess the functional status of adult patients ous seven days. The questions are asked three with arthritis living in the community (1). In- times. First, to assess dependency (or level of as- tended both as a clinical and an evaluative tool, sistance used) the respondent is asked: “How the scale measures the degree of dependence, much help did you use to do ___, on average, pain, and difficulty experienced in performing during the past week?” A five-point rating scale ADLs (2). runs from independent to unable to do the activ- ity. Second, the same items are used to assess the Conceptual Basis level of pain experienced when performing each The FSI was developed to evaluate a Pilot Geri- activity; four-point scales are used for the pain atric Arthritis Program (PGAP) that sought to rating, and also for the third rating of the improve the quality of life of elderly patients amount of difficulty experienced. Alternatively, with arthritis (3). The goals of the program were 0 to 13 or 0 to 7 ladder scales have been used, to prevent disability, restore activity, reduce but the 4-point rating is the standard approach pain, and encourage social and emotional adjust- (Dr. A. M. Jette, personal communication, ment (3). Previous instruments were criticized 1993). Cue cards may be used to show the an- Exhibit 3.17 The Functional Status Index Functional Dependence “In this first section of the interview, we are trying to measure the degree to which you used help to perform your daily activities, on the average, during the past 7 days. By help, I mean the extent to which you used equipment (such as a cane), whether you used human assistance (such as a friend or relative), and whether you used both equipment and human assistance to do certain activities. “I would now like you to tell me how much help you used, on the average, during the past week, to do each activity I will read to you. Tell me if you did the activity without help, used equipment, used human assistance, used both equipment and human assistance, or if you were unable or it was unsafe to do it. “Do you have any questions before we begin? “How much help did you use when ______, on average, during the past week?” (Repeat for each item). Items: Gross Mobility Hand Activities Walking inside Writing Climbing up stairs Opening a container Rising from a chair Dialing a phone Personal Care Home Chores Putting on pants Vacuuming a rug Buttoning a shirt/blouse Reaching into low cupboards Washing all parts of the body Doing laundry Putting on a shirt/blouse Doing yardwork Social/Role Activities Performing your job Driving a car Attending meetings/appointments Visiting with friends and relatives Functional Pain “In this section of the interview, we are trying to measure the amount of pain you experienced when you performed your daily activities during the past week. For each activity performed during the past 7 days, I would like you to judge the amount of pain you experienced when doing it. For each activity you performed, please judge whether you experienced no, mild, moderate, or severe pain when performing the activity. “By pain, I mean the discomfort or sensation of hurting you experienced when doing the activity. “Do you have any questions before we start? “How much pain did you experience, on average, during the past week when ———? Would you say no, mild, moderate or severe pain?” (Repeat for each item, except those that the person has said they did not attempt). Functional Difficulty “In this section of the interview, we are trying to measure how difficult it was to perform each activity, on average, during the past 7 days. By difficulty, we mean how easy or hard it was to do the activity. For each activity you performed, please tell me whether you experienced no, mild, moderate, or severe difficulty in doing it. “Do you have any questions? “How much difficulty did you have in ______, on average, during the previous 7 days? Would you say no, mild, moderate or severe difficulty?” (Repeat for each item, except those that the person has said they did not attempt). Adapted from an original sent by Dr. AM Jette. With permission. 101 102 Measuring Health swer categories to the respondent. The 18-item and a self-rating of ability to deal with disease- version of the FSI takes 20 to 30 minutes to ad- related problems averaged 0.24; correlations minister (5). with a self-rating of joint condition averaged 0.39 (3, Table 1). Correlations with ratings made Reliability by the staff were lower, ranging from 0.11 to forty-five-item version. Jette and Deniston 0.22 (3, Table 4). studied inter-rater reliability in assessing 19 pa- tients and found that as the degree of pain and eighteen-item version. A factor analysis iden- difficulty increased, agreement between raters tified the five factors shown in Exhibit 3.17 (5). decreased (1, Table 3). The agreement among In a different study analyzing 36 items, Jette nine raters yielded intraclass correlations averag- again obtained five factors, accounting for ing 0.78 for the dependence dimension, 0.61 for 58.5% of the variance (2, Table 2). There was difficulty, and 0.75 for pain (1, Table 4). Liang some contrast in the factor structure for the and Jette reported equivalent figures of 0.72, pain, difficulty, and dependency ratings, but Jette 0.75, and 0.78 (6). concluded that five functional categories are Liang and Jette reported test-retest reliability common to the three dimensions (2). of 0.75 in the dependence dimension, 0.77 in the In a sample of 80 patients with rheumatoid difficulty dimension, and 0.69 in the pain di- arthritis, Shope et al. obtained correlations rang- mension (6, p83). ing from 0.40 to 0.43 with the American Rheumatism Association functional class and eighteen-item version. Jette compared the in- from 0.40 to 0.47 with a physician assessment ternal consistency reliability for the five dimen- of functional ability (8, Table V). The FSI has sions of this version and also examined three been used in evaluating change following treat- different response modes: defined response op- ment; in a small study of 15 patients with arthri- tions (of the type listed above), a ladder scale, tis, change in FSI scores correlated 0.49 with and a Q-sort technique (5). For 149 patients, the improvement in muscle strength, 0.53 with im- internal consistency of the mobility and personal provement in endurance, and 0.67 in time taken. care sections ranged from 0.70 to over 0.90 (5, These variables predicted 77% of the variance in Tables 5–7). Similar reliability results were FSI change scores (9). Liang et al. provide a achieved with each of the three response modes, more comprehensive comparison of sensitivity except that in assessing pain levels the fixed an- to change of five scales in evaluating change after swer categories proved less reliable than the hip or knee surgery (10). The FSI proved to be other scaling techniques (5). A subsequent publi- the least sensitive of the measures, in some com- cation repeated the same analyses, but added parisons requiring a sample size three or four test-retest results (ranging from 0.69 to 0.88) times greater than those needed by the Arthritis and inter-observer agreement (0.64 to 0.89) (7, Impact Measurement Scales to demonstrate a Tables 2–4). significant improvement (10, Table 3). Validity Alternative Forms forty-five-item version. Deniston and Jette Harris et al. and Jette et al. have described a dif- compared responses to the 45-item FSI with rat- ferent instrument that they also named the Func- ings made by hospital staff and with self-ratings tional Status Index (11, 12). Rudimentary made by 95 elderly patients with arthritis. Cor- validity data for a 17-item version were reported relations between the patients’ judgments of by Harris et al. for 47 elderly patients with hip their “number of good days in the past week” fractures (11). and their FSI scores were 0.14 for the depend- A 12-item modification of the FSI showed an ence dimension, 0.41 for difficulty, and 0.46 for alpha reliability of 0.91 and correlated 0.46 with pain (3, Table 2). Correlations of the FSI scores the Quality of Well-Being Scale (13, p962).  Physical Disability and Handicap 103 Commentary instrument. Arch Phys Med Rehabil This instrument is similar to the Kenny scale in its 1980;61:395–401. aim of providing a more detailed disability rating (6) Liang MH, Jette AM. Measuring functional than competing scales. The FSI is well-founded on ability in chronic arthritis: a critical review. a conceptual analysis of disability; the distinction Arthritis Rheum 1981;24:80–86. (7) Jette AM. The Functional Status Index: it makes between difficulty, dependence, and pain reliability and validity of a self-report is innovative and may prove helpful. These dimen- functional disability measure. J Rheumatol sions have received some empirical support 1987;14(suppl 15):15–19. through factor analyses and contrasting correla- (8) Shope JT, Banwell BA, Jette AM, et al. tions with other scales. Deniston and Jette noted Functional status outcome after treatment that the distinction between dependence and the of rheumatoid arthritis. Clin Rheumatol other two dimensions is meaningful, but the con- Pract 1983;1: 243–248. trast between pain and difficulty remains equivo- (9) Fisher NM, Pendergast DR, Gresham GE, cal (3). Jette does not report the correlation et al. Muscle rehabilitation: its effect on between the pain and difficulty dimensions; it is muscular and functional performance of to be hoped that future studies will examine the patients with knee osteoarthritis. Arch Phys Med Rehabil 1991;72:367–374. necessity of keeping these two dimensions sepa- (10) Liang MH, Fossel AH, Larson MG. rate. Reliability results for the FSI are good. Comparisons of five health status The existence of different versions (15 items instruments for orthopedic evaluation. Med reported by Shope, 17 by Harris, and the stan- Care 1990;28:632–642. dard 18 items) is a problem shared by several (11) Harris BA, Jette AM, Campion EW, et al. other health measures. The suggestion (11, p35; Validity of self-report measures of 12, p736) that reliability data for the 18-item functional disability. Top Geriatr Rehabil version also hold for the very different 17-item 1986;1:31–41. version is misleading. Some validity results may (12) Jette AM, Harris BA, Cleary PD, et al. cause concern: several of the criterion correla- Functional recovery after hip fracture. tions are lower than those obtained with other Arch Phys Med Rehabil 1987;68:735–740. (13) Ganiats TG, Palinkas LA, Kaplan RM. scales that we review. It is desirable that more ev- Comparison of Quality of Well-Being Scale idence on validity be accumulated, including and Functional Status Index in patients testing on conditions other than arthritis, before with atrial fibrillation. Med Care this scale can be fully recommended. 1992;30:958–964. References The Patient Evaluation (1) Jette AM, Deniston OL. Inter-observer Conference System reliability of a functional status assessment (Richard F. Harvey and Hollis M. instrument. J Chronic Dis 1978;31:573–580. Jellinek, 1981) (2) Jette AM. Functional capacity evaluation: an empirical approach. Arch Phys Med Purpose Rehabil 1980;61:85–89. The Patient Evaluation Conference System (3) Deniston OL, Jette A. A functional status (PECS) rates the functional and psychosocial assessment instrument: validation in an el- status of rehabilitation patients. It is intended for derly population. Health Serv Res use in defining treatment goals and in evaluating 1980;15:21–34. (4) Jette AM. Health status indicators: their progress toward them. utility in chronic-disease evaluation research. J Chronic Dis 1980;33:567–579. Conceptual Basis (5) Jette AM. Functional Status Index: Although no formal conceptual basis was given reliability of a chronic disease evaluation to justify the content of the instrument, Harvey 104 Measuring Health and Jellinek described several principles that proved by the omission of items on psychologi- guided the design of the PECS. These included cal distress (4, Tables 2–6). its need to be able to reflect minor changes in In a comparison of the PECS and the FIM us- functional level, its multidisciplinary scope (cov- ing a Rasch analysis, the PECS cognition scale ering medical, psychological, social, and voca- showed wider coverage than the corresponding tional topics), and its simplicity of application, FIM scale, whereas the two motor scales had a scoring, and interpretation (1). comparable range of coverage (4, Figures 2 and 3). An abbreviated, self-administered version of Description the PECS was compared with the Brief Symptom The PECS is a broad-ranging instrument con- Inventory, which measures emotional distress, taining 79 functional assessment items grouped for 22 brain-injured patients. Significant correla- into 15 sections with an additional three sec- tions in the range 0.38 to 0.47 were obtained tions pertaining to the results of a case confer- with the self-care, mobility, living arrangements, ence. Each section is completed by the staff and communication scales on the PECS (5, Table member who has primary responsibility for 2). Two PECS scales, bladder and skin care, that aspect of care. The ratings made by each were assessed at time of discharge on 28 pa- therapist are collated onto a master form that tients. The results correlated with several depres- summarizes the rehabilitation goals. This is sion scores recorded at admission to the used in case conferences to record the patient’s rehabilitation program, with coefficients be- progress. The PECS form shown in Exhibit tween 0.37 and 0.39 (6, p361). 3.18 was obtained from Dr. Harvey and is a A study of 30 head trauma patients com- slightly expanded version of that shown in (2, pared change in PECS scores between admission Figure 2). and discharge with the results of computed to- Eight-point responses are used for most mography (CT) scans (7). Three patients found items, with 0 representing unmeasured or by CT scan to have no lesions achieved complete unmeasurable function and 7 representing full recovery in four out of five PECS scales; the ten independence. Scores are comparable across dif- patients with unilateral lesions achieved indepen- ferent scales: a cutting-point of 5 distinguishes dence in at least two areas, whereas for 17 pa- between a need for human assistance and man- tients with bilateral lesions, there were no areas aging independently (with or without aids). A in which all patients recovered completely (7, few items use four-point scales. Table 2). In a discriminant analysis, PECS scores cor- Reliability rectly categorized 75% of patients in three con- Inter-rater reliability for different sections of the trasting levels of rehabilitation program (8). PECS ranged from 0.68 to 0.80 for 125 patients (1, p459). Commentary The PECS is an older scale that has passed from Validity the mainstream of measures. Although it has A factor analysis of the PECS items produced some distinctive features, such as its use of a eight factors, covering cognition, motor control, goal-attainment approach, few recent publica- self-care, communication, physical impairment, tions have reviewed the validity and reliability assistive devices, social interaction, and family of the PECS. Harvey et al. expanded the original support (3, Table 4). This was followed by PECS to include documentation for outpatient analyses to determine whether the scales fit a evaluation, generating reports to referring Rasch measurement model (4). The motor com- physicians, and a graphical presentation of data petence, self-care, and impairment severity scales has been developed (2). However, most valida- of the PECS did fit a unidimensional Rasch tion studies used very small samples and addi- model, but the cognitive scale would be im- tional evidence on the quality of this scale Exhibit 3.18 The Patient Evaluation Conference System Scores range from 0 to 7, with 0 being the lowest score, or not assessed, and 7 being the highest score, such as normal or independent. Scores of 1 to 4 indicate dependent function. Scores of 5 or more indicate independent function. Keys to scores are available in each participating discipline. Instructions: Circle (0) the goal score example X the current status score 0 1 2 3 4 5 6 7 Dependent Independent I. Rehabilitation Medi- 6. Driving mobility 0 1 2 3 4 5 6 7 cine (MED) 7. Assumes responsi- 0 1 2 3 4 5 6 7 1. Motor loss (includ- 0 1 2 3 4 5 6 7 bility for mobility ing muscle weak- ness and limb defi- IV. Activities of Daily Liv- ciency) ing (ADL) 2. Spasticity/ 0 1 2 3 4 5 6 7 1. Performance in fee- 0 1 2 3 4 5 6 7 involuntary ding movement 2. Performance in hy- 0 1 2 3 4 5 6 7 (including dystonia giene/grooming and ataxia) 3. Performance in 0 1 2 3 4 5 6 7 3. Joint limitations 0 1 2 3 4 5 6 7 dressing 4. Autonomic dis- 0 1 2 3 4 5 6 7 4. Performance in 0 1 2 3 4 5 6 7 turbance home management 5. Sensory deficiency 0 1 2 3 4 5 6 7 5. Performance of 0 1 2 3 4 5 6 7 6. Perceptual and cog- 0 1 2 3 4 5 6 7 mobility in home nitive deficit environment (in- 7. Associated medical 0 1 2 3 4 5 6 7 cluding utilization problems of environmental adaptations for II. Rehabilitation Nurs- communication) ing (NSG) 1. Performance of 0 1 2 3 4 5 6 7 V. Communication bowel program (COM) 2. Performance of uri- 0 1 2 3 4 5 6 7 1. Ability to compre- 0 1 2 3 4 5 6 7 nary program hend spoken language 3. Performance of 0 1 2 3 4 5 6 7 2. Ability to produce 0 1 2 3 4 5 6 7 skin program language 4. Assumes responsi- 0 1 2 3 4 5 6 7 3. Ability to read 0 1 2 3 4 5 6 7 bility for self-care 4. Ability to produce 0 1 2 3 4 5 6 7 5. Performs assigned 0 1 2 3 4 5 6 7 written language interdisciplinary 5. Ability to hear 0 1 2 3 4 5 6 7 activities 6. Ability to compre- 0 1 2 3 4 5 6 7 III. Physical Mobility (PHY) hend and use gestures 1. Performance of 0 1 2 3 4 5 6 7 7. Ability to produce 0 1 2 3 4 5 6 7 transfers speech 2. Performance of am- 0 1 2 3 4 5 6 7 bulation VI. Medications (DRG) 3. Performance of 0 1 2 3 4 5 6 7 1. Knowledge of med- 0 1 2 3 4 5 6 7 wheelchair mobil- ications ity (primary mode) 2. Skill with medica- 0 1 2 3 4 5 6 7 4. Ability to handle 0 1 2 3 4 5 6 7 tions environmental 3. Utilization of medi- 0 1 2 3 4 5 6 7 barriers (e.g., cations stairs, rugs, eleva- tors) VII. Nutrition (NUT) 5. Performance of car 0 1 2 3 4 5 6 7 1. Nutritional 0 1 2 3 4 5 6 7 transfer status—body weight (continued) 105 Exhibit 3.18 (continued) 2. Nutritional 0 1 2 3 4 5 6 7 3. Family understand- 0 1 2 3 4 5 6 7 status—lab values ing of disability 3. Knowledge of nu- 0 1 2 3 4 5 6 7 4. Economic resources 0 1 2 3 4 5 6 7 trition and/or mod- 5. Ability to live inde- 0 1 2 3 4 5 6 7 ified diet pendently 4. Skill with nutrition 0 1 2 3 4 5 6 7 6. Living arrange- 0 1 2 3 4 5 6 7 and diet (adherence ments to nutritional plan) 5. Utilization of nutri- 0 1 2 3 4 5 6 7 XII. Vocational- tion and diet (nutri- Educational Activity tional health) (V/E) 1. Active participation 0 1 2 3 4 5 6 7 VIII. Assistive Devices in realistic voc/ed (DEV) planning 1. Knowledge of as- 0 1 2 3 4 5 6 7 2. Realistic perception 0 1 2 3 4 5 6 7 sistive devices(s) of work-related ac- 2. Skill with assuming 0 1 2 3 4 5 6 7 tivity operating position 3. Ability to tolerate 0 1 2 3 4 5 6 7 of assistive device(s) planned number of 3. Utilization of as- 0 1 2 3 4 5 6 7 hours of voc/ed ac- sistive device(s) tivity/day 4. Vocational/ 0 1 2 3 4 5 6 7 IX. Psychology (PSY) educational place- 1. Distress/comfort 0 1 2 3 4 5 6 7 ment 2. Helplessness/self- 0 1 2 3 4 5 6 7 5. Physical capacity 0 1 2 3 4 5 6 7 efficacy for work 3. Self-directed learn- 0 1 2 3 4 5 6 7 ing skills XIII. Recreation (REC) 4. Skill in self-man- 0 1 2 3 4 5 6 7 1. Participation in 0 1 2 3 4 5 6 7 agement of behav- group activities ior and emotions 2. Participation in 0 1 2 3 4 5 6 7 5. Skill in interper- 0 1 2 3 4 5 6 7 community activi- sonal relations ties X. Neuropsychology (NP) 3. Interaction with 0 1 2 3 4 5 6 7 1. Impairment of 0 1 2 3 4 5 6 7 others short-term memory 4. Participation and 0 1 2 3 4 5 6 7 2. Impairment of 0 1 2 3 4 5 6 7 satisfaction with in- long-term memory dividual leisure ac- 3. Impairment in at- 0 1 2 3 4 5 6 7 tivities tention-concentra- 5. Active participation 0 1 2 3 4 5 6 7 tion skills in sports 4. Impairment in ver- 0 1 2 3 4 5 6 7 bal linguistic pro- XIV. Pain (consensus) (PAI) cessing 1. Pain behavior 0 1 2 3 4 5 6 7 5. Impairment in vi- 0 1 2 3 4 5 6 7 2. Physical inactivity 0 1 2 3 4 5 6 7 sual spatial pro- 3. Social withdrawal 0 1 2 3 4 5 6 7 cessing 4. Pacing 0 1 2 3 4 5 6 7 6. Impairment in basic 0 1 2 3 4 5 6 7 5. Sitting 0 1 2 3 4 5 6 7 intellectual skills 6. Standing tolerance 0 1 2 3 4 5 6 7 7. Walking endurance 0 1 2 3 4 5 6 7 XI. Social Issues (SOC) 1. Ability to problem 0 1 2 3 4 5 6 7 XV. Pulmonary Rehabilita- solve and utilize tion (PUL) resources 1. Knowledge of pul- 0 1 2 3 4 5 6 7 2. Family: communi- 0 1 2 3 4 5 6 7 monary rehabilita- cation/resource tion program 106  Physical Disability and Handicap 107 Exhibit 3.18 2. Skill with pulmo- 0 1 2 3 4 5 6 7 8. Therapy schedule N Y nary rehabilitation set according to program priority 3. Utilization of pul- 0 1 2 3 4 5 6 7 9. Rehab. Med. Clinic N Y monary rehabilita- standard follow-up tion program (1, 3, 6, 12 mo.) XVI. Patient Participation 1 2 XVIII. Specialty Program at Conference (Specify): 1. Attended N Y 1. __________________ 0 1 2 3 4 5 6 7 2. Participated in goal N Y 2. __________________ 0 1 2 3 4 5 6 7 setting 3. __________________ 0 1 2 3 4 5 6 7 3. Family (significant N Y 4. __________________ 0 1 2 3 4 5 6 7 other) attended 5. __________________ 0 1 2 3 4 5 6 7 6. __________________ 0 1 2 3 4 5 6 7 XVII. Preparation Com- 7. __________________ 0 1 2 3 4 5 6 7 pleted for Pass and/or Discharge Summary 1. Recreational Ther- N Y apy pass 2. P.R.N. pass N Y 3. T.L.O.A pass N Y 4. Is this the dis- N Y charge conference? 5. Equipment ordered N Y 6. Type of assistive 0 1 2 3 4 5 6 7 device 7. Phase of device 0 1 2 3 4 5 6 7 development Reproduced from the Patient Evaluation Conference System rating form provided by Dr. Richard F Harvey. With permission. would be required if it is to reach its potential as Defining interval measures. Arch Phys Med a clinical measurement system for rehabilitation Rehabil 1992;73:507–518. settings. (5) Jellinek HM, Torkelson RM, Harvey RF. Functional abilities and distress levels in brain injured patients at long-term follow- up. Arch Phys Med Rehabil References 1982;63:160–162. (1) Harvey RF, Jellinek HM. Functional per- (6) Malec J, Neimeyer R. Psychologic formance assessment: a program approach. prediction of duration of inpatient spinal Arch Phys Med Rehabil 1981;62:456–461. cord injury rehabilitation and performance (2) Harvey RF, Jellinek HM. Patient profiles: of self-care. Arch Phys Med Rehabil utilization in functional performance 1983;64:359–363. assessment. Arch Phys Med Rehabil (7) Rao N, Jellinek HM, Harvey RF, et al. 1983;64:268–271. Computerized tomography head scans as (3) Silverstein B, Kilgore KM, Fisher WP, et al. predictors of rehabilitation outcome. Arch Applying psychometric criteria to functional Phys Med Rehabil 1984;65:18–20. assessment in medical rehabilitation: I. (8) Harvey RF, Silverstein B, Venzon MA, et Exploring unidimensionality. Arch Phys al. Applying psychometric criteria to Med Rehabil 1991;72:631–637. functional assessment in medical (4) Silverstein B, Fisher WP, Kilgore KM, et al. rehabilitation: III. Construct validity and Applying pychometric criteria to functional predicting level of care. Arch Phys Med assessment in medical rehabilitation: II. Rehabil 1992;73:887–892. 108 Measuring Health The Functional Activities were mostly in excess of 0.70; the lowest correla- Questionnaire tion (0.41) was with Raven’s matrices. The high- (Robert I. Pfeffer, 1982, est correlation (0.83) was with a neurologist’s Revised 1984) global rating on a Scale of Functional Capacity designed by Pfeffer (1, Table 3). The validity co- Purpose efficients obtained for the FAQ were consistently This is a screening tool for assessing indepen- higher than those obtained for the Lawton and dence in ADLs designed for community studies Brody scale. The FAQ and the IADL were used in of normal aging and mild senile dementia (1). multiple regression analyses as predictors of mental status and functional assessments; the Conceptual Basis FAQ performed better. The FAQ was found to The scale was intended to cover universal skills correlate 0.76 with a Mental Function Index de- among older adults. Pfeffer et al. followed the veloped by Pfeffer (2), and -0.60 with the Cogni- intuitively appealing concept of a hierarchy of tive Capacity Screening Examination (CCSE) (3, skills proposed by Lawton and Brody and thus Table 4). The FAQ also appears to identify agita- concentrated on higher level skills such as man- tion in patients with dementia: after controlling aging one’s financial affairs and reading, which for scores on the Mini-Mental State Exam they had termed “social functions” (1). (MMSE), FAQ scores differed significantly be- tween those with, and those without, symptoms Description of agitation (4, p17). Karagiozis et al. compared The Functional Activities Questionnaire (FAQ) FAQ scores with direct observations of IADL is not self-administered but is completed by a lay performance. Agreement for cognitively normal informant such as the spouse, a relative, or a subjects was high, but patients with dementia close friend. The original version described in tended to exaggerate their self-reported abilities the literature is shown in Exhibit 3.19; it has ten compared with observational data; such overesti- items concerned with performing daily tasks mation increased with the severity of dementia. necessary for independent living. In 1984, the Informants, by contrast, tended to under-rate the questionnaire was slightly expanded by adding ability of people with dementia (5). four ADL items and an item on initiative; the A review by the U.S. Preventive Services Task first ten items are the same in both versions. For Force on screening for dementia concluded that each activity, four levels ranging from depend- the FAQ is as sensitive as the MMSE; they re- ence (scored 3) to independence (scored 0) are ported sensitivity and specificity figures of 90% specified. For activities not normally undertaken for each scale (6). A Chilean study found that a by the person, the informant must specify combination of the FAQ and the MMSE was whether the person would be unable to under- ideal; the FAQ mainly contributed to specificity. take the task if required (scored 1) or could do Sensitivity for the MMSE was 93.6% at a speci- so if required (0). The total score is the sum of ficity of 46%; figures for the FAQ (at a cutting- individual item scores; higher scores reflect point of 5/6) were 89% and 71%. The greater dependency. combination of MMSE plus FAQ provided a sensitivity of 94% at a specificity of 83% (7, Reliability Table 3). Comparing the FAQ with diagnoses The item-total correlations for all items ex- made by attending neurologists, sensitivity was ceeded 0.80 (1, Table 4). 85% at a specificity of 81% (1). In screening for vascular dementia, sensitivity was 92% and Validity specificity 87%; equivalent results for the Cogni- In a study of 158 elderly people living in the tive Capacity Screening Examination were 85% community, ratings on the FAQ correlated 0.72 sensitivity and 87% specificity (3, Tables 2 and with Lawton and Brody’s IADL scale (1, Table 3). Scores on the FAQ in a longitudinal study re- 3). Correlations with mental functioning tests flected both clinical judgments of change and Exhibit 3.19 The Functional Activities Questionnaire Activities questionnaire to be completed by spouse, child, close friend or relative of the participant. Instructions: The following pages list ten common activities. For each activity, please read all choices, then choose the one statement which best describes the current ability of the participant. Answers should apply to that person’s abilities, not your own. Please check off a choice for each activity; do not skip any. 1. Writing checks, paying bills, balancing checkbook, keeping financial records _____ A. Someone has recently taken over this activity completely or almost completely. _____ B. Requires frequent advice or assistance from others (e.g., relatives, friends, business associates, banker), which was not previously necessary. _____ C. Does without any advice or assistance, but more difficult than used to be or less good job. _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to start now. _____ F. Didn’t do regularly but can do normally now with a little practice if they have to. 2. Making out insurance or Social Security forms, handling business affairs or papers, assembling tax records _____ A. Someone has recently taken over this activity completely or almost completely, and that someone did not used to do any or as much. _____ B. Requires more frequent advice or more assistance from others than in the past. _____ C. Does without any more advice or assistance than used to, but finds more difficult or does less good job than in the past. _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to start now, even with practice. _____ F. Didn’t do routinely, but can do normally now should they have to. 3. Shopping alone for clothes, household necessities and groceries _____ A. Someone has recently taken over this activity completely or almost completely. _____ B. Requires frequent advice or assistance from others. _____ C. Does without advice or assistance, but finds more difficult than used to or does less good job. _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to start now. _____ F. Didn’t do routinely but can do normally now should they have to. 4. Playing a game of skill such as bridge, other card games or chess or working on a hobby such as painting, photography, woodwork, stamp collecting _____ A. Hardly ever does now or has great difficulty. _____ B. Requires advice, or others have to make allowances. _____ C. Does without advice, or assistance, but more difficult or less skillful than used to be. _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to start now. _____ F. Didn’t do regularly, but can do normally now should they have to. 5. Heat the water, make a cup of coffee or tea, and turn off the stove _____ A. Someone else has recently taken over this activity completely, or almost completely. _____ B. Requires advice or has frequent problems (for example, burns pots, forgets to turn off stove). _____ C. Does without advice or assistance but occasional problems. _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to start now. _____ F. Didn’t usually, but can do normally now, should they have to. 6. Prepare a balanced meal (e.g., meat, chicken or fish, vegetables, dessert) _____ A. Someone else has recently taken over this activity completely or almost completely. _____ B. Requires frequent advice or has frequent problems (for example, burns pots, forgets how to make a given dish). _____ C. Does without much advice or assistance, but more difficult (for example, switched to TV dinners most of the time because of difficulty). _____ D. Does without any difficulty or advice. _____ E. Never did and would find quite difficult to do now even after a little practice. _____ F. Didn’t do regularly, but can do normally now should they have to. (continued) 109 110 Measuring Health Exhibit 3.19 (continued) 7. Keep track of current events, either in the neighborhood or nationally _____ A. Pays no attention to, or doesn’t remember outside happenings. _____ B. Some idea about major events (for example, comments on presidential election, major events in the news or major sporting events). _____ C. Somewhat less attention to, or knowledge of, current events than formerly. _____ D. As aware of current events as ever was. _____ E. Never paid much attention to current events, and would find quite difficult to start now. _____ F. Never paid much attention, but can do as well as anyone now when they try. 8. Pay attention to, understand, and discuss the plot or theme of a one-hour television program; get something out of a book or magazine _____ A. Doesn’t remember, or seems confused by, what they have watched or read. _____ B. Aware of the general idea, characters, or nature while they watch or read, but may not recall later; may not grasp theme or have opinion about what they saw. _____ C. Less attention, or less memory than before, less likely to catch humor, points which are made quickly, or subtle points. _____ D. Grasps as quickly as ever. _____ E. Never paid much attention to or commented on T.V., never read much and would probably find very difficult to start now. _____ F. Never read or watched T.V. much, but read or watch as much as ever and get as much out of it as ever. 9. Remember appointments, plans, household tasks, car repairs, family occasions (such as birthdays or anniversaries), holidays, medications _____ A. Someone else has recently taken this over. _____ B. Has to be reminded some of the time (more than in the past or more than most people). _____ C. Manages without reminders but has to rely heavily on notes, calendars, schemes. _____ D. Remembers appointments, plans, occasions, etc. as well as they ever did. _____ E. Never had to keep track of appointments, medications or family occasions, and would probably find very difficult to start now. _____ F. Didn’t have to keep track of these things in the past, but can do as well as anyone when they try. 10. Travel out of neighborhood; driving, walking, arranging to take or change buses and trains, planes _____ A. Someone else has taken this over completely or almost completely. _____ B. Can get around in own neighborhood but gets lost out of neighborhood. _____ C. Has more problems getting around than used to (for example, occasionally lost, loss of confidence, can’t find car, etc.) but usually O.K. _____ D. Gets around as well as ever. _____ E. Rarely did much driving or had to get around alone and would find quite difficult to learn bus routes or make similar arrangements now. _____ F. Didn’t have to get around alone much in past, but can do as well as ever when has to. Reproduced from the Functional Activities Questionnaire provided by Dr. Robert I. Pfeffer. With permission. cognitive measures in 54 elderly patients (2, builds. The method differs somewhat from other Table 8). The FAQ also showed significant con- IADL instruments in that the scale levels are de- trasts between normal, depressed, and demented fined primarily in terms of social function rather respondents in a study of 195 respondents (8, than physical capacities. This brings the scale Table 6). close to some of the social health measurements described in Chapter 4. Commentary The FAQ continues to be used in assessing func- tional status in studies of dementia. The validity References results appear good, and apparently superior to (1) Pfeffer RI, Kurosaki TT, Harrah CH, et al. those of Lawton and Brody’s IADL, on which it Measurement of functional activities in  Physical Disability and Handicap 111 older adults in the community. J Gerontol verse) effects of treatment, and medical costs 1982;37:323–329. (1–5). Except for death, these dimensions are di- (2) Pfeffer RI, Kurosaki TT, Chance JM, et al. vided into subdimensions, such as upper and Use of the Mental Function Index in older lower limb problems for the disability dimen- adults: reliability, validity and measurement sion, and physical and psychological problems of change over time. Am J Epidemiol for the discomfort dimension. These sub dimen- 1984;120:922–935. sions are then divided into components, which (3) Hershey LA, Jaffe DF, Greenough PG, et al. Validation of cognitive and functional are further divided into individual question top- assessment instruments in vascular ics (1, Fig. 1; 2, Fig. 1). Fries et al. followed a dementia. Int J Psychiatry Med parsimonious approach in selecting questions, 1987;17:183–192. noting that there may be no need to measure ap- (4) Senanarong V, Cummings JL, Fairbanks L, parently distinct aspects of disability that are et al. Agitation in Alzheimer’s disease is a correlated. This allows an instrument to repre- manifestation of frontal lobe dysfunction. sent a content area without addressing every Dement Geriatr Cogn Disord possible question (4). The hierarchic model ex- 2004;17:14–20. presses results at various levels of generality: (5) Karagiozis H, Gray S, Sacco J, et al. The question scores may be combined to form com- Direct Assessment of Functional Abilities ponent (e.g., eating or dressing) and dimension (DAFA): a comparison to an indirect mea- sure of instrumental activities of daily (e.g., disability) scores (6). However, Fries ar- living. Gerontologist 1998;38:113–121. gued against adding dimension scores because (6) Boustani M, Peterson B, Hanson L, et al. this would involve value judgments of the rela- Screening for dementia in primary care: tive importance of dimensions that may not hold summary of the evidence for the U.S. across patients. Empirically, correlations across Preventive Services Task Force. Ann Intern dimensions are lower than within dimensions, so Med 2003;138:927–937. Fries argued that “[d]isability, discomfort, psy- (7) Quiroga P, Albala C, Klaasen G. chologic outcomes, cost, and death have been [Validation of a screening test for age identified as separable outcomes. The full num- associated cognitive impairment, in Chile]. ber of dimensions seems likely to be between 5 Rev Med Chil 2004;132:467–478. and 8” (6, p701). (8) Pfeffer RI, Kurosaki TT, Harrah CH, et al. A survey diagnostic tool for senile dementia. The HAQ model also considers the economic Am J Epidemiol 1981;114:515–527. costs of disease and the possible side-effects of treatment. A separate dimension considers med- ical and surgical complications (e.g., gastroin- The Health Assessment testinal problems, infection). These are recorded Questionnaire from an audit of hospital records and death cer- (James F. Fries, 1980) tificates; weights have so far been developed for rating several possible side effects (1, p120). The Purpose economic impact of disease is assessed through The Stanford Health Assessment Questionnaire direct (cost of drugs and doctor visits) and indi- (HAQ) measures difficulty in performing ADLs. rect effects such as work loss. Costs can be rated It was originally designed for the clinical assess- using standard computations based on average ment of adult patients with arthritis, but it has costs for various types of disease (1, p121). been used in a wide range of research settings to Although the full HAQ instrument covers the evaluate care. five dimensions mentioned by Fries (6), develop- ment work has concentrated on the disability and Conceptual Basis discomfort dimensions; these are the most com- The HAQ is based on a hierarchical model that monly used and are the two described in detail in considers the effects of a disease in terms of this review. They are referred to as the “Short or death, disability, discomfort, the side (i.e., ad- 2-Page HAQ” in the Stanford University HAQ 112 Measuring Health documentation (aramis.stanford.edu/downloads/ The discomfort dimension of the HAQ in- HAQ%20Instructions.pdf ). The full instrument cludes a single question on physical pain in the is available from the Aramis Web site, which will past week. It uses a 15-cm visual analogue pain include occasional updates. scale, with the end-points labeled “No pain” and “Very severe pain.” Scores are measured in cm Description from the left and are multiplied by 0.2 to give a The disability dimension of the HAQ includes range from 0 to 3; scores are rounded to two 20 questions on daily functioning during the decimal places. Fries and Spitz noted: “Attempts past week. These cover eight component areas: to elaborate pain activity by part of the body in- dressing and grooming, arising, eating, walking, volved, times during the day which were painful, hygiene, reach, grip, and outdoor activities. Ear- and severity of pain in different body parts failed lier versions also included sexual activity. Each to yield indexes that outperformed a simple ana- component includes two or three questions log scale” (3, p31). The HAQ also includes a drawn from previous measures (2, p138); a de- global health analogue scale, which is a 15-cm scription of the development of the HAQ is given horizontal visual analogue scale that runs from by Fries et al. (4). The scale may be self- “very well” to “very poor.” The HAQ is consid- administered, or it may be applied in a telephone ered to be in the public domain, but permission or personal interview (3, p30). It can be com- must be obtained to use it; this is intended to en- pleted in five to eight minutes, and scored in less sure standardization of the instrument. than one (4). Wolfe et al. found that where pa- The HAQ has been very widely used, and tients had previously completed the HAQ, 88% considerable evidence for reliability and validity completed it in less than three minutes; it took has accumulated. Reviews by Ramey and Fries 15 to 22 seconds to score (7, p1485). The ques- and by Bruce and Fries, summarize evidence tions are shown in Exhibit 3.20. from more than 200 articles (1; 5). Each response is scored on a four-point scale of ability patterned after the American Rheuma- Reliability tism Association functional classification (3, Fries compared interview and self-administered p31). The response scales range from “without versions of the disability scale (N = 20). The any difficulty” to “unable to do” and a check-list Spearman correlation for the disability index records any aids used or assistance received. The was 0.85, whereas correlations for individual highest score in each of the eight components is sections ranged from 0.56 (IADL activities and added to form a total (range, 0–24); this is di- hygiene) to 0.85 (eating) (2, Table 1). vided by 8 to provide a 0 to 3 continuous score, During the development of the HAQ, Fries termed the Functional Disability Index (7). abbreviated the questionnaire and removed Scoring instructions are given on the Aramis questions that correlated highly with others in web site (aramis.stanford.edu/downloads/ the scale. Not surprisingly, therefore, item-total HAQ%20Instructions.pdf ). Siegert et al. sug- correlations are modest, ranging from 0.51 to gested the following interpretations of overall 0.81 (2, Table 3). Pincus et al. reported some- scores: “0.0–0.5: the patient is completely self- what higher alpha coefficients for the questions sufficient . . . 0.5–1.25: the patient is reasonably in each category, ranging from 0.71 (reaching) self-sufficient and experiences some minor and to 0.89 (eating) (10, Table 2). Milligan et al. even major difficulties in performing ADL; found an alpha coefficient of 0.94 for the com- 1.25–2.0: the patient is still self-sufficient but plete instrument, with maximum inter-item cor- has many major problems with ADL; 2.0–3.0: relations of 0.75 (11). the patient may be called severely handicapped” Two-week test-retest reliability of the disabil- (8). Tennant et al. used Rasch analysis to exam- ity section was investigated with 37 patients ine the scale characteristics of the original scor- with rheumatoid arthritis, showing no signifi- ing approach, showing that it does not possess cant difference by t-test and a Spearman correla- interval-scale qualities (9). tion of 0.87 (3, p31). Goeppinger et al. reported Exhibit 3.20 The Health Assessment Questionnaire Please tell us how your arthritis affects your ability to carry out your daily activities. Please place an ‘x’ in the box ( ) that best describes your usual abilities OVER THE PAST WEEK: Without ANY With SOME With MUCH UNABLE Difficulty Difficulty Difficulty To Do DRESSING & GROOMING Are you able to: —Dress yourself, including shoelaces and buttons? —Shampoo your hair? ARISING Are you able to: —Stand up from a straight chair? —Get in and out of bed? EATING Are you able to: —Cut your meat? —Lift a full cup or glass to your mouth? —Open a new milk carton? WALKING Are you able to: —Walk outdoors on flat ground? —Climb up five steps? • Please check any AIDS OR DEVICES that you usually use for any of the above activities: Cane Devices Used for Dressing (button hook, zipper pull, etc.) Walker Built Up or Special Utensils Crutches Special or Built Up Chair Wheelchair • Please check any categories for which you usually need HELP FROM ANOTHER PERSON: Dressing & Grooming Eating Arising Walking • Please place an ‘x’ ( ) in the box which best describes your usual abilities OVER THE PAST WEEK: Without ANY With SOME With MUCH UNABLE Difficulty Difficulty Difficulty To Do HYGIENE Are you able to: —Wash and dry your body? —Take a tub bath? —Get on and off the toilet? (continued) 113 Exhibit 3.20 (continued) Without ANY With SOME With MUCH UNABLE Difficulty Difficulty Difficulty To Do REACH Are you able to: —Reach and get down a 5 pound object (such as a bag of sugar) from above your head? —Bend down to pick up clothing from the floor? GRIP Are you able to: —Open car doors? —Open previously opened jars? —Turn faucets on and off? ACTIVITIES Are you able to: —Run errands and shop? —Get in and out of a car? —Do chores such as vacuuming or yardwork? • Please check any AIDS OR DEVICES that you usually use for any of these activities: Raised Toilet Seat Bathtub Bar Bathtub Seat Long-Handled Appliances for Reach Jar Opener (for jars Long-Handled Appliances in Bathroom previously opened) • Please check any categories for which you usually need HELP FROM ANOTHER PERSON: Hygiene Gripping and Opening Things Reach Errands and Chores We are also interested in learning whether or not you are affected by pain because of your illness. • How much pain have you had because of your arthritis IN THE PAST WEEK? PLACE A SINGLE VERTICAL MARK THROUGH THE LINE TO INDICATE THE SEVERITY OF THE PAIN VERY NO PAIN SEVERE PAIN 0 100 • In general, how would you rate your current health? PLACE A SINGLE VERTICAL MARK THROUGH THE LINE TO INDICATE YOUR CURRENT HEALTH. VERY WELL VERY POOR Reproduced from the Stanford Arthritis Center Disability and Discomfort Scales, 1981, with format changes from the Aramis web site. With permission. 114  Physical Disability and Handicap 115 a one-week test-retest reliability of 0.95 (N = 30 several studies that have used it as an outcome rheumatoid arthritis patients) and 0.93 (N = 30 measure in randomized trials (1, Tables 6 and 8). osteoarthritis patients) (12). Fries et al. adminis- Brown et al. compared the HAQ with the tered the HAQ on successive occasions and ob- Arthritis Impact Measurement Scales (AIMS). tained a re-test correlation of 0.98 after 6 The correlations were 0.91 for the disability di- months (4, p791). mension and 0.64 for the pain questions (14, p160). The two scales correlated 0.89 in another Validity study (16). Liang et al. compared responses of Principal component analyses have broadly con- 50 patients with arthritis on the HAQ, the Sick- firmed the dimensions originally postulated by ness Impact Profile (SIP), the Functional Status Fries: one main factor underlay 15 of the disabil- Index (FSI), the AIMS, and the Quality of Well- ity questions (2, Table 4). The eight disability Being Scale (QWB). The overall score on the subscales are substantially correlated with each HAQ correlated 0.84 with the AIMS, 0.78 with other: a median correlation of 0.44 among them the SIP, 0.75 with the FSI, and 0.60 with the has been reported (13, p948). Brown et al. tested QWB (17, Table 4). For the mobility scale, cor- the factorial structure of the HAQ, showing a relations of the HAQ with the other instruments two-factor solution with the eight disability were lower than the correlations among the components loading on the first factor and the other four scales. pain scale on the second in a small study of 48 Liang et al. compared the relative efficiency patients with rheumatoid arthritis (14). Milligan of five measures, indicating their ability to iden- et al. obtained one factor relating to movements tify intra-subject change before and after hip or involving the large limbs (rising, walking) and a knee surgery. The rank order of five measures in second for fine movements such as grasping and terms of this statistic placed the HAQ in fourth eating (11). place (17, Table 5). They subsequently replicated Fries et al. compared self-administered HAQ broadly similar findings: the overall and mobility responses to observations of performance made scores on the HAQ would require larger sample during a home visit (N = 25). The Spearman cor- sizes to demonstrate a significant effect of treat- relation for the overall score was 0.88, whereas ment than equivalent scores from the AIMS or correlations for component scores ranged from the SIP would (18, Table 3). However, the HAQ 0.47 (arising) to 0.88 (walking) (2, Table 2). may be more sensitive to change than physical Fitzpatrick et al. compared the HAQ to indica- measures such as ESR, grip strength, or morning tors of disease activity in 105 patients with stiffness. Hawley and Wolfe found the HAQ to arthritis. Correlations of the overall score were be more responsive than physical measures or highest with grip strength (−0.73) and with the depression following methotrexate treatment for Ritchie articular index (0.69). The HAQ overall rheumatoid arthritis; the HAQ pain score was score correlated 0.38 with erythrocyte sedimen- especially responsive (19, p133). The HAQ also tation rate (ESR) and 0.41 with a rating of reflected the progressive nature of the condition morning stiffness (15, Table 2). Wolfe et al. also better than did physical indicators: at five- and showed significant associations between the ten-year follow-up assessments, the HAQ HAQ disability score and joint count, grip showed large declines in function (effect sizes of strength, anxiety and depression, and ESR (7). −1.6 and −2.4, respectively). In a study of pa- They demonstrated the validity of the HAQ in tients with polymyalgia rheumatica, the HAQ predicting health services utilization, clinical gave a standardized response mean of 3.0, com- progression, and mortality. For predicting mor- pared with 1.7 for morning stiffness, 1.8 for a tality, the relative risk associated with a one- visual analogue scale measure of pain, and 1.6 point increase in baseline disability score was for C-reactive protein (20, Table 1). Fitzpatrick 1.77 (7, p1484). Ramey et al. listed several et al. found sensitivity to improvement in disease dozen studies that have compared the HAQ with state over 15 months to be modest, at 65% clinical and laboratory variables; they also cited (specificity, 61%), whereas sensitivity to deterio- 116 Measuring Health ration was 60% (specificity 73%) (15). Hawley this version has also been modified for use in and Wolfe’s findings may suggest, however, that Costa Rica (28). A version for juvenile arthritics the problem lies not with the HAQ so much as showed an alpha of 0.94, and a correlation of with the lack of sensitivity of the traditional cri- 0.67 with number of affected joints (29). terion standard measures. In a study of arthritis The “AIDS-HAQ” includes 14 items from patients, the AIMS2 physical function score pro- the Medical Outcomes Study instruments and vided slightly greater sensitivity to change than 16 items from the HAQ. The items cover physi- the modified HAQ (21, Table 2). cal function, mental health, cognitive function, energy levels, and general health (30; 31, p94; Alternative Forms 32). An instrument called the Fibromyalgia Pincus et al. abbreviated the HAQ by retaining HAQ was developed using a subset of eight only one question in each of the eight disability HAQ items using Rasch analysis (33). components; they also added questions on satis- The HAQ has been adapted for use in many faction and change in activities. This has been countries; the MAPI Institute offers translations called the Modified HAQ (MHAQ). The test- into more than two dozen languages, and a sum- retest reliability at one month was higher for the mary is given by Ramey et al. (34, Table 2). revised version (0.91) than for the original HAQ Translations into about 50 languages are also (0.78) (10, p1350); item-total correlations listed, with references, in the article by Bruce ranged from 0.52 to 0.74 (22, Table 2). A fur- and Fries (5, p172). Published translations in- ther test of the eight-item HAQ showed correla- clude those for Great Britain (35), Sweden (36), tions of −0.53 with grip strength, 0.44 with Spanish-speaking countries (37–41), Germany walking time, and 0.60 with the American (42), the Netherlands (8; 43; 44) and China (45). Rheumatism Association functional class (23, French translations are available from France Table 1). Correlations with joint tenderness (46; 47) and Canada (48). The one-week test- (0.57) and joint swelling (0.33) were reported in retest reliability of the Swedish version was 0.91, a clinical trial (22, p1913), and convergent cor- and results correlated 0.76 with observational relations ranged from −0.46 to −0.61 for compa- ratings of the patients carrying out the activities rable items on dressing, walking, and bending in included in the scale (36, p267). Test-retest reli- the SF-36 (22, p1912). Callahan and Pincus also ability of the Italian version ranged from 0.81 to evaluated the ratio of the pain score to the eight- 0.99 in several centers (49). item disability score as an approach to distin- guishing early rheumatoid arthritis and other Commentary diffuse musculoskeletal pain (24). The HAQ has become the most widely used in- Ziebland et al. also proposed “transition ques- strument in a field that pays close attention to tions” such as “Compared with three months rigor in measurement. It has been included in the ago, how difficult is it now (this week) to . . . American Rheumatism Association Medical In- [Dress yourself, Get in and out of bed . . .]” (25, formation System, and the National Health and Table 1). Subsequent testing showed these ques- Nutrition Survey in the U.S. (3, p30). Three re- tions to be more sensitive to detecting change in view articles have described the instrument and status in patients with rheumatoid arthritis than provide citations to hundreds of references to it the overall HAQ scores (25). (1; 5; 34). The design of the HAQ offers a scale A modified version of the HAQ has been pro- that is broad in scope yet brief enough to be posed for patients with spondylitis. This adds completed by patients while waiting to see their five questions covering handicaps arising from physician. The available evidence shows the health problems of the spine and back (13). A HAQ to have strong reliability and validity. A children’s version of the HAQ has also been pro- further strength lies in the continued involve- posed, the Childhood HAQ (1, p122). This has ment of the originator of the scale in coordinat- been translated into Norwegian (26); validity re- ing its development; this has helped to control sults for a Spanish version are available (27), and the proliferation of different versions that typi-  Physical Disability and Handicap 117 fies other scales. The reader of the review by status and review. Arthritis Care Res Ramey et al., for example, gains the impression 1992;5:119–129. of a well-planned development effort (1). (2) Fries JF, Spitz P, Kraines RG, et al. Mea- In terms of improvements to the HAQ, it surement of patient outcome in arthritis. would be valuable to see a fuller exploitation of Arthritis Rheum 1980;23:137–145. (3) Fries JF, Spitz PW. The hierarchy of patient the large studies in which the method has been outcomes. In: Spilker B, ed. Quality of life used, for example to provide population refer- assessments in clinical trials. New York: ence standards for people with a range of disabil- Raven Press, 1990:25–35. ities; we also have little information on its (4) Fries JF, Spitz PW, Young DY. The adequacy in patients with diseases other than dimensions of health outcomes: the Health arthritis. Assessment Questionnaire, disability and The criticisms of the HAQ have focused on pain scales. J Rheumatol 1982;9:789–793. its scoring, which was designed for simplicity, (5) Bruce B, Fries JF. The Stanford Health which may have been achieved at the cost of re- Assessment Questionnaire: a review of its duced precision. By counting only the highest history, issues, progress, and score in each section, the HAQ summarizes the documentation. J Rheumatol 2002;30:167–178. patient’s major difficulty but does not use all the (6) Fries JF. Toward an understanding of information collected. In comparing scores over patient outcome measurement. Arthritis time, therefore, improvements in less severely af- Rheum 1983;26:697–704. fected areas of functioning may be missed, (7) Wolfe F, Kleinheksel SM, Cathey MA, et which may account for the high test-retest relia- al. The clinical value of the Stanford bility of the HAQ, combined with its compara- Health Assessment Questionnaire tive insensitivity to measuring change (50). functional disability index in patients with Liang et al. concluded that “the HAQ and Index rheumatoid arthritis. J Rheumatol of Well-Being are judged to be poor candidates 1988;15:1480–1488. for use when mobility change is a major func- (8) Siegert CEH, Vleming L-J, van den tional outcome” (17, p547). Certainly, it seems Broucke JP, et al. Measurement of disability in Dutch rheumatoid arthritis curious to ask questions that are not incorpo- patients. Clin Rheumatol 1984;3:305–309. rated into the scoring system; to include all ques- (9) Tennant A, Hillman M, Fear J, et al. Are tions would increase sensitivity. The study by we making the most of the Stanford Health Ziebland et al. raised the possibility that asking Assessment Questionnaire? Br J Rheumatol patients to rate their own progress may form a 1996;35:574–578. valuable adjunct to repeated administration of (10) Pincus T, Summey JA, Soraci SA, Jr., et al. the basic HAQ (25). We conclude that the scale Assessment of patient satisfaction in is a good descriptive instrument but may be less activities of daily living using a modified appropriate as a tool for measuring clinical Stanford Health Assessment Questionnaire. change in outcome studies. Arthritis Rheum 1983;26:1346–1353. (11) Milligan SE, Hom DL, Ballou SP, et al. An Address assessment of the Health Assessment Questionnaire functional ability index Stanford University School of Medicine has pro- among women with systemic lupus duced an informative web site describing the erythematosus. J Rheumatol HAQ at aramis.stanford.edu/HAQ.html. The 1993;20:972–976. site provides copies of the HAQ and administra- (12) Goeppinger J, Doyle M, Murdock B. Self- tion instructions. administered function measures: the impossible dream? Arthritis Rheum 1985;28(suppl):145. References (13) Daltroy LH, Larson MG, Roberts WN, et (1) Ramey DR, Raynauld J-P, Fries JF. The al. A modification of the Health Health Assessment Questionnaire 1992: Assessment Questionnaire for the 118 Measuring Health spondyloarthropathies. J Rheumatol rheumatoid arthritis compared with 1990;17:946–950. traditional physical, radiographic, and (14) Brown JH, Kazis LE, Spitz PW, et al. The laboratory measures. Ann Intern Med dimensions of health outcomes: a cross- 1989;110:259–266. validated examination of health status mea- (24) Callahan LF, Pincus T. A clue from a self- surement. Am J Public Health report questionnaire to distinguish 1984;74:159–161. rheumatoid arthritis from noninflammatory (15) Fitzpatrick R, Newman S, Lamb R, et al. A diffuse musculoskeletal pain. Arthritis comparison of measures of health status in Rheum 1990;33:1317–1322. rheumatoid arthritis. Br J Rheumatol (25) Ziebland S, Fitzpatrick R, Jenkinson C, et 1989;28:201–206. al. Comparison of two approaches to mea- (16) Hakala M, Nieminen P, Manelius J. Joint suring change in health status in impairment is strongly correlated with rheumatoid arthritis: the Health Assessment disability measured by self-report Questionnaire (HAQ) and modified HAQ. questionnaires. Functional status Ann Rheum Dis 1992;51:1202–1205. assessment of individuals with rheumatoid (26) Flato B, Sorskaar D, Vinje O, et al. Mea- arthritis in a population based series. J suring disability in early juvenile Rheumatol 1994;21:64–69. rheumatoid arthritis: evaluation of a (17) Liang MH, Larson MG, Cullen KE, et al. Norwegian version of the childhood Health Comparative measurement efficiency and Assessment Questionnaire. J Rheumatol sensitivity of five health status instruments 1998;25:1851–1858. for arthritis research. Arthritis Rheum (27) Goycochea-Robles MV, Garduno-Espinosa 1985;28:542–547. J, Vilchis-Guizar E, et al. Validation of a (18) Liang MH, Fossel AH, Larson MG. Spanish version of the Childhood Health Comparisons of five health status Assessment Questionnaire. J Rheumatol instruments for orthopedic evaluation. Med 1997;24:2242–2245. Care 1990;28:632–642. (28) Arguedas O, Andersson-Gare B, Fasth A, et (19) Hawley DJ, Wolfe F. Sensitivity to change al. Development of a Costa Rican version of the Health Assessment Questionnaire of the Childhood Health Assessment (HAQ) and other clinical and health status Questionnaire. J Rheumatol measures in rheumatoid arthritis: results of 1997;24:2233–2241. short-term clinical trials and observational (29) Singh G, Athreya BH, Fries JF, et al. Mea- studies versus long-term observational surement of health status in children with studies. Arthritis Care Res juvenile rheumatoid arthritis. Arthritis 1992;5:130–136. Rheum 1994;37:1761–1769. (20) Kalke S, Mukerjee D, Dasgupta B. A study (30) Lubeck DP, Fries JF. Changes in quality of of the Health Assessment Questionnaire to life among persons with HIV infection. evaluate functional status in polymyalgia Qual Life Res 1992;1:359–366. rheumatica. Rheumatology (31) Hays RD, Shapiro MF. An overview of 2000;39:883–885. generic health-related quality of life mea- (21) Taal E, Rasker JJ, Riemsma RP. Sensitivity sures for HIV research. Qual Life Res to change of AIMS2 and AIMS2-SF 1992;1:91–97. components in comparison to M-HAQ and (32) Skevington SM, O’Connell KA. Measuring VAS-pain. Ann Rheum Dis quality of life in HIV and AIDS: a review 2004;63:1655–1658. of the recent literature. Psychol Health (22) Tuttleman M, Pillemer SR, Tilley BC, et al. 2003;18:331–350. A cross sectional assessment of health (33) Wolfe F, Hawley DJ, Goldenberg DL, et al. status instruments in patients with The assessment of functional impairment rheumatoid arthritis participating in a in fibromyalgia (FM): Rasch analyses of 5 clinical trial. J Rheumatol functional scales and the development of 1997;24:1910–1915. the FM Health Assessment Questionnaire. (23) Pincus T, Callahan LF, Brooks RH, et al. J Rheumatol 2000;27:1989–1999. Self-report questionnaire scores in (34) Ramey DR, Fries JF, Singh G. The Health  Physical Disability and Handicap 119 Assessment Questionnaire 1995 - status distinct aspects of physical function in and review. In: Spilker B, ed. Quality of life rheumatoid arthritis. Ann Rheum Dis and pharmacoeconomics in clinical trials. 1993;52:497–502. Philadelphia: Lippincott-Raven, (45) Koh ET, Seow A, Pong LY, et al. Cross 1996:227–237. cultural adaptation and validation of the (35) Kirwan JR, Reeback JS. Stanford Health Chinese Health Assessment Questionnaire Assessment Questionnaire modified to for use in rheumatoid arthritis. J assess disability in British patients with Rheumatol 1998;25:1705–1708. rheumatoid arthritis. Br J Rheumatol (46) Guillemin F, Brainéon S, Pourel J. Mea- 1986;25:206–209. surement of the functional capacity in (36) Ekdahl C, Eberhardt K, Andersson SI, et rheumatoid polyarthritis: a French al. Assessing disability in patients with adaptation of the Health Assessment rheumatoid arthritis. Scand J Rheumatol Questionnaire (HAQ). Rev Rhum Mal 1988;17:263–271. Osteoartic 1991;58:459–465. (37) Bosi-Ferraz M, Oliveira LM, Araujo PMP, (47) Guillemin F, Briançon S, Pourel J. Validity et al. Cross-cultural reliability of the and discriminant ability of a French version physical ability dimension of the Health of the Health Assessment Questionnaire in Assessment Questionnaire. J Rheumatol early RA. Disabil Rehabil 1992;14:71–77. 1990;17:813–817. (48) Raynauld J-P, Singh G, Shiroky JB, et al. A (38) Cardiel MH, Abello-Banfi M, Ruiz- French-Canadian version of the Health Mercado R, et al. Quality of life in Assessment Questionnaire. Arthritis rheumatoid arthritis: validation of a Rheum 1992;35(suppl 9):S125. Spanish version of the disability index of (49) Ranza R, Marchesoni A, Calori G, Bianchi the Health Assessment Questionnaire. G, et al. The Italian version of the Arthritis Rheum 1991;34(suppl 9):S183. Functional Disability Index of the Health (39) Perez ER, MacKenzie CR, Ryan C. Assessment Questionnaire. A reliable Development of a Spanish version of the instrument for multicenter studies on Modified Health Assessment Questionnaire. rheumatoid arthritis. Clin Exp Rheumatol Arthritis Rheum 1990;33(suppl 9):S100. 1993;11:123–128. (40) Esteve-Vives J, Batelle-Gualda E, Reig A. (50) Gardiner PV, Sykes HR, Hassey GA, et al. Spanish version of the Health Assessment An evaluation of the Health Assessment Questionnaire: reliability, validity and Questionnaire in long-term longitudinal transcultural equivalency. J Rheumatol follow-up of disability in rheumatoid 1993;20:2116–2122. arthritis. Br J Rheumatol (41) Gonzalez VM, Stewart A, Ritter PL, et al. 1993;32:724–728. Translation and validation of arthritis outcome measures into Spanish. Arthritis Rheum 1995;38:1429–1446. The MOS Physical (42) Bruhlmann P, Stucki G, Michel BA. Functioning Measure Evaluation of a German version of the (Anita Stewart, 1992) physical dimensions of the Health Assessment Questionnaire in patients with Purpose rheumatoid arthritis. J Rheumatol The Medical Outcomes Study (MOS) measure- 1994;21:1245–1249. ment of physical functioning offers an extended (43) van der Heijde DM, van Riel PLCM, van ADL scale sensitive to variations at relatively de Putte LBA. Sensitivity of a Dutch Health high levels of physical function. It is suitable for Assessment Questionnaire in a trial use in health surveys and in outcome assessment comparing hydroxychloroquine vs for outpatient care. sulphasalazine. Scand J Rheumatol 1990;19:407–412. (44) van der Heide A, Jacobs JW, van Albada- Conceptual Basis Kuipers GA, et al. Self report functional As part of the comprehensive measurement bat- disability scores and the use of devices: two tery designed for the MOS, several considera- 120 Measuring Health tions guided the design of the physical func- responses from items 3 and 4 only. A missing tioning scale. First, an attempt was made to in- score is given if either question was not an- clude activities that reflect physical disabilities swered (1, p97). For other analyses, the item on rather than social roles. Stewart argued that use of transportation is dichotomized so that ADLs (e.g., shopping, cleaning house) reflect a 0 = unable to use transport for health reasons, blend of physical functioning and social roles, and 1 = all other replies (1, pp93–94). so there may be reasons other than physical limitations why some respondents do not cook Reliability or clean house. The MOS team developed sepa- Eight of ten physical function items correlated rate scales for physical function and role per- 0.70 or greater with the overall physical scale formance (1; 2). A second issue concerned the score; the vigorous activity item correlated 0.62 level of disability implicit in the questions. and the bathing or dressing item showed a lower Most ADL questions reflect relatively severe correlation of 0.48 (1, Table 6-3). Internal con- disabilities and are insensitive to variations at sistency for the functioning score was 0.92; for higher levels of functioning, where most people the mobility scale it was 0.71 (1, p98). score. For use with relatively healthy patients, The alpha internal consistency of a slightly the MOS instrument included items on more modified version of the scale was 0.92 in a sam- strenuous physical activities while still covering ple of 1,054 elderly respondents; intraclass test- basic ADLs such as dressing and walking. Fi- retest reliability was 0.93 on a subset of 52 (3, nally, the MOS team argued that people’s dif- Table 4). fering values for functional ability should be recognized: some people may not wish to per- Validity form certain activities (e.g., running). Accord- The physical functioning scale scores correlated ingly, the MOS instrument includes a question 0.58 with the mobility scores and 0.63 with the on satisfaction with performance, which was satisfaction scores (1, Table 6-6). A factor analy- expected to be somewhat independent of the sis identified a single factor accounting for 70% level of functioning (1, p89). of the variance. The instrument described here is an extension of the six-item physical functioning scale in- Alternative Forms cluded in the Short-Form-20 Health Survey. Pilot The same ten physical functioning items appear studies suggested that a longer battery would in the SF-36 instrument reviewed in Chapter 10. have higher sensitivity in detecting disabilities (1, p90). Commentary Overlapping with the content of the SF-36, this Description brief instrument offers a well-established set of The MOS Physical Functioning Measure in- ADL and mobility items. It seeks to provide a cludes ten items on functioning, one on satisfac- relatively pure measure of functional ability, in- tion with physical activity, and three on mobility dependent of role functioning, which was cov- (see Exhibit 3.21). ered in a separate MOS instrument (2). The Three scores are derived. A physical function physical functioning measure has 21 scale levels, score is formed by averaging non missing items all of which were represented in preliminary from question 1; the score is transformed to a 0 testing (1, p100). The inclusion of an item cover- to 100 scale in which a higher score indicates ing satisfaction with function is innovative, and better function. People omitting more than five serves to extend the scope of the functioning items receive a missing score. A satisfaction items by identifying people who report no dis- score is based on item 2, transformed to a 0 to ability on the items listed but are still dissatis- 100 scale. Stewart and Kamberg tested several fied. Stewart and Kamberg also found the reverse approaches to scoring the mobility items 3 to 5 to be significant and noted that 31% of those re- and found that the best approach was to sum the porting some level of physical disability nonethe- Exhibit 3.21 The Medical Outcomes Study Physical Functioning Measure 1. The following items are activities you might do during a typical day. Does your health limit you in these activities? (Circle One Number on Each Line) Yes, Yes, No, not limited a limited a limited ACTIVITIES lot little at all a. Vigorous activities, such as running, lifting heavy objects, participating in strenuous sports . . . . . . . . . . . . . . . . . . . . . . 1 2 3 b. Moderate activities, such as moving a table, pushing a vacuum cleaner, bowling, or playing golf . . . . . . . . . . . . . . 1 2 3 c. Lifting or carrying groceries . . . . . . . . . . 1 2 3 d. Climbing several flights of stairs . . . . . . 1 2 3 e. Climbing one flight of stairs . . . . . . . . . . 1 2 3 f. Bending, kneeling or stooping . . . . . . . . . 1 2 3 g. Walking more than one mile . . . . . . . . . . 1 2 3 h. Walking several blocks . . . . . . . . . . . . . . . 1 2 3 i. Walking one block . . . . . . . . . . . . . . . . . . . . 1 2 3 j. Bathing or dressing yourself . . . . . . . . . . . 1 2 3 2. How satisfied are you with your physical ability to do what you want to do? (Circle One) Completely satisfied . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Very satisfied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Somewhat satisfied. . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Somewhat dissatisfied . . . . . . . . . . . . . . . . . . . . . . . . 4 Very dissatisfied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Completely dissatisfied. . . . . . . . . . . . . . . . . . . . . . . . 6 3. When you travel around your community, does someone have to assist you because of your health? (Circle One) Yes, all of the time . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Yes, most of the time . . . . . . . . . . . . . . . . . . . . . . . . 2 Yes, some of the time . . . . . . . . . . . . . . . . . . . . . . . . 3 Yes, a little of the time . . . . . . . . . . . . . . . . . . . . . . . 4 No, none of the time . . . . . . . . . . . . . . . . . . . . . . . . 5 4. Are you in bed or in a chair most or all of the day because of your health? (Circle One) Yes, every day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Yes, most days . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Yes, some days . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Yes, occasionally . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 No, never . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 5. Are you able to use public transportation? (Circle One) No, because of my health . . . . . . . . . . . . . . . . . . . . . 1 No, for some other reason . . . . . . . . . . . . . . . . . . . . 2 Yes, able to use public transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 From Stewart AL, Ware JE Jr. Measuring functioning and well-being: the Medical Outcomes Study approach. Durham, North Carolina: Duke University Press, 1992:375–376. With permission. 121 122 Measuring Health less said they were very or completely satisfied: person may have, and the resources (social or their level of functioning appeared to allow them material) they have to manage their condition (4; to do what they wanted to do (1, p101). 5). Providing support to maintain an elderly per- This measure should be considered for use in son at home illustrates the use of social resources relatively healthy populations, such as those see- to offset disability; institutional admission typi- ing primary care doctors and those included in cally results from an imbalance between disabili- social surveys. ties and compensating social resources (4, p1039). The SMAF therefore extends the com- mon medical concept of disability to include cul- References tural or personal choices that limit autonomy. (1) Stewart AL, Kamberg CJ. Physical Hébert illustrates: “a man who cannot perform functioning measures. In: Stewart AL, domestic tasks, regardless of the reason, is dis- Ware JE, Jr., eds. Measuring functioning abled and must rely on his social resources, usu- and well-being: the Medical Outcomes ally represented by his wife, to compensate for Study approach. Durham, North Carolina: the disabilities. These social and cultural disabil- Duke University Press, 1992:86–101. ities are real, since with the loss of the resource, (2) Sherbourne CD, Stewart AL, Wells KB. the handicaps generated are often sufficient to Role functioning measures. In: Stewart AL, Ware JE, Jr., eds. Measuring functioning justify admission to an institution.” (4, p1040) and well-being: the Medical Outcomes By also assessing the stability of a person’s re- Study approach. Durham, North Carolina: sources for support, the SMAF conveys a prog- Duke University Press, 1992:205–219. nostic dimension that extends the notion of (3) Raina P, Bonnett B, Waltner-Toews D, et handicap to indicate a person’s autonomy, or so- al. How reliable are selected scales from cial vulnerability. population-based health surveys? An analysis among seniors. Can J Public Description Health 1999;90:60–64. The SMAF (the acronym comes from the French “Système de mesure de l’autonomie fonction- nelle”) is a 29-item rating scale applied by a The Functional Autonomy physician, nurse, or social worker (6). It records Measurement System (SMAF) functional disabilities and the available material (Réjean Hébert, 1984, and social resources that may compensate for the revised 1993, 2001) disabilities. It assesses ADLs (7 items), mobility Purpose (6 items), communication (3 items), mental The SMAF is a clinical rating scale that measures function (5 items), and IADLs (8 items). Scale the functional autonomy of elderly patients. It is content was based on that of previous instru- used to make routine assessments, to guide deci- ments and on the 1980 WHO classification of sions about allocating home care, or in deciding impairments, disabilities, and handicaps (1, on institutional admission (1, p301). By exten- p294). Hébert shows the correspondence be- sion, it can be used in needs-based health care tween the items in the scale and the WHO classi- planning, evaluation, and cost-benefit analyses fication (5, Table 1; 6, Table 1). Items refer to (2; 3). present function and assess actual performance rather than potential. Information for making Conceptual Basis the ratings is obtained by interviewing the pa- The SMAF was designed to be used in assessing tient or a relative, or through observation and needs and in planning care for elderly people. Its testing the person (5). This typically takes about design reflects the WHO distinction between dis- 42 minutes (1, p297; 6, p163). ability and handicap, but this is extended in sev- For each item, disability is rated on a scale eral innovative ways. Handicap is conceptualized running from independent to dependent on assis- in terms of the shortfall between any disability a tance; the 1993 revision added intermediate  Physical Disability and Handicap 123 scale levels to some items, as seen in Exhibit into five professional categories. The overall 3.22. Because disability represents a deficit, kappa was 0.75, with coefficients for the sub- scores are oriented negatively, to a maximum of scales ranging from 0.58 (mental function) to −87. Hébert et al. discussed the option of using 0.76 (IADLs) (5, Table 2; 6, Table 2). Perfect nursing time as a criterion for individually agreement was achieved for 61% of items (1, weighting each item. However, because the ADL p298). Hébert et al. also noted that agreement and IADL items lie in a hierarchy such that an was as good for the earlier assessments by each ADL disability generally entails IADL problems, pair of raters as for the later ones, so concluded the value of differential item weights is minimal that special training is not necessary for using (7, p1307). Where a disability is identified, the the SMAF (6, pp163–4). assessor asks whether human resources (e.g., Inter-rater agreement was also estimated in a family members, volunteers, paid staff ) are study of pairs of nurses who assessed 45 elderly available to compensate for the disability. If so, people in residential care. The intraclass correla- the handicap score is zero, but otherwise the tion (ICC) for the overall score was 0.96 and the handicap score equals the disability score. More ICCs for the section scores ranged from 0.74 complex systems were considered for scoring (communication) to 0.95 (ADL) and 0.96 help that partially compensates for a disability, (IADL). Equivalent kappa values were 0.68 over- but these proved impractical (Hébert, personal all, with section values ranging from 0.61 to 0.81 communication). In such cases, the SMAF hand- (9, Table II). The same study also assessed two- icap scores are somewhat exaggerated, but this week test-retest reliability, yielding weighted bias was considered acceptable (5, p142). The kappa coefficients for the items ranging from people who assist are identified, and an estimate 0.45 to 0.95. The ICC for the overall score was is made of the stability of this arrangement over 0.95, with values for the subscales ranging from the coming 3 to 4 weeks (see Exhibit 3.22). 0.78 (communication) to 0.96 (ADL); (5, Table An administration manual is available (8), as 3; 9, p404). ICC values between 0.97 and 1.00 is a Web-based training program (details avail- were reported for the inter-rater reliability of able from Dr. Hébert). For people in institutional most of the SMAF sections in a study of emer- care, a pictoral summary sheet has been designed gency patients; the ICC for the communication for inclusion in the patient’s chart (5, Figure 4; subscale was lower, at 0.72 (10, p1026). 6, Figure 5). Here, small color-coded stickers re- place the 0 to −3 scores for each item, and pro- Validity vide an immediate visual picture of the patient’s In a study of 146 long-stay patients, SMAF dis- areas of disability; there is space to add suc- ability ratings were compared with an estimate cessive stickers to illustrate improvement or de- of the nursing time required to care for each pa- terioration over time. The SMAF has been tient. The overall SMAF score correlated 0.88 computerized for use in planning home support with nursing time; the correlation for the ADL services; the profiles are kept on a central com- section was 0.89; the coefficient for the mobility puter accessible by administrators, by staff at the score was 0.83, whereas lower correlations were health center, and by physicians’ offices (5, obtained for communication and mental func- p146). tions (1, Table IV). In a replication on 1,997 subjects, SMAF scores correlated 0.92 with Reliability nursing care time (5, p144; 11, Figure 2). This Because the SMAF is a rating scale, it is impor- close agreement led Hébert et al. to derive re- tant to assess whether equivalent results are ob- gression equations to estimate the nursing time tained by different types of rater (e.g., nurses, required for people with varying levels of disabil- social workers). An early study blended inter- ity (6, p164; 11, p10). From the equation, a rater and test-retest reliability by comparing SMAF score of 20 would predict roughly 40 pairs of ratings of 146 community patients made minutes of daily nursing care, and a score of 40 24 hours apart by a total of 30 raters, grouped would predict roughly two hours of care (5,  Exhibit 3.22 The Functional Autonomy Measurement System A u t o n o m y a s s e s s m e n t s c a l e Name : _________________________________________________________________ Dossier : ________________________________________________________________ Date : _____________________________________ Assessment #: ________________ DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 124 1. Family 3. Employee 5. Nurse 7. Other A. ACTIVITIES OF DAILY LIVING (ADL) 1. EATING 0 Feeds self independently − −0,5 With difficulty ➤ Does the subject presently have the human resources (help or supervision) 0 −1 Feeds self but needs stimulation or supervision + ➤ necessary to overcome this disability? OR food must be prepared or cut or puréed first −2 Needs some help to eat Yes . OR dishes must be presented one after another −1 −3 Must be fed totally by another person No −2 OR has a naso-gastric tube or a gastrostomy −3 naso-gastric tube gastrostomy Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 2. WASHING 0 Washes self independently (including getting in or out of the bathtub or shower) ➤ −0,5 With difficulty − Does the subject presently have the human resources (help or supervision) 0 −1 Washes self but needs cueing ➤ necessary to overcome this disability? + OR needs supervision Yes OR needs preparation OR needs help for the complete weekly bath −1 . only (including washing feet and hair) No −2 −3 −2 Needs help for the daily wash but participates Resources: actively 125 −3 Must be washed by another person 3. DRESSING (all seasons) 0 Dresses self independently −0,5 With difficulty ➤ −1 Dresses self but needs cueing Does the subject presently have the human resources (help or supervision) 0 − ➤ OR needs supervision necessary to overcome this disability? OR clothing must be readied and Yes + presented OR needs help for finishing touches −1 . (buttons, laces) No −2 −3 −2 Needs help dressing Resources: −3 Must be dressed by another person support hose/stocking * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 4. GROOMING (brushes teeth or combs hair or shaves or trims finger or toenails or puts on make-up) 0 Grooms self independently −0,5 With difficulty − ➤ −1 Grooms self but needs cueing or supervision Does the subject presently have the human resources (help or supervision) 0 + ➤ −2 Needs help for grooming necessary to overcome this disability? −3 Must be groomed by another person Yes . −1 No −2 −3 126 Resources: 5. URINARY FUNCTION 0 Normal voiding −1 Occasional incontinence ➤ OR dribbling Does the subject presently have the human resources (help or supervision) 0 − OR needs frequent cueing to avoid ➤ necessary to overcome this disability? incontinence Yes + −2 Frequent urinary incontinence −1 −3 Complete urinary incontinence No −2 . OR wears a diaper or an indwelling catheter or a −3 urinary condom diaper, indwelling catheter, urinary condom Resources: da day incontinence night incontinence * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 6. BOWEL FUNCTION 0 Normal bowel function −1 Occasional incontinence ➤ OR needs cleansing enema occasionally Does the subject presently have the human resources (help or supervision) 0 − ➤ −2 Frequent incontinence necessary to overcome this disability? OR needs cleansing enema regularly + Yes −3 Always incontinent OR wears a diaper or an ostomy −1 . No −2 diaper, ostomy −3 day incontinence night incontinence Resources: 127 7. TOILETTING 0 Toilets self (including getting on/off toilet, wiping self and managing clothing) ➤ −0,5 With difficulty Does the subject presently have the human resources (help or supervision) 0 − ➤ −1 Needs supervision for toiletting necessary to overcome this disability? OR uses commode, bedpan or urinal Yes + −2 Needs help to go to the toilet −1 OR use commode, bedpan or urinal No −2 . −3 Does not use toilet, commode, bedpan or urinal −3 commode, bedpan, urinal Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other B. MOBILITY 1. TRANSFERS (bed to chair or wheelchair and to stand, and vice-versa) 0 Gets in and out of bed or chair alone −0,5 With difficulty ➤ − −1 Gets in and out of bed/chair alone but needs cueing, Does the subject presently have the human resources (help or supervision) 0 ➤ supervision or guidance necessary to overcome this disability? + specify: ______________________________________ Yes −2 Needs help to get in and out of bed/chair −1 . specify: ______________________________________ No −2 −3 Bedridden (must be lifted in and out of bed) −3 particular positioning: Resources: 128 lift transfer board 2. WALKING INSIDE (including in the building and going to the elevator) 0 Walks independently (with or without cane, prosthesis, orthosis or walker) ➤ −0,5 With difficulty Does the subject presently have the human resources (help or supervision) 0 − ➤ −1 Walks independently but needs guidance, cueing or necessary to overcome this disability? supervision in certain circumstances OR unsafe gait + Yes −2 Needs help of another person to walk −1 . −3 Does not walk No −2 cane, tripod, quadripod, 1 walker −3 *Distance of at least 10 meters Resources: *STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 3. DONNING PROSTHESIS OR ORTHOSIS 0 Does not wear prosthesis or orthosis −1 Dons prosthesis or orthosis independently ➤ − −0,5 With difficulty Does the subject presently have the human resources (help or supervision) 0 ➤ necessary to overcome this disability? −2 Donning of prosthesis of orthosis needs checking + OR needs partial help Yes −3 Prosthesis or orthosis must be put on by another −1 . person Type of prosthesis or orthosis: No −2 ____________________________________________ −3 ____________________________________________ Resources: 129 4. PROPELLING A WHEELCHAIR (W/C) INSIDE 0 Does not need a wheelchair −1 Propels wheelchair independently Does the subject’s present residence allow for W/C −0,5 With difficulty or scooter mobility? ➤ Yes ➤ −2 Needs to have wheelchair pushed 0 − ➤ −3 Unable to use wheelchair (must be transported No on stretcher) ➤ + standard wheelchair Does the subject presently have the human resources (help or supervision) wheelchair with unilateral axis necessary to overcome this disability? . motorized wheelchair Yes three wheeled scooter −1 four wheeled scooter No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 5. NEGOTIATING STAIRS 0 Goes up and down stairs independently −0,5 With difficulty Does the subject have to negotiate stairs? ➤ −1 Requires cueing, supervision or guidance to No negotiate stairs OR does not safely negotiate stairs 0 Yes − ➤ −2 Needs help to go up and down stairs ➤ −3 Does not negotiate stairs Does the subject presently have the human resources (help or supervision) + necessary to overcome this disability? Yes . −1 No −2 −3 Resources: 130 6. MOVING AROUND OUTSIDE 0 Walks independently (with or without cane, prosthesis, orthosis or walker) −0,5 With difficulty *Does the outside environment where the subject lives allow for W/C or scooter access and mobility? ➤ −1 Uses a wheelchair or scooter independently −1,5 W/C with difficulty Yes ➤ 0 − ➤ ➤ OR walks independently but needs guidance, cueing or supervision in certain circumstances No + ➤ OR unsafe gait1 Does the subject presently have the human resources (help or supervision) −2 Nees help of another person to walk1 necessary to overcome this disability? . OR to use W/C −3 Cannot move around outside (must be transported Yes on a stretcher) −1 1 Distance of at least 20 meters No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other C. COMMUNICATION 1. VISION 0 Sees adequately with or without corrective lenses −1 Visual problems but sees enough to do ADLs − ➤ −2 Only sees outlines of objects and needs guidance in Does the subject presently have the human resources (help or supervision) 0 ➤ ADLs necessary to overcome this disability? + −3 Blind Yes corrective lenses magnifying glass −1 . No −2 −3 131 Resources: 2. HEARING 0 Hears adequately with or without hearing aid −1 Hears if spoken to in a loud voice ➤ OR needs hearing aid put in by another person Does the subject presently have the human resources (help or supervision) 0 − ➤ −2 Only hears shouting or certain words necessary to overcome this disability? OR reads lips + Yes OR understands gestures −3 Completely deaf and unable to understand what is −1 . No −2 said to him/her −3 hearing aid Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 3. SPEAKING 0 Speaks normally −1 Has a speech/language problem but able to ➤ express him/herself Does the subject presently have the human resources (help or supervision) 0 − ➤ −2 Has a major speech/language problem but able to necessary to overcome this disability? express basic needs Yes + OR answer simple questions (yes, no) −1 OR uses sign language No −2 . −3 Does not communicate technical aid: −3 computer Resources: communication board 132 D. MENTAL FUNCTIONS 1. MEMORY 0 Normal memory −1 Minor recent memory deficit (names, appointments, ➤ − etc.) but remembers important facts Does the subject presently have the human resources (help or supervision) 0 ➤ −2 Serious memory lapses (shut off stove, medications, necessary to overcome this disability? + putting things away, eating, visitors) Yes −3 Almost total memory loss or amnesia −1 . No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 2. ORIENTATION 0 Oriented to time, place and persons −1 Sometimes disoriented to time, place and persons ➤ −2 Only oriented for immediate events (i.e., time of day) Does the subject presently have the human resources (help or supervision) 0 − and in the usual living environment and with familiar ➤ necessary to overcome this disability? persons + Yes −3 Complete disorientation −1 . No −2 −3 133 Resources: 3. COMPREHENSION 0 Understands instructions and requests −1 Slow to understand instructions and requests ➤ −2 Partial understanding even after repeated Does the subject presently have the human resources (help or supervision) 0 − ➤ instructions OR is incapable of learning necessary to overcome this disability? −3 Does not understand what goes on around him/her Yes + −1 . No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 4. JUDGMENT 0 Evaluates situations and makes sound decisions −1 Evaluates situations but needs help in making ➤ sound decisions Does the subject presently have the human resources (help or supervision) 0 − ➤ −2 Poorly evaluates situations and only makes necessary to overcome this disability? sound decisions with strong suggestions + Yes −3 Does not evaluate situations and dependent on −1 . others for decision making No −2 −3 134 Resources: 5. BEHAVIOUR 0 Appropriate behaviour −1 Minor behavioural problems (whimpering, ➤ emotional lability, stubbornness, apathy) requiring Does the subject presently have the human resources (help or supervision) 0 − occasional supervision or a reminder or stimulation ➤ necessary to overcome this disability? −2 Major behavioural problems requiring more + Yes intensive supervision (aggressive towards self or −1 . others, disturbs others, wanders, screams out constantly) No −2 −3 −3 Dangerous, requires restraint OR harmful to others or self-destructive Resources: OR tries to run away * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other E. INSTRUMENTAL ACTIVITIES OF DAILY LIVING 1. HOUSEKEEPING 0 Does housekeeping alone (including daily housework and occasional big jobs) −0,5 With difficulty − ➤ −1 Does housekeeping but needs supervision or cueing to Does the subject presently have the human resources (help or supervision) 0 + ➤ ensure cleanliness (including washing the dishes) OR needs necessary to overcome this disability? help for big jobs (floors, windows, painting, Yes . lawn, clearing the snow, etc.) −1 −2 Needs help for daily housework No −2 −3 Does not do housework −3 135 Resources: 2. MEAL PREPARATION 0 Prepares own meals independently −0,5 With difficulty ➤ −1 Prepares meals but needs cueing to maintain Does the subject presently have the human resources (help or supervision) 0 − ➤ adequate nutrition necessary to overcome this disability? Only prepares light meals Yes + −2 OR heats up pre-prepared meals −1 . (including handling the plates) No −2 −3 Does not prepare meals −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 3. SHOPPING 0 Plans and does shopping independently (food, clothes) −0,5 With difficulty ➤ −1 Plans and shops independently but needs delivery Does the subject presently have the human resources (help or supervision) 0 − ➤ service necessary to overcome this disability? + −2 Needs help to plan or shop Yes . −3 Does not shops −1 No −2 −3 136 Resources: 4. LAUNDRY 0 Does all laundry independently −0,5 With difficulty ➤ −1 Does laundry but needs cueing or supervision to maintain Does the subject presently have the human resources (help or supervision) 0 − ➤ standards of cleanliness necessary to overcome this disability? + −2 Needs help to do laundry Yes −3 Does not do laundry −1 . No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 5. TELEPHONE 0 Uses telephone independently (including use of directory) −0,5 With difficulty ➤ Does the subject presently have the human resources (help or supervision) 0 − −1 Answers telephone but only dials a few memorized ➤ numbers or emergency numbers necessary to overcome this disability? + −2 Communicates by telephone but does not dial Yes numbers or lift the receiver off the hook . −1 −3 Does not use the telephone No −2 −3 137 Resources: 6. TRANSPORTATION 0 Able to use transportation alone (car, adapted vehicle, taxi, bus, etc.) −0,5 With difficulty ➤ −1 Must be accompanied to use transportation Does the subject presently have the human resources (help or supervision) 0 − ➤ OR uses paratransit independently necessary to overcome this disability? + −2 Uses car or paratransit only if accompanied Yes and has help getting in and out of the vehicle . −1 −3 Must be transported on a stretcher No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply. (continued)  Exhibit 3.22 (continued) DISABILITIES RESOURCES HANDICAP STABILITY* 0. Subject himself 2. Neighbour 4. Aides 6. Volunteer 1. Family 3. Employee 5. Nurse 7. Other 7. MEDICATION USE 0 Takes medication according to prescription OR does not need medication −0,5 With difficulty ➤ −1 Needs weekly supervision (including supervision Does the subject presently have the human resources (help or supervision) 0 − ➤ by telephone) to ensure compliance to prescription necessary to overcome this disability? + OR uses a medication dispenser aid Yes (prepared by someone else) −1 . −2 Takes medication if prepared daily No −2 −3 Must be given each dosage of medications −3 138 (as prescribed) Resources: medication dispenser aid 8. BUDGETING 0 Manages budget independently (including banking) −0,5 With difficulty ➤ −1 Needs help for certain major transactions Does the subject presently have the human resources (help or supervision) − ➤ −2 Needs help for some regular transactions (cashing necessary to overcome this disability? checks, paying bills) but uses pocket money wisely Yes + −3 Does not manage budget −1 . No −2 −3 Resources: * STABILITY: In the next 3 or 4 weeks, it is foreseeable that these resources will: − lessen, + increase, . remain stable or does not apply.  Physical Disability and Handicap 139 p144; 11, Figure 2). Hébert et al. also emphasized Reference Standards that such estimates may be valid for a group of Hébert et al. provide an interesting analysis of people but not for an individual (6, p164). the smallest change in function that can reliably SMAF scores have been linked to the cost of be detected by the SMAF (7). They compared care for home care, intermediate-level care, and empirical and statistical approaches to estimat- long-term care institutions (5, Table 4). Estimat- ing this threshold and concluded that the ran- ing the fit of regression equations for the three dom variation of an underlying stable SMAF settings, the variance explained (R2) was 0.57 for score lies within ± 5 points, so that a change in home care costs and 0.70 for institutional care. score over time of five points or more represents Because of the considerable variability in types a reliable change. They reported a standard devi- of intermediate care settings, the R2 was only ation in scores of 9.4, so that if the reliable 0.22 (5, Table 4). change score is 5, the SMAF can detect a moder- SMAF scores distinguished significantly ate effect size (difference divided by standard de- (p<0.01) between elderly people living in differ- viation) of roughly 0.5 (7, p1308). In a ent types of institution; the main contrast in subsequent study, it was found that all people scores was for the mobility and ADL sections (6, admitted to institutional care over a one-year pe- Figure 3; 12, Figure 1). Rai et al. showed signifi- riod had showed a decline exceeding five points cantly greater improvements in SMAF scores for (16, p164). people discharged home compared with others Based on 1,997 interviews, reference stan- who remained in continuing care (13). dards have been calculated for people living at In a study of 80 patients, the responsiveness home, in intermediate-level care, and long-term (see Glossary) of the SMAF (Guyatt index score care institutions (5, Figure 3). On average, 14.5) was higher than that for the Barthel Index healthy elderly people lose 2.9 SMAF points per (12.8) or the FIM (13.7). The correlation between year (17). SMAF and FIM scores was 0.94; that between SMAF and Barthel was 0.92 (14, p145). The Commentary SMAF overall score has a Spearman correlation of The SMAF is innovative in several ways. It inte- 0.80 with the Older Americans Resources and Ser- grates disability and handicap in a single instru- vices ADL/IADL questionnaire; the correlation ment. The ratings of support in effect make the was 0.63 for the ADL section alone, and 0.77 for SMAF a family-level measure, conceptually re- the IADL section (10, Table 3). Mercier et al. used lated to research on caregiving, whereas the rat- a LISREL analysis to study the relative contribu- ing of the stability of that support adds a tions of measures of motor (divided into balance prognostic dimension. This is pertinent to as- and upper limb functions), perceptual and cogni- sessing unmet needs and in allocating health care tive deficits to the SMAF scores. The four factors resources, given that these address disabilities explained 93% of variance in SMAF scores; the that are not alleviated by the family’s own re- strongest link was with balance (standardized sources (5, p145; 6). Reflecting this practical ap- weight 0.64), followed by cognitive (0.24), upper plication, the instrument has been designed to limbs (0.17) and perceptual (0.16) (14, p2604). make it practically applicable, and it is being routinely used in health care planning in the Alternative Forms province of Québec in Canada. An abbreviated version with 20 items is intended As indicators of need, the focus on the profile for use in institutional settings; scores are out of of disability scores (rather than the conventional 60 (15). It omits items on household tasks and overall scores) is also innovative. This led to the on walking outside; a single-page form showing identification of patterns of disability that held the items is available from Dr. Hébert. similar implications care needs: the so-called ISO- The SMAF was developed in Canada in SMAF patient profiles (5, p145). These comprise French, and is available in English, Dutch and groups of people with similar patterns of disability Spanish translations. (or handicap) who therefore require similar types 140 Measuring Health of care. The profiles were based on cluster analysis of a large population sample, guided by review of References the resulting groups to ensure that they made clin- (1) Hébert R, Carrier R, Bilodeau A. The ical sense (18). Fourteen such groups were identi- Functional Autonomy Measurement fied, ranging from people with loss of IADL System (SMAF): description and validation abilities (group 1) to bedridden people who re- of an instrument for the measurement of quire total care in group 14. Groups 2 to 13 repre- handicaps. Age Ageing 1988;17:293–302. sent intermediate steps, at each of which roughly (2) Hébert R, Brayne C, Spiegelhalter D. equivalent resources are required (18, Table 1). Factors associated with functional decline and improvement in a very elderly The focus on profiles is reminiscent of the econo- community-dwelling population. Am J metric approach used in instruments such as the Epidemiol 1995;150:501–510. Quality of Well-Being Index or the Health Utilities (3) Robichaud L, Hébert R, Roy P-M, et al. A Index. However, in contrast to these, Hébert’s preventive program for community- team has taken the SMAF further into the field of dwelling elderly at risk of functional cost analysis. Average costs of care for each profile decline: a pilot study. Arch Gerontol have been calculated for various types of care set- Geriatr 2000;30:73–84. ting (19, Table 1). From prevalence estimates of (4) Hébert R. Functional decline in old age. the ISO-SMAF groups estimates can be made of Can Med Assoc J 1997;157:1037–1045. the health system resources required in a region (5) Hébert R, Guilbeault J, Desrosiers J, et al. (18, p256). Finally, the early studies recorded the The Functional Autonomy Measurement System (SMAF): a clinical-based instrument typical disability profiles that each category of in- for measuring disabilities and handicaps in stitution could manage (6, Figure 4), so that the older people. Geriatrics Today: J Can clinician can judge which level of care would suit a Geriatrics Soc 2001;4:141–147. particular patient. An example of applying the (6) Hébert R, Carrier R, Bilodeau A. Le SMAF in a cost-benefit analysis is given by Tousig- système de mesure de l’autonomie nant et al. (17; 19); this illustrates analyses of the fonctionnelle (SMAF). Revue Geriatr relative costs of caring for people with given levels 1988;13:161–167. of disability in different settings. Likewise, Hébert (7) Hébert R, Spiegelhalter DJ, Brayne C. et al. have illustrated the differences, for example, Setting the minimal metrically detectable between analysing total costs (in which admission change on disability rating scales. Arch to institutional care becomes cheaper than home Phys Med Rehabil 1997;78:1305–1308. (8) Hébert R, Guilbeault J. Functional care at certain levels of disability) versus public Autonomy Measuring System: user guide. costs (in which home care is always cheaper) (11, Sherbrooke, Québec: Sherbrooke pp13–14). University Geriatric Institute (E-mail: The SMAF is chiefly distinguished by its in- [email protected]), 2002. novative design and the thoroughness with (9) Desrosiers J, Bravo G, Hébert R, et al. which its application in health care planning has Reliability of the revised Functional been pursued. Compared with other measure- Autonomy Measurement System (SMAF) ment scales, there has been relatively little testing for epidemiological research. Age Ageing of its reliability and validity, and it has seen little 1995;24:402–406. application outside of Québec. This scale de- (10) McCusker J, Bellavance F, Cardin S, et al. serves to be better known and to see more wide- Validity of an activities of daily living questionnaire among older patients in the spread application and testing. emergency department. J Clin Epidemiol 1999;52:1023–1030. Address (11) Hébert R, Dubuc N, Buteau M, et al. Dr. Réjean Hébert, Institut universitaire de géri- Resources and costs associated with atrie de Sherbrooke, 1036 Belvédère Sud, Sher- disabilities of elderly people living at home brooke, Québec, Canada J1J 4C4 and in institutions. Can J Aging E-mail : [email protected]. 2001;20:1–22.  Physical Disability and Handicap 141 (12) Hébert R, Bilodeau A. Profil d’autonomie tion. It is a rating scale applicable to patients of fonctionnelle des personnes agées all ages and diagnoses, by clinicians or by non hébergées en institution. Cahiers de clinicians, and has been widely adopted by reha- l’ACFAS 1986;46:66–79. bilitation facilities in the United States and else- (13) Rai GS, Gluck T, Wientjes HJ, et al. The where (1). Functional Autonomy Measurement System (SMAF): a measure of functional Conceptual Basis change with rehabilitation. Arch Gerontol Geriatr 1996;22:81–85. To simplify medical payments in the United (14) Mercier L, Audet T, Hébert R, et al. States, a standard remuneration system bases Impact of motor, cognitive, and perceptual payment for acute care on diagnosis rather than disorders on ability to perform activities of on the care actually provided. However, because daily living after stroke. Stroke the amount of care required for rehabilitation is 2001;32:2602–2608. based on level of disability rather than on diag- (15) Desrosiers J, Hébert R. Principaux outils nosis, an alternative assessment system was d’évaluation en clinique et en recherche. In: needed to form the basis for estimating payments Arcand M, Hébert R, eds. Précis pratique in rehabilitation medicine. In 1983, a national de gériatrie. St.-Hyacinthe et Paris: task force designed a Uniform Data System for Edisem/Malone, 1997:78–107. Medical Rehabilitation (UDS) to achieve uni- (16) Hébert R, Bravo G, Korner-Bitensky N, et al. Predictive validity of a postal form definitions and measurements of disability; questionnaire for screening community- the FIM is the central measurement of this dwelling elderly individuals at risk of scheme (2; 3). In addition, the UDS includes fur- functional decline. Age Ageing ther items covering demographic characteristics, 1996;25:159–167. diagnoses, impairment groups, length of hospi- (17) Tousignant M, Hébert R, Desrosiers J, et tal stay, and hospital charges. al. Economic evaluation of a geriatric day The UDS distinguishes between alterations in hospital: cost-benefit analysis based on structure or function (“impairment,” in WHO functional autonomy changes. Age Ageing terms), activity (disability), and role (handicap). 2003;32:53–59. The FIM covers the activity and role levels, (18) Dubuc N, Hébert R, Desrosiers J, et al. termed “life functions” (1, p141). Life functions Système de classification basé sur le profil d’autonomie fonctionnelle. In: Hébert R, are reduced by a disabling condition and rehabili- Kouri K, eds. Autonomie et vieillissement. tation seeks to restore them. The FIM is not seen St-Hyacinthe, Québec: Edisem, as comprehensive, but as a basic indicator, collect- 1999:255–272. ing the minimum information required for assess- (19) Tousignant M, Hébert R, Dubuc N, et al. ing disability (4). The FIM focuses on burden of Application of a case-mix classification care: the level of a patient’s disability indicates the based on the functional autonomy of the burden of caring for them and items are scored on residents for funding long-term care the basis of how much assistance is required for facilities. Age Ageing 2003;32:60–66. the individual to carry out activities of daily living (4). As human or physical resources have to be used to substitute for the individual’s reduced The Functional Independence function, disability entails an opportunity cost to Measure society, because these resources cannot be applied (Carl V. Granger and Byron B. to other uses. “The helper cost is measured in Hamilton, 1987) hours or energy consumed (e.g., heavy lifting), Purpose stress of concern or responsibility for the individ- The FIM is a clinical rating scale that assesses ual’s safety (e.g., falling), and the frustration of physical and cognitive disability in terms of level being on call constantly” (1, p142). of care required. It is used to monitor patient The UDS identifies several stages of rehabili- progress and to assess outcomes of rehabilita- tation and efficiency of care may be estimated by 142 Measuring Health dividing the increase in life function (e.g. mea- based on over 27,000 patients that translates this sured by improvement in FIM scores) by the cost ordinal score into an interval scale; they provide of the rehabilitation services. charts that show the conversion and the rank or- der of severity of each item (7, Figures 2, 5, & 6; Description 8, Figure 4). As an alternative scoring approach, The FIM includes 18 items covering indepen- the 13 physical items may be scored separately dence in self-care, sphincter control, mobility, from the five cognitive items in the communica- locomotion, communication, and cognition (5). tion and social cognition groups. The use of The physical items were based on the Barthel In- summated motor and cognitive scores is gener- dex and cover self-care, sphincter control, mo- ally upheld (8–10). A method for translating bility, and locomotion. Three cognition items FIM scores into scores on the Minimum Data cover social interaction, problem solving, and Set for rehabilitation is available (11). memory (see Exhibit 3.23). A pocket-sized chart summarizing the items and scoring system is Reliability shown in Exhibit 3.24. Ratings consider perfor- Numerous reports about inter-rater reliability mance rather than capacity and may be based on have been published. Ottenbacher et al. reviewed observation, a patient interview, or medical rec- eleven such studies, and calculated mean values ords. A decision tree is available from the UDS of inter-rater ICCs and of test-retest coefficients; that indicates the questions to ask to rate each both were 0.92 and the corresponding medians item in a telephone interview. Evaluators are were 0.95 (12, Table 4). Based on data from usually physicians, nurses, or therapists, but pooled samples from several studies, Otten- they may include laypeople. It takes about one bacher et al. estimated the standard error of hour to train raters to use the FIM and about 30 measurement for the FIM at 4.7 points (12, minutes to administer and score the scale for p1230). each patient (1, p145). Training workshops can During original development work on the be arranged through the UDS group at Buffalo. FIM, inter-rater tests were carried out on pa- The seven-point ratings reflect the amount tients from 25 facilities by physicians, nurses, of assistance a patient requires (see the end of and therapists. The ICC for an earlier, four-point Exhibit 3.23). For each item, two levels of in- rating version was 0.86 for 303 pairs of clinical dependent functioning distinguish complete in- assessments at admission and 0.88 for 184 pairs dependence from modified independence, when at discharge (1, p145; 2, p871). Kappa coeffi- the activity is performed with some delay, safety cients of agreement for the 18 items averaged risk, or use of an assistive device. Two levels of 0.54. Turning to the seven-point rating version, dependency refer to the provision of assistance: ICCs for pairs of clinicians rating 263 patients “modified dependence” is when the assistant ranged from 0.93 (locomotion subscale) to 0.96 provides less than half the effort required to (self-care and mobility). The mean kappa index complete the task, and “complete dependence” is of agreement between ratings for each item was defined by the assistant providing more than half 0.71 (13). A comparison of two physiotherapist the effort. Within each level are finer gradations ratings of 81 multiple sclerosis patients gave of assistance. A summary of the rating scale in kappa coefficients between 0.50 and 0.70 for the shown in Exhibit 3.24; fuller details and illustra- 11 items in the self-care, sphincter, and mobility tions are contained in the guide (6), whereas sections; kappa coefficients were lowest for the details of how to rate unusual cases are con- social cognition section, ranging from 0.14 to tained in the periodic UDS Update publication 0.32 (14, Table 3). The ICC agreement between available from the UDS group (address follows). the raters was 0.83, and the alpha coefficient was A total score sums the individual ratings; higher 0.95 (14, p110). Hamilton et al. analysed data scores indicate more independent function. from 1018 patients drawn from UDS participat- Scores range from a low of 18 to a maximum of ing centers; inter-rater reliability ICCs were 0.96 126. Granger et al. outline a Rasch analysis for the total score on the seven-point version of  Physical Disability and Handicap 143 Exhibit 3.23 The Coverage of the Functional Independence Measure SELF-CARE Check and evaluate the most usual mode of comprehension. If both are about equally used, check A Eating. Includes use of suitable utensils to bring and V. food to mouth, chewing and swallowing, once meal ( )A = Auditory ( )V = Visual is appropriately prepared. Expression. Includes clear vocal or non-vocal Grooming. Includes oral care, hair grooming, expression of language. This item includes both washing hands and face, and either shaving or intelligible speech or clear expression of language using applying makeup. writing or a communication device. Bathing. Includes bathing the body from the neck Check and evaluate the most usual mode of expression. down (excluding the back), either tub, shower or If both are about equally used, check V and N. sponge/bed bath. Performs safely. ( )V = Vocal ( )N = Nonvocal Dressing—Upper Body. Includes dressing above the waist as well as donning and removing prosthesis or orthosis when applicable. SOCIAL COGNITION Dressing—Lower Body. Includes dressing from the Social Interaction. Includes skills related to getting waist down as well as donning or removing along and participating with others in therapeutic and prosthesis or orthosis when applicable. social situations. It represents how one deals with one’s Toileting. Includes maintaining perineal hygiene own needs together with the needs of others. and adjusting clothing before and after toilet or bed Problem Solving. Includes skills related to solving pan use. Performs safely. problems of daily living. This means making reasonable, safe, and timely decisions regarding financial, social and SPHINCTER CONTROL personal affairs and initiating, sequencing and self- Bladder Management. Includes complete correcting tasks and activities to solve the problems. intentional control of urinary bladder and use of Memory. Includes skills related to recognizing and equipment or agents necessary for bladder control. remembering while performing daily activities in an Bowel Management. Includes complete intentional institutional or community setting. It includes ability to control of bowel movement and use of equipment or store and retrieve information, particularly verbal and agents necessary for bowel control. visual. A deficit in memory impairs learning as well as performance of tasks. MOBILITY DESCRIPTION OF THE LEVELS OF FUNCTION Transfers: Bed, Chair, Wheelchair. Includes all AND THEIR SCORES aspects of transferring to and from bed, chair, and INDEPENDENT—Another person is not required for wheelchair, and coming to a standing position, if the activity (NO HELPER). walking is the typical mode of locomotion. 7 Complete Independence—All of the tasks described as Transfer: Toilet. Includes getting on and off a making up the activity are typically performed safely, toilet. without modification, assistive devices, or aids, and Transfers: Tub or Shower. Includes getting into within a reasonable time. and out of a tub or shower stall. 6 Modified Independence—Activity requires any one or more than one of the following: an assistive device, more LOCOMOTION than reasonable time, or there are safety (risk) Walking or Using Wheelchair. Includes walking, considerations. once in a standing position, or using a wheelchair, DEPENDENT—Another person is required for either once in a seated position, on a level surface. supervision or physical assistance in order for the Check most frequent mode of locomotion. If activity to be performed, or it is not performed both are about equal, check W and C. If initiating a (REQUIRES HELPER). rehabilitation program, check the mode for which MODIFIED DEPENDENCE—The subject expends training is intended. half (50%) or more of the effort. The levels of assistance ( ) W = Walking ( )C = Wheelchair required are: Stairs. Goes up and down 12 to 14 stairs (one 5 Supervision or setup—Subject requires no more help flight) indoors. than standby, cuing or coaxing, without physical contact. Or, helper sets up needed items or applies COMMUNICATION orthoses. Comprehension. Includes understanding of either 4 Minimal contact assistance—With physical contact auditory or visual communication (e.g. writing, sign the subject requires no more help than touching, or language, gestures). subject expends 75% or more of the effort. (continued) 144 Measuring Health Exhibit 3.23 (continued) 3 Moderate assistance—Subject requires more help activity is not performed. The levels of assistance than touching, or expends half (50%) or more (up required are: to 75%) of the effort. 2 Maximal assistance—Subject expends less than 50% COMPLETE DEPENDENCE—The subject of the effort, but at least 25%. expends less than half (less than 50%) of the effort. 1 Total assistance—Subject expends less than 25% of Maximal or total assistance is required, or the the effort. Adapted from Guide for the use of the Uniform Data Set for Medical Rehabilitation. Version 3.0. Buffalo, New York: Uniform Data System for Medical Rehabilitation, The Buffalo General Hospital, 1990. the FIM, 0.96 for the motor subscale, and 0.91 88). Stineman et al. also showed this using factor for the cognitive scale (15, Table III). analysis, and proposed a hierarchical view of the Alpha coefficients of 0.93 (admission) and FIM, with a single, overall score that is divided 0.95 (discharge) were found in a study of into motor and cognitive dimensions, which are 11,102 rehabilitation patients. The internal con- in turn sub divided into finer patterns of impair- sistency of the locomotion sub-scale was lower, ments (21, Figure 1). Jette et al. provided a most at 0.8 (16, p533). Alpha coefficients for the interesting analysis in which they combined overall scale were 0.92 at admission and 0.96 at items from the FIM, the 10 items of the physical discharge from rehabilitation (17, p638). Across function scale from the SF-36, the Minimum a range of different diagnostic categories, alpha Data Set (MDS) and the Outcome and Assess- values ranged from 0.88 to 0.97 (10, Table 5). ment Information Set for Home Health Care (OASIS). Using Rasch analysis, they compared Validity the scope of coverage of the four scales on a 0 During the early development of the FIM, con- (greatest disability) to 100 (no disability) scale. tent validity was tested by asking clinicians to The coverage of the FIM was narrow, running judge its scope and ease of administration. This from 25 to 86; the SF-36 physical function scale led to the addition of two new items (social ad- covered higher levels of function, running from justment and cognition) and the expansion of 50 to 100 (22, Figure 2). The MDS and OASIS the answer categories to include modified de- scales had a far broader coverage, almost the en- pendence and complete dependence (18, p12). tire range. Although relatively narrow in scope, Factor analyses have identified three factors: however, the FIM obtained the greatest measure- handicap, disability, and lower limb problems ment precision within that scope owing to its (17, Table 3), or two factors (originally termed larger number of items. ADLs versus neuropsychological abilities, but Granger et al. recorded the time required to now commonly termed motor versus cognitive) provide help for personal care tasks for 24 mul- (19, Table 2). Subsequent factor analyses sup- tiple sclerosis patients over a seven-day period. ported the division into motor and cognitive The FIM items predicted this rating (R2 = 0.77); scores (10, p1106), and this was supported by correlations for several items exceeded 0.80; a Rasch analysis (7). There may be minor varia- change of one point on the FIM total score rep- tions: in one Rasch analysis, eating, bowel and resented 3.8 minutes of care per day (2, Tables bladder (and to a lesser extent, walking) did not 2, 3). The R2 improved to 0.99 when five pa- fit the scale (20, Table 3), and two of the cogni- tients with visual impairments were omitted tion items also failed to meet item response the- from the analyses: the FIM does not consider the ory scaling criteria. In a more thorough amount of time required to care for someone be- investigation, Rasch analyses showed that con- cause of visual handicap. Similar analyses for 21 trasting patterns of responses arise for different stroke patients yielded an R2 of 0.65. A change patient groups, reflecting the types of disability of one point in the FIM score was related to an to be expected from their diagnoses (7, pp86, average of 2.2 minutes of help per day (23, Exhibit 3.24 Functional Independence Measure (FIM): Summary Chart 145 146 Measuring Health pp136–137). Disler et al. found FIM scores to on the FIM, Hall proposed the Functional As- correlate −0.39 with an estimate of the hours of sessment Measure (FAM), which extends the care required for 75 neurological patients; after range of difficulty (31). It was intended for pa- removing three patients with cognitive or visual tients with brain injury. The FAM includes the impairments, the Pearson correlation rose to FIM items but adds 12 new items mainly cover- −0.76 (24, p141). Each point of the FIM total ing aspects of cognition such as community inte- score reflected 4.1 minutes of care per day gration, emotional status, orientation, attention, (p142). reading and writing skills, and employability An attempt to evaluate the FIM cognitive (32, Table 1; 33, p64). Two studies have used items by comparison with a neuropsychological Rasch analysis to assess whether the FAM suc- test battery failed because almost all 41 spinal ceeds in improving the coverage of the FIM (20; cord injury patients achieved maximum scores 32). Linn et al. found that most of the FAM on the FIM items (25). Davidoff et al. com- items overlapped in difficulty with the FIM mented: “These findings underscore the poten- items; only two were useful in reducing ceiling tially misleading nature of a ‘normal’ score of 6 effects (20). Tesio and Cantagallo reported item or 7 on the Social Cognition and Communica- and person reliabilities of 0.91 and 0.93, but tion subscale items of the FIM. The false nega- again found that the FAM items added little to tive rate for detection of cognitive deficit using the FIM (32). the FIM varied from 0% to 63% for each neu- A version for children aged 6 months to 7 ropsychologic test” (25, p328). years (the “WeeFIM”) is a direct adaptation of Correlations with other measures include 0.84 the adult measure, with 18 items covering six with the Barthel Index, 0.68 with Katz’s Index of domains (34). It may be administered by observ- ADL, and 0.45 with Spitzer’s Quality of Life In- ing the child’s performance, or by interviewing a dex (26, Table 2). In a study of stroke patients, parent; the intraclass correlation between the correlations with the PULSES Profile ranged from two was found to be 0.93 (35). A Japanese ver- −0.82 to −0.88. ROC curves for the two scales in sion of the WeeFIM has been described (36). A predicting discharge to community versus long- basic description of the WeeFIM is available term care were virtually identical (27, p763). In a from the UDS web site (www.udsmr.org). study of predicting discharge destination, Sand- A version for telephone administration, the strom et al. obtained low-to-moderate predictive FONE-FIM, has been shown to generate slightly validity correlations for the FIM scores (28). lower estimates of disability than the observa- In a large study of 11,102 patients, Dodds et tional version (37). The reliability of a self- al. found that FIM scores improved between ad- report version has been described (38). mission and discharge and reflected the patient’s The FIM has been translated into French destination. Scores also reflected the presence of (19), German, Swedish, and Japanese (36). coexisting conditions and the severity of impair- ments (16, pp533–534). In a study of recovering Reference Standards stroke patients, the Barthel and the motor com- Granger and Hamilton have produced annual ponent of the FIM proved equally responsive to reports from the UDS that show mean scores and change (29, Tables 4 and 5). sub-scores at admission and discharge for vari- The FIM, although designed for adults, has ous categories of rehabilitation patients. These also been used with children as young as 7 years; are based on large numbers of patients: 44,997 overall and component scores showed significant in the 1991 report alone (5). associations with clinical prediction of duration Transition norms, describing the progress of disability (30). measured in FIM scores made by patients during rehabilitation, have been produced (39; 40). Alternative Forms These results might also be used in computing Because of problems with a ceiling effect the transition probabilities for prognostic in- whereby many patients achieve maximum scores dices such as the Quality of Well-Being scale.  Physical Disability and Handicap 147 Commentary Address The FIM was based on the well-established Information and guidelines for using the FIM are Barthel Index and developed with the consensus available from the Uniform Data System for of a national advisory committee that continues Medical Rehabilitation, 270 Northpointe Park- to oversee its refinement. It was carefully de- way Suite 300, Amherst, NY 14228 www signed, with close attention to item definitions, .udsmr.org/. standard administration procedures, and relia- bility. Documentation on the FIM is outstand- ing. The manual is thorough, and regular “UDS References Update” newsletters are written in an upbeat (1) Hamilton BB, Granger CV, Sherwin FS, et style that conveys the sense of participating in a al. A uniform national data system for large family of users. They provide information medical rehabilitation. In: Fuhrer MJ, ed. on training workshops, updates on validity re- Rehabilitation outcomes: analysis and mea- sults and scoring of unusual cases, and answer surement. Baltimore: Paul H. Brookes, readers’ questions. Training videos are available. 1987:137–147. A data management service oversees the colla- (2) Granger CV, Cotter AC, Hamilton BB, et tion of data from user groups, and these data al. Functional assessment scales: a study of form the basis for many validity and reliability persons with multiple sclerosis. Arch Phys studies. It is evident that considerable resources Med Rehabil 1990;71:870–875. are being channeled into developing and stan- (3) Granger CV, Hamilton BB, Keith RA, et al. Advances in functional assessment for dardizing this instrument. medical rehabilitation. Top Geriatr Rehabil A major strength of the FIM lies in the size of 1986;1:59–74. the UDS enterprise. As of 1990, 140 rehabilita- (4) UDS Data Management Service. Uniform tion facilities were participating in the data man- Data Set for Medical Rehabilitation: agement service, and an estimated 100 additional update. Buffalo, New York: UDS Data facilities use UDS in the United States, Canada, Management Service, SUNY, 1990. Australia, France, Japan, Sweden, and Germany. (5) Granger CV, Hamilton BB. The Uniform Several of the validation studies report data from Data System for Medical Rehabilitation 10,000 or more cases, and the study of 93,829 report of first admissions for 1991. Am J patients takes the gold medal in sample size Phys Med Rehabil 1993;72:33–38. among the measures reviewed in this book (21). (6) UDS Data Management Service. Guide for use of the Uniform Data Set for Medical In the United States, patient classification system Rehabilitation including the Functional In- called the FIM-Function Related Groups (FIM- dependence Measure (version 3.1). 3rd ed. FRGs) has been developed as a basis for health Buffalo: State University of New York at care payments or reimbursements (10). Buffalo, 1990. The physical components of the FIM appear (7) Granger CV, Hamilton BB, Linacre JM, et comparable with the best among other ADL in- al. Performance profiles of the Functional struments. The cognitive and social communica- Independence Measure. Am J Phys Med tion dimensions may have low sensitivity (25); Rehabil 1993;72:84–89. refinement may be desirable. Limitations in the (8) Linacre JM, Heinemann AW, Wright BD, et FIM include somewhat inflexible rules for scor- al. The structure and stability of the ing: where an assessment cannot be made, the Functional Independence Measure. Arch Phys Med Rehabil 1994;75:127–132. patient is rated as disabled, which is sometimes (9) Heinemann AW, Linacre JM, Wright BD, et inappropriate. Overall, the FIM is a sound in- al. Relationships between impairment and strument that benefits from outstanding support physical disability as measured by the services. Viewed as a brief disability measure functional independence measure. Arch rather than a general health instrument, it de- Phys Med Rehabil 1993;74:566–573. serves close consideration as a patient assessment (10) Stineman MG, Shea JA, Jette A, et al. The tool and also as an evaluative instrument. Functional Independence Measure: tests of 148 Measuring Health scaling assumptions, structure, and (21) Stineman MG, Jette A, Fiedler R, et al. reliability across 20 diverse impairment Impairment-specific dimensions within the categories. Arch Phys Med Rehabil Functional Independence Measure. Arch 1996;77:1101–1108. Phys Med Rehabil 1997;78:636–643. (11) Williams BC, Li Y, Fries BE, et al. (22) Jette AM, Haley SM, Ni P. Comparison of Predicting patient scores between the functional status tools used in post-acute Functional Independence Measure and the care. Health Care Financ Rev Minimum Data Set: development and per- 2003;24:13–24. formance of a FIM-MDS “crosswalk”. (23) Granger CV, Cotter AC, Hamilton BB, et Arch Phys Med Rehabil 1997;78:48–54. al. Functional assessment scales: a study of (12) Ottenbacher KJ, Hsu Y, Granger CV, et al. persons after stroke. Arch Phys Med The reliability of the Functional Indepen- Rehabil 1993;74:133–138. dence Measure: a quantitative review. Arch (24) Disler PB, Roy CW, Smith BP. Predicting Phys Med Rehabil 1996;77:1226–1232. hours of care needed. Arch Phys Med (13) Hamilton BB, Laughlin JA, Granger CV, et Rehabil 1993;74:139–143. al. Interrater agreement of the seven level (25) Davidoff GN, Roth EJ, Haughton JS, et al. Functional Independence Measure (FIM). Cognitive dysfunction in spinal cord injury Arch Phys Med Rehabil 1991;72:790. patients: sensitivity of the Functional Inde- (14) Brosseau L, Wolfson C. The inter-rater pendence Measure subscales vs reliability and construct validity of the neuropsychologic assessment. Arch Phys Functional Independence Measure for Med Rehabil 1990;71:326–329. multiple sclerosis subjects. Clin Rehabil (26) Rockwood K, Stolee P, Fox RA. Use of 1994;8:107–115. goal attainment scaling in measuring (15) Hamilton BB, Laughlin JA, Fiedler RC, et clinically important change in the frail el- al. Interrater reliability of the 7-level derly. J Clin Epidemiol functional independence measure (FIM). 1993;46:1113–1118. Scand J Rehabil Med 1994;26:115–119. (27) Marshall SC, Heisel B, Grinnell D. Validity (16) Dodds TA, Martin DP, Stolov WC, et al. A of the PULSES Profile compared with the validation of the Functional Independence Functional Independence Measure for mea- Measurement and its performance among suring disability in a stroke rehabilitation rehabilitation inpatients. Arch Phys Med setting. Arch Phys Med Rehabil Rehabil 1993;74:531–536. 1999;80:760–765. (17) Demers L, Giroux F. Validité de la mesure (28) Sandstrom R, Mokler PJ, Hoppe KM. de l’indépendence fonctionnelle (MIF) pour Discharge destination and motor function les personnes âgées suivies en réadaptation. outcome in severe stroke as measured by Can J Aging 1997;16:626–646. the functional independence mea- (18) Keith RA, Granger CV, Hamilton BB, et sure/function-related group classification al. The Functional Independence Measure: system. Arch Phys Med Rehabil a new tool for rehabilitation. In: Eisenberg 1998;79:762–765. MG, Grzesiak RC, eds. Advances in (29) Wallace D, Duncan PW, Lai SM. clinical rehabilitation. Vol. 1. New York: Comparison of the responsiveness of the Springer, 1987:6–18. Barthel Index and the motor component of (19) Brosseau L, Potvin L, Philippe P, et al. The the Functional Independence Measure in construct validity of the Functional Inde- stroke: the impact of using different pendence Measure as applied to stroke methods for measuring responsiveness. J patients. Physiother Theory Pract Clin Epidemiol 2002;55:922–928. 1996;12:161–171. (30) Di Scala C, Grant CC, Brooke MM, et al. (20) Linn RT, Blair RS, Granger CV, et al. Does Functional outcome in children with the Functional Assessment Measure (FAM) traumatic brain injury. Am J Phys Med extend the Functional Independence Mea- Rehabil 1992;71:145–148. sure (FIM) instrument? A Rasch analysis of (31) Hall KM, Mann N, High WM, et al. stroke inpatients. J Outcome Meas Functional measures after traumatic brain 1999;3:339–359. injury: ceiling effects of FIM, FIM+FAM,  Physical Disability and Handicap 149 DRS, and CIQ. J Head Trauma Rehabil measures of social functioning, many of which 1996;11:27–39. assess a person’s ability to perform normal so- (32) Tesio L, Cantagallo A. The Functional cial roles (see Chapter 4). We criticized the ADL Assessment Measure (FAM) in closed scales on several grounds: most were developed traumatic brain injury outpatients: a in relative isolation from other methods, and Rasch-based psychometric study. J few were founded on a clear conceptual basis or Outcome Meas 1998;2:79–96. critique of earlier work. Because of their con- (33) Hall K. The Functional Assessment Mea- sure (FAM). J Rehabil Outcomes Meas centration on basic functions, ADL measures 1997;1:63–65. suffer a ceiling effect when used with popula- (34) McCabe MA, Granger CV. Content validity tions living in the community. The broader of a pediatric functional independence mea- scope of the newer IADL instruments is increas- sure. Appl Nurs Res 1990;3:120–122. ingly supplanting the older scales; they are more (35) Sperle PA, Ottenbacher KJ, Braun SL, et al. sensitive to minor variations in a patient’s con- Equivalence reliability of the functional in- dition and have often been more thoroughly dependence measure for children (WeeFIM) tested. However, little work has yet been done administration methods. Am J Occup Ther to establish the formal correspondence among 1997;51:35–41. the various disability scales, with the exception (36) Liu M, Toikawa H, Seki M, et al. of the work of Jette et al. Research that com- Functional Independence Measure for Children (WeeFIM): a preliminary study in pares different scales forms a crucial stage in nondisabled Japanese children. Am J Phys consolidating a field of health measurement; Med Rehabil 1998;77:36–44. this has been achieved for measures of psycho- (37) Chang W-C, Chan C, Slaughter SE, et al. logical well-being (see Chapter 5) and for gen- Evaluating the FONE FIM: Part II. eral health measures (see Chapter 10), but not Concurrent validity & influencing factors. J yet in the field of functional disability measure- Outcome Meas 1997;1:259–285. ment. (38) Hoenig H, McIntyre L, Sloane R, et al. The Several themes emerge from our review. Per- reliability of a self-reported measure of haps because these topics are inherently subjec- disease, impairment, and function in tive and rely on self-report, more attention has persons with spinal cord dysfunction. Arch been paid to establishing the validity and relia- Phys Med Rehabil 1998;79:378–387. (39) Long WB, Sacco WJ, Coombes SS, et al. bility of IADL scales than is the case with the Determining normative standards for older ADL methods. Because they are sensitive Functional Independence Measure to lower levels of disablement, the IADL scales transitions in rehabilitation. Arch Phys are more suited to use as survey methods for Med Rehabil 1994;75:144–148. general population studies. It is also plausible (40) Hamilton BB, Granger CV. Disability that the IADL approach will come to rival, and outcomes following inpatient rehabilitation perhaps replace, the traditional ADL scales in for stroke. Phys Ther 1994;74:494–503. clinical studies. In their turn, however, the IADL scales may come to be replaced by the broader- ranging general measurement methods described Conclusion in Chapter 10. There is no essential distinction between the mixed ADL/IADL scales described The IADL scales reviewed in this chapter repre- here and the functional component of several of sent a bridge between the traditional physical the general health measurements covered in measurements represented by ADL scales, and Chapter 10.  4 Social Health he theme of social health may seem less fa- T miliar and is less frequently discussed and studied than physical or mental health. Being Since the 1947 World Health Organization (WHO) definition of health, an emphasis on treating patients as social beings who live in a less familiar, several potential misconceptions complex social context has been prominent in must be addressed at the outset. Because the medicine. People who are well-integrated into word “social” does not refer to a characteristic their communities tend to live longer and have a of individuals, it may not be immediately clear greater capacity to recover from disease; con- how a person can be rated in terms of social versely, social isolation is a risk factor for sick- health. Indeed, there is an important tradition of ness. Moreover, people with serious disease or regarding social health as a characteristic of so- disability need social support to remain in the ciety rather than of individuals: “A society is community, and the social view of medicine healthy when there is equal opportunity for all holds that the ultimate aim of care should be to and access by all to the goods and services essen- reintegrate people into productive lives in society tial to full functioning as a citizen” (1, p75). In- rather than merely to treat their medical symp- dicators of social health in this sense might toms. Beyond the philosophical appeal of con- include the distribution of economic wealth, sidering social adjustment as a component of public access to the decision-making process, health, there are practical reasons for measuring and the accountability of public officials. This an individual’s social well-being and adjustment. book, however, does not review indicators of the The expense of institutional care and the result- health of a society or population; it considers ing emphasis on discharging patients as early as only measurements of the rather less intuitively possible implies a need to assess their readiness obvious concept of the social health of individu- to live independently in the community. This als. A representative definition might describe theme was seen in the instrumental activities of social health as “that dimension of an individ- daily living (IADL) measures reviewed in Chap- ual’s well-being that concerns how he gets along ter 3. The movement away from institutional with other people, how other people react to care in the mental health field, which has been him, and how he interacts with social institu- responsible for partially emptying and some- tions and societal mores” (1, p75). The defini- times closing large mental hospitals, has fostered tion is broad; it incorporates elements of studies of the quality of adjustment to commu- personality, sociability, and social skills, and it nity living or social functioning, especially also in part reflects the norms of the society in among older patients (2). Studies of this type are which the individual finds himself. In fact, most equally relevant in the area of physical rehabili- measures of the social health of individuals do tation, and social function measures can be used not employ the word “health,” but speak instead to evaluate rehabilitation outcomes in terms of of “well-being,” “adjustment,” “performance,” social restoration: has the individual returned to or “social functioning.” Why, then, should we a productive and stable position in society (3)? regard this sphere of human interaction as a part The theme of social roles returns in some of the of health at all? quality of life measures reviewed in Chapter 10. 150  Social Health 151 A further reason to measure social health, albeit considering a person’s satisfaction with his rela- in a slightly different sense, is to examine the in- tionships or by studying his performance of var- fluence of social support and social ties on a per- ious social roles. The subjective approach son’s physical and psychological well-being. records affective responses such as discontent, This treats the social adjustment not as the de- unhappiness, or anxiety (e.g., Linn’s Social Dys- pendent variable, but as a predictor of health. function Rating Scale). This area of measure- Reviews of this long-established field have been ment is diffuse, and there are no clear given by Antonovsky (4), Berkman and Breslow boundaries between subjective measurements of (5), and Murawski et al. (6). social adjustment and measurements of life satis- These contrasting ways of defining social faction, happiness, or quality of life. Such scales health—in terms of adjustment, social support, are often subsumed under the general heading of or the ability to perform normal roles in soci- “subjective well-being,” but we have attempted ety—and the measurements that have been de- to form a finer classification. Measurement of veloped for each are further examined in the happiness and general affective well-being that following sections. are not specifically related to social relation- ships, such as Bradburn’s Affect Balance scale, are included in Chapter 5, which addresses psy- Social Adjustment and chological well-being. Measurements of affec- Social Roles tive responses that focus on social relationships are included in this chapter, whereas quality of The conception of social health in terms of so- life scales are described in Chapter 10. cial or community adjustment derived primarily One major challenge in measuring social ad- from the work of sociologists and, in the health justment lies selecting an appropriate standard field, of psychiatrists. Psychiatric interest in so- against which to evaluate adjustment. Norms cial health arose because fracture of personal or vary greatly from one culture to another, ranging social relationships is a common reason for seek- from an emphasis on “oneness with nature” and ing care for nonpsychotic mental disorders. The rejection of worldly values in Asian cultures to adequacy of a person’s social adjustment or in- an emphasis on material possessions in some teraction may therefore indicate a need for care; sectors of contemporary Western society. Expec- they may also form indicators of its outcome, es- tations also vary among social classes within a pecially that of psychotherapy. The development culture, making it difficult to compare adjust- of adjustment scales coincided with a gradual ment among times, places, and groups. The most shift in psychiatry away from medical concep- common way to avoid these problems is to focus tions of mental illness that emphasized disease the measurement on specific social roles for or deviance toward a view of mental distress in which there is some agreement about appropri- terms of inadequate social integration: does the ate behavior. The social role approach to assess- individual function adequately in personal rela- ing adjustment is based loosely on role theory tionships? This is most commonly expressed as and implies a valuation: how adequately is the social adjustment, broadly definable in terms of person performing compared with social expec- the interplay between the individual and her so- tations? A person who cannot function in a way cial environment and her success in chosen so- that meets the normal demands of his situation cial roles (7). Linn has viewed adjustment in a may be considered socially disabled (8). Al- dynamic sense, covering the person’s equilibrium though this does not completely overcome the or success in reducing tensions and in satisfying problem of defining what is normal, there are, at needs (2). Interest in social adjustment is, of least, recognized norms for many roles. They course, not specific to psychiatry: elementary may be formally couched in law, or in less for- school stresses the importance of learning to mal regulations, traditions, or agreements function as a social being. among individuals. Although approaches based Social adjustment may be measured either by on norms seem to offer promise, they are not 152 Measuring Health sufficiently refined to specify what should be in- may evidently be influenced by many factors cluded in a social health questionnaire, and ulti- other than health status. Although respondents mately the selection of topics appears to be more may be asked to identify only health-related or less arbitrary. Most operational definitions of problems, this is often a very complex judgment social roles consider housework, occupation, to make because problems rarely have a single community involvement, roles as spouse and cause. Nor are changes in role function specific parent, and leisure activities. Most of these top- to any one type of health problem: similar social ics are also covered in the IADL scales (Chapter and role limitations can be caused by depression 3), so the role approach to measuring social or physical disability. It can, therefore, be diffi- health brings it conceptually very close to in- cult to know how to classify indicators of social dices of functional handicap. A social role ap- or role functioning: as indicators of physical, proach was used in the measurements developed mental, or social health? by Weissman and by Gurland, reviewed else- Recognizing these potential problems, several where in this chapter. scales (such as that of Remington and Tyrer) There are several conceptual problems in us- avoid imposing fixed definitions of what consti- ing role theory as an approach to measuring so- tutes normal or adequate performance. The Katz cial health. Assumptions have to be made Adjustment Scales use another approach that regarding how to evaluate performance: should combines the objective assessments of the role it be compared to some ideal, to the person’s approach with subjective evaluations of satisfac- own aspirations, or to other peoples’ expecta- tion made by the respondent: to evaluate how tions of her performance? The first tends to be important it is that the individual does or does insufficient: although there are recognized not fulfill her social role requires information as norms for much behavior, there is little consen- to how she views that role. This reflects the con- sus over what constitutes a socially “correct” cept of the person-environment fit; rather than definition of the marital role, for example. stressing adherence to somewhat arbitrary prin- Norms vary between social strata and there is lit- ciples of behavior, the socially healthy person tle agreement over the relative importance of dif- would be one who has found a comfortable ferent roles. Alternative approaches also suffer niche in which to operate to the best of her ca- problems: comparing a person’s performance to pacities, and to the approval of those around her. the aspirations of her spouse, for example, makes it hard to evaluate social functioning be- cause the partner may have unrealistic expecta- Social Support tions. The role approach has been criticized as being rigid and conservative; it may be impracti- Studies in the field of social epidemiology have cal to evaluate the legitimacy of reasons for not long highlighted the importance of social sup- behaving in a “normal” way. Platt argued that port in attenuating the effects of stressful events the role approach implies viewing the “ideal in- and thereby reducing the incidence of disease dividual as an object which passively shapes it- (10–12). In addition, social support contributes self to the culture and the external environment. to positive adjustment in the child and adult and He should be satisfied with his situation and if he encourages personal growth. Because of the im- is not, then he is not fully adjusted” (9, p103). portance of social integration and social sup- Pursuing this further, one might argue that a so- port, we review some social support scales in cially healthy world would be “characterized by this chapter. harmony, happiness and consensus, and is in- “Social support” is generally defined in terms habited by men and women who are consistently of the availability of people whom the individual interested, active, friendly, adequate, guilt-free, trusts, on whom he can rely, and who make him nondistressed and so on. If they show anything feel cared for and valued as a person. Social sup- less than interest in their work they are malad- port may be distinguished from the related con- justed” (9, p106). Functioning in social roles cept of social networks, which refers to the roles  Social Health 153 and ties that link people along definable paths of phasis now commonly lies with assessing the kinship, friendship, or acquaintance. Social net- functional and qualitative aspects of relation- works may be seen as the structure through ships rather than their number or type. A re- which support is provided (13), whereas most search agenda might include evaluating how measures of social support record the function- these different dimensions of social support re- ing (process and outcome) of support. late to outcomes. An early scale that covered aspects of social support was the Berle Index, published in 1952 (14). For 30 years there were few other formal Scope of the Chapter measurements of social support, and many stud- ies relied on indirect structural indicators such as We present measures of social support and social marital status or other sociodemographic vari- adjustment. The theme of social support is rep- ables (13). The field has, however, become an resented here by McFarlane’s Social Relation- important area of growth in sociomedical mea- ship Scale, by Sarason’s Social Support sures and many scales have been proposed. Im- Questionnaire, by two Duke scales, and by two portant stimuli for the development of more scales from the RAND/MOS group: the RAND formal scales came from conceptual discussions Social Health Battery, which measures social in- of support, including Bowlby’s theories of at- teraction, and the social support scale of Sher- tachment (15) and Weiss’s functional analysis bourne and Stewart. There are also relevant (16). Weiss saw social support as performing in- scales in other chapters in the book, such as the strumental and expressive functions for the indi- Functional Assessment Inventory of Crewe and vidual: it provides for social integration, Athelstan in Chapter 10, which review the re- nurturance, alliance, and guidance; it also fosters sources that may assist a patient in coping with feelings of worth and intimacy. physical handicaps. Support may be of various types and Sher- The topic of social adjustment is treated in a bourne and Stewart distinguished five: providing sequence running from scales suited for general emotional support, love and empathy; providing application toward instruments designed for use instrumental or tangible support; providing in- with psychiatric and other patient groups. We formation, guidance, advice or feedback on be- begin with the Katz Adjustment Scales, followed havior; offering appraisal support which helps by the Social Functioning Schedule, which cov- the person to evaluate themselves; and giving ers problems in social functioning, the Interview companionship in leisure and recreational activi- Schedule for Social Interaction of Henderson et ties (17, p705). Support can also be classified by al., and Weissman’s Social Adjustment Scale. We the way in which it is experienced. Thus, a gen- then review the Social Maladjustment Schedule eral sense of belonging may be contrasted with of Clare, Linn’s Social Dysfunction Rating Scale perceived support, which refers to the availabil- and finally Gurland’s Structured and Scaled In- ity of particular people who can provide assis- terview to Assess Maladjustment. Table 4.1 pro- tance as required, and with enacted support, vides a quick reference comparison of the format which refers to specific supportive actions. Mea- and psychometric quality of these scales. The sures of structural support cover the existence conclusion to the chapter mentions other scales and quantity of social relationships (e.g., num- that were considered for inclusion; it may be of bers of relatives or friends) and the interconnect- value to researchers unable to find what they re- edness of the person’s network (how closely the quire in the main review section. person’s friends know each other). Perennial is- sues in the measurement of social support in- clude whether it is the number of social contacts References a person has, or their quality, that is more im- (1) Russell RD. Social health: an attempt to portant; and how to compare the value of formal clarify this dimension of well-being. Int J affiliations and informal friendships. The em- Health Educ 1973;16:74–82.  Table 4.1 Comparison of the Quality of Social Health Measurements* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurement Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Social Relationship Scale ordinal 6 research self, few * ** * * (McFarlane, 1981) interviewer Social Support Questionnaire ordinal 27 research self several ** ** ** ** (Sarason, 1983) RAND Social Health Battery ordinal 11 survey self few * * * * (RAND, 1978) MOS Social Support Survey ordinal 20 survey self few * ** ** ** (Sherbourne and Stewart, 1991) Duke-UNC Functional Social ordinal 8 clinical self (3 min) few * * ** * Support Questionnaire (Broadhead, 1988) Duke Social Support and Stress ordinal 24 research self (15 min) several ** ** ** * Scale (Parkerson, 1989) 154 Katz Adjustment Scales ordinal 205** clinical self many ** ** ** ** (Katz, 1963) (45–60 min) Social Functioning Schedule ordinal 121 clinical expert few * * ** ** (Remington and Tyrer, 1979) (20 min) Interview Schedule for Social ordinal 52 research, interviewer several ** ** *** ** Interaction (Henderson, 1980) clinical (45 min) Social Adjustment Scale—Self ordinal 42 clinical self (15–20 min) many ** ** ** ** Report (Weissman, 1971) interviewer (45–60 min) Social Maladjustment ordinal 42 clinical, interviewer few * * * * Schedule (Clare, 1978) survey (45 min) Social Dysfunction Rating ordinal 21 research staff (30 min) several * ** * ** Scale (Linn, 1969) Structured & Scaled Interview ordinal 60 clinical, interviewer several * ** * * to Assess Maladjustment research (SSIAM) (Gurland, 1972) * For an explanation of the categories used, see Chapter 1, pages 6–7. ** There are 205 items in the five sections of this instrument, but the questions can be answered twice, once by the patient and once by a relative.  Social Health 155 (2) Linn MW. Assessing community (17) Sherbourne CD, Stewart AL. The MOS adjustment in the elderly. In: Raskin A, Social Support Survey. Soc Sci Med Jervik LF, eds. Assessment of psychiatric 1991;32:705–714. symptoms and cognitive loss in the elderly. Washington, DC: Hemisphere Press, 1979:187–204. The Social Relationship Scale (3) Berger DG, Rice CE, Sewall LG, et al. (Allan H. McFarlane, 1981) Posthospital evaluation of psychiatric patients: the Social Adjustment Inventory Purpose Method. Psychiatr Studies Projects The Social Relationship Scale (SRS) was devel- 1964;2:2–30. oped to measure the extent of an individual’s (4) Antonovsky A. Health, stress, and coping. network of social relationships and its perceived San Francisco: Jossey-Bass, 1980. helpfulness in cushioning the effects of life (5) Berkman LF, Breslow L. Health and ways stresses on health (1). This social support scale of living: the Alameda County Study. New was intended primarily as a research instrument York: Oxford University Press, 1983. for use in studies of life events in general popula- (6) Murawski BJ, Penman D, Schmitt M. tion samples. Social support in health and illness: the concept and its measurement. Cancer Nurs Conceptual Basis 1978;1:365–371. The notion that social support is a buffer against (7) Weissman MM, Sholomskas D, John K. The assessment of social adjustment: an update. disease formed the stimulus for the development Arch Gen Psychiatry 1981;38:1250–1258. of this scale: social bonds are considered neces- (8) Ruesch J, Brodsky CM. The concept of sary for the individual to cope with adverse social disability. Arch Gen Psychiatry events. The scale was designed to summarize the 1968;19:394–403. qualitative and quantitative aspects of a person’s (9) Platt S. Social adjustment as a criterion of network of relationships that help him to deal treatment success: just what are we mea- with stresses (1). suring? Psychiatry 1981;44:95–112. (10) Broadhead WE, Kaplan BH, James SA, et Description al. The epidemiologic evidence for a The SRS is a self-administered scale that is intro- relationship between social support and duced by a trained interviewer who orients the health. Am J Epidemiol respondent and who prompts the respondent at 1983;117:521–537. (11) Mitchell RE, Billings AG, Moos RH. Social the end to review any relationships that may support and well-being: implications for have been forgotten. The scale originally formed prevention programs. J Primary Prev one section in a larger questionnaire concerned 1982;3:77–98. with life changes and emotional well-being in (12) Bruhn JG, Philips BU. Measuring social which the respondent was asked to identify the support: a synthesis of current approaches. people who supported him in each of six areas in J Behav Med 1984;7:151–169. which he had experienced life changes (2). The (13) Lin N, Dean A, Ensel WM. Social support SRS can also be used as a social support indica- scales: a methodological note. Schizophr tor on its own. Bull 1981;7:73–89. The scale covers six areas of life change, us- (14) Berle BB, Pinsky RH, Wolf S, et al. A ing the same question stem and response scale clinical guide to prognosis in stress for each. The six areas of life change include: diseases. JAMA 1952;149:1624–1628. (15) Bowlby J. Attachment and loss. Vol. I, work-related events, changes in monetary and fi- Attachment. London: Hogarth, 1969. nancial situation, events in the home and family, (16) Weiss RS. The provisions of social personal health events, personal and social relationships. In: Rubin Z, ed. Doing unto events, and society in general. The format of the others. Englewood Cliffs, NJ: Prentice- scale is shown in Exhibit 4.1. The scale shown in Hall, 1974:17–26. the Exhibit is applied six times, referring to each Exhibit 4.1 Format of the Social Relationship Scale Example 1: Home and Family Please list the people with whom you generally discuss home and family, using the first name or initials only. After each name or set of initials fill in a one- or two-word description of the relation each person has to you. Then go on to check the circle which indicates the degree of helpfulness or unhelpfulness of your discussions with each person, and lastly, check off yes or no if you feel this person would come to you to discuss home and family. Don’t feel you have to fill up all the spaces provided. If you find you need more spaces, please inform the interviewer. Reproduced from McFarlane AH, Neale KA, Norman GR, Roy RJ, Streiner DL. Methodological issues in developing a scale to measure social support. Schizophr Bull 1981;7:91. By permission of Oxford University Press. 156  Social Health 157 of the six topics. Respondents record the initials Reference Standards of the person they talked to and indicate the McFarlane et al. provided descriptive statistics by type of relationship (e.g., spouse, close family, sex and marital status for the SRS scores, derived distant family, friend, fellow worker, profes- from a general population sample of 518 respon- sional). They then rate the helpfulness of the dents (1, Table 5). discussion on a seven-point scale. They also rate whether that person would come to them to dis- Commentary cuss similar problems to indicate reciprocity in This brief rating scale provides more informa- the relationship (2). tion than most social support measures. It covers Three scores may be calculated. The quality both the quantity of social contacts and their of the network is estimated from the average of supportive quality and deals with giving as well the seven-point helpfulness ratings, while the ex- as receiving support. It also covers potential neg- tent of the network is estimated from a count of ative aspects of relationships and satisfaction. the total number of different individuals the re- The structure of the questionnaire is similar to spondent mentions (3). A score reflecting the de- that used in Part I of the Personal Resource gree of reciprocity is established by counting the Questionnaire developed by Brandt and Weinert number of people named who the respondent (4). McFarlane et al. used the SRS in a study of thinks would come to him to discuss similar reactions to life events and drew several conclu- problems. McFarlane et al. designated a relation- sions concerning the role of social support. For ship as multiplex if a support person was named example, the quality of social supports (i.e., the in three separate problem areas (2). helpfulness of relationships) had greater impact than quantity (2); Henderson made the same Reliability point in discussing the Interview Schedule for Test-retest reliability was assessed on 73 students Social Interaction (5). Those who felt least after a one-week interval. Reliability correlations helped by their social networks had larger net- for the size of network ranged from 0.62 to 0.99, works, made more contact with them, and re- with a median of 0.91 (1, p92). Correlations for ported more stressful events in their current, as the quality score were lower, ranging from 0.54 well as their past life (2). This is a well-designed to 0.94, giving a median of 0.78 (1, p93). and promising scale that has unfortunately not been tested further. Validity Content validity was ensured through a review by four psychiatrists whose recommendations for im- References provements were incorporated in the scale. Dis- (1) McFarlane AH, Neale KA, Norman GR, et criminant validity was assessed by comparing 15 al. Methodological issues in developing a couples with known marital or family problems scale to measure social support. Schizophr with 18 couples judged to communicate effectively Bull 1981;7:90–100. with each other. The scale showed significant dif- (2) McFarlane AH, Norman GR, Streiner DL, ferences in ratings between these groups (1). et al. Characteristics and correlates of Response bias was also examined to ascertain effective and ineffective social supports. J whether respondents tended to give socially de- Psychosom Res 1984;28:501–510. sirable replies. This was tested on 19 postgradu- (3) McFarlane AH, Norman GR, Streiner DL, et al. The process of social stress: stable, ate students by altering the question stem so as reciprocal, and mediating relationships. J to deliberately encourage a biased response, and Health Soc Behav 1983;24:160–173. then assessing how far this differed from the re- (4) Brandt PA, Weinert C. The PRQ—a social sponses given with the standard question stem. support measure. Nurs Res The results suggested that the standard wording 1981;30:277–280. showed significantly less bias toward a socially (5) Henderson AS, Brown GW. Social support: desirable response in all areas (1). the hypothesis and the evidence. In: 158 Measuring Health Henderson AS, Burrows GD, eds. structions and questions are shown in Exhibit Handbook of social psychiatry. 4.2. Amsterdam: Elsevier, 1988:73–85. A support score for each item is the number of support persons listed (the “number score”). The mean of these scores across the 27 items The Social Support Questionnaire gives an overall support score (SSQN). An over- (Irwin G. Sarason, 1983) all satisfaction score (SSQS) is based on the Purpose mean of the 27 satisfaction scores. The Social Support Questionnaire (SSQ) is in- tended to quantify the availability of, and satis- Reliability faction with, social support (1). It was designed For the number scores, inter-item correlations primarily as a research instrument and can be ranged from 0.35 to 0.71, with a mean of 0.54. used with any type of respondent. Corrected item-total correlations ranged from 0.51 to 0.79; the alpha coefficient of internal re- Conceptual Basis liability was 0.97. For the satisfaction scores, As with McFarlane’s scale, development of this inter-item correlations ranged from 0.21 to instrument was stimulated by the numerous 0.74, with an alpha of 0.94 (N = 602) (1, p130). studies that link social support with health. Four-week test-retest correlations of 0.90 for Sarason et al. noted that social support con- the SSQN score and 0.83 for the SSQS were ob- tributes to positive adjustment and personal de- tained from 105 students (1, p130). A further velopment and provides a buffer against the study of undergraduate students examined retest effects of stress (1). After reviewing alternative correlations 2 and 36 months after the initial as- conceptual approaches to social support, Sara- sessment. Results for the SSQN were 0.78 and son et al. focused on two central elements in the 0.67; figures for the SSQS were 0.86 and 0.55 concept: the perception that there are sufficient (3, Table 1). people available to help in times of need and the person’s degree of satisfaction with the support Validity available (1, 2). Sarason et al. also acknowl- Separate factor analyses were performed for the edged that perceptions of social support reflect two scores. In both cases, a strong first factor aspects of personality, correlating positively with was identified, accounting for 82% of the vari- extraversion and negatively with neuroticism, ance in the number score and 72% in the satis- depression and hostility, for example (3, p845). faction score (1). Sarason et al. concluded that the two scores represent different dimensions of Description social support. Supporting this view, the correla- The SSQ is a 27-item self-administered scale; a tion between the number and satisfaction scores homogeneous set of items was drawn from a has been studied in several samples and is low, larger pool by discarding those with low inter- ranging from 0.21 to 0.34 (1, pp130–131). Mc- correlations (1). Each question requires a two- Cormick et al. ran factor analyses of the SSQS part answer: respondents are asked to list people and SSQN scores along with scores from other to whom they could turn and on whom they support scales. The network size and satisfaction could rely in specified sets of circumstances scores fell on separate factors (4, Table 1). (availability of support), and to rate how satisfied Criterion validity was studied in samples of they are with the available support (satisfaction). psychology students. Significant negative corre- A maximum of nine people can be listed as sup- lations were obtained between the SSQ and a de- ports for each topic, their identity being indicated pression scale (correlations ranged from −0.22 by their initials and relationship to the respon- to −0.43) (1, Table 2). For females only, both dent. The satisfaction rating is the same for each scales of the SSQ correlated negatively with hos- item and uses a six-point scale running from tility and lack of protection scales; for both “very satsified” to “very dissatisfied.” The in- sexes, there was a slight, but not significant, cor- Exhibit 4.2 The Social Support Questionnaire Note: The answer categories and the satisfaction rating are the same for all questions and are therefore shown only for the first question in the exhibit. The following questions ask about people in your environment who provide you with help or support. Each question has two parts. For the first part, list all the people you know, excluding yourself, whom you can count on for help or support in the manner described. Give the person’s initials and their relationship to you (see example). Do not list more than one person next to each of the letters beneath the question. For the second part, circle how satisfied you are with the overall support you have. If you have no support for a question, check the words “No one,” but still rate your level of satisfaction. Do not list more than nine persons per question. Please answer all questions as best you can. All your responses will be kept confidential. EXAMPLE Who do you know whom you can trust with information that could get you in trouble? No one 1) T.N. (brother) 4) T.N. (father) 7) 2) L.M. (friend) 5) L.N. (employer) 8) 3) R.S. (friend) 6) 9) How satisfied? 6—very 5—fairly 4—a little 3—a little 2—fairly 1—very satisfied satisfied satisfied dissatisfied dissatisfied dissatisfied 1. Whom can you really count on to listen to you when you need to talk? No one 1) 4) 7) 2) 5) 8) 3) 6) 9) How satisfied? 6—very 5—fairly 4—a little 3—a little 2—fairly 1—very satisfied satisfied satisfied dissatisfied dissatisfied dissatisfied 2. Whom could you really count on to help you if a person whom you thought was a good friend insulted you and told you that he/she didn’t want to see you again? 3. Whose lives do you feel that you are an important part of? 4. Whom do you feel would help you if you were married and had just separated from your spouse? 5. Whom could you really count on to help you out in a crisis situation, even though they would have to go out of their way to do so? 6. Whom can you talk with frankly, without having to watch what you say? 7. Who helps you feel that you truly have something positive to contribute to others? 8. Whom can you really count on to distract you from your worries when you feel under stress? 9. Whom can you really count on to be dependable when you need help? 10. Whom could you really count on to help you out if you had just been fired from your job or expelled from school? 11. With whom can you totally be yourself? 12. Whom do you feel really appreciates you as a person? 13. Whom can you really count on to give you useful suggestions that help you to avoid making mistakes? 14. Whom can you count on to listen openly and uncritically to your innermost feelings? 15. Who will comfort you when you need it by holding you in their arms? (continued) 159 160 Measuring Health Exhibit 4.2 16. Whom do you feel would help if a good friend of yours had been in a car accident and was hospitalized in serious condition? 17. Whom can you really count on to help you feel more relaxed when you are under pressure or tense? 18. Whom do you feel would help if a family member very close to you died? 19. Who accepts you totally, including both your worst and your best points? 20. Whom can you really count on to care about you, regardless of what is happening to you? 21. Whom can you really count on to listen to you when you are very angry at someone else? 22. Whom can you really count on to tell you, in a thoughtful manner, when you need to improve in some way? 23. Whom can you really count on to help you feel better when you are feeling generally down-in-the-dumps? 24. Whom do you feel truly loves you deeply? 25. Whom can you count on to console you when you are very upset? 26. Whom can you really count on to support you in major decisions you make? 27. Whom can you really count on to help you feel better when you are very irritable, ready to get angry at almost anything? Reproduced from the Social Support Questionnaire obtained from Dr. Irwin G Sarason. With permission. relation (range, 0.16–0.24) between the satisfac- on the SSQN and SSQS scales; they are items 9, tion score and a social desirability scale (1, Table 17, 19, 20, 23, and 25 in Exhibit 4.2. Alpha in- 2). A correlation of 0.57 was obtained between ternal consistency of the SSQ-6 was 0.90 and the satisfaction score and an optimism scale, 0.93 in two samples; correlations with the full while the number score correlated 0.34 (1, SSQ (less the six common items) were 0.95 for p132). Correlations between the SSQN and a SSQN and 0.96 for SSQS. Correlations with the measure of parental support ranged from 0.26 Beck Depression Inventory and with other social to 0.42 in a sample of undergraduates; equiva- support measures were high and similar to those lent results for the SSQS ranged from 0.28 to of the full SSQ. The abbreviated version appears 0.52 (3, Table 4). appropriate for use when time constraints do not With 295 students, SSQN and SSQS scores permit use of the complete scale. Translations rose with the numbers of positive life events (1). exist in Dutch, German, Spanish, Chinese, and Those with more social support also felt more Japanese. able to control the occurrence of life events (1, As a by-product of their work on developing Table 3). In a study of 163 men in military train- the SSQ, Sarason et al. developed a Dyadic Effec- ing, respondents who had many negative life tiveness Scale that records judgments of how ef- events and less support showed a higher fre- fective a person would be in forming social quency of chronic illness than other groups (2). relationships (3, Table 5). Sarason et al. found significant agreement be- tween an experimenter’s rating of the respon- Commentary dent’s social competence and the number score; The SSQ seems to be a valid and reliable scale, those with high and SSQ scores differed signifi- although the evidence is not extensive and most cantly in ratings of loneliness and social compe- validation studies have been undertaken by the tence (5). original authors. Considerable reliance was placed on psychology students in testing the in- Alternative Forms strument and it will be important to assess how In 1987, Sarason et al. described a six-item ab- it performs with other samples and how it corre- breviation of the SSQ (6). The six items were se- lates with other social support scales. The item lected through factor analyses as loading highly selection was based largely on internal consis-  Social Health 161 tency, which may provide a coherent instrument (4) McCormick IA, Siegert RJ, Walkey FH. at the expense of breadth of coverage, as is sug- Dimensions of social support: a factorial gested by the single factor result in the factor an- confirmation. Am J Commun Psychol alytic study. The SSQ principally covers 1987;15:73–77. appraisal and emotional support, and has little (5) Sarason BR, Sarason IG, Hacker TA, et al. Concomitants of social support: social coverage of instrumental or practical support. skills, physical attractiveness, and gender. J The response categories used in assessing so- Pers Soc Psychol 1985;49:469–480. cial support vary from instrument to instrument. (6) Sarason IG, Sarason BR, Shearin EN, et al. The Medical Outcomes Study (MOS) Social A brief measure of social support: practical Support Scale asks how much of the time each and theoretical implications. J Soc Personal form of support is available; McFarlane’s scale Relations 1987;4:497–510. counts the helpfulness of each supportive per- son, whereas Sarason counts the number of peo- The RAND Social Health Battery ple available to help and the perceived adequacy (RAND Corporation, 1978) of this support. This diversity reflects the diffi- culty of selecting appropriate answer categories; Purpose it is not certain that counting numbers of people The RAND Social Health Battery records re- available (number score) is the most relevant in- sources for social support and the frequency of dicator. Perhaps the link between numbers of social interactions; it does not rate the subjective contacts and perceived support is not linear, in experience of support. It is intended for use in that having too few people and also reporting general population surveys. large numbers under every category might both indicate problems. Furthermore, the number Conceptual Basis score does not reflect the extent of overlap be- Originally, Donald and Ware used the concepts tween people identified in different questions, so of social well-being and support interchange- does not indicate the overall size of the network ably (1, 2), but later distinguished between so- or capture McFarlane’s theme of multiplexity. cial functioning, role functioning, and social Even though asking about who provides support support; a series of measures was developed to and about satisfaction for each question length- assess with each of these constructs. The in- ens the instrument, it is likely that having the re- strument reviewed here forms an overall mea- spondent think about all the people who provide sure of social functioning, defined as “the support improves the accuracy of reports of sat- ability to develop, maintain, and nurture major isfaction. social relationships” (3, p173). This may be The SSQ appears to offer a sound, but longer, measured in terms of relatively objective behav- alternative to the MOS instrument described ioral indicators such as the numbers of social later in this chapter. resources a person has, or the frequency of con- tact with friends and relatives (1, 2). Social sup- port may be independent of social functioning, References because a person may have good social func- (1) Sarason IG, Levine HM, Basham RB, et al. tioning yet derive little support, although con- Assessing social support: the Social versely a chronically ill person who is unable to Support Questionnaire. J Pers Soc Psychol function socially may receive strong support 1983;44:127–139. from family or relatives (3). Likewise, social (2) Sarason IG, Sarason BR, Potter EH, et al. functioning in personal relationships was dis- Life events, social support, and illness. Psychosom Med 1985;47:156–163. tinguished from role functioning; a separate (3) Sarason IG, Sarason BR, Shearin EN. Social measure of role functioning was developed (4). support as an individual difference variable: Finally, a separate four-item scale focused on its stability, origins, and relational aspects. J restrictions in social functioning produced by Pers Social Psychol 1986;50:845–855. illness (3). 162 Measuring Health Description group participation, and 0.68 for the overall in- This self-administered scale was developed along dex (1, Table 11). The corresponding one-year with the RAND physical and psychological test-retest coefficients were 0.55, 0.68, and 0.68 scales as an outcome measurement for the (coefficients for individual items ranged from Health Insurance Experiment. The 11 items in- 0.23–0.80). clude predominantly objective indicators cover- ing social resources (e.g., number of friends) and Validity contacts (e.g., the frequency of seeing friends or Preliminary validation results were drawn from involvement in group activities). The scale cov- 4,603 interviews in the Health Insurance Exper- ers home and family, friendships, and social and iment. Correlations were calculated between community life; it specifically excludes work- each item and three criterion scores: a nine-item related performance and activities that need not self-rating of health in general, a three-item involve interaction, such as attending sports measure of emotional ties, and a nine-item psy- events (2). The scale does not cover satisfaction chological well-being scale. The correlations with relationships either. The development of the were low, with only three of 33 correlations questionnaire is described by Donald and Ware equal to or above 0.20 (1, Table 4). Correlations (2), and it is shown in Exhibit 4.3. for the three aggregated indices were somewhat Forced choice and open-ended responses are higher; the overall index correlated 0.32 with used. A scoring format developed by Donald and the psychological well-being scale and 0.20 Ware is used to recode the printed response op- with emotional ties (1, Table 13). The overall tions: this is shown in Exhibit 4.4. High scores score was found to explain 12% of the variance indicate more extensive social contacts, although in mental health as measured by the RAND the authors give no guidance about threshold Mental Health Inventory (5). The question on scores that would distinguish good from poor writing letters did not correlate with the crite- adjustment. On the basis of factor analyses, the rion scores and was not used further in analyses items may be grouped to form two subscales and of the scale. (Perhaps a question on e-mail cor- an overall score (1). The first subscale, social respondence might succeed in contemporary so- contacts, includes the third to fifth items from ciety?) Exhibit 4.3; a group participation scale includes For a sample of 256 patients with multiple items ten and 11. An overall score uses all the sclerosis, the social index scores showed moder- items except for seven (writing letters) and eight ate deterioration as disease severity increased (getting along with others), although the authors (Spearman rho −0.31) (6, p307). recommend using scores for individual items and subscales rather than the overall score, pending Reference Standards additional validity studies (1). They also recom- Table 7 in Donald and Ware’s report shows the mend standardizing items to a mean of zero and response patterns for ten items for 4,603 respon- a standard deviation of one before forming sub- dents from the RAND study (1). scores (2). Item seven (writing letters) was dropped from analyses of scale results because Commentary few people answered affirmatively and it did not This scale was based on clear conceptual design correlate with the total score. It could be deleted and on an extensive review of social health mea- (C. D. Sherbourne, personal communication, surements and was designed to reflect areas iden- 1994). tified as important by then current literature (7). It is one of the few scales we review that was not Reliability designed for use with patients, and the authors The inter-item correlations are low, with only 5 made some interesting observations on the point of 45 correlations exceeding 0.40 (1, Table 8). beyond which an increase in social contacts may Internal consistency coefficients for the three not bring additional benefits to a person’s well- subscores were 0.72 for social contacts, 0.84 for being. Exhibit 4.3 The RAND Social Health Battery 1. About how many families in your neighborhood are you well enough acquainted with, that you visit each other in your homes? ______ families 2. About how many close friends do you have—people you feel at ease with and can talk with about what is on your mind? (You may include relatives.) (Enter number on line) ______ close friends 3. Over a year’s time, about how often do you get together with friends or relatives, like going out together or visiting in each other’s homes? (Circle one) Every day 1 Several days a week 2 About once a week 3 2 or 3 times a month 4 About once a month 5 5 to 10 times a year 6 Less than 5 times a year 7 4. During the past month, about how often have you had friends over to your home? (Do not count relatives.) (Circle one) Every day 1 Several days a week 2 About once a week 3 2 or 3 times in past month 4 Once in past month 5 Not at all in past month 6 5. About how often have you visited with friends at their homes during the past month? (Do not count relatives.) (Circle one) Every day 1 Several days a week 2 About once a week 3 2 or 3 times in past month 4 Once in past month 5 Not at all in past month 6 6. About how often were you on the telephone with close friends or relatives during the past month? (Circle one) Every day 1 Several times a week 2 About once a week 3 2 or 3 times 4 Once 5 Not at all 6 7. About how often did you write a letter to a friend or relative during the past month? (Circle one) Every day 1 Several times a week 2 About once a week 3 2 or 3 times in past month 4 Once in past month 5 Not at all in past month 6 8. In general, how well are you getting along with other people these days—would you say better than usual, about the same, or not as well as usual? (Circle one) Better than usual 1 About the same 2 Not as well as usual 3 (continued) 163 164 Measuring Health Exhibit 4.3 9. How often have you attended a religious service during the past month? (Circle one) Every day 1 More than once a week 2 Once a week 3 2 or 3 times in past month 4 Once in past month 5 Not at all in past month 6 10. About how many voluntary groups or organizations do you belong to—like church groups, clubs or lodges, parent groups, etc. (“Voluntary” means because you want to.) ______ groups or organizations (Write in number. If none, enter “0.”) 11. How active are you in the affairs of these groups or clubs you belong to? (If you belong to a great many, just count those you feel closest to. If you don’t belong to any, circle 4.) (Circle one) Very active, attend most meetings 1 Fairly active, attend fairly often 2 Not active, belong but hardly ever go 3 Do not belong to any groups or clubs 4 Reproduced from Donald CA, Ware JE, Jr. The measurement of social support. Res Commun Ment Health 1984;4:334–335. With permission. The preliminary testing of the method had does not correlate with something it is supposed the advantage of a large representative sample, to differ from does not prove that it would cor- but the design of the scale complicated the vali- relate with another scale closer in meaning. The dation process. That is, items were deliberately results so far published do not suggest high levels chosen to represent a concept of social health in- of validity or reliability, and further studies are dependent of physical and psychological well- required to indicate how the instrument com- being, so that the low concurrent validity pares to alternative social health measurements, correlations may be expected. This is a dilemma and how well it agrees with assessments made by of discriminant validity: showing that a scale independent observers. Given the current slender Exhibit 4.4 Scoring Method for the RAND Social Health Battery Abbreviated item content Recoding rule Neighborhood family acquaintances (0 = 0) (1 = 1) (2 = 2) (3 = 3) (4 = 4) (5 thru 10 = 5) (11 or higher = 6) Close friends and relatives (0 = 0) (1 = 1) (2 = 2) (3 = 3) (4 = 4) (5 thru 9 = 5) (10 thru 20 = 6) (21 thru 25 = 7) (26 thru 35 = 8) (36 or higher = 9) Visits with friends/relatives (1 thru 3 = 4) (4= 3) (5,6 = 2) (7 = 1) Home visits by friends (1 thru 4 = 3) (5 = 2) (6 = 1) Visits to homes of friends (1 thru 3 = 3) (4,5 = 2) (6 = 1) Telephone contacts (1 = 5) (2 = 4) (3,4 = 3) (5 = 2) (6 = 1) Getting along (1 = 3) (2 = 2) (3 = 1) Attendance at religious services (1,2 = 5) (3 = 4) (4 = 3) (5 = 2) (6 = 1) Voluntary group membership (0 = 0) (1 = 1) (2 = 2) (3 = 3) (4 = 4) (5 or higher = 5) Level of group activity (1 = 4) (2 = 3) (3 = 2) (4 = 1) Reproduced from Donald CA, Ware JE, Jr. The measurement of social support. Res Commun Ment Health 1984;4:350, Table 6. With permission.  Social Health 165 evidence for validity and reliability, we recom- categories of social support. It is intended for use mend that potential users of the scale first check in survey research with people with chronic ill- for additional evidence on its quality. Readers ness, but it can be used with general population may also consider the MOS Social Support Sur- samples (1). vey that we review next. It was developed by the same team and examines the functional and Conceptual Basis structural aspects of support. The RAND and Medical Outcomes Study (MOS) teams developed several measures of so- cial health, including the social interaction mea- References sure described earlier in this chapter, a measure (1) Donald CA, Ware JE, Jr. The measurement of social role functioning (2), and the present of social support. Res Community Ment measure of social support. Social functioning Health 1984;4:325–370. refers to the ability to establish and maintain (2) Donald CA, Ware JE, Jr. The quantification major social relationships; a person may derive of social contacts and resources. (R-2937- support from these, but the connection is not HHS). Santa Monica, CA: RAND strong because social relationships need not al- Corporation, 1982. ways be supportive (3). Support includes tangi- (3) Sherbourne CD. Social functioning: social ble and emotional support, and empirical activity limitations measure. In: Stewart evidence suggests that these help people cope AL, Ware JE, Jr, eds. Measuring with stress or illness, although the mechanisms functioning and well-being: the Medical Outcomes Study approach. Durham, NC: involved are not fully clear. Duke University Press, 1992:173–181. Existing measures of social support generally (4) Sherbourne CD, Stewart AL, Wells KB. cover structural aspects (e.g., size of social net- Role functioning measures. In: Stewart AL, work, how closely the friends know each other) Ware JE, Jr, eds. Measuring functioning or the functional aspects (e.g., perception of be- and well-being: the Medical Outcomes ing supported). Functional support appears the Study approach. Durham, NC: Duke most important and can be of various types: pro- University Press, 1992:205–219. viding emotional support, love and empathy; (5) Williams AW, Ware JE, Jr, Donald CA. A providing instrumental or tangible support; pro- model of mental health, life events, and social viding information, guidance, or feedback; ap- supports applicable to general populations. J praisal support that helps the person evaluate Health Soc Behav 1981;22:324–336. (6) Harper AC, Harper DA, Chambers LW, et herself; and companionship in leisure and recre- al. An epidemiological description of ational activities (1, p705). The focus on per- physical, social and psychological problems ceived support is justified because “the fact that in multiple sclerosis. J Chronic Dis a person does not receive support during a given 1986;39:305–310. time period does not mean that the person is un- (7) Donald CA, Ware JE, Jr, Brook RH, et al. supported. Received support is confounded with Conceptualization and measurement of need and may not accurately reflect the amount health for adults in the Health Insurance of support that is available to a person” (1, Study. Vol. IV, Social health. Santa Monica, p706). CA: RAND Corporation, 1978. Description An initial pool of 50 functional support items The MOS Social Support Survey was reduced to 19 that were posited to cover (Cathy Sherbourne and Anita five dimensions: emotional support, informa- Stewart, 1991) tional support, tangible support, positive social Purpose interaction, and affection. To reduce respon- The Social Support Survey offers a brief, self- dent burden, the scale does not ask about who administered indicator of the availability of four provides the support; each question asks about 166 Measuring Health how often each form of support is available to that the four subscales were internally consistent them. One structural support item asks about and distinct from each other. Correlations be- the respondent’s number of close friends or rel- tween the four subscales ranged from 0.69 to atives. 0.82 (1, p710). The first item, the single measure The instrument is self-administered and uses of structural support, showed low correlations five-point answer scales (Exhibit 4.5). Empirical with the four subscale scores (range, 0.18–0.24) analyses indicated that the emotional and infor- (1, p709). mational support items should be scored to- gether, so four subscales are derived: tangible Reference Standards support (items 2, 5, 12, 15), affectionate (items Sherbourne and Stewart reported mean scores 6, 10, 20), positive social interaction (items 7, and standard deviations for each item taken 11, 14, 18), and emotional or informational sup- from 2,987 MOS participants; these represent port (items 3, 4, 8, 9, 13, 16, 17, and 19). Sub- an ambulatory sick population, each of whom scale scores sum the responses checked for the had screened positive for one of four medical relevant items; scores are rescaled to a 0 to 100 conditions (1, Table 2). range for each subscale, with higher scores indi- cating more support. A total score is calculated Commentary from the mean of the subscale scores, although The MOS questionnaire was carefully devel- Sherbourne and Stewart recommend using the oped from previous instruments and was based subscale scores rather than the total (1, p712). on a sound theoretical formulation. Sherbourne Item 1 is not included in the subscores. Further and Stewart criticized existing support scales as information is available from the RAND web weak in design, narrow in content, and unim- site (www.rand.org/health/surveys/mos.descrip pressive in psychometric properties. There is .html). still a need for multidimensional instruments that are psychometrically sound yet relatively Reliability short. The preliminary evidence for reliability Internal consistency for the overall scale was and validity is impressive. The criterion validity high (alpha = 0.97) and values for the subscales coefficients are logical and higher than those ranged from alpha = 0.91 to 0.96 in the MOS. for other scales. Although the scale was de- Item-scale correlations all exceeded 0.72 (1, signed for use in a study of chronically ill pa- pp709–710). One year test-retest reliability was tients living in the community, items are also high at 0.78 (0.72–0.76 for each subscale) universally applicable. We do not yet have in- (1, Table 3). formation on the validity of the scale in a gen- eral population sample, but it should be Validity carefully considered for use in surveys and Criterion validity was tested using variables in- epidemiological studies of chronic disease etiol- cluded in the MOS. The Social Support Survey ogy. This instrument demonstrates that func- showed significant convergent correlations with tional social support is distinct from the loneliness (r = −0.53 to −0.69), marital and fam- structural aspects of support, a distinction like ily functioning (0.38–0.57), and mental health that between availability of support and its ade- (0.36–0.45) (1, Table 4). Discriminant correla- quacy (see the Interview Schedule for Social In- tions ranged from −0.14 to −0.30 with physical teraction). symptoms and role limitations and −0.14 to −0.21 with pain severity. Correlations with indi- cators of social activity were intermediate, rang- References ing from 0.24 to 0.33 (1, Table 4). (1) Sherbourne CD, Stewart AL. The MOS Factor analyses confirmed that the 19 items Social Support Survey. Soc Sci Med could reasonably form an overall index and also 1991;32:705–714.  Social Health 167 Exhibit 4.5 The Medical Outcomes Study Social Support Survey Next are some questions about the support that is available to you. 1. About how many close friends and close relatives do you have (people you feel at ease with and can talk to about what is on your mind)? Write in number of close friends and close relatives: People sometimes look to others for companionship, assistance, or other types of support. How often is each of the following kinds of support available to you if you need it? (Circle one number on each line) None A Little Some Most All of of the of the of the of the the time time time time time 2. Someone to help you if you were confined to bed . . . . . . . 1 2 3 4 5 3. Someone you can count on to listen to you when you need to talk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 4. Someone to give you good advice about a crisis . . . . . . . . . 1 2 3 4 5 5. Someone to take you to the doctor if you needed it . . . . . . 1 2 3 4 5 6. Someone who shows you love and affection . . . . . . . . . . . . 1 2 3 4 5 7. Someone to have a good time with . . . . . . . . . . . . . . . . . . 1 2 3 4 5 8. Someone to give you information to help you understand a situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 9. Someone to confide in or talk to about yourself or your problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 10. Someone who hugs you . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 11. Someone to get together with for relaxation . . . . . . . . . . . 1 2 3 4 5 12. Someone to prepare your meals if you were unable to do it yourself . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 13. Someone whose advice you really want . . . . . . . . . . . . . . . 1 2 3 4 5 14. Someone to do things with to help you get your mind off things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 15. Someone to help with daily chores if you were sick . . . . . . 1 2 3 4 5 16. Someone to share your most private worries and fears with . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 17. Someone to turn to for suggestions about how to deal with a personal problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 18. Someone to do something enjoyable with . . . . . . . . . . . . . 1 2 3 4 5 19. Someone who understands your problems . . . . . . . . . . . . . 1 2 3 4 5 20. Someone to love and make you feel wanted . . . . . . . . . . . . 1 2 3 4 5 Sherbourne CD, Stewart AL. The MOS Social Support Survey. Soc Sci Med 1991;32:713–714. With permission. (2) Sherbourne CD, Stewart AL, Wells KB. (3) Sherbourne CD. Social functioning: social Role functioning measures. In: Stewart AL, activity limitations measure. In: Stewart Ware JE Jr, eds. Measuring functioning and AL, Ware JE, Jr, eds. Measuring well-being: the Medical Outcomes Study functioning and well-being: the Medical approach. Durham, NC: Duke University Outcomes Study approach. Durham, NC: Press, 1992:205–219. Duke University Press, 1992:173–181. 168 Measuring Health The Duke-University of North correlations were 0.62 for confidant support and Carolina (UNC) Functional Social 0.64 for affective support (1, pp714–715). Support Questionnaire (W.E. Broadhead, 1988) Validity Discriminant validity correlations for each item Purpose were derived from a sample of 401 family prac- The Duke-UNC Functional Social Support tice patients. Divergent correlations between Questionnaire (DUFSS) measures a person’s sat- confidant and affective support and physical isfaction with the functional and affective as- function ranged from 0.08 to 0.17, correlations pects of social support. It was intended for with symptom status ranged from 0.18 to 0.30, clinical use in family practice settings to identify whereas convergent correlations with emotional people at risk of isolation, and in research appli- function ranged from 0.34 to 0.41 (1, Table 4). cations to examine the interactions between so- Curiously, however, correlations with a social cial support and other determinants of health. function scale derived from the Duke Health Profile were low, at 0.15 for the affective scale Conceptual Basis and 0.17 for the confidant support scale (1, Social support has direct and buffering effects Table 6). Correlations with social functioning on health, and most research has demonstrated measures drawn from the RAND studies were that the quality of social relationships better pre- also low, although some showed an appropriate dicts health and well-being than the number of pattern. For example, the correlation of the con- friends or frequency of contact. Previous work fidant scale with a measure of social contacts had shown that quantity and quality of support (r = 0.35) was higher than that of the affective “are minimally intercorrelated and that it may support score (0.17); the equivalent correlations be inappropriate to combine them into summary with a question on socializing with other people measures” (1, p710). The DUFSS covers the were 0.29 and 0.22 (1, Table 6). The overall im- qualitative, or functional, aspects of support. It pression is of very low associations: correlations was originally designed to cover four content ar- with group participation for both scales were eas: relations with confidants (a relationship in 0.08, lower than the correlations with physical which important life concerns can be discussed), or mental health measures. affective support (an emotional form of caring), Factor analyses confirmed the presence of quantity of support, and instrumental assistance. two factors, with loadings ranging from 0.52 to Fourteen items were tested but those covering in- 0.72 (1, Table 3). strumental assistance and quantity of support Construct validity was assessed in various were found unreliable and were deleted, leaving ways. Level of support was found to be linked to eight items in the final instrument. number of office visits to general practitioners, such that those with low support made more vis- Description its (2, Table 3). In particular, those with low Of the eight items in the DUFSS, numbers one, confidant support made more longer-than- two, and eight cover affective support and the average office visits (2, Table 4). The association remainder cover confidant support. The five- was stronger than that between use and struc- point answer scales range from “as much as I tural measures of support (e.g., numbers of would like” to “much less than I would like” friends). The DUFSS scores showed no correla- (Exhibit 4.6). A summary score is formed by tion with demographic variables (e.g., race, age, adding item scores; subscores for affective and employment) but were correlated with whether confidant support can also be formed. the respondent lived alone (1, Table 5). Reliability Commentary Two-week test-retest reliability for the items This Duke questionnaire is the briefest of the so- ranged from 0.50 to 0.77. The average item-total cial support measures we review; its practicality  Social Health 169 Exhibit 4.6 The Duke-UNC Functional Social Support Questionnaire Here is a list of some things that other people do for us or give us that may be helpful or supportive. Please read each statement carefully and place a check (✔) in the blank that is closest to your situation. Much less Here is an example: As much as than I I get . . . I would like would like enough vacation time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . If you put a check where we have, it means that you get almost as much vacation time as you would like, but not quite as much as you would like. Answer each item as best you can. There are no right or wrong answers. Much less I get . . . As much as than I I would like would like 1. people who care what happens to me . . . . . . . . . . . . . . . . . . . . . 2. love and affection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. chances to talk to someone about problems at work or with my housework . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. chances to talk to someone I trust about my personal and family problems . . . . . . . . . . . . . . . . . . . . . . . . . . 5. chances to talk about money matters . . . . . . . . . . . . . . . . . . . . . 6. invitations to go out and do things with other people . . . . . . . . . 7. useful advice about important things in life . . . . . . . . . . . . . . . . 8. help when I’m sick in bed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adapted from Broadhead WE, Gehlbach SH, de Gruy FV, Kaplan BH. The Duke-UNC Functional Social Support Questionnaire: measurement of social support in family medicine patients. Med Care 1988;26:722–723. in a clinical setting is a strong asset. Its focus on et al. did not address this issue but presented the the quality rather than the amount of support re- correlations as statistically significant and as in- flects the trend of previous research; the items dicating that “the new scales are measuring con- are very similar to those in the MOS Social Sup- structs similar, but not identical, to the existing port Survey. The preliminary results show ade- scales” (1, p718). Perhaps the criterion measures quate reliability: item-total correlations of 0.62 against which the DUFSS was tested are not will translate into an alpha of about 0.80 to ideal, but nonetheless the shared variance be- 0.85; the retest correlations are appropriate. The tween the new and existing scales ranged from divergent validity correlations with symptoms only 0.5% to a high of 12%. Internal consis- are very similar to those reported for the MOS tency indicates that this scale is measuring some- Social Support Survey. thing, but the validity results do not clarify The convergent validity findings, however, exactly what this is. Broadhead et al. noted that cause concern: associations with other social two items (“help when sick in bed” and “invita- support indicators were low, even if statistically tions to go out”) were not predicted to load on significant owing to the large sample. Broadhead the factors on which empirical analyses placed 170 Measuring Health them. The extent to which the two factors were Description distinct was not reported, and future analyses The DUSOCS is self-administered; 12 items cov- may show less separation between them. Some ering perceived support and 12 covering stress of the items are grammatically awkward (e.g., “I are rated on three-point scales for six categories get people who care what happens to me . . . As of family members and four categories of non- much as I would like”). family members. In addition, the most support- Before this instrument can be recommended ive and most stressful relationships are identified for general use, further examination of its agree- (Exhibit 4.7). ment with other measures of social support Four scores are created: family support and should be undertaken. Readers are referred to stress, and nonfamily support and stress. Total other scales developed by this team; a review of support and stress scores can be created by the Duke Social Support and Stress Scale fol- adding family and nonfamily scores. Responses lows, and the Duke general health measurement are coded as follows: “none” = 0, “some” = 1, “a is reviewed in Chapter 10. lot” = 2, “yes” = 2, “no” = 0, and “there is no such person” = 0. Blank responses are consid- ered as 0 unless all items in the entire section (A, References B, or C) are left blank, in which case no score (1) Broadhead WE, Gehlbach SH, de Gruy FV, can be generated for that section. The family et al. The Duke-UNC Functional Social support score is calculated by summing the six Support Questionnaire: measurement of responses in section A; if the reply to section C social support in family medicine patients. identified a family member, 2 is added to the Med Care 1988;26:709–723. family support score. The resulting total is di- (2) Broadhead WE, Gehlbach SH, de Gruy FV, et al. Functional versus structural social vided by 14 and multiplied by 100 to give a 0 to support and health care utilization in a 100 score. The same approach is used in scoring family medicine outpatient practice. Med family stress and for the nonfamily scores (B), Care 1989;27:221–233. except that the total nonfamily score is divided by 10 and multiplied by 100 to provide a 0 to 100 score. The total stress and support scores The Duke Social Support are calculated by summing the raw scores in sec- and Stress Scale tions A, B, and C and dividing this total by 22 (George R. Parkerson, 1989) (i.e., 6 × 2 for A, plus 4 × 2 for B, plus 2 for C) Purpose and multiplying by 100. Higher scores indicate The Duke Social Support and Stress Scale (DU- more stress or support for the four scales. The SOCS) rates family and nonfamily relationships items on the scale are shown in Exhibit 4.7; in terms of the amount of support they provide copies of self- and interviewer-administered ver- and the amount of stress they cause. It is a fam- sions are given in the Manual of the DUSOCS ily practice research instrument to be used in (2, Figures 12 to 16). Copies of the scale, plus studying the family environment as a determi- full scoring instructions, are also available on nant of health (1). the web, at http://healthmeasures.mc.duke.edu/ images/ScoreDoc.pdf. SAS computer code for Conceptual Basis scoring is included in the User’s Manual (2, Ap- The links between stress, social support, and pendix J). health have been extensively studied, and Park- erson et al. addressed the role that family mem- Reliability bers play in this process, placing an emphasis Two-week test-retest correlations were 0.76 for on the person’s perceptions of the supportive- family support, 0.67 for nonfamily support, ness or stressfulness of their relationships (1, 0.68 for nonfamily stress, but only 0.40 for fam- p218). ily stress (1, p222). The sample was, however, Exhibit 4.7 The Duke Social Support and Stress Scale I. People Who Give Personal Support (A supportive person is one who is helpful, will listen to you or who will back you up when you are in trouble.) Instructions: Please look at the following list and decide how much each person (or group of persons) is supportive for you at this time in your life. Check your answer. A. Family Members How supportive are these people now: There is No Such None ____________ Some ____________ A Lot _____________ Person ____________ 1. Your wife, husband, or significant other person . . . . . . . . . ____________ ____________ _____________ ____________ 2. Your children or grandchildren . . . ____________ ____________ _____________ ____________ 3. Your parents or grandparents . . . . ____________ ____________ _____________ ____________ 4. Your brothers or sisters . . . . . . . . . ____________ ____________ _____________ ____________ 5. Your other blood relatives . . . . . . . ____________ ____________ _____________ ____________ 6. Your relatives by marriage (for example: in-laws, ex-wife, ex-husband) . . . . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ B. Non-Family Members 7. Your neighbors . . . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ 8. Your co-workers . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ 9. Your church members . . . . . . . . . . ____________ ____________ _____________ ____________ 10. Your other friends . . . . . . . . . . . . . ____________ ____________ _____________ ____________ C. Special Supportive Person Yes No 11. Do you have one particular person whom you trust and to whom you can go with personal difficulties? . . . . . . . . . . __________ __________ 12. If you answered “yes”, which of the above types of person is he or she? (for example: child, parent, neighbor) ____________ II. People Who Cause Personal Stress (A person who stresses you is one who causes problems for you or makes your life more difficult.) Instructions: Please look at the following list and decide how much each person (or group of persons) is a stress for you at this time in your life. Check your answer. A. Family Members How stressed do you feel by these people now: There is No Such None ____________ Some ____________ A Lot _____________ Person ____________ 1. Your wife, husband, or significant other person . . . . . . . . . ____________ ____________ _____________ ____________ 2. Your children or grandchildren . . . ____________ ____________ _____________ ____________ 3. Your parents or grandparents . . . . ____________ ____________ _____________ ____________ 4. Your brothers or sisters . . . . . . . . . ____________ ____________ _____________ ____________ 5. Your other blood relatives . . . . . . . ____________ ____________ _____________ ____________ (continued) 171 172 Measuring Health Exhibit 4.7 (continued) 6. Your relatives by marriage (for example: in-laws, ex-wife, ex-husband) . . . . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ B. Non-Family Members 7. Your neighbors . . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ 8. Your co-workers . . . . . . . . . . . . . . ____________ ____________ _____________ ____________ 9. Your church members . . . . . . . . . . ____________ ____________ _____________ ____________ 10. Your other friends . . . . . . . . . . . . . ____________ ____________ _____________ ____________ C. Most Stressful Person Yes No 11. Is there one particular person who is causing you the most personal stress now? . . . . . . . . . . . . . . . . . . . . . . . . . . . __________ __________ 12. If you answered “yes”, which of the above types of person is he or she? (for example: child, parent, neighbor) __________ Copyright © 1986 and 1995 by the Department of Community and Family Medicine, Duke University Medical Center, Durham, NC. Reproduced with permission. selected as experiencing low family stress. Test- Strengths measure, however, were higher, at retest correlations were reported from a sample 0.29 and 0.59 (1, Table 3). In a separate study, of 314 ambulatory patients: 0.58 for family the Olson instrument again correlated more stress and 0.27 for nonfamily stress, 0.73 for strongly (r = 0.17 to 0.53) with scales from the family support and 0.50 for nonfamily support. Duke measure than did the DUSOCS family sup- Alpha coefficients ranged from 0.53 to 0.70 (3, port measure (r = 0.07 to 0.33) (6, Table 1). Table 1). Retest intraclass correlations after six Again, the DUSOCS stress measure showed days were 0.92 for a self-completed version of stronger (but reversed) correlations than the sup- DUSOCS, and 0.80 for an interviewer adminis- port scores: rho ranging from 0.07 to 0.43 (6, tered version (4). Table 1). The comparison between family and nonfam- Validity ily measures showed inconsistent trends: in some Initial validity findings were based on a sample instances nonfamily scores showed stronger as- of 249 adults who went to a family medicine sociations with criterion measures whereas in center. A Spearman rho correlation of 0.43 was others the family measures were the more obtained between the DUSOCS family support strongly associated (1, Table 4; 6, Table 2). In score and Olson’s Family Strength measure; a the study of 314 ambulatory patients, DUSOCS rho of 0.45 was found between the family stress family stress scores were significantly associated score and an independent measure of intrafamily with all subscores of the DUHP (2, Table 22; 7, and marital strains (1, p222). Equivalent figures Table 1). Family stress scores also predicted from a subsequent study were 0. 51 and 0.33 (5, subsequent health care use over an 18-month pe- p690). The family stress measure correlated riod (7, Tables 2–5). A factor analysis identified −0.32 with symptom status from the Duke-UNC two factors: confidant support and affective sup- Health Profile (DUHP); the correlation with port (4). emotional function was −0.44. Correlations for the family support measures were somewhat Alternative Forms weaker, at 0.20 and 0.37, respectively. The A Portuguese translation is included in the User’s equivalent correlations for the Olson Family Manual (2, Appendix I).  Social Health 173 Reference Standards should also consider the other brief social sup- Reference values by age and gender for primary port instrument developed by the same team (see care patients were reported by Parkerson, along previous review) and their multidimensional with a table for converting raw scores into per- measurement included in Chapter 10. centile values (2, Appendices K and L). Commentary References The DUSOCS is an innovative and somewhat (1) Parkerson GR Jr, Michener JL, Wu LR, et provocative scale in combining support and al. Associations among family support, stress in the same measure. The developers con- family stress, and personal functional sider its simplicity and wide applicability as the health status. J Clin Epidemiol main advantages of the DUSOCS. 1989;42:217–229. A few concerns arise from the results of the (2) Parkerson GR Jr. User’s guide for Duke preliminary psychometric testing. The retest reli- health measures. Durham, NC: Duke ability coefficients are somewhat low, suggesting University Department of Community and Family Medicine, 2002. that perceptions of support change over time. (3) Parkerson GR Jr, Broadhead WE, Tse C- Note, however, that the items are phrased in KJ. Quality of life and functional health of terms of “this person is supportive now”: this is primary care patients. J Clin Epidemiol an acute measure of support. The correlations 1992;45:1303–1313. with the DUHP social function score and with (4) Bellon Saameno JA, Delgado, SA, Luna del the Duke social health score (see Chapter 10) do Castillo JD, et al. [Validity and reliability not give strong indications of convergent valid- of the Duke-UNC-11 questionnaire of ity. The correlations between the criterion mea- functional social support]. Aten Primaria sures and the Olson support scale exceeded 1996;18:153–163. those for the DUSOCS, suggesting that the Ol- (5) Parkerson GR Jr, Michener JL, Wu LR, et son scale may be superior. The stress component al. The effect of a telephone family assessment intervention on the functional of the DUSOCS may be more adequate than the health of patients with elevated family support component. stress. Med Care 1989;27:680–693. The potential of combining support and (6) Parkerson GR Jr, Broadhead WE, Tse C- stress indicators does not seem to have been fully KJ. Validation of the Duke Social Support exploited. The authors do not, for example, dis- and Stress Scale. Fam Med cuss the possibility of forming balance scores in 1991;23:357–360. which the stress scores are subtracted from the (7) Parkerson GR Jr, Broadhead WE, Tse C- support scores, along the lines used in Brad- KJ. Perceived family stress as a predictor of burn’s Affect Balance Scale, to provide an indi- health-related outcomes. Arch Fam Med cator of the net supportive effects of family and 1995;4:253–260. acquaintances. The resulting scores might corre- late more strongly with health outcomes than The Katz Adjustment Scales either component alone. Correlations between (Martin M. Katz, 1963) support and stress scores were not reported, so we do not know if these are the obverse of each Purpose other, or whether respondents with intense fam- Katz developed this set of scales to measure the ily relationships can report feeling both sup- social adjustment of psychiatric patients follow- ported and also stressed in their relationships. ing treatment; the scales incorporate judgments This seems a potentially exciting scale, and made by the patient and by a relative. The as- further testing, including comparisons with sessments cover psychiatric symptoms, social other social support scales, is desirable before behavior, and home and leisure activities. The the DUSOCS can be recommended as a rival to scales have also been used in population sur- some of the other scales we review. The reader veys. 174 Measuring Health Conceptual Basis ties: social responsibilities, self-care, and com- Katz and Lyerly presented the aims of psychi- munity activities. Three-point frequency re- atric treatment in terms of enhancing the pa- sponse scales are used. The items and response tient’s adjustment to living in the community. scales on form R3 are identical to those on R2, Adjustment was defined as a balance between save that the relative now indicates his expecta- the individual and his environment; it is a posi- tion of the patient’s level of performance in tive concept, implying more than the absence of these activities. A score indicating the relative’s negative behaviors. It includes not only freedom satisfaction with the patient’s performance may from symptoms of psychopathology and ab- be derived from the discrepancy between expec- sence of personal distress, but also suitable pat- tations (form R3) and actual performance (form terns of social interaction and adequate R2) (1). In a similar manner, a pair of 23-item performance in social roles (1). This conceptual forms (R4 and R5) cover the relative’s ratings of approach brings the adjustment scales close to the patient’s level of free-time activities and the indexes of positive mental well-being. In addi- relative’s expectations for this. Items cover so- tion to the person’s own feelings of well-being cial, community, and self-improvement activi- and satisfaction, the assessment of social adjust- ties and hobbies. ment must reflect the view of other people in his The patient also completes five forms. Form milieu: “the extent to which persons in the pa- S1 contains 55 items derived from the Hopkins tient’s social environment are satisfied with his Symptom Checklist on somatic symptoms and type and level of functioning” (1). This led to mood from which a total score may be calcu- the approach of basing the measurement on lated. The other four forms are equivalent to the judgments made by the individual and also by relative’s rating forms R2 to R5 and include the those close to him. same items with minor changes in wording. Be- cause the wording of the fourth set of forms Description (free-time activities) is identical, Katz et al. re- Two sets of scales are used, one completed by ferred to these as RS4. the patient (S scales) and the other by a relative The forms are too long to reproduce here; a (R scales). There are five scales in each set. The summary of the items is given in Katz and Ly- scales are introduced by an interviewer but are erly’s article (1, Tables 1–3). Because the ques- completed by the patient or a relative; the ques- tions in form R2 cover social health, we include tions use nontechnical language. The patient re- a summary of them in Exhibit 4.8. ports on his or her somatic symptoms, mood, level of activities, and personal satisfaction. A Reliability relative who has been in close contact with the Kuder-Richardson internal consistency coeffi- patient reports on the patient’s behavior and in- cients for eleven subscores on form R1 were cal- dicates the extent to which other people are sat- culated on two samples of patients. Coefficients isfied with his functioning. Full administration ranged from 0.41 to 0.87, with a median coeffi- takes 45 to 60 minutes but the scales need not be cient of 0.72 (N = 315) (1, Table 7). Six of the administered as a complete set. scales did not reach the level of 0.70 normally On form R1 (127 items) the patient’s relative used as indicative of an acceptable internal con- rates the patient’s psychiatric symptoms (e.g., sistency. Alpha internal consistency values for 13 “looks worn out,” “laughs or cries for no rea- scores derived from scale R1 were only moder- son”). Form R1 also covers social behavior ately high, ranging from 0.61 to 0.87 in a U.S. (e.g., “dependable” or “gets into fights with study; similar figures were reported from Japan- people”). Four-point scales indicate the fre- ese and West African studies (2, Table 3). A quency of each symptom and these may be more recent study reported alpha coefficients for summed into 12 or more scales and an overall 12 R1 scores ranging from 0.57 to 0.89; seven score. Form R2 contains 16 items on the indi- scales failed to meet the criterion of 0.70 (3, vidual’s performance of socially expected activi- Table 1).  Social Health 175 Exhibit 4.8 The Katz Adjustment Scale Form R2 (Level of Performance of Socially Expected Activities) and Form R3 (Level of Expectations for Performance of Social Activities) Note: The two forms are used to derive separate measures and in combination, to provide a “level of satisfaction” measure. 1. Helps with household chores 9. Goes to parties and other social activities 2. Visits his friends 10. Gets along with neighbors 3. Visits his relatives 11. Helps with family shopping 4. Entertains friends at home 12. Helps in the care and training of children 5. Dresses and takes care of himself 13. Goes to church 6. Helps with the family budgeting 14. Takes up hobbies 7. Remembers to do important things on time 15. Works 8. Gets along with family members 16. Supports the family The response scales for form R2 include three categories: “is not doing,” “is doing some,” and “is doing regularly.” For form R3, the response are: “did not expect him to be doing,” “expected him to be doing some” and “expected him to be doing regularly.” Reproduced with permission from Katz MM, Lyerly SB Methods for measuring adjustment and social behavior in the community: I. Rationale, description, discriminative validity and scale development. Psychological Reports, 1963;13:503–535, Table 2. (Monogr. Suppl. 4-V13) © Southern Universities Press 1963. Inter-rater agreement has been reported for Scores for the 127 items in form R1 were fac- the R forms. Agreement between fathers and tor analyzed and found to fall on 13 factors (2, mothers in rating adult patients with schizo- Table 2). A profile analysis of variance was used phrenic disorders showed median correlations of to compare responses to the 13 scales between 0.71 for the R1 rating; figures for role perfor- subtypes of schizophrenia; the contrast was mance and recreation were 0.85 and 0.47, re- highly significant for a sample in India but not in spectively (4, p213). Katz et al. reported an Nigeria (2, p343). The R1 Adjustment Scale indirect measure of inter-rater agreement, show- showed significant differences between the In- ing that there was no difference between ratings dian and Nigerian samples that broadly corre- made by different types of rater (father, mother sponded to differences in presenting symptoms or spouse) for most ratings; significant differ- observed with the Present State Examination (2, ences were found on only 4 out of 26 compar- p347). Goran and Fabiano obtained a 10-factor isons (2, pp340–341). result, with alpha values ranging from 0.78 to 0.94 (7, p221). Jackson et al. found a more com- Validity plex structure which they presented as seven sec- The relative’s forms discriminated significantly ond order factors, each comprising between between patients judged on clinical grounds to three and 18 first-order factors (8, Table 2). The be well-adjusted and those who were poorly ad- authors commented that, although numerous, justed. Hogarty quotes multiple correlations the factors were readily interpretable. They then ranging from 0.70 to 0.83 between the scale used the resulting factor scores to discriminate scores and global ratings made by a psychologist between four categories of head and spinal in- and a social worker (4, p214). The data pub- jury patients; the overall percentage correctly lished by Hogarty and Katz indicate consistent classified was 47%, but this rose to 61% using a contrasts in responses between psychiatric pa- different grouping of items (8, Tables 4 and 5). tients and a general population sample (5). There Correlations with the Sickness Impact Profile is some evidence from a study of patients with were 0.45 for the overall score, 0.57 for the schizophrenic disorders that the R scale is sensi- withdrawal subscale, and 0.36 for the psychoti- tive to change in health status (6). cism scale (9, Table 5). 176 Measuring Health Alternative Forms the negative aspects of adjustment that are scored The Adjustment Scales have been translated and (10, p616). The scales also permit comparison of used in a variety of languages and settings, in- actual and expected levels of performance, and cluding most of Europe, Japan, India, Hong between the relative’s and the patient’s assess- Kong, Turkey, and West Africa. References to ments. The use of the relative’s forms permits as- these studies are given by Katz et al. (2). sessment of patients with cognitive problems. The R1 form includes 127 items, of which Strengths of the Katz instrument include its em- Katz and Lyerly originally scored only 76, form- phasis on observable behavior, its wide range of ing 12 subscales (1). Vickrey et al. evaluated a topics, its emphasis on ordinary community be- modified scoring procedure that increased the havior, its discriminant validity, and that it can number of items scored to 126 items, forming be used by nonprofessional raters (8, p111). 14 subscales. The alpha reliability of these Given the widespread use of the scales, it is ranged from 0.66 to 0.88, considerably higher surprising how little evidence has been published than the figures obtained using the original scor- on their validity. Many studies have used the ing approach (3, Table 2). The revised scale dis- Katz scales, but virtually none presents data criminated between different grades of epilepsy from which conclusions may be drawn concern- (3, p68). ing their reliability and validity; many validity Two, separate modifications of the R scale and reliability results reviewed in one report have been described for use with patients with come from unpublished studies (4). We cannot, head injuries (7; 8). Goran and Fabiano also therefore, fully agree with the conclusion of identified 37 items that contributed only margin- Chen and Bryant, who claimed that “extensive ally to their scales; they show the revised 79-item efforts were made to establish the different inventory in their report (7, Table 2). The ten forms of reliabilities and validities, all of which component scales showed alpha coefficients were found satisfactory” (11). Nonetheless, the ranging from 0.75 to 0.93 (7, Table 3). The ten Katz scales have had a considerable impact on scales were then grouped under two second- the design of subsequent measurements. For ex- order factors. The first represented emotional ample, the approach of combining ratings by a sensitivity (which included scales on antisocial patient and a relative was followed in subsequent behavior, belligerence, verbal expansiveness, scales, including that of Clare and Cairns re- bizarre behavior, paranoid ideation, emotional viewed elsewhere in this chapter. Considering sensitivity, and social irresponsibility). The sec- the age of the Katz scales, their length and the ond factor represented physical and intellectual scant evidence for their reliability and validity, components (comprising speech and cognition, we recommend their use only if none of the orientation and apathy) (7, Table 4). other scales we describe applies. Reference Standards References Hogarty and Katz reported reference standards for the 13 factor scores on form R1, derived (1) Katz MM, Lyerly SB. Methods for measur- from a sample of 450 community respondents ing adjustment and social behavior in the aged 15 years and older. The standards are pre- community. I. Rationale, description, sented by age, sex, social class, and marital sta- discriminative validity and scale development. Psychol Rep tus (5, Table 1). 1963;13:503–535. (2) Katz MM, Marsella A, Dube KC, et al. On Commentary the expression of psychosis in different The Katz Adjustment Scales were long taken as cultures: schizophrenia in an Indian and in the standard approach to measuring social ad- a Nigerian community. Cult Med justment and continue to be used. They benefit Psychiatry 1988;12:331–355. from a clear conceptual foundation, but although (3) Vickrey BG, Hays RD, Brook RH, et al. this mentions positive adjustment, it is primarily Reliability and validity of the Katz  Social Health 177 Adjustment Scales in an epilepsy sample. standards to define adequate performance. In- Qual Life Res 1992;1:63–72. stead, the patient is asked to record difficulties (4) Hogarty GE. Informant ratings of experienced, in effect comparing the person’s so- community adjustment. In: Waskow IE, cial functioning with his personal expectations. Parloff MB, eds. Psychotherapy change The questions are phrased in terms of difficul- measures. Rockville, MD: National ties; they do not specify what form the difficul- Institute of Mental Health (DHEW ties may take (1). Publication No. (ADM) 74–120), 1975:202–221. (5) Hogarty GE, Katz MM. Norms of Description adjustment and social behavior. Arch Gen The SFS is intended for use by a psychologist or Psychiatry 1971;25:470–480. physician in clinical practice settings. It is a rat- (6) Parker G, Johnston P. Reliability of ing based on “a number of suggested questions parental reports using the Katz Adjustment designed to encompass the range of difficulty en- Scales: before and after hospital admission countered with neurotic out-patients. The exam- for schizophrenia. Psychol Rep iner is free to adapt and add questions where this 1989;65:251–258. is necessary to gain sufficient information” (2, (7) Goran DA, Fabiano RJ. The scaling of the p1). The schedule includes a total of 121 ques- Katz Adjustment Scale in a traumatic brain tions, not all of which would be relevant to every injury rehabilitation sample. Brain Inj 1993;7:219–229. respondent. The complete SFS includes 12 sec- (8) Jackson HF, Hopewell CA, Glass CA, et al. tions: employment, household chores, contribu- The Katz Adjustment Scale: modification tion to household, money, self-care, marital for use with victims of traumatic brain and relationship, care of children, patient-child rela- spinal injury. Brain Inj 1992;6:109–127. tionships, patient-parent and household relation- (9) Temkin NR, Dikmen S, Machamer J, et al. ships, social contacts, hobbies, and spare time General versus disease-specific measures: activities. The first two, the fourth, and the last further work on the Sickness Impact Profile sections are subdivided into problems in manag- for head injury. Med Care ing activities in that area and the feelings of dis- 1989;27(suppl):S44–S53. tress that result. Sections that do not apply to a (10) Linn MW. A critical review of scales used particular respondent may be omitted. The SFS to evaluate social and interpersonal adjustment in the community. is too long to reproduce here; as an illustration, Psychopharmacol Bull 1988;24:615–621. the section on work problems is shown in Ex- (11) Chen MK, Bryant BE. The measurement of hibit 4.9. Note that the version shown is a slight health-a critical and selective overview. Int revision of that originally described by Reming- J Epidemiol 1975;4:257–264. ton and Tyrer (1). Copies of the complete instru- ment may be obtained from Dr. Tyrer. After asking the questions in each section, The Social Functioning Schedule the interviewer rates the intensity of problems (Marina Remington and in that area on a visual analogue scale running P.J. Tyrer, 1979) from “none” to “severe difficulties.” The rat- Purpose ings cover only difficulties reported by the pa- The Social Functioning Schedule (SFS) is a semi- tient; the rater avoids making normative structured interview designed to assess the prob- judgments. Ratings refer to the past four weeks, lems a person experiences in 12 areas of normal with the exception of six months for the em- social functioning. The scale was designed for ployment questions, and the interview takes ten evaluating treatment of neurotic outpatients. to 20 minutes. Numerical scores are derived from the analogue scales by measurement, and Conceptual Basis an overall score is calculated as the average of The SFS uses a role performance approach to as- the relevant subsections; lower scores represent sess functioning but does not impose external better adjustment. Exhibit 4.9 Example of a Section from the Social Functioning Schedule 1. Work problems—behavior 1a Performance: As far as you know, how has S [the Subject] been coping with work? Does S have any difficulties? (Rate performance at work tasks.) 0 1 2 Not known no problems reduced output/given unable to perform Not applicable easier job his job/others have taken over 1b Time keeping: Does S usually get to work on time? 0 1 2 Not known usually arrives at a has occasionally has been more than Not applicable reasonable time missed 1⁄2–1 hour, or 1 hour late on more been more than 1 than two occasions hour late in last 4 weeks 1c Overactivity: Does S take on too much? (Is he rushed? Does he miss breaks or work late a lot?) 0 1 2 Not known does a day’s work rushes to complete work frequently Not applicable but no more— jobs, on a tight schedule occupies evenings work does not intrude and/or occasionally and weekends on personal time works late or brings work home Other problems (specify) ____________________________________ ____________________________________ 1. Rate work problems—behaviour none severe difficulties ___________________________________________________________________________________________ 2. Work problems—stress Does S talk about work? Has S complained about work recently? Has S seemed upset about work or under strain because of work? 2a Interest and satisfaction: Does S say that he likes his work? Has S complained that he is bored or fed up with work? 0 1 2 Not known S seems reasonably S reports that he is S indicates that he is Not applicable satisfied with work disinterested or somewhat utterly bored or situation dissatisfied with work dissatisfied with work 2b Distress: Does S seem to take work in his stride, or does work get him down? Does he appear troubled when he gets home from work? Does S complain that he has lost confidence? (Exclude boredom and dissatisfaction; include worry, strain, anxiety and anger). 0 1 2 Not known no noticeable some degree of distress S reports extreme Not applicable discomfort due occasionally reported distress or informant to work or observed observes this most of the time 2c Work relationships—friction: Has S talked about other people at work? In general how does he get on with them? Has S mentioned any quarrels or friction recently? (Include overt interpersonal difficulty with both clients and colleagues regardless of degree of associated distress). 0 1 2 Not known generally, smooth some friction or friction or quarrelling Not applicable easy relationships quarrelling during is a constant feature each week of work situation 178  Social Health 179 Exhibit 4.9 2d Work relationships—exploitation: Has S complained that he is treated unfairly at work? Has he complained that he feels put upon or dominated? 0 1 2 Not known S reports no S reports occasional S complains of Not applicable exploitation injustices or extreme exploitation exploitation Other problems (specify) ____________________________________ ____________________________________ 1. Rate work problems—strees none severe difficulties ___________________________________________________________________________________________ Reproduced from the Social Functioning Schedule obtained from Dr. PJ Tyer. With permission. Reliability gram for 27 patients (4, Table 1). The SFS Intraclass inter-rater reliability figures ranged showed significant differences between patients from 0.45 to 0.81 for different sections (average, with phobia, anxiety, and depression before 0.62) (1, p153). Ratings based on interviews treatment (5, p60). Finally, Casey and Tyrer with patients were compared with those with compared SFS scores in randomly drawn rural spouses; correlations ranged from 0.45 to 0.80 and urban community samples. Social function- (1, Table 1). ing was significantly worse in urban than rural settings, worse among people defined as psychi- Validity atric cases using the PSE, and significantly worse Virtually all evidence comes from studies of dis- among people with depression than those with criminant validity. This was originally tested by anxiety (6, p367). The PSE score correlated 0.75 comparing ratings of patients with personality with the SFS scores (6, Table 4). disorders, other psychiatric outpatients (mainly patients with psychotic disorders on maintenance Commentary pharmacotherapy), and the spouses of patients. The SFS covers a patient’s problems in social in- The SFS distinguished between personality- teraction and role performance, and the patient’s disordered patients and other groups, but not be- satisfaction with this, in a manner similar to that tween unaffected people and other psychiatric of Weissman’s and Gurland’s scales. It does not cases. In a study of 171 general practice patients indicate the level of social support available to referred for psychiatric consultation, the SFS the patient, nor does it cover positive levels of showed significant differences between patients functioning: the highest rating in each section is rated with various levels of certainty by a psychi- expressed in terms of the absence of identifiable atrist as psychiatric cases (3). The scale was not, problems. This is comparable to the approach however, able to differentiate among clinically used in several of the scales we review, as is the defined categories of personality disorder. In an- semistructured interview format. The scale is be- other study, the SFS overall score correlated ing used in current research and preliminary va- strongly (r = 0.69) with the total score of the Pre- lidity and reliability analyses are available. sent State Examination (PSE) (3, p7). A correla- Potential criticisms of the SFS include the tion of 0.65 was obtained with a five-point possibility of interviewer bias in translating re- indicator of level of alcohol consumption used as sponses into the visual analogue ratings. This the outcome of an alcohol detoxification pro- may have caused the relatively low inter-rater re- 180 Measuring Health liability results, although the intraclass correla- availability and supportive quality of social rela- tions used here give a coefficient as much as 0.20 tionships. The interview was designed as a sur- lower than a Pearson correlation computed vey method to measure social factors associated based on the same data. More reliability testing with the development of neurotic illness; it may is desirable, particularly because the rating sys- also be used to evaluate the outcomes of care for tem depends on the judgment of the interviewer. psychiatric patients (1). Because the scale is broad in scope, it naturally sacrifices detail in most areas when compared Conceptual Basis with alternative scales. Nonetheless, where an Henderson’s approach to measuring social rela- expert rater is available to make the ratings and tionships was guided by his research goal of where summary ratings of a patient’s problems identifying how social bonding and support pro- (rather than assets) are required, this scale tect against neurotic disorders in the presence of should be considered for use. adversity (2). For this, he required a measure of the independent variable: the supportive quality Address of relationships (3). Following the conceptual Professor P.J. Tyrer, Department of Psychologi- work of Robert Weiss, Henderson et al. identi- cal Medicine, Imperial College, London, W6 fied six benefits that are offered by lasting social 8LH, UK [email protected] relationships: a sense of attachment and security, social integration, the opportunity to care for References others, the provision of reassurance as to one’s personal worth, a sense of reliable alliance, and (1) Remington M, Tyrer P. The Social the availability of help and guidance when Functioning Schedule—a brief semi- needed (1). The first of these themes, attach- structured interview. Soc Psychiatry ment, was considered especially important, and 1979;14:151–157. (2) Tyrer PJ. Social Functioning Schedule— Henderson et al. here drew on the concepts of short version. (Manuscript, Mapperely Bowlby. Attachment refers to “that attribute of Hospital, Nottingham, UK, c 1984). relationships which is characterized by affection (3) Casey PR, Tyrer PJ, Platt S. The and which gives the recipient a subjective sense relationship between social functioning and of closeness. It is also pleasant and highly val- psychiatric symptomatology in primary ued, commonly above all else” (4, p725). care. Soc Psychiatry 1985;20:5–9. Social ties may be evaluated in terms of their (4) Griggs SMLB, Tyrer PJ. Personality objective availability, or in terms of the person’s disorder, social adjustment and treatment subjective assessment of their adequacy. The ISSI outcome in alcoholics. J Stud Alcohol “seeks to establish the availability of most of the 1981;42:802–805. six provisions proposed by Weiss by ascertaining (5) Tyrer P, Remington M, Alexander J. The outcome of neurotic disorders after out- the availability of persons in specified roles. patient and day hospital care. Br J Questions about adequacy follow each of the Psychiatry 1987;151:57–62. availability items” (1, p34). (6) Casey PR, Tyrer PJ. Personality, functioning and symptomatology. J Description Psychiatr Res 1986;20:363–374. The ISSI is a 45-minute interview that records details of a person’s network of social attach- ments, covering both quantity and quality of so- The Interview Schedule for cial support in the last 12 months. The questions Social Interaction cover close intimate relationships such as those (Scott Henderson, 1980) with family, parents, and very close friends. It Purpose also covers more diffuse ties, such as those with The Interview Schedule for Social Interaction neighbors, acquaintances, or work associates. (ISSI) is a research instrument that assesses the Four principal indices are formed: the availabil-  Social Health 181 ity of close and emotionally intimate relation- it is available, along with guide notes, in the ap- ships, their adequacy, the availability of more pendices of the book by Henderson et al. (1, diffuse relationships and friendships that provide pp203–230). As an illustration, Exhibit 4.10 social integration, and the adequacy of such rela- shows one question from the ISSI. tionships. The interviewer mentions a particular Detailed discussions of scoring procedures type of social relationship and asks the respon- are available (1, pp37–39; 5). The scores are dent if he has such a relationship; the interviewer complex in that they reflect the idea that scores then asks whether the amount of this type of re- should not necessarily increase or decrease mo- lationship is adequate. Adequacy covers friend- notonically: both too much and too little sup- ship, attachment, nurturance, reassurance of port may constitute less than ideal replies. worth, and reliable alliances (5). In addition, the However, initial analyses indicated that the re- respondent is asked to name the main person sults obtained from the questionnaire did not who provides each of several different types of fully reflect the complexity of the conceptual attachment relationships. This information is formulation, so a simplified scoring system was summarized on an attachment table that records proposed. Four scores summarize the extent and the degree of closeness of the respondent to each adequacy of social support: of the people she cites as emotionally close to her. Details of the identity of these individuals availability of attachment (AVAT, 8 items) are recorded, as is an indication of their accessi- adequacy of attachment (ADAT, 12 items) bility to the respondent. The table is also used to availability of social integration (AVSI, 16 indicate the extent to which social provisions are items) concentrated on few or many people. adequacy of social integration (ADSI, 17 The instrument is too long to reproduce here; items) Exhibit 4.10 Example of an Item from the Interview Schedule for Social Interaction 33. At present, do you have someone you can share your most private feelings with (confide in) or not? No one (Go to Q. 33D) 0 Yes 1 A. Who is this mainly? (Fill in only one on Attachment Table) B. Do you wish you could share more with _______________ or is it about right? About right 1 Depends on the situation 2 More 3 Not applicable 9 C. Would you like to have someone else like this as well, would you prefer not to use a confidant, or is it just about right for you the way it is? Prefers no confidant 1 About right 2 Depends on the situation 3 Like someone else as well 4 Not applicable 9 (Go to Q. 34) (If no one) D. Would you like to have someone like this or would you prefer to keep your feelings to yourself? Keep things to self 1 Like someone 2 Not applicable 9 Reprinted from Neurosis and the social environment by S Henderson with DG Byrne and P Duncan-Jones, Academic Press, Sydney, 1981: 214. With permission. 182 Measuring Health Details of this scoring system and the questions respondents (1). Henderson et al. described the that are included in forming the four scores are pattern of associations as “coherent.” Correla- given in the Henderson et al. book (1, Appendix tions between the respondent’s scores and an in- III). formant’s score reflecting his perception of the respondent’s social world ranged from 0.26 to Reliability 0.59 (N = 114) (4, p731). To estimate the effect The alpha internal consistency of four scores of response sets, the scale was correlated with ranged from 0.67 to 0.79 (N = 756), and 18-day two lie scales, and a maximum of 10.6% of vari- test-retest correlations ranged from 0.71 to 0.76 ance in the ISSI scores could be explained by so- (N = 51) (1, p47). For 221 respondents, stability cially desirable response styles (4). Other correlations were calculated using a structural validity data include a comparison with the modeling approach that corrected for the imper- Health Locus of Control Scale: rho 0.40 with fect internal consistency of the scores. The sta- the availability of social integration score, show- bility results at 4, 8, and 12 months ranged from ing greater social integration with greater inter- 0.66 to 0.88 (1; 4). Alpha values were higher in nality (9). a Swedish study, including 0.77 (AVSI), 0.80 In a study of predictive validity over four (AVAT), 0.86 (ADSI) and 0.94 (ADAT) (6, Table months, Henderson showed that 30% of the 3). variance in the General Health Questionnaire (GHQ) was shared by the ISSI in a population Validity experiencing many life changes (10). Concurrent Preliminary comparisons were made between correlations between the GHQ-30 and the four the structure of the scale and the conceptual def- ISSI scores ranged from −0.16 to −0.38 (10, inition of its content. A detailed presentation Table 1). A significant association was found be- was given by Duncan-Jones (7). Henderson et al. tween the GHQ-12 and the ISSI AVAT score, concluded that the dimensions of availability such that those with lower support tended to be and adequacy of “reliable alliance” and “reas- more disturbed emotionally (11, Table 3). surance of worth” could be distinguished but Changes in health may not be reflected in could not be accurately distinguished from changes in ISSI scores: in a 12-month study of friendship. The attempt to measure Weiss’s con- anxious patients, reductions in the level of anxiety cept of “opportunity for nurturing” was not were not mirrored by changes in ISSI scores (12). successful and, finally, the results showed that a more general dimension of “social integration” Alternative Forms could be formed by combining acquaintance, Henderson et al. made minor changes to item friendship, reassurance of worth, and reliable al- wording to suit the ISSI to elderly respondents liance (1, p38). (8, p381). A self-rating version has been de- The ISSI was shown to discriminate signifi- scribed (13). A 12-item abbreviation has been cantly between groups that would be expected to developed for survey use. Alpha reliability values differ in social adjustment: recent arrivals in a were typically about ten points lower than those city compared with residents, and separated or for the long form and correlations with physical divorced people compared with those who were activities were lower, but correlations with so- married (1). Similar analyses compared scores cial activities were, if anything, higher (6, Tables by living arrangements, marital status, and the 3 and 4). presence of an extended family; there were clear and logical associations with the ISSI scores (8, Reference Standards pp383–384). Henderson et al. reported mean scores and stan- Correlations between the four scores and trait dard deviations for population samples in Can- neuroticism measured by the Eysenck Personal- berra, Australia, by age and marital status (1, ity Inventory ranged from 0.18 to 0.31 for 225 Tables 3.1–3.3; 8, Table 1).  Social Health 183 Commentary (2) Henderson S. A development of social The ISSI is one of the few scales that measures psychiatry: the systematic study of social social support rather than social roles. Like bonds. J Nerv Ment Dis 1980;168(2):63–69. Linn’s Social Dysfunction Rating Scale (reviewed (3) Henderson AS, Brown GW. Social support: in this chapter), and Brandt and Weinert’s Per- the hypothesis and the evidence. In: Henderson AS, Burrows GD, eds. sonal Resource Questionnaire (14), the ISSI as- Handbook of social psychiatry. sesses both availability and adequacy of Amsterdam: Elsevier, 1988:73–85. relationships. It is one of the few instruments that (4) Henderson S, Duncan-Jones P, Byrne DG, assesses unwanted (“too much”) support. It has et al. Measuring social relationships: the offered stimulating insights into the relationships Interview Schedule for Social Interaction. among support, life change, coping, and morbid- Psychol Med 1980;10:723–734. ity (3). Thus, for example, Henderson and Brown (5) Duncan-Jones P. The structure of social were able to show that the quality, rather than relationships: analysis of a survey the quantity, of support provided the best predic- instrument, part 1. Soc Psychiatry tor of resistance to psychological disorder. 1981;16:55–61. Henderson has reviewed some criticisms of (6) Undén A-L, Orth-Gomér K. Development of a social support instrument for use in the ISSI and has drawn an interesting compari- population surveys. Soc Sci Med son with the measurement approach used by 1989;29:1387–1392. George Brown in London (3). The ISSI covers (7) Duncan-Jones P. The structure of social feelings of attachment more than the actual pro- relationships: analysis of a survey vision of support, and Henderson noted that the instrument, part 2. Soc Psychiatry two may not completely correspond. Brown’s 1981;16:143–149. Social Evaluation and Social Support Schedule, (8) Henderson AS, Grayson DA, Scott R, et al. by contrast, collects more detailed information Social support, dementia and depression on the nature of support provided by each per- among the elderly living in the Hobart son (3, p75). community. Psychol Med It is noticeable that, as was the case with 1986;16:379–390. (9) Thomas PD, Hooper EM. Healthy elderly: Weissman’s scale, empirical analyses of the social bonds and locus of control. Res structure of the scale do not match the concep- Nurs Health 1983;6:11–16. tual framework that it was designed to reflect. (10) Henderson S. Social relationships, adversity This seems to be a problem in this field. In com- and neurosis: an analysis of prospective parison with the validity and reliability results of observations. Br J Psychiatry the other scales that we review, evidence for reli- 1981;138:391–398. ability and validity is quite good, and hopefully (11) Singh B, Lewin T, Raphael B, et al. Minor will continue to accumulate. The ISSI is suffi- psychiatric morbidity in a casualty ciently successful that we recommend its use in population: identification, attempted studies where a 45-minute interview is practical. intervention and six-month follow- up. Where a shorter rating of social support is re- Aust NZ J Psychiatry 1987;21:231–240. (12) Parker G, Barnett B. A test of the social quired, we recommend McFarlane’s or Sarason’s support hypothesis. Br J Psychiatry scales, reviewed elsewhere in this chapter, or one 1987;150:72–77. of the scales developed by Brandt (14) or Nor- (13) Furukawa T. Factor structure of social beck (15). support and its relationship to minor psychiatric disorders among Japanese adolescents. Int J Soc Psychiatry References 1995;41:88–102. (1) Henderson S, Byrne DG, Duncan-Jones P. (14) Brandt PA, Weinert C. The PRQ—a social Neurosis and the social environment. support measure. Nurs Res Sydney, Australia: Academic Press, 1981. 1981;30(5):277–280. 184 Measuring Health (15) Norbeck JS, Lindsey AM, Carrieri VL. The Both the interview and self-report versions development of an instrument to measure contain 42 questions covering role performance social support. Nurs Res in six areas of role functioning: work (as em- 1981;30:264–269. ployee, housewife, or student, questions 1–18); social leisure activities (questions 19–29); rela- tionships with extended family (questions The Social Adjustment Scale 30–37); and roles as spouse (questions 38–46); (Myrna M. Weissman, 1971) parent (questions 47–50), and member of the Purpose family unit (questions 51–54). The method pro- The Social Adjustment Scale (SAS) was designed vides alternative questions on work relations for as an outcome measurement to evaluate drug students, housewives, and employed people, so treatment and psychotherapy for depressed pa- the scale includes a total of 54 questions, of tients. It has since been used in studying a which respondents answer 42. In each role area, broader range of patients and healthy respon- questions cover the patient’s performance over the dents. past two weeks, the amount of friction he experi- ences with others, finer aspects of interpersonal Conceptual Basis relationships (e.g., level of independence), inner The development of this self-report scale re- feelings (e.g., shyness, boredom), and satisfaction. flected a growing interest in measuring success- Five- and six-point response scales are used; ful adjustment to community living, as distinct higher scores represent increasing impairment. from problems in role performance. This ap- Two scoring methods are used: a mean score for proach is particularly relevant to patients receiv- each section (e.g., work, leisure), or an overall ing psychotherapy who do not, for the most score obtained by summing the item scores and part, present with clinical symptoms (1). The dividing by the number of items checked. conceptual approach and item content were de- The self-report version takes 15 to 20 min- rived from Gurland’s Structured and Scaled In- utes to complete, whereas the interview version terview to Assess Maladjustment and from prior takes about 45 to 60 minutes and includes an ad- empirical studies by Paykel and Weissman. The ditional six global judgments (1). The SAS-SR is scale assesses interpersonal relationships in vari- usually completed in the presence of a research ous roles, covering feelings, satisfaction, friction, assistant, who can explain the format, answer and performance. The structure reflects two sep- questions, and check on the completeness of arate dimensions: six role areas (e.g., work, fam- replies. ily) and five aspects of adjustment that are applied, depending on appropriateness, to each Reliability role area (2). For 15 depressed patients, the correlation be- tween the patient’s replies on the self-report in- Description strument and a rating made by the spouse or The SAS was originally developed as an inter- other informant was 0.74; the correlation be- view schedule, which was then turned into a tween patient and interviewer assessments was self-report version, the SAS-SR, shown in Ex- 0.70 (1, Table 5). Patients rated themselves as hibit 4.11. The self-report version has the advan- more impaired than the interviewer did. Scores tage of being inexpensive to administer and free on the self-report and interview versions of the from interviewer bias (1). It is generally com- SAS correlated 0.72 for 76 depressed patients; pleted by the respondent but can also be com- agreement for the various sections ranged from pleted by a relative. An updated version of the 0.40 to 0.76 (1, Table 3). Agreement between scale was produced in 1999 and is copyrighted raters was assessed for the interview version for and marketed by Multi-Health Systems (www 31 patients. The raters agreed completely on .mhs.com), who provide a manual and scoring 68% of all items, with a further 27% of ratings sheets. falling within one point of each other (3). Inter- Exhibit 4.11 The Social Adjustment Scale—Self-Report Social Adjustment Self-Report Questionnaire We are interested in finding out how you have been doing in the last two weeks. We would like you to answer some questions about your work, spare time and your family life. There are no right or wrong answers to these questions. Check the answers that best describe how you have been in the last two weeks. WORK OUTSIDE THE HOME 4 I felt ashamed most of the time. Please check the situation that best describes you. 5 I felt ashamed all the time. I am 1 a worker for pay 4 retired 2 a housewife 5 unemployed 4. Have you had any arguments with people at work in the last 2 weeks? 3 a student 1 I had no arguments and got along very Do you usually work for pay more than 15 hours per well. week? 2 I usually got along well but had minor 1 YES 2 NO arguments. Did you work any hours for pay in the last two 3 I had more than one argument. weeks? 4 I had many arguments. 1 YES 2 NO 5 I was constantly in arguments. Check the answer that best describes how you have 5. Have you felt upset, worried, or uncomfortable been in the last two weeks while doing your work during the last 2 weeks? 1 I never felt upset. 1. How many days did you miss from work in the last two weeks? 2 Once or twice I felt upset. 1 No days missed. 3 Half the time I felt upset. 2 One day. 4 I felt upset most of the time. 3 I missed about half the time. 5 I felt upset all of the time. 4 Missed more than half the time but did 6. Have you found your work interesting these last make at least one day. two weeks? 5 I did not work any days. 1 My work was almost always interesting. 8 On vacation all of the last two weeks. 2 Once or twice my work was not interesting. If you have not worked any days in the last two 3 Half the time my work was uninteresting. weeks, go on to Question 7. 4 Most of the time my work was 2. Have you been able to do your work in the last 2 uninteresting. weeks? 5 My work was always uninteresting. 1 I did my work very well. WORK AT HOME—HOUSEWIVES ANSWER 2 I did my work well but had some minor QUESTIONS 7–12. OTHERWISE, GO ON TO problems. QUESTION 13. 3 I needed help with work and did not do well about half the time. 7. How many days did you do some housework during the last 2 weeks? 4 I did my work poorly most of the time. 1 Every day. 5 I did my work poorly all the time. 2 I did the housework almost every day. 3. Have you been ashamed of how you do your work 3 I did the housework about half the time. in the last 2 weeks? 4 I usually did not do the housework. 1 I never felt ashamed. 5 I was completely unable to do housework. 2 Once or twice I felt a little ashamed. 8 I was away from home all of the last two 3 About half the time I felt ashamed. weeks. (continued) 185 Exhibit 4.11 (continued) 8. During the last two weeks, have you kept up with FOR STUDENTS your housework? This includes cooking, cleaning, Answer Questions 13–18 if you go to school half time laundry, grocery shopping, and errands. or more. Otherwise, go on to Question 19. 1 I did my work very well. What best describes your school program? (Choose 2 I did my work well but had some minor one) problems. 1 Full Time 3 I needed help with my work and did not do it well about half the time. 2 3/4 Time 4 I did my work poorly most of the time. 3 Half Time 5 I did my work poorly all of the time. Check the answer that best describes how you have been the last 2 weeks. 9. Have you been ashamed of how you did your 13. How many days of classes did you miss in the last housework during the last 2 weeks? 2 weeks? 1 I never felt ashamed. 1 No days missed. 2 Once or twice I felt a little ashamed. 2 A few days missed. 3 About half the time I felt ashamed. 3 I missed about half the time. 4 I felt ashamed most of the time. 4 Missed more than half time but did make 5 I felt ashamed all the time. at least one day. 5 I did not go to classes at all. 10. Have you had any arguments with salespeople, 8 I was on vacation all of the last two weeks. tradesmen or neighbors in the last 2 weeks? 1 I had no arguments and got along very 14. Have you been able to keep up with your class well. work in the last 2 weeks? 2 I usually got along well, but had minor 1 I did my work very well. arguments. 2 I did my work well but had minor 3 I had more than one argument. problems. 4 I had many arguments. 3 I needed help with my work and did not do 5 I was constantly in arguments. well about half the time. 4 I did my work poorly most of the time. 11. Have you felt upset while doing your housework 5 I did my work poorly all the time. during the last 2 weeks? 15. During the last 2 weeks, have you been ashamed 1 I never felt upset. of how you do your school work? 2 Once or twice I felt upset. 1 I never felt ashamed. 3 Half the time I felt upset. 2 Once or twice I felt ashamed. 4 I felt upset most of the time. 3 About half the time I felt ashamed. 5 I felt upset all of the time. 4 I felt ashamed most of the time. 5 I felt ashamed all of the time. 12. Have you found your housework interesting these last 2 weeks? 16. Have you had any arguments with people at 1 My work was almost always interesting. school in the last 2 weeks? 2 Once or twice my work was not 1 I had no arguments and got along interesting. very well. 3 Half the time my work was uninteresting. 2 I usually got along well but had minor 4 Most of the time my work was arguments. uninteresting. 3 I had more than one argument. 5 My work was always uninteresting. 4 I had many arguments. 186 Exhibit 4.11 5 I was constantly in arguments. 21. How many times in the last two weeks have you 8 Not applicable; I did not attend school. gone out socially with other people? For example, visited friends, gone to movies, bowling, church, restaurants, invited friends to your home? 17. Have you felt upset at school during the last 1 More than 3 times. 2 weeks? 2 Three times. 1 I never felt upset. 3 Twice. 2 Once or twice I felt upset. 4 Once. 3 Half the time I felt upset. 5 None. 4 I felt upset most of the time. 5 I felt upset all of the time. 22. How much time have you spent on hobbies or 8 Not applicable; I did not attend spare time interests during the last 2 weeks? For school. example, bowling, sewing, gardening, sports, reading? 18. Have you found your school work interesting 1 I spent most of my spare time on hobbies these last 2 weeks? almost every day. 1 My work was almost always interesting. 2 I spent some spare time on hobbies some of the days. 2 Once or twice my work was not interesting. 3 I spent a little spare time on hobbies. 3 Half the time my work was uninteresting. 4 I usually did not spend any time on hobbies but did watch TV. 4 Most of the time my work was uninteresting. 5 I did not spend any spare time on hobbies or watch TV. 5 My work was always uninteresting. 23. Have you had open arguments with your friends in the last 2 weeks? SPARE TIME—EVERYONE ANSWER QUESTIONS 19–27. 1 I had no arguments and got along very well. Check the answer that best describes how you have 2 I usually got along well but had minor been in the last 2 weeks. arguments. 3 I had more than one argument. 19. How many friends have you seen or spoken to on the telephone in the last weeks? 4 I had many arguments. 1 Nine or more friends. 5 I was constantly in arguments. 2 Five to eight friends. 8 Not applicable; I have no friends. 3 Two to four friends. 24. If your feelings were hurt or offended by a friend 4 One friend. during the last two weeks, how badly did you take 5 No friends. it? 1 It did not affect me or it did not happen. 20. Have you been able to talk about your feelings and 2 I got over it in a few hours. problems with at least one friend during the last 2 3 I got over it in a few days. weeks? 4 I got over it in a week. 1 I can always talk about my innermost 5 It will take me months to recover. feelings. 8 Not applicable; I have no friends. 2 I usually can talk about my feelings. 3 About half the time I felt able to talk about 25. Have you felt shy or uncomfortable with people in my feelings. the last 2 weeks? 4 I usually was not able to talk about my 1 I always felt comfortable. feelings. 2 Sometimes I felt uncomfortable but could 5 I was never able to talk about my feelings. relax after a while. 8 Not applicable; I have no friends. 3 About half the time I felt uncomfortable. (continued) 187 Exhibit 4.11 (continued) 4 I usually felt uncomfortable. 30. Have you had open arguments with your relatives 5 I always felt uncomfortable. in the last 2 weeks? 8 Not applicable; I was never with people. 1 We always got along very well. 2 We usually got along very well but had 26. Have you felt lonely and wished for more friends some minor arguments. during the last 2 weeks? 3 I had more than one argument with at least 1 I have not felt lonely. one relative. 2 I have felt lonely a few times. 4 I had many arguments. 3 About half the time I felt lonely. 5 I was constantly in arguments. 4 I usually felt lonely. 31. Have you been able to talk about your feelings and 5 I always felt lonely and wished for more problems with at least one of your relatives in the friends. last 2 weeks? 27. Have you felt bored in your spare time during the 1 I can always talk about my feelings with at last 2 weeks? least one relative. 1 I never felt bored. 2 I usually can talk about my feelings. 2 I usually did not feel bored. 3 About half the time I felt able to talk about 3 About half the time I felt bored. my feelings. 4 Most of the time I felt bored. 4 I usually was not able to talk about my feelings. 5 I was constantly bored. 5 I was never able to talk about my feelings. Are you a Single, Separated, or Divorced Person not living with a person of opposite sex; please answer 32. Have you avoided contacts with your relatives below: these last two weeks? 1 YES, Answer questions 28 & 29. 1 I have contacted relatives regularly. 2 NO, go to question 30. 2 I have contacted a relative at least once. 28. How many times have you been with a date these 3 I have waited for my relatives to contact last 2 weeks? me. 1 More than 3 times. 4 I avoided my relatives, but they contacted me. 2 Three times. 5 I have no contacts with any relatives. 3 Twice. 33. Did you depend on your relatives for help, advice, 4 Once. money or friendship during the last 2 weeks? 5 Never. 1 I never need to depend on them. 29. Have you been interested in dating during the last 2 I usually did not need to depend on them. 2 weeks? If you have not dated, would you have liked to? 3 About half the time I needed to depend on them. 1 I was always interested in dating. 4 Most of the time I depend on them. 2 Most of the time I was interested. 5 I depend completely on them. 3 About half of the time I was interested. 34. Have you wanted to do the opposite of what your 4 Most of the time I was not interested. relatives wanted in order to make them angry 5 I was completely uninterested. during the last 2 weeks? 1 I never wanted to oppose them. FAMILY 2 Once or twice I wanted to oppose them. Answer Questions 30–37 about your parents, brothers, sisters, in laws, and children not living at 3 About half the time I wanted to oppose home. Have you been in contact with any of them in them. the last two weeks? 4 Most of the time I wanted to oppose them. 1 YES, Answer questions 30–37. 5 I always opposed them. 2 NO, Go to question 36. 188 Exhibit 4.11 35. Have you been worried about things happening 39. Have you been able to talk about your feelings and to your relatives without good reason in the last problems with your partner during the last 2 2 weeks? weeks? 1 I have not worried without reason. 1 I could always talk freely about my 2 Once or twice I worried. feelings. 3 About half the time I worried. 2 I usually could talk about my feelings. 4 Most of the time I worried. 3 About half the time I felt able to talk about my feelings. 5 I have worried the entire time. 4 I usually was not able to talk about my 8 Not applicable; my relatives are no longer feelings. living. 5 I was never able to talk about my feelings. EVERYONE answer Questions 36 and 37, even if your relatives are not living. 40. Have you been demanding to have your own way at home during the last 2 weeks? 36. During the last two weeks, have you been thinking 1 I have not insisted on always having my that you have let any of your relatives down or own way. have been unfair to them at any time? 2 I usually have not insisted on having my 1 I did not feel that I let them down at all. own way. 2 I usually did not feel that I let them down. 3 About half the time I insisted on having my 3 About half the time I felt that I let them own way. down. 4 I usually insisted on having my own way. 4 Most of the time I have felt that I let them 5 I always insisted on having my own way. down. 5 I always felt that I let them down. 41. Have you been bossed around by your partner these last 2 weeks? 37. During the last two weeks, have you been thinking that any of your relatives have let you down or 1 Almost never. have been unfair to you at any time? 2 Once in a while. 1 I never felt that they let me down. 3 About half the time. 2 I felt that they usually did not let me down. 4 Most of the time. 3 About half the time I felt they let me down. 5 Always. 4 I usually have felt that they let me down. 5 I am very bitter that they let me down. 42. How much have you felt dependent on your partner these last 2 weeks? Are you living with your spouse or have you been 1 I was independent. living with a person of the opposite sex in a 2 I was usually independent. permanent relationship? 3 I was somewhat dependent. 1 YES, Please answer questions 38–46. 4 I was usually dependent. 2 NO, Go to question 47. 5 I depended on my partner for everything. 38. Have you had open arguments with your partner in the last 2 weeks? 43. How have you felt about your partner during the 1 We had no arguments and we got along last 2 weeks? well. 1 I always felt affection. 2 We usually got along well but had minor 2 I usually felt affection. arguments. 3 About half the time I felt dislike and half 3 We had more than one argument. the time affection. 4 We had many arguments. 4 I usually felt dislike. 5 We were constantly in arguments. 5 I always felt dislike. (continued) 189 Exhibit 4.11 (continued) 44. How many times have you and your partner had 48. Have you been able to talk and listen to your intercourse? children during the last 2 weeks? Include only 1 More than twice a week. children over the age of 2. 2 Once or twice a week. 1 I always was able to communicate with them. 3 Once every two weeks. 2 I usually was able to communicate with 4 Less than once every two weeks but at least them. once in the last month. 3 About half the time I could communicate. 5 Not at all in a month or longer. 4 I usually was not able to communicate. 45. Have you had any problems during intercourse, 5 I was completely unable to communicate. such as pain these last two weeks? 8 Not applicable; no children over the age of 2. 1 None. 2 Once or twice. 49. How have you been getting along with the 3 About half the time. children during the last 2 weeks? 4 Most of the time. 1 I had no arguments and got along very well. 5 Always. 2 I usually got along well but had minor 8 Not applicable; no intercourse in the last arguments. two weeks. 3 I had more than one argument. 4 I had many arguments. 46. How have you felt about intercourse during the 5 I was constantly in arguments. last 2 weeks? 1 I always enjoyed it. 50. How have you felt toward your children these last 2 I usually enjoyed it. 2 weeks? 3 About half the time I did and half the time 1 I always felt affection. I did not enjoy it. 2 I mostly felt affection. 4 I usually did not enjoy it. 3 About half the time I felt affection. 5 I never enjoyed it. 4 Most of the time I did not feel affection. 5 I never felt affection toward them. CHILDREN FAMILY UNIT Have you had unmarried children, stepchildren, or foster children living at home during the last Have you ever been married, ever lived with a person two weeks? of the opposite sex, or ever had children? Please check 1 YES, Answer questions 47–50. 1 YES, Please answer questions 51–53. 2 NO, Go to question 51. 2 NO, Go to question 54. 47. Have you been interested in what your children 51. Have your worried about your partner or any of are doing—school, play or hobbies during the last your children without any reason during the last 2 2 weeks? weeks, even if you are not living together now? 1 I was always interested and actively 1 I never worried. involved. 2 Once or twice I worried. 2 I usually was interested and involved. 3 About half the time I worried. 3 About half the time interested and half the time not interested. 4 Most of the time I worried. 4 I usually was disinterested. 5 I always worried. 5 I was always disinterested. 8 Not applicable; partner and children not living. 190  Social Health 191 Exhibit 4.11 52. During the last 2 weeks have you been thinking 4 I usually felt they let me down. that you have let down your partner or any of your 5 I feel bitter that they have let me down. children at any time? 1 I did not feel I let them down at all. FINANCIAL—EVERYONE PLEASE ANSWER QUESTION 54. 2 I usually did not feel that I let them down. 54. Have you had enough money to take care of your 3 About half the time I felt I let them down. own and your family’s financial needs during the 4 Most of the time I have felt that I let them last 2 weeks? down. 1 I had enough money for needs. 5 I let them down completely. 2 I usually had enough money with minor 53. During the last 2 weeks, have you been thinking problems. that your partner or any of your children have let 3 About half the time I did not have enough you down at any time? money but did not have to borrow money. 1 I never felt that they let me down. 4 I usually did not have enough money and 2 I felt they usually did not let me down. had to borrow from others. 3 About half the time I felt they let me down. 5 I had great financial difficulty. Reproduced from Social Adjustment Scale obtained from Dr. Myrna M Weissman. With permission. rater Pearson correlations across all items aver- were recorded in all six areas covered in the aged 0.83 (3, Table 3). questionnaire (1). Applied to samples of commu- Item-total correlations for the various role ar- nity residents, patients with depressive disorders, eas ranged between 0.09 and 0.83 for the patients with alcoholism, and those diagnosed interviewer-administered SAS (2, Table 2). An with schizophrenia, the SAS demonstrated con- alpha internal consistency coefficient of 0.74 sistent, although not strong, contrasts in re- and a mean test-retest coefficient of 0.80 were sponses (4). In an earlier study, significant reported (4, p324). Alpha was 0.73 for the SAS- differences had been shown between depressed SR in a Japanese study (5, Table 1). patients and nonpatients for 40 out of the 48 items (7). Validity Weissman et al. presented correlations with The SAS scores did not correlate significantly independent ratings for various subsamples. with age, social class, sex, or history of previous Table 4.2 shows the resulting correlations with depression (N = 76), suggesting that scores are four independent assessments: the Hamilton unaffected by sociodemographic status (1). Rating Scale for Depression and the Raskin De- Women were more impaired than men on the pression Scale, both applied by a clinician, and SAS-SR, but there were no differences in scores two self-administered scales: the Center for Epi- by age or race (6, p462). demiologic Studies Depression Scale and the A factor analysis applied to the interview ver- Symptom Checklist-90. A correlation of 0.42 sion produced six factors: work performance, was obtained with the Brief Psychiatric Rating interpersonal friction, inhibited communication, Scale and one of 0.53 was found with a clinical submissive dependency, family attachment, and rating of irritability (8, Table 2). Further details anxiety (2). These factors cut across the two- of the validation results are given in a review by dimensional conceptual framework on which Weissman et al. (3). Suzuki et al. reported a cor- the method was constructed. relation of 0.56 with the General Health Ques- For 76 depressed patients, the self-report tionnaire (GHQ-30); this was an average across method was administered before and after four several socioeconomic groups, with a range from weeks of treatment. Significant improvements 0.11 to 0.36 (5, Table 2). Weissman et al. re- 192 Measuring Health Table 4.2 Correlation of the SAS and Independent Rating Scales Correlation with Sample Hamilton Raskin CES-D SCL-90 (N) Community sample ... 0.44 0.57 0.59 (482) Acute depressive patients 0.36 0.18 0.49 0.66 (191) Alcoholic patients 0.67 0.65 0.74 0.76 (54) Schizophrenic patients 0.72 0.75 0.85 0.84 (47) Adapted from Weissman MM, Prusoff BA, Thompson WD, Harding PS, Myers JK. Social adjustment by self-report in a com- munity sample and in psychiatric outpatients. J Nerv Ment Dis 1978;166:324, Table 5. ported a correlation of 0.57 with the Social version of the SAS-II containing 24 items (13). It Adaptation Self-Evaluation Scale and of 0.42 covers seven themes; overall alpha was 0.79 and with the social functioning scale of the SF-36 (6, 0.80 in two samples, with 2-week test-retest reli- Table 1). ability of 0.83 (13, Table 3). Concurrent validity appears good (13, Table 6). Reference Standards A British version of the SAS-SR modified Weissman et al. reported mean scores from a item wording and standardized the rating scale sample of 482 community respondents and 191 for each question (14). Agreement between self- patients with diagnosed depression (4). Richman administered and interviewer versions was close: reported scores by employment and marital sta- Pearson correlations of 0.63 for women who tus (9, Tables 2 and 3). Japanese reference stan- screened negative on the General Health Ques- dards for various categories of psychiatric tionnaire (GHQ) and 0.80 for women who patients are available (5, Table 3). Suzuki et al. screened positive. Interestingly, the women’s compared scores for several socioeconomic cate- husbands appeared to be less familiar with their gories from Japan, the United States, and Brazil spouse’s relative social adjustment: the equiva- (5, Table 5). lent correlations were lower, at 0.45 and 0.70 re- spectively (14, Table I). Correlations with the Alternative Forms Present State Examination scores ranged from An enlarged version (SAS-II), a semistructured 0.33 to 0.64 among cases and from 0.17 to 0.53 interview containing 56 items in eight role areas, among the noncases; correlations with the Pro- has been developed for patients with schizophre- file of Mood States ranged from 0.35 to 0.74 for nia. The scale takes about one hour to complete both groups (14, p72). and information may be obtained either from the Through Multi-Health Systems, translations patient or from a significant other. Agreement are available into Afrikaans, Chinese, Czech, between self-report and ratings by significant Danish, Dutch, Finnish, French (European and others was studied for 56 patients with schizo- Canadian) (12; 15; 16), German, Greek, He- phrenia, giving intraclass correlations from 0.27 brew, Hungarian, Italian, Japanese (5), Norwe- to 0.81 (10, Table 2). The multiple correlation of gian, Portuguese (17), Russian, Spanish the SAS-II with the section scores from the Brief (European and Latin American), and Swedish. A Psychiatric Rating Scale was 0.58 for 98 schizo- version for children has been tested (18). phrenic patients (11, Table 1). Data on inter- rater reliability of this version are given by Commentary Glazer et al. (10), whereas Toupin et al. reported The SAS continues to be the most widely used of on internal consistency (alpha, 0.61–0.81 for the all the scales reviewed in this chapter. It was subscales) and interrater reliability (0.74–0.94) based on a clearly defined conceptual approach (12). The Social Adjustment Scale for the Se- to the topic and drew items from another well- verely Mentally Ill (SAS-SMI) is an abbreviated established scale, the SSIAM. Its emphasis on  Social Health 193 successful adjustment places it in contrast with (2) Paykel ES, Weissman M, Prusoff BA, et al. the maladjustment measures of Linn, Gurland, Dimensions of social adjustment in and Remington. The SAS has been extensively depressed women. J Nerv Ment Dis used in psychiatric research, and Weissman pro- 1971;152:158–172. vided a lengthy discussion of the dimensions of (3) Weissman MM, Paykel ES, Prusoff BA. Social Adjustment Scale handbook: social health and of the components that may be rationale, reliability, validity, scoring, and modifiable through therapy for depression. The training guide. New Haven, CT: Yale scale is one of the few designed to measure the University School of Medicine outcomes of psychotherapy, which may seek less (Manuscript), 1985. to alleviate clinical symptoms than to improve (4) Weissman MM, Prusoff BA, Thompson interpersonal skills and relationships. Informa- WD, et al. Social adjustment by self-report tion on administering, scoring, and interpreting in a community sample and in psychiatric the SAS are available from Dr. Weissman, along outpatients. J Nerv Ment Dis with a bibliography of studies that have used the 1978;166:317–326. instrument. (5) Suzuki Y, Sakurai A, Yasuda T, et al. Weissman et al. have reviewed some limita- Reliability, validity and standardization of the Japanese version of the Social tions of the scale, including the difficulty of scor- Adjustment Scale-Self Report. Psychiatry ing patients who are too sick to undertake some Clin Neurosci 2003;57:441–446. of the roles (e.g., work). As originally proposed, (6) Weissman MM, Olfson M, Gameroff MJ, sections that are not applicable to an individual et al. A comparison of three scales for are omitted, but this means that a patient who assessing social functioning in primary subsequently assumes a role (such as starting to care. Am J Psychiatry 2001;158:460–466. work, perhaps at a low level) may receive a low (7) Weissman MM, Paykel ES, Siegel R, et al. score, thereby appearing to have deteriorated The social role performance of depressed (3). A more adequate scoring approach is women: comparisons with a normal group. required for such instances. Factor analyses sug- Am J Orthopsychiatry 1971;41:390–405. gested a grouping that cut across the two- (8) Weissman MM, Klerman GL, Paykel ES. Clinical evaluation of hostility in dimensional conceptual schema on which the depression. Am J Psychiatry instrument was constructed. It is therefore not 1971;128:261–266. clear that providing scores for each role area as (9) Richman J. Sex differences in social Weissman et al. did subsequently (1; 3) is the op- adjustment: effects of sex role socialization timal way to score the SAS; further examination and role stress. J Nerv Ment Dis of this issue would seem to be indicated if the 1984;172:539–545. SAS is to reach its potential as an outcome (10) Glazer WM, Aaronson HS, Prusoff BA, et measurement. al. Assessment of social adjustment in chronic ambulatory schizophrenics. J Nerv Address Ment Dis 1980;168:493–497. United States: Multi-Health Systems Inc. 908 (11) Glazer W, Prusoff B, John K, et al. Depression and social adjustment among Niagara Falls Blvd, North Tonawanda, NY chronic schizophrenic outpatients. J Nerv 14120-2060 (1-800-456-3003) Ment Dis 1981;169:712–717. Canada: Multi-Health Systems Inc. 3770 Vic- (12) Toupin J, Cyr M, Lesage A, et al. toria Park Ave, Toronto On M2H 3M6 (1-800- Validation d’une questionnaire d’évaluation 268-6011) (www.mhs.com) du fonctionnement social des personnes ayant des troubles mentaux chroniques. Can J Commun Ment Health References 1993;12:143–156. (1) Weissman MM, Bothwell S. Assessment of (13) Wieduwilt KM, Jerrell JM. The reliability social adjustment by patient self-report. and validity of the SAS-SMI. J Psychiatr Arch Gen Psychiatry 1976;33:1111–1115. Res 1999;33:105–112. 194 Measuring Health (14) Cooper P, Osborn M, Gath D, et al. ing, occupation and social roles, economic situa- Evaluation of a modified self-report mea- tion, leisure and social activities, family relation- sure of social adjustment. Br J Psychiatry ships, and marriage. Questions in each domain 1982;141:68–75. cover three themes that were described by Clare (15) Achard S, Chignon JM, Poirier-Littre MF, and Cairns as follows: et al. Social adjustment and depression: value of the SAS-SR (Social Adjustment the social schedule examines each individ- Scale Self-Report). Encephale 1995;21:107–116. ual’s life from 3 main standpoints: first, it at- (16) Waintraud L, Guelfi JD, Lancrenon S, et al. tempts to assess what the individual has, in [Validation of M. Weissman’s social terms of his living conditions, money, social adaptation questionnaire in its French opportunities in a number of areas; secondly, version]. Ann Med Psychol (Paris) it measures what he does with his life, what 1995;153:274–277. use he makes of his opportunities, how well (17) Gorenstein C, Moreno RA, Bernik MA, et he copes; finally, it measures what he feels al. Validation of the Portuguese version of about it, that is to say how satisfied he is with the Social Adjustment Scale on Brazilian various aspects of his social situation. (1, samples. J Affect Disord 2002;69:167–175. p592) (18) Clemente C, Tsiantis J, Kolvin I, et al. Social adjustment in three cultures: data from families affected by chronic blood A trained interviewer administers the semistruc- disorders. A sibling study. Haemophilia tured interview in the respondent’s home; the in- 2003;9:317–324. terviewer may also incorporate information collected from the spouse. The interview re- quires about 45 minutes. The content of the The Social Maladjustment Schedule schedule is summarized in Exhibit 4.12. (Anthony W. Clare, 1978) From the individual’s responses the inter- Purpose viewer makes a total of 42 ratings on four-point This rating form was designed to measure social scales that describe the extent of maladjustment maladjustment among adults with chronic neu- in each of the three areas. Ten ratings cover ma- rotic disorders. Originally intended for use in terial conditions, 14 refer to management of so- psychiatric research, it has also been used in cial opportunities and activities, and 18 cover studies in family practice and with general popu- satisfaction (2). The ratings concentrate on mal- lation samples. adjustment; no gradation is made of levels of sat- isfactory functioning. An overall score may also Conceptual Basis be used, with higher scores indicating poorer ad- Clare and Cairns argued that scales measuring justment. social adjustment in terms of conformity to so- cial roles and norms will not permit compar- Reliability isons across social groups among which norms Inter-rater reliability was assessed using analyses and social expectations differ. They designed of variance; agreement was close with the excep- their scale to combine an interviewer’s objective tion of 3 of the 25 items tested, for which signif- assessment of the patient’s material circum- icant differences were obtained (1, Table 3). stances and performance with the patient’s own Weighted kappas ranged from 0.55 to 0.94 with ratings of satisfaction. The topics covered in the most coefficients falling above 0.70 (1, Table 4). scale were derived from a review of previous measurements (1). Validity Factor analyses were applied to various samples, Description but the results did not clearly replicate the di- The Social Maladjustment Schedule (SMS) is a mensions around which the schedule was con- 26-page interview that covers six domains: hous- structed, and from this Clare and Cairns inferred Exhibit 4.12 Structure and Content of the Social Maladjustment Schedule Rating category with each item rated shown below the appropriate category Subject area Material conditions Social management Satisfaction Housing Housing conditions Household care Satisfaction with housing Residential stability Management of housekeeping Occupation/social role Occupational stability Quality of personal Satisfaction with interaction with occupation/social role workmates (includes housewives, unemployed, disabled, retired) Opportunities for Satisfaction with interaction with personal interaction workmates* with workmates Economic situation Family income Management of income Satisfaction with income Leisure/social activities Opportunities for leisure Extent of leisure and Satisfaction with leisure and social activities* social activities and social activities Opportunities for Quality of interaction Satisfaction with interaction with neighbors with neighbors interaction with neighbors Satisfaction with heterosexual role Family and domestic Opportunities for Quality of interaction Satisfaction with relationships interaction with relatives* with relatives interaction with relatives Quality of solitary living Satisfaction with solitary living Opportunities for Quality of domestic Satisfaction with domestic interaction (i.e., interaction (i.e., with domestic interaction with unrelated others or unrelated others or adult offspring in adult offspring in household) household) Situational handicaps to Child management Satisfaction with child management* parental role Marital Fertility and family planning Sharing of responsibilities Satisfaction with marital and decision-making harmony Sharing of interests and Satisfaction with sexual activities compatibility *This group of items rates objective restrictions which might be expected to impair functioning in the appropriate area. “Situa- tional handicaps to child management” assesses difficulties likely to exacerbate normal problems of child-rearing, e.g., inadequate living space, an absent parent. Objective restrictions on leisure activities include extreme age, physical disabilities, heavy domestic or work commitments, isolated situation of the home, etc. Reproduced with permission from Clare AW, Cairns VE, Design, development, and use of a standardized interview to assess social maladjustment and dysfunction in community studies. Psychol Med 1978;8:592, Table 1. Copyright Cambridge University Press. Reprinted with the permission of Cambridge University Press. 195 196 Measuring Health the need to calculate an overall score. This over- use the instrument should be taken in the light of all maladjustment score was associated (at future evidence on its validity and reliability. p < 0.05) with a rating made using Goldberg’s standardized psychiatric interview. From Clare and Cairns’s data, a sensitivity of 30% and a References specificity of 80% may be calculated, compared (1) Clare AW, Cairns VE. Design, development with the Goldberg rating (1, Table 7). and use of a standardized interview to Correlations between the SF-36 and various assess social maladjustment and scales on the SMS have been reported. These dysfunction in community studies. Psychol were generally low, the highest being 0.36 (3, Med 1978;8:589–604. Table 1). Convergent correlations between SF-36 (2) A manual for use in conjunction with the social functioning and social contacts on the General Practice Research Unit’s standardized social interview schedule. Maladjustment Schedule of 0.31; the correlation London: Institute of Psychiatry, 1979. between general mental health and the leisure (3) Stansfield SA, Roberts R, Foot SP. score was 0.32. Assessing the validity of the SF-36 general health survey. Qual Life Res Alternative Forms 1997;6:217–224. A 41-item (later abridged to 33-item) self-report (4) Corney RH, Clare AW, Fry J. The Social Problem Questionnaire has been derived development of a self-report questionnaire from the SMS (4; 5). Murray et al. modified the to identify social problems: a pilot study. SMS by omitting items not applicable to some Psychol Med 1982;12:903–909. subjects (e.g., living alone) and by collecting (5) Corney RH, Clare AW. The construction, development and testing of a self-report more detailed information on family and social questionnaire to identify social problems. relationships (6). Psychol Med 1985;15:637–649. (6) Murray D, Cox JL, Chapman G, et al. Commentary Childbirth: life event or start of a long- Clare and Cairns (1) offered a thorough concep- term difficulty? Br J Psychiatry tual discussion of the development of measures 1995;166:595–600. of social health by reviewing the problems of (7) Clare AW. Psychiatric and social aspects of comparing social behavior across cultural pre-menstrual complaint. Psychol Med groups and discussing the balance to be main- 1983;13(suppl 4):1–58. tained between recording objective life circum- (8) Corney RH. Social work effectiveness in stances and personal satisfaction. Although their the management of depressed women: a clinical trial. Psychol Med scale was explicitly designed to reflect this dis- 1981;11:417–423. tinction, empirical data from factor analyses did not confirm the intended conceptual structure. The scale seems well-designed, and has seen The Social Dysfunction Rating Scale occasional use, mainly in Great Britain (4–8). (Margaret W. Linn, 1969) However, beyond the initial development work, little further evidence for the reliability and valid- Purpose ity of the scale has been published. The scale cov- The Social Dysfunction Rating Scale (SDRS) as- ers only the negative aspects of social adjustment; sesses the negative aspects of a person’s social for assessing social support or to cover positive adjustment. This rating scale is applied by a cli- indications of integration, a different scale, such nician and is intended as a research instrument, as that of Henderson or McFarlane, would be mainly for use with the elderly. needed. The SMS may find a role in studies that require detailed information on problems in so- Conceptual Basis cial adjustment and that have the resources to Effective social functioning, in the Linn et al. carry out home interviews, but the decision to conceptual formulation,  Social Health 197 would suggest equilibrium within the per- the respondent’s own self-evaluation (1, p301). son and in his interaction with his environ- For instance, the interviewer rates the availabil- ment. . . . Dysfunction, on the other hand, ity of friends and social contacts, after which the implies discontent and unhappiness, accom- respondent is asked if he feels a need for more panied by negative self-regarding attitudes. friends. Hence, the person who has few friends It furthermore suggests handicapping anxi- and is discontented about this will receive a ety and other pathological interpersonal lower rating than the person with few friends functions that reduce flexibility in coping but who is not concerned about it. The interview with stressful situations or achieving self- lasts about 30 minutes (2, p617). actualization in what is to that person a Linn et al. provide definitions of the items significant role. . . . From this standpoint, and instructions for completing the scale. As dysfunction is seen as coping with either an example, comments on item 4 read as fol- personal, interpersonal, or geographic envi- lows: ronment in a maladaptive manner. In this re- spect, the SDRS seeks to quantify the 4. Self-health concern. The frequency and objective observations of man’s dysfunc- severity of complaints of body illness are tional interaction with his environment. (1, rated. Evaluation is based on degree to which p299) the person believes that physical symptoms are an important factor in his total well- Linn viewed adjustment as a process of coping, being. No consideration is given for actual problem-solving, and achieving personal goals organic basis of illness. Only the frequency (2, p617). As a dysfunction scale, however, the and severity of complaints are rated. (1, SDRS concentrates on symptoms of low morale p301) and reduced social participation; it does not as- sess positive adjustment. The assessments do not Higher scores on the scale reflect greater dys- emphasize particular roles, which makes the in- function. Items are not weighted differentially, strument suitable for elderly people. although Linn et al. considered using discrimi- nant function coefficients as item weights (1, The SDRS is applicable to older patients, par- p305). ticularly with respect to assessing the mean- ingfulness of their life, their goals, and their Reliability satisfactions. It does not provide descriptive The agreement between two raters in scoring 40 assessments of different kinds of activities. subjects was measured; intraclass correlations Work is rated on the basis of productive ac- for the 21 items ranged from 0.54 to 0.86 (1, tivities . . . and whether these generate feel- Table 1). The agreement between seven raters, ings of usefulness. (2, p617) who independently rated ten schizophrenics in group interviews, yielded a Kendall index of Description concordance of 0.91(1, p303). The SDRS is applied by an interviewer, generally a social worker or other therapist familiar with Validity the patient. The scale, shown in Exhibit 4.13, in- The scale was applied to schizophrenic outpa- cludes 21 symptoms of social and emotional tients and nonpsychiatric respondents. Using problems, each judged on a six-point severity discriminant function analysis, it correctly classi- scale. The ratings are grouped into three classes: fied 92% of the 80 respondents (1, Table 2). In four items refer to the respondent’s self-image, the same study, a correlation of 0.89 was ob- six refer to interpersonal relationships, and 11 tained between the total scale scores and a concern lack of success and dissatisfaction in so- global judgment of adjustment made by a social cial situations. The questions are semistructured worker who interviewed the respondents (1, and combine the interviewer’s evaluations with p305). The data were factor analyzed, produc- 198 Measuring Health Exhibit 4.13 The Social Dysfunction Rating Scale Directions: Score each of the items as follows: 1. Not Present 3. Mild 5. Severe 2. Very Mild 4. Moderate 6. Very severe Self system 1. ____________ Low self concept (feeling of inadequacy, not measuring up to self ideal) 2. ____________ Goallessness (lack of inner motivation and sense of future orientation) 3. ____________ Lack of a satisfying philosophy or meaning of life (a conceptual framework for integrating past and present experiences) 4. ____________ Self-health concern (preoccupation with physical health, somatic concerns) Interpersonal system 5. ____________ Emotional withdrawal (degree of deficiency in relating to others) 6. ____________ Hostility (degree of aggression toward others) 7. ____________ Manipulation (exploiting of environment, controlling at other’s expense) 8. ____________ Over-dependency (degree of parasitic attachment to others) 9. ____________ Anxiety (degree of feeling of uneasiness, impending doom) 10. ____________ Suspiciousness (degree of distrust or paranoid ideation) Performance system 11. ____________ Lack of satisfying relationships with significant persons (spouse, children, kin, significant persons serving in a family role) 12. ____________ Lack of friends, social contacts 13. ____________ Expressed need for more friends, social contacts 14. ____________ Lack of work (remunerative or non-remunerative, productive work activities which normally give a sense of usefulness, status, confidence) 15. ____________ Lack of satisfaction from work 16. ____________ Lack of leisure time activities 17. ____________ Expressed need for more leisure, self-enhancing and satisfying activities 18. ____________ Lack of participation in community activities 19. ____________ Lack of interest in community affairs and activities which influence others 20. ____________ Financial insecurity 21. ____________ Adaptive rigidity (lack of complex coping patterns to stress) Reproduced from Linn MW, Sculthorpe WB, Evje M, Slater PH, Goodman SP. A Social Dysfunction Rating Scale. J Psychiatr Res 1969;6:300. Copyright (1969), with permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK. ing five factors: apathetic-detachment, dissatis- ment when there are no serious limitations on faction, hostility, health-finance concern, and staff and patient time” (2, p617). manipulative-dependency (1). Commentary Alternative Forms The SDRS was based on considerable conceptual A self-report version has been used by Linn, al- work on the theme of social adjustment among though she argued that “the original version of the elderly (1–4). It is a broad-ranging instru- the scale provides a better assessment of adjust- ment, overlapping in content with the morale  Social Health 199 and life satisfaction scales described in Chapter The Structured and Scaled 5. Inter-rater reliability can be high for overall Interview to Assess Maladjustment scores, although agreement for individual scales (Barry J. Gurland, 1972) shows a wide variation. There is little evidence on validity. Although it was first described in Purpose 1969 and has been used in several studies (5–8), The Structured and Scaled Interview to Assess the only validity results come from a single study Maladjustment (SSIAM) provides a detailed of 80 subjects. The question of how best to score clinical assessment of social role performance as the scale also requires clarification, especially be- an outcome indicator for psychotherapy. It has cause the empirical factor analysis results do not been used in both clinical and research applica- match the three subdivisions built into the scale tions (1). (e.g., self-perceptions, interpersonal relations and social performance). It is not clear whether a Conceptual Basis total score or subscores offer a better way to Gurland et al. held that the relevance of measur- summarize the results. The SDRS offers a brief ing social maladjustment derives from the find- and rather narrower alternative to Clare’s Social ing that much psychiatric treatment seeks to Maladjustment Schedule. assist people in becoming more socially effective and in reducing distress, deviant behavior, and References friction with others (2). The questions in the SSIAM “cover those aspects of social adjustment (1) Linn MW, Sculthorpe WB, Evje M, et al. A which are of interest to a clinician” (2). The social dysfunction rating scale. J Psychiatr scale was derived from Parloff ’s 1954 Social In- Res 1969;6:299–306. effectiveness Scale. Gurland et al. distinguished (2) Linn MW. A critical review of scales used to evaluate social and interpersonal between objective and subjective facets of mal- adjustment in the community. adjustment. Objectively, maladjustment is Psychopharmacol Bull 1988;24:615–621. viewed as ineffective performance of social roles; (3) Linn MW. Studies in rating the physical, subjectively, it refers to a failure to obtain satis- mental, and social dysfunction of the faction from one’s social activities (2). The chronically ill aged. Med Care SSIAM covers both facets of maladjustment, in- 1976;14(suppl 5):119–125. dicating levels of distress, deviant behavior, and (4) Linn MW. Assessing community adjustment friction. Assessments of maladjustment must in the elderly. In: Raskin A, Jervik LF, eds. also consider the patient’s environment, because Assessment of psychiatric symptoms and an unfavorable environment may in part explain cognitive loss in the elderly. Washington, distress or disturbed behavior. To cover this, the DC: Hemisphere Press, 1979:187–204. (5) Linn MW, Caffey EM Jr. Foster placement SSIAM includes a rating by the interviewer in for the older psychiatric patient. J Gerontol each section indicating to what degree the rat- 1977;32:340–345. ings are due to a currently unfavorable environ- (6) Linn MW, Caffey EM Jr, Klett CJ, et al. ment. Hospital vs community (foster) care for psychiatric patients. Arch Gen Psychiatry Description 1977;34:78–83. The instrument contains 60 items, 45 of which (7) Linn MW, Klett CJ, Caffey EM Jr. Foster are grouped into five “fields”: work, social rela- home characteristics and psychiatric patient tions, family, marriage, and sex. The remaining outcome: the wisdom of Gheel confirmed. Arch Gen Psychiatry 1980;37:129–132. 15 items are used to record the interviewer’s (8) Linn MW, Caffey EM, Klett CJ, et al. Day judgments of the level of stress in the patient’s treatment and psychotropic drugs in the environment; of his prognosis and willingness to aftercare of schizophrenic patients. Arch change; and of aspects of positive mental health Gen Psychiatry 1979;36:1055–1066. such as personality strengths. 200 Measuring Health Within each of the five fields the assessments Reliability follow a standard order: five deal with the pa- Fifteen patients were interviewed by one of three tient’s deviant behavior, one deals with friction psychiatrists; all three then rated each patient, ei- between the patient and others, and three deal ther during the interview or from a tape record- with the patient’s distress (2). Questions refer to ing of it. Intraclass correlations among raters behavior over the past four months (1). Gurland were calculated for six factor scores and ranged et al. describe the structure of the questions as from 0.78 to 0.97 (3, p265). Analyses of vari- follows: ance showed no significant differences among the raters, but small differences among them were obtained for the scores on social isolation Each item has a caption indicating the distur- and friction in relationships with people other bance covered, a question which the rater than family members (3). asks the patient, and a continuous scale with five anchoring definitions. The highest an- Validity choring definition describes the maximum Using a sample of 164 adults “considered suit- disturbance likely to be found in an outpa- able for outpatient psychotherapy” (70% of tient psychoneurotic population. The lowest whom were students), 33 of the 45 subjective describes reasonable adjustment. The remain- items were factor analyzed. Twelve items were ing three definitions represent successive lev- found not to load on factors; the remaining 21 els of disturbance between the extremes. (2, items loaded on six factors covering social isola- pp261–262) tion, work inadequacy, friction with family, de- pendence on family, sexual dissatisfaction, and The questions are asked open-ended and the in- friction outside the family (3). For 89 patients, a terviewer matches the reply to the closest phrase relative or close friend of the patient was inter- printed on the interview schedule. If there is viewed to provide independent ratings of these doubt about which rating best matches the reply, six themes. There was significant agreement be- the interviewer reads the two most applicable tween the SSIAM and the informants’ ratings for categories (in a preset order), effectively imple- all factors except sexual dissatisfaction (3). menting a forced-choice response (1). The scale Serban and Gidynski have reported on the positions of the defining phrases were deter- performance of the SSIAM with 100 schizo- mined by four psychotherapists in a scaling task phrenic patients (4; 5). The correlations between (2). The response categories are unique to each the SSIAM scores and the total score derived item, thereby reducing the likelihood that an in- from a psychiatrist’s evaluation ranged from 0.21 terviewer will use a particular response category and 0.41 (4, Table 1). The SSIAM correlated across several questions. The interview takes 0.45 with the Social Stress and Functioning In- about 30 minutes. The questionnaire is too long ventory for Psychotic Disorders (4, p950). Serban to reproduce here, but an indication of the scope and Gidynski also showed that the SSIAM dis- of the instrument is given in Exhibit 4.14. Defin- criminated significantly between different types itions of terms are given on the rating form as of schizophrenic patients (5). The SSIAM has part of the item: an illustration is given in Ex- also been shown capable of identifying significant hibit 4.15. An instruction manual is included in changes before and after psychotherapy (6). the 30-page interview booklet (1). Raw scores from 0 to 10 for each scale may Commentary be summed across each of the five fields, or each The descriptions of the SSIAM given by Gurland field may be scored in terms of deviant behavior, et al. are extremely clear. Great care was evi- friction between the patient and others, and the dently taken in the design of the questionnaire patient’s distress. Alternatively, factor scores and the interviewer instructions are exemplary. may be used (see Validity heading in this sec- The conceptual basis for this scale, contrasting tion). objective and subjective indexes of adjustment, Exhibit 4.14 Scope of the SSIAM Showing Arrangement of Items within Each Section Fields of maladjustment Type of item Caption of items { Behavior Unstable, inefficient, unsuccessful, over-working, over-submissive Friction Friction WORK Distress Disinterested, distressed, feeling inadequate Inferential Rater’s assessment of environmental stress { Behavior Isolated, constrained, unadaptable, apathetic in leisure, unconforming Friction Friction SOCIAL Distress Distressed by company, lonely, bored by leisure Inferential Rater’s assessment of environmental stress { Behavior Reticent, over-compliant, rebellious, family-bound, withdrawn FAMILY Friction Friction Distress Guilt-ridden, resentful, fearful Inferential Rater’s assessment of environmental stress { Behavior Constrained, submissive, domineering, neglectful, over-dependent Friction Friction MARRIAGE Distress Distressed, feeling deprived, feeling inadequate Inferential Rater’s assessment of environmental stress { Behavior Undesirous, inadequate, inactive, cold, promiscuous SEX Friction Rejected by partner Distress Tension, feeling deprived, unwanted urges Inferential Rater’s assessment of environmental stress { Global Extent of patient’s distress, exaggerating, minimizing Prognostic Duration, contrast with previous state, OVERALL willingness to change, pressure from others to change Positive mental Strengths and assets, resourcefulness, health constructive effort Reproduced from Gurland BJ, Yorkston NJ, Stone AR, Frank JD, Fleiss JL. The Structured and Scaled Interview to Assess Malad- justment (SSIAM): I. Description, rationale, and development. Arch Gen Psychiatry 1972;27:263. Copyright © 1972, American Medical Association. With permission. 201 202 Measuring Health Exhibit 4.15 An Example of Two Items Drawn from the SSIAM: Social and Leisure Life Section Friction Distress S6 FRICTION *S7 DISTRESSED BY COMPANY Q: How smoothly and well do you get along with Q: Are you ill at ease, tense, shy or upset when your friends and close acquaintances? with friends? 1) Rate overt behavior between the patient and others. 1) Only rate distress occurring in friendly and The patient’s subjective responses are rated under #S7 informal company. 2) Mild initial shyness or mild anticipatory anxiety should be rated as “reasonable.” 3) Include distress from any other source which interferes with the enjoyment of company. Higher first Lower first Frequently has furious clashes or is studiously Company is mainly a source of agonizing avoided by others. distress. Often irritates others or is treated with reserve Company is mainly a source of marked by them. distress. Sometimes relationships with others somewhat Company is sometimes a source of distress but uneasy and tense. often enjoyable. Not provocative but can not handle delicate Company is unnecessarily distressing only in social situations. special circumstances but usually enjoyable. Reasonably diplomatic. Company is enjoyed with reasonable ease. — Not known — Not known — Not applicable — Not applicable scope scope The frequency and intensity of his aggressive The frequency and intensity of distress when in actions towards others, and the seriousness of the company, and enjoyment of company. reaction he provokes in others. Reproduced from Gurland BJ, Yorkston NJ, Stone AR, Frank JD. Structured and Scaled Interview to Assess Maladjustment (SSIAM). New York: Springer, 1974:10. With permission. corresponds well with other available ap- depression (7) and as an outcome indicator for proaches, and the approach used in the SSIAM psychotherapy (6). has influenced the design of subsequent measure- The expectation that three manifestations of ments such as the OARS Multidimensional Func- maladjustment (i.e., behavior, friction, distress) tional Assessment Questionnaire and Weissman’s would appear across all five fields of maladjust- Social Adjustment Scale. The SSIAM is one of ment received little empirical support from the the more widely used of the social health indexes, factor analytic study, so careful consideration having been applied, for example, in studies of must be given to how the instrument is scored.  Social Health 203 We would also like to see considerably more evi- conceptual discussion of social disability was dence for the reliability and validity of the in- given by Ruesch in presenting his Rating of Social strument, including correlations with other Disability (5; 6). The scale itself summarizes phys- social health measurement scales. With these ical and emotional impairment and describes the reservations, we recommend the SSIAM where resulting impact on social role functioning; it is time permits a thorough assessment of a broad completed by a psychiatrist, social worker, or psy- range of types and levels of disorder. chologist. Ruesch’s scale has seldom been reported in the literature and there is no published evidence for its validity. References Roen’s Community Adaptation Schedule was (1) Gurland BJ, Yorkston NJ, Stone AR, et al. developed to evaluate the success of aftercare Structured and Scaled Interview to Assess programs for mental patients discharged to the Maladjustment (SSIAM). New York: community (7; 8). It is a 202-item interview that Springer, 1974. covers work, family relationships, social interac- (2) Gurland BJ, Yorkston NJ, Stone AR, et al. tion, social activities, and ADLs. It employs an The Structured and Scaled Interview to interesting manner of collecting information in Assess Maladjustment (SSIAM): I. that each question is asked in three modes. The Description, rationale, and development. first records factual information on circum- Arch Gen Psychiatry 1972;27:259–264. stances or describes behavior, the second covers (3) Gurland BJ, Yorkston NJ, Goldberg K, et affective responses to these, and the third as- al. The Structured and Scaled Interview to Assess Maladjustment (SSIAM): II. Factor sesses the patient’s cognitive responses: for ex- analysis, reliability, and validity. Arch Gen ample, plans the patient has made or his Psychiatry 1972;27:264–267. understanding of how other people feel about (4) Serban G. Mental status, functioning, and him. The scale fell into disuse after a spate of stress in chronic schizophrenic patients in validation studies in the early 1970s, most of community care. Am J Psychiatry which showed only modest agreement with 1979;136:948–952. other scales (results that did not, however, deter (5) Serban G, Gidynski CB. Relationship the authors from inferring good construct valid- between cognitive defect, affect response ity for the method) (9; 10). The interpretation of and community adjustment in chronic Roen’s scale remains unclear and more evidence schizophrenics. Br J Psychiatry is required on its association with other social 1979;134:602–608. (6) Cross DG, Sheehan PW, Khan JA. Short- scales—such as those considered in this chap- and long-term follow-up of clients ter—before we can recommend its use. Further receiving insight-oriented therapy and references to validation studies can be found in behavior therapy. J Consult Clin Psychol the article by Harris and Brown (10). 1982;50:103–112. The Social Disability Questionnaire by (7) Paykel ES, Weissman M, Prusoff BA, et al. Branch deserves mention as one of the few scales Dimensions of social adjustment in designed for use in general population surveys. It depressed women. J Nerv Ment Dis is a self-report instrument that estimates need for 1971;152:158–172. help in performing daily tasks among elderly people in the general population (11). An inno- vative feature is its provision of a high-risk score Conclusion that anticipates the possible development of fu- ture problems. Termed a social disability ques- In addition to the scales reviewed here, a large tionnaire, the scale resembles the IADL scales number of other measures were considered for in- discussed in Chapter 3, but it also considers so- clusion in this chapter. Several have been described cial support and social interaction. The scale in review articles by Linn (1), Donald et al. (2), lacks evidence on reliability and validity and has and Weissman et al. (3; 4). An extensive and useful seen only limited use. 204 Measuring Health The Community Adjustment Profile System derly people. It should not be confused with the was designed for use in the long-term monitoring other Duke social support measures reviewed of patients’ adjustment. Using 60 questions, it elsewhere in this chapter. The 35-item DSSI cov- covers ten aspects of adjustment and was designed ers social network, social interactions, subjective for computer scoring. Test-retest reliability was feelings of social support, and instrumental or quoted at 0.83 and internal consistency of the practical support. Abbreviated versions (23- and scales ranged from 0.70 to 0.92 (12, p533). A 11-items) have been produced (19). The DSSI newer instrument, the Social Functioning Scale, was used in two Australian studies: The Preven- was developed for use with psychiatric patients tive Care Trial for veterans and the Australian and assesses strengths and weaknesses in the pa- Longitudinal Study on Women’s Health, a 20- tient’s social functioning. Preliminary evidence year cohort study of 12,000 elderly women. We for reliability and validity is promising (13). recommend that readers planning to measure so- An early scale occasionally mentioned in in- cial support obtain recent evidence on the fur- troductory discussions, but seldom used in pub- ther testing of these scales. Finally, as a totally lished studies, is the 1968 Personality and Social different approach to measuring social function- Network Adjustment Scale by Clark (14). This ing, Norton and Hope review the validity of was designed to evaluate social adjustment role-play assessments as an alternative to inter- among severely ill psychiatric patients receiving view measures (20). These are primarily used for treatment in a therapeutic community. There is clinical purposes and typically occur in a struc- relatively good evidence for the validity and reli- tured and standardized role-play situation ability of this scale and an abbreviated version is (“analogue assessment”), which is more practi- shown in Clark’s report; it is worth considera- cal than observing the person’s interactions in tion where a brief and simple rating is required. naturally occurring encounters. Norton and Another scale often cited as one of the seminal Hope report a wide range of reliability and valid- efforts in the field was developed by Renne (15). ity studies of these assessments. We have not described this as it has virtually no Two types of scale have been reviewed in this published validation data and has seldom been chapter: social adjustment scales and measure- used. ments of social support. Among the former, the A promising social support scale has been de- Social Adjustment Scale of Weissman is the most veloped by Norbeck (16). The Norbeck Social carefully developed and shows the highest levels Support Questionnaire is based on an explicit of validity and reliability. Henderson’s social in- conceptual discussion of support; it showed test- teraction scale also shows attention to concep- retest reliability coefficients between 0.85 and tual and empirical development, and study 0.92 as well as high internal consistency (16, results have made contributions to the literature p267). Another instrument, the Personal Re- on social aspects of disease. Among the social source Questionnaire developed by Brandt and support scales, none is clearly superior, mainly Weinert, is in two sections. The first provides de- because few have been widely tested. Of those scriptive information on the person’s resources we review, Sarason’s scale appears to be the and satisfaction with these, and the second sec- most promising, but readers should search for tion includes questions that reflect Weiss’s di- more recent validity reports on the scales. mensions of social support (17). Alpha internal consistency for the second part is 0.89 and valid- ity coefficients ranged from 0.30 to 0.44. Valid- References ity coefficients for the first part were somewhat (1) Linn MW. Assessing community lower, ranging from 0.21 to 0.23 (17, p279). adjustment in the elderly. In: Raskin A, The Inventory of Socially Supportive Behaviors Jervik LF, eds. Assessment of psychiatric (ISSB) measures the actual provision of support symptoms and cognitive loss in the elderly. (18). Finally, the Duke Social Support Index Washington, DC: Hemisphere Press, (DSSI) is designed for use with chronically ill el- 1979:187–204.  Social Health 205 (2) Donald CA, Ware JE, Jr, Brook RH, et al. the aging. Am J Public Health Conceptualization and measurement of 1981;71:1202–1210. health for adults in the Health Insurance (12) Evenson RC, Sletten IW, Hedlund JL, et al. Study. Vol. IV, Social Health. Santa CAPS: an automated evaluation system. Monica, CA: RAND Corporation, 1978. Am J Psychiatry 1974;131:531–534. (3) Weissman MM. The assessment of social (13) Birchwood M, Smith J, Cochrane R, et al. adjustment: a review of techniques. Arch The Social Functioning Scale: the Gen Psychiatry 1975;32:357–365. development and validation of a new scale (4) Weissman MM, Sholomskas D, John K. of social adjustment for use in family The assessment of social adjustment: an intervention programmes with update. Arch Gen Psychiatry schizophrenic patients. Br J Psychiatry 1981;38:1250–1258. 1990;157:853–859. (5) Reusch J, Brodsky CM. The concept of (14) Clark AW. The Personality and Social social disability. Arch Gen Psychiatry Network Adjustment Scale: its use in the 1968;19:394–403. evaluation of treatment in a therapeutic (6) Ruesch J, Jospe S, Peterson HW Jr, et al. community. Hum Relat 1968;21:85–96. Measurement of social disability. Compr (15) Renne KS. Measurement of social health in Psychiatry 1972;13:507–518. a general population survey. Soc Sci Res (7) Roen SR, Ottenstein D, Cooper S, et al. 1974;3:25–44. Community adaptation as an evaluative (16) Norbeck JS, Lindsey AM, Carrieri VL. The concept in community mental health. Arch development of an instrument to measure Gen Psychiatry 1966;15:36–44. social support. Nurs Res 1981;30:264–269. (8) Burnes AJ, Roen SR. Social roles and (17) Brandt PA, Weinert C. The PRQ—a social adaptation to the community. Community support measure. Nurs Res Ment Health J 1967;3:153–158. 1981;30:277–280. (9) Cook PE, Looney MA, Pine L. The (18) Barrera MJ, Sandler IN, Ramsay TB. Community Adaptation Schedule and the Preliminary development of a scale of Adjective Check List: a validational study social support: studies of college students. with psychiatric inpatients and outpatients. Am J Community Psychol Community Ment Health J 1973;9:11–17. 1981;9:435–447. (10) Harris DE, Brown TR. Relationship of (19) Koenig HG, Westlund RE, George LK, et al. the Community Adaptation Schedule and Abbreviating the Duke Social Support Index the Personal Orientation Inventory: two for use in chronically ill elderly individuals. measures of positive mental health. Psychosomatics 1993;34:61–69. Community Ment Health J (20) Norton PJ, Hope DA. Analogue 1974;10:111–118. observational methods in the assessment of (11) Branch LG, Jette AM. The Framingham social functioning in adults. Psychol Assess Disability Study: 1. social disability among 2001;13:59–72.  5 Psychological Well-being his brief chapter serves as a prelude to the sary because people were expected to be reticent T more detailed chapters on anxiety, depres- sion, and mental status measurement that follow. about reporting their true feelings. Indeed, it was estimated that underreporting of emotional It reviews scales that provide broad summaries of problems in surveys might be as high as 60% psychological well-being, including positive men- (2). Thus, for example, Macmillan deliberately tal states. This book does not review psychiatric named his screening scale the “Health Opinion diagnostic instruments nor methods used in eval- Survey” to conceal its intent. Conversely, symp- uating severe psychiatric disorders, such as tom checklists almost certainly misclassify some schizophrenia. Measurements reviewed in this physical disorders as psychological; they can chapter cover transitory psychological states detect only more severe forms of disorder and rather than more persistent traits and describe they cannot identify emotional distress unless it human psychological responses in adapting to is manifested somatically or behaviorally. the inherent environmental challenges. Psycho- Questions on feelings are needed to encompass logical well-being is frequently recorded in social the positive end of the mental health spectrum, surveys, both as a component of subjective qual- because they can be phrased to differentiate lev- ity of life and as an outcome in studies of stress, els of health among asymptomatic people. social support, and coping. Subsequently, the argument that people will The review is not exhaustive. The presenta- not respond honestly to direct questions about tion traces the historical evolution in the their emotional well-being largely passed from approach to measurement, beginning with favor. Influenced by criticisms of the symptom checklists that recorded symptoms of distress checklist approach and by acceptance in the (1). Measurement scales used in the 1930s and 1950s of the potential accuracy of subjective 1940s took the form of checklists that included reports, Gurin, and later Bradburn, led a move- behavioral and somatic symptoms of distress. ment in the United States toward surveying feel- Feelings of distress have long been considered a ings of happiness and emotional well-being. nonspecific indicator of mental health; because This trend also reflected the theme of positive distress is often a stimulus to seek care, measures mental health, a concept that may be traced of distress represent a clinical orientation. back, through Jahoda’s work (3), to the 1947 However, the scales we review stop short of World Health Organization (WHO) conception making diagnostic classifications; they offer gen- of health. These measures recorded the affec- eral screens. They can also only indicate well- tive responses to experience—the feeling states being in terms of the absence of distress, and so inspired by daily experience. They approached later gave way to an approach that asked direct- psychological well-being largely as a cognitive ly about positive and negative feelings of well- process in which people compare their percep- being. The arguments in support of the symp- tions of their current situation with their aspi- tom checklist held that it is more objective, and rations. This led to approaching well-being in that it more adequately conceals the intent of the terms of life satisfaction, represented here by measurement; this was formerly deemed neces- measures such as the Life Satisfaction Index 206  Psychological Well-being 207 and the Philadelphia Geriatric Center Morale reflects the difficulty of establishing firm concep- Scale. tual definitions—a problem that we have seen These scales can show strong intercorrela- over the past ten years in the area of quality of tions but appear less well suited to screening for life measurement. Dohrenwend et al. long ago psychological or psychiatric disorders, and later commented critically on indicators of “non- work suggested that the reaction against the specific psychological distress”: symptom checklist approach may have been too strong. Although these checklists have limita- As might be expected given the actuarial pro- tions, many validation studies have shown that cedures and undifferentiated patient criterion they can achieve high sensitivity and specificity groups used to construct them, none of the when compared with psychiatric ratings. screening scales reflects a clearly specified Checklists do detect mental disorders, so that conceptual domain. Thus, there is no ready more recent scales, such as those developed by correspondence between the content of the Dupuy and by Goldberg, have combined the scales and conceptions of major dimensions checklist and questionnaire approaches to form or types of psychopathology such as mania, a hybrid. This evolution has formed a dialectical depression, hallucinations, or antisocial be- process, with the more recent methods repre- havior. (4, p1229) senting a synthesis of the earlier approaches. The newer scales may offer the best of both worlds Dohrenwend et al. suggested that these scales and can be used in studies of the protective give general indications of distress, analogous to impact of positive mental health (e.g., in pre- the measurement of body temperature: elevated venting cancer) and in studies of “wellness.” scores tell you that something is wrong, but not There have been many attempts to specify what (4). In this chapter, we refer to distress what is being measured—to distinguish, for rather than disorder, and we use the broad term example, between “distress” and “disorder,” “psychological” to connote a general level of among “feelings,” “mood,” and “affect,” and discussion, commonly including emotional, and, among “psychological,” “emotional,” and at times, mental problems. Because distress cov- “mental” well-being. The attempts have not ers only the negative end of the spectrum, many always been successful and the development of scales use the broader term “affect.” Affect this field has been distracted by disputes over the refers to positive or negative subjective feelings intent and conceptual interpretation of the and moods that are genuine and personal (as scales. Unfortunately, this was exacerbated, in when someone says “I am feeling over- the earlier scales at least, by the authors’ failure whelmed”). It does not cover thoughts or cogni- to explain conceptually what they were attempt- tive reactions to situations or objects (“I hate my ing to measure. Many earlier scales were devel- job”) (5). oped empirically by selecting questions that dis- A related area of enduring debate has con- tinguished between mentally well and emotion- cerned whether distress and well-being lie at op- ally distressed patients. This fostered consider- posite ends of a single continuum, or whether able dispute over the correct way to interpret the they form separate dimensions of feelings: a di- distinction and hence the measurement itself, as mensional versus a categorical view of affect. seen with Langner’s 22-item scale. This same set From the (initially surprising) finding of Brad- of questions has been said to indicate “mental burn that positive and negative affect formed health,” “emotional adjustment,” “psychologi- two separate dimensions, there was a shift to re- cal disturbance or disorder,” “psychiatric or psy- garding them as distinct but correlated aspects of chological symptoms,” and “mental illness,” emotion, rather than forming polar opposites and has even been described as a “psychiatric (see, for example, Zautra et al., 6, Figure 1), and case identification instrument.” This disagree- then back to viewing them as bipolar opposites ment indicates the disadvantage of the empirical (5). Negative affect refers to feelings of being approach to developing questionnaires, but also upset, angry, guilty, sad, afraid, or worried; be- 208 Measuring Health ing calm or relaxed indicates a lack of negative many distress and depression scales lies in the affect (7, p321). Positive affect refers to feelings southeast corner of the diagram (see Chapter 7); of energy and zest for life, active engagement, in- anxiety scales (see Chapter 6) may lie in the terest, pride, or delight; absence of positive af- northeast quadrant and the morale scales re- fect is indicated by fatigue or tiredness, although viewed in this chapter cover the northwest quad- this classification somewhat confounds positive rant. The southwest sector may be covered by and negative affect with level of activation. The some of the social health measures described in themes of positive and negative affect, and the Chapter 4. This is only one model, however, and relations between them, have assumed a central there are others. Russell and Carroll, for exam- position in the extensive discussions of the asso- ple, presented a similar model, but with valence ciations among anxiety, depression and mood (pleasantness-unpleasantness) on the horizontal disorders. Clark and Watson, for example, pro- axis, and activation on the vertical (5, Figure 1). posed a tripartite model comprising a general The activation dimension of affect is useful in distress or negative affect dimension that is highlighting the distinction among negative af- shared by depression and anxiety, plus physio- fects such as feeling upset, dissatisfied, sad, or logical hyperarousal that is particular to anxiety, depressed. These differ in terms of the level of and the absence of positive affect, which charac- activation: an upset person would be more likely terizes depression (7). Negative affect may re- to do something about his negative feelings than flect a vulnerability factor for the development of a depressed person. anxiety or depression; a tendency toward nega- Perhaps because of these conceptual com- tive affect may be heritable and form a stable plexities, considerable attention has been paid to trait (8, p180). The debate over the structure of testing the validity of psychological measures. affect remains active, and Mehrabian proposed a Often, this has resulted in scales of high quality, different model that contrasts three axes: Plea- but there are instances in which the critical inter- sure-Displeasure, Arousal-Nonarousal, and est in a scale has backfired. This most commonly Dominance-Submissiveness (9). Historical re- occurs where there is a lack of leadership by the views of the debate over whether positive and originator of a method and is accentuated where negative affect form bipolar opposites, or there is also no clear conceptual definition of the whether they form independent dimensions of precise purpose of the instrument. The result can affect were given by Russell and Carroll (5) and be a series of well-intended but uncoordinated by Watson and Clark (10, pp282–288). This attempts to improve the scale, and Macmillan’s theme is picked up in the reviews of Bradburn’s Health Opinion Survey (HOS) illustrates this. scale and of the Positive and Negative Affect Looking ahead in the chapter, Exhibit 5.3 com- Scales in this chapter. The more general themes pares seven different, yet widely used, versions of the relationships among affect, distress, and of this instrument and to complete the confu- neurotic disorders are discussed in more detail in sion, these versions all bear the same name. This the introductions to Chapter 6 on anxiety, and problem is most acute with the older scales; the to Chapter 7 on depression. more recent methods reviewed in this chapter These distinctions between positive and nega- have somewhat clearer explanations of their pur- tive affect suggest a conceptual map of mood pose. states, such as that illustrated in Exhibit 5.1. This is a general representation of a variety of such diagrams, which have been proposed since Scope of the Chapter the 1970s, and are often termed “circumplex models” of affect. Several are reviewed by Wat- The scales we review in this chapter fall into son et al., who also discuss limitations to the four categories; all are suited for use in popula- model and propose a refinement (11). Note that tion surveys. The first category comprises brief Bradburn’s scale covers the north-south and east- screening scales for psychological distress that west axes of the diagram, whereas the content of use a symptom checklist approach; we illustrate  Psychological Well-being 209 Exhibit 5.1 An Example of a Two-Dimensional Conceptual Model of Mood High positive affect active, elated, Strong Pleasantness excited engagement content, aroused, happy, astonished, satisfied concerned Low relaxed, distressed, High negative calm, fearful, negative affect placid hostile affect inactive, sad, still, lonely, quiet sluggish, withdrawn dull, Disengagement drowsy Unpleasantness Low positive affect Adapted from Tellegan A. Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self- report. In Tuma AH, Master JD (eds.) Anxiety and the anxiety disorders. Hillsdale, NJ: Erlbaum, 1985. this type of instrument by describing the Health final category of measures represents the current Opinion Survey and Langner’s 22-Item Screen- trend in this domain of measurement and in- ing Scale. These do not cover positive well-being, cludes scales that combine elements of survey which is a feature of the second category of measures and clinical instruments. Their items scales, illustrated by Bradburn’s Affect Balance are more clearly grouped into symptom areas: Scale and the Positive and Negative Affect Scales. anxiety, depression, and other categories. The In the second edition of this book, there was a General Well-Being Schedule, the Mental Health description of Frank Andrews’s single-item well- Inventory, and the Health Perceptions Question- being measures; this entry has been expanded naire cover both positive and negative feelings; and is now included in Chapter 10 on quality of this more clinical orientation is then pursued in life. The third category of scales covers life satis- Goldberg’s General Health Questionnaire, the fi- faction (which refers to feelings about the past) nal scale in the chapter. This method is explicitly and morale (which refers to optimism about the designed to detect acute psychiatrically diagnos- future). These are illustrated here by the Life Sat- able disorders in population studies. It has seen isfaction Index and the Morale Scale from the widespread use in many parts of the world and it Philadelphia Geriatric Center. Both are intended serves to introduce the clinical scales in Chapters for elderly populations and cover some of the 6 to 8. A comparative summary of the quality of negative feelings that may occur with aging. The the measurements in this chapter is shown in 210 Measuring Health Table 5.1. Readers searching for a broad- (4) Dohrenwend BP, Shrout PE, Egri G, et al. ranging instrument should also consider scales Nonspecific psychological distress and reviewed in other chapters. These include the other dimensions of psychopathology. Arch Depression, Anxiety, and Stress Scale or the Gen Psychiatry 1980;37:1229–1236. Hospital Anxiety and Depression Scale (both in (5) Russell JA, Carroll JM. On the bipolarity of positive and negative affect. Psychol Bull Chapter 6), while several of the scales in Chapter 1999;125:3–30. 10 include significant coverage of psychological (6) Zautra AJ, Guarnaccia CA, Reich JW. well-being. Factor structure of mental health measures Several scales we review share questions; for older adults. J Consult Clin Psychol most of the symptom checklists drew items from 1988;56:514–519. the U.S. Army’s neuropsychiatric screening in- (7) Clark LA, Watson D. Tripartite model of strument (12). This covered symptoms of ad- anxiety and depression: psychometric verse reactions to stressful situations, selected evidence and taxonomic implications. J empirically as identifying recruits who subse- Abnorm Psychol 1991;100:316–336. quently performed poorly in military combat. (8) Brown TA, Chorpita BF, Barlow DH. Although these questions were originally de- Structural relationships among dimensions of the DSM-IV anxiety and mood disorders signed for use with healthy young adult males, and dimensions of negative affect, positive they were adapted by Macmillan, Langner, and affect, and autonomic arousal. J Abnorm others for use in community surveys, forming Psychol 1998;107:179–192. the first generation of psychological well-being (9) Mehrabian A. Framework for a scales. Despite widespread criticism of these comprehensive description and measure- scales, they still see occasional use: Langner’s ment of emotional states. Genet Soc Gen questions, for example, are quite frequently used Psychol Monogr 1995;121:339–361. in studies of the impact of life events, stress, and (10) Watson D, Clark LA. Measurement and social support on emotional health. We have re- mismeasurement of mood: recurrent and viewed the Macmillan and Langner scales for emergent issues. J Pers Assess this reason and also to provide a historical intro- 1997;68:267–296. (11) Watson D, Wiese D, Vaidya J, et al. The duction to the field. More recent scales also two general activation systems of affect: share items in common: the RAND Mental structural findings, evolutionary Health Inventory incorporates many of Dupuy’s considerations, and psychobiological questions from the General Well-Being Schedule, evidence. J Pers Soc Psychol and these bear a strong family resemblance to 1999;76:820–838. items in Goldberg’s General Health Question- (12) Stouffer SA, Guttman L, Suchman EA, et naire. al. Measurement and prediction. Studies in social psychology in World War II. Volume IV. Princeton, NJ: Princeton University References Press, 1950. (1) Campbell A. Subjective measures of well- being. Am Psychol 1976;31:117–124. (2) U. S. Department of Health, Education, The Health Opinion Survey and Welfare. Net differences in interview (Allister M. Macmillan; first used in data on chronic conditions and 1951, published in 1957) information derived from medical records (DHEW Publication No. [HSM] 73-1331. Purpose [Vital and Health Statistics, Series 2, No. Macmillan developed the Health Opinion Sur- 57]). Washington, DC: U. S. Government vey (HOS) as a “psychological screening test for Printing Office, 1973. adults in rural communities” (1). It was designed (3) Jahoda M. Current concepts of positive to identify “psychoneurotic and related types of mental health. New York: Basic Books, disorder.” Subsequently, the HOS has been 1958. widely used in epidemiological studies, in esti-  Table 5.1 Comparison of the Quality of Psychological Indexes* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurement Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Health Opinion Survey ordinal 20 survey self (5 min) many ** ** *** ** (Macmillan, 1957) 22-Item Screening Score ordinal 22 survey self (5 min) many ** ** ** ** (Langner, 1962) Affect Balance Scale ordinal 10 survey self (4 min) many ** ** ** ** (Bradburn, 1965) Positive and Negative ordinal 20 survey self (4 min) many *** *** *** *** Affect Scale (Watson, Clark, and Tellegen, 1988) 211 Life Satisfaction Index A ordinal 20 survey self (5 min) many ** *** *** ** (Neugarten and Havighurst, 1961) Philadelphia Geriatric ordinal 22 clinical, self few ** ** ** ** Center Morale Scale survey (Lawton, 1972) General Well-Being Schedule ordinal 18 survey self (5 min) several *** *** *** *** (Dupuy, 1977) RAND Mental Health ordinal 38 survey self few * ** ** ** Inventory (Ware, 1979) Health Perceptions ordinal 33 survey self several *** ** ** ** Questionnaire (Ware, 1976) General Health Questionnaire ordinal 60 survey self many *** *** *** *** (Goldberg, 1972) * For an explanation of the categories used, see Chapter 1, pages 6–7. ** Andrews describes several, single-item rating scales. 212 Measuring Health mating need for psychiatric services, and in eval- who have used it and this serves to illustrate the uating their impact. problems of uncoordinated development of a measure. Indeed, the original 20 items were even Conceptual Basis altered during the course of the studies in which No relevant information is available. Macmillan Macmillan participated. Seven questions were used the title “Health Opinion Survey” to dis- deleted and replaced by questions on other topics; guise the purpose of the scale and to make re- nine others were reworded. Subsequent users have spondents less reticent about reporting not adhered strictly to either version; we present a emotional problems (1, 2). comparison of the main variants in Exhibit 5.3, which gives Macmillan’s original question topics Description and shows the variations made to his wording. The HOS comprises 20 items that were found to This means that extreme caution is needed in in- discriminate among 78 people clinically diag- terpreting results obtained with the scale, because nosed with a neurosis and 559 community re- it is seldom clear which version was used. Unfor- spondents in a pilot study in Nova Scotia, Canada tunately, neither the results of the validation stud- (1, 3). The 20 items are shown in Exhibit 5.2 (1). ies reported here nor the cutting points selected to More than is the case with other instruments, distinguish sick from well respondents are strictly the HOS has frequently been modified by those comparable among different studies. Exhibit 5.2 The Original Version of the Health Opinion Survey Note: The questions are not presented in the order as asked in the interview, but in decreasing rank order of their derived weights. 1. Do you have loss of appetite? 2. How often are you bothered by having an upset stomach? 3. Has any ill health affected the amount of work you do? 4. Have you ever felt that you were going to have a nervous breakdown? 5. Are you ever troubled by your hands sweating so that they feel damp and clammy? 6. Do you feel that you are bothered by all sorts (different kinds) of ailments in different parts of your body? 7. Do you ever have any trouble in getting to sleep and staying asleep? 8. Do your hands ever tremble enough to bother you? 9. Do you have any particular physical or health trouble? 10. Do you ever take weak turns? 11. Are you ever bothered by having nightmares? (Dreams that frighten or upset you very much?) 12. Do you smoke a lot? 13. Have you ever had spells of dizziness? 14. Have you ever been bothered by your heart beating hard? 15. Do you tend to lose weight when you have important things bothering you? 16. Are you ever bothered by nervousness? 17. Have you ever been bothered by shortness of breath when you were not exercising or working hard? 18. Do you tend to feel tired in the mornings? 19. For the most part, do you feel healthy enough to carry out the things that you would like to do? 20. Have you ever been troubled by “cold sweats”? (NOT a hot-sweat—you feel a chill, but you are sweating at the same time.) Reproduced from Macmillan, AM. The Health Opinion Survey: technique for estimating prevalence of psychoneurotic and related types of disorder in communities. Psychological Reports, 1957;3:325–329. (Monogr. Suppl. 7). © Southern Universities Press 1957.  Psychological Well-being 213 Exhibit 5.3 Main Variants of the Health Opinion Survey Note: A blank indicates that the question was omitted, “=” indicates identical wording, “V” indicates minor variation in wording, “R” indicates the question was reworded. Macmillan Leighton Denis Butler Gunderson Spiro Gurin Schwartz No. of items: 20 18 18 20 13 20 20 1. Loss of appetite R = R R R = = 2. Upset stomach = R = = = = R 3. Ill health affected work R = = R = = 4. Nervous breakdown V = = = 5. Hands sweating R R R R R R R 6. Bothered by ailments V = R R R R R 7. Trouble sleeping R R R R R R R 8. Hands tremble = R = = = = 9. Particular health trouble R R R R R 10. Weak turns R R R 11. Nightmares V R R R R R 12. Smokes R R R 13. Spells of dizziness R R R = = 14. Heart beating hard = = = = R = = 15. Lose weight R R R V V 16. Bothered by nervousness R R R 17. Shortness of breath R R R R = = 18. Tired in mornings = = = V R R R 19. Feel healthy enough = = V R = V 20. “Cold sweats” V V V R The HOS may be self- or interviewer- Reliability administered. A three-point answer scale (“of- Leighton et al. reported a test-retest correlation ten,” “sometimes,” “hardly ever or never”) was of 0.87 “after a few weeks or months” (3, used originally; other versions have employed p208). Tousignant et al. obtained a coefficient of four- or five-point scales. Macmillan proposed a 0.78 for 387 respondents after a ten-month de- scoring system by which the questions may be lay (4, p243). Schwab et al. showed a remark- weighted to discriminate maximally between able degree of stability between surveys in 1970 neurotic and mentally healthy respondents (1). and 1973 for 517 respondents (6). There was no Other users have reported high correlations be- difference in the mean scores at the two times of tween weighted and unweighted scores; the ad- administration; 53.4% of the variance in 1973 vantage of weighted scores seems slight (4). scores was attributable to the 1970 score. When Macmillan suggested a cutting point of 60.0 for classified into normal and abnormal scores, the weighted scoring system to distinguish be- 81.5% of respondents did not change their clas- tween neurotic and non neurotic patients; for the sification between the two surveys (6, p183). unweighted score, 29.5 was optimal (4, p244). Butler and Jones reported item-total correla- A computer scoring system derives depression tions ranging from 0.20 to 0.62; the coefficient and anxiety scores from the HOS (5). alpha for their 18-item version of the question- 214 Measuring Health naire was 0.84 (7, p557). Tousignant et al. re- tween the computer scoring system and a psychi- ported item-total correlations ranging from 0.21 atrist’s rating is reported by Murphy et al. (5). to 0.60 (4, p244). The sensitivity was 89% for depression and 96% for anxiety; the specificity was 79% for de- Validity pression and 48% for anxiety (5, Table 4). Re- There are many validation studies of the variants ceiver Operating Characteristic (ROC) curves of the HOS so that this review is not exhaustive. were reported for the HOS scores for 154 pa- In a study of in-patients, Macmillan reported a tients diagnosed by psychiatrists as having neu- sensitivity of 92% (at a cutting-point of 60.0 for rotic disorders and 787 people designated as the weighted scoring system), and specificity lev- psychiatrically well. Using a dichotomous scor- els ranging from 75% to 88% according to the ing for each item, the area under the ROC curves socioeconomic status of the presumedly healthy was 0.90; this rose to 0.91 using scores that used population (1, p332). Eleven HOS questions dis- the frequency responses. The area under the tinguished between people receiving outpatient curve rose to 0.97 for the computer scoring psychotherapy and others who were not. Four method (10, Figures 3–5). Schwartz et al. corre- questions were included in a discriminant func- lated the HOS with the New Haven Schizophre- tion that provided a sensitivity of 63% and a nia Index (r = 0.39) and with the Psychiatric specificity of 89% (8, p111). Whereas Macmil- Evaluation Form (r = 0.55) (11, p268). The lan’s patients were hospitalized, none in Spiro’s HOS was found to measure neurotic traits only; study was, and this may account for the lower it did not cover the range of psychotic symptoms sensitivity level. Leighton et al. reported correla- exhibited by schizophrenics. tions between psychiatric ratings and the HOS Tousignant et al. showed that replies to the ranging from 0.37 to 0.57 (3, pp208–210). The HOS were associated with use of medications, HOS discriminated adequately between the ex- psychological symptoms, reports of behavioral tremes of mentally well and psychiatrically sick disturbances, and judgments of disorder made but less adequately at intermediate stages of dis- by interviewers (4, Table 3). Denis et al. showed tress (3). highly significant variations in HOS scores by Tousignant et al. administered the HOS to 88 age, sex, occupation, marital status, education, psychiatric patients and to 88 matched commu- income, language, and geographical location (9). nity controls (4). All of the items discriminated at Butler and Jones obtained significant correla- p < 0.001; sensitivity was 80.7%. A cutting- tions with estimates of role conflict, family point of 29.5 identified as sick 90% of the neu- strain, and frequency of illness (7). rotic patients in the study, all the patients with Three factor analyses have identified factors alcoholism, 70% of the 13 patients with diag- representing physical and psychological prob- nosed schizophrenia, 71% of 45 patients with lems (7, 8, 12). There was, however, no clear psychosis, but only 30% of those with clinical correspondence between the factor placement of manic depression (9, p391). Gunderson et al. those questions common to the three studies. evaluated the ability of the HOS to classify over 4,000 Navy personnel into the categories “fit” Alternative Forms and “not fit for duty” (2). Thirteen items showed In addition to the variants noted, Murphy used a significant differences between the groups. questionnaire that included some of the HOS Macmillan compared the HOS scores with items in a study in Vietnam (13). A French ver- judgments of caseness by a psychiatrist; for 64 sion was used in the Stirling County studies (3) respondents, he reported a 14% disagreement and in Quebec by Tousignant et al. (4; 9). (1, p335). However, the disagreement may be much higher depending on how the substantial Commentary number of cases rated “uncertain” by the psy- The HOS was extensively used during the 1960s chiatrist are classified (4). The agreement be- and 1970s, including cross-cultural studies in  Psychological Well-being 215 Africa and North America (14; 15). There have the development of so many versions of the been several validation studies, and considerable HOS illustrates the confusion that can arise evidence suggests that it succeeds in its purpose when health indexes are modified piecemeal and as a screening test for neurotic disorders. without clear conceptual guidelines to define Nonetheless, few would now recommend that their content. Other more recent scales seem to the scale be used and, for the purposes of our re- have avoided this pitfall. view, there are several lessons to be learned from Ultimately, history may reject use of the the story of the HOS. HOS, not because it does not work, but for rea- The first illustrates our theme that measures sons that relate to the uncertainty of exactly why should have a clear conceptual basis. Although it works and of how it should be interpreted. the HOS can distinguish between neurotic pa- The problems with the HOS highlight the need tients and people without psychiatric diagnoses, for measurement methods to be founded on a se- it is not clear what a high score actually indi- cure conceptual basis that explains what they cates: mental disorder or normal reactions to measure and how they should be interpreted. We stress? Dohrenwend and Dohrenwend suggested cannot recommend the HOS for these reasons that the symptoms covered in measures such as and because other scales, such as those of Gold- the HOS may reflect normal processes of re- berg and Dupuy, offer better alternatives. sponding to temporary stressors, rather than neurotic disorders (16). Butler and Jones, indeed, References commented that “continued use of the HOS and related mental health indices appears to offer (1) Macmillan AM. The Health Opinion greater potential if they are approached more as Survey: technique for estimating prevalence stress indicators than as general indices of men- of psychoneurotic and related types of tal health” (7). Alternatively, the HOS has been disorder in communities. Psychol Rep 1957;3:325–339. said to measure a general demoralization, rather (2) Gunderson EKE, Arthur RJ, Wilkins WL. than diagnosable mental disorders, an interpre- A mental health survey instrument: the tation that Murphy, however, denies (17). Em- Health Opinion Survey. Milit Med pirically, the high test-retest reliability results 1968;133:306–311. obtained by Tousignant et al., by Schwab et al., (3) Leighton DC, Harding JS, Macklin DB, et and by Leighton et al. suggest that the HOS is al. The character of danger: psychiatric measuring a stable construct rather than a tran- symptoms in selected communities. Vol. III sient state. of the Stirling County Study of Psychiatric The second comment is that the empirical Disorder and Sociocultural Environment. way in which the HOS was developed com- New York: Basic Books, 1963. pounded the interpretation problem further. (4) Tousignant M, Denis G, Lachapelle R. Some considerations concerning the The tactic of using physical symptoms to dis- validity and use of the Health Opinion guise the intent of the scale complicates inter- Survey. J Health Soc Behav pretation and may not have worked anyway: 1974;15:241–252. Tousignant et al. reported correlations between the HOS and a “lie scale” that indicated a ten- dency to avoid admitting to socially undesirable *On a humorous note, the principle of trying to ob- attributes (4; 9).* The studies that showed sepa- scure the intent of a question was carried to its logical rate physical and psychological factors suggest conclusion in the No-Nonsense Personality Inventory that the HOS may reflect purely physical com- (a spoof on the MMPI). Concealed among items such as “Sometimes I find it hard to conceal the fact that I plaints as well as psychosomatic problems. am not angry,” or “Weeping brings tears to my eyes” is Wells and Strickland have studied this bias and question 69. Question 69 is entirely blank. In: Scherr have suggested an approach to remove it from GH, ed. The Best of the Journal of Irreproducible Re- the scale (18). Finally, the unfortunate history of sults. New York: Workman Publishing, 1983. 216 Measuring Health (5) Murphy JM, Neff RK, Sobol AM, et al. Computer diagnosis of depression and The Twenty-Two Item Screening anxiety: the Stirling County Study. Psychol Score of Psychiatric Symptoms Med 1985;15:99–112. (Thomas S. Langner, 1962) (6) Schwab JJ, Bell RA, Warheit GJ, et al. Social order and mental health: the Florida Purpose Health Study. New York: Brunner/Mazel, The 22-item scale is a screening method to pro- 1979. vide a “rough indication of where people lie on a (7) Butler MC, Jones AP. The Health Opinion continuum of impairment in life functioning due Survey reconsidered: dimensionality, to very common types of psychiatric symptoms” reliability, and validity. J Clin Psychol (1, p269). The scale is intended to identify men- 1979;35:554–559. tal illness, but not to specify its type or degree; (8) Spiro HR, Siassi I, Crocetti GM. What gets nor does it detect organic brain damage, mental surveyed in a psychiatric survey? A case study of the Macmillan index. J Nerv Ment retardation, or sociopathic traits (1). Dis 1972;154:105–114. (9) Denis G, Tousignant M, Laforest L. Conceptual Basis Prévalence de cas d’intéret psychiatrique No information is available. dans une région du Québec. Can J Public Health 1973;64:387–397. Description (10) Murphy JM, Berwick DM, Weinstein MC, The items in the screen were mainly taken from et al. Performance of screening and the U.S. Army’s Neuropsychiatric Screening Ad- diagnostic tests. Arch Gen Psychiatr 1987;44:550–555. junct and from the Minnesota Multiphasic Per- (11) Schwartz CC, Myers JK, Astrachan BM. sonality Inventory; the scale was developed for Comparing three measures of mental the Midtown Manhattan Study of the social status: a note on the validity of estimates of context of mental disorder (2). Of 120 items psychological disorder in the community. J originally tested, 22 were found to discriminate Health Soc Behav 1973;14:265–273. most adequately between people classified as (12) Gurin G, Veroff J, Feld S. Americans view well by a psychiatrist and a group of psychiatric their mental health: a nationwide interview patients. survey. New York: Basic Books, 1960. Closed-ended questions cover somatic symp- (13) Murphy JM. War stress and civilian Viet- toms of anxiety, depression, and other neurotic namese: a study of psychological effects. disturbances and also record subjective judg- Acta Psychiatr Scand 1977;56:92–108. (14) Beiser M, Benfari RC, Collomb H, et al. ments of emotional states (3). Fabrega and Measuring psychoneurotic behavior in McBee (4) and Muller (5) concluded that the cross-cultural surveys. J Nerv Ment Dis scale assesses mild neurotic and psychosomatic 1976;163:10–23. symptoms. In Langner’s original work the ques- (15) Jegede RO. Psychometric characteristics of tionnaire was administered by an interviewer; the Health Opinion Survey. Psychol Rep self-completed (6, 7) and telephone versions (8, 1977;40:1160–1162. 9) have also been used. The self-administered (16) Dohrenwend BP, Dohrenwend BS. The version requires few instructions and takes un- problem of validity in field studies of der five minutes to complete. psychological disorder. J Abnorm Psychol The items and response categories are shown 1965;70:52–69. in Exhibit 5.4. The score consists of the total (17) Murphy JM. Diagnosis, screening, and ‘demoralization’: epidemiologic number of responses that indicate sickness implications. Psychiatr Dev (termed “pathognomonic responses”) as desig- 1986;2:101–133. nated by asterisks in the exhibit. Differential (18) Wells JA, Strickland DE. Physiogenic bias weights were not used in the original, but Haese as invalidity in psychiatric symptom scales. and Meile proposed a scoring system that pro- J Health Soc Behav 1982;23:235–252. vided a different weight for each item based on Exhibit 5.4 Langner’s Twenty-Two Item Screening Score of Psychiatric Symptoms Note: An asterisk indicates the scored or pathognomonic responses. DK indicates Don’t Know. NA indicates No Answer. Item Response 1. I feel weak all over much of the time. *1. Yes 2. No 3. DK 4. NA 2. I have had periods of days, weeks or months when I couldn’t take care *1. Yes of things because I couldn’t “get going.” 2. No 3. DK 4. NA 3. In general, would you say that most of the time you are in high 1. High (very good) spirits, good spirits, low spirits, or very low spirits? 2. Good *3. Low *4. Very Low 5. DK 6. NA 4. Every so often I suddenly feel hot all over. *1. Yes 2. No 3. DK 4. NA 5. Have you ever been bothered by your heart beating hard? Would you *1. Often say: often, sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 6. Would you say your appetite is poor, fair, good or too good? *1. Poor 2. Fair 3. Good 4. Too Good 5. DK 6. NA 7. I have periods of such great restlessness that I cannot sit long in a chair *1. Yes (cannot sit still very long). 2. No 3. DK 4. NA 8. Are you the worrying type (a worrier)? *1. Yes 2. No 3. DK 4. NA 9. Have you ever been bothered by shortness of breath when you *1. Often were not exercising or working hard? Would you say: often, 2. Sometimes sometimes, or never? 3. Never 4. DK 5. NA 10. Are you ever bothered by nervousness (irritable, fidgety, tense)? *1. Often Would you say: often, sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 11. Have you ever had any fainting spells (lost consciousness)? Would you 1. Never say: never, a few times, or more than a few times? 2. A few times *3. More than a few times 4. DK 5. NA (continued) 217 Exhibit 5.4 (continued) Item Response 12. Do you ever have any trouble in getting to sleep or staying asleep? *1. Often Would you say: often, sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 13. I am bothered by acid (sour) stomach several times a week. *1. Yes 2. No 3. DK 4. NA 14. My memory seems to be all right (good). 1. Yes *2. No 3. DK 4. NA 15. Have you ever been bothered by “cold sweats”? Would you say: often, *1. Often sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 16. Do your hands ever tremble enough to bother you? Would you say: *1. Often often, sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 17. There seems to be a fullness (clogging) in my head or nose much *1. Yes of the time. 2. No 3. DK 4. NA 18. I have personal worries that get me down physically (make me *1. Yes physically ill). 2. No 3. DK 4. NA 19. Do you feel somewhat apart even among friends (apart, isolated, *1. Yes alone)? 2. No 3. DK 4. NA 20. Nothing ever turns out for me the way I want it to (turns out, happens, *1. Yes comes about, i.e., my wishes aren’t fulfilled). 2. No 3. DK 4. NA 21. Are you ever troubled with headaches or pains in the head? Would you *1. Often say: often, sometimes, or never? 2. Sometimes 3. Never 4. DK 5. NA 22. You sometimes can’t help wondering if anything is worthwhile anymore. *1. Yes 2. No 3. DK 4. NA Reproduced from Langer TS. A twenty-two item screening score of psychiatric symptoms indicating impairment. J Health Hum Behav 1962;3:271–273. 218  Psychological Well-being 219 the conditional probability of having a particu- the grouping made by psychiatrists in Crandell lar diagnosis with a certain symptom pattern and Dohrenwend’s study. Johnson and Meile (10). A comparison of this technique with the factor analyzed the scale using data from a simpler summative scoring system showed few large community study. Three factors were differences in the classification of patients and identified, reflecting physical symptoms, psy- healthy respondents. Langner recommended that chological stress, and psychophysiological re- four or more symptoms provided a “convenient sponses (9, Table 2). Johnson and Meile, as cutting point” for distinguishing well and sick well as De Marco, concluded that the physical groups (1). Twenty-eight percent of non patients component in the scale did not function inde- reported four or more symptoms, compared pendently of the psychological or psychophysi- with 50% of former patients, and 60% of outpa- ological components but rather contributed to tients. Other commentators have set scores of 7 the overall impression (9, 17). or 10 as cutting-points (7, 8, 10). Using information obtained from an inter- view that included the 22 items among 100 Reliability psychiatric symptoms, two psychiatrists inde- From a survey of over 11,000 respondents, pendently rated 1,660 respondents on their de- Johnson and Meile obtained alpha reliability gree of psychiatric impairment in the Midtown coefficients of 0.77 and an omega coefficient of Manhattan Study (1). Each of the 22 questions 0.80 (this estimates internal consistency where was then compared with this rating; correlations items fall on more than one factor) (9). They ranged from 0.41 to 0.79, suggesting that the found little variation in these results across age, psychiatrists had relied on these items in form- sex, and educational categories (9, Table 1). ing their overall judgment (1, p273). All 22 Cochrane reported relatively low item-total items in the scale distinguished between patients correlations ranging from 0.17 to 0.54 (11, newly admitted to a mental hospital and samples Table 3). He also reported an alpha of 0.83 and drawn from the community (7, Table 1). How- a one-week test-retest reliability of 0.88 (11, ever, sensitivity and specificity were relatively Table 4). low, at 67% and 63%, respectively, at a cutting- Wheaton studied two samples (N = 613 and point of four. A cutting-point of ten gave a sensi- 250) over four years and reported path coeffi- tivity of 20% and a specificity of 96% (7, p111). cients of 0.68 and 0.81 between the initial and These values suggest the instrument has limita- subsequent scores for ten items that Crandell tions as a screening tool. A score derived from and Dohrenwend (12) recommended be taken to the nine most discriminative questions per- form a psychological subscale (13, p399). formed almost as well as the full scale (7). The positive predictive value of the test in the Mid- Validity town study was also low: around 13% for the Several studies have reviewed the meaning of cutting-point of four, 21% for the cutting-point the items in the Langner scale. Crandell and of seven. Dohrenwend asked a sample of psychiatrists Manis et al. reported a correlation of 0.65 and internists to judge the content of each with a 45-item scale of behavioral symptoms of item. Ten items were judged to reflect psycho- mental health (7). Shader et al. reported a corre- logical symptoms, five were psychophysiologi- lation of 0.77 between the scale and Taylor’s cal, three were physical, and four could not be Manifest Anxiety Scale (N = 566), and of 0.72 classified. Responses to these four types of with a Minnesota Multiphasic Personality Inven- items reveal variations by age, sex, and socio- tory depression score, and of 0.72 with economic status (14; 15). A similar analysis Eysenck’s Neuroticism Scale (6, Table 8). Fab- was carried out by Seiler and Summers (16). rega and McBee obtained correlations of 0.50 Three studies of the structure of the scale used with psychiatrists’ ratings of depression and cluster analysis (14; 15; 17). They identified anxiety, and 0.30 with scores indicating neuroti- between three and five clusters that cut across cism (4). 220 Measuring Health Commentary some validity as a community survey technique Langner’s scale, although widely used, has also but cannot indicate the health of individuals (7). received considerable unfavorable critical atten- The scale contains no items covering positive tion. These criticisms, although now old, are re- mental health, so a low score will not distinguish viewed here briefly to illustrate an important between the absence of sickness and more posi- phase in the development of psychological in- tive states of well-being (3). dices. Wheaton’s review of the Langner scale pro- As with the HOS, there was active debate vided a balanced summary; he concluded that over precisely what the 22-item scale measures. the psychological items provide a good indicator The questions have been variously said to indi- of the likelihood that a person scoring highly has cate “psychiatric or psychological symptoms,” a psychiatric disorder. The psychophysiological “psychological disturbance or disorder,” “psy- items (numbers 1, 4, 15, 18, and 21 in our ex- chophysiological symptoms,” “emotional ad- hibit) are more problematic: their interpretation justment,” “mental health,” or “mental illness” varies from group to group, they are often (3). The method has even been termed a “psychi- closely associated with physical illness, and they atric case identification instrument” (10, p335). are not strongly associated with the chances of The debate is unlikely to be resolved, although receiving a psychiatric diagnosis (18, p50). Be- Seiler’s conclusion that the scale is partly an indi- cause the 22-item scale does not claim to cover cator of psychological stress and partly of physi- several important psychiatric problems, the low- ological malaise (16) is supported by several scoring group may include mentally healthy peo- commentators. The interpretation of a high ple and those suffering various types of mental score may also not be clear: does this suggest an illness not identified by the items. These prob- increasing probability of disorder or does it im- lems in interpreting the Langner scale have led to ply a more severe disorder? Wheaton answers its virtual replacement by newer scales that may this in terms of increasing scores indicating a be more reliably interpreted. higher probability of impairment (18, p28). Both the HOS and Langner scales may falsely References interpret purely physical symptoms as reflecting a psychological disorder (8; 9; 12; 14; 16). So- (1) Langner TS. A twenty-two item screening matic symptoms may also not provide a consis- score of psychiatric symptoms indicating tent indicator of psychological distress across impairment. J Health Hum Behav 1962;3:269–276. different social groups: respondents of a lower (2) Srole L, Langner TS, Michael ST, et al. social class may both suffer more physical illness Mental health in the metropolis: the and tend to express psychological disorders in Midtown Manhattan Study. New York: physical, rather than psychological, terms (3; 8; New York University Press, 1978. 12). However, Meile has dissented and argued (3) Seiler LH. The 22-item scale used in field on the basis of large studies that the physical studies of mental illness: a question of items did not provide evidence that diverged method, a question of substance, and a from that offered by the other questions in the question of theory. J Health Soc Behav scale (8; 9). Several studies have shown a higher 1973;14:252–264. symptom reporting among women than men (4) Fabrega H Jr, McBee G. Validity features (15; 16; 19); this may reflect a reporting bias be- of a mental health questionnaire. Soc Sci Med 1970;4:669–673. cause women are less inhibited about reporting (5) Muller DJ. Discussion of Langner’s their symptoms (12). Clancy and Gove, however, psychiatric impairment scale. Am J showed that males and females did not differ in Psychiatry 1971;128:601. their bias toward acquiescing to the items and (6) Shader RI, Ebert MH, Harmatz JS. that the difference in responses seemed to reflect Langner’s psychiatric impairment scale: a a true difference in symptoms experienced (19). short screening device. Am J Psychiatry Manis et al. commented that the scale holds 1971;128:596–601.  Psychological Well-being 221 (7) Manis JG, Brawer MJ, Hunt CL, et al. The Affect Balance Scale (ABS) Validating a mental health scale. Am Sociol (Norman M. Bradburn, 1965, Rev 1963;28:108–116. Revised 1969) (8) Meile RL. The 22-item index of psychophysiological disorder: Purpose psychological or organic symptoms? Soc The ten questions developed by Norman Brad- Sci Med 1972;6:125–135. burn were designed to indicate the positive and (9) Johnson DR, Meile RL. Does negative psychological reactions of people in the dimensionality bias in Langner’s 22-item general population to events in their daily lives. index affect the validity of social status Bradburn described his scale as an indicator of comparisons? An empirical investigation. J Health Soc Behav 1981;22:415–433. happiness or of general psychological well- (10) Haese PN, Meile RL. The relative being; these reflect an individual’s ability to cope effectiveness of two models for scoring the with the stresses of everyday living. The scale is mid-town psychological disorder index. not concerned with detecting psychological or Community Ment Health J 1967;3:335–342. psychiatric disorders, which Bradburn viewed as (11) Cochrane R. A comparative evaluation of reactions that persist after removal of the stress- the Symptom Rating Test and the Langner ful conditions or that are out of proportion to 22-item index for use in epidemiological the magnitude of the stress (1). surveys. Psychol Med 1980;10:115–124. (12) Crandell DL, Dohrenwend BP. Some Conceptual Basis relations among psychiatric symptoms, From their early studies, Bradburn and organic illness, and social class. Am J Caplovitz suggested that subjective feelings of Psychiatry 1967;123:1527–1537. (13) Wheaton B. The sociogenesis of well-being could be indicated by a person’s posi- psychological disorder: reexamining the tion on two independent dimensions, termed causal issues with longitudinal data. Am positive and negative affect (2). Overall well- Sociol Rev 1978;43:383–403. being is expressed as the balance between these (14) Roberts RE, Forthofer RN, Fabrega H Jr. two compensatory forces: an “individual will be Further evidence on dimensionality of the high in psychological well-being in the degree to index of psychophysiological stress. Soc Sci which he has an excess of positive over negative Med 1976;10:483–490. affect and will be low in well-being in the degree (15) Roberts RF, Forthofer RN, Fabrega H Jr. to which negative affect predominates over posi- The Langner items and acquiescence. Soc tive” (1, p9). Positive factors (e.g., being compli- Sci Med 1976;10:69–75. mented) can compensate for the negative feelings (16) Seiler LH, Summers GF. Toward an interpretation of items used in field studies to keep the overall sense of well-being at a con- of mental illness. Soc Sci Med stant level. The “affect balance score” represents 1974;8:459–467. this theme. (17) De Marco R. Relationships between Beyond simply compensating for each other, physical and psychological Bradburn and Caplovitz found that positive and symptomatology in the 22-item Langner’s negative feelings varied independently of one an- scale. Soc Sci Med 1984;19:59–65. other: they were not simply the opposite ends of (18) Wheaton B. Uses and abuses of the a single dimension of well-being. To illustrate Langner Index: a reexamination of findings the independence of the dimensions, Bradburn on psychological and psychophysiological and Caplovitz cited the example of a man who distress. In: Mechanic D, ed. Symptoms, has an argument with his wife, which may in- illness behavior and help-seeking. New York: Prodist, 1982:25–53. crease their negative feelings without changing (19) Clancy K, Gove W. Sex differences in their underlying positive feelings. Different cir- mental illness: an analysis of response bias cumstances were found to contribute to the pres- in self-reports. Am J Sociol ence of positive and negative affects. That, at 1974;80:205–216. least, was the argument until more detailed item 222 Measuring Health analyses suggested differing response tendencies tested but did not significantly alter the results to the two types of item, as discussed in the and so are not used (6). Positive and negative Commentary section. scores (PAS, NAS) are generally calculated, and the affect balance score is the positive score mi- Description nus the negative (zero represents balance). The Bradburn’s research formed part of the National resulting balance scale has occasionally been col- Opinion Research Center’s investigations into lapsed into one with fewer categories (4; 5). mental health at about the same time that Macmillan, Leighton, Gurin, and Langner were Reliability working on similar themes. The original scale de- Bradburn reported test-retest reliability results veloped by Bradburn consisted of 12 questions, over three days for 174 respondents. The result- seven measuring positive affect (PA) and five ing retest associations (Yule’s Q) exceeded 0.90 measuring negative affect (NA). Responses were for nine of the items; “excited or interested” had coded on a frequency scale (“once,” “some- a reliability of 0.86 (1). times,” “often”). Four questions were deleted Internal consistency results from several sub- and two others were added to give the five posi- samples ranged from 0.55 to 0.73 for the PAS tive and five negative questions that have been and from 0.61 to 0.73 for the NAS (7, p196). widely used. They are shown in Exhibit 5.5. Himmelfarb and Murrell reported alpha coeffi- The wording of the questions has remained cients of 0.65 (community sample) and 0.70 constant in most studies, but the question stem (clinical sample) (8, Table 1). Watson et al. criti- has changed. Bradburn specified a time referent cised the low internal consistency of the two (originally “the past week” and subsequently scales (alpha = 0.52 for NA and 0.54 for PA), us- “the past few weeks”); some users have changed ing this as partial justification for their develop- this to “the past few months” (3), whereas oth- ment of the Positive and Negative Affect Scales ers have asked, “How often do you feel each of (PANAS) (9, p1064). Warr obtained median these ways?” (4; 5). item-total correlations of 0.47 for the positive The scale is self-administered, and replies may scale and 0.48 for the negative scale (10, p114). use a dichotomous yes/no reply or a scale of Correlations among the items in the two scales three, four, or five points representing the fre- were modest, in the range of 0.24 to 0.26 (10). quency of experiencing the feelings; a three-point Warr also summarized the response patterns to scale (“often,” “sometimes,” “never”) has been the questions from five studies; although the ab- most commonly used. Differential weights were solute rates of affirmative replies varied between Exhibit 5.5 The Affect Balance Scale During the past few weeks, did you ever feel _________ (Yes/No) A. Particularly excited or interested in something? B. Did you ever feel so restless that you couldn’t sit long in a chair? C. Proud because someone complimented you on something you had done? D. Very lonely or remote from other people? E. Pleased about having accomplished something? F. Bored? G. On top of the world? H. Depressed or very unhappy? I. That things were going your way? J. Upset because someone criticized you? Reproduced from Bradburn NM. The structure of psychological well-being. Chicago: Aldine, 1969: 267. With permission.  Psychological Well-being 223 studies, the rank ordering of the questions by re- “proud,” and “restless.” A significant number of sponse rates was remarkably consistent. respondents found the idiom of these items inap- propriate, and it appeared that a negative answer Validity to them implied discontent with the question Bradburn provided extensive evidence of agree- rather than the absence of the affect in question ment between the questions and other self- (16, pS273). reported indexes of well-being. Discriminant validity was inferred from contrasts in response Alternative Forms patterns between employed and unemployed, be- A French translation was made in Canada (17); a tween rich and poor, and by occupational level German version has been published (18). Castil- (1). Positive affect was shown to be related to so- ian and Catalan Spanish versions are shown in cial participation, satisfaction with social life, an article by Stock et al. (19, pp230–1). Alpha and engaging in novel activities. Several of these reliability for the Catalan version was 0.72 for findings have been confirmed in subsequent PAS, and 0.64 for NAS. PA correlated 0.35 with studies. The independence of positive and nega- the Philadelphia Morale Scale and 0.43 with the tive affect scores and their lack of association Life Satisfaction Index; the NAS correlated with age have been widely replicated (7; 10–13). −0.62 and −0.61 (19, Table 1). Equivalent fig- Similarly, correlations have frequently been re- ures for the Castilian version (more relevant for ported with ratings of overall happiness (11; 12), Latin America) were lower: reliability 0.5 and employment status (10; 14), and social partici- 0.68; PAS correlations were 0.2 and 0.42 with pation (3; 7; 12; 14). Kushman and Lane re- the PGCMS and LSI; NAS correlations were ported significant associations with minority −0.59 and −0.4 (19, Table 1). Cantonese, Lao- status and the sex of the respondent (14). tian, and Cambodian translations have been Berkman used eight of the questions in the compared, and two-factor solutions were ob- Alameda County survey and reported a correla- tained in each; internal consistency scores tion of 0.48 with a 20-item Index of Neurotic ranged from an alpha of 0.62 to 0.72 for the Traits (5). Warr reported significant correlations PAS, and 0.62 to 0.70 for the NAS (20). between the affect scales and an anxiety rating and a scale of feelings about one’s present life among steelworkers who had been laid off (10). Reference Standards The NAS correlated 0.42 with a psychiatrist’s Reference standards for the Canadian popula- rating of “psychiatric caseness” (12). The PAS tion were produced from the 1978–1979 correlated 0.35 with a single-item scale measur- Canada Health Survey (N = 23,000) (17; 21). ing life satisfaction; the correlation with the NAS was −0.40. These values were lower than Commentary correlations obtained using other scales. The There are several important strengths in Brad- positive score correlated −0.30 with the 12-item burn’s scale. The inclusion of both positive and General Health Questionnaire, and −0.25 with negative questions was a major innovation, plac- Beck’s Depression Inventory. It correlated −0.17 ing the ABS among the most influential of all with Spielberger’s state anxiety scale (15, Table health measures. The questions have been used 1). with consistent phrasing in many large surveys, Cherlin and Reeder reported that positive and so that findings can be compared across studies. negative items formed two clearly distinct facto- The clear conceptual description of the purpose rial groups, although they questioned whether of the scale seems to have prevented some of the these measured affect (see Commentary section) misconceptions and disputes over interpretation (7). A detailed analysis of the items was gained that have characterized the HOS and Langner by interviewing respondents about their re- scales. sponses (16). Three items, in particular, were At the same time, the ABS has been closely identified as problematic: “on top of the world,” scrutinized and detailed criticisms have been 224 Measuring Health made of the scale, for example by Cherlin and and negative scores separately (7). Using a LIS- Reeder (7), Beiser et al. (3; 12), and Brenner REL analysis, Benin et al. showed that the best- (13). Because it is brief yet broad in scope, the fitting model involved allowing positive and ABS inevitably suffers some psychometric weak- negative items to correlate and giving different ness: its internal consistency, for example, is low weights to each item. Correlations between posi- compared with that of the HOS and Langner tive and negative factors varied by age-group, scales. The interpretation of the questions has from 0.33 to 0.49. Optimal item weights ranged been challenged; Cherlin and Reeder argued that from 1 to 4 (25, pp173–174). Mirowsky and the questions cover a broader theme than that Ross offered an insight into the way that re- implied by Bradburn’s term “affect”: the posi- sponse biases may confound the construct valid- tive dimension also covers activation or partici- ity of questionnaires, using the well-recognized pation (7). Reflecting this, Beiser altered the difference between men and women in reporting term “positive affect” to “pleasurable involve- emotional distress. They suggested that response ment” to reflect the item content more ade- to questions such as Bradburn’s are influenced by quately; he also discarded the item “on top of the person’s position on the positive–negative the world” (3; 12). continuum and by a response tendency that runs Behind criticisms of individual questions lies from emotional reticence or detachment to emo- the general issue of the adequacy of Bradburn’s tional dynamism or expressiveness (23, p593). two-component model of emotional well-being. The latter is influenced by culture and by gender; Reality appears to be more complex (7), and the men may be culturally reticent to report negative somewhat surprising finding of statistical inde- feelings, whereas both sexes appear to report pendence between PA and NA may merely be an similar levels of positive well-being. artifact of the question phrasing, a possibility The Bradburn scale was instrumental in stim- that Bradburn had recognized (1). For example, ulating research in the measurement of subjec- some positive and some negative questions refer tive well-being and happiness. It served to to specific events (e.g., “upset because someone demonstrate that these qualities can be mea- criticized you”) and quite reasonably these do sured, a claim that was disputed when the scale seem to be independent of one another (7). The was introduced. Nonetheless, the scale is 40 positive and negative questions covering more years old, and despite its historic significance, general feelings tend, however, to show a com- users should seriously consider applying an alter- paratively strong inverse relationship (13). This is native scale such as the General Well-Being not unique to the ABS: Goodchild and Duncan- Schedule, the PANAS, or the RAND Mental Jones noted that positively worded items the Health Inventory. General Health Questionnaire often tap transient feelings, whereas negative items reflect more sta- ble states (22), and similar findings have been re- References ported for Rotter’s Locus of Control Scale (23). (1) Bradburn NM. The structure of Kammann et al. (24) also contributed to the de- psychological well-being. Chicago: Aldine, bate over the independence of PA and NA; the 1969. theme is discussed further in the review of the (2) Bradburn NM, Caplovitz D. Reports on RAND Mental Health Inventory in this chapter. happiness: a pilot study of behavior related to mental health. Chicago: Aldine, 1965. Indeed, the issue has engrossed psychologists for (3) Beiser M, Feldman JJ, Egelhoff CJ. Assets much of the 40 years since Bradburn’s original and affects: a study of positive mental findings; some more recent discussions are sum- health. Arch Gen Psychiatry marized in the review of the PANAS. Because of 1972;27:545–549. these criticisms of the Bradburn scale, Cherlin (4) Berkman PL. Life stress and psychological and Reeder questioned the affect balance score as well-being: a replication of Langner’s the summary statistic because it may entail a loss analysis in the Midtown Manhattan Study. of information compared with reporting positive J Health Soc Behav 1971;12:35–45.  Psychological Well-being 225 (5) Berkman PL. Measurement of mental (20) Devins GM, Beiser M, Dion R, et al. health in a general population survey. Am J Cross-cultural measurements of Epidemiol 1971;94:105–111. psychological well-being: the psychometric (6) Bradburn NM, Miles C. Vague quantifiers. equivalence of Cantonese, Vietnamese, and Public Opin Q 1979;43:92–101. Laotian translations of the Affect Balance (7) Cherlin A, Reeder LG. The dimensions of Scale. Am J Public Health psychological well-being: a critical review. 1997;87:794–799. Sociol Methods Res 1975;4:189–214. (21) McDowell I, Praught E. On the measure- (8) Himmelfarb S, Murrell SA. Reliability and ment of happiness: an examination of the validity of five mental health scales in older Bradburn scale in the Canada Health persons. J Gerontol 1983;38:333–339. Survey. Am J Epidemiol (9) Watson D, Clark LA, Tellegen A. 1982;116:949–958. Development and validation of brief mea- (22) Goodchild ME, Duncan-Jones P. sures of positive and negative affect: the Chronicity and the General Health PANAS scales. J Pers Soc Psychol Questionnaire. Br J Psychiatry 1988;54:1063–1070. 1985;146:55–61. (10) Warr P. A study of psychological well- (23) Mirowski J, Ross CE. Fundamental being. Br J Psychol 1978;69:111–121. analysis in research on well-being: distress (11) Gaitz CM, Scott J. Age and the measure- and the sense of control. Gerontologist ment of mental health. J Health Soc Behav 1996;36:584–594. 1972;13:55–67. (24) Kammann R, Farry M, Herbison P. The (12) Beiser M. Components and correlates of analysis and measurement of happiness as a mental well-being. J Health Soc Behav sense of well-being. Soc Indicat Res 1974;15:320–327. 1984;15:91–115. (13) Brenner B. Quality of affect and self- (25) Benin MH, Stock WA, Okun MA. Positive evaluated happiness. Soc Indicat Res and negative affect: a maximum-likelihood 1975;2:315–331. approach. Soc Indicat Res (14) Kushman J, Lane S. A multivariate analysis 1988;20:165–175. of factors affecting perceived life satisfaction and psychological well-being among the elderly. Soc Sci Q 1980;61:264–277. The Positive and Negative Affect (15) Headey B, Kelley J, Wearing A. Dimensions Scale (PANAS) of mental health: life satisfaction, positive (D. Watson, L.A. Clark, and affect, anxiety and depression. Soc Indicat A. Tellegen, 1988) Res 1993;29:63–82. (16) Perkinson MA, Albert SM, Luborsky M, et Purpose al. Exploring the validity of the Affect The PANAS was developed as a brief measure of Balance Scale with a sample of family the two primary facets of mood, positive and caregivers. J Gerontol 1994;49:S264–S275. negative affect. It has been used mainly in re- (17) Health and Welfare Canada. The health of search studies of mood states (1). Canadians: report of the Canada Health Survey (Catalogue No. 82-538E). Ottawa, Conceptual Basis Ontario: Ministry of Supply and Services, Mood may be measured either in term of specific 1981. types of affect, as with the depression or anxiety (18) Noelle-Neumann E. Politik und Gluck. In: scales reviewed in Chapters 6 and 7, or it may be Baier H, ed. Freiheit und Schwang Beitrage measured in a nonspecific manner, the approach zu Ehren Helmut Schelskys. Opladen: West Deutscher Verlag, 1977:207–262. taken with the scales described in this chapter. (19) Stock WA, Okun MA, Gómez Benito J. Support for the nonspecific approach derives Subjective well-being measures: reliability from the finding of strong interrelationships be- and validity among Spanish elders. Int J tween specific scales, even when these purport to Aging Hum Devel 1994;38:221–235. measure different topics, suggesting that much 226 Measuring Health of the variance in such scales can be attributed to laxed” at the lower pole (6, p195). A model of underlying positive and negative affect. the various combinations of high and low PA Nonetheless, Watson and Clark argued that the and NA is found in Exhibit 5.1 in the introduc- specific and nonspecific approaches are not in- tion to this chapter. compatible but instead represent different levels Subsequent to the development of the in a hierarchical structure in which positive af- PANAS, Watson and Clark expanded their mod- fect (PA) and negative affect (NA) underlie more els of PA and NA. Within NA, they distin- specific representations, such as anxiety, depres- guished four relatively distinct facets: fear, sion, or fear (2). They further distinguish be- hostility, guilt, and sadness. Three facets of PA tween short-term emotional states (e.g., thrilled, were distinguished: joviality, self-assurance, and joyful) that are typically intense and fleeting, attentiveness (7; 8). versus longer-lasting, lower-intensity mood states (e.g., alert, active) (2, p276). In practice, Description however, this distinction did not appear to be The PANAS contains ten PA and ten NA items, valid empirically and the PANAS combines taken from a longer list of 60 descriptors used in mood and emotional terms under the general studies of mood. They were selected as being heading of “affect.” Hence, Watson’s conception specific in terms of loading on only one of the of affect defines PA in terms of the extent to PA or NA factors (7, p2). The items also covered which a person feels enthusiastic, alert, active, a range of themes (e.g., distressed, angry, guilty) and positively engaged, whereas NA reflects within each of the main categories (1, p1064). aversive moods such as distress, anger, guilt, Ratings use a five-point Likert scale, and fear, or nervousness (3, p602). These transient cover the extent to which the respondent has ex- feeling states are related to longer-term mood perienced each feeling in a particular time pe- traits that reflect a person’s characteristic ways riod. Positive and negative scores are formed by of reacting to situations, such as extroversion summing the responses to the 10 items in each versus neuroticism (1, p1063). Watson et al. scale, giving a range from 10 to 50. Watson et al. have outlined the links between neuroticism and tested six alternative time periods (as shown at extraversion and the dimensions of affect cov- the end of Exhibit 5.6) and they also evaluated a ered in the PANAS (4). In theory, underlying frequency response scale (“a little of the time,” personality characteristics are linked to the per- etc.) which gave comparable results to the scale son’s sensitivity to signals of reward and punish- shown in the Exhibit (1, Table 6). ment. In this conceptualization that defines PA and Reliability NA in terms of activation, they are only weakly Watson et al. reported alpha values for the PA related, such that a person can simultaneously scale ranging from 0.83 to 0.90 for six samples feel both alert and angry, or both active and dis- of undergraduate students (using different time- tressed (5). However, Watson and Tellegen argue frames for the PANAS responses). Alpha values that pleasantness and unpleasantness represent for NA ranged from 0.84 to 0.93 (7, Table 4). different aspects of affect and form the opposite Crocker reported an alpha of 0.88 for PA and ends of a single continuum and thus are nega- 0.79 for NA (9), Crawford and Henry reported tively correlated in empirical studies (3, 0.89 for PA and 0.85 for NA (10, p257), pp602–4). Hence, their overall model of affect whereas Huebner and Dew obtained values of includes both independent dimensions and bipo- 0.85 for PA and 0.84 for NA (11). In a sample lar scales. Empirical studies of the adjectives of 61 psychiatric patients, alpha was 0.85 for used to describe PA and NA suggest that each PA and 0.91 for NA, with an intercorrelation of has two poles. PA has adjectives such as “active” −0.27 between the scales (1, p1066). or “energetic” at its positive end, and “tired” or The intercorrelations between PA and NA “sluggish” at the negative end. NA would run scales are generally low and negative, ranging from “jittery” or “nervous” to “calm” and “re- from −0.12 to −0.23, with a slight tendency for  Psychological Well-being 227 Exhibit 5.6 The Positive and Negative Affect Scale This scale consists of a number of words that describe different feelings and emotions. Read each item and then mark the appropriate answer in the space next to that word. Indicate to what extent [you feel this way right now, that is, at the present moment.] Use the following scale to record your answers. 1 2 3 4 5 very slightly a little moderately quite a bit extremely or not at all ____________ interested ____________ irritable ____________ distressed ____________ alert ____________ excited ____________ ashamed ____________ upset ____________ inspired ____________ strong ____________ nervous ____________ guilty ____________ determined ____________ scared ____________ attentive ____________ hostile ____________ jittery ____________ enthusiastic ____________ active ____________ proud ____________ afraid (Note that alternative time frames may be substituted for the text in square braces in the Introduction. The alternatives include: “you have felt this way today”; “you have felt this way during the past few days”; “you have felt this way during the past week”; “you have felt this way during the past few weeks”; “you have felt this way during the past year” and “you generally feel this way, that is, how you feel on the average”) Reproduced form Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Personality Soc Psychol 1988;54:1070. the correlations to rise as the length of the re- NA. Watson’s original analyses showed a clear sponse time-frame increased from “today” to discrimination between the factors; the highest “the past year” (1, Table 2). Subsequent analyses loading of any item on the non dominant factor combined data from several studies, pooling re- was −0.14, whereas the lowest loading on the peated measures within subjects, giving correla- dominant factor was 0.52 (1, Table 5). A slightly tions of −0.23 for feelings right now (24,637 less clear two-factor solution was found in an observations on 533 subjects) and −0.32 for af- Australian study; alpha values were 0.89 for fect over the past few days (588 respondents and the PA factor and 0.87 for the NA (13, 26,833 observations) (3, p604). pp1210–1211). Killgore replicated these analy- An eight-week test-retest reliability study ses and found that the NA factor correlated 0.57 showed higher values as the time-instruction in- with the Beck Depression Inventory (14, Table creased from feelings at the moment to feelings 1). However, he then specified a three-factor so- in general: 0.47 to 0.68 for PA and 0.39 to 0.71 lution, which split the NA items into two com- for NA (1, Table 3). Eight-week retest reliability ponents, labeled upset and afraid, which had in another study ranged from 0.45 to 0.71 (12, been previously identified by Mehrabian (15). p102). These subcategories of the NA dimensions showed a clarified relationship with the Beck In- Validity ventory: 0.69 for the upset factor and 0.40 for Several studies have examined the factor struc- the afraid factor (14, Table 1). Finally, Crawford ture of the PANAS and in general two relatively and Henry applied confirmatory factor analysis independent factors emerge, representing PA and in a large general population sample and found 228 Measuring Health that two dimensions exist, but that they are earlier, and four other unrelated themes: shyness, moderately intercorrelated (r = −0.30) (10, fatigue, serenity, and surprise. The items are pp253–254). This two-factor model fit the data shown in the Manual (7, Table 2), which is avail- better than did Mehrabian’s three factor model. able from www.psychology.uiowa.edu/Faculty PANAS scores reflect emotionality and per- /Watson/PANAS-X.pdf. The PANAS-X includes sonality: the correlation between PA and extra- the 20 items from the PANAS, which are termed version was 0.51, whereas NA correlated 0.58 “general positive and negative affect scales.” with neuroticism scores (4, p48). Cronbach’s alpha for the PANS-X in a sample of Watson et al. reported correlations of 0.74 undergraduates was 0.82 (8, p1336). Alpha co- and 0.65 between NA and the Hopkins Symp- efficients for the individual scales ranged from tom Checklist in two studies; figures for the PA 0.76 to 0.93 (2, Table 7). All but two of the 11 were −0.19 and −0.29 (1, Table 7). Correla- subscales distinguished significantly between tions with Beck’s Depression Inventory (BDI) people with internal and external locus of con- were 0.56 and 0.58 for the NA, and −0.35 and trol (8, Table 1). Convergent correlations with −0.36 for PA. Correlations with Spielberger’s scales from the Profile of Mood States (POMS) State Anxiety scale were 0.51 for NA and were very high, ranging from 0.85 to 0.91, but −0.35 for PA (1, Table 7). Correlations with the the intercorrelations among the PANAS scales Hospital Anxiety and Depression Scale in- were lower than those among the POMS scales, cluded 0.44 for NA with depression and 0.65 suggesting that they are more discriminating (7, between NA and anxiety; correlations for PA Table 15). The PANAS-X Sadness scale corre- were −0.52 with depression and −0.31 with lated 0.59 with the BDI and 0.95 with the Center anxiety. Correlations with the Depression Anx- for Epidemiologic Studies Depression Scale; the iety Stress Scales (DASS) were 0.60 for NA Manual shows a range of other validity correla- with both DASS depression and anxiety scores, tions (7, p18). Norms are also shown in the and 0.67 with the stress score (10, Table 6). Manual (7, Tables 12 and 13). Watson et al. also reported significant associa- A children’s version has been produced, the tions between self-reports of stress and NA PANAS-C (17). The instructions were simpli- scores, but not PA (1, p1068). Similarly, NA fied and seven items were altered: “hostile” was was significantly correlated with somatic symp- changed to “angry”; “inspired” was altered to toms, including pain but PA was not (16, “lively”; “attentive” became “paying good at- p231). PA appears to show a consistent diurnal tention”; “jittery” was changed to “jumpy”; variation, rising through the morning, then re- “distressed” became “stressed out”; enthusias- maining steady for the day until declining again tic” became “eager”, and “determined” was re- during the evening. NA, meanwhile, did not ex- placed by “satisfied” (17, p403). Factor analysis hibit a significant diurnal pattern (1, showed a two-factor solution with a correlation pp1068–1069). of −0.13 between the factors (11). Alpha values Convergent correlations with scales of the have been reported from several studies: 0.84 Mental Health Inventory (MHI) included 0.70 for PA and 0.80 for NA (17, p403); 0.86 for between PANAS negative affect and MHI de- both PA and NA, and 0.89 at a second adminis- pression scales; 0.65 between negative affect and tration two weeks later (18, p339). Retest relia- anxiety, and 0.59 between the PANAS positive bility estimates include 0.72 and 0.79 for NA, score and the MHI general positive affect (12, and 0.67 and 0.82 for PA (18, p340). The NA Table 5). score correlated 0.68 with Taylor’s Manifest Anxiety Scale for children; the correlation with Alternative Forms the PA scale was −0.34. Equivalent correlations In 1994 Watson and Clark proposed an ex- with the Children’s Depression Inventory were panded version, the PANAS-Extended Form, or 0.68 and −0.50 (18, Table V). A different chil- PANAS-X (7). This has 60 items covering the dren’s version has also been described (see 17, three facets of PA and four of NA mentioned p403).  Psychological Well-being 229 A Spanish version of the PANAS has been de- (20). Watson and Clark have addressed this com- scribed (19). ment, acknowledging that the PANAS scales are not truly bipolar, but arguing that this does not Reference Standards seem to matter empirically. They tested a revised The data collected in the original development version of the PANAS that included items that studies of the PANAS provide norms for college were the reverse of selected items in each scale students (1, Table 1; 7, Table 3). Crawford and but found that this served mainly to reduce the Henry’s U.K. study provides median and mean distinction between PA and NA, thus damaging scores, standard deviations and percentiles for the clarity of the factorial structure. They like- the PA and NA scales (10, Tables 4 and 5). wise tested the advantage of including items on happiness and sadness, also finding that “includ- Commentary ing such items would have raised the correlation The PANAS offers a more recent and apparently between the scales and lessened their discrimi- superior alternative to Bradburn’s Affect Balance nant validity” (2, pp277, 280). Through their Scale. It avoids the criticisms of the heterogene- analyses of the PANAS, they also devoted con- ity in the Bradburn items; it does not include so- siderable attention to refining the underlying cir- matic items that may be confounded with cumplex conceptual structure of affect that was medical conditions and has become widely used, described in the introduction to this chapter (5). at least in psychological research. Growing psy- Several authors have also commented on the chometric evidence is accumulating on the scales; restricted range of the coverage of PA and NA in there has been extensive examination of the in- the PANAS (6; 9; 14; 15). Nemanick and Munz, ternal structure of the PANAS, its reliability ap- for example, commented that the adjectives in- pears appropriate and its correlations with other, cluded in the PANAS do not cover the complete more clinical scales are strong. As a measure- conceptual model that underlies it: there are no ment of health, limitations include the lack of in- adjectives describing the low poles of either PA formation on criterion validity and on sensitivity or NA, and this might create a floor effect, trun- to change. We do not yet know whether the cating each scale in the middle (6, p196). They PANAS will be useful as a screening or as an therefore made an empirical comparison of the evaluative instrument. PANAS and Thayer’s Activation Deactivation Among other criticisms of the PANAS is the Adjective Check List (ADACL), a scale that does observation that the negative item set does not in- cover the full range of PA and NA. Factor analy- clude items that are the opposites of the positive sis confirmed that the ADACL covered positive items: there is no item “bored” to represent the and negative ends of the PA and NA factors, opposite of “interested”; no “weak” as the op- with items that loaded both positively and nega- posite of “strong” (20, p11). Russell and Carroll tively on each, whereas the PANAS only in- noted that items representing high arousal pre- cluded adjectives that loaded positively on each dominate (active, alert, attentive for PA, and dis- factor (6, Table 1). Russell and Carroll gave a tressed, jittery and upset for NA). They somewhat abrupt summary of the critiques of accordingly argued that the PANAS scales actu- the PANAS: ally indicate a combination of PA or NA and high activation (20, p12). It follows that the More generally, Watson and Tellegen might PANAS cannot be used to test the hypothesis want to reevaluate their PANAS scales. The that PA and NA are polar opposites. Russell and response format used is ambiguous. These Carroll gave an extended discussion of the rela- scales do not measure the bipolar opposites tionship between PA and NA, showing that it of pleasant versus unpleasant affect that their varies widely between studies, influenced by a title might suggest. These scales do not mea- combination of the form of the response scale, sure strictly independent dimensions of posi- the time frame of the question, and characteris- tive activated and negative activated affect. tics of the actual items representing PA and NA (21, p615) 230 Measuring Health Studies using the PANAS have fueled the considerations, and psychobiological continuing discussions over the structure of evidence. J Pers Soc Psychol mood that began with Bradburn’s scale (2; 10; 1999;76:820–838. 20). In many ways, it is a question of the balance (6) Nemanick RC, Jr., Munz DC. Measuring to establish between lumping and splitting. Wat- the poles of negative and positive mood using the Positive Affect Negative Affect son, Clark, and Tellegen deliberately created a Schedule and Activation Deactivation simplified two-component model, but in some Adjective Check List. Psychol Rep situations this may prove inadequate. For exam- 1994;74:195–199. ple, in studies of stress related to sporting com- (7) Watson D, Clark LA. The PANAS-X: petition, it appears that summarizing negative Manual for the Positive and Negative mood into a single factor is less predictive of Affect Schedule—Expanded Form. 2nd ed. performance than subdividing it further (22). Iowa: University of Iowa, 1999. Hence, the 1994 revision to Watson and Clark’s (8) Henson HN, Chang EC. Locus of control conceptual model that split the two dimensions and the fundamental dimensions of moods. into eleven may prove more adequate for certain Psychol Rep 1998;82:1335–1338. predictive analyses. (9) Crocker PRE. A confirmatory factor analysis of the Positive Affect Negative The PANAS has proved itself a popular in- Affect Schedule (PANAS) with a youth strument that has been subjected to more de- sport sample. J Sport Exerc Psychol tailed conceptual and structural examination 1997;19:331–357. than most other measures of well-being. It de- (10) Crawford JR, Henry JD. The Positive and serves serious consideration as a measure of gen- Negative Affect Schedule (PANAS): eral affect, and it is to be hoped that more construct validity, measurement properties information will accumulate on the relationship and normative data in a large non-clinical of scores to other measures of psychopathology. sample. Br J Clin Psychol 2004;43:245–265. (11) Huebner ES, Dew T. Preliminary validation of the Positive and Negative Affect References Schedule with adolescents. J Psychoeduc (1) Watson D, Clark LA, Tellegen A. Assess 1995;13:286–293. Development and validation of brief mea- (12) Manne S, Schnoll R. Measuring cancer sures of positive and negative affect: the patients’ psychological distress and well- PANAS scales. J Pers Soc Psychol being: a factor analytic assessment of the 1988;54:1063–1070. Mental Health Inventory. Psychol Assess (2) Watson D, Clark LA. Measurement and 2001;13:99–109. mismeasurement of mood: recurrent and (13) Melvin GA, Molloy GN. Some emergent issues. J Pers Assess psychometric properties of the Positive and 1997;68:267–296. Negative Affect Schedule among Australian (3) Watson D, Tellegen A. Issues in the youth. Psychol Rep 2000;86:1209–1212. dimensional structure of affect—effects of (14) Killgore WDS. Evidence for a third factor descriptors, measurement error, and on the Positive and Negative Affect response formats: comment on Russell and Schedule in a college student sample. Carroll (1999). Psychol Bull Percept Mot Skills 2000;90:147–152. 1999;125:601–610. (15) Mehrabian A. Comparison of PAD and (4) Watson D, Gamez W, Simms LJ. Basic PANAS as models for describing emotions dimensions of temperament and their and for differentiating anxiety from relation to anxiety and depression: a depression. J Psychopathol Behav Assess symptom-based perspective. J Res Pers 1997;19:331–357. 2005;39:46–66. (16) Kvaal SA, Patodia S. Relations among (5) Watson D, Wiese D, Vaidya J, et al. The positive affect, negative affect, and somatic two general activation systems of affect: symptoms in a medically ill patient sample. structural findings, evolutionary Psychol Rep 2000;87:227–233.  Psychological Well-being 231 (17) Joiner TE, Jr., Catanzaro SJ, Laurent J. The term Life Satisfaction was finally adopted Tripartite structure of positive and negative on the grounds that, although it is not alto- affect, depression, and anxiety in child and gether adequate, it comes close to represent- adolescent psychiatric inpatients. J Abnorm ing the five components (1, p137). Psychol 1996;105:401–409. (18) Crook K, Beaver BR, Bell M. Anxiety and Neugarten et al. criticized earlier, single- depression in children: a preliminary dimensional approaches to measuring morale or examination of the utility of the PANAS-C. J Psychopathol Behav Assess well-being; from a review of previous measure- 1998;20:333–350. ment instruments they identified five compo- (19) Joiner TE, Jr., Sandin B, Chorot P, et al. nents of life satisfaction which the LSI was Development and factor analytic validation intended to measure. These include zest (as op- of the SPANAS among women in Spain: posed to apathy), resolution and fortitude, con- (more) cross-cultural convergence in the gruence between desired and achieved goals, structure of mood. J Pers Assess positive self-concept, and mood tone (1). Posi- 1997;68:600–615. tive well-being is indicated by the someone’s tak- (20) Russell JA, Carroll JM. On the bipolarity ing pleasure to his daily activities, finding life of positive and negative affect. Psychol Bull meaningful, reporting a feeling of success in 1999;125:3–30. achieving major goals, having a positive self- (21) Russell JA, Carroll JM. The phoenix of bipolarity: reply to Watson and Tellegen image, and maintaining optimism (1). (1999). Psychol Bull 1999;125:611–617. (22) Lane AM, Lane HJ. Predictive effectiveness Description of mood measures. Percept Mot Skills Several versions of the LSI exist. The original, 2002;94:785–791. Version A (LSIA), comprises 20 items, of which 12 are positive and eight are negative. An agree/disagree response format is used. A second The Life Satisfaction Index and little used version, the LSIB, contains 12 (Bernice L. Neugarten and Robert J. questions using three-point answer scales (1). A Havighurst, 1961) third version, the LSIZ, was proposed by Wood et al. as a refinement of the LSIA and contains Purpose 13 of the 20 items (2). Finally, Adams recom- The Life Satisfaction Index (LSI) covers general mended deleting items 11 and 14 from the LSIA, feelings of well-being among older people to forming an 18-item version, which he confus- identify “successful” aging (1). ingly called the LSIA (3). This was later used by Harris in two large national surveys, although he Conceptual Basis renamed it the LSIZ (4). Exhibit 5.7 shows the As used by Neugarten et al. the concept of life original 20-item LSIA. satisfaction is closely related to morale, adjust- The LSIA was developed empirically by ad- ment, and psychological well-being. Discussing ministering a draft version of the questionnaire these terms, they noted: to two groups of people considered to differ in their level of life satisfaction. This difference had The term “adjustment” is unsuitable because been established on the basis of the Life Satisfac- it carries the implication that conformity is tion Rating Scale, also developed by Neugarten the most desirable pattern of behavior. “Psy- et al. The Rating Scale is scored by a profes- chological well-being” is, if nothing else, an sional and also reflects the five components of awkward phrase. “Morale,” in many ways, life satisfaction hypothesized by the authors (1). captures best the qualities here being de- Questions in the draft scale that differentiated scribed, but there was the practical problem successfully between high and low scorers on the that there are already in use in gerontological Rating Scale were selected for the LSIA, which is research two different scales entitled Morale. self-administered. 232 Measuring Health Exhibit 5.7 The Life Satisfaction Index A Here are some statements about life in general that people feel differently about. Would you read each statement in the list, and if you agree with it, put a check mark in the space under “AGREE.” If you do not agree with a statement, put a check mark in the space under “DISAGREE.” If you are not sure one way or the other, put a check mark in the space under “?”. Please be sure to answer every question on the list. Agree Disagree ? 1. As I grow older, things seem better than I thought they would 2 0 1 be. 2. I have gotten more of the breaks in life than most of the people I know. 2 0 1 3. This is the dreariest time of my life. 0 2 1 4. I am just as happy as when I was younger. 2 0 1 5. My life could be happier than it is now. 0 2 1 6. These are the best years of my life. 2 0 1 7. Most of the things I do are boring or monotonous. 0 2 1 8. I expect some interesting and pleasant things to happen to me in the future. 2 0 1 9. The things I do are as interesting to me as they ever were. 2 0 1 10. I feel old and somewhat tired. 0 2 1 11. I feel my age, but it does not bother me. 2 0 1 12. As I look back on my life, I am fairly well satisfied. 2 0 1 13. I would not change my past life even if I could. 2 0 1 14. Compared to other people my age, I’ve made a lot of foolish decisions in my life. 0 2 1 15. Compared to other people my age, I make a good appearance. 2 0 1 16. I have made plans for things I’ll be doing a month or a year from now. 2 0 1 17. When I think back over my life, I didn’t get most of the important things I wanted. 0 2 1 18. Compared to other people, I get down in the dumps too often. 0 2 1 19. I’ve gotten pretty much what I expected out of life. 2 0 1 20. In spite of what people say, the lot of the average man is getting worse, not better. 0 2 1 Reproduced from Neugarten BL, Havighurst RJ, Tobin SS. The measurement of life satisfaction. J Gerontol 1961;16:141. With permission. Scoring system based on Wood V, Wylie ML, Sheafor B. An analysis of a short self-report measure of life satisfcation correlation with rater judgments. J Gernotol 1969;24:467. There are two ways to score the LSI. In the response as 0, giving a range of 0 to 40 (3). original method, a two-point agree/disagree This approach was used by Harris in his na- score rated items 0 for a response indicating tional surveys and is shown in Exhibit 5.7. In- dissatisfaction and 1 for satisfaction (range ternal consistency has been reported to be 0 to 20). Problems with coding “undecided” marginally higher with the three-point ap- responses then prompted the use of a three- proach (5, p377), but Ray showed little advan- point scale, rating a satisfied response as 2, an tage of this approach over the two-point uncertain response as 1, and a dissatisfied method (6).  Psychological Well-being 233 Reliability found to vary across age groups, whereas the Adams calculated item-total correlations for his other factors remained stable (17). Extensive ex- 18-item LSIA but only reported results for a few ploratory and confirmatory factor analyses were (2). The alpha internal consistency of Wood’s carried out by Hoyt and Creech (N = 2,651), 13-item LSIZ was 0.79 (3, p467). Stock and again using 11 items (18). The best model they Okun obtained an alpha of 0.80 with 325 older identified was a three-factor solution that resem- persons (7, p626). In a study of 1,288 older bled Adams’s results: congruence, mood tone, men, Dobson et al. used the 13-item LSIZ and and optimism (18). Using multiple regression reported alphas of 0.70 for two-point responses, analyses, Knapp found that different demo- and 0.76 for five-point answer scales (8, p571). graphic and health variables predicted each fac- Internal consistency appears to improve for a tor score, thus confirming the multidimensional subset of ten items identified by Adams as load- nature of the scale (19). Although the LSI does ing significantly on a factor analysis: Edwards cover several dimensions, the empirical findings and Klemmack obtained an alpha of 0.90 (9, did not closely replicate the original conceptual p498). Himmelfarb et al. reported an alpha of formulation given by Neugarten and Havighurst 0.74 for 264 community subjects, and 0.84 for (18, p115). A more recent report by Helmes et 101 patients (10, Table 1). By contrast, Abra- al. used confirmatory factor analysis to assess ham found lower internal consistency, citing the fit of results obtained in a range of previous Kuder-Richardson-20 coefficients ranging from studies (5). They found that oblique factor rota- 0.11 to 0.60 in a study of depressed elderly pa- tions fit better than orthogonal, and that a five- tients (11, Table 1). In a study in Spain, alpha factor solution for all 20 items was reasonable, values were 0.75 for a Catalan Spanish version, and that the results of Liang and of Hoyt and and 0.74 for a Castilian version (12, Table 1). Creech also provided an acceptable fit for the Test-retest reliability for the LSIZ ranged 11-item version of the LSI (5, Table III). from 0.80 to 0.90 in three samples of patients Convergent validity has been studied exten- with chronic disease (13, p352). sively. Neugarten et al. reported a correlation of 0.55 between the LSIA and the fuller Life Satis- Validity faction Rating Scale for 92 respondents aged be- Because the LSI was based on a five-component tween 50 and 90 years and of 0.39 with a conception of life satisfaction, several studies psychologist’s clinical assessment of 51 respon- have examined its factor structure empirically. dents (1, p142). A separate study again com- Interpretation is complicated because different pared the LSIA and the Rating Scale, reporting a studies used different subsets of LSI items and virtually identical correlation of 0.56 (3, p467). the factor structure seems to vary from sample Lohmann compared the LSIA with other indica- to sample (14; 15). Adams identified three inter- tors of life satisfaction administered to 259 el- pretable factors from a sample of 508 commu- derly people (20). The scales included the LSIB, nity respondents, but used only 18 items. The the LSIZ, the Philadelphia Geriatric Center results showed an important general factor Morale Scale, the Kutner Morale Scale, and a (34% of the variance) reflecting mood tone. The global life satisfaction rating: “How satisfied are second factor corresponded to the original con- you with your life?” The results of the analyses cept of zest for life; the third reflected congru- are shown in Table 5.2. The Kutner Morale Scale ence between desired and achieved goals. The shares four items with the LSIB, contributing to interpretation of a fourth factor was unclear, the high correlation between them. The LSIZ and two items did not fall on any factor (2). correlated 0.74 with the Philadelphia Geriatric Liang analyzed only 11 items and identified Center Morale Scale in a nursing-home sample three first-order factors (mood tone, zest, and from a long-term care facility (21, pS163). Stock congruence between hopes and reality) and one and Okun reported correlations of 0.33 with the second order factor representing general subjec- positive affect score on the Bradburn scale and tive well-being (16). The congruence factor was of −0.39 with the negative affect score (7, Table 234 Measuring Health Table 5.2 Correlations of the LSIA with Other Scales LSIA LSIB LSIZ Kutner PGC LSIB 0.63 LSIZ 0.94 0.64 Kutner 0.65 0.88 0.67 PGC 0.76 0.74 0.79 0.74 Global rating 0.41 0.40 0.40 0.40 0.47 Adapted from Lohmann N. Correlations of life satisfaction, morale and adjustment measures. J Gerontol 1977;32:74, Table 1. 1). The LSIZ correlated −0.31 with the psycho- Reference Standards logical dimension scores on the Sickness Impact Neugarten et al. obtained a mean LSIA score of Profile (21, pS163). In a Spanish study, the corre- 12.4 (SD, 4.4) using the two-point responses (1, lations with the Philadelphia Morale Scale were p142). Very similar results have been obtained 0.62 for the Catalan version and 0.58 for the by other users: 11.6 (3, p466), 12.5 (2, p470), Castilian. Correlations with Bradburn’s positive and 12.1 (SD, 3.9) (22, p76). For the 18-item affect score were 0.43 and 0.42 for the two lan- scale, using three-point responses, Harris re- guages, whereas correlations with the negative af- ported mean scores of 26.7 for those aged 18 to fect score were −0.61 and −0.40 (12, Table 1). 64 years (N = 1,457) and 24.4 for those older Neugarten and Havighurst, and also Lieber- than 65 (N = 2,797) (4, p159). man showed that replies to the LSIA did not cor- relate with sex, socioeconomic status, age, or Commentary geographical location, concluding that the scale The LSI has been extensively used and has sev- does not merely indicate objective environmental eral virtues, including reliability, strong correla- circumstances (1; 22). Other studies have not tions with other scales, and the availability of replicated this finding, however: Cutler obtained some reference standards. The consistency of the significant correlations with socioeconomic sta- validity findings and, in particular, of the factor tus (23). Harris found positive correlations with structure, is striking: many other scales reviewed income, employment, and education (4). Using in this book show much less consistency between multiple regression analysis, Edwards and Klem- samples. mack showed that socioeconomic status, per- Despite these strengths, there have been a ceived health status, and social participation number of critical reviews of the LSI from together explained 24% of the variance in LSIA which several points emerge. The question of scores (9). By contrast, Markides and Martin precisely what the scale does measure is open to found that a self-rating health score, income, and debate. It is agreed that the scale does not fully education explained 50% of the variance on the reflect the subtleties implied in the original five- LSIA scores for men and 40% for women (24). component conceptual model of life satisfac- tion. Helmes et al. noted that “the scale should Alternative Forms not be conceived of as strictly unidimensional, An eight item Life Satisfaction Index—Well- even though the scoring procedure typically em- Being (LSIW) was derived from the LSIA and has ployed is unidimensional.” The items also did been used in Great Britain. The eight items load not reflect the five-dimensional model, and the on two factors with alpha coefficients of 0.65 original item sampling did not appear to address and 0.41 (25, p649). these five dimensions adequately (5, pp384–5). Translations include Castilian Spanish (12, Hoyt and Creech were critical: their results pp228–229), Catalan (12, pp228–229), and “raise serious questions about the structure and Greek (26). interpretation of the measures in the LSIA”  Psychological Well-being 235 (18). Indeed, measurement techniques (such as scales, most of which have been less thoroughly those of the LSI) have not managed to reflect the evaluated. Its psychometric properties rival those conceptual distinctions that have been drawn of the best among comparable indexes; the task between concepts such as quality of life, is to identify clearly what, in conceptual terms, anomie, happiness, and morale (8; 27). Klem- the scale measures. mack et al. noted: References Although the distinction between life satis- faction and social isolation may have some (1) Neugarten BL, Havighurst RJ, Tobin SS. justification on theoretic grounds, there is no The measurement of life satisfaction. J Gerontol 1961;16:134–143. reason to anticipate, on the basis of our (2) Wood V, Wylie ML, Sheafor B. An analysis data, that the subtleties between the two of a short self-report measure of life concepts are reflected on an empirical level satisfaction: correlation with rater (27, p270). judgments. J Gerontol 1969;24:465–469. (3) Adams DL. Analysis of a Life Satisfaction The failure of the measurement methods to re- Index. J Gerontol 1969;24:470–474. flect the distinctions among these concepts is (4) Harris L. The myth and reality of aging in shown by Lohmann’s findings of strong associa- America. Washington, DC: National tions between the LSI and morale scales, al- Council on the Aging, 1975. though both were only weakly associated with a (5) Helmes E, Goffin RD, Chrisjohn RD. global life satisfaction rating scale of the type Confirmatory factor analysis of the Life Satisfaction Index. Soc Indicat Res used by Andrews. Some commentators have at- 1998;45:371–390. tempted to modify Neugarten and Havighurst’s (6) Ray RO. The Life Satisfaction Index-Form conceptual formulation to bring it more into line A as applied to older adults: technical note with empirical evidence. Lieberman noted: on scoring patterns. J Am Geriatr Soc 1979;27:418–420. Life satisfaction, rather than being merely a (7) Stock WA, Okun MA. The construct reflection of a person’s current level of goal validity of life satisfaction among the elder- achievement, is more like a set or orientation ly. J Gerontol 1982;37:625–627. to one’s environment which is acquired fairly (8) Dobson C, Powers EA, Keith PM, et al. early and remains moderately stable through- Anomie, self-esteem, and life satisfaction: out life (22, p75). interrelationships among three scales of well-being. J Gerontol 1979;34:569–572. (9) Edwards JN, Klemmack DL. Correlates of It is clear that the scale is multidimensional, so life satisfaction: a re-examination. J the single, overall score would appear inade- Gerontol 1973;28:479–502. quate. (10) Himmelfarb S, Murrell SA. Reliability and Finally, there have been criticisms of the validity of five mental health scales in older wording of some of the items. Connidis sug- persons. J Gerontol 1983;38:333–339. gested that values implicit in the wording may (11) Abraham IL. Longitudinal reliability of the lead some respondents to disagree with the item Life Satisfaction Index (short form) with even though they were not personally dissatisfied nursing home residents: a cautionary note. (28). Helmes et al. noted that a negative re- Percept Mot Skills 1992;75:665–666. sponse to items such as “I am just as happy as (12) Stock WA, Okun MA, Gómez Benito J. Subjective well-being measures: reliability when I was younger” may occur not because the and validity among Spanish elders. Int J person is unhappy, but because they are now Aging Hum Devel 1994;38:221–235. happier than they were (5, pp384–385). (13) Burckhardt CS, Woods SL, Schultz AA, et Despite the conceptual uncertainties over the al. Quality of life of adults with chronic LSI and despite its age, we do not recommend illness: a psychometric study. Res Nurs discarding it in favor of other life satisfaction Health 1989;12:347–354. 236 Measuring Health (14) Wilson GA, Elias JW, Brownlee LJ, Jr. Life Satisfaction Index A and Affect Factor invariance and the Life Satisfaction Balance Scales: a serendipitous analysis. Index. J Gerontol 1985;40:344–346. Soc Indicat Res 1984;15:117–129. (15) Cutler NE. Age variations in the dimensionality of life satisfaction. J Gerontol 1979;34:573–578. The Philadelphia Geriatric Center (16) Liang J. Dimensions of the Life Satisfaction Morale Scale Index A: a structural formulation. J (M. Powell Lawton, 1972) Gerontol 1984;39:613–622. (17) Liang J, Tran TV, Markides KS. Purpose Differences in the structure of Life The Philadelphia Geriatric Center Morale Scale Satisfaction Index in three generations of (PGCMS) was designed to measure three dimen- Mexican Americans. J Gerontol sions of emotional adjustment in people aged be- 1988;43:S1–S8. tween 70 and 90 years. It is applicable both to (18) Hoyt DR, Creech JC. The Life Satisfaction community residents and to people in institu- Index: a methodological and theoretical tions. critique. J Gerontol 1983;38:111–116. (19) Knapp MRJ. Predicting the dimensions of life satisfaction. J Gerontol Conceptual Basis 1976;31:595–604. Lawton viewed morale as “a generalized feeling (20) Lohmann N. Correlations of life of well-being with diverse specific indicators” satisfaction, morale and adjustment mea- (1). The indicators of morale include: sures. J Gerontol 1977;32:73–75. (21) Rothman ML, Hedrick S, Inui T. The freedom from distressing symptoms, satisfac- Sickness Impact Profile as a measure of the tion with self, feeling of syntony between self health status of noncognitively impaired and environment, and ability to strive appro- nursing home residents. Med Care priately while still accepting the inevitable 1989;27(suppl):S157–S167. (22) Lieberman LR. Life satisfaction in the (1). young and the old. Psychol Rep 1970;27:75–79. The interrelationship among these compo- (23) Cutler SJ. Voluntary association nents may or may not be close: a pessimistic ide- participation and life satisfaction: a ology “may or may not accompany an ability to cautionary research note. J Gerontol accept the status quo.” Morale is viewed as a 1973;28:96–100. feeling that is not necessarily related to behavior; (24) Markides L, Martin HW. A causal model the relationship resembles that between attitudes of life satisfaction among the elderly. J and behavior (1). Gerontol 1979;34:86–93. (25) James O, Davies ADM, Ananthakopan S. The Life Satisfaction Index-Well-being: its The person of high morale has a feeling of internal reliability and factorial having attained something in his life, of being composition. Br J Psychiatry useful now, and thinks of himself as an ade- 1986;149:647–650. quate person. . . . High morale also means a (26) Malikois-Loizos M, Anderson LR. feeling that there is a place in the environ- Reliability of a Greek translation of the ment for oneself . . . a certain acceptance of Life Satisfaction Index. Psychol Rep what cannot be changed. (1, p148) 1994;74:1319–1322. (27) Klemmack DL, Carlson JR, Edwards JN. Measures of well-being: an empirical and Description critical assessment. J Health Soc Behav The morale scale is one of several geriatric as- 1974;15:267–270. sessment scales developed at the Philadelphia (28) Connidis I. The construct validity of the Geriatric Center; Lawton and Brody’s Physical  Psychological Well-being 237 Self-Maintenance Scale (PSMS) is described in 0.81. Internal consistency reliability for the three Chapter 3, whereas the Multilevel Assessment subscales ranged from 0.57 to 0.61 (5, p80). Instrument is described in Chapter 10. Others Differences were found between black and white include the Mental Status Questionnaire re- respondents in the reliability of only two items: viewed in Chapter 7, the Instrumental Role “I am afraid of a lot of things” and “Life is hard Maintenance Scale, and the Minimal Social Be- for me” (6, p427). In a study in Spain, alpha val- havior Scale (2). ues were 0.65 for a Catalan version, and 0.60 A preliminary version of the morale scale for a Castilian Spanish version (7, Table 1). containing 41 items was tested on 300 healthy people with an average age of 78 years. Twenty- Validity two items that were significantly associated For 199 elderly respondents, the 22-item scale with an independent ranking of the respondents correlated 0.47 with an independent ranking of according to morale, and also loaded on a factor their morale. Because of the low reliability of the analysis, were retained for the main version of independent ranking, this result probably repre- the scale, shown in Exhibit 5.8). Lawton subse- sents an underestimate of the validity of the quently recommended a further abbreviation of scale. A correlation of 0.57 was obtained with the scale to 17 items, indicated by asterisks in the Life Satisfaction Index (LSI) (1, p151). The the exhibit and termed the Revised PGCMS; Morale Scale correlated 0.74 with the LSIZ in a this version is now normally used (3). The mixed community and hospital sample (8, Table method can be self- or interviewer-administered 1); a correlation of 0.74 was also obtained in a and most items have a dichotomous response. sample in a long-term care facility; the correla- The self-administered version uses the first per- tion with the psychological dimension of the son phrasing shown in the Exhibit (except for Sickness Impact Profile was −0.40 (9, pS163). In item 6); the interview version uses the second a Spanish study, the correlations with the LSI person (“Do you . . . ?”). Each high morale re- were 0.62 for the Catalan version and 0.58 for sponse receives a score of 1, giving a range from the Castilian. Correlations with Bradburn’s posi- 0 to 17 for the abbreviated version. Liang and tive affect score were 0.35 and 0.2 for the two Bollen suggested that scores be calculated to languages, whereas correlations with the nega- form three subscales (agitation, dissatisfaction, tive affect score were −0.62 and −0.59 (7, Table and attitudes toward one’s own aging) and this 1). has been widely followed; an overall score re- Much attention has been paid to the factor flecting global life satisfaction can also be structure of the morale scale. From the replies of formed (4). Although there are no formal 300 subjects in Lawton’s original study, six fac- cutting-points for interpreting scores, the man- tors were extracted: surgency, defined as a feel- ual of the PGCMS suggests that scores of 10 to ing of optimism and willingness to be involved; 17 are high; 10 to 12 midrange, and scores of 9 attitudes toward one’s own aging; satisfaction or less “are at the low end of the scale” (see with the status quo; anxiety; depression versus www.abramsoncenter.org/PRI/documents/PGC optimism; and loneliness and dissatisfaction (1). _morale_scale.pdf ). Test-retest reliability on the factor scores ranged from 0.75 to 0.80. Morris and Sherwood exam- Reliability ined the factor structure in two samples of el- Lawton studied retest reliability for several derly and moderately handicapped patients (5). groups of respondents following varying delays. Similar results were obtained from the two sam- Test-retest correlations ranged from 0.91 after ples, but they differed from those of Lawton: five weeks to 0.75 after three months (1, p150). satisfaction with the status quo and surgency For 300 respondents, a split-half reliability of were not replicated, leaving three factors (atti- 0.79 was obtained with the 22-item scale (1); tudes toward aging, agitation and loneliness) in Kuder-Richardson internal consistency was all samples. 238 Measuring Health Exhibit 5.8 The Philadelphia Geriatric Center Morale Scale Note: Asterisks indicate the 17 items retained for the shortened version. Responses indicating satisfaction are shown on the right. Item Positive response * 1. Things keep getting worse as I get older no * 2. I have as much pep as I did last year yes * 3. How much do you feel lonely? (not much, a lot) not much * 4. Little things bother me more this year no * 5. I see enough of my friends and relatives yes * 6. As you get older you are less useful no 7. If you could live where you wanted, where would you live? here * 8. I sometimes worry so much that I can’t sleep no * 9. As I get older, things are (better, worse, same) than/as I thought they would be better *10. I sometimes feel that life isn’t worth living no *11. I am as happy now as I was when I was younger yes 12. Most days I have plenty to do no *13. I have a lot to be sad about no 14. People had it better in the old days no *15. I am afraid of a lot of things no 16. My health is (good, not so good) good *17. I get mad more than I used to no *18. Life is hard for me most of the time no *19. How satisfied are you with your life today? (satisfied, not satisfied) satisfied *20. I take things hard no 21. A person has to live for today and not worry about tomorrow yes *22. I get upset easily no Derived from Lawton MP. The dimensions of morale. In: Kent DP, Kastenbaum R, Sherwood S, eds. Research planning and action for the elderly: the power and potential of social science. New York: Behavioral Publications, 1972:152–153. Also from Lawton MP. The Philadelphia Geriatric Center Morale Scale: a revision. J Gerontol 1975;30:78, Table 1. Morris and Sherwood factor analyzed an ab- 3,996 elderly respondents. Using a structural breviated version of the scale. Three factors were equation modeling approach, they identified obtained with internal consistencies ranging three first-order factors (e.g., agitation, dissatis- from 0.62 to 0.76 (5, p81). Lawton replicated faction, and attitudes toward one’s own aging) this analysis on 828 elderly community residents and one second-order factor (e.g., global life sat- (3). Seventeen items formed three factors that isfaction), which linked the three first-order fac- were comparable with those obtained by Morris tors (4). They subsequently reported that this and Sherwood: agitation (six items), attitude to- structure applied well to both males and females ward one’s own aging (five items), and lonely (10). The same three-factor structure was fur- dissatisfaction (six items). They obtained alpha ther replicated by McCulloch using a confirma- internal consistency coefficients were 0.85, 0.81, tory factor analysis. The model did not hold and 0.85, respectively (3, p87). These are the constant over time, however (11, p256). For a items in the “Revised PGC Morale Scale.” sample of 4,000 people aged 65 and older, Liang and Bollen analyzed the factor struc- Schooler factored a pool of morale-related items ture of the scale for a community sample of that included 21 of the original 22 items of the  Psychological Well-being 239 morale scale. The results “closely reproduced inally developed at the Philadelphia Center and the three factors” previously obtained by Law- includes a manual for the PGCMS at www ton (3). Liang et al. further showed that the three .abramsoncenter.org/PRI/documents/PGC_mora factor solution applied in a Japanese study (12). le_scale.pdf. Alternative Forms References Translations have been made into Castilian Spanish and Catalan (7, pp226–227). (1) Lawton MP. The dimensions of morale. In: Kent DP, Kastenbaum R, Sherwood S, eds. Research planning and action for the Reference Standards elderly: the power and potential of social The manual for the PGCMS (see Address sec- science. New York: Behavioral tion) reports mean scores for the three factors Publications, 1972:144–165. (agitation, mean 4.38; attitudes toward aging, (2) Lawton MP. Assessing the competence of mean 2.17; and lonely dissatisfaction, mean older people. In: Kent DP, Kastenbaum R, 4.81) from a community sample of elderly peo- Sherwood S, eds. Research planning and ple in the United States. action for the elderly: the power and potential of social science. New York: Commentary Behavioral Publications, 1972:122–143. The PGCMS appears to be a reliable and inter- (3) Lawton MP. The Philadelphia Geriatric Center Morale Scale: a revision. J Gerontol nally consistent scale that correlates with the 1975;30:85–89. most comparable alternative, the Life Satisfac- (4) Liang J, Bollen KA. The structure of the tion Index. The manual of the scale notes that its Philadelphia Geriatric Center Morale Scale: use in routine practice promotes communication a reinterpretation. J Gerontol between clinician and client. More data are, 1983;38:181–189. however, needed on the validity of the scale in (5) Morris JN, Sherwood S. A retesting and terms of its prediction and correlation with modification of the Philadelphia Geriatric other quality of life scales. Nevertheless, the con- Center Morale Scale. J Gerontol sistency of results across several studies suggests 1975;30:77–84. that Lawton’s scale offers a reliable measure- (6) Liang J, Lawrence RH, Bollen KA. Race ment of a relatively stable concept. differences in factorial structures of two measures of subjective well-being. J Opinion is divergent about how many items Gerontol 1987;42:426–428. to include: Morris and Sherwood and Liang and (7) Stock, WA, Okun MA, Gómez Benito J. Bollen found that two questions (numbers 3 and Subjective well-being measures: reliability 5 in Exhibit 5.8) were conceptually different and validity among Spanish elders. Int J from the rest of the scale and should be omitted, Aging Hum Devel 1994;38:221–235. but Lawton recommends retaining them. As (8) Kozma A, Stones MJ. Social desirability in Lawton noted, the PGCMS might benefit from measures of subjective well-being: a the addition of more positive affect items (3). systematic evaluation. J Gerontol Liang and Bollen reviewed the scale in some de- 1987;42:56–59. tail and provide a thoughtful discussion of the (9) Rothman ML, Hedrick S, Inui T. The alternative ways of scoring and interpreting the Sickness Impact Profile as a measure of health status of noncognitively impaired instrument (4). nursing home residents. Med Care 1989;27(suppl):S157–S167. Address (10) Liang J, Bollen KA. Sex differences in the The Philadelphia Geriatric Center has been re- structure of the Philadelphia Geriatric named the Abramson Center for Jewish Life, Center Morale Scale. J Gerontol which is part of the Polisher Research Institute; 1985;40:468–477. www.abramsoncenter.org/PRI/scales.htm. The (11) McCulloch BJ. A longitudinal investigation Web site includes information on the scales orig- of the factor structure of subjective well- 240 Measuring Health being: the case of the Phailadelphia resents more severe distress. Dupuy used a total Geriatric Center Morale Scale. J Gerontol score running from 0 to 110, and for this 14 is 1991;46:P251–P258. subtracted from the score derived from the codes (12) Liang J, Asano H, Bollen KA, et al. Cross- shown in Exhibit 5.9. Dupuy proposed cutting- cultural comparability of the Philadelphia points to represent three levels of disorder: Geriatric Center Morale Scale: an scores of 0 to 60 reflect “severe distress,” 61 to American-Japanese comparison. J Gerontol 72 “moderate distress,” and 73 to 110 “positive 1987;42:37–43. well-being” (1). Six subscores may be formed as shown in Exhibit 5.10. Using labels proposed by Brook et al. (3), the subscores measure anxiety, The General Well-Being Schedule depression, positive well-being, self-control, vi- (Harold J. Dupuy, 1977) tality, and general health. Purpose The General Well-Being Schedule (GWB) offers Reliability a brief but broad-ranging indicator of subjective Using the HANES data, Monk reported three- feelings of psychological well-being and distress month test-retest reliability coefficients of 0.68 for use in community surveys. and 0.85 for “two different groups” (2, p183). Fazio reported a retest coefficient of 0.85 after Conceptual Basis three months for 195 college students (4, p10). The conceptual description of the content of the Edwards et al. obtained a retest coefficient of GWB is contained in an unpublished report by 0.69 for 98 college graduates (5, Table 3). Dupuy (1). Reflecting the theories of Kurt The internal consistency of the GWB is very Lewin, the scale is designed to assess how the in- high: in Fazio’s study, the coefficients were 0.91 dividual feels about his “inner personal state,” for 79 men and 0.95 for 116 women (4, p11). rather than about external conditions such as in- Three other studies reported internal consis- come, work environment, or neighborhood (1). tency coefficients over 0.9 (6). Other figures in- The scale reflects both positive and negative feel- clude 0.93 from the HANES study (N = 6,913) ings; six dimensions assessed include positive (1, p7); 0.92 with black women (7, p33); 0.88 well-being, self-control, vitality, anxiety, depres- in a community sample, and 0.92 in a clinical sion, and general health. sample (8, Table 1), whereas Edwards et al. re- ported an alpha of 0.95 (5, Table 2). The Inter- Description national Quality of Life Outcomes Database The GWB is a self-administered questionnaire group (IQOD) has pooled data from 18 coun- that was developed for the U.S. Health and Nu- tries (16 languages; N = 8,536) and obtained al- trition Examination Survey (HANES I) (2). The pha values for the six dimensions of the GWB draft instrument contained 68 items, 18 of above 0.7; all of the inter-item correlations ex- which were used in the HANES study and form ceeded 0.40 (9). Fazio reported correlations the usual set of questions referred to as the among the subscores ranging from 0.16 to 0.72 GWB. They are shown in Exhibit 5.9. (4, Table 6). The GWB includes positive and negative questions. Each item has the time frame “during Validity the last month” and the first 14 questions use From the HANES data, Wan and Livieratos re- six-point response scales representing intensity ported factor analyses of the GWB items, provid- or frequency. The ordinal qualities of these re- ing three factors that explained 51% of the sponse options were checked empirically (1). variance. These factors were labeled depressive The remaining four questions use 0 to 10 rating mood, health concern, and life satisfaction (10, scales defined by adjectives at each end. In scor- Table 2; 11, Table 2). A similar result was ob- ing replies, the polarity of items 1, 3, 6, 7, 9, 11, tained by Taylor et al., using an oblique rotation; 15, and 16 is reversed, so that a lower score rep- they labeled the three factors as psychological Exhibit 5.9 The General Well-Being Schedule READ —This section of the examination contains questions about how you feel and how things have been going with you. For each question, mark (X) beside the answer which best applies to you. 1. How have you been feeling in general? 1 In excellent spirits (DURING THE PAST MONTH) 2 In very good spirits 3 In good spirits mostly 4 I have been up and down in spirits a lot 5 In low spirits mostly 6 In very low spirits 2. Have you been bothered by nervousness 1 Extremely so—to the point where I could not or your “nerves”? (DURING THE PAST work or take care of things MONTH) 2 Very much so 3 Quite a bit 4 Some—enough to bother me 5 A little 6 Not at all 3. Have you been in firm control of your behavior, 1 Yes, definitely so thoughts, emotions, OR feelings? (DURING THE 2 Yes, for the most part PAST MONTH) 3 Generally so 4 Not too well 5 No, and I am somewhat disturbed 6 No, and I am very disturbed 4. Have you felt so sad, discouraged, hopeless, or had 1 Extremely so—to the point that I have just so many problems that you wondered if anything was about given up worthwhile? (DURING THE PAST MONTH) 2 Very much so 3 Quite a bit 4 Some—enough to bother me 5 A little bit 6 Not at all 5. Have you been under or felt you were under any Yes—almost more than I could bear or 1 strain, stress, or pressure? (DURING THE PAST stand MONTH) 2 Yes—quite a bit of pressure 3 Yes—some, more than usual 4 Yes—some, but about usual 5 Yes—a little 6 Not at all 6. How happy, satisfied, or pleased have you been 1 Extremely happy—could not have been with your personal life? (DURING THE PAST more satisfied or pleased MONTH) 2 Very happy 3 Fairly happy 4 Satisfied—pleased 5 Somewhat dissatisfied 6 Very dissatisfied 7. Have you had any reason to wonder if you 1 Not at all were losing your mind, or losing control over the 2 Only a little way you act, talk, think, feel, or of your memory? 3 Some—but not enough to be concerned or (DURING THE PAST MONTH) worried about 4 Some and I have been a little concerned 5 Some and I am quite concerned 6 Yes, very much so and I am very concerned (continued) 241 Exhibit 5.9 (continued) 8. Have you been anxious, worried, or upset? 1 Extremely so—to the point of being sick or (DURING THE PAST MONTH) almost sick 2 Very much so 3 Quite a bit 4 Some—enough to bother me 5 A little bit 6 Not at all 9. Have you been waking up fresh and rested? 1 Every day (DURING THE PAST MONTH) 2 Most every day 3 Fairly often 4 Less than half the time 5 Rarely 6 None of the time 10. Have you been bothered by any illness, bodily 1 All the time disorder, pains, or fears about your health? 2 Most of the time (DURING THE PAST MONTH) 3 A good bit of the time 4 Some of the time 5 A little of the time 6 None of the time 11. Has your daily life been full of things that 1 All the time were interesting to you? 2 Most of the time (DURING THE PAST MONTH) 3 A good bit of the time 4 Some of the time 5 A little of the time 6 None of the time 12. Have you felt down-hearted and blue? 1 All the time (DURING THE PAST MONTH) 2 Most of the time 3 A good bit of the time 4 Some of the time 5 A little of the time 6 None of the time 13. Have you been feeling emotionally stable and 1 All the time sure of yourself? (DURING THE PAST MONTH) 2 Most of the time 3 A good bit of the time 4 Some of the time 5 A little of the time 6 None of the time 14. Have you felt tired, worn out, used-up, or 1 All the time exhausted? (DURING THE PAST MONTH) 2 Most of the time 3 A good bit of the time 4 Some of the time 5 A little of the time 6 None of the time For each of the four scales below, note that the words at each end of the 0 to 10 scale describe opposite feelings. Circle any number along the bar which seems closest to how you have generally felt DURING THE PAST MONTH 15. How concerned or worried about your HEALTH 0 1 2 3 4 5 6 7 8 9 10 have you been? (DURING THE PAST MONTH) Not Very concerned concerned at all 242  Psychological Well-being 243 Exhibit 5.9 16. How RELAXED or TENSE have you been? 0 1 2 3 4 5 6 7 8 9 10 (DURING THE PAST MONTH) Very Very relaxed tense 17. How much ENERGY, PEP, VITALITY have 0 1 2 3 4 5 6 7 8 9 10 you felt? (DURING THE PAST MONTH) No energy Very AT ALL, ENERGETIC, listless dynamic 18. How DEPRESSED or CHEERFUL have 0 1 2 3 4 5 6 7 8 9 10 you been? (DURING THE PAST MONTH) Very Very depressed cheerful Reproduced from Fazio AF. A concurrent validational study of the NCHS General Well-Being Schedule. Hyattsville, Maryland: U.S. Department of Health, Education and Welfare, National Center for Health Statistics, 1977:34–36. With permission. distress, well-being and vitality, and general health (7, Table 3). Given that the factors corre- lated quite strongly they concluded that using a Exhibit 5.10 The General Well-Being Schedule: single overall score was justified (7, p37). A four- Subscore Labels and Question Topics factor solution replicated Taylor’s study, but sep- arated well-being from vitality (12, Table 1). Subscore labels Question topics Considerable evidence confirms the correla- Anxiety 2. nervousness tional validity of the GWB. In Fazio’s validation 5. strain, stress, or pressure study, the GWB total score correlated 0.47 with 8. anxious, worried, upset an interviewer’s rating of depression, 0.66 with 16. relaxed, tense Zung’s Self-Rating Depression Scale, and 0.78 Depression 4. sad, discouraged, hopeless with the Personal Feelings Inventory—Depres- 12. down-hearted, blue 18. depressed sion (4, Table B). The average correlation of the Positive well-being 1. feeling in general GWB and six independent depression scales was 6. happy, satisfied with life 0.69; the average correlation was 0.64 with 11. interesting daily life three anxiety scales (4, p10). Simpkins and Self-control 3. firm control of behavior, Burke obtained correlations of 0.70 with a ten- emotions item depression score, 0.58 with the Lubin De- 7. afraid losing mind, or pression Adjective Checklist, and 0.80 with losing control Zung’s Self-Rating Depression Scale (6, Table 9; 13. emotionally stable, sure of self 13). Brook et al. reported correlations between Vitality 9. waking fresh, rested the GWB subscales and reports of stress at home 14. feeling tired, worn out and at work ranging from 0.17 to 0.59 (3, Table 17. energy level 12). A Japanese study reported a correlation of General health 10. bothered by illness −0.76 with the General Health Questionnaire; 15. concerned, worried about −0.67 with the state anxiety scale of the State- health Trait Anxiety Inventory, and −0.66 for the trait 244 Measuring Health scale; −0.59 with the Center for Epidemiologic plained, perceived health status explained the Studies Depression scale, and −0.55 with Zung’s most, followed by symptoms of a so-called ner- Self-Rating Depression Scale (14, Table 5). vous breakdown, number of consultations for Correlations between individual GWB sub- counseling, use of headache or sleep medica- scales and criterion ratings were reported by tions, weight loss, followed by having consulted Fazio (4) and by Ware et al. (6). In the main, a psychiatrist. In a similar analysis, physical con- such correlations were high, frequently falling ditions explained 20 to 24% of the variance (11, between 0.65 and 0.90. Using an interviewer’s Table 6). Stephens analyzed U.S. and Canadian rating as the criterion, Fazio noted that three survey data to show a significant association be- short subscales of the GWB correlated with the tween mental well-being measured by the GWB criterion about as well or better than many and the level of physical activity, controlling for longer scales (4, p8 and Table A). Correlations education, age, and physical health status (15). with use of services were summarized by Ware The GWB was only weakly related to sociode- et al. and fell in the range of 0.09 to 0.48 (6, mographic status (r = 0.25), and this association pp48–49). The draft version of the GWB con- disappeared when the somatic and the psycho- tained a validation question: “Have you had se- logical problem indexes were controlled for (1). vere enough personal, emotional, behavior or An analysis using automatic interaction detec- mental problems that you felt you needed help tion confirmed that gender only accounted for 1 during the past year?” Dupuy showed a correla- to 4% of variance in GWB scores in the various tion of 0.53 between the GWB total score and age groups, and education explained even less this question (N = 2,007 from the HANES sur- variance (11, Table 3). vey); Simpkins and Burke reported a correlation Edwards et al. showed a significant contrast of 0.67 (13, Table 9). The GWB total score cor- in GWB scores between psychiatric day patients related −0.46 with the Beck Depression Inven- and nonpatient volunteers (5, Table 1). They tory (7, Table 5). also showed the GWB detected progress made To test discriminant validity, Dupuy used the by 21 psychiatric day patients after two weeks HANES data to construct a sociodemographic of treatment. Simpkins and Burke’s comparison index (reflecting social class and size of house- of community and psychiatric patient samples hold), a somatic index (which covered use of yielded a point biserial correlation of 0.56 with medications, self-report of symptoms of anxiety, GWB scores (13, p38). and a self-rating of general health), and a psy- chological problem index. The multiple correla- Alternative Forms tion between the GWB overall score and the The GWB items shown in Exhibit 5.9 were somatic and psychological indexes was 0.73 (1, modified for inclusion in the draft version of the p9). Fifty-five questions covering clinical symp- RAND Mental Health Inventory and extensive toms and self-perceptions of health in the scaling, reliability, and validity tests were carried HANES study explained 31% of the variance in out for the Health Insurance Study. This instru- GWB scores (10). Indicators of psychological ment was named the “HIS-GWB.” This has led factors (e.g., perceived nervousness, use of med- us to include some of the RAND findings here as ications, consultations) were responsible for be- suggestive of the quality of the GWB questions. tween 30 and 21% of the variance, the amount The same 22 items were later termed the Psycho- falling with rising age. Physical well-being ex- logical General Well-Being Index (PGWBI) by plained between 3 and 17% of variance, rising Dupuy (16). with age (11, Table 4). Finally, multiple classifi- The validity of Dupuy’s hypothesized group- cation analysis showed the relative contribution ing of the PGWB items into six subscales was of various mental health indicators in explaining evaluated empirically on 1,209 respondents us- variance in the GWB scores. Overall, 36% of ing multitrait and factor analyses (6). A six- variance was explained, almost constant across factor solution provided results that agreed age groups (11, Table 5). In order of variance ex- closely with the structure hypothesized by  Psychological Well-being 245 Dupuy, providing scores indicating anxiety, de- correlated 0.93 with the total GWB score (7, pression, self-control, positive well-being, gen- p37). eral health, and vitality. Internal consistency A British version of the GWB has been pro- coefficients for these six subscales ranged from posed (19) and validated (20). It is called the 0.72 to 0.88 (6, Table 20; 16, Table II). One- adapted GWB index (AGWBI); a copy is pro- week test-retest reliability estimates were made vided in the article by Hopton et al. (20, Appen- for the anxiety and positive well-being scales, dix 3). Other translations of the GWB into 16 and were 0.70 and 0.74, respectively (N = 437) languages have been coordinated through the (6, Table 20). Reliability declined to 0.50 when Mapi Research Institute (address below), and the retest interval exceeded one month (6, p77). summary information is available from Validity was examined by correlating GWB sub- www.iqod.org. Psychometric information is scales and overall scores with 24 validating vari- available for the Spanish (21), French (22) and ables covering stress, recognition of mental Japanese versions (14). problems, life satisfaction and use of mental health care. Of 192 correlations, 158 were sta- Reference Standards tistically significant at p ≤ 0.01 in the hypothe- Dupuy derived U.S. national reference standards sized direction. The nonsignificant associations from the HANES data (16, Table II). Seventy- pertained to stressful life events, which occurred one percent of the adult population fell into the rarely in the sample under study (6, Table 22). positive well-being category (scores 73 to 110), Dupuy reported correlations between the PGWB 15.5% showed moderate distress (scores 61 to and the Center for Epidemiologic Studies De- 72), and 13.5% were classified as experiencing pression Scale (−0.72), the Beck Depression In- severe distress (scores 0 to 60) (1, p10). About ventory (−0.68), the Zung depression scale 60% of the population were both free from se- (−0.75), the Langner 22-item score (−0.77), and vere problems over the past year and in a state of the Health Opinion Survey (−0.59) (16, Table positive well-being during the past month. Mean III). Alpha reliability for the PGWB Index scores for each item by age-group are available ranged from 0.90 to 0.94 in seven samples (16, from the 1971–1975 HANES data (11, Table 1). Table VI). Fifteen of the GWB items were re- Other reference figures were presented by Fazio tained for use in the final version of the RAND (4, Table 1). Mental Health Inventory, which we describe in the following review (6, p94; 17). Commentary A ten-item version of the GWB is called the The GWB Schedule improves on the older meth- Psychological Mental Health Index (18). It in- ods reviewed in this chapter in several respects. cludes four subscales: positive well-being (items Like the Bradburn scale, it includes positive 1 and 6 from Exhibit 5.9), depressed mood well-being but, reflecting a criticism of the Brad- (items 4, 12), behavioral-emotional control burn scale, it divides positive questions into sep- (items 3, 7, and 13), and tension-anxiety (items arate dimensions. It avoids reference to physical 2, 5, and 8). Administered to patients with long- symptoms of emotional distress and so avoids term psychosis, a retest coefficient of 0.27 was the interpretation problems seen with the Health obtained; internal consistency alpha scores Opinion Survey and Langner scales. The avail- ranged from 0.69 to 0.85. The item-total corre- able reliability and validity tests show extremely lations ranged from 0.38 to 0.64 (18, p233). good results—internal consistency is higher than The ten-item version correlated 0.45 with a for other scales and there is wide evidence of therapist-rated symptom score. In the light of agreement with other purpose-built depression these low coefficients, Ulin concluded that fur- and anxiety scales. Fazio’s study indicated that ther research is needed to test this abbreviation the GWB performed as well as several other of the GWB before it can be recommended for leading scales in assessing emotional distress in a general use. Taylor et al. noted that items 1, 12, student sample. He concluded: “the GWB 16, and 18 formed an abbreviated version that emerged as the single most useful instrument in 246 Measuring Health measuring depression” (4). A possible weakness adults. Los Angeles, California: Paper in the performance of the scale was noted by Ed- presented at American Public Health wards et al., who noted that, although internal Association Meeting, 1978. consistency was excellent, test-retest reliability (2) Monk M. Blood pressure awareness and was low (5). This may, of course, be because it is psychological well-being in the Health and Nutrition Examination Survey. Clin Invest sensitive to change, but no data have been re- Med 1981;4:183–189. ported on the responsiveness of the GWB. Given (3) Brook RH, Ware JE, Jr., Davies-Avery A, et the quality of the GWB Schedule, it is unfortu- al. Overview of adult health status mea- nate that so many of the validation studies are sures fielded in Rand’s Health Insurance unpublished. The most useful document summa- Study. Med Care 1979;17(suppl):1–131. rizing the unpublished material is the RAND re- (4) Fazio AF. A concurrent validational study view by Ware et al. (6). It is disappointing that a of the NCHS General Well-Being Schedule. scale of the potential of the GWB Schedule does Hyattsville, MD: U.S. DHEW, National not benefit from a user’s manual such as that Center for Health Statistics, 1977. produced by Goldberg for the General Health (5) Edwards DW, Yarvis RM, Mueller DP, et Questionnaire. al. Test-taking and the stability of adjustment scales: can we assess patient Although Dupuy’s description of the concep- deterioration? Eval Q 1978;2:275–291. tual structure of the GWB is vague, the results of (6) Ware JE, Jr., Johnston SA, Davies-Avery a large factor analytic study supported the di- A, et al. Conceptualization and measure- mensions originally built into the scale. How- ment of health for adults in the Health ever, more recent commentators have suggested Insurance Study. Vol. III, Mental health. that the GWB is primarily unidimensional, not- Santa Monica, CA: RAND Corporation ing the high internal consistency and inconsis- (Publication No. R-1987/3-HEW), tent results of factor analyses (7). Some debate 1979. has arisen over the most useful way to score the (7) Taylor JE, Poston II WSC, Haddock CK, et GWB. Because internal consistency is high, sub- al. Psychometric characteristics of the scores may be redundant. Wan and Livieratos ar- General Well-Being Schedule (GWB) with African-American women. Qual Life Res gued that with so few items, subscores would 2003;12:31–39. provide only crude measurements and that an (8) Himmelfarb S, Murrell SA. Reliability and overall score was better (11). Fazio’s results did, validity of five mental health scales in older indeed, show lower internal consistency for the persons. J Gerontol 1983;38:333–339. subscales than for the instrument as a whole. (9) Lobo-Luppi L, Mouly M. The Because of its strong reliability and validity International health-related Quality of life results, we recommend that the GWB Schedule Outcomes Database (IQOD) programme— be seriously considered for use where a general WHQ and PGWBI databases. Reference population indicator of subjective well-being is values for cross-cultural comparisons. required. We know less about its adequacy as a Lyon, France: MAPI Research Institute, case-detection instrument, for which the General 2002. (10) Wan TTH, Livieratos B. A validation of the Health Questionnaire is recommended. General Well-Being Index: a two-stage multivariate approach. Paper presented at: Address American Public Health Association Translations are available through the Mapi Re- Meeting; November, 1977; Washington, search Institute, 27 rue de la Villette, 69003 DC. Lyon, France (www.iqod.org). (11) Wan TTH, Livieratos B. Interpreting a general index of subjective well-being. Milbank Mem Fund Q 1978;56:531–556. References (12) Poston II WSC, Olvera NE, Yanez C, et al. (1) Dupuy HJ. Self-representations of general Evaluation of the factor structure and psycholgical well-being of American psychometric characteristics of the General  Psychological Well-being 247 Well-Being Schedule (GWB) with Mexican The RAND Mental Health Inventory American women. Women Health (RAND Corporation and John E. 1998;27:51–64. Ware, 1979) (13) Simpkins C, Burke FF. Comparative analyses of the NCHS General Well-Being Schedule: Purpose response distributions, community vs. The Mental Health Inventory (MHI) measures patient status discriminations, and content mental health in terms of psychological distress relationships. Nashville, TN: Center for and well-being, focusing on affective states (1, Community Studies, George Peabody p105). It was developed for use in population College (Contract No. HRA 106-74-13), surveys. 1974. (14) Nakayama T, Toyoda H, Ohno K, et al. Validity, reliability and acceptability of the Conceptual Basis Japanese version of the General Well-Being Veit and Ware discussed the limitations of early Schedule (GWBS). Qual Life Res screening tests such as the Health Opinion Sur- 2000;9:529–539. vey and the Langner scale. Reliance on somatic (15) Stephens T. Physical activity and mental symptoms of distress means that such methods health in the United States and Canada: may not be able to distinguish changes in mental evidence from four population surveys. health from changes in physical health, and Prev Med 1988;17:35–47. these symptoms they include are rarely encoun- (16) Dupuy HJ. The Psychological General tered in the general population (2). Veit and Well-Being (PGWB) Index. In: Wenger NK, Mattson ME, Furberg CD, et al., eds. Ware noted: Assessment of quality of life in clinical trial of cardiovascular therapies. New York: Le a substantial proportion of people in a gen- Jacq, 1984:170–183. eral population rarely or never report occur- (17) Veit CT, Ware JE, Jr. The structure of rences of even the most prevalent psychological distress and well-being in psychological distress symptoms. To increase general populations. J Consult Clin Psychol measurement precision, it may be necessary 1983;51:730–742. to extend the definition of mental health . . . (18) Ulin PR. Measuring adjustment in to include characteristics of psychological chronically ill clients in community mental well-being (e.g., feeling cheerful, interest in health care: an assessment of the and enjoyment of life). Psychological well- Psychological Mental Health Index. Nurs being items have the potential to improve the Res 1981;30:229–235. (19) Hunt SM, McKenna SP. A British precision of mental health measurement by adaptation of the General Well-Being distinguishing among persons who receive Index: a new tool for clinical research. Br J perfect scores on measures of psychological Med Econ 1992;2:49–60. distress. (2, p730) (20) Hopton JL, Hunt SM, Shiels C, et al. Mea- suring psychological well-being. The As well as being developed as a screening in- adapted General Well-being Index in a strument, the MHI was also used to examine the primary care setting: a test of validity. Fam structure of mental health: are distress and posi- Pract 1995;12:452–460. tive well-being separate dimensions (as argued (21) Badia X, Gutierrez F, Wiklund I, et al. by Bradburn), and are these concepts themselves Validity and reliability of the Spanish multidimensional, implying that they should be version of the Psychological General Well- further subdivided (2)? To develop an instrument Being Index. Qual Life Res 1996;5:101–108. that could reflect the multidimensional nature of (22) Bravo G, Hébert R. Validation d’une psychological well-being, Veit and Ware incor- échelle de bien-être général auprès d’une porated four factors hypothesized by Dupuy— population francophone agée de 50 à 75 anxiety, depression, loss of behavioral/ ans. Can J Aging 1996;15:112–118. emotional control, and general positive affect— 248 Measuring Health and added a fifth factor, emotional ties, to form site provides further scoring information and a the basis for the MHI (2). The behavioral con- user’s manual (www.rand.org/health/surveys/ trol dimension covers emotional stability and section5.html). control of behavior or thoughts and feelings, in- cluding fear of losing one’s mind. A fuller dis- Reliability cussion of the relationships among these The MHI was tested on a representative popula- constructs is given by Stewart et al. (1, tion sample of 5,089 respondents in the RAND pp106–107). Health Insurance Experiment. One-year test- retest results were based on 3,525 respondents, Description and coefficients ranged from 0.56 (for the de- The MHI formed the primary mental health pression scale) to 0.63 (for anxiety). Test-retest measurement in the RAND Health Insurance reliability of the overall score was 0.64 (2). Inter- Experiment. It focuses on mood and symptoms nal consistency coefficients ranged from 0.83 to of anxiety and of loss of control over feelings, 0.92 for the five scales; the coefficient was 0.96 thoughts, and behavior (3). The MHI used 15 for the overall score (2, Table 6). items from Dupuy’s General Well-Being (GWB) Schedule: the GWB items covering general Validity health and vitality were discarded because they Veit and Ware presented an extensive discussion failed discriminant tests of validity (2). Twenty of the factorial structure of the MHI, from items were drawn from other scales to cover which they derived a hierarchical model of the anxiety, depression, general positive affect, and structure of the scores it provides. The items loss of behavioral or emotional control; three were found to fall onto the five factors indicated items were written to cover the fifth hypothe- in Exhibit 5.11. Correlations among the factors sized factor, emotional ties. To these 38 items, ranged from −0.39 to +0.77 and the five factors another eight may be added to assess a socially were, in turn, grouped into two higher-order desirable response set (3). Details of the develop- factors termed “psychological distress” (incor- ment of the MHI are given in several sources porating the negative items) and “psychological (1–3). The questions and response scales are well-being.” These factors correlated −0.75, and shown in Exhibit 5.11 along with the factor may be regarded as forming a bipolar distress placement of each item. The questionnaire is versus well-being measurement of general men- self-administered, taking about ten minutes (4, tal health. These results supported the hypothe- p182S), and items refer to the past month. Most sized multidimensional model of emotional of the response scales have six options. For com- well-being, although a strong general factor un- parability, the response options were kept close derlies the instrument (2). This factor structure to those used in the questionnaires from which has been replicated, with minor modifications, the items were originally drawn. by Zautra et al. (5). Manne and Schnoll repli- As well as an overall score known as the cated these analyses in a sample of cancer pa- MHI, subscores are available for psychological tients, showing that a correlated five-factor distress and psychological well-being (see the fi- model best fit the data (6, Figure 1). nal column of Exhibit 5.11). Three distress Ware et al. showed a strong association be- scores include anxiety, depression, and loss of tween MHI scores and the use of ambulatory behavioral or emotional control; two well-being mental health services in a prospective study (2; scores represent general positive affect and emo- 7). Correlations between the MHI and criterion tional ties (1, p105). The subscales can be scored measurements were shown by Ware et al. (8). and interpreted separately or scores can be ag- The correlations between a life events scale and gregated into the MHI (2). When combining all the various sections of the MHI ranged from the scores, it is necessary to reverse the scoring of 0.12 to 0.26; correlations with life satisfaction the positive section. This may be done by sub- ran from 0.40 to 0.51, and correlations with an tracting the raw score from 77. The RAND web indicator of severe emotional problems ranged Exhibit 5.11 The RAND Mental Health Inventory, Showing Response Scales and Factor Placement of Each Item Note: The answer scales vary from question to question and are shown at the foot of the table. Letters to the right of each question indicate the response scale that is applicable: T refers to the answer scale indicating time or frequency, AN indicates the scale running from always to never, and U indicates a unique answer category. The factor placement of each item is shown in the right-hand column: Anx = anxiety, Dep = depression, Behave = behavioral/emotional control, Pos = general positive affect and Emotion = emotional ties. Question Response scale Factor How happy, satisfied, or pleased have you been with your personal life during U1 Pos the past month? How much of the time have you felt lonely during the past month? T Emotion How often did you become nervous or jumpy when faced with excitement or AN Anx unexpected situations during the past month? During the past month, how much of the time have you felt that the future looks T Pos hopeful and promising? How much of the time, during the past month, has your daily life been full of T Pos things that were interesting to you? How much of the time, during the past month, did you feel relaxed and free of T Pos tension? During the past month, how much of the time have you generally enjoyed the T Pos things you do? During the past month, have you had any reason to wonder if you were losing U2 Behav your mind, or losing control over the way you act, talk, think, feel, or of your memory? Did you feel depressed during the past month? U3 Dep During the past month, how much of the time have you felt loved and wanted? T Emotion How much of the time, during the past month, have you been a very nervous T Anx person? When you got up in the morning, this past month, about how often did you AN Pos expect to have an interesting day? During the past month, how much of the time have you felt tense or “high- T Anx strung”? During the past month, have you been in firm control of your behavior, thoughts, U4 Behav emotions, feelings? During the past month, how often did your hands shake when you tried to do AN Anx something? During the past month, how often did you feel that you had nothing to look AN Behav forward to? How much of the time, during the past month, have you felt calm and peaceful? T Pos How much of the time, during the past month, have you felt emotionally stable? T Behav How much of the time, during the past month, have you felt downhearted and blue T Dep How often have you felt like crying, during the past month? AN Behav During the past month, how often did you feel that others would be better off if AN Behav you were dead? How much of the time, during the past month, were you able to relax without T Anx difficulty? During the past month, how much of the time did you feel that your love T Emotion relationships, loving and being loved, were full and complete? How often, during the past month, did you feel that nothing turned out for you AN Behav the way you wanted it to? (continued) 249 Exhibit 5.11 (continued) Question Response scale Factor How much have you been bothered by nervousness, or your “nerves,” during the U5 Anx past month? During the past month, how much of the time has living been a wonderful T Pos adventure for you? How often, during the past month, have you felt so down in the dumps that AN Behav nothing could cheer you up? During the past month, did you ever think about taking your own life? U6 Behav During the past month, how much of the time have you felt restless, fidgety, or T Anx impatient? During the past month, how much of the time have you been moody or brooded T Dep about things? How much of the time, during the past month, have you felt cheerful, T Pos lighthearted? During the past month, how often did you get rattled, upset, or flustered? AN Anx During the past month, have you been anxious or worried? U7 Anx During the past month, how much of the time were you a happy person? T Pos How often during the past month did you find yourself having difficulty trying to AN Anx calm down? During the past month, how much of the time have you been in a low or very T Dep low spirits? How often, during the past month, have you been waking up feeling fresh and U8 Pos rested? During the past month, have you been under or felt you were under any strain, U9 Dep stress, or pressure? Response scales and scores: T (1) All of the time (4) Some of the time (2) Most of the time (5) A little of the time (3) A good bit of the time (6) None of the time AN (1) Always (4) Sometimes (2) Very often (5) Almost never (3) Fairly often (6) Never U1 (1) Extremely happy, could not have been more satisfied or pleased (2) Very happy most of the time (3) Generally satisfied, pleased (4) Sometimes fairly satisfied, sometimes fairly unhappy (5) Generally dissatisfied, unhappy (6) Very dissatisfied, unhappy most of the time U2 (1) No, not at all (2) Maybe a little (3) Yes, but not enough to be concerned or worried about it (4) Yes, and I have been a little concerned (5) Yes, and I am quite concerned (6) Yes, and I am very much concerned about it 250 Exhibit 5.11 U3 (1) Yes, to the point that I did not care about anything for days at a time (2) Yes, very depressed almost every day (3) Yes, quite depressed several times (4) Yes, a little depressed now and then (5) No, never felt depressed at all U4 (1) Yes, very definitely (2) Yes, for the most part (3) Yes, I guess so (4) No, not too well (5) No, and I am somewhat disturbed (6) No, and I am very disturbed U5 (1) Extremely so, to the point where I could not take care of things (2) Very much bothered (3) Bothered quite a bit by nerves (4) Bothered some, enough to notice (5) Bothered just a little by nerves (6) Not bothered at all by this U6 (1) Yes, very often (2) Yes, fairly often (3) Yes, a couple of times (4) Yes, at one time (5) No, never U7 (1) Yes, extremely so, to the point of being sick or almost sick (2) Yes, very much so (3) Yes, quite a bit (4) Yes, some, enough to bother me (5) Yes, a little bit (6) No, not at all U8 (1) Always, every day (2) Almost every day (3) Most days (4) Some days, but usually not (5) Hardly ever (6) Never wake up feeling rested U9 (1) Yes, almost more than I could stand or bear (2) Yes, quite a bit of pressure (3) Yes, some, more than usual (4) Yes, some, about normal (5) Yes, a little bit (6) No, not at all Adapted from Veit CT, Ware JE, Jr. The structure of psychological distress and well-being in general populations. J Consult Clin Psychol 1983;51:733,Table 1. Also from Ware JE Jr. Johnston SA, Davies-Avery A, Brook RH. Conceptualization and measure- ment of health for adults in the Health Insurance Study: Vol. III, Mental Health. Santa Monica, California: RAND Corporation, 1979:Table 27 and Appendix E. 251 252 Measuring Health from 0.48 to 0.58 (8, Table 4). Convergent cor- performed almost as well as the MHI18 in de- relations with scales of the Positive and Negative tecting any DIS disorder, with areas under the Affect Scale (PANAS) included 0.70 between ROC curve of 0.79 and 0.80, respectively (9, MHI depression and PANAS negative affect; Table 1). The GHQ performed slightly less well 0.65 between anxiety and negative affect, and (0.77), and the SSI was the least accurate (0.71) 0.59 between MHI general positive affect and (9, Fig 1). The two versions of the MHI also per- the PANAS positive score (6, Table 5). formed as well as the GHQ in detecting depres- sive symptoms, whereas the SSI was again less Alternative Forms successful. The item on feeling downhearted and As with all RAND scales, several abbreviated blue detected nearly three-quarters of the DIS versions of the MHI have been developed. We disorders, with only a 5% false-positive rate, list a select few of these, reassuring readers that forming “a powerful nonspecific detector for all there are yet others; fuller details are given by of the five diagnostic clusters.” (9, p173). In a Stewart et al. (1, Table 7–2), and the RAND web separate analysis, the MHI performed better site provides further information (www.rand.org/ than the GHQ, showing an area under the curve health/surveys/section5.html). of 0.76 compared with 0.68 for the GHQ (10, A five-item version (the MHI5) is quite Table 3). widely used (9; 10). The items are introduced by A subsequent version of the MHI18 omitted the question, “How much of the time, during the the question on ability to relax without diffi- last month, have you . . .” culty, making a 17-item version (1, Table 7-2). This had an internal consistency between 0.94 and 0.96 in different samples (1, Table 7-13). 1. “. . . been a very nervous person?” A new, revised version of the MHI was cre- 2. “. . . felt calm and peaceful?” ated for use in the Medical Outcomes Study. 3. “. . . felt downhearted and blue?” This contained 33 items, of which 24 are identi- 4. “. . . been a happy person?” cal to items in the 38-item MHI described here 5. “. . . felt so down in the dumps that noth- and six others contain slight variations in word- ing could cheer you up?” ing or response categories (1, Table 7-2). This version was tested, along with the MHI5 and The first four items use the “T” response scale MHI17, in various parts of the Medical Out- shown at the foot of Exhibit 5.11, whereas the comes Study. Item-total correlations ranged fifth uses scale “AN.” The MHI5 was subse- from 0.50 to 0.87 (1, p128). quently incorporated into the Short-Form-20 A validation of a Chinese version of the MHI and -36 instruments described in Chapter 10. has been reported (11). Item-total correlations ranged from 0.54 to 0.81; alpha was 0.90 in one study and 0.86 in Reference Standards another (1, pp128, 134). Veit and Ware presented mean scores for each An 18-item version of the MHI was devel- section of the MHI, although these figures are oped for administration by telephone. It con- based on slightly different numbers of items tained four or more items from each of the from those shown in Exhibit 5.11 (2, Table 6). anxiety, depression, behavioral control, and pos- itive affect subscales of the MHI, including the Commentary five items just listed (1, p110). Berwick et al. The MHI incorporates the most adequate ques- compared the MHI5 and the MHI18, the 30- tions from some of the leading mental health item General Health Questionnaire (GHQ), and scales; it has been carefully constructed and ap- a 28-item Somatic Symptom Inventory (SSI) pears to have been used without alterations to against a criterion diagnosis using the Diagnos- the wording of the questions. The MHI deliber- tic Interview Schedule (DIS) (9; 10). The MHI5 ately focused on affective indicators of well-  Psychological Well-being 253 being and therefore avoided the problems that of health for adults in the Health Insurance beset the earlier scales such as the Health Opin- Study. Vol. III, Mental health. Santa ion Survey and the Langner scale. The MHI and Monica, CA: RAND Corporation its derivatives have been used in several large (Publication No. R-1987/3-HEW), studies and to predict service use (2). 1979. (4) Smith GR, Witt AS, Golding JM. The Stewart et al. provided an insightful discus- relationship of function-specific mental sion of the relative merits of the various abbrevi- health measures to psychiatric diagnoses. ations of the MHI and of the concepts the scales Control Clin Trials 1991;12:180S–188S. assess (1, p139–141). They also mentioned the (5) Zautra AJ, Guarnaccia CA, Reich JW. possible problem of response bias whereby some Factor structure of mental health measures people underreport distress and others overre- for older adults. J Consult Clin Psychol port. It was not feasible to create a balanced set 1988;56:514–519. of positively and negatively worded items for the (6) Manne S, Schnoll R. Measuring cancer MHI (1, p141). patients’ psychological distress and well- The MHI should be seriously considered as being: a factor analytic assessment of the an alternative to the General Well-Being Sched- Mental Health Inventory. Psychol Assess 2001;13:99–109. ule in general population surveys, because more (7) Ware JE, Jr., Manning WG, Jr., Duan N, et published material is available and because it has al. Health status and the use of ambulatory extended the scope of Dupuy’s scale. A direct mental health services. Am Psychol comparison of the sensitivity and specificity of 1984;39:1090–1100. the two methods would be beneficial. The abbre- (8) Ware JE, Jr., Davies-Avery A, Brook RH. viated versions of the scale perform well, at least Conceptualization and measurement of when presented as overall scores. The stability of health for adults in the Health Insurance the subscores from the abbreviated instruments Study. Vol. VI, Analysis of relationships is likely to be limited, however. The success of among health status measures. Santa the single question on feeling downhearted and Monica, CA: RAND Corporation blue commends it as an option when a single- (Publication No. R-1987/6-HEW), 1980. (9) Berwick DM, Murphy JM, Goldman PA, item screen for mental distress is required. The et al. Performance of a five-item mental strength of single screening items is again illus- health screening test. Med Care trated in the review of single-item scales in 1991;29:169–176. Chapter 10, and supports the approach of the (10) Weinstein MC, Berwick DM, Goldman PA, Dartmouth COOP Charts described in Chapter et al. A comparison of three psychiatric 10. screening tests using Receiver Operating Characteristic (ROC) analysis. Med Care 1989;27:593–607. References (11) Liang J, Wu SC, Krause NM, et al. The (1) Stewart AL, Ware JE, Jr., Sherbourne CD, structure of the Mental Health Inventory et al. Psychological distress/well-being and among Chinese in Taiwan. Med Care cognitive functioning measures. In: Stewart 1992;30:659–676. AL, Ware JE, Jr., eds. Measuring functioning and well-being: the Medical Outcomes Study approach. Durham, NC: The Health Perceptions Questionnaire Duke University Press, 1992:102–142. (J.E. Ware, 1976) (2) Veit CT, Ware JE, Jr. The structure of psychological distress and well-being in Purpose general populations. J Consult Clin Psychol The Health Perceptions Questionnaire (HPQ) is 1983;51:730–742. a self-report instrument that records perceptions (3) Ware JE, Jr., Johnston SA, Davies-Avery A, of past, present, and future health; resistance to et al. Conceptualization and measurement illness; and attitudes toward sickness (1). It is a 254 Measuring Health survey instrument that has been used as an out- subscales: current health (nine items), prior come measurement in the Health Insurance Ex- health (three items), health outlook (four items), periment (HIE) and as a predictor of use of care resistance to illness (four items), health (2). worry/concern (five items), and sickness orienta- tion (two items). Six items are not used in the Conceptual Basis subscales; these cover rejection of the sick role “Health perceptions are personal beliefs and and attitudes toward going to the doctor (7). The evaluations of general health status” and refer to items comprising each scale are indicated in the whether people see themselves as well or unwell right-hand column of Exhibit 5.12. The 22 (3, p143). This is a subjective concept and per- items used in forming an overall General Health ceptions may reflect a person’s feelings and be- Rating Index (GHRI) are also identified. Where liefs more than her actual physical health. Ware an indication of general health is required but wrote: space does not permit fielding the 22-item GHRI, either the nine items forming the current Measures of general health perceptions differ health subscale can be used, or a four-item ver- from other health status measures in that they sion can be used which includes items I, V, Q, do not specify one or more components of and Z (7, p103). The questions use five-point, health (physical, mental, or social). Rather, Likert-type responses and the full instrument is respondents are asked only for an assessment self-administered in 7 to 11 minutes (8). of their “health.” In theory, this difference in Summated scores are calculated for each of measurement strategy makes it possible to the six subscales and an overall score is derived achieve two important goals. First, general for the General Health Rating Index. For this it health ratings may constitute one kind of is first necessary to reverse the scores on items C, overall health status index if respondents con- E, F, I, K, L, R, T, Z, CC, DD by subtracting sider all health components when they make each response from 6; the score for question 6 is their ratings. Second, general health ratings also reversed by subtracting it from 5. Davies et may reflect the objective information people al. handled missing data for individual items by have about their health status as well as their substituting the mean score on the remaining evaluation of that information and may, items of that scale, after the appropriate item therefore, help solve the problem of aggregat- scores have been reversed (7, p227). Because ing the two kinds of health status data. (4, item-total correlations are similar, differential page v) weights are not used in computing scores (9). Raw subscale scores and the GHRI may be This is relevant because subjective perceptions of transformed to a 0 to 100 scale. The formula is: health, rather than objective measures of health (Actual raw score − status, predict use of care; health perceptions fit Lowest possible within the conceptual framework of the Health Transformed = raw score) × 100 Belief Model of health behavior (3, p144). score (Highest possible Whereas previous measurement of health per- raw score − Lowest ceptions generally used single items, the RAND possible raw score) group developed a multi-item scale to test hy- pothesized dimensions of the overall concept (5). Full details of scoring are given by Davies et al. (7). Description The HPQ contains the 33 items shown in Ex- Reliability hibit 5.12. The questions were originally tested There are extensive data on internal consistency for the RAND Health Insurance Experiment; and test-retest reliability from several large pop- further items did not satisfy scaling criteria and ulation samples. Typical results are summarized were discarded (6, Table 1). The items form six in Table 5.3. The first column shows median al- Exhibit 5.12 The Health Perceptions Questionnaire, Showing the Items Included in Each Subscore Note: In copy given to respondent, the two columns on the right are omitted. Please read each of the following statements, and then circle one of the numbers on each line to indicate whether the statement is true or false for you. There are no right or wrong answers. If a statement is definitely true for you, circle 5. If it is mostly true for you, circle 4. If you don’t know whether it is true or false, circle 3. If it is mostly false for you, circle 2. If it is definitely false for you, circle 1. Some of the statements may look or seem like others. But each statement is different, and should be rated by itself. Definitely Mostly Don’t Mostly Definitely true true know false false GHRI Subscale* A. According to the doctors 5 4 3 2 1 • CH I’ve seen, my health is now excellent B. I try to avoid letting illness 5 4 3 2 1 interfere with my life C. I seem to get sick a little 5 4 3 2 1 • RI easier than other people D. I feel better now than I ever 5 4 3 2 1 • CH have before E. I will probably be sick a lot 5 4 3 2 1 • HO in the future F. I never worry about my 5 4 3 2 1 • HW health G. Most people get sick a little 5 4 3 2 1 • RI easier than I do H. I don’t like to go to the 5 4 3 2 1 doctor I. I am somewhat ill 5 4 3 2 1 • CH J. In the future, I expect to 5 4 3 2 1 • HO have better health than other people I know K. I was so sick once I thought 5 4 3 2 1 • PH I might die L. I’m not as healthy now as I 5 4 3 2 1 • CH used to be M. I worry about my health 5 4 3 2 1 • HW more than other people worry about their health N. When I’m sick, I try to just 5 4 3 2 1 keep going as usual O. My body seems to resist 5 4 3 2 1 • RI illness very well (continued) 255 Exhibit 5.12 (continued) Definitely Mostly Don’t Mostly Definitely true true know false false GHRI Subscale* P. Getting sick once in a while 5 4 3 2 1 SO is a part of my life Q. I’m as healthy as anybody I 5 4 3 2 1 • CH know R. I think my health will be 5 4 3 2 1 • HO worse in the future than it is now S. I’ve never had an illness that 5 4 3 2 1 • PH lasted a long period of time T. Others seem more con- 5 4 3 2 1 HW cerned about their health than I am about mine U. When I’m sick, I try to keep 5 4 3 2 1 it to myself V. My health is excellent 5 4 3 2 1 • CH W. I expect to have a very 5 4 3 2 1 • HO healthy life X. My health is a concern to 5 4 3 2 1 HW my life Y. I accept that sometimes I’m 5 4 3 2 1 SO just going to be sick Z. I have been feeling bad lately 5 4 3 2 1 • CH AA. It doesn’t bother me to go 5 4 3 2 1 to a doctor BB. I have never been seriously ill 5 4 3 2 1 • PH CC. When there is something go- ing around, I usually catch it 5 4 3 2 1 • RI DD. Doctors say that I am now in 5 4 3 2 1 • CH poor health EE. When I think I am getting 5 4 3 2 1 sick, I fight it FF. I feel about as good now as I 5 4 3 2 1 • CH ever have During the past 3 months, how much has your health worried or concerned you? (circle one) A great deal . . . . . . . . . . . . . . . . . . 1 Somewhat . . . . . . . . . . . . . . . . . . . . 2 A little . . . . . . . . . . . . . . . . . . . . . . 3 Not at all . . . . . . . . . . . . . . . . . . . . 4 *Subscale labels: CH = Current Health; PH = Prior Health; HO = Health Outlook; RI = Resistance to Illness; HW = Health Worry/ Concern; SO = Sickness Orientation. Derived from Davies AR, Sherbourne CD, Peterson JR, Ware JE, Jr. Scoring manual: adult health status and patient satisfaction measures used in Rand’s Health Insurance Experiment. Santa Monica, California: RAND Corporation, 1988: Figures 5.1 and 5.2. 256  Psychological Well-being 257 Table 5.3 Reliability of Health Perceptions Questionnaire Scales Alpha One-year stability Scale (N = 1,790) (N = 4,700) (N = 1,200) Current health (9 items) 0.91 0.88 0.58 Current health (4 items) (NA) 0.81 (NA) Prior health 0.73 0.65 0.67 Health outlook 0.75 0.73 0.59 Resistance to illness 0.71 0.70 0.65 Health worry/concern 0.60 0.64 0.50 Sickness orientation 0.59 0.53 0.55 General Health Rating Index (NA) 0.89 0.67 pha coefficients from four field tests (4, p42; 6, with variables defining favorable health states, Table 4). The second column gives alpha coeffi- and negative relationships with variables defin- cients reported by Davies et al., using data from ing poor health. Coefficients typically fell in the main HIE. The right column gives one year the range of 0.30 to 0.60 (4, p52). Correla- test-retest figures, drawn from the Seattle center tions with health behaviors (e.g., visits to the of the HIE (7, p107). doctor) were lower than those with health sta- Reliability results were slightly lower in tus variables. Associations between age and poorer people, in those with less education, and health perceptions were generally negative, in older people (4, p43; 7, p97) but nonetheless whereas greater income and higher education remain adequate for group comparisons. were generally positively correlated with gen- Test-retest reliability figures for the GHRI at eral health perceptions (5, p48). Connelly et al. intervals of one, two, and three years for adults controlled for differences in level of physical in HIE were 0.67, 0.59, and 0.56, respectively health and found significant associations be- (1, p185). tween health perceptions scores and anxiety, depression, worry, and health care behaviors, Validity such as numbers of physician visits and tele- The HPQ was designed to measure six postu- phone calls to the physician (10). Indeed, 5% lated aspects of health perceptions. Factor of office visits were attributed to poor health analyses confirmed the existence of six main perceptions alone in the absence of any med- factors and indicated that each scale con- ical indication for the visit (10, pS107). tributed some unique information about health The GHRI correlated 0.46 with the Quality perceptions (6). Furthermore, the results were of Well-Being Scale (QWB) and 0.52 with the similar across different samples. The Resistance Sickness Impact Profile (SIP) (8, Table 4). Con- to Illness subscale showed the least clear sepa- vergent validity correlations suggest that the ration, and a higher order factor analysis GHRI offers a general indicator of self-perceived showed that it overlapped with several other health that shows relatively weak associations factors (4, p49). with more objective indicators. Scores on the Ware and others have presented numerous GHRI correlated 0.71 with the question “How correlations between the subscales and crite- would you rate your overall health: Excellent, rion variables such as disability days, number Good, Fair, or Poor?” The equivalent correla- of chronic problems, pain level, and psycholog- tions for the QWB and the SIP were lower, at ical well-being (4; 6, Tables 6 to 8; 9, 0.43 and 0.51, respectively (8, Table 4). Con- pp49–60). Most coefficients were as hypothe- versely, the QWB and SIP showed generally sized; for example, current health and prior stronger correlations than the GHRI with other health showed significant positive relationships indicators, such as numbers of health problems 258 Measuring Health recorded in the patient’s chart, use of care, level of disability, a mental health indicator, and em- References ployment status (8, Table 5). (1) Ware JE, Jr. General health rating index. In: Wenger NK, Mattson ME, Furberg CD, et al., eds. Asssessment of quality of life in Alternative Forms clinical trials for cardiovascular disease. A modified version of the HPQ was developed New York: LeJacq, 1984:184–188. for the Medical Outcomes Study and is de- (2) Ware JE, Jr., Manning WG, Jr., Duan N, et scribed by Stewart et al. (3, Table 8-8). This con- al. Health status and the use of ambulatory tains 36 items that cover the same dimensions as mental health services. Am Psychol the original scale, except that sickness orienta- 1984;39:1090–1100. tion was replaced by a health distress scale, and (3) Stewart AL, Hays RD, Ware JE, Jr. Health a new energy/fatigue scale was developed. perceptions, energy/fatigue, and health distress measures. In: Stewart AL, Ware JE, A general health perceptions measurement Jr., eds. Measuring functioning and well- for children has been described by Eisen et al. being: the Medical Outcomes Study (11). Seven questions cover current health, resis- approach. Durham, NC: Duke University tance to illness, and prior health. Press, 1992:143–172. (4) Ware JE, Jr., Davies-Avery A, Donald CA. Conceptualization and measurement of Reference Standards health for adults in the Health Insurance Means, standard deviations, and complete fre- Study. Vol. V, General health perceptions. quency tables are available for each subscale and Santa Monica, CA: RAND Corporation, for the overall GHRI from Appendix A of the 1978. publication by Davies et al. (7, pp184–194). (5) Brook RH, Ware JE, Jr., Davies-Avery A, et al. Overview of adult health status mea- sures fielded in RAND’s Health Insurance Commentary Study. Med Care 1979;17(suppl):1–131. The HPQ is an important extension of the (6) Ware JE, Jr. Scales for measuring general single-item measures with the format, “How health perceptions. Health Serv Res would you rate your health today: Good, Fair, or 1976;11:396–415. Poor?” It was carefully developed and has been (7) Davies AR, Sherbourne CD, Peterson JR, et widely tested in large national studies. al. Scoring manual: adult health status and patients satisfaction measures used in Evidence for reliability and validity is prom- RAND’s Health Insurance Experiment. Santa ising. The hypothesized subscores are largely Monica, CA: RAND Corporation (Pub. No. supported by empirical evidence and, although N-2190-HHS), 1988. all six scales are related positively, the correla- (8) Read JL, Quinn RJ, Hoefer MA. Measur- tions among them are low enough to suggest ing overall health: an evaluation of three that they tap separate dimensions (5, p35). The important approaches. J Chronic Dis high stability over time suggests that it may be 1987;40(suppl 1):S7–S21. more suitable as a trait indicator than as an out- (9) Davies AR, Ware JE, Jr. Measuring health come measure that is sensitive to changes. The perceptions in the Health Insurance correlation of 0.71 with an overall self-rating Experiment. Santa Monica, CA: RAND question both confirms the general nature of the Corporation (Publication No.R-2711- HHS), 1981. GHRI and suggests that the single question may (10) Connelly JE, Philbrick JT, Smith GR, et al. offer an effective alternative. Most concurrent Health perceptions of primary care patients validity correlations were drawn from survey and the influence on health care utilization. measurements of disability or health care use; it Med Care 1989;27(suppl):S99–S109. would be valuable to see correlations between (11) Eisen M, Ware JE, Jr., Donald CA, et al. the HPQ subscales and other established health Measuring components of children’s health indexes. status. Med Care 1979;17:902–921.  Psychological Well-being 259 identified on the basis of checking any 12 or The General Health Questionnaire more of the 60 symptoms included, and the re- (David Goldberg, 1972) sults express the likelihood of psychiatric disor- der. Purpose The General Health Questionnaire (GHQ) is a Description self-administered screening instrument designed The GHQ was designed for use in population to detect current diagnosable psychiatric disor- surveys, in primary medical care settings, or ders. The GHQ may be used in surveys or in clin- among general medical outpatients (3). It was ical settings to identify potential cases, leaving the meant to be a first-stage screening instrument task of diagnosing actual disorder to a psychiatric for psychiatric illness that could then be verified interview (1). and diagnosed. The questions ask whether the respondent has recently experienced a particular Conceptual Basis symptom (e.g., abnormal feelings or thoughts) The GHQ is designed to identify two main or type of behavior. Emphasis is on changes in classes of problem: “inability to carry out one’s condition, not on the absolute level of the prob- normal ‘healthy’ functions, and the appearance lem, so items compare the present state to the of new phenomena of a distressing nature” (2). person’s normal situation with responses rang- It focuses on breaks in normal functioning ing from “less than usual” to “much more than rather than on life-long traits; therefore it only usual” (3). The questionnaire begins with rela- covers personality disorders or patterns of ad- tively neutral questions and leads to the more justment where these are associated with dis- overtly psychiatric items toward the end. The tress. Nor was the GHQ intended to detect questions were drawn from existing instruments severe illness such as schizophrenia or psychotic or were created especially for this application, depression, although subsequent experience with and those that discriminated between severely ill the scale suggests that these conditions are de- and mildly ill psychiatric patients and mentally tected (3). healthy people were retained. Details of the item The GHQ was designed to cover four ele- selection procedure are given by Goldberg (1; ments of distress: depression, anxiety, social im- 3). The GHQ is normally completed by the pa- pairment, and hypochondriasis (chiefly indicated tient; Goldberg reports that more than 95% of by organic symptoms) (1). Subsequent empirical respondents could do this and were remarkably analyses of the factor structure of the GHQ have frank in admitting symptoms. largely confirmed this coverage (4). Goldberg The main version of the GHQ, shown in Ex- suggests that his approach to psychiatric disor- hibit 5.13, contains 60 items, and Goldberg rec- der is close to the lowest level of the hierarchy of ommends using this version (where possible) mental illness outlined by Foulds and Bedford, because of its superior validity. However, he pro- which they term dysthymic states. “An individ- posed alternative, shorter versions for use where ual falling into any of these states might be said all 60 questions could not be asked. These in- to be disturbed, emotionally stirred up, altered clude 30-, 20-, and 12-item abbreviations, and in this respect from his normal self ” (3). Such the GHQ-28 or “Scaled GHQ” that contains individuals will be prone to minor somatic four scales derived from factor analyses (2). The symptoms and may show outwardly observable items in the abbreviated versions are shown in changes in social behaviors. Exhibit 5.14. Note that the exhibits give the Although the GHQ does cover separate types original wording of the items; some may be re- of distress, it was not intended to distinguish phrased using a U.S. idiom: these are shown un- among psychiatric disorders or to be used in der Alternative Forms. The GHQ-28 provides making diagnoses. No assumptions were made four scores, measuring somatic symptoms, anxi- concerning a hierarchy among the symptoms in- ety and insomnia, social dysfunction, and severe cluded in the questionnaire: probable cases are depression. It is intended for studies in which an  Exhibit 5.13 The General Health Questionnaire (60-Item Version) Please read this carefully: We should like to know if you have had any medical complaints, and how your health has been in general, over the past few weeks. Please answer ALL the questions on the following pages simply by underlining the answer which you think most nearly applies to you. Remember that we want to know about present and recent complaints, not those that you had in the past. It is important that you try to answer ALL the questions. Thank you very much for your co-operation. Have you recently: 1. been feeling perfectly well and in good Better than usual Same as usual Worse than usual Much worse than usual health? 2. been feeling in need of a good tonic? Not at all No more than usual Rather more than usual Much more than usual 3. been feeling run-down and out of sorts? Not at all No more than usual Rather more than usual Much more than usual 4. felt that you are ill? Not at all No more than usual Rather more than usual Much more than usual 5. been getting any pains in your head? Not at all No more than usual Rather more than usual Much more than usual 6. been getting a feeling of tightness or Not at all No more than usual Rather more than usual Much more than usual pressure in your head? 7. been able to concentrate on whatever Better than usual Same as usual Less than usual Much less than usual 260 you’re doing? 8. been afraid that you were going to collapse Not at all No more than usual Rather more than usual Much more than usual in a public place? 9. been having hot or cold spells? Not at all No more than usual Rather more than usual Much more than usual 10. been perspiring (sweating) a lot? Not at all No more than usual Rather more than usual Much more than usual 11. found yourself waking early and unable to Not at all No more than usual Rather more than usual Much more than usual get back to sleep? 12. been getting up feeling your sleep hasn’t Not at all No more than usual Rather more than usual Much more than usual refreshed you? 13. been feeling too tired and exhausted even Not at all No more than usual Rather more than usual Much more than usual to eat? 14. lost much sleep over worry? Not at all No more than usual Rather more than usual Much more than usual 15. been feeling mentally alert and wide awake? Better than usual Same as usual Less alert than usual Much less alert 16. been feeling full of energy? Better than usual Same as usual Less energy than usual Much less energetic 17. had difficulty in getting off to sleep? Not at all No more than usual Rather more than usual Much more than usual 18. had difficulty in staying asleep once you Not at all No more than usual Rather more than usual Much more than usual are off?  19. been having frightening or unpleasant Not at all No more than usual Rather more than usual Much more than usual dreams? 20. been having restless, disturbed nights? Not at all No more than usual Rather more than usual Much more than usual 21. been managing to keep yourself busy and More so than usual Same as usual Rather less than usual Much less than usual occupied? 22. been taking longer over the things you do? Quicker than usual Same as usual Longer than usual Much longer than usual 23. tended to lose interest in your ordinary Not at all No more than usual Rather more than usual Much more than usual activities? 24. been losing interest in your personal Not at all No more than usual Rather more than usual Much more than usual appearance? 25. been taking less trouble with your clothes? More trouble than usual About same as usual Less trouble than usual Much less trouble 26. been getting out of the house as much as More than usual Same as usual Less than usual Much less than usual usual? 27. been managing as well as most people Better than most About the same Rather less well Much less well would in your shoes? 28. felt on the whole you were doing things Better than usual About the same Less well than usual Much less well well? 261 29. been late getting to work, or getting started Not at all No later than usual Rather later than usual Much later than usual on your housework? 30. been satisfied with the way you’ve carried More satisfied About same as usual Less satisfied than usual Much less satisfied out your task? 31. been able to feel warmth and affection for Better than usual About same as usual Less well than usual Much less well those near to you? 32. been finding it easy to get on with other Better than usual About same as usual Less well than usual Much less well people? 33. spent much time chatting with people? More time than usual About same as usual Less than usual Much less than usual 34. kept feeling afraid to say anything to people Not at all No more than usual Rather more than usual Much more than usual in case you made a fool of yourself? 35. felt that you are playing a useful part in More so than usual Same as usual Less useful than usual Much less useful things? 36. felt capable of making decisions about More so than usual Same as usual Less so than usual Much less capable things? 37. felt you’re just not able to make a start on Not at all No more than usual Rather more than usual Much more than usual anything? (continued)  Exhibit 5.13 (continued) 38. felt yourself dreading everything that you Not at all No more than usual Rather more than usual Much more than usual have to do? 39. felt constantly under strain? Not at all No more than usual Rather more than usual Much more than usual 40. felt you couldn’t overcome your difficulties? Not at all No more than usual Rather more than usual Much more than usual 41. been finding life a struggle all the time? Not at all No more than usual Rather more than usual Much more than usual 42. been able to enjoy your normal day-to-day More so than usual Same as usual Less so than usual Much less than usual activities? 43. been taking things hard? Not at all No more than usual Rather more than usual Much more than usual 44. been getting edgy and bad-tempered? Not at all No more than usual Rather more than usual Much more than usual 45. been getting scared or panicky for no good Not at all No more than usual Rather more than usual Much more than usual reason? 46. been able to face up to your problems? More so than usual Same as usual Less able than usual Much less able 47. found everything getting on top of you? Not at all No more than usual Rather more than usual Much more than usual 48. had the feeling that people were looking at Not at all No more than usual Rather more than usual Much more than usual you? 49. been feeling unhappy and depressed? Not at all No more than usual Rather more than usual Much more than usual 50. been losing confidence in yourself? Not at all No more than usual Rather more than usual Much more than usual 262 51. been thinking of yourself as a worthless Not at all No more than usual Rather more than usual Much more than usual person? 52. felt that life is entirely hopeless? Not at all No more than usual Rather more than usual Much more than usual 53. been feeling hopeful about your own More so than usual About same as usual Less so than usual Much less hopeful future? 54. been feeling reasonably happy, all things More so than usual About same as usual Less so than usual Much less than usual considered? 55. been feeling nervous and strung-up all the Not at all No more than usual Rather more than usual Much more than usual time? 56. felt that life isn’t worth living? Not at all No more than usual Rather more than usual Much more than usual 57. thought of the possibility that you might Definitely not I don’t think so Has crossed my mind Definitely have make away with yourself? 58. found at times you couldn’t do anything Not at all No more than usual Rather more than usual Much more than usual because your nerves were too bad? 59. found yourself wishing you were dead and Not at all No more than usual Rather more than usual Much more than usual away from it all? 60. found that the idea of taking your own life Definitely not I don’t think so Has crossed my mind Definitely has kept coming into your mind? Reproduced from Goldberg DP. The detection of psychiatric illness by questionnaire. London: Oxford University Press, 1972: Appendix 8. With permission. Copyright © General Practice Research Unit 1968 (de Crespigny Park, London, S.E. 5). Exhibit 5.14 Abbreviated Versions of the General Health Questionnaire Note: Using the item numbers from Exhibit 5.13 the contents of the shortened versions are as follows: GHQ-12 7. able to concentrate 42. enjoy normal activities 14. lost sleep over worry 46. face up to problems 35. playing a useful part 49. unhappy and depressed 36. capable of making decisions 50. losing confidence in yourself 39. constantly under strain 51. thinking of yourself as worthless 40. couldn’t overcome difficulties 54. feeling reasonably happy GHQ-20 In addition to the 12 items above, the 20-item version includes: 21. busy and occupied 43. taking things hard 26. getting out of house as usual 47. everything on top of you 28. doing things well 55. nervous and strung-up 30. satisfied with carrying out task 58. nerves too bad Note: Item 30 is replaced by item 15 for use in the United States (3, p19). GHQ-30 In addition to the 20 items above, the 30-item version includes: 20. restless, disturbed nights 41. life a struggle all the time 27. managing as well as most people 45. scared or panicky 31. feel warmth and affection 52. life entirely hopeless 32. easy to get on with others 53. hopeful about your future 33. much time chatting 56. life not worth living Note: Item 33 is replaced by item 16 for use in the United States (3, p19). GHQ-28 The 28-item version is as follows: Scale A Scale B Somatic Symptoms Anxiety and Insomnia 1. feeling perfectly well 14. lost sleep over worry 2. in need of a good tonic 18. difficulty staying asleep 3. run down 39. constantly under strain 4. felt that you are ill 44. edgy and bad-tempered 5. pains in head 45. scared or panicky 6. pressure in your head 47. everything on top of you 9. hot or cold spells 55. nervous and strung-up Scale C Scale D Social Dysfunction Severe Depression 21. busy and occupied 51. thinking of yourself as worthless 22. taking longer over things 52. life entirely hopeless 28. doing things well 56. life not worth living 30. satisfied with carrying out task 57. make away with yourself 35. playing a useful part 58. nerves too bad 36. capable of making decisions 59. dead and away from it all 42. enjoy normal activities 60. idea of taking your life Adapted from Goldberg DP. The detection of psychiatric illness by questionnaire. London: Oxford University Press, 1972: Appendix 6. 263 264 Measuring Health investigator requires more information than is and 0.59 for the CGHQ (7, Table IV). Sensitivity provided by a single severity score (2). There is and specificity may also be improved (7, p57; only a partial overlap between the GHQ-28 and 10, Table III). An empirical comparison of three the GHQ-30, which share just 14 items. The scoring approaches suggested that CGHQ ap- 30-, 20-, and 12-item versions are balanced in proach was best, followed by the Likert scoring terms of “agreement sets”—that is, half of the and then the binary GHQ scoring method (8, questions are worded positively and half nega- Table 2). Meanwhile, other reviews have found tively. The shortened versions also discard 12 little difference between the GHQ and the questions that were answered positively by phys- CGHQ (11, p1012; 12; 13, p94). Goldberg, for ically ill patients (1). It takes six to eight minutes example, compared the Likert, the standard to complete the GHQ-60 and three to four min- scoring (0-0-1-1), and the CGHQ in a study un- utes for the GHQ-30 (3). dertaken in 15 countries. The area under the Items may be scored using conventional 0-1- ROC curve was fractionally higher for the stan- 2-3 Likert scores for the response categories dard scoring (0.88 versus 0.86 for the CGHQ shown in Exhibit 5.13. Alternatively, a two- and 0.85 for the Likert for the GHQ-12), but point score rates each problem as present or ab- there was less difference for the GHQ-28 (14, sent, ignoring the intensity categories (3). In this Table 4). approach, known as “the GHQ score,” replies Scores can be interpreted as indicating the are coded 0-0-1-1. Goldberg found little advan- severity of psychological disturbance on a con- tage to the Likert scoring, with correlations be- tinuum; as a screening test, the score expresses tween the two scoring methods between 0.92 the probability of a patient’s needing psychiatric and 0.94 (5, Table 3) and so recommended the care. Any 12 positive answers on the GHQ-60 binary system (2). Differential weights for each identify a probable case (at the cutting-point, the item held little advantage and were discarded in probability of being a case is 0.5). The threshold the interests of simplicity, although a variant of scores may have to be altered depending on the this approach was revived by Surtees and Miller expected prevalence of disorder and according to (6). A scoring approach that has, however, the purpose of the study: prevalence surveys ver- gained currency was proposed by Goodchild and sus detection of severe disorders, for example. Duncan-Jones (originally for the GHQ-30). Thus, cutting-points of 9/10, 10/11, or 11/12 They argued that ratings should cover changes have been used for the GHQ-60 (3). Unfortu- from what is normal in the population rather nately, cutting-points for the other versions have than what is normal for this respondent. Their varied considerably. In Goldberg’s initial studies, scoring system treats the response “no more they were five positive answers for the GHQ-30 than usual” as an indicator of a health problem and GHQ-28, and four for the GHQ-20, for negatively worded items (e.g., “been having whereas a cutting-point of 2/3 was commonly restless, disturbed nights”), which are therefore used for the GHQ-12 (5). However, other com- scored 0-1-1-1. The “no more than usual” re- mon choices for the GHQ-12 include 1/2, 3/4 sponse for positive items was taken as indicating and 4/5 (14, Table 1), and on reviewing 17 stud- a healthy response, so these items are scored us- ies, Goldberg recorded a range from 0/1 to 5/6 ing the conventional 0-0-1-1 (7, p56; 8, p326). (15, p916). The optimal threshold from the This scoring method is termed the Corrected Goldberg et al. study in 15 cities was 3/4, giving GHQ, or CGHQ (9). The alternative scoring an area under the ROC curve of 0.95 (15, Table methods have been compared in several studies 3). and no clear conclusion has been reached. Some authors have found the CGHQ to be an im- Reliability provement: it may reduce the floor effect of the The test-retest coefficient after six months was overall GHQ score and provide less skewed re- 0.90 (N = 20) when the stability of the patient’s sponses (7; 9, Table III). The correlation with the condition was confirmed by repeating a standard Zung depression scale was 0.50 for the GHQ psychiatric examination (3, p15). For another 65  Psychological Well-being 265 patients who judged their own condition as hav- 8). Results from studies in England, Australia, ing remained “about the same,” the retest coeffi- and Spain were consistent, with correlations be- cient was 0.75 (3). tween the two scales ranging from 0.76 to 0.81. Split-half reliability on the 60-item version Sensitivity values ranged from 81 to 91%; speci- was 0.95 for 853 respondents (3). The equiva- ficity results in four of the studies ranged be- lent value for the GHQ-30 was 0.92; for the tween 88 and 94%, whereas in the remaining GHQ-20, 0.90, and for the GHQ-12, 0.83 (1, study it was 73% (3, Table 4.1). Subsequent Table 27). Chan and Chan reported an alpha of studies have also given comparable results: 0.85 for the GHQ-30 (16, Table 1). Hobbs et al. reported a correlation of 0.72 with Inter-rater reliability for 12 interviews the CIS and sensitivity results between 84 and showed a disagreement on only 4% of symptom 96% at specificities from 70 to 91% (20, Table scores (17, p 410). IV). Slightly lower figures were obtained by Nott Internal consistency estimates include split- and Cutts with women postpartum (17). Some- half figures of 0.95 for the GHQ-60, 0.92 for what less adequate results were obtained by Ben- the GHQ-30, 0.90 for the GHQ-20, and 0.83 jamin et al., who applied the GHQ-60 to 92 for the GHQ-12 (18, p75). Alpha coefficients women aged between 40 and 49 years (22). for the GHQ-12 ranged from 0.82 to 0.90 in They obtained a sensitivity of 54.5% at a speci- four studies (18, p75; 14, p192). ficity of 91.5%, and a Spearman correlation of Reliability calculated using LISREL modeling 0.63 with the CIS (22). On further examination, was 0.74 for the GHQ-12 (19, p862). the false-negative findings proved to be in those who had long-standing mental disorders (see Validity Commentary). Comparisons of the GHQ and The GHQ is among the most thoroughly tested the Present State Examination were made in En- of all health measures. Validation studies have gland and India, giving correlations between been undertaken in many different countries and 0.71 and 0.88 (3, Table 4.1). most have used directly comparable procedures. In a metaanalysis of studies that evaluated de- Because of the size of this literature, we have pression screening instruments, Mulrow et al. been strictly selective in our presentation; a re- showed the GHQ to have almost equal sensitiv- view by Vieweg and Hedlund presented several ity to the BDI, but better specificity, in detecting pages of validity results (18, pp75–78). major depression (23, Figures 2 and 3). The va- The factor structure of the GHQ was origi- lidity results of the GHQ may be compared with nally studied by Goldberg and used as a basis for those obtained using rival screening tests. Gold- abbreviating the scale. Several analyses produced berg provided a table summarizing the results relatively consistent factors: somatic symptoms, from which we drew the comparisons shown in sleep disturbance (sometimes combined with Table 5.4 (1, Table 32). Comparisons with the anxiety), social dysfunction, and severe depres- Hospital Anxiety and Depression Scale (HADS) sion (3; 4, Table 5). In subsequent studies, suggest little difference overall, although with Hobbs et al. extracted three factors, covering de- the suggestion that the HADS may perform bet- bility (failure to cope), depression, and somatic ter in detecting anxiety than the GHQ (8; 24). In symptoms (20). The area under the ROC curve other studies, the GHQ-28 performed somewhat for the depressive factor scores compared with a less well than the HADS, but somewhat better clinical diagnosis of depression was 0.75; the ar- than the Rotterdam Symptom Checklist (25, eas under the curve for other factors ranged from Table 1). 0.50 (sleep) to 0.70 (social function) (21, Table Screening tests should not be influenced by 5). age, sex, or educational status of the respon- Goldberg provided a table summarizing five dents. Goldberg et al. studied the validity of the studies that compared the GHQ-60 with the GHQ-28 and GHQ-12 in 15 countries and standardized psychiatric interview that he devel- found no significant differences in the validity oped, the Clinical Interview Schedule (CIS) (3; results by age, sex, education, or in the contrast 266 Measuring Health Table 5.4 A Comparison of the Sensitivity and Specificity Results for Four Versions of the General Health Questionnaire (GHQ) General practice patients Hospital outpatients Sensitivity % Specificity % Sensitivity % Specificity % GHQ-60 95.7 87.8 80.6 93.3 GHQ-30 91.4 87.0 64.5 91.6 GHQ-20 88.2 86.0 64.5 96.7 GHQ-12 93.5 78.5 74.2 95.0 Note: The patients completed the GHQ-60. Validity estimates for the shortened versions were calculated by analyzing subsets of questions from the 60-item version. between developing and developed countries (14, the sensitivity was 78%. Using statistical manip- pp195–196). In another study, however, female ulations, Tarnopolsky estimated that the sensi- respondents tended to show higher scores; there tivity would fall to 54% as the ratio of high to was little association between age and GHQ low scoring cases falls to 22%. Kendall’s tau scores, and there was a significant tendency for correlations with the interview schedule ranged respondents of lower social classes to have from 0.34 to 0.45 (26, Table V). The correlation higher scores (3). of the GHQ-30 with the Hopkins Symptom Turning to the abbreviated versions of the Checklist of physical and psychological symp- GHQ, perhaps the salient finding is that the ab- toms was 0.78 (27, p65); the GHQ showed breviated versions perform remarkably well slightly higher sensitivity and specificity values compared with the full 60-item version. Correla- (3). The RAND Mental Health Inventory (MHI) tions among the three abbreviated GHQ scales had higher sensitivity and specificity than the fell between 0.85 and 0.97 (5, Table 3). We pre- GHQ-30: the area under the ROC curve was sent comparative data on the sensitivity and 0.76 for the MHI and 0.68 for the GHQ in specificity of four versions of the GHQ in our detecting any disorder; the figures were 0.76 and Table 5.5 (adapted from 1, Table 27). Slightly 0.73 for affective disorders and 0.70 compared lower figures were reported by Banks (5, Table to 0.65 for anxiety (28, Table 3). 2), whereas a similar pattern of slightly declining Several studies of the factor structure of the validity for the abbreviated versions was found 30-item version give comparable results. One by Clarke et al. (8, Table 3). A comparison of identified factors covering depression and anxi- the GHQ-28, GHQ-30, and GHQ-12 adminis- ety, insomnia and lack of energy, social function- tered to young respondents showed the GHQ-28 ing, and anhedonia (unhappiness) (27). Cleary et was superior when compared with the Present al. reported similar findings from analyses of State Examination (5). 1,072 respondents in Wisconsin (29). Berwick et al. identified six factors (30) whereas the Hup- ghq-30. Four studies of the 30-item scale have pert et al. study of 6,317 respondents in Great shown sensitivity values between 71 and 91%, Britain identified five factors, covering anxiety, with specificities in the same range (3, p11). feelings of incompetence, depression, difficulty Tarnopolsky et al. examined the sensitivity and in coping, and social dysfunction (31, p182). A specificity of the 30-item GHQ compared with Chinese study also identified five factors: anxi- the CIS, producing results that appear to be at ety, inadequate coping, depression, insomnia, odds with those already reviewed. The sensitiv- and social dysfunction (16). ity results varied according to the prevalence of the disorder in the study population. When half ghq-28. Scales in the GHQ-28 (selected via of the population scored above the cutting-point, factor analysis) measure somatic symptoms,  Psychological Well-being 267 Table 5.5 Comparison of the Validity of the General Health Questionnaire (GHQ) with That of Other Scales Sensitivity % Specificity % Overall misclassification % GHQ-60 95.7 87.8 10.3 GHQ-30 85.0 79.5 19.1 Cornell Index 73.5 81.7 17.8 HOS (Macmillian) 75–84 54–68 22–40 22-Item Scale (Langner) 73.5 81.7 17.8 anxiety and insomnia, social dysfunction, and score correlated 0.36 with the Clinical Anxiety severe depression. These scales are not indepen- Scale; the depression score correlated 0.66 with dent of each other: correlations range from 0.33 the Montgomery-Åsberg Depression Rating to 0.58 (2, Table 9). Scale (35, p264). Goldberg et al. compared the GHQ-28 against diagnoses based on the Composite In- ghq-12. Goldberg et al. summarized the results ternational Diagnostic Interview (CIDI) in a from a number of validation studies of the study in 15 countries. The mean sensitivity (for GHQ-12. The median sensitivity drawn from 17 all CIDI diagnoses) was 79.7% at a mean speci- studies was 83.7% and the median specificity ficity of 79.2%. The area under the ROC curve was 79.0% (14, Table 1). The area under the was 0.88 (14, Table 3). Other figures include a ROC curve was estimated in studies in 15 sensitivity of 85.6% at a specificity of 86.8% countries (N = 5,438) in a comparison against (3, p22). Studying general practice patients in the CIDI rating of current mental status. The Sydney, Australia, Tennant reported sensitivities mean area under the curve was 0.88 (range, ranging from 86.6 to 90% and specificities 0.83–0.95). Overall sensitivity was 83.4% and ranging from 90 to 94.4% (32, Table 1). Rather specificity 76.3% (14, p194). These figures were lower figures were obtained comparing GHQ- very comparable to those obtained for the GHQ- 30 scores with the Schedule for Affective Disor- 28. ders and Schizophrenia—a sensitivity of 68% The area under the ROC curve for the and a specificity of 81% (29, Table 3). The GHQ-12 was 0.87 in a Brazilian sample; the re- GHQ-28 was found to be sensitive to depres- sult was comparable with that obtained using sion even when used in patients with dementia Harding’s 20-item Self Report Questionnaire (10). The area under the ROC curve for the (36). Using the GHQ-12 to identify depression, GHQ-28 was 0.88 in a sample of inpatients the area under the ROC curve was 0.81, com- with neurological disorders (33, p550). The va- pared with 0.92 for the Hospital Anxiety and lidity of the GHQ-28 was reviewed by Gold- Depression scale (37, p403). A factor analysis berg and Hillier. The correlation of the overall of the GHQ-12 from an Australian sample score with the CIS was 0.76 (2, Table 5). A cor- identified three factors: anhedonia and sleep relation of 0.73 was obtained with the CIS de- disturbance, social performance, and loss of pression rating and of 0.67 with the anxiety confidence (4). The evidence available suggests rating. Using a cutting-point of 4/5, the sensitiv- that several of the GHQ factors are stable ity was 88%, the specificity 84.2%, and the across samples and among different versions of overall misclassification rate was 14.5% (2). the questionnaire. Poorer results were obtained in a small study of The GHQ-12 has been compared with other 56 pain patients; the Spearman correlation with scales, and appears to correlate highly with both the CIS was 0.47 and sensitivity was 71% at a measures of well-being and measures of distress. specificity of 63% (34, p199). In a mixed sam- The correlation with a single-item “delightful- ple of hospital outpatients, the GHQ-28 anxiety terrible” scale measuring “your life as a whole” 268 Measuring Health was −0.50; the correlation with Bradburn’s posi- oped. It was founded on a clear conceptual ap- tive affect scale was −0.30, and that with the proach, the initial item selection and item analy- negative affect scale, 0.40. The correlation with ses are fully documented, and the questions have Beck’s Depression Inventory was 0.49, and that not been revised by subsequent users. Goldberg’s with Spielberger’s state anxiety scale was 0.38 book (1) is a model of clarity and thoroughness; (38, Table 1). The correlation with the 5-item unfortunately, the manual of the GHQ does not Mental Health Inventory was 0.64, and kappa contain copies of the questionnaire (3). The vali- agreement on dichotomous scoring the two dation studies have been thorough and exten- scales (MHI 72/73 and GHQ ≥ 2) was 0.49 (39). sive; they have used comparable approaches and Both scales predicted physician visits and psy- have consistently indicated a high degree of va- chiatric consultations. Sensitivity to change was lidity, markedly higher than that of rival meth- reviewed by Ormel et al., and was found to be ods. The scale has been tested in numerous good (11). countries and shows remarkably consistent va- lidity results. The validation reports offer inter- Alternative Forms esting insights into test development. To give just The GHQ was developed in England, but with one example, Goldberg studied the validity of the aim of making comparative studies of psy- the GHQ-12 in 15 cities (and 11 languages) and chiatric illness in England and the United States. showed that the relative sensitivity and speci- Several of the items have been rephrased for U.S. ficity of the items varied from place to place. use (1; 3): However, there was no item that could be dis- carded because each proved to be highly valid 2. _____ been feeling in need of some medi- somewhere in the world (51, Table 5). Hence, an cine to pick you up? even shorter version might be feasible, but only 18. _____ had difficulty staying asleep? if it were to be used in a single location. 27. _____ been managing as well as most Most criticisms that have been raised over the people would in your place? GHQ reflect limitations imposed by the deliber- 47. _____ found everything getting too much ate design of the instrument. The response cate- for you? gories ask whether each symptom is worse than 55. _____ been feeling nervous and uptight usual, and if a person has suffered a symptom (or hung up) all the time? for a long time and has come to consider it 57. _____ thought of the possibility that you “usual,” the scale will not identify this as a might do away with yourself? problem. Benjamin et al. viewed this as a limita- tion of the scale, although Goldberg developed The GHQ has been used across the world the GHQ to measure changes in a person’s con- and versions exist in a large number of lan- dition and not the absolute level of the problem guages; many of these versions have been vali- (22). It screens, therefore, for acute rather than dated. Validated examples include Italian (40), chronic conditions. This issue may have been re- Cambodian (41), Mexican Spanish (42), Japan- solved by the “corrected GHQ” scoring proce- ese (43), Chinese (16; 44–46), Turkish (47; 48) dure proposed by Goodchild and Duncan-Jones and Urdu (49). Consistency of experience with (7). the GHQ in India and in Brazil led Sen and Mari There has also been some debate over the to conclude that the GHQ taps “an inner core of suitability of the items in the GHQ-60 that re- human suffering which can be reliably detected flect physical symptoms (“e.g., feeling of tight- by suitably modified instruments developed in ness or pressure in your head,” “perspiring a the West” (50, p277). lot”). The physical items were excluded from the abbreviated versions of the GHQ because Commentary they produced several false-positive responses, The GHQ offers a preeminent example of how although the problem seems to be far less seri- health measurement methods should be devel- ous than it is with Macmillan’s Health Opinion  Psychological Well-being 269 Survey. Tennant, however, noted that “all false patients. Int J Psychiatry Med positives were subjects with substantial physi- 1993;23:323–337. cal illness” (32). Other studies of patients with (9) Huppert FA, Gore M, Elliott BJ. The value physical illness also suggest that the GHQ iden- of an improved scoring system (CGHQ) tifies false-positive responses (34). The diffi- for the General Health Questionnaire in a representative community sample. Psychol culty of using somatic questions to screen for Med 1988;18:1001–1006. psychiatric disorders may still not have been re- (10) O’Riordan TG, Hayes JP, O’Neill D, et al. solved. The main dissonant note in the valida- The effect of mild to moderate dementia on tion studies comes from Tarnopolsky, who the Geriatric Depression Scale and on the obtained lower sensitivity rates than those ob- General Health Questionnaire. Age Ageing tained by Goldberg (26). However, Tarnopol- 1990;19:57–61. sky’s study was small and used estimation (11) Ormel J, Koeter MWJ, van den Brink W, et procedures to model changes in sensitivity al. Concurrent validity of GHQ-28 and rather than actual empirical evidence, so it PSE as measures of change. Psychol Med needs to be replicated before we accept that its 1989;19:1007–1013. results are valid. The GHQ is most useful as (12) Newman SC, Bland RC, Orn H. A comparison of methods of scoring the part of a medical consultation and has seen General Health Questionnaire. Compr widespread use in general practice for screening Psychiatry 1988;29:402–408. for mental disorders. We highly recommend it (13) Vázquez-Barquero JL, Diéz-Manrique JF, for these applications. Peña C, et al. Two stage design in a community survey. Br J Psychiatry 1986;149:88–97. References (14) Goldberg DP, Gater R, Sartorius N, et al. (1) Goldberg DP. The detection of psychiatric The validity of two versions of the GHQ in illness by questionnaire. London: Oxford the WHO study of mental illness in general University Press (Maudsley Monograph health care. Psychol Med 1997;27:191–197. No. 21), 1972. (15) Goldberg DP, Oldehinkel T, Ormel J. Why (2) Goldberg DP, Hillier VF. A scaled version GHQ threshold varies from one place to of the General Health Questionnaire. another. Psychol Med 1998;28:915–921. Psychol Med 1979;9:139–145. (16) Chan DW, Chan TSC. Reliability, validity (3) Goldberg D. Manual of the General Health and the structure of the General Health Questionnaire. Windsor, UK: NFER Questionnaire in a Chinese context. Publishing, 1978. Psychol Med 1983;13:363–371. (4) Worsley A, Gribbin CC. A factor analytic (17) Nott PN, Cutts S. Validation of the 30- study of the twelve item General Health item General Health Questionnaire in Questionnaire. Aust NZ J Psychiatry postpartum women. Psychol Med 1977;11:269–272. 1982;12(2):409–413. (5) Banks MH. Validation of the General (18) Vieweg BW, Hedlund JL. The General Health Questionnaire in a young Health Questionnaire (GHQ): a community sample. Psychol Med comprehensive review. J Operat Psychiatry 1983;13:349–353. 1983;14:74–85. (6) Surtees PG, Miller PM. The interval (19) Lewis G, Wessely S. Comparison of the General Health Questionnaire. Br J General Health Questionnaire and the Psychiatry 1990;157:679–685. Hospital Anxiety and Depression Scale. Br (7) Goodchild ME, Duncan-Jones P. J Psychiatry 1990;157:860–864. Chronicity and the General Health (20) Hobbs P, Ballinger CB, Smith AHW. Questionnaire. Br J Psychiatry Factor analysis and validation of the 1985;146:55–61. General Health Questionnaire in women: a (8) Clarke DM, Smith GC, Herrman HE. A general practice survey. Br J Psychiatry comparative study of screening instruments 1983;142:257–264. for mental disorders in general hospital (21) Vázquez-Barquero JL, Williams P, Diéz- 270 Measuring Health Manrique JF, et al. The factor structure of in neurological in-patients. Br J Psychiatry the GHQ-60 in a community sample. 1986;148:548–553. Psychol Med 1988;18:211–218. (34) Benjamin S, Lennon S, Gardner G. The (22) Benjamin S, Decalmer P, Haran D. validity of the General Health Community screening for mental illness: a Questionnaire for first-stage screening for validity study of the General Health mental illness in pain clinic patients. Pain Questionnaire. Br J Psychiatry 1991;47:197–202. 1982;140:174–180. (35) Aylard PR, Gooding JH, McKenna PJ, et (23) Mulrow CD, Williams JW, Jr., Gerety MB, al. A validation study of three anxiety and et al. Case-finding instruments for depression self-assessment scales. J depression in primary care settings. Ann Psychosom Res 1987;31:261–268. Intern Med 1995;122:913–921. (36) Mari JJ, Williams P. A comparison of the (24) Chandarana PC, Eals M, Steingart AB, et validity of two psychiatric screening al. The detection of psychiatric morbidity questionnaires (GHQ-12 and SRQ-20) in and associated factors in patients with Brazil, using Relative Operating rheumatoid arthritis. Can J Psychiatry Characteristic (ROC) analysis. Psychol 1987;32:356–361. Med 1985;15:651–659. (25) Ibbotson T, Maguire P, Selby P, et al. (37) Le Fevre P, Devereux J, Smith S, et al. Screening for anxiety and depression in Screening for psychiatric illness in the cancer patients: the effects of disease and palliative care inpatient setting: a treatment. Eur J Cancer 1994;30A:37–40. comparison between the Hospital Anxiety (26) Tarnopolsky A, Hand DJ, McLean EK, et and Depression Scale and the General al. Validity and uses of a screening Health Questionnaire-12. Palliat Med questionnaire (GHQ) in the community. Br 1999;13:399–407. J Psychiatry 1979;134:508–515. (38) Headey B, Kelley J, Wearing A. Dimensions (27) Goldberg DP, Rickels K, Downing R, et al. of mental health: life satisfaction, positive A comparison of two psychiatric screening affect, anxiety and depression. Soc Indicat tests. Br J Psychiatry 1976;129:61–67. Res 1993;29:63–82. (28) Weinstein MC, Berwick DM, Goldman PA, (39) Hoeymans N, Garssen AA, Westert GP, et et al. A comparison of three psychiatric al. Measuring mental health of the Dutch screening tests using Receiver Operating population: a comparison of the GHQ-12 Characteristic (ROC) analysis. Med Care and the MHI-5. Health Qual Life 1989;27:593–607. Outcomes 2004;2:23–27. (29) Cleary PD, Goldberg ID, Kessler LG, et al. (40) Piccinelli M, Bisoffi G, Bon MG, et al. Screening for mental disorder among Validity and test-retest reliability of the primary care patients. Arch Gen Psychiatry Italian version of the 12-item General 1982;39:837–840. Health Questionnaire in general practice: a (30) Berwick DM, Budman S, Damico-White J, comparison between three scoring methods. et al. Assessment of psychological Compr Psychiatry 1993;34:198–205. morbidity in primary care: explorations (41) Cheung P, Spears G. Reliability and with the General Health Questionnaire. J validity of the Cambodian version of the Chronic Dis 1987;40:S71–S79. 28-item General Health Questionnaire. Soc (31) Huppert FA, Walters DE, Day NE, et al. Psychiatry Psychiatr Epidemiol The factor structure of the General Health 1994;29:95–99. Questionnaire (GHQ-30): a reliability (42) Medina-Mora ME, Padilla GP, Campillo- study on 6317 community residents. Br J Serrano C, et al. The factor structure of the Psychiatry 1989;155:178–185. GHQ: a scaled version for a hospital’s (32) Tennant C. The General Health general practice service in Mexico. Psychol Questionnaire: a valid index of Med 1983;13:355–361. psychological impairment in Australian (43) Kitamura T, Sugawara M, Aoki M, et al. populations. Med J Aust 1977;2:392–394. Validity of the Japanese version of the (33) Bridges KW, Goldberg DP. The validation GHQ among antenatal clinic attendants. of the GHQ-28 and the use of the MMSE Psychol Med 1989;19:507–511.  Psychological Well-being 271 (44) Chong M-Y, Wilkinson G. Validation of ments. Newer methods have generally been used 30- and 12-item versions of the Chinese with consistent question wording, thus enhanc- Health Questionnaire (CHQ) in patients ing comparability of results across studies. The admitted for general health screening. conceptual basis of the more recent scales, their Psychol Med 1989;19:495–505. statement of purpose, and their interpretation (45) Cheng T-A, Williams P. The design and are all far more clearly spelled out than was the development of a screening questionnaire case with the early methods. Empirical analyses (CHQ) for use in community studies of mental disorders in Taiwan. Psychol Med using some of the measures has informed subse- 1986;16:415–422. quent conceptual discussion, forming an ideal it- (46) Chan DW. The two scaled versions of the erative loop. Chinese General Health Questionnaire: a The major lessons to be learned relate to the comparative analysis. Soc Psychiatry conceptual formulation of an index. The concept Psychiatr Epidemiol 1995;30:85–91. of psychological well-being is inherently less (47) Kilic C, Rezaki M, Rezaki B, et al. General specific than that of physical disability, and im- Health Questionnaire (GHQ12 & mense effort has been expended on debating GHQ28): psychometric properties and what the scales measure. While the results of val- factor structure of the scales in a Turkish idation studies show the Health Opinion Survey primary care sample. Soc Psychiatry or the Langner scale capable of screening for Psychiatr Epidemiol 1997;32:327–331. (48) Kihc C, Rezaki M, Rezaki B, et al. General clinically identifiable disorders, they do not offer Health Questionnaire (GHQ12 & differential diagnoses in the traditional way that GHQ28): psychometric properties and psychiatrists classify mental disorders. As noted factor structure of the scales in a Turkish earlier, they represent the psychological counter- primary care sample. Soc Psychiatry part of Selye’s notion of stress, the nonspecific Psychiatr Epidemiol 1997;32:327–331. element common to diverse disorders that warns (49) Riaz H, Reza H. The evaluation of an Urdu the observer that something is wrong without version of the GHQ-28. Acta Psychiatr specifying what it might be. This idea may have Scand 1998;97:427–432. been clear to the originators of the early scales, (50) Sen B, Mari JJ. Psychiatric research but if so, it was not made sufficiently explicit to instruments in the transcultural setting: prevent subsequent users from misinterpreting experiences in India and Brazil. Soc Sci Med 1986;23:277–281. or overinterpreting the scales. Newer measure- (51) Rost K, Burnam MA, Smith GR. ments have provided more explicit definitions of Development of screeners for depressive how they should and should not be interpreted, disorders and substance disorder history. and of what high scores do and do not indicate. Med Care 1993;31:189–200. We have also seen that there are serious dis- advantages to making piecemeal alterations to the questions in a scale. If changes become nec- Conclusion essary, it would be well to indicate this by alter- ing the title, perhaps by adding a version The measurements presented in this chapter il- number, similar to the approach used by Gold- lustrate several salient lessons learned in the evo- berg. Confusion can also arise from the early lution of health measurements. Early scales publication of draft questionnaires. Because of sought objectivity and could only indicate well- the pressure to publish in academic circles, draft being by the absence of symptoms of distress. forms of a measure are frequently published and This was replaced by measures that include a it then often becomes difficult to ensure that combination of symptoms and direct reports of users apply the final definitive version. feelings. Early methods often lacked a clear defi- The current status of this area of general psy- nition of precisely what they were intended to chological measurement is best summarized in measure and the uncertainty this engendered led terms of the individual measurement methods. to criticisms and variant forms of the instru- The Goldberg scale provides a good method for 272 Measuring Health screening for general psychological and psychi- the RAND scale, which expands the scope of the atric disorder. It has been used internationally, General Well-Being Schedule with additional and many validation studies have demonstrated positive items. This provides a good example of its psychometric qualities. The field of more sub- the planned and systematic development of a jective feelings of well-being is currently repre- measurement instrument that reflects the current sented by Dupuy’s scale, although the Bradburn state of conceptual development and builds de- questions continue to see some use. The gap in liberately on existing measurements. Readers the field of subjective well-being covered for so should also look at the more recent Depression, many years by the Bradburn scale was only par- Anxiety and Stress Scale (DASS) which is re- tially covered by the Goldberg and Dupuy meth- viewed in Chapter 6 and which covers much of ods. More recently, this gap has been filled by the scope of the more general scales reviewed the Positive and Negative Affect Scales and by here.  6 Anxiety n popular language, the term anxiety is used to ciated with a “fight-or-flight” response to imme- I refer to various things: a mental state; a drive, such as, being anxious to please; a response to a diate danger; it typically arises after the expo- sure, whereas anxiety is apprehension over a particular situation, for example, being anxious potential future danger. Anxiety need not be about a new job; a personality trait, as in an anx- negative: it may increase vigilance and arousal, ious person; the cause of a behavior (e.g., a per- and thereby enhance performance and learning. son who smokes out of anxiety), and a It is only intense anxiety or anxiety with an psychiatric disorder (1). Translating this range of inappropriate focus that comes to the attention common parlance into more clinical terms, of clinicians. Anxiety disorders are relatively evi- Hamilton distinguished between anxiety as a dent. The anxious patient appears apprehensive, normal reaction to danger, anxiety as a patho- sweats, and complains of nervousness, palpita- logical mood, and anxiety as a neurotic state or tions, and faintness; somatic signs include rapid syndrome. Anxiety in reaction to danger is breathing, tachycardia, and labile blood pres- milder but more prolonged than fear, and com- sure. Most clinical descriptions include four prises biological changes in the organism that main components: an affective or emotional re- prepare it to handle stress (2). By contrast, sponse; cognitive changes such as confusion, pathological anxiety arises not in reaction to an poor decision-making, memory problems, or external threat but to an internal stimulus; the fearful thoughts; behavioral symptoms such as relationship between this condition and anxiety agitated movements like pacing, or wringing neurosis was not made fully clear (2). Anxiety is hands, and physiological symptoms of hyper- a normal response to threats or challenges, espe- arousal: sweating, palpitations, muscle tension cially those that are perceived to be uncontrol- or gastrointestinal symptoms (6). Under the lable. It involves a loose cognitive and affective rubric of anxiety disorders, the DSM-IV includes structure of negative feelings that blends appre- panic disorders, phobias, obsessive-compulsive hension with a state of physical readiness to disorders, generalized anxiety disorder, and post- cope with upcoming negative events. The affec- traumatic stress disorder (4). tive component may include a feeling of helpless- Anxiety disorders are common, with a life- ness due to perceived inability to predict or time prevalence perhaps reaching 25% of the control the anticipated events (3). population (5, p32). The World Health Organiz- The Diagnostic and Statistical Manual of ation (WHO) undertook a 14-nation study of Mental Disorders (DSM-IV) defined anxiety as psychological problems and set the prevalence of “apprehensive anticipation of future danger or generalized anxiety, defined by ICD-10 criteria, misfortune accompanied by a feeling of dyspho- at 7.9%; a further 1.1% had a panic disorder, ria or somatic symptoms of tension” (4). It is and 1.5% suffered agoraphobia. Although com- thus a future-oriented state, motivating the per- mon, anxiety is underdiagnosed in the general son to avoid the perceived danger; worry may be population; nearly half the anxious patients in seen as a cognitive manifestation of anxiety (5). the WHO study had not been identified by their Fear, by contrast, is a basic emotion that is asso- primary care physicians (7, pp39–40). 273 274 Measuring Health It is important to recognize a distinction be- response involving common signs such as heart tween psychological and medical models of anx- palpitations, dilation of the pupils, and increased iety. The medical approach is categorical; to perspiration. Freud subsequently distinguished receive a diagnosis of anxiety, a patient must between objective and neurotic anxiety, based meed specified criteria, as laid out, for example, on whether the source of anxiety was external or in the DSM-IV (4). The categorical approach is internal. The feelings of apprehension, irritabil- practical and provides a basis for deciding ity, and physiological arousal are the same in whether or not to treat a patient. The underlying both conditions. Objective anxiety, synonymous assumption is that there is a qualitative distinc- with fear, is an internal reaction to a real exter- tion between those who are well and those who nal threat, whereas Freud characterized internal are sick; although sickness can vary in severity, or neurotic anxiety as a reaction to the person’s cases either do not lie on the same continuum as own repressed sexual or aggressive impulses that noncases, or at least form a distinctive cluster at threaten to enter consciousness (9, p117). Neu- one end of a continuum. This conception is rotic anxiety arises especially in response to un- widely challenged, however, in many areas of acceptable impulses such as oedipal conflicts or psychiatry. Psychologists take a dimensional ap- sexual feelings that may have been punished in proach that treats anxiety as a continuum of childhood and are accordingly repressed. If a severity (or, in some models, a set of continua) person’s repression of their impulses should par- with no intrinsic threshold. The arguments for a tially break down, placing them in danger of re- dimensional conception point out that there does experiencing repressed psychological trauma, not seem to be a bimodal distribution of scores they may experience a free-floating anxiety that representing well and sick groups; there also appears to have no specific object save a fear of seems to be a continuum of impairments due to punishment if the inner impulses are expressed anxiety, with no clear threshold beyond which openly. Neurotic anxiety may be inferred when rising anxiety scores would indicate an anxiety anxiety reactions appear disproportionate to the disorder. Similarly, a dimensional model has level of threat (10). Freud also described moral been proposed in mental status testing (Chapter anxiety, in which the conflict lies between the 8), where there is a diagnostic dilemma of how person’s impulses or unconscious desires and ex- to classify people who do not meet the criteria ternal prohibitions as perceived through the per- for dementia, but who are also not normal. son’s conscience. Thus, a high school student Most of the anxiety scales we review provide in- who is attracted to a teacher will feel anxious tensity or severity scores that reflect an underly- about meeting the teacher in the corridor. ing dimensional model of anxiety. A limitation of Freud’s theory is that it did not adequately distinguish among the resulting feelings of stress, guilt, anxiety, or depression, Theoretical Approaches which tended to be grouped under the general la- to Anxiety bel of “neurotic” symptoms. Freud’s perspective was also strictly clinical and he opposed formal Spielberger (8; 9), Reiss (10), and Endler and measurement; this orientation seemed to delay Kocovski (11) have offered helpful summaries of the development of anxiety scales for roughly 30 the evolution of scientific interest in anxiety, years. which has long been recognized but only rela- The 1950s saw the development of an experi- tively recently studied systematically. Fear was mental tradition in studying anxiety. Laboratory apparently portrayed in ancient Egyptian hiero- studies assessed the links among personal drive, glyphics, and the Roman orator Cicero distin- anxiety, the complexity of an experimental task guished between a character predisposition to and feelings of fear and frustration. In 1953, anxiety (a– nxieta– s) and emotional responses to Taylor presented her Manifest Anxiety Scale situations (angor). Nineteen hundred years later, (TMAS) that built on Freud’s theme of neurotic Darwin analyzed the role of fear as an adaptive anxiety. The TMAS was widely used in experi-  Anxiety 275 mental research; common findings were that berger’s contribution was to clarify further the people with higher drive, or “manifest anxiety,” distinction between trait and state anxiety, lead- showed superior performance in simple response ing to his 1968 State-Trait Anxiety Inventory tasks, but less adequate performance in complex (STAI). State and trait anxiety have been likened tasks that included many possible types of error to kinetic and potential energy (11, p232), but (9, p119). Subsequent work on the link between defining trait anxiety in terms of a general ten- learning and anxiety during the mid 1950s re- dency to respond anxiously to stress does not de- vealed limitations in the drive theory that under- fine either the general tendency or the types of lay Taylor’s early work; Spielberger has reported threat (10, p204). Anxiety can feed on itself, so how this led to a revised model developed by a subsequent development was to try and sepa- K.W. Spence (who was Taylor’s husband) (9). rate feelings of anxiety from feelings about anx- Anxiety came to be viewed more comprehen- iety. The Anxiety Sensitivity Index (ASI) was sively as a process that involves stressful threats, proposed to record individual differences in fear personality characteristics and defences, and be- of experiencing anxiety. A person who is sensi- havioral reactions. tive to anxiety would tend, for example, to antic- During the 1960s, this reference to personal- ipate that a rapidly beating heart presages a ity led to Spielberger’s empirical demonstration heart attack; a person with low sensitivity might of a distinction between anxiety as a reaction perceive stress as a transient nuisance (10, versus an underlying tendency to respond to p206). Anxiety sensitivity appears similar to threats. Cattell and others had applied newly de- trait anxiety, except that it refers less to past ten- veloped multivariate analysis techniques to mea- dencies than to future fears about the conse- sures of anxiety, thus also showing two distinct quences of anxiety. More recently, Endler has facets of anxiety, state, and trait, which were not presented a multidimensional model of anxiety included in Taylor’s measure (8). Traits refer to that maintains the state-trait distinction, but enduring and general dispositions to react to sit- subdivides each component (11–13). This is por- uations in a consistent manner; trait anxiety in- trayed in the Endler Multidimensional Anxiety volves a tendency to experience anxious Scales (EMAS), which divides state scores into symptoms in nonthreatening situations; it im- cognitive-worry and autonomic-emotional com- plies vulnerability to stress. State anxiety is a dis- ponents (14). crete response to a specific threatening situation: Etiological theories of anxiety are diverse but Freud’s objective anxiety. State anxiety involves may be grouped into biological theories that transitory unpleasant feelings of apprehension, emphasize the relevance of hormone levels, neuro- tension, nervousness, or worry, often accompa- chemical patterns, and genetics, versus cognitive- nied by activation of the autonomic nervous sys- behavioral theories that argue that such biological tem. It presumably forms a natural defence and changes may result from psychological reactions. adaptation mechanism in the face of threat. Peo- A synthesis between these perspectives has also ple with high trait anxiety are assumed to be been proposed (5). Behavioral theorists tend to more prone to experiencing state anxiety, per- emphasize the relevance of parenting styles and haps to excess. Freud had anticipated this in his early learning experiences that may foster a fear recognition of the variations in response to ob- response and a sense of powerlessness. Cognitive jective threats among normal people and those theorists point out the relevance of beliefs and with neurotic disorders (8, pp7–8). perceptions for the maintenance of anxiety reac- These conceptual developments were re- tions. Clark, for example, showed that panic at- flected in the world of health measurements, and tacks may be triggered by a misinterpretation of in 1963 Cattell and Scheier developed the Anxi- normal physical sensations as presaging a threat; ety Scale Questionnaire (ASQ) to measure trait a vicious circle involves a reaction of heightened anxiety. It was distributed by the Institute for anxiety that produces more physical sensations Personality Assessment and Testing, so it was leading to more catastrophic interpretations, spi- also called the IPAT Anxiety Scale (8, p9). Spiel- raling into a panic attack (15). Many other vari- 276 Measuring Health ants have been described (16). Biological theorists with it, and it’s probably my fault anyway so have tended to focus on the role of particular neu- there’s really nothing I can do.” (3, p14). rotransmitter systems in particular anxiety disor- A 1991 paper by Clark and Watson formed a ders; the noradrenergic system may be linked to watershed in formulating the conceptual distinc- panic disorder, whereas the serotonergic system tion between anxiety and depression (20). Based appears relevant in obsessive-compulsive disor- on an analysis of patterns of association between ders, dopamine may be relevant in social phobia, measures of anxiety and depression, they pro- for example (17, p10). posed a tripartite hierarchical model that holds that anxiety and depression have common, but also unique, features. Common to both condi- Anxiety and Depression tions is general affective distress or negative af- fect, plus symptoms such as sleep disturbances, Much debate in psychology has been devoted to irritability, and loss of appetite. Beyond these how finely to draw distinctions among condi- nonspecific symptoms, depression is uniquely tions that are clearly related. Anxiety and de- characterized by anhedonia and low levels of pression share common symptoms and can positive affect. These refer to a loss of pleasure result from similar circumstances, and there is and interest in life, a lack of enthusiasm, slug- evidence that treatments for each condition can gishness, apathy, social withdrawal, and disinter- benefit patients with the other (18). But in the- est. Thus, negative affect is nonspecific, whereas ory, at least, the two are distinguishable and anx- positive affect (or, in this case, its absence) is iety is not generally seen as an aspect of specific to depression (20, p330). Anxiety, depression. Conversely, discriminating between meanwhile, is uniquely characterized by physio- them may focus attention on the trees rather logical hyperarousal, exhibited in racing heart, than the forest; Brown et al. noted, “Of further sweating, shakiness, trembling, shortness of concern is the possibility that our classification breath, and feelings of panic (20, p331). The systems have become overly precise to the point nonspecific distress factor reflects a “tempera- that they are now erroneously distinguishing mental sensitivity to negative stimuli” that is symptoms and disorders that actually reflect in- related to neuroticism and predicts the develop- consequential variations of broader, underlying ment of either depression or anxiety disorders syndromes” (19, p170). There is a dialectical (21, p81). Clark et al. reviewed evidence for this conflict in perspectives on the links between idea and noted that people who score highly on anxiety and depression: a unitary theory sees trait neuroticism appear more likely to develop them as expressions of the same pathology; the depression, concluding that neuroticism forms a opposing perspective sees them as fundamentally vulnerability factor to depression (22). Empiri- different, whereas the compromise is to view cal support for the Clark and Watson tripartite them as having common roots but different ex- model comes largely from confirmatory factor pressions for a variety of reasons (19). Probably analyses of measurement scales that identified they are linked, but anxiety suggests arousal and first- and second-order factors (19; 21), although an attempt to cope with the situation; depression there has been criticism of the model. For exam- suggests lack of arousal and withdrawal. Refer- ple, in factor analyses, the three dimensions, ring back to Exhibit 5.1, the contrast lies be- although distinguishable, have been highly inter- tween the northeast and southeast quadrants of correlated, with a strong inverse correlation be- the compass. Barlow characterized the contrast tween positive and negative affect (r = −0.81 to with two quotations. An anxious person might −0.86) (23). Further developments to the con- say “That terrible event is not my fault but it ceptual model of affect, anxiety, and depression may happen again, and I may not be able to cope have been described by Watson et al. (24). with it but I’ve got to be ready to try.” A de- Although conceptual distinctions are drawn pressed person might say “That terrible event between anxiety and depression, the two share may happen again and I won’t be able to cope many common symptoms and may result from  Anxiety 277 similar circumstances; there may also be bidirec- measures may also be used, for example using tional causal links between them (25, p145), so the electroencephalogram for recording central that they quite commonly occur together (7; 26). nervous system responses; the electrocardiogram For example, the limitations in everyday function for cardiovascular system responses; measures of brought on by anxiety may lead to pessimism respiration rate and depth; measures of gastroin- and general despondency; conversely, the lack of testinal system responses such as stomach pH or energy and the poor self-esteem of depression stomach motility; and measures of skin potential may undermine the sense of self-efficacy and or palmar sweating responses (29, p349). These thereby lead to anxiety. Further, the validation of types of measure are not covered in this book. anxiety measures is complicated by a tendency Table 6.1 summarizes the evidence for the quality for clinicians to follow a hierarchical diagnostic of the scales reviewed. approach in which they may overlook symptoms of anxiety in a depressed patient, so that depres- sion tends to exclude anxiety (27, p137). Because References of the overlap of symptoms, it proves much easier (1) Zung WWK. How normal is anxiety? to develop sensitive tests than specific ones. De- Kalamazoo, MI: The Upjohn Company, spite the attempts of several authors to write 1980. items that are unique to anxiety or to depression, (2) Hamilton M. Diagnosis and rating of measures of the two seem to correlate about anxiety. Br J Psychiatry 1969; Special +0.50. The Depression Anxiety Stress Scales, for Publication #3:76–79. (3) Barlow DH. The nature of anxiety: anxiety, example, included only items that loaded depression, and emotional disorders. In: uniquely on either depression or anxiety factors, Rapee RM, Barlow DH, eds. Chronic and yet the correlation between DASS depression anxiety: generalized anxiety disorder and and anxiety scales was 0.56 (28, Table 2). mixed anxiety-depression. New York: Guilford, 1991:1–28. (4) American Psychiatric Association. Diagnostic Anxiety Measurements and statistical manual of mental disorders, 4th ed. (DSM-IV). Washington, DC: The options for measurement include subjective American Psychiatric Association, 1994. ratings, either by the person herself, or by a clini- (5) Antony MM, Swinson RP. Anxiety cian, the latter being based on her interpretation disorders: future directions for research and treatment. A discussion paper. Ottawa, of the patient’s report in an interview. Self- Ontario: Health Canada, 1996. ratings are illustrated here by Taylor’s Manifest (6) Creamer M, Foran J, Bell R. The Beck Anxiety Inventory, Zung’s Self-Rating Anxiety Anxiety Inventory in a non-clinical sample. Scale, Spielberger’s State-Trait Anxiety Inven- Behav Res Ther 1995;33:477–485. tory, the Beck Anxiety Inventory, and the Hospi- (7) Sartorius N, Üstün TB, Lecrubier Y, et al. tal Anxiety and Depression Scale.* The Depression comorbid with anxiety: results Depression Anxiety Stress Scales provide an ex- from the WHO study on psychological ample of a clinician-rating method. Semiobjec- disorders in primary health care. Br J tive ratings can also be used, covering specific Psychiatry 1996; 168(suppl 30):38–43. cues for particular signs of anxiety; the Hamilton (8) Spielberger CD. Assessment of state and rating scale and Zung’s Anxiety Status Inventory trait anxiety: conceptual and methodological issues. South Psychol are examples reviewed in this chapter. Objective 1985;2:6–16. (9) Spielberger CD. Anxiety: state-trait-pro- *Note that the HADS and the DASS could equally well cess. In: Spielberger CD, Sarason IG, eds. have been included in Chapter 7 on Depression; the Stress and anxiety. Vol. I. Washington, DC: DASS might also fit in Chapter 5 on general well-being Hemisphere Publishers, 1975: 115–143. measures; their inclusion in the present chapter is rela- (10) Reiss S. Trait anxiety: it’s not what you tively arbitrary. think it is. J Anx Disord 1997; 11:201–214.  Table 6.1 Comparison of the Quality of Anxiety Measurements* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurement Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Manifest Anxiety Scale ordinal 50 experimental self (5–10 min) many ** ** ** ** (Taylor, 1953) research The Hamilton Anxiety Rating ordinal 14 clinical clinicin rating many ** *** ** *** Scale (Hamilton, 1959) scale (20 min) Hospital Anxiety and ordinal 14 clinical self (2–5 min) many *** *** *** *** Depression Scale (Zigmond 278 and Snaith, 1983) Self-Rating Anxiety Scale; ordinal 20 clinical, self; clinician several * ** * ** Anxiety Status Inventory screening rating (Zung, 1971) (5–15 min) Beck Anxiety Inventory ordinal 21 survey, self or inter- many *** *** *** *** (Beck, 1988) clinical viewer (5 min) Depression Anxiety Stress ordinal 42 survey, interviewer few ** *** *** *** Scales (Lovibond, 1993) clinical (10 min) State-Trait Anxiety Inventory ordinal 40 research, self (10 min) many *** *** *** *** (Spielberger, 1977) screening * For an explanation of the categories used, see Chapter 1, pages 6–7.  Anxiety 279 (11) Endler NS, Kocovski NL. State and trait depressive and hypertensive patient groups. anxiety revisited. Anx Disord J Pers Assess 2003;80:139–153. 2001;15:231–245. (24) Watson D, Gamez W, Simms LJ. Basic (12) Endler NS. A person-situation interaction dimensions of temperament and their model for anxiety. In: Spielberger CD, relation to anxiety and depression: a Sarason IG, eds. Stress and anxiety. Vol. 1. symptom-based perspective. J Res Pers Washington, DC: Hemisphere Publishing, 2005; 39:46–66. 1975:145–164. (25) de Beurs E, Wilson KA, Chambless DL, et (13) Endler NS, Parker J, Bagby RM, et al. al. Convergent and divergent validity of the Multidimensionality of state and trait Beck Anxiety Inventory for patients with anxiety: factor structure of the Endler panic disorder and agoraphobia. Dep Anx Multidimensional Anxiety Scales. J Pers 1997;6:140–146. Soc Psychol 1991;60:919–926. (26) Burns DD, Eidelson RJ. Why are (14) Endler NS, Edwards JM, Vitelli R. Endler depression and anxiety correlated? A test Multidimensional Anxiety Scales: manual. of the tripartite model. J Consult Clin 1st ed. Los Angeles: Western Psychological Psychol 1998;66:461–473. Services, 1991. (27) Bramley PN, Easton AME, Morley S, et al. (15) Clark DM. A cognitive approach to panic. The differentiation of anxiety and Behav Res Ther 1986;24:461–470. depression by rating scales. Acta Psychiatr (16) Rapee RM. Current controversies in the Scand 1988;77:133–138. anxiety disorders. New York: Guilford (28) Lovibond PF, Lovibond SH. The structure Press, 1996. of negative emotional states: comparison of (17) Antony MM, Swinson RP. Anxiety the Depression Anxiety Stress Scales disorders and their treatment: a critical (DASS) with the Beck Depression and review of the evidence-based literature. Anxiety Inventories. Behav Res Ther Ottawa, Ontario: Health Canada, 1996. 1995;33:335–343. (18) Lipman RS, Covi L, Downing RW, et al. (29) Zung WWK, Cavenar JO, Jr. Assessment Pharmacotherapy of anxiety and scales and techniques. In: Kutash IL, depression. Psychopharmacol Bull Schlesinger LB, eds. Handbook on stress 1981;17:91–103. and anxiety. San Francisco: Jossey-Bass, (19) Brown TA, Chorpita BF, Barlow DH. 1980:348–363. Structural relationships among dimensions of the DSM-IV anxiety and mood disorders and dimensions of negative affect, positive The Manifest Anxiety Scale affect, and autonomic arousal. J Abnorm (Janet Taylor, 1953) Psychol 1998;107:179–192. (20) Clark LA, Watson D. Tripartite model of Purpose anxiety and depression: psychometric Taylor’s Manifest Anxiety Scale (TMAS) was evidence and taxonomic implications. J originally developed as a device for selecting sub- Abnorm Psychol 1991;100:316–336. jects for inclusion in psychological experiments (21) Dunbar M, Ford G, Hunt K, et al. A on stress, motivation, and human performance confirmatory factor analysis of the Hospital (1). It has subsequently been used as a general Anxiety and Depression Scale: comparing indicator of anxiety as a personality trait; it is empirically and theoretically derived not intended as a specific measure of anxiety as a structures. Br J Clin Psychol clinical entity (2). 2000;39:79–94. (22) Clark LA, Watson D, Mineka S. Temperament, personality, and the mood Conceptual Basis and anxiety disorders. J Abnorm Psychol Taylor conducted a series of experimental stud- 1994;103:103–116. ies in the 1950s, initially designed to test a the- (23) Marshall GN, Sherbourne CD, Meredith ory about the effects of level of drive on human LS, et al. The tripartite model of anxiety performance in tasks of different levels of and depression: symptom structure in complexity. She assumed that within the context 280 Measuring Health of these studies, drive level would be reflected in filler items does not affect response to the items, the intensity of what she termed “manifest which are therefore normally presented alone anxiety”—that is, anxiety that was evident and (8). self-perceived (J. Taylor Spence, personal com- munication, 2005). The theory predicted that on Reliability simple tasks, performance would be improved For the 50-item version, Taylor reported retest by higher levels of drive, as reflected on her correlations of 0.89, 0.82, and 0.81 over inter- measure of anxiety. In her initial experiment (3), vals of three weeks, five months and nine to 17 which involved classical conditioning of eyelid months (1, p286). For the 28-item version, she response to a puff of air, she confirmed this pre- reported a four-week retest correlation of 0.88 diction. The theory further predicted that on (1, p289). The 50- and 28-item versions corre- more complex tasks, anxiety level would be neg- lated 0.85, with an interval of three weeks atively related to performance; this was later between administrations (1, p288). confirmed (4). Item-total correlations appear low: one study The concept of manifest anxiety was derived found only 20 of the 50 items to have item-total from Freud’s idea of neurotic anxiety noted in a correlations above 0.4 (9); another study found 1947 text on behavioral psychology by only 16 (10). The range of item-total correla- Cameron. It refers to a general tendency to expe- tions is wide: 0.01 to 0.70 in one study (11, rience anxiety in the face of stress and is exhib- p625). Correlations may also vary by ethnic ited in traits such as giving exaggerated and group and educational level (9). inappropriate reactions on slight provocation, Kuder-Richardson internal consistency esti- expressing fatigue not explained by the person’s mates were 0.78 and 0.84 in two samples (11, physical condition, or being easily upset or p626), whereas Bendig reported a median alpha tremulous (5, p430). To develop the scale of of 0.82 in an unspecified number of studies (12). manifest anxiety, Taylor selected items that cor- A coefficient alpha of 0.70 was obtained from a responded to Cameron’s 1947 formulation of sample of graduate students (13, p259). Internal chronic anxiety reactions (1). consistency rose from 0.81 to 0.90 when a six- point scoring was used for each item in place of Description the traditional dichotomous scoring (7, Table 1). Items judged by clinicians as being indicative of manifest anxiety were selected from the Validity Minnesota Multiphasic Personality Inventory. The resulting scale included 50 items, but Taylor Results of factor analytic studies suggest that the presented a 28-item abbreviation that also sim- TMAS has a broad and diffuse coverage, includ- plified the wording of several of the original ing some dimensions that appear unrelated to items (1, Table 2). Both versions have been used anxiety (14). One very small study identified no in subsequent studies, and both are shown in less than 15 factors for the 50-item version, in- Exhibit 6.1. Taylor’s empirical testing of the cluding dimensions such as ‘general apprehen- TMAS was based on trials with undergraduate sion,’ ‘perceived self-effectiveness,’ ‘lack of self students undertaken between 1948 and 1951 confidence,’ and ‘social confidence’ (15). A study (1). True-false responses are used for each item, of graduate students found that 18 factors had and the replies indicating anxiety (shown in eigenvalues over 1.0 but chose to extract four in- parentheses in the Exhibit) are counted, giving a terpretable factors. These factors, however, ex- score from 0 to 28 or 50. Other users have occa- plained only 22.6% of the common variance; all sionally substituted five- or six-point intensity 18 factors explained only 45.9% (13, response scales (6; 7). Taylor originally inter- pp259–260). Studies by Khan, by O’Connor et spersed the 50 items among a number of other al., and by Moore et al. all reported five-factor ‘filler’ items to disguise the intent of the ques- solutions (6; 16; 17), although the results are not tionnaire; it appears, however, that omitting the closely comparable between the studies. Khan Exhibit 6.1 The Taylor Manifest Anxiety Scale (with answers indicating anxiety shown in parentheses) Original 50-item version: 1. I do not tire quickly. (False) 2. I am troubled by attacks of nausea. (True) 3. I believe I am no more nervous than most others. (False) 4. I have very few headaches. (False) 5. I work under a great deal of tension. (True) 6. I cannot keep my mind on one thing. (True) 7. I worry over money and business. (True) 8. I frequently notice my hand shakes when I try to do something. (True) 9. I blush no more often than others. (False) 10. I have diarrhea once a month or more. (True) 11. I worry quite a bit over possible misfortunes. (True) 12. I practically never blush. (False) 13. I am often afraid that I am going to blush. (True) 14. I have nightmares every few nights. (True) 15. My hands and feet are usually warm enough. (False) 16. I sweat very easily even on cool days. (True) 17. Sometimes when embarrassed, I break out in a sweat which annoys me greatly. (True) 18. I hardly ever notice my heart pounding and I am seldom short of breath. (False) 19. I feel hungry almost all the time. (True) 20. I am very seldom troubled by constipation. (False) 21. I have a great deal of stomach trouble. (True) 22. I have had periods in which I lost sleep over worry. (True) 23. My sleep is fitful and disturbed. (True) 24. I dream frequently about things that are best kept to myself. (True) 25. I am easily embarrassed. (True) 26. I am more sensitive than most other people. (True) 27. I frequently find myself worrying about something. (True) 28. I wish I could be as happy as others seem to be. (True) 29. I am usually calm and not easily upset. (False) 30. I cry easily. (True) 31. I feel anxiety about something or someone almost all of the time. (True) 32. I am happy most of the time. (False) 33. It makes me nervous to have to wait. (True) 34. I have periods of such great restlessness that I cannot sit long in a chair. (True) 35. Sometimes I become so excited that I find it hard to get to sleep. (True) 36. I have sometimes felt that difficulties were piling up so high that I could not overcome them. (True) 37. I must admit that I have at times been worried beyond reason over something that really did not matter. (True) 38. I have very few fears compared to my friends. (False) 39. I have been afraid of things or people that I know could not hurt me. (True) 40. I certainly feel useless at times. (True) (continued) 281 Exhibit 6.1 (continued) 41. I find it hard to keep my mind on a task or job. (True) 42. I am unusually self-conscious. (True) 43. I am inclined to take things hard. (True) 44. I am a high-strung person. (True) 45. Life is a strain for me much of the time. (True) 46. At times I think I am no good at all. (True) 47. I am certainly lacking in self-confidence. (True) 48. I sometimes feel that I am about to go to pieces. (True) 49. I shrink from facing a crisis or difficulty. (True) 50. I am entirely self-confident. (False) 28-item version: (note, the items are numbered here as in the 50-item version to facilitate comparison of the phrasing) 2. I am often sick to my stomach. (True) 3. I am about as nervous as other people. (False) 5. I work under a great deal of strain. (True) 9. I blush as often as others. (False) 10. I have diarrhea (“the runs”) once a month or more. (True) 11. I worry quite a bit over possible troubles. (True) 17. When embarrassed I often break out in a sweat which is very annoying. (True) 18. I do not often notice my heart pounding and I am seldom out of breath. (False) 20. Often my bowels don’t move for several days at a time. (True) 22. At times I lose sleep over worry. (True) 23. My sleep is restless and disturbed. (True) 24. I often dream about things I don’t like to tell other people. (True) 26. My feelings are hurt easier than most people. (True) 27. I often find myself worrying about something. (True) 28. I wish I could be as happy as others. (True) 31. I feel anxious about something or someone almost all of the time. (True) 34. At times I am so restless that I cannot sit in a chair for very long. (True) 36. I have often felt that I faced so many difficulties I could not overcome them. (True) 37. At times I have been worried beyond reason about something that really did not matter. (True) 38. I do not have as many fears as my friends. (False) 42. I am more self-conscious than most people. (True) 43. I am the kind of person who takes things hard. (True) 44. I am a very nervous person. (True) 45. Life is often a strain for me. (True) 47. I am not at all confident of myself. (True) 48. At times I feel that I am going to crack up. (True) 49. I don’t like to face a difficulty or make an important decision. (True) 50. I am very confident of myself. (False) From Taylor JA. A personality scale of manifest anxiety. J Abnormal and Social Psychology 1953;48:286, 288. With permission. 282  Anxiety 283 concluded that “the total score on the MA scale when scores hold consequences for the respon- is a composite of dissimilar traits and hence is dent, more intelligent people may be more apt to not meaningful” (6, p227). fake good scores than less intelligent respon- To demonstrate discriminative ability, Taylor dents (J. Taylor Spence, personal communica- compared scores obtained from a mixed sample tion, 2005) of neuroses and psychoses patients with (median Because Taylor’s original theory held that score, 34) with scores from a sample of univer- anxiety might be related to certain physiological sity students (median 13). The patient median measures, several studies have tested its validity fell at the 99th percentile of the university sam- in this way. Jessor and Hammond failed to find ple (1, p290). Moore et al. found significant dif- an association between TMAS scores and elec- ferences in TMAS scores between patient trophysiological activity (22); Neva and Hicks groups, both for overall scores and for factor found no association with heart rate or galvanic scores (17, p1432). Kendall compared TMAS skin response (23). Other studies of criterion va- scores with independent anxiety ratings made by lidity have related TMAS scores to academic nurses for a sample of patients; the agreement achievement; Khan reported only one significant was weak and only significant if a most patients correlation out of 10 between factor scores on with intermediate TMAS scores were omitted the TMAS and university grades (6, Table 2). from the analysis (5). Similarly, a correlation be- tween TMAS scores and anxiety ratings made by Alternative Forms psychiatrists was only 0.34 (18, p138). TMAS Bendig proposed a 20-item abbreviation that scores did, however, discriminate significantly eliminated items of low internal consistency; al- between patients with anxiety and a range of pha was 0.76, compared with 0.82 for the 50- other diagnostic groups (18, Table 2). item version, and the correlation between them Construct validity has been widely studied. was 0.93 (12). Subsequently Hicks et al. pro- Correlations of 0.72 and 0.75 were reported be- posed a different 20-item version that more tween the TMAS and Eysenck’s measure of neu- closely approximated a single dimension. The roticism in two samples; correlations with the test-retest reliability was 0.88, and they also pro- psychoticism scale were 0.26 and 0.21 (11, vided percentile reference values for this version p626). Similarly, high correlations of 0.81 and (14, Table 2). 0.92 were reported between the TMAS and the A 1956 adaptation for children was called the Psychasthenia scale of the MMPI (19, Table 1). Children’s Manifest Anxiety Scale, and this was The same study found correlations of 0.74 and subsequently revised (24–26) Note that the 0.60 with the MMPI depression score. Other Reynolds and Richmond article was published correlations with depression scores include 0.64 twice). The revised children’s scale has 37 items with the Beck Depression Inventory, and 0.58 covering four dimensions: physiological anxiety with Zung’s Self-Rating Depression Scale (20, (10 items), worry/oversensitivity (11 items), so- Table 2). A correlation of −0.72 with a self- cial concerns (7 items), a lie detection scale (9 esteem score was reported (18, p141). As the items such as “I am always kind,” “I never get TMAS correlated 0.72 with Eysenck’s neuroti- angry.”). The children’s scale has been fre- cism index, Meites et al. concluded that the quently examined for validity (23; 27–29) and is TMAS taps a general emotionality trait (20, still in common use as a screening and outcomes p430). Several studies tested the hypothesis that measure (30; 31). TMAS scores may be confounded by intelli- gence; results show a wide range of correlations, Commentary from −0.40 to +0.19. The results appear to vary The TMAS played an important role in the his- according to the testing conditions, especially tory of research on anxiety; it was one of the according to the level of threat implied by the earliest psychometric measures of anxiety and its test results (21, p402). Taylor suggested that, content influenced the design of the State-Trait 284 Measuring Health Anxiety Inventory. It was the leading anxiety esteem (34). Finally, it is not clear what the opti- measure until the 1970s but has since fallen mal level of manifest anxiety should be. If anxi- from favor in the English-speaking world as a ety enhances performance, presumably some measure for use with adults. However, the chil- anxiety is desirable, but presumably excess anxi- dren’s version is still frequently used, especially ety would be handicapping. in studies of anxiety toward dental treatment; The results of factor analyses have shown in- the adult TMAS is still used in non-English- consistent results. Reasons for this include the speaking countries. use of dichotomous responses (which can make Several concerns led to the decline in popular- the item intercorrelations unstable), the broad ity of the TMAS. It was developed for a specific content of the TMAS, possible confounding by application in experimental psychology and was social desirability in some samples, and the pos- designed to reflect a particular conceptual ap- sible influence of situational factors on self- proach to anxiety that subsequently fell from fa- reported anxiety (13, pp260–261). The use of vor. It appears to be useful in assessing the level dichotomous items has been consistently criti- of drive in experimental subjects (6, p223), al- cized (7; 35); Khan noted that “there seems to be though some studies have questioned its validity no logical justification for the use of a dichoto- as an experimental measure (22; 23). The theory mous response scale because the purpose of Tay- that underlay the TMAS has been subtly refined. lor’s MA scale is to arrive at the intensity of an Taylor held that high scoring individuals were individual’s drive level.” (6, p226). more predisposed to reacting to stressful situa- Despite these criticisms, the TMAS was the tions with anxiety (3), whereas more recent in- first anxiety measurement to see international terpretations hold that high scores identify use, and it formed an important milestone in the people who react with more drive in stressful sit- development of the field. uations, but not in the absence of stress (32, p8). Therefore, the TMAS is an indirect measure of anxiety and, although it may hold relevance in References studies of motivation, it may be less adequate as a pure measure of anxiety. The concept of mani- (1) Taylor JA. A personality scale of manifest fest anxiety does not distinguish between trait anxiety. J Abnorm Soc Psychol anxiety in mentally healthy people and patholog- 1953;48:285–290. ical anxiety in people with mental illness (33, (2) Zung WWK. The measurement of affects: depression and anxiety. Mod Probl p203). Hence, TMAS scores do not correlate Pharmacopsychiatry 1974;7:170–188. well with psychiatric assessments of anxiety (3) Taylor JA. The relationship of anxiety to (18), nor do its scores predict performance in the conditioned eyelid response. J Exp settings such as school (6, p223). Even though it Psychol 1951;41:81–92. appears to work quite well with students, the (4) Taylor JA. Drive theory and manifest TMAS performed inconsistently when applied to anxiety. Psychol Bull 1956;53:303–320. patient samples; for example, mean scores ob- (5) Kendall E. The validity of Taylor’s tained from different types of patient do not ap- Manifest Anxiety Scale. J Consult Psychol pear to correspond to their clinical condition 1954;18:429–432. (19). As with other anxiety scales, the TMAS is (6) Khan SB. Dimensions of manifest anxiety not specific in defining anxiety, at least as it is and their relationship to college achievement. J Consult Clin Psychol conceptualized in contemporary psychiatry. In- 1970;35:223–228. stead, the TMAS appears to cover a broad theme (7) Salisbury JL, Sherrill D, Friedman ST, et al. of neuroticism or social performance anxiety. Comparison of two scoring methods for Spielberger’s analyses showed that the TMAS the short form of the Manifest Anxiety did not predict anxiety reactions to electric Scale and Eysenck’s Extraversion (E) and shock but did predict reactions to psychologi- Neuroticism (N) scales. Psychol Rep cally threatening situations or threats to self- 1968;22:1235–1236.  Anxiety 285 (8) McCreary JB, Bendig AW. Comparison of validity and the Taylor Manifest Anxiety two forms of the Manifest Anxiety Scale. J Scale. Psychol Bull 1957;54:161–170. Consult Psychol 1954;18:206. (23) Neva E, Hicks RA. A new look at an old (9) Moerdyk AP, Spinks PM. Preliminary issue: Manifest Anxiety Scale validity. J cross-cultural validity study of Taylor Consult Clin Psychol 1970;35:406–408. Manifest Anxiety Scale. Psychol Rep (24) Reynolds CR, Richmond BO. Revised 1979;45:663–664. Children’s Manifest Anxiety Scale (10) Hoyt DP, Magoon TM. A validation study (RCMAS) manual. Los Angeles: Western of the Taylor Manifest Anxiety Scale. J Psychological Services, 1985. Clin Psychol 1954;10:357–361. (25) Reynolds CR, Richmond BO. What I think (11) Hojat M, Shapurian R. Anxiety and its and feel: a revised measure of children’s measurement: a study of the psychometric manifest anxiety. J Abnorm Child Psychol characteristics of a short form of the Taylor 1978;6:271–280. Manifest Anxiety Scale in Iranian college (26) Reynolds CR, Richmond BO. What I students. J Soc Behav Pers Think and Feel: a revised measure of 1986;1:621–630. Children’s Manifest Anxiety. J Abnorm (12) Bendig AW. The development of a short Child Psychol 1997;25:15–20. form of the Manifest Anxiety Scale. J (27) Reynolds CR, Richmond BO. Factor Consult Psychol 1956;20:384. structure and construct validity of “what I (13) Livneh H, Redding CA. A factor analytic think and feel”: The Revised Children’s study of manifest anxiety: a transsituational, Manifest Anxiety Scale. J Pers Assess transtemporal investigation. J Psychol 1979;43:281–283. 1986;120:253–263. (28) Wisniewski JJ, Mulick JA, Genshaft JL, et (14) Hicks RA, Ostle JR, Pellegrini RJ. A al. Test-retest reliability of the revised unidimensional short form of the TMAS. Children’s Manifest Anxiety Scale. Percept Bull Psychosom Soc 1980;16:447–448. Mot Skills 1987;65:67–70. (15) Reynolds SL, Burdsal C. A factor analytic (29) Joiner TE, Jr., Catanzaro SJ, Laurent J. study of manifest anxiety and abstract- Tripartite structure of positive and negative concrete word recall. J Multivar Exp affect, depression, and anxiety in child and Personality Clin Psychol 1975;1:150–164. adolescent psychiatric inpatients. J Abnorm (16) O’Connor JP, Lorr M, Stafford JW. Some Psychol 1996;105:401–409. patterns of manifest anxiety. J Clin Psychol (30) Dierker LC, Albano AM, Clarke GN, et al. 1956;12:160–165. Screening for anxiety and depression in (17) Moore PN, Kinsman RA, Dirks JF. early adolescence. J Am Acad Child Subscales to the Taylor Manifest Anxiety Adolesc Psychiatry 2001;40:929–936. Scale in three chronically ill populations. J (31) Crook K, Beaver BR, Bell M. Anxiety and Clin Psychol 1984;40:1431–1433. depression in children: a preliminart (18) Siegman AW. Cognitive, affective, and examination of the utility of the PANAS-C. J psychopathological correlates of the Taylor Psychopathol Behav Assess 1998;20:333–350. Manifest Anxiety Scale. J Consult Psychol (32) Spielberger CD. Assessment of state and 1956;20:137–141. trait anxiety: conceptual and (19) Brackbill G, Little KB. MMPI correlates of methodological issues. South Psychol the Taylor scale of manifest anxiety. J 1985;2:6–16. Consult Psychol 1954;18:433–436. (33) Reiss S. Trait anxiety: it’s not what you think (20) Meites K, Lovallo W, Pishkin V. A it is. J Anxiety Disord 1997;11:201–214. comparison of four scales for anxiety, (34) Spielberger CD. Anxiety: state-trait-pro- depression, and neuroticism. J Clin Psychol cess. In: Spielberger CD, Sarason IG, eds. 1980;36:427–432. Stress and anxiety, Vol. I. Washington, DC: (21) Mayzner MS, Jr., Sersen E, Tresselt ME. Hemisphere Publishers, 1975:115–143. The Taylor Manifest Anxiety Scale and (35) Gilbert AR. Increased diagnostic value of intelligence. J Consult Psychol the Taylor Manifest Anxiety Scale by use 1955;19:401–403. of response latency. Psychol Rep (22) Jessor R, Hammond KR. Construct 1967;20:63–67. 286 Measuring Health The Hamilton Anxiety Rating Scale p55). Bech et al. offered detailed descriptions of (Max Hamilton, 1959) the scale levels for each symptom (3, pp19–22). The 14 scores are summed to give an overall Purpose score ranging from 0 to 56, with higher scores This is a clinical rating of the severity of anxiety indicating higher levels of anxiety. Occasionally, symptoms; it was designed for use with patients separate scores for psychic items (clusters 1–6) already diagnosed with anxiety neurosis, rather and somatic items (7–13) are calculated (4), and than for detecting anxiety in patients with other some have argued that the behavioral ratings diagnoses (1, p50). should not be included in the overall score (5, p76). A total score between zero and five repre- Conceptual Basis sents no anxiety; six to 14 suggests minor anxi- Hamilton developed the HARS within a diag- ety, whereas scores of 15 or higher indicate nostic tradition that distinguished between anxi- major anxiety (3, p10). If a single cutting-point ety as a normal reaction to danger, anxiety as a is required, scores of 14 and higher are often pathological condition not related to stress, and taken to designate a case. anxiety as a state or broad syndrome that he termed “anxiety neurosis” (2). Although in- tended as a severity rating, Hamilton also used Reliability his scale to distinguish empirically between anx- In Hamilton’s original study, three psychiatrists iety as a pathological mood, versus a state (or worked in pairs to rate patients; the mean of the neurosis). Evidence from his early studies sug- correlations between them was 0.89; t-tests gested a quantitative, rather than a qualitative, showed no significant differences between the distinction between them in terms of their sever- raters (1, p51). Other studies have reported sim- ity (2). ilar findings: inter-rater agreement coefficients of 0.82 (6, p285), 0.92 (7, p356), 0.92 (8, pS347), and an intraclass correlation of 0.98 (9, p176). Description In a study of eight raters, Spearman correlations In the original version of the HARS, a clinician among the ratings ranged from 0.74 to 0.93; interviewed the patient and rated the presence among the four most experienced raters, the av- and severity of 13 categories of anxiety symp- erage correlation was 0.90 (10, p166). The only toms and then rated the patient’s behavior at the outlier from the picture of very high inter-rater interview (1, Table 2). The scale content was se- agreement was a study producing an intra-class lected on the basis of clinical experience; roughly coefficient of 0.74 (4, p63). equal weight was given to psychic and somatic Estimates of internal consistency include al- symptoms (1, p53). Hamilton listed between pha coefficients of 0.92 (11, p489), 0.85 (12, two and eight examples of symptoms to consider Table 3), 0.86 (7, p356), and 0.89 (8, pS347). under each heading—see Exhibit 6.2. Hamilton Beneke reported a Spearman-Brown reliability later split the category of general somatic symp- coefficient of 0.83 (13, p252). Item-total corre- toms into muscular and sensory classes (see cate- lations, however, vary widely across the sections. gories 7 and 8 in Exhibit 6.2), making 14 In a 1983 study, Gjerris et al. obtained item- categories (the two behavioral ratings in the Ex- total correlations ranging from −0.04 to 0.67 hibit are scored together). (10, Table 2). Other estimates include a range The sequence of the interview, and the pre- from 0.20 to 0.52 (13, Table 3), 0.29 to 0.68 (7, cise question phrasing are left to the clinician. p356), and 0.76 to 0.98 (9, p176). Each cluster of symptoms is rated on a five- Test-retest reliability estimates include 0.92 point scale running from grade 0 (none), 1 (8, pS347), 0.96 (11, p489), and an intra-class (mild), 2 (moderate intensity), 3 (severe), to correlation of 0.86 (9, p176). Stability after one grade 4 (very severe or grossly disabling) (1, year was 0.64 (7, p358). Exhibit 6.2 The Hamilton Anxiety Rating Scale Symptoms of anxiety states 1. Anxious mood 8. General somatic (sensory) Tendency to sweat Worries Tinnitus Giddiness Anticipation of the worst Blurring of vision Tension headache Apprehension (fearful Hot and cold flushes Raising of hair anticipation) Feelings of weakness Irritability Pricking sensations 14. Behaviour at interview (general) 2. Tension 9. Cardiovascular symptoms Tense, not relaxed Feelings of tension Tachycardia Fidgeting: hands, Fatiguability Palpitations picking fingers, Inability to relax Pain in chest clenching, tics, Startle response Throbbing of vessels handkerchief Moved to tears easily Fainting feelings Restlessness: pacing Trembling Missing beat Tremor of hands Feelings of restlessness Furrowed brow 10. Respiratory symptoms Strained face 3. Fears Pressure or constriction in chest Increased muscular tone Of dark Choking feelings Sighing respirations Strangers Sighings Facial pallor Being left alone Dyspnoea Large animals, etc. (psychological) Traffic 11. Gastro-Intestinal Symptoms Swallowing Crowds Difficulty in swallowing Wind Belching Dyspepsia: High resting pulse rate 4. Insomnia pain before and after meals Respiration rate over 20/ min. Difficulty in falling asleep burning sensations Brisk tendon jerks Broken sleep fullness Tremor Unsatisfying sleep and fatigue waterbrash Dilated pupils on waking nausea Exophthalmos Dreams vomiting Sweating Nightmares sinking feelings Eye-lid twitching Night terrors ‘Working’ in abdomen Borborygmi 5. Intellectual (cognitive) Looseness of bowels Difficulty in concentration Loss of weight Poor memory Constipation 6. Depressed mood 12. Genito-urinary symptoms 再 Loss of interest Frequency of micturition Lack of pleasure in hobbies Urgency of micturition Depression Amenorrhea Early waking Menorrhagia Diurnal swing Development of frigidity Ejaculation praecox 7. General somatic (muscular) Loss of erection Muscular pains and aches Impotence Muscular stiffness Muscular twitchings 13. Autonomic symptoms Clonic jerks Dry mouth Grinding of teeth Flushing Unsteady voice Pallor From Hamilton M. The assessment of anxiety stated by rating. British Journal of Medical Psychology 1959;32:54–55. With permission. 287 288 Measuring Health Validity trait subscale in one study (7, p357), but only Hamilton’s original factor analysis identified 0.23 in another (12, Table 4). three factors of which only the first two were in- Responsiveness to change was higher for the terpretable (1; 2). They reflected psychic and so- psychic items than the somatic; items 11 and 12 matic anxiety (1, Table 2). This two-factor in particular were not sensitive to change follow- pattern has been supported in several subsequent ing treatment (4, p64). Maier et al. pointed out analyses (14, Table 2; 15, Table 2; 7, Table 5). that the somatic items may reflect side effects of One study subdivided the somatic factor into medication or change in physical health rather two components (13, Table 4), whereas a four- than in anxiety (4, p67). factor solution has been published that also sep- arated symptoms of depression (16, Table 1). Alternative Forms Reflecting the two-factor solution, and also the Because of the lack of precision in the original wide range of item-total correlations, Maier et administration instructions, several authors al. found that the HARS did not fit a single- have suggested clarifications of the HARS (3; 9; dimensional Rasch measurement model (4, 19; 20). The ECDEU assessment manual in- p64). However, the psychic and somatic sub- cludes a very slightly modified version of the scales, when tested separately, also did not fit the original scale that is commonly used in North Rasch model (15, p208). America (19, p194). Bruss et al. developed an Because the HARS includes symptoms of de- interview guide for the HARS that standardizes pression, there are evident concerns over its the probe questions and clarifies the scoring specificity; an equivalent concern arises for the (20, pp199–202). Correlations between raters Hamilton Rating Scale for Depression (HRSD). were 0.99 for the new form, compared to 0.74 Correlations between Hamilton’s anxiety and for the original version (20, Table 1). Shear et depression scales vary according to the sample al. developed a structured interview guide studied and include 0.62 (17, p478) for a mixed (SIGH-A) that is shown in Exhibit 6.3 (9). sample of anxiety and depression patients; 0.67 Scores on the SIGH-A were consistently about for panic patients (15, p209); 0.78 for a sample four points higher (more anxiety) than on the of anxiety patients (6, p286), and 0.92 for a original scale, although the test-retest reliability mixed sample of unaffected and anxiety patients was similar for each (intraclass correlation of (6, p286). Clark and Watson quoted an average 0.89 for the SIGH-A compared to 0.86 for the correlation of 0.77 (18, p323). Riskind et al. original). Inter-rater reliability for the SIGH-A analysed the items from the HARS and HRSD was 0.99, and the item-total correlations for together in a factor analysis that produced clear items 1–13 were high, ranging from 0.91 to anxiety and depression factors; however, several 0.99 (9, p176). The SIGH-A and the original items from each scale loaded on the wrong fac- scale correlated 0.77. tor—notably items 4, 5, and 6 from the HARS Because coverage overlaps between Hamil- (17, Table 1). ton’s anxiety and depression scales, Riskind et Sensitivity was 74% for detecting anxiety al. reconstructed both to create more specific disorders, and 87% for affective disorders in forms (17). This led to a 16-item anxiety scale general, whereas specificity was 100% (11, which retained items 1–3, 7–11, 13 and 14 from Table 5). Eight items contributed significantly the HARS, and added six others from the HRSD to distinguishing between anxious and de- (see 21, Table 1). In the Riskind et al. study, the pressed patients in a discriminant analysis (17, correlation between the anxiety and depression Table 2). scales fell from 0.62 to 0.15 for the recon- Convergent validity correlations include 0.53 structed versions, whereas coefficient alpha for with the Beck Anxiety Inventory (9, p177); 0.63 the HARS rose from 0.78 to 0.83 and its dis- with the Covi anxiety scale, and 0.75 with a criminant ability also improved (17, p478). Sub- clinician’s global rating of anxiety (4, p63). sequent studies have not fully replicated these HARS total scores correlated 0.58 with the STAI findings, however. Beck et al. compared the orig- Exhibit 6.3 The Structured Interview Guide for the Hamilton Anxiety Rating Scale To elicit the information necessary for assigning severity ratings, the interviewer must assess the frequency of occurrence, degree of distress, and degree of interference associated with symptoms. The following questions are recommended for this assessment: A. Have you had the symptom every day? If NO, Have you had the symptom more days than not? B. How much does the symptom bother you? C. How much does it interfere with your life? Starting the interview. Begin the interview with an introduction, describing the scale and its purpose in a way that is relevant for the specific patient and for the specific assessment. For example, for the first administration, one might say: “As you know, we have diagnosed your condition as an anxiety disorder. We are now going to be asking you a number of questions about different aspects of your anxiety. Together, they allow us to rate as accurately as possible the overall severity of your anxiety state. We will be rating anxiety severity in this way at different points in your treatment in order to decide how much the treatment is helping you.” This example is not meant as a script. The interviewer should introduce the scale in a way judged most comfortable for the patient and for his/her style. It is assumed that the interviewer has completed a previous diagnostic interview and is familiar with the patient’s general range of symptoms. If this is not true, the interviewer should preface the Hamilton Anxiety Scale by asking for a summary (five or ten minutes) of the patient’s specific worries, disturbing physical symptoms, duration of the syndrome, and its characteristics over time (e.g., Does it tend to wax and wane or has it been per- sistent since the onset?); The interviewer should also obtain a global statement on distress and impairment during the last week, and the cause of this distress. This information will provide the rater with a background or framework from which to conduct ratings. Although there are differences between studies, it is assumed that all ratings for the Hamilton Anxiety Rating Scale for patients with Panic Disorder will focus exclusively on times other than panic episodes. 1. What’s your mood been like this past week? Have you been anxious, nervous? Have you been worrying? Feeling something bad may happen? Feeling irritable? 0 – No anxious mood 1 – Mild worry or anxiety indicated only on questioning; no change in functioning 2 – Preoccupation with minor events, anxiety on as many days as not 3 – Near daily episodes of anxiety/worry with disruption of daily activities; daily preoccupation 4 – Nearly constant anxiety; significant role disruption 2. Have you been feeling tense? Do you startle easily? Cry easily? Easily fatigued? Have you been trembling or feeling restless or unable to relax? 0 – No tension 1 – Several days of mild tension or occasional (e.g., 1–2) episodes of exaggerated startle or labile mood 2 – Muscle tension or fatigue 50% of the time, or repeated (>2) episodes pf trembling, exaggerated startle, etc. 3 – Near daily muscle tension, fatigue and/or restlessness >75% of the time or persistent, disruptive symptoms 4 – Constant tension, restlessness, agitation, unable to relax in the interview 3. Have you been feeling fearful (phobic) of situations or events? For example, have you been afraid of the dark? Of strangers? Of being left alone? Of animals? Of being caught in traffic? Of crowds? Other fears? 0 – No fears 1 – Mild phobic concerns that do not cause significant distress or disrupt functioning 2 – Fears lead to distress or avoidance on one or more occasions 3 – Fears are an object of concern on a near daily basis (75%); patient may need to be accompanied by others to a fearful event 4 – Fears or avoidance that markedly affect function. Patient may avoid multiple situations even if accompanied; extensive agoraphobia 4. How has your sleeping been this week? Any difficulties falling asleep? Any problems with waking during the night? Waking early and not being able to return to sleep? Do you feel rested in the morning? Do you have disturbing dreams or nightmares? 0 – No sleep disturbance 1 – Mildly disrupted sleep (e.g., one to two nights of difficulties falling asleep or nightmares) (continued) 289 Exhibit 6.3 (continued) 2 – Several episodes of sleep disturbance that is regular but not persistent (e.g., over one-half hour falling asleep, nightmare or excessive AM fatigue) 3 – Persistent sleep disruption (more days than not), characterized by difficulty falling (e.g., over one hour) or staying asleep, restlessness, unsatisfying sleep or frequent nightmare, or fatigue 4 – Nightly difficulties with sleep onset or maintenance, or daily severe fatigue on waking in the AM 5. Have you had trouble concentrating or remembering things? 0 – No difficulties 1 – Infrequent episodes of forgetfulness or difficulty concentrating that are not distressing to the patient 2 – Recurrent episodes of forgetfulness or difficulty concentrating, or episodes of sufficient intensity to cause the patient recurrent concern 3 – Persistent concentration or memory impairment interferes with daily tasks 4 – Significant role impairment due to concentration difficulties 6. Have you been feeling depressed? Have you lost interest in things? Do you get pleasure from friends or hobbies? 0 – No depression 1 – Occasional or mild blue or sad mood, or reports of decreased enjoyment of activities 2 – Sad or blue mood or disinterest 50% of the time, mood does not generally interfere with functioning 3 – Persistent depressed mood or loss of pleasure, mood is significantly distressing to the patient or may be evident to others 4 – Daily evidence of severe depression with significant role impairment 7. Have you been experiencing aches, pains or stiffness in your muscles? Have you experienced muscle twitching or sudden muscle jerks? Have you been grinding your teeth? Have you had an unsteady voice? 0 – No muscular symptoms 1 – Infrequent presence of one or two symptoms, no significant distress 2 – Mild distress over several symptoms or moderate distress over a single symptom 3 – Symptoms occur on more days than not, symptoms are associated with moderate to severe distress and/or regular attempts at symptom control by limiting activities or taking medications 4 – Daily or near daily episodes of symptoms that cause the patient significant distress and lead to restriction of activities or repeated visits for medical attention 8. Have you been experiencing ringing in your ears, blurred vision, hot or cold flashes, feelings of weakness or prickling sensations? (Has this occurred at times other than during a panic attack?) 0 – No symptoms 1 – Infrequent presence of one or two symptoms, no significant distress 2 – Mild distress over several symptoms or moderate distress over a single symptom 3 – Symptoms occur on more days than not, symptoms are associated with moderate to severe distress and/or regular attempts at symptom control by limiting activities or taking medications 4 – Daily or near daily episodes of symptoms that cause the patient significant distress and lead to restriction of activities or repeated visits for medical attention 9. Have you had episodes of a racing, skipping or pounding heart? How about pain in your chest or fainting feelings? (Has this occurred at times other than during a panic attack?) [Same answer scale as for Question 8] 10. Have you been having trouble with your breathing? For example, pressure or constriction in your chest, choking feelings, sighing or feeling like you can’t catch your breath? (Has this occurred at times other than during a panic attack?) [Same answer scale as for Question 8] 11. Have you had any difficulties with stomach pain or discomfort? Nausea or vomiting? Burning or rumbling in your stomach? Heartburn? Loose bowels? Constipation? Sinking feeling in your stomach? (Has this occurred at times other than during a panic attack?) 0 – No symptoms 1 – Infrequent and minor episodes of gastric discomfort, constipation, or loosening of bowels, fleeting nausea 2 – An episode of vomiting or recurrent episodes of abdominal pain, loosening of bowels, difficulty swallowing, etc. 290  Anxiety 291 Exhibit 6.3 3 – Symptoms more days than not that are very bothersome to the patient or lead to concerns over eating, bathroom availability, or use of medication 4 – Daily or near daily episodes of symptoms that cause the patient significant distress and lead to restriction of activities or visits for medical attention 12. Have you been experiencing urinary difficulties? For example, have you had to urinate more frequently than usual? Have you had more urgency to urinate? Have you had decreased sexual interest? FOR WOMEN: Have your periods been regular? Have you experienced a change in your ability to have an orgasm? FOR MEN: Have you had trouble maintaining an erection? Ejaculating prematurely? 0 – No symptoms 1 – Infrequent and minor episodes of urinary symptoms or mild changes in sexual interest 2 – Urinary symptoms several days during the week, occasional difficulties with sexual function 3 – Urinary or sexual symptoms more days than not, amenorrhea 4 – Daily urinary or sexual symptoms that lead to distress and medical care seeking 13. Have you been experiencing flushing in your face? Getting pale? Lightheadedness? Have you been having tension headaches? Have you felt the hair rise on your arms, the back of your neck or head, as though something had frightened you? (Has this occurred at times other than during a panic attack?) 0 – No symptoms 1 – Mild symptoms occurring infrequently 2 – Symptoms occurred several times during the week and were bothersome 3 – Near daily symptoms with distress or embarrassment about the symptoms 4 – Daily symptoms that are a focus of distress and impair function (e.g., daily headaches or lightheadedness leading to limitation of activities) 14. Rate Interview Behaviour: Fidgeting, restlessness or pacing, tremor of hands, furrowed brow, strained face, sighing or rapid respirations, facial pallor, frequent swallowing, etc. 0 – No apparent symptoms 1 – Presence of one or two symptoms to a mild degree 2 – Presence of several symptoms of mild intensity or one symptoms of moderate intensity 3 – Persistent symptoms throughout the interview 4 – Agitation, hyperventilation, difficulty completing the interview From Shear MK et al. Reliability and validity of a structured interview guide for the Hamilton Anxiety Rating Scale (SIGH-A). Depression and Anxiety 2001;13:166–178. With permission. inal and revised versions of the scales; the revised also undertook replication analyses of the re- versions showed a lower correlation (0.74 com- vised version, again without fully replicating pared with 0.92), but still much higher than Riskind’s results. There was no difference in al- Riskind’s result. The revised version provided pha coefficients between the original and revised slightly better discrimination between anxious anxiety scales; the revised anxiety and depres- and normal respondents (6, p286). Moras et al. sion scales correlated 0.58; correlations with subsequently tested the reconstructed scales on a other anxiety measures were similar for original sample of patients with anxiety disorders and and revised versions. However, the discriminant confirmed the improvements in internal consis- ability of the revised scale was superior (12, tency of each scale but obtained a correlation of pp120–122). 0.61 between the scales (21, p225). They also Bech et al. noted the absence of items cover- found little advantage in the ability of the recon- ing panic symptoms, making it difficult to distin- structed anxiety scale to discriminate patients guish between panic attacks and generalized or with anxiety from others with both anxiety and persistent anxiety (3, p9). Accordingly, they pro- depression (Table 2). Finally, Diefenbach et al. posed modifications to the Hamilton anxiety 292 Measuring Health scale to cover panic attacks (the HAS-P); this in- Criticisms focus on shortcomings of the dicates whether the DSM-III criteria for anxiety HARS as a classification instrument. Its scope is have been met. They also proposed a version to limited and the conceptual framework on which assess generalized anxiety (the HAS-G) (3, it is based is no longer current. The HARS was pp9–14). developed many years before criteria such as the Snaith et al. proposed a six-item abbreviation DSM-III, and changes in diagnostic tradition that retained the most discriminating items from have replaced the concept of anxiety neurosis the HARS and added one on feelings of being with a set of disorders that include panic, phobic keyed up (psychic tension); they named this in- disorders, and more generalized anxiety; these strument the Clinical Anxiety Scale (22). The six entities are not distinguished in the HARS. items are: psychic tension, inability to relax, Nonetheless, the instrument appears to work on startle response, worries, apprehension, and patients with panic disorder; a study of panic feelings of restlessness (22, pp519–20). These disorder patients revealed the same factor struc- items proved more sensitive to change than the ture as obtained with other patients (15, p210). overall HARS (22, Table 4). Without doubt, the scale confounds symp- Kobak et al. described a computerized, self- toms of anxiety and depression, and whether administered version that contains 31 questions this limits its value as a severity measure contin- and produces scores for the 14 categories in the ues to be debated. Hamilton deliberately in- original scale (11). Ratings of the occurrence and cluded depression items in the HARS, arguing severity of each symptom are included in the fi- that anxiety often produces depressive symp- nal score. Scores on the computerized and the toms, and that these would help to indicate the clinical versions were found to be equivalent for severity of the anxiety. Because this contami- several, but not all, categories of respondents nates the scale and makes it hard to interpret the (11, Table 2), and sensitivity and specificity were scores, Riskind’s “reconstruction” of the HARS slightly lower for the computerized version (11, to eliminate the depression items appeared sensi- Tables 5 and 6). The computerized version mini- ble, and initial results showed that the correla- mizes clinician time, offers a more standardized tion between the revised Hamilton anxiety and instrument, and provides automated data cap- depression scales was greatly reduced (18, ture. p328). However, two subsequent studies showed Translations of the HARS include German substantial correlations between the recon- (13), Spanish (8), and Polish (23). structed scales, which seems to suggest that (as Hamilton originally argued) anxiety and depres- Commentary sion often do occur together, so that he may Hamilton’s anxiety scale has been widely used have been right to include the depression items and holds an historical place in the field of anxi- in a scale that measures the severity of anxiety. ety research. In reviewing the commentaries on Other criticisms of the HARS point out that the the HARS, it is important to remember that its somatic items (categories 7–12 in the exhibit) purpose was to measure the severity of anxiety in may reflect change in physical health rather than previously diagnosed patients. It was not intended in level of anxiety, although the findings appear to discriminate between anxiety and depression, inconsistent (10; 13; 17). Probably these items or to diagnose anxiety. When considered as a have low specificity but may form sensitive indi- severity measure, the HARS has several positive cators of anxiety for some patients and not for features. As a clinical rating scale, it is inherently others. Perhaps because it is a rating scale that is more accurate than a self-report questionnaire. used in clinical research, there are no population Inter-rater reliability is often high, especially for reference values. the structured forms, and it shows reasonable The lack of clarity in the original instructions sensitivity to change over time (4). It can be used and the overlap between items (e.g., item 6 on with elderly respondents (6) and adolescents (7) early waking and item 4 on broken sleep) have alike. stimulated various revisions of the HARS.  Anxiety 293 Hamilton (who wrote in an endearingly self- Anxiety Rating Scale (HARS) and the deprecating style) recognized the limitations of Montgomery-Asberg depression rating scale his scale, noting that “Some of the variables are (MADRS). Eur Neuropsychopharmacol obviously a rag-bag of oddments and need fur- 2001;11(suppl 3):S346–S347. ther investigation” (1, p53). (9) Shear MK, Vander Bilt J, Rucci P, et al. Reliability and validity of a structured In its structured versions, the HARS is likely interview guide for the Hamilton Anxiety to continue to be used as a rating scale for assess- Rating Scale (SIGH-A). Dep Anx ing the severity of anxiety, although as it requires 2001;13:166–178. a 30-minute interview, it appears relatively im- (10) Gjerris A, Bech P, Bøjholm S, et al. The practical. The HARS should not be interpreted as Hamilton Anxiety Scale. Evaluation of a classification or a diagnostic instrument to dis- homogeneity and inter-observer reliability tinguish anxiety from depression, for which the in patients with depressive disorders. J Beck Anxiety Inventory or Hospital Anxiety and Affect Disord 1983;5:163–170. Depression Scales perform better. (11) Kobak KA, Reynolds WM, Greist JH. Development and validation of a computer- administered version of the Hamilton Anxiety Scale. Psychol Assess References 1993;5:487–492. (1) Hamilton M. The assessment of anxiety (12) Diefenbach GJ, Stanley MA, Beck JG, et al. states by rating. Br J Med Psychol Examination of the Hamilton scales in 1959;32:50–55. assessment of anxious older adults: a (2) Hamilton M. Diagnosis and rating of replication and extension. J Psychopathol anxiety. Br J Psychiatry 1969; Special Behav Assess 2001;23:117–124. Publication #3:76–79. (13) Beneke M. Methodological investigations (3) Bech P, Kastrup M, Rafaelsen OJ. Mini- of the Hamilton Anxiety Scale. compendium of rating scales for states of Pharmacopsychiatry 1987;20:249–255. anxiety, depression, mania, schizophrenia, (14) Beck AT, Steer RA. Relationship between with corresponding DSM-III syndromes. the Beck Anxiety Inventory and the Acta Psychiatr Scand 1986;73(suppl Hamilton Anxiety Rating Scale with 326):1–37. anxious outpatients. J Anx Disord (4) Maier W, Buller R, Philipp M, et al. The 1991;5:213–223. Hamilton Anxiety Scale: reliability, validity (15) Bech P, Allerup P, Maier W, et al. The and sensitivity to change in anxiety and Hamilton scales and the Hopkins Symptom depressive disorders. J Affect Disord Checklist (SCL-90): a cross-national 1988;14:61–68. validity study in patients with panic (5) Bech P. Psychometric development of the disorders. Br J Psychiatry Hamilton scales: the spectrum of 1992;160:206–211. depression, dysthymia, and anxiety. In: (16) Pancheri P, Picardi A, Pasquini M, et al. Bech P, Coppen A, eds. The Hamilton Psychopathological dimensions of scales. Berlin: Springer-Verlag, depression: a factor study of the 17-item 1990:72–79. Hamilton depression rating scale in (6) Beck JG, Stanley MA, Zebb BJ. unipolar depressed outpatients. J Affect Effectiveness of the Hamilton Anxiety Disord 2002;68:41–47. Rating Scale with older generalized anxiety (17) Riskind JH, Beck AT, Brown G, et al. patients. J Clin Geropsychology Taking the measure of anxiety and 1999;5:281–290. depression: validity of the reconstructed (7) Clark DB, Donovan JE. Reliability and Hamilton scales. J Nerv Ment Dis validity of the Hamilton Anxiety Rating 1987;175:474–479. Scale in an adolescent sample. J Am Acad (18) Clark LA, Watson D. Tripartite model of Child Adolesc Psychiatry 1994;33:354–360. anxiety and depression: psychometric (8) Lobo A, Chamorro L, Luque A, et al. evidence and taxonomic implications. J Validation into Spanish of the Hamilton Abnorm Psychol 1991;100:316–336. 294 Measuring Health (19) Guy W. Early clinical drug evaluation they did not indicate the nature of any disorder (ECDEU) assessment manual for so identified (3, p361). Given that it was in- psychopharmacology. DHEW Pub No. 76- tended for use with general medical patients, the 338 ed. Rockville, MD: US Department of HADS focused on relatively mild degrees of dis- Health, Education, and Welfare National order (items on suicidal thoughts were ex- Institute of Mental Health, 1976. cluded), and somatic items such as dizziness or (20) Bruss GS, Gruenberg AM, Goldstein headaches were deliberately not used (1; 2). At- RD, et al. Hamilton Anxiety Rating Scale Interview guide: joint interview tention was paid to selecting items that could and test-retest methods for interrater distinguish between anxiety and depression; to reliability. Psychiatry Res avoid overlaps, the depression scale was built on 1994;53:191–202. the concept of anhedonia: five of seven items re- (21) Moras K, DiNardo PA, Barlow DH. fer to loss of pleasure. As outlined in the intro- Distinguishing anxiety and depression: duction to this chapter, anhedonia is central to reexamination of the reconstructed the concept of depression and is amenable to Hamilton scales. Psychol Assess treatment (3, p362). Zigmond and Snaith also 1992;4:224–227. excluded items relating to the cognitive compo- (22) Snaith RP, Baugh SJ, Clayden AD, et al. nents of depression (e.g., hopelessness, low self- The Clinical Anxiety Scale: an instrument esteem, suicidal ideation), because these may derived from the Hamilton Anxiety Scale. Br J Psychiatry 1982;141:518–523. reflect cognitive processes rather than depression (23) Malyszczak K, Kiejna A, Grzesiak M. itself (4). The content of the anxiety scale was [Factorial structure of Hamilton Anxiety guided by the Present State Examination (3, Rating Scale]. Psychiatr Pol p362); three items refer to fear or panic and four 1998;32:771–779. cover generalized anxiety. Description The Hospital Anxiety and The HADS is a patient-completed, 14-item scale, Depression Scale with seven items measuring anxiety (HADS-A) (A. Zigmond and R.P. Snaith, 1983) and seven measuring depression (HADS-D). Scores range from 0 to 21 for each scale; higher Purpose scores represent more distress. The time frame The Hospital Anxiety and Depression Scale refers to mood during the past week so as to (HADS) was designed as a self-assessment in- avoid the influence of possible immediate strument for detecting clinically significant de- changes, such as those due to the stress of at- pression and anxiety in patients attending tending the clinic appointment (3, pp365–66). outpatient medical clinics, and for discriminating The scale is shown in Exhibit 6. 4. between anxiety and depression. It has been Various cutting-points have been quoted. In widely used as a screening instrument outside of the original description of the HADS, Zigmond the hospital setting, and also for rating psychi- and Snaith proposed scores of eight to ten (on atric patients (1). each scale) to represent possible cases, and 11 or more for definite cases (3, p363). A cutting- Conceptual Basis point of 14/15 may be used to represent severe The HADS originated from a deceptively simple disorder (5, p21). In subsequent publications, a request from a physician for a way to identify cutting-point of 8/9 has most often been taken those seemingly depressed patients whose condi- to indicate mild intensity for the two scales; tion might be expected to improve with antide- 11/12 indicates severe. The Bjelland et al. review pressant medication (2, p262). The available reported that 8/9 for both anxiety and depres- screening instruments, such as the General sion scales represented the optimal cutting point Health Questionnaire, were inadequate because (6, p71). Exhibit 6.4 The Hospital Anxiety and Depression Scale Note: the right-hand column is not shown to the respondent. It indicates the sub-scale to which each item belongs and the score for each response Doctors are aware that emotions play an important part in most illnesses. If your doctor knows about these feelings he will be able to help you more. This questionnaire is designed to help your doctor to know how you feel. Read each item and underline the reply which comes closest to how you have been feeling in the past week. Don’t take too long over your replies; your immediate reaction to each item will probably be more accurate than a long thought out response. 1 I feel tense or ‘wound up’: A 8 I feel as if I am slowed down: D Most of the time 3 Nearly all the time 3 A lot of the time 2 Very often 2 From time to time, occasionally 1 Sometimes 1 Not at all 0 Not at all 0 2 I still enjoy the things I used to enjoy: D 9 I get a sort of frightened feeling like ‘butterflies’ in the stomach: A Definitely as much 0 Not at all 0 Not quite so much 1 Occasionally 1 Only a little 2 Quite often 2 Hardly at all 3 Very often 3 3 I get a sort of frightened feeling as if something awful is about to happen: A 10 I have lost interest in my appearance: D Very definitely and quite badly 3 Definitely 3 Yes, but not too badly 2 I don’t take so much care as I should 2 A little, but it doesn’t worry me 1 I may not take quite as much care 1 Not at all 0 I take just as much care as ever 0 11 I feel restless as if I have to be on the move: A 4 I can laugh and see the funny side of things: D Very much indeed 3 As much as I always could 0 Quite a lot 2 Not quite so much now 1 Not very much 1 Definitely not so much now 2 Not at all 0 Not at all 3 12 I look forward with enjoyment to things: D 5 Worrying thoughts go through my mind: A As much as I ever did 0 A great deal of the time 3 Rather less than I used to 1 A lot of the time 2 Definitely less than I used to 2 From time to time but not too often 1 Hardly at all 3 Only occasionally 0 13 I get sudden feelings of panic: A 6 I feel cheerful: D Very often indeed 3 Not at all 3 Quite often 2 Not often 2 Not very often 1 Sometimes 1 Not at all 0 Most of the time 0 14 I can enjoy a good book or radio or TV 7 I can sit at ease and feel relaxed: A programme: D Definitely 0 Often 0 Usually 1 Sometimes 1 Not often 2 Not often 2 Not at all 3 Very seldom 3 Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavica 1983;67:367–370. With permission. 295 296 Measuring Health Table 6.2 Alpha Internal Consistency Coefficients for the Hospital Anxiety and Depression Scale Sample Studied Anxiety Depression Reference Cancer patients 0.93 0.90 (7, p257) Cancer patients 0.87 0.83 (8, p29) Breast cancer patients 0.82 0.79 (9, Table 2) Depressed patients 0.76 0.77 (1, p121) Self-help depression group 0.84 0.83 (4, p132) HIV patients 0.89 0.89 (10, p359) Myocardial infarction patients 0.85 0.82 (9, Table 2) Female cardiac patients 0.85 0.80 (11, p379) Myocardial infarction patients 0.76 0.72 (12, p196) Stroke patients 0.76 0.70 (9, Table 2) General population 0.84 0.82 (13, p283) General population 0.80 0.76 (14, Table 4) Reliability 10; 13; 14; 17). A consistent finding is that item Bjelland et al. reported alpha internal consis- seven (‘I can sit at ease and feel relaxed’) shows tency results from 18 studies (6, Table 1); Table cross-loadings on the depression factor. Item six 6.2 offers an abbreviated summary. (‘I feel cheerful’) also loaded on both factors in Item 8 has shown a low item-scale correla- several analyses. The variance explained is gener- tion; removing it improved the alpha from 0.77 ally somewhat low, about 50%, although this is to 0.80 in one study (1, p121). Item-total corre- comparable with results for other scales (7, lations for the anxiety items ranged from 0.36 to p258). The variance explained may be higher in 0.64; for depression the range was from 0.45 to patients with mental problems (70%) than in the 0.72 (4, p132). In a second study, rho values for general population (57%) (14, p543). Finally, an item-total correlations on the anxiety scale oblique factor rotation may be appropriate as ranged from 0.41 to 0.76; values for the depres- the two factors correlate (1; 7; 11; 13). A review sion items ranged from 0.30 to 0.60 (15, p357). of 21 studies found a mean correlation of 0.56 Two-month test-retest reliability was 0.79 between the anxiety and depression subscales (6, for the anxiety scale, 0.63 for the depression, p71); Herrmann concluded that the correlation and 0.78 for the full scale (11, p379). Reliability represented an overlap between anxiety and de- calculated using LISREL modeling was 0.74 for pression rather than weakness in the scales (5). the HADS total score (16, p862). Herrmann re- Exceptions to the two-factor pattern included ported retest reliability (anxiety scale) of 0.84 at a Belgian study that identified only a single fac- up to two weeks, 0.73 at two to six weeks, and tor (18); this may have reflected a flawed transla- 0.70 at over six weeks. The equivalent figures for tion of the questions (7; 10, p362). One the depression scale were 0.85, 0.76, and 0.70, three-factor solution split the anxiety items into respectively (5, Table 2). state and trait aspects (12, Table 3). Similarly, a French study (N = 2,669) split the anxiety items Validity into Psychic anxiety (“I get a sort of frightened factorial validity. Of 19 studies reporting fac- feeling;” “I get sudden feelings of panic”) and tor analyses, 11 (total N = 14,588) obtained a Psychomotor agitation (“I can sit at ease;” “I two-factor solution; five studies found three- feel restless”) (19, Table 1). Dunbar et al. tested factor solutions (N = 3,459) and two found four- a three-factor solution based on Clark and Wat- factor solutions (6, p70). Thus, the general son’s tripartite model of anxiety and depression. finding is that the items fall onto two factors cor- The results lent some support for the notion of a responding to anxiety and depression (1; 5; 7; hierarchical model in which a general negative  Anxiety 297 affectivity factor predicted separate anxiety and der the curve 0.93 versus 0.84) (24, p403). depression components (20). Although the Bjelland et al. summarized sensitivity and HADS was never designed with this model in specificity results from 24 studies and estimated mind, Dunbar et al. suggested that deriving three a mean sensitivity and specificity of about 80% scores rather than two may be valuable (20, (6, p71). The optimal cutting point was 8/9, al- p92). A four-factor solution derived from a though 9/10 might appear superior for cancer small sample of mentally healthy people in Swe- patients (6, p73). However, sensitivity varied den did not provide easily interpretable results widely between studies, from about 65% to well (21, Table 3); another Swedish study identified over 90% (Table 2); no clear pattern emerged by three or four factors in different subgroups (22). patient type or diagnostic criterion, save that the HADS performs well as a case-finding instru- criterion validity. During the original develop- ment for general practice patients, with areas un- ment of the HADS, Zigmond and Snaith re- der the ROC curve of 0.84 to 0.96 (6, p74). The ported correlations between HADS scores and scale appeared to function less well for palliative clinical ratings. Convergent correlations were care patients, in whom it may be difficult to dis- significant (r = 0.54 for anxiety and 0.79 for de- tinguish what may be considered appropriate pression), whereas discriminant correlations sadness in a person whose life is ending (25). To were not (r = 0.19 for HADS anxiety and clini- illustrate the disparate results, a range of studies cal depression, and 0.08 for HADS depression may be cited. Using a cutting-point of 8/9 with with clinical anxiety) (3, Table 2). Clarke et al. hospital outpatients, sensitivity was 82% at a compared the HADS, the General Health Ques- specificity of 94% for depression; equivalent fig- tionnaire (GHQ), and the Beck Depression In- ures were 70% and 68% for anxiety (26, Table ventory (BDI) against DSM-III-R diagnoses 1). For depression, sensitivity was 88% and among hospitalized medical patients (23). The specificity 78% in a sample of people presenting results of the comparison varied according to the with evidence of deliberate self-harm (27). In statistic used; sensitivity and specificity figures a sample of breast cancer patients, sensitivity suggested that the GHQ was superior in detect- for both the anxiety and depression scales was ing depression, followed by the BDI and the 75%, equal to that of the Rotterdam Symptom HADS (23, Table 3). If the area under the ROC Checklist (RSCL). Specificity was 90% for the curve or kappa statistics are considered, how- HADS-A, but 75% for the depression scale; it ever, the HADS total score performed slightly was 80% for the RSCL (28, Table 1). Using the less well than the GHQ in detecting depression 8/9 cutting-point, Silverstone found a sensitivity and diagnoses of all types, but considerably bet- of 100% and specificity of 73% for the HADS- ter in identifying anxiety (23, Table 2). The D in medical patients, but sensitivity was only HADS depression score also performed better 80% and specificity was 28% for psychiatric pa- than the BDI in the same analysis. A similar tients (29, Table 2). In a study of stroke patients, analysis again illustrates how the choice between the area under the ROC curve for the depression instruments can vary according to the statistic subscale was 0.83, slightly higher than the value used to compare them; using the area under the of 0.79 for the BDI. Results were better for men ROC curve, the HADS total score was slightly (0.87) than for women (0.77) (30, pp389–90). superior to that of the GHQ-12, whereas sensi- tivity and specificity figures showed a slight ad- concurrent validity. Bjelland et al. reviewed vantage to the GHQ. The sensitivity figures were six studies that reported correlations between 72.3% for the HADS and 78.7% for the GHQ the HADS and the BDI. Correlations for the (16, Table 1). In a separate study of depression, HADS-A ranged from 0.61 to 0.83; curiously, the area under the ROC curve was 0.92 for the figures for the depression scale were lower, rang- HADS, compared with 0.81 for the GHQ-12; ing from 0.62 to 0.73 (6, Table 3). The same re- curiously, the anxiety scale of the HADS per- view reported five comparisons of the HADS formed better than the depression scale (area un- and the STAI, with results ranging from correla- 298 Measuring Health tions of 0.64 to 0.81 (HADS-A) and 0.52 to ety scores should be expected to correlate more 0.65 (HADS-D) (6, Table 3). strongly with another anxiety scale than with For hospital outpatients, the HADS depres- scores from depression scales. One would also sion score correlated 0.77 with the expect correlations between anxiety and depres- Montgomery-Åsberg Depression Rating Scale sion scores on the HADS to be lower than their (2, p264); equivalent figures were 0.70 in a sam- correlations with alternative scales measuring ple of psychiatric patients (31, Table 1) and 0.54 the same trait. A correlation between anxiety and 0.79 in elderly depressed patients (32). In a and depression scores may, of course, arise if the study of patients with rheumatoid arthritis, the two conditions coexist in the patients being HADS anxiety scale correlated 0.48 with the studied. anxiety score from the General Health Ques- Figures for the correlation between the anxi- tionnaire (GHQ-28), whereas the depression ety and depression scales vary widely, in part ac- scales correlated 0.46 (15, Table 1). For derma- cording to the sample studied. In one of the tology outpatients, overall scores from the original studies, Aylard et al. reported indepen- HADS correlated 0.75 with overall scores from dence of the two scales (correlation −0.04), the GHQ-12 (16, p862). Other findings also whereas they reported a correlation for the suggest variability: Bjelland’s review article cites equivalent scales of the GHQ-28 of 0.54 (2, correlations with the GHQ-28 ranging from p264). Clark and Watson reported an average 0.50 to 0.68; results were similar for the anxiety correlation for the HADS scales, based on two and depression scales (6, Table 3). studies, of only 0.11, which they attributed to A frequent finding is that the HADS total the inclusion of items related to anhedonia in the score shows higher correlations with anxiety or depression scale (35, p321). Numerous subse- depression criterion measures than do the sub- quent studies have found more substantial corre- scales (13; 18; 29; 33). For example, in a general lations between the HADS anxiety and population sample, the HADS-D correlated 0.71 depression scales, however. Among patients with with scores on the BDI, whereas the correlation rheumatoid arthritis, the correlation was 0.53; for the HADS total score was higher, at 0.73 (13, the equivalent correlation between the GHQ p283). Similarly, the HADS total score corre- anxiety and depression scales was 0.37 (15, lated more strongly with the two STAI scores Table 1). In a study of depressed patients, the (0.68 and 0.71) than the HADS anxiety scale did correlation was 0.46; the anxiety scale correlated (0.64 and 0.66). Finally, the correlations be- 0.47 with the Hamilton Rating Scale for Depres- tween the HAD-D and the STAI scores were sion (1, p119). In a sample of psychiatric pa- identical to those obtained with the HADS-A tients, the correlation between the anxiety and (13, p283). depressions scales was 0.49, whereas the anxiety In a mixed sample of psychiatric patients, the scale correlated 0.47 with the Montgomery- HADS-A correlated 0.70 with the anxiety scale Åsberg Depression Rating Scale (31, p135). of the Irritability-Depression-Anxiety (IDA) Other correlations between the anxiety and de- scale; the HADS-D correlated 0.78 with the cor- pression scores include 0.54 (12, p196). In a responding IDA scale. Equivalent correlations study of HIV patients, anxiety and depression with the Zung scales were lower: 0.54 for the sores correlated 0.63, a figure similar to the dis- anxiety scales and 0.65 for the depression scales criminant correlation of 0.68 between the (34, Table 3). HADS-A and the BDI, and the coefficient of 0.65 between the HADS-D with the STAI-S (10, Fig- discriminant validity. A goal of the HADS ure 1). In the same study, the convergent correla- was to distinguish anxiety from depression, and tions were only slightly higher, at 0.70 between several studies allow us to compare convergent HADS-D and the BDI, and 0.78 between HADS- and discriminant correlations. If the HADS can A and STAI-S. In a study of melanoma patients, distinguish anxiety from depression, then anxi- the patterns of correlations between the anxiety  Anxiety 299 and depression scales and a cancer disease- guish between members of a self-help depression specific scale showed logical differences. Depres- group who were taking antidepression medica- sion, for example, was more highly associated tions and others who were not (4, p134). than the anxiety score with physical problems and with difficulties with sexual function, Alternative Forms whereas the anxiety scale correlated with adverse Versions are available in Arabic (33; 37), Can- effects of treatment more strongly than did the tonese (38), Japanese (39), Dutch, French, He- depression scale (36, Table 5). brew, Swedish (13), Italian, Spanish (3), To address the question of whether responses Norwegian (14), and Canadian French (10). to the HADS may be influenced by physical ill- Herrmann provided references to studies validat- ness, Zigmond and Snaith compared scores for ing different linguistic versions (5, Table 6). patients with physical illness who were judged A computer-administered version using a by a psychiatrist to have no emotional illness touch-screen has been used (40) and was found with a comparison sample of healthy controls; to give results as valid as the paper version (41). there were no significant differences (3, p364). A study of patients with rheumatoid arthritis al- Reference Standards lowed for comparison between measures of pain Herrmann’s review refers to German norms (5, and joint stiffness and the anxiety and depres- p22). sion scores from the HADS and the GHQ-28. Three of four correlations were nonsignificant Commentary for the HADS (overall mean 0.18), whereas The HADS has become widely used, and two ar- three of four were significant for the GHQ ticles offer systematic reviews of its performance (mean 0.25), suggesting that the HADS is less in- (5; 6). A major attraction of the HADS is that it fluenced by physical health problems than is the was designed for use with clinical populations, GHQ (15, Table 3). Scores on the HADS ap- so it excludes items that might reflect physical peared to be somewhat less affected by the pres- illness. Savard et al., for example, underscored ence of varying levels of somatic illness than was this potential issue among HIV patients, for the BDI in a study of HIV patients (10, Table 5). whom both the illness and its treatment produce Finally, the HADS was included in a factor symptoms such as fatigue, sleep difficulties, or analysis with the Rotterdam Symptom Check- weight loss that mimic somatic symptoms of List, which contains 22 items on somatic symp- anxiety or depression (10). toms. The results produced psychological and The general consensus is that the HADS somatic factors; the HADS items all loaded on achieves its purpose. It is brief and simple to ad- the psychological factor, although three items minister, it does seem to separate anxiety from also cross-loaded on the somatic factor (9, Table depression, and it is not confounded by physical 1). illness. Convergent correlations with other scales Patients with medical problems may experi- appear good: it performs as well as the longer ence depressive symptoms as a result, and HADS BDI or GHQ instruments. Although developed depression scores do tend to increase with level for use in the hospital setting, the HADS also ap- of physical disability (15, Table 4). However, in pears to perform well in unselected populations, the same study there was no association between such as in general practice. To achieve this, it disability and the anxiety scale, suggesting that uses a narrow definition of depression that omits the HADS anxiety and depression scales do, in- concepts such as low self-esteem, hopelessness, deed, measure different reactions. or guilt (7, p258). Although the focus on anhe- The depression scale discriminated signifi- donia may be appropriate for the original pur- cantly between anxious and depressed psychi- pose of identifying treatable depression, it atric patients, but the anxiety scale did not (31, probably limits the validity of the HADS as a Table 2). The depression scale did not distin- general assessment of depression or as an out- 300 Measuring Health come measurement (7). The focus on anhedonia down” which could reflect physical illness, but may also limit applicability with palliative care empirical analyses suggest that they perform ad- patients for whom not experiencing pleasure equately (9). Item 7 (“I can sit at ease and feel re- does not imply morbidity (25, p995). There has, laxed”) appears to be nonspecific, in that it however, been debate over the use of the HADS consistently loads on both anxiety and depres- as a screening instrument in general practice; a sion factors. problem of high screen-positive rates has been noted (42), and Snaith has energetically re- Address minded us of the original purpose of the HADS, The Manual for the HADS is available from arguing that it should not be considered as a NFER Publications in England: www.nfer- general screening instrument, for which a scale nelson.co.uk. such as the General Health Questionnaire would be more appropriate (43; 44). The anxiety items include few somatic symptoms; the items on restlessness, agitation, tension, and worry may References represent Clark and Watson’s concept of negative (1) Flint AJ, Rifat SL. Factor structure of affect, rather than specifically anxiety (20, p81). the Hospital Anxiety and Depression Another curiosity has been a finding that breast Scale in older patients with major cancer patients report surprisingly low levels of depression. Int J Geriatr Psychiatry anxiety and depression; an investigation showed 2002;17:117–123. (2) Aylard PR, Gooding JH, McKenna PJ, et that this did not appear to be due to shortcom- al. A validation study of three anxiety and ings of the HADS (17). depression self-assessment scales. J Although the factor analyses generally distin- Psychosom Res 1987;31:261–268. guish the anxiety and depression subscales, these (3) Zigmond AS, Snaith RP. The Hospital intercorrelate and often show similar correla- Anxiety and Depression Scale. Acta tions with criterion variables: convergent and Psychiatr Scand 1983;67:361–370. discriminant correlations are often quite close. (4) Dagnan D, Chadwick P, Trower.P. This, of course, may mean merely that many pa- Psychometric properties of the Hospital tients experience both anxiety and depression, or Anxiety and Depression Scale with a that a general distress dimension underlies both; population of members of a depression indeed, the correlation between HADS-A and self-help group. Br J Med Psychol 2000;73:129–137. HADS-D was lower in a sample of strictly de- (5) Herrmann C. International experiences pressed (i.e., nonanxious) patients, although no with the Hospital Anxiety and Depression criterion validity was reported in that study (1). Scale—a review of validation data and Alternatively, Clarke et al. note that anhedonia, clinical results. J Psychosom Res on which the HADS depression scale is based, is 1997;42:17–41. not particularly sensitive or specific for depres- (6) Bjelland I, Dahl AA, Haug TT, et al. The sion among hospitalized medical patients, which validity of the Hospital Anxiety and explains why the HADS total score may perform Depression Scale. An updated literature better in detecting major depression than did the review. J Psychosom Res 2002;52:69–77. depression scale (23, p333). As proposed by (7) Moorey S, Greer S, Watson M, et al. The Clark and Watson (35), symptoms of anxiety factor structure and factor stability of the Hospital Anxiety and Depression Scale in and depression may also be considered as a spec- patients with cancer. Br J Psychiatry trum running from pure anxiety measures, 1991;158:255–259. through a mixed area in which symptoms over- (8) Skarstein J, Aass N, Fosså SD, et al. lap (as with HADS item 7), to items that reflect Anxiety and depression in cancer patients: pure depression (11, p380). relation between the Hospital Anxiety and Some doubts have been expressed over the in- Depression Scale and the European Organ- clusion of items such as “I feel as if I am slowed ization for Research and Treatment of  Anxiety 301 Cancer Core Quality of Life Questionnaire. Hospital Anxiety and Depression Scale in a J Psychosom Res 2000;49:27–34. large French primary care population (9) Johnston M, Pollard B, Hennessey P. suffering from major depression. Construct validation of the Hospital Psychiatry Res 2001;104:247–257. Anxiety and Depression Scale with clinical (20) Dunbar M, Ford G, Hunt K, et al. A populations. J Psychosom Res confirmatory factor analysis of the Hospital 2000;48:579–584. Anxiety and Depression Scale: comparing (10) Savard J, Laberge B, Gauthier JG, et al. empirically and theoretically derived Evaluating anxiety and depression in HIV- structures. Br J Clin Psychol infected patients. J Pers Assess 2000;39:79–94. 1998;71:349–367. (21) Andersson E. The Hospital Anxiety and (11) Roberts SB, Bonnici DM, Mackinnon AJ, Depression Scale. Homogeneity of the et al. Psychometric evaluation of the subscales. Soc Behav Pers Hospital Anxiety and Depression Scale 1993;21:197–204. (HADS) among female cardiac patients. Br (22) Brandberg Y, Bolund C, Sigurdardóttir V, J Health Psychol 2001;6:373–383. et al. Anxiety and depressive symptoms at (12) Martin CR, Thompson DR. A different stages of malignant melanoma. psychometric evaluation of the Hospital Psycho-Oncology 1992;1:71–78. Anxiety and Depression Scale in coronary (23) Clarke DM, Smith GC, Herrman HE. A care patients following acute myocardial comparative study of screening instruments infarction. Psychology Health Med for mental disorders in general hospital 2000;5:193–201. patients. Int J Psychiatry Med (13) Lisspers J, Nygren A, Söderman E. 1993;23:323–337. Hospital Anxiety and Depression Scale (24) Le Fevre P, Devereux J, Smith S, et al. (HAD): some psychometric data for a Screening for psychiatric illness in the Swedish sample. Acta Psychiatr Scand palliative care inpatient setting: a 1997;96:281–286. comparison between the Hospital Anxiety (14) Mykletun A, Stordal E, Dahl AA. Hospital and Depression Scale and the General Anxiety and Depression (HAD) scale: Health Questionnaire-12. Palliat Med factor structure, item analysis and internal 1999;13:399–407. consistency in a large population. Br J (25) Lloyd-Williams M, Friedman T, Rudd N. Psychiatry 2001;179:540–544. An analysis of the validity of the Hospital (15) Chandarana PC, Eals M, Steingart AB, et Anxiety and Depression Scale as a al. The detection of psychiatric morbidity screening tool in patients with advanced and associated factors in patients with metastatic cancer. J Pain Symptom Manage rheumatoid arthritis. Can J Psychiatry 2001;22:990–996. 1987;32:356–361. (26) Barczak P, Kane N, Andrews S, et al. (16) Lewis G, Wessely S. Comparison of the Patterns of psychiatric morbidity in a General Health Questionnaire and the genito-urinary clinic. A validation of the Hospital Anxiety and Depression Scale. Br Hospital Anxiety Depression Scale (HAD). J Psychiatry 1990;157:860–864. Br J Psychiatry 1988;152:698–700. (17) Osborne RH, Elsworth GR, Sprangers (27) Hamer D, Sanjeev D, Butterworth E, et al. MAG, et al. The value of the Hospital Using the Hospital Anxiety and Depression Anxiety and Depression Scale (HADS) for Scale to screen for psychiatric disorders in comparing women with early onset breast people presenting with deliberate self- cancer with population-based reference harm. Br J Psychiatry 1991;158:782–784. women. Qual Life Res 2004;13:191–206. (28) Hopwood P, Howell A, Maguire P. (18) Razavi D, Delvaux N, Farvacques C, et al. Screening for psychiatric morbidity in Screening for adjustment disorders and patients with advanced breast cancer: major depressive disorders in cancer in- validation of two self-report questionnaires. patients. Br J Psychiatry 1990;156:79–83. Br J Cancer 1991;64:353–356. (19) Friedman S, Samuelian J-C, Lancrenon S, (29) Silverstone PH. Poor efficacy of the et al. Three-dimensional structure of the Hospital Anxiety and Depression Scale in 302 Measuring Health the diagnosis of major depressive disorder computers the way forward in a clinical in both medical and psychiatric patients. J setting? Qual Life Res 2002;11:27–35. Psychosom Res 1994;38:441–450. (41) Velikova G, Wright EP, Smith AB, et al. (30) Aben I, Verhey F, Lousberg R, et al. Automated collection of quality-of-life Validity of the Beck Depression Inventory, data: a comparison of paper and computer Hospital Anxiety and Depression Scale, touch-screen questionnaires. J Clin Oncol SCL-90, and Hamilton Depression Rating 1999;17:998–1007. Scale as screening instruments for (42) Dowell AC, Biran LA. Problems in using depression in stroke patients. the Hospital Anxiety and Depression Scale Psychosomatics 2002;43:386–393. for screening patients in general practice. (31) Bramley PN, Easton AME, Morley S, et al. Br J Gen Pract 1990;40:27–28. The differentiation of anxiety and (43) Snaith RP. The Hospital Anxiety and depression by rating scales. Acta Psychiatr Depression Scale. Br J Gen Pract Scand 1988;77:133–138. 1990;40:305. (32) Flint AJ, Rifat SL. Validation of the (44) Snaith RP. The GHQ and the HAD. Br J Hospital Anxiety and Depression Scale as a Psychiatry 1991;158:433. measure of severity of geriatric depression. Int J Geriatr Psychiatry 1996;11:991–994. (33) Malasi TH, Mirza IA, El-Islam MF. The Self-Rating Anxiety Scale, and Validation of the Hospital Anxiety and The Anxiety Status Inventory Depression Scale in Arab patients. Acta (W.W.K. Zung, 1971) Psychiatr Scand 1991;84:323–326. (34) Snaith RP, Taylor CM. Rating scales for Purpose depression and anxiety: a current Zung’s two anxiety scales were intended as brief perspective. Br J Clin Pharmacol quantitative rating instruments for assessing 1985;19:17S–20S. anxiety as a psychiatric disorder rather than as a (35) Clark LA, Watson D. Tripartite model of trait or feeling state. They “would be inclusive anxiety and depression: psychometric with respect to symptoms of anxiety as a psychi- evidence and taxonomic implications. J atric disorder.” The same items are presented in Abnorm Psychol 1991;100:316–336. two formats so that self- and clinician- (36) Sigurdardóttir V, Brandberg Y, Sullivan M. evaluations may be directly compared (1, p180). Criterion-based validation of the EORTC QLQ-C36 in advanced melanoma: the CIPS questionnaire and proxy raters. Qual Conceptual Basis Life Res 1996;5:375–386. Zung distinguished between several common (37) El-Rufaie OE, Absood GH. Retesting the uses of the term “anxiety” and focused his anxi- validity of the Arabic version of the ety scales on the measurement of generalized Hospital Anxiety and Depression (HAD) anxiety disorders as defined in the DSM-III and scale in primary health care. Soc Psychiatry elsewhere (2, pp5–7). The symptoms included in Psychiatr Epidemiol 1995;30:26–31. the instrument are those most commonly found (38) Leung CM, Wing YK, Kwong PK, et al. in anxiety disorder (3, Table 1). Zung noted that Validation of the Chinese-Cantonese anxiety disorder typically arises as a response to version of the Hospital Anxiety and coping with stress and the signs and symptoms Depression Scale and comparison with the are largely without cultural attributes. Hamilton Rating Scale of Depression. Acta Psychiatr Scand 1999;100:456–461. Description (39) Kugaya A, Akechi T, Okuyama T, et al. Screening for psychological distress in The 20-item clinician-rating Anxiety Status In- Japanese cancer patients. Jpn J Clin Oncol ventory (ASI) and the self-administered Self- 1998;28:333–338. rating Anxiety Scale (SAS) were developed (40) Boyes A, Newell S, Girgis A. Rapid concurrently and have the same coverage. The assessment of psychosocial well-being: are first five items record affective symptoms and the  Anxiety 303 Exhibit 6.5 The Zung Anxiety Status Inventory (ASI) Note: A four-point rating is used for each set of symptoms: 1 = none or insignificant in intensity or duration, and present none or only a little of the time in frequency; 2 = mild in intensity or duration, present some of the time in frequency; 3 = of moderate severity, present a good part of the time in frequency, and 4 = severe in intensity or duration, present most or all of the time in frequency. The raw scores are summed and converted to a 25-100 scale by dividing the total by 0.8. Affective and somatic symptoms of anxiety Interview guide for Anxiety Status Inventory (ASI) 1. Anxiousness Do you ever feel nervous and anxious? 2. Fear Have you ever felt afraid? 3. Panic How easily do you get upset? Ever had panic spells or feel like it? 4. Mental disintegration Do you ever feel like you’re falling apart? 5. Apprehension Have you ever felt uneasy? Or that something terrible was going to happen? 6. Tremors Have you ever had times when you felt yourself trembling? Shaking? 7. Body aches and pains Do you have headaches? Neck or back pains? 8. Easy fatigability, weakness How easily do you get tired? Ever have spells of weakness? 9. Restlessness Do you find yourself restless and can’t sit still? 10. Palpitation Have you ever felt that your heart was running away? 11. Dizziness Do you have dizzy spells? 12. Faintness Do you have fainting spells? Or feel like it? 13. Dyspnea Ever had trouble with your breathing? 14. Paresthesias Ever had feelings of numbness and tingling in your fingertips? Or around your mouth? 15. Nausea and vomiting Do you ever feel sick to your stomach or feel like vomiting? 16. Urinary frequency How often do you need to empty your bladder? 17. Sweating Do you ever get wet, clammy hands? 18. Face flushing Do you ever feel your face getting hot and blushing? 19. Insomnia, init. How have you been sleeping? 20. Nightmares Do you have dreams that scare you? Adapted from Zung WWK. A rating scale for anxiety disorders. Psychosomatics 1971;12:371–379. Reprinted with permission from Psychosomatics, copyright (1971). American Psychiatric Association. remaining 15 comprise physiological symptoms p202). The ratings cover the week prior to the of anxiety. evaluation. Each item is rated for severity, in Ratings in the ASI, based on an interview terms of the intensity, duration and frequency of with the patient, combine clinical observations each symptom. A four-point scale is used: and reports by the patient. The ASI shown in one = none or insignificant in intensity or dura- Exhibit 6.5 forms a guide, but the clinician may tion, and present none or only a little of the ask additional questions or probe for details (4, time in frequency; two = mild in intensity or du- 304 Measuring Health ration, present some of the time in frequency; of 0.56 for the Hospital Anxiety and Depres- three = symptoms of moderate severity, present a sion Scale anxiety score) (7, Table 1). Zung re- good part of the time in frequency, and ported a series of correlations between the SAS four = severe in intensity or duration, present and the Hamilton Anxiety Rating Scale; these most or all of the time in frequency (1, p182; 4, ranged from 0.56 to 0.81, with a median of p203). Additional questions may be required to 0.74 (2, p9). establish the severity of each item, such as Some attention has been paid to the discrimi- “How bad was it?”; “How long did it last?” or nant validity of the SAS. The SAS correlated “How much of the time did you feel that way?” 0.53 with Zung’s Self-rating Depression Scale (3, p373; 4, p203). (SDS) (7, p135), but the anxiety scale did not The Self-rating Scale uses the same 20 diag- discriminate significantly between anxious and nostic criteria. Items are scored as shown in Ex- depressed psychiatric patients, whereas the SDS hibit 6.6; note the reverse scoring of positively did (7, Table 2). In another study, Zung reported worded items. SAS scores of 58.7 for anxious patients and 50.7 For both ASI and SAS, raw scores are added for depressed patients; the corresponding SDS (range, 20–80) and the total is transformed by scores were 52.5 and 65.3, again suggesting that dividing by 80 and multiplying by 100, giving a the SDS discriminates better between the diag- range from 25 (low anxiety) to 100 (1, p182). nostic groups (8, Table IV). Nonetheless, each Zung referred to the resulting scores as a “Z item is capable of distinguishing significantly be- score” for the ASI, and an “Index score” for the tween patients with anxiety and unaffected SAS (3, p376). He proposed a cutting-point of adults (2, Table 3). 44/45 to indicate clinically significant anxiety (2, p18). Scores of 45 to 59 indicate “minimal to Alternative Forms moderate anxiety”; 60 to 74 suggests “marked The National Institutes of Health Early Clinical to severe anxiety” and 75 or higher indicates ex- Drug Evaluation (ECDEU) assessment manual treme anxiety (5, p356). presents a slightly modified version, the main changes being in the tense of the verbs (“ever Reliability had” becomes “ever have” in items 13 and 14) A split-half coefficient of 0.71 was reported, and (4, p200). A modified version has been proposed alpha was 0.85 (2, p10). Alpha was 0.69 for un- for people with mental handicaps that clarifies affected subjects, and 0.81 for psychiatric pa- phrasing of the items and response categories tients in a Nigerian study (6). (9). The MAPI Research Institute clearinghouse Validity offers translations of the SAS into Dutch, Zung studied 225 psychiatric patients and 343 Finnish, French, German, Italian, Japanese, Nor- nonpatients (1). The correlation between the wegian, Russian and Spanish (www.qolid.org/ self- and clinician-administered versions was public/ZungSAS.html). It has also been trans- 0.66 overall, rising to 0.74 for patients with a lated into Yoruba (6), Korean, and Taiwanese diagnosis of anxiety disorder (3, p378). Corre- Chinese (2). lations with the Taylor Manifest Anxiety Scale (TMAS) were 0.30 for the SAS and 0.33 for the Reference Standards ASI (1, p185). Both the ASI and SAS discrimi- ASI reference values for selected diagnostic nated (p < 0.05) between anxiety patients and groups were: patients with anxiety disorders = patients with other psychiatric conditions; the mean 62.0 (standard deviation [SD], 13.8); TMAS did not discriminate significantly (1, schizophrenia, mean = 49.4 (SD, 15.9); depres- p184). In a subsequent study of cardiac pa- sive disorders 49.9 (SD, 12.5); personality disor- tients, the correlation with the TMAS was 0.62 ders 52.6 (SD, 13.6), and transient situational (2, p9). The correlation with the Clinical Anxi- disturbances 42.0 (SD, 8.1) (3, Table VII; 4, ety Scale was 0.52 (comparable with the value p201). The equivalent values for the SAS were:  Anxiety 305 Exhibit 6.6 The Zung Self-Rating Anxiety Scale (SAS) Note: The scores shown here are not included in the version presented to the respondent; they are replaced by boxes in which the respondent places check marks. The raw scores are summed and converted to a 25–100 scale by dividing the total by 0.8. None or a Some of Good part Most or little of the time of the all of the the time time time 1. I feel more nervous and anxious than usual 1 2 3 4 2. I feel afraid for no reason at all 1 2 3 4 3. I get upset easily or feel panicky 1 2 3 4 4. I feel like I’m falling apart and going to 1 2 3 4 pieces 5. I feel that everything is all right and 4 3 2 1 nothing bad will happen 6. My arms and legs shake and tremble 1 2 3 4 7. I am bothered by headaches, neck and back 1 2 3 4 pains 8. I feel weak and get tired easily 1 2 3 4 9. I feel calm and can sit still easily 4 3 2 1 10. I can feel my heart beating fast 1 2 3 4 11. I am bothered by dizzy spells 1 2 3 4 12. I have fainting spells or feel like it 1 2 3 4 13. I can breathe in and out easily 4 3 2 1 14. I get feelings of numbness and tingling in 1 2 3 4 my fingers, toes 15. I am bothered by stomach aches or 1 2 3 4 indigestion 16. I have to empty my bladder often 1 2 3 4 17. My hands are usually dry and warm 4 3 2 1 18. My face gets hot and blushes 1 2 3 4 19. I fall asleep easily and get a good night’s 4 3 2 1 rest 20. I have nightmares 1 2 3 4 Adapted from Zung WWK. How normal is anxiety? Kalamazoo, MI: The Upjohn Company, 1980. Used with permission from Pfizer, Inc. All rights reserved. anxiety disorders 58.7 (SD, 13.5); schizophrenia of patients: 58.8 for patients with anxiety disor- 46.4 (SD, 12.9); depressive disorders 50.7 (SD, ders, 48.8 for other psychiatric patients, and 13.4); personality disorders 51.2 (SD, 13.2), and 41.8 for general practice patients (2, Tables 7 transient situational disturbances 45.8 (SD, and 8). 11.9). Zung reported mean scale and item scores Zung summarized mean scores from various for healthy adolescents (aged 14–19 years) and countries, showing broadly comparable results, adults (2, Tables 4 to 6). The mean for adoles- with the exception of a Korean study of men- cents was 45.1 (SD, 8.9); that for adults aged 20 tally healthy people in which several item scores to 64 was 34.4 (SD, 6.9) and for adults 65 and fell outside a range of one SD of the U.S. mean over the mean was 36.9 (SD, 8.8). Zung also scores (2, pp16–17). Agreement for patient presented mean scale scores for various groups groups, however, was very close. 306 Measuring Health Commentary (7) Bramley PN, Easton AME, Morley S, et al. The Zung anxiety scales offer a potentially use- The differentiation of anxiety and ful resource where it is necessary to assess anxi- depression by rating scales. Acta Psychiatr ety by either self-report or clinical rating. The Scand 1988;77:133–138. self-rating scale, in particular, is used quite com- (8) Zung WWK. The differentiation of anxiety and depressive disorders: a biometric monly in studies of anxiety. The scales were in- approach. Psychosomatics corporated into the ECDEU protocol and appear 1971;12:380–384. useful as outcome measures (4). Their clinical (9) Lindsay WR, Michie AM. Adaptation of orientation distinguishes the Zung scales from the Zung self-rating anxiety scale for more general measures such as Taylor’s Manifest people with a mental handicap. J Ment Anxiety Scale. The Zung scale has also achieved Defic Res 1988;32:485–490. popular attention with a Web site that allows (10) Beck AT, Epstein N, Brown G, et al. An users to score their own anxiety on the SAS: inventory for measuring clinical anxiety: www.anxietyhelp.org/information/sas_zung.html psychometric properties. J Consult Clin #noteszung_sas. Psychol 1988;56:893–897. The phrasing of some of the items appears awkward (“I feel more nervous and anxious than usual . . . None or a little of the time”). Several The Beck Anxiety Inventory SAS items had higher correlations with the BDI (A. Beck, 1988) than with the total SAS score, suggesting that Purpose these items may not be specific to anxiety states Beck’s self-report Anxiety Inventory (BAI) mea- (10, p897). sures the severity of self-reported anxiety in adults and adolescents; it was especially designed to minimize confounding with symptoms of de- pression. References (1) Zung WWK. The measurement of affects: Conceptual Basis depression and anxiety. Mod Probl A central stimulus for developing the BAI arose Pharmacopsychiatry 1974;7:170–188. during studies of the link between anxiety and (2) Zung WWK. How normal is anxiety? depression. During the 1980s, concern arose Kalamazoo, MI: The Upjohn Company, over the high correlations between anxiety and 1980. depression scales, typically between 0.6 and 0.7, (3) Zung WWK. A rating instrument for anxiety disorders. Psychosomatics which are almost as high as correlations among 1971;12:371–379. anxiety scales (1). It was not clear whether this (4) Guy W. Early clinical drug evaluation association between anxiety and depression re- (ECDEU) assessment manual for flected a biological link between them, or shared psychopharmacology. DHEW Pub No. 76- etiologies, or was merely an artefact due to non- 338 ed. Rockville, Maryland: US specificity of the measures (2, p196). As with Department of Health, Education, and measures of psychological well-being discussed Welfare National Institute of Mental in Chapter 5, it was also possible that anxiety Health, 1976. and depression scales tap into a nonspecific di- (5) Zung WWK, Cavenar JO, Jr. Assessment mension of general psychological distress or neg- scales and techniques. In: Kutash IL, ative affectivity (1). This involves a personal Schlesinger LB, eds. Handbook on stress and anxiety. San Francisco: Jossey-Bass, tendency to be worried, anxious, and self-critical 1980:348–363. that arises in both anxiety and depression. In (6) Jegede RO. Psychometric attributes of the contrast with this, Beck et al. proposed a model self-rating anxiety scale. Psychol Rep of psychopathology that suggested that anxiety 1977;40:303–306. and depression could, indeed, be distinguished  Anxiety 307 by the form and content of dysfunctional cogni- 63 indicate severe anxiety (6, p5). Slightly differ- tions, with anxiety referring to negative feelings ent cut-points were given by Steer and Beck who that are specific to certain situations, whereas cited 0–7 as indicating minimal anxiety; eight to depression involves more absolute and pervasive 15 as mild; 16 to 25 as moderate anxiety, and 26 negative feelings (3, p646). To analyze these al- to 63 as severe anxiety (7, p36). ternatives, Beck et al. conceived the BAI so as to Recent changes to copyright have prevented be as uncontaminated with items reflecting de- reproduction of the scale here, but the items are pression as possible; this would permit distinc- shown in Beck’s article in the Journal of Con- tion between the common disorder and the sulting and Clinical Psychology 1988, vol. 56, measurement artefact hypotheses (4). page 895 (Table 1), as well as in references (9–13) listed below. Copies of the scale are avail- Description able through Harcourt (http://harcourtassess The BAI is a patient-completed 21-item measure ment.com). in which 14 items cover somatic symptoms and seven reflect subjective aspects of anxiety or Reliability panic. The items were drawn from three of Numerous studies have reported alpha internal Beck’s earlier instruments, the Anxiety Check- consistency and the results are consistent across list, the Physician’s Desk Reference checklist, a wide range of respondents (see Table 6.3). Al- and the Situational Anxiety Checklist (4, p894). pha values for sub scales identified through fac- The items were chosen to measure symptoms tor analysis range from 0.7 to 0.92 (19, Table specific to anxiety and unrelated to depression; 2). they were also intended to measure general Item-total correlations range from 0.41 (for symptoms, rather than anxiety due to specific the item on numbness) to 0.68 (terrified) (2, disorders such as phobias or panic disorders (5). p198). A median value was 0.56 (11, p347). The After initial item analyses, the 21-item version ranges in other studies include 0.30 (numbness) was established (4). to 0.71 (unsteady), with a median of 0.60 (4, The BAI may be self-administered or may be Table 1), or 0.37 to 0.69 (10, p37). administered and scored by an interviewer with One-week test-retest reliability results in- appropriate training, but Beck et al. stressed that clude 0.75 (4, p895) and 0.73 (18, p59). An 11- the interpretation of scores should be under- day retest figure of 0.67 was reported (18, p59); taken by a clinician able to bear in mind the im- a three-week retest coefficient was 0.83 (12, plications of comorbid conditions or suicidal p143); a seven-week retest correlation was 0.62 ideation (6). Administration instructions are (12, p480). given in the manual (6). The BAI takes five to ten minutes to complete (6; 7). The respondent rates Validity how much he or she has been bothered by each content validity. Steer and Beck argued that symptom of anxiety over the past week (includ- the BAI covers most DSM-III-R symptoms of ing the day of testing) on a four-point intensity generalized anxiety; it also covers 11 of 13 DSM scale: 0 (Not at all), 1 (Mildly—it did not bother symptoms of panic disorders. However, the BAI me much), 2 (Moderately—it was very unpleas- does not cover the avoidant behavior that under- ant but I could stand it) to 3 (Severely—I could lies DSM criteria for conditions such as social barely stand it). anxiety and phobias (7, p27). A total score is the sum of the item scores; so- matic and subjective subscores are sometimes factorial validity. The factor structure of the used (8). Overall scores of 0 and nine points are BAI has been examined in various samples. Most interpreted as normal anxiety; ten to 18 repre- analyses have identified two factors: cognitive sent mild-to-moderate anxiety; 19 to 29 indicate (or subjective) complaints and somatic symp- moderate-to-severe anxiety, and scores of 30 to toms (4, p895; 8, Table 1; 10, p37; 12, Table 3; 308 Measuring Health Table 6.3 Alpha Internal Consistency Coefficients for the Beck Anxiety Inventory Type of Study Subjects Alpha Coefficient Reference Elderly medical outpatients 0.86 (5, p348) Representative sample of elderly people 0.89 (9, p23) Elderly psychiatric outpatients 0.90 (5, p348) Adult psychiatric outpatients 0.90 (10, p37) Young psychiatric outpatients 0.90 (6) Undergraduates 0.90 (14, p9) Panic disorder patients 0.90 (15, p143) Adolescent inpatients 0.91 (16, p129) Undergraduate students 0.91 (11, p347) Undergraduate students 0.91 (12, p480) Adult psychiatric outpatients 0.92 (4, p895) Adult psychiatric outpatients 0.92 (2, p198) Adult psychiatric outpatients 0.92 (8, p409) Elderly medical patients 0.92 (16, p138) Outpatients with panic disorder 0.93 (13, p951) Outpatients with anxiety disorder 0.94 (18, p59) 16, Table 1; 20, Table 1; 21). The two factors Several analyses suggest that the BAI can also be tend to be correlated, coefficients typically viewed as four factors subsumed under a single falling between 0.56 and 0.63 (10, Table 3; 20, second-order factor that describes severity of gen- p538). Although the two-factor solution appears eral anxiety (14; 17). Variants include a five- consistent, there are some variations in item factor solution, labeled Subjective Fear, Somatic placement. The item “Scared,” for example, nervousness, Neurophysiological, Muscular/Mo- loaded highly on the somatic factor in the Beck toric, and Respiration (11, Table 2), and a six- et al. original analysis (4, Table 1), whereas sub- factor solution (9, Table 3). sequent studies have found it to fit on the subjec- tive factor (12). convergent validity. Many studies have com- Other studies have split the somatic items pared the BAI with other scales; a selection of re- onto separate factors, generally resulting in four- sults is included in Table 6.4. A factor analysis factor solutions. These include subjective com- that included both BAI and State-Trait Anxiety plaints (e.g., unable to relax, nervous, fear of Inventory (STAI) items showed the two to load losing control); neurophysiological or neuromo- on separate factors, suggesting that they may tor aspects (e.g., numbness, unsteady, hands represent separate concepts (12, Table 4). trembling), autonomic aspects (e.g., feeling hot, Because of the physiological symptoms it in- indigestion, sweating), and a panic component cludes, responses to the BAI have been compared (e.g., heart pounding, choking, difficulty breath- with physical measures (e.g., electromyographi- ing) (2, p199; 19, Table 1; 22, Table 1). Subse- cally evident muscle activity, skin temperature, quent confirmatory factor analyses have and heart rate). The results suggested a modest supported this general conclusion with only mi- general correspondence, but no clear pattern of nor variation in the placement of a few items agreement has been determined between the BAI (11, Figure 1; 17, Figure 1; 23, Figure 1). Beck somatic factor scores and the corresponding has suggested that the more specific, four-factor physical measures (11, p352). In fact, a self- solution may obtain with anxiety patients, report measure of overall subjective distress cor- whereas fewer factors may appear with psychi- related more strongly with the BAI somatic atric patients with mixed diagnoses (22, p221). factors than physical measures did.  Anxiety 309 Table 6.4 Convergent Validity Correlations for the Beck Anxiety Index Comparison scale: Coefficient Reference State-Trait Anxiety Inventory (State scale) 0.52 (10, Table 3) 0.47 (18, Table 1) 0.56 (12, p482) 0.64 (12, p482) State-Trait Anxiety Inventory (Trait scale) 0.44 (10, Table 3) 0.58 (18, Table 1) 0.57 (12, p482) 0.68 (12, p482) Hamilton Rating Scale for Anxiety 0.47 (9, Table 2) 0.51 (6, p13) 0.56 (22, p216) 0.67 (3, Table 6) Depression Anxiety Stress Scales (Anxiety scale) 0.81 (24, Table 2) 0.84 (25, Table 3) (Anxiety scale abbreviated version) 0.85 (25, Table 3) Brief Symptom Inventory (Anxiety scale) 0.69 (14, Table 3) 0.73 (9, Table 2) 0.78 (15, Table 1) Cognition Checklist (Anxiety scale) 0.51 (14, Table 3) Symptom Check-List-90 (Anxiety scale) 0.81 (2, Table 2) (Phobic anxiety scale) 0.57 (2, Table 2) (Somatization) 0.75 (2, Table 2) (Symptom Check-List 0.75 (2, Table 2) global severity) discriminant validity. Several studies have (BDI); the BAI also performed somewhat less ad- demonstrated that the BAI can discriminate sig- equately than the Hamilton Anxiety Rating nificantly between outpatients diagnosed with Scale (26, Table 2). In a study of mixed diagno- mood, anxiety, and other types of psychiatric sis psychiatric outpatients, the BAI distinguished disorder (2). Because sample sizes are often those with an anxiety disorder from those with- large, mere statistical significance provides little out better than the STAI Trait score did, whereas information. In a study of psychiatric outpa- the STAI State score failed to discriminate at all tients, sensitivity for anxiety (contrasted with (10, p38). Sensitivity to change after treatment other psychiatric conditions) was 94%, and appears good: a Cohen’s d statistic of 1.1 was re- specificity was 45% at a cutting-point of 9/10 ported; this was superior to the equivalent result (10, p41). Raising the cutting-point to 18/19 of 0.95 for the Brief Symptom Inventory Anxi- gave a sensitivity of 56% and a specificity of ety scale (15, Table 4). 69%. The same study found that including only The validity of the BAI as a screening instru- the subjective anxiety items gave a sensitivity ment in unselected populations has received less that was actually superior to that of the complete attention, but it has been found to perform rela- scale (10, p41 and Table 4). tively poorly in two studies of primary care pa- In a detailed study of the BAI’s ability to dis- tients. In one study, a cutting-point of ten gave a criminate among psychiatric diagnoses, Somoza sensitivity of 40% at a specificity of 89% (27, et al. found the BAI to be significantly less able Table 3); the authors concluded that the Symp- to discriminate between depressive and panic tom Check List Anxiety subscale (which is disorders than either the Hamilton Rating Scale briefer than the BAI) performed equally well. In for Depression or the Beck Depression Inventory a second study, the BAI was not independently 310 Measuring Health predictive of any disorder after controlling for medical patients; the items “Fear of the worst its correlations with depression screening scales happening” and “Unsteady” were the most dis- (28, p26). criminating (5, p348). The authors did, however, Many studies have addressed how well the note that six of the somatic items did not distin- BAI distinguishes anxiety from depression. Beck guish between the two groups. et al. compared psychiatric outpatients with anx- Cox et al. have argued that many of these so- iety disorders and others with depression, show- matic items correspond to symptoms of panic ing that the mean BAI scores were 6.5 to 12 rather than anxiety. They included the BAI items points higher for the anxious than the depressed with questions from a panic scale in a factor patients (4, Table 2). (In the same study the BDI analysis, arguing that if the analysis separated showed differences in mean scores in the reverse items from the two scales, this would reassure direction of only 2 to 6 points). If it does dis- that the BAI covers anxiety in general, rather criminate, the correlation between the BAI and then panic. The results did not identify a sepa- depression measures should be low to moderate; rate BAI factor (13, Table 1). However, further in the early Beck et al. study of adult outpatients, investigation has not fully replicated this finding, the correlation between the BAI and the BDI was showing that although the BAI does correlate 0.48 (4, p895). Likewise, a small study of pa- with measures of panic, it does not correlate tients with panic disorders gave a correlation of with measures of fear of fear, which is a hall- 0.45 (15, Table 1), comparable with a value of mark of panic disorder (15, Table 2). 0.44 in a general sample of elderly people (9, The associations with sex, age, and ethnicity Table 2) and 0.50 for patients with anxiety dis- were found to be nonsignificant in a study of orders (18, Table 1). More typical results, how- adolescents (16, p127). However, several other ever, are higher: several studies reported a studies have found that scores for women were correlation of 0.62 between the BAI and the BDI significantly higher than those for men, by 3.2 (2, p201; 16, p127; 23, Table 1), or 0.63 (8, points (9, p24), 3.8 points (19, Table 2); 4.2 p409; 12, p482). Intermediate figures include points (5, p348); 4.8 points (2, p198) and 5.5 0.58 (3, p648), or 0.56 (14, Table 3; 16, p140). points (8, p409). In a general population sample, However, it appears that other anxiety measures age correlated −0.20 with BAI total scores (19, (e.g., the STAI) show higher correlations with Table 2), but this seems to be an exception and the BDI (12). A correlation of 0.65 was obtained in most other studies, there was little correlation between the BAI and the Geriatric Depression with age. Scale (17, p140). Studies that have factor- analyzed items from the BAI and the BDI to- Alternative Forms gether consistently show them as falling on A Spanish version has been tested for reliability separate factors (3; 4; 8; 17), although it has also and validity (29). A French version showed an been shown that a second-order general distress internal consistency of 0.85 and four-week retest or negative affect factor explains twice as much stability of 0.63 (30). variance as the first-order factors (3, p649). Steer A computer-administered version has been and Beck concluded that about 20% of variance used with psychiatric patients (20). Mean scores in the BAI reflects unique aspects of anxiety, from the computer administered version were whereas the remainder reflects general negative 16.3, compared to 17.0 for a paper and pencil affect, neuroticism or depression (7, p31). version (r = 0.93) (20, p536). The question of whether the somatic symp- Beck et al. proposed an abbreviated version toms in the BAI may led to false-positive re- of the BAI to identify anxiety in primary care sponses from people with physical conditions settings. The BAI-PC includes seven items: un- has been addressed. Steer et al. showed that the able to relax, fear of the worst happening, terri- BAI could distinguish between elderly medical fied, nervous, fear of losing control, fear of and psychiatric outpatients; the mean score for dying, and scared. Questions refer to symptoms the psychiatric patients was 2.7 times that of the in “the past two weeks, including today.” Each  Anxiety 311 item is rated on a 0 to 3 scale, and scores be subsumed under a general distress factor. summed (31, p213). Internal consistency alpha Variations in the results of factor analyses ap- was 0.90, with item-total correlations running pear to reflect differences in sample composi- from 0.59 to 0.81 (31, p214). The area under a tion; some studies have used homogeneous ROC curve for detecting generalized anxiety or samples, whereas others used a wide range of pa- panic was 0.87 (31, p216). tient types. Creamer et al. offered an interesting insight from their study of undergraduates who Reference Standards were assessed first under low-stress conditions The BAI manual quotes mean scores for several and then again at examination time. A different subcategories of anxiety patients, based on small pattern of responses emerged at the two times, samples ranging in size from 26 to 95 (6, Tables rather than merely the endorsement of more 2 and 4). Beck and Steer reported a mean score items. They suggested that the distinction be- of 22.4 for 160 adult outpatients diagnosed with tween physiological and cognitive symptoms of mixed psychiatric disorders (4). A mean score of anxiety becomes more apparent under stressful 15.9 was reported for 108 adolescent inpatients conditions (12, p484). Analyses of the factor (16, p127). scores show that these measure a different aspect Women commonly report higher scores than of anxiety than subjective responses do. This has men; the BAI manual suggests that a difference led to a debate over the breadth of the BAI: does of four points is typical and should be taken into it work best for disorders that exhibit physiolog- consideration when interpreting scores (6, p5). ical symptoms (e.g., panic disorder), and perhaps less well in anxiety-related conditions with a Commentary large cognitive or social component (e.g., social In the relatively brief period since its introduc- phobias, obsessive-compulsive disorders)? (12, tion, the BAI has become well established (32). p484). Cox et al. extended this argument to con- The goal of developing a scale that distinguished clude that “the BAI is likely confounded with, or anxiety from depression seems to have been in large part measures, panic symptoms.” (13, largely achieved, although a common underlying p953). In response, Steer and Beck argued that negative affect factor underlies both conditions, these symptoms are relevant both to panic at- so it will never be possible to separate them com- tacks and to anxiety in general, and that most pletely. The correlations between the BAI and panic patients also exhibit general anxiety. Fea- the BDI do not prove that the scale is flawed, for tures such as the pattern of onset distinguish be- many of the study samples included patients tween the conditions; this may be therefore with general mood disorders who exhibited better assessed using a clinical interview than a symptoms of both anxiety and depression. In- self-report instrument (33). In a counter- deed, the correlations appear lower when only argument, Cox et al. acknowledged that the BAI patients with anxiety are included (2, p203). was developed under the third edition of the Di- Symptoms of hyperarousal (e.g., inability to re- agnostic and Statistical Manual, whereas the lax, heart pounding, hands trembling) appear fourth edition (DSM-IV) has deleted several of the most specific to anxiety, although these items the symptoms from the definition of generalized also loaded on the common negative affect fac- anxiety but retained them under the criteria for tor (3, p652). Clark and Watson, in particular, diagnosing a panic attack (34, p960). Hence, in praised the specificity of the BAI, noting that terms of the newer DSM-IV criteria, the BAI Beck’s anxiety and depression scales achieve a may, indeed, blend anxiety and panic attack clearer discrimination between the two condi- symptoms. tions than the Hamilton scales or the STAI; it Although Beck does not refer to state and appears to achieve this by virtue of its inclusion trait anxiety, the one-week time-frame used in of physiological symptoms (1, p327). answering the BAI would seem to correspond to The general results of factor analyses appear a state measure. It is somewhat surprising, there- to suggest two main factors, both of which can fore, that correlations with the trait scale of the 312 Measuring Health STAI are somewhat higher than those with the psychiatric outpatients with the Beck state scale (see Table 6.4, above). Anxiety Inventory. Assessment Overall, the BAI appears to be a good quality 1994;1:345–351. measure; it seems to achieve its goal of discrimi- (6) Beck AT, Steer RA. Beck Anxiety Inventory nating anxiety from depression better than rival manual. 1st ed. San Antonio, TX: Psychological Corporation, 1990. scales. Being structured for this purpose, conver- (7) Steer RA, Beck AT. Beck Anxiety gent correlations with other measures may be Inventory. In: Zalaquett CP, Wood RJ, eds. somewhat reduced, and it may perform best Evaluating stress: a book of resources. with anxiety disorders that include a physiologi- Lanham, MD: Scarecrow Press, 1997: cal component. Consequently it may underrate 23–40. anxiety that presents with low levels of psy- (8) Hewitt PL, Norton GR. The Beck Anxiety chophysiological symptoms but higher levels of Inventory: a psychometric analysis. Psychol cognitive or behavioral symptoms, and it does Assess 1993;5:408–412. not cover avoidant behaviors, specific phobias, (9) Morin CM, Landreville P, Colecchi C, et or feelings about reexperiencing trauma (7). The al. The Beck Anxiety Inventory: psychometric properties of the BAI are well es- psychometric properties with older adults. tablished; it will be valuable now to see more ex- J Clin Geropsychology 1999;5:19–29. (10) Kabacoff RI, Segal DL, Hersen M, et al. amination of its accuracy as a screening or Psychometric properties and diagnostic general survey tool in unselected populations. utility of the Beck Anxiety Inventory and the State-Trait Anxiety Inventory with Address older adult psychiatric outpatients. J Anx Copies of the Beck inventory are available from Disord 1997;11:33–47. Harcourt Assessment http://harcourtassessment (11) Borden JW, Peterson DR, Jackson EA. The .com Beck Anxiety Inventory in nonclinical Information on the Beck Institute is available samples: initial psychometric properties. J from http://www.beckinstitute.org/, while Dr. Psychopathol Behav Assess 1991;13:345–356. Beck’s home page is at http:// mail.med.upenn (12) Creamer M, Foran J, Bell R. The Beck .edu/~abeck/index .html Anxiety Inventory in a non-clinical sample. Behav Res Ther 1995;33:477–485. (13) Cox BJ, Cohen E, Direnfeld DM, et al. References Does the Beck Anxiety Inventory measure (1) Clark LA, Watson D. Tripartite model of anything beyond panic attack symptoms? anxiety and depression: psychometric Behav Res Ther 1996;34:949–954. evidence and taxonomic implications. J (14) Osman A, Kopper BA, Barrios FX, et al. Abnorm Psychol 1991;100:316–336. The Beck Anxiety Inventory: reexamination (2) Steer RA, Ranieri WF, Beck AT, et al. of factor structure and psychometric Further evidence for the validity of the properties. J Clin Psychol 1997;3:7–14. Beck Anxiety Inventory with psychiatric (15) de Beurs E, Wilson KA, Chambless DL, et outpatients. J Anx Disord 1993;7:195–205. al. Convergent and divergent validity of the (3) Clark DA, Steer RA, Beck AT. Common Beck Anxiety Inventory for patients with and specific dimensions of self-reported panic disorder and agoraphobia. Dep Anx anxiety and depression: implications for 1997;6:140–146. the cognitive and tripartite models. J (16) Kumar G, Steer RA, Beck AT. Factor Abnorm Psychol 1994;103:645–654. structure of the Beck Anxiety Inventory. (4) Beck AT, Epstein N, Brown G, et al. An Anxiety, Stress, and Coping inventory for measuring clinical anxiety: 1993;6:125–131. psychometric properties. J Consult Clin (17) Wetherell JL, Areán PA. Psychometric Psychol 1988;56:893–897. evaluation of the Beck Anxiety Inventory (5) Steer RA, Willman M, Kay PAJ, et al. with older medical patients. Psychol Assess Differentiating elderly medical and 1997;9:136–144.  Anxiety 313 (18) Fydrich T, Dowdall D, Chambless DL. Psychometric comparability of English- and Reliability and validity of the Beck Anxiety Spanish-language measures of anxiety and Inventory. J Anx Disord 1992;6:55–61. related affective symptoms. Psychol Assess (19) Osman A, Barrios FX, Aukes D, et al. The 2001;13:347–355. Beck Anxiety Inventory: psychometric (30) Freeston MH, Ladouceur R, Thibodeau N, properties in a community population. J et al. [The Beck Anxiety Inventory. Psychopathol Behav Assess Psychometric properties of a French 1993;15:287–297. translation]. Encephale 1994;20:47–55. (20) Steer RA, Rissmiller DJ, Ranieri WF, et al. (31) Beck AT, Steer RA, Ball R, et al. Use of the Structure of the computer-assisted Beck Beck Anxiety and Depression Inventories Anxiety Inventory with psychiatric for primary care with medical outpatients. inpatients. J Pers Assess 1993;60:532–542. Assessment 1997;4:211–219. (21) Steer RA, Kumar G, Ranieri WF, et al. Use (32) Piotrowski C. The status of the Beck of the Beck Anxiety Inventory with Anxiety Inventory in contemporary adolescent psychiatric outpatients. Psychol research. Psychol Rep 1999;85:261–262. Rep 1995;76:459–465. (33) Steer RA, Beck AT. Generalized anxiety (22) Beck AT, Steer RA. Relationship between and panic disorders: response to Cox, the Beck Anxiety Inventory and the Cohen, Direnfeld, and Swinson (1996). Hamilton Anxiety Rating Scale with Behav Res Ther 1996;34:955–957. anxious outpatients. J Anx Disord (34) Cox BJ, Cohen E, Direnfeld DM, et al. 1991;5:213–223. Reply to Steer and Beck: panic disorder, (23) Enns MW, Cox BJ, Parker JDA, et al. generalized anxiety disorder, and Confirmatory factor analysis of the Beck quantitative versus qualitative differences in Anxiety and Depression Inventories in anxiety assessment. Behav Res Ther patients with major depression. J Affect 1996;34:959–961. Disord 1998;47:195–200. (24) Lovibond PF, Lovibond SH. The structure of negative emotional states: comparison of The Depression Anxiety Stress Scales the Depression Anxiety Stress Scales (P.F. Lovibond, 1993) (DASS) with the Beck Depression and Purpose Anxiety Inventories. Behav Res Ther The Depression Anxiety Stress Scales (DASS) are 1995;33:335–343. designed to assess the severity of core symptoms (25) Antony MM, Bieling PJ, Cox BJ, et al. of depression, anxiety, and tension (or stress) Psychometric properties of the 42-item and 21-item versions of the Depression Anxiety over the previous week. Together, the scales pro- Stress scales in clinical groups and a vide a broad-spectrum measure of psychological community sample. Psychol Assess distress, indicating the severity and frequency of 1998;10:176–181. symptoms. Originally developed for use with (26) Somoza E, Steer RA, Beck AT, et al. population samples, they can also be used in Differentiating major depression and panic clinical research. Their main clinical application disorders by self-report and clinical rating is to identify the locus of an emotional distur- scales: ROC analysis and information bance as part of a broader clinical assessment. theory. Behav Res Ther 1994;32:771–782. They are also suitable for tracking change in (27) Hoyer J, Becker ES, Neumer S, et al. severity over time, for example before and after Screening for anxiety in an epidemiological therapeutic intervention. sample: predictive accuracy of questionnaires. J Anx Disord 2002;16:113–134. Conceptual Basis (28) McQuaid JR, Stein MB, McCahill M, et al. The DASS has similarities to, and differences Use of brief psychiatric screening measures from, the tripartite conceptual model proposed in a primary care sample. Dep Anx by Clark and Watson (1). Lovibond and Lovi- 2000;12:21–29. bond used confirmatory factor analysis to iden- (29) Novy DM, Stanley MA, Averill P, et al. tify specific symptoms of anxiety and depression 314 Measuring Health (2). The aim was to maximize discrimination be- guilt, appetite change) that were rejected during tween the depression and anxiety scales while re- developmental work as not being specific to de- taining the breadth of these constructs. pression. Finally, the DASS Stress scale corre- Unexpectedly, a third factor was identified dur- sponds quite closely to the DSM-IV symptom ing the analysis, and it was labelled ‘stress’ be- criteria for Generalized Anxiety Disorder (P. cause of its focus on symptoms of tension, Lovibond, personal communication, 2005). irritability, and difficulty concentrating. Nonspe- cific items that loaded on all three factors (e.g., sleep problems, appetite change) were discarded. Description Lovibond and Lovibond used empirical analyses The DASS is a revised version of a scale origi- of the responses of nonclinical samples to refine nally described in 1983. The revised scale in- the DASS before testing it on clinical samples. cludes 42 negative symptoms; 14 each cover This reflected their assumption that clinical and depression (DASS-D), anxiety (DASS-A), and nonclinical samples experience the same under- stress (DASS-S). These main themes are divided lying syndromes, differing only in severity. into a total of 16 subthemes (see Exhibit 6.7). Hence, “The DASS is based on a dimensional The depression scale focuses mainly on symp- rather than a categorical conception of psycho- toms of dysphoric mood such as sadness, feel- logical disorder” (DASS Web site www.psy.unsw. ings of worthlessness, and inertia (2, p342). The edu.au/dass/). anxiety scale mainly includes symptoms of phys- As with the Clark and Watson model, the ical arousal, panic attacks, and fear, whereas the DASS emphasizes symptoms of autonomic stress scale covers symptoms of nonspecific arousal in defining anxiety and differentiating it arousal: tension, irritability, and overreaction. from depression. Both models emphasize anhe- Each item uses a four-point response scale and donia in defining depression, although the DASS separate depression, anxiety, and stress scores gives equal weight to other features such as are formed by summing item scores. The 16 sub- hopelessness and devaluation of life. However, scales identified in the Exhibit are not normally the models differ in terms of symptoms of ten- scored separately but may have clinical utility if sion and irritability. The DASS assumes that used in that manner. these symptoms reflect a third coherent syn- The DASS was not intended to allocate pa- drome that is distinct from both anxiety and de- tients to discrete DSM or ICD diagnostic cate- pression. The tripartite model, by contrast, gories, although the manual does recommend groups these symptoms together with other cut-offs for conventional severity labels (e.g., symptoms that both models agree are nonspe- normal, moderate, severe) (3; see also the DASS cific into a “General Distress” or “Negative Af- Web site). Based on percentiles derived from a fect” category. Finally, the DASS assumes that normative sample of students, the manual sug- the syndromes it attempts to measure are inter- gests that scores of 0 to 77 represent normal correlated not because they share common states, 78 to 86 represent mild disorder, 87 to 94 symptoms (these were excluded), but because represent moderate disorder, 95 to 97 ‘severe,’ they share common causes (e.g., genetic and en- and 98 to 100 represent extremely severe dis- vironmental vulnerabilities, environmental trig- tress (3). gers). The scales were developed for people aged 17 In terms of traditional diagnostic classifica- or older but may be suitable for younger adoles- tions such as the DSM-IV, the DASS Anxiety cents. The DASS can be administered and scored scale corresponds most closely to the criteria for by non psychologists, but inferences from partic- diagnoses of the various Anxiety Disorders, with ular score profiles should be made by experi- the exception of Generalized Anxiety Disorder. enced clinicians who have carried out an The DASS Depression scale corresponds fairly appropriate clinical examination. closely to the Mood Disorders, although diag- A 21-item abbreviation (DASS-21) requires nostic criteria for these include symptoms (e.g., less time to administer; the items are indicated in Exhibit 6.7 The Depression Anxiety Stress Scales Note: Asterisks identify the 21 items included in the shortened version. In practice, the items are presented in a random order, but are printed here under their respective scales and subscales For each of the statements below, please circle the number which best indicates how much the statement applied to you OVER THE PAST WEEK. There are no right or wrong answers. Do not spend too much time on any one statement. Response scale (applies to all items): (0) Did not apply to me at all (1) Applied to me to some degree, or some of the time (2) Applied to me to a considerable degree, or a good part of the time (3) Applied to me very much, or most of the time DEPRESSION Dysphoria I felt downhearted and blue* I felt sad and depressed Hopelessness I could see nothing in the future to be hopeful about I felt that I had nothing to look forward to* Devaluation of life I felt that life was meaningless* I felt that life wasn’t worthwhile Self-deprecation I felt I was pretty worthless I felt I wasn’t worth much as a person* Lack of interest / involvement I felt that I had lost interest in just about everything I was unable to become enthusiastic about anything* Anhedonia I couldn’t seem to experience any positive feeling at all* I couldn’t seem to get any enjoyment out of the things I did Inertia I just couldn’t seem to get going I found it extremely difficult to work up the initiative to do things* ANXIETY Autonomic arousal I was aware of the action of my heart in the absence of physical exertion (e.g., sense of heart rate increase, heart missing a beat)* I perspired noticeably (e.g., hands sweaty) in the absence of high temperatures or physical exertion I was aware of dryness of my mouth* I experienced breathing difficulty (e.g., excessively rapid breathing, breathlessness in the absence of physical exertion)* I had difficulty in swallowing Skeletal musculature effects I had a feeling of shakiness (e.g., legs going to give way) I experienced trembling (e.g., in the hands)* Situational anxiety I was worried about situations in which I might panic and make a fool of myself* I found myself in situations which made me so anxious I was most relieved when they ended I feared that I would be “thrown” by some trivial but unfamiliar task (continued) 315 316 Measuring Health Exhibit 6.7 (continued) Subjective experience of anxious affect I felt I was close to panic* I felt terrified I felt scared without any good reason* I had a feeling of faintness STRESS Difficulty relaxing I found it hard to wind down* I found it hard to calm down after something upset me I found it difficult to relax* Nervous arousal I felt that I was using a lot of nervous energy* I was in a state of nervous tension Easily upset / agitated I found myself getting upset rather easily I found myself getting upset by quite trivial things I found myself getting agitated* Irritable / over-reactive I tended to over-react to situations* I found that I was very irritable I felt that I was rather touchy* Impatient I was intolerant of anything that kept me from getting on with what I was doing* I found myself getting impatient when I was delayed in any way (e.g., lifts†, traffic lights, being kept waiting) I found it difficult to tolerate interruptions to what I was doing Reproduced from Lovibond PF and Lovibond SH. Behavior Research and Therapy 1995;33:339. With permission. † Elevators the Exhibit. Scores from this version can be mul- depression, 0.92 for anxiety, and 0.95 for stress. tiplied by two to simulate the full-scale scores. Equivalent values for the DASS-21 scales were The DASS is in the public domain and can be 0.94, 0.87, and 0.91 (5, p179). Brown et al. downloaded (together with a scoring template) studied a mixed sample of outpatients and from www.psy.unsw.edu.au/dass/. The manual recorded an alpha of 0.96 for depression, 0.89 can also be ordered through the site. for anxiety, and 0.93 for stress (6, Table 1). A Dutch study obtained alpha values of 0.94 for Reliability depression, 0.88 for anxiety, and 0.93 for stress In a sample of healthy volunteers, alpha internal (7, pi79). A British study reported alphas of consistency coefficients were 0.91 for the de- 0.90 for anxiety, 0.95 for depression, and 0.93 pression score, 0.84 for the anxiety score, and for stress (0.97 overall) (8, p122). 0.90 for the stress score (2, p337). For a sample Two-week retest correlations were 0.71 for of psychiatric outpatients, the alpha values were depression, 0.79 for anxiety, and 0.81 for stress 0.89 for anxiety, 0.96 for depression, and 0.94 (6, Table 2). Lovibond reported retest correla- for stress; equivalent values for the 21-item ver- tions at intervals ranging from three to eight sion of the DASS were 0.81, 0.92, and 0.88, years using the trait wording rather than the respectively (4, p62). A second study of psychi- usual past week (state) version. At three years, atric outpatients gave alpha values of 0.97 for the correlations were 0.47 for depression, 0.46  Anxiety 317 for anxiety, and 0.34 for stress (9, Table 2). A studies have reported correlations between the structural equation modeling analysis suggested two scores. The results include 0.54 (2, Table 2), that depression was the least stable scale and that 0.49 and 0.62 (9, Table 1), 0.44 (5, Table 3), anxiety was the most stable (9, p524). 0.48 (6, pp85–86), 0.45 (6, Table 6), 0.58 (7, Table 3), and 0.70 (8, Table 7). This range is Validity broadly comparable with equivalent correlations Several studies have reported exploratory and/or between Beck’s anxiety and depression measures confirmatory factor analyses. All support the (see page 310). three-factor structure originally proposed, al- All three subscales proved capable of discrim- though certain items show cross-loadings in inating significantly between psychiatric outpa- some analyses but not in others (2; 5; 7; 8). tients with different diagnoses, and between Lovibond’s original exploratory factor analysis patients and unaffected people (5, Table 4). in a healthy sample specified a three-factor solu- Brown et al. showed significant contrasts on all tion and confirmed the placement of all items three DASS scales between samples of panic dis- except for the situational anxiety item “I feared order patients, generalized anxiety disorders, that I might be ‘thrown’ . . . ,” which loaded people with social phobias, and mood disorders more highly on the stress factor. A confirmatory such as depression (6, Table 4). The DASS de- factor analysis showed that a three-factor solu- pression scale had a sensitivity of 0.91 in detect- tion fit significantly better than a two-factor ing depression, but a specificity of 0.46; model; a second-order common factor ac- equivalent figures for the anxiety scale (in detect- counted for 50% of the variance in the depres- ing an anxiety disorder) were 0.92 and 0.40 (7, sion score, 74% of the anxiety score variance, Table 5). and 77% of the stress score (2, p338). Crawford and Henry reported modest corre- Subsequent analyses have proposed only mi- lations between DASS scores and age (−0.4 for nor variations in the placement of items (6, anxiety; −0.11 for depression and −0.18 for Table 3). Confirmatory factor analysis led stress). There was no correlation with occupa- Brown et al. to propose a slightly revised factor tional class and a negligible correlation with ed- placement that provided improved fit over Lovi- ucation and sex (8, Table 1). bond’s original. In this, depression and anxiety scores correlated 0.48, whereas stress scores cor- Alternative Forms related 0.68 with anxiety and 0.69 with depres- The DASS has been translated into Arabic, Chi- sion (6, pp85–86). Other studies have found a nese, Dutch, German, Hungarian, Japanese, Per- slight advantage of Brown’s result over the origi- sian, Spanish, and Vietnamese. Details and links nal solution (4; 8). Two studies of psychiatric to translated versions are given on the DASS outpatients also found that the 21-item abbrevia- website. tion showed a better fit to the three-factor model than the full 42-item version (4, p65; 5, Table 2). Reference Standards Several studies have reported convergent and General population norms (means and per- discriminant correlations between the DASS sub- centiles) are available from a British sample (8, scales and other measures. A summary of results Tables 2 and 3). The mean total score was 18.4 is provided in Table 6.5. (standard deviation, 18.8). The main surprise appears to be the low cor- relations between the anxiety scale and the STAI: Commentary 0.44 for the DASS-A full version, and 0.55 for The DASS is one of the newest anxiety or de- the abbreviated version, much lower than corre- pression scales; early results of psychometric lations with the Beck Anxiety Inventory (5, testing are extremely positive. The instrument Table 3). appears to provide a good indicator of the over- As one purpose of the DASS was to distin- all severity of negative emotions that correlates guish between anxiety and depression, several in a logical manner with other established scales. 318 Measuring Health Table 6.5 Correlations between the Depression Anxiety Stress Scales (DASS) and Other Measures Comparison Scale; Sample Depression Anxiety Stress Reference Beck Depression Inventory Undergraduates 0.74 0.58 0.60 (2, Table 2) Psychiatric outpatients (42-item DASS) 0.77 0.57 0.62 (5, Table 3) Psychiatric outpatients (21-item DASS) 0.79 0.62 0.69 (5, Table 3) Phobia and anxiety 0.75 0.49 0.61 (6, Table 6) Beck Anxiety Inventory Undergraduates 0.54 0.81 0.64 (2, Table 2) Psychiatric outpatients (42-item DASS) 0.42 0.84 0.64 (5, Table 3) Psychiatric outpatients (21-item DASS) 0.51 0.85 0.70 (5, Table 3) Phobia and anxiety 0.40 0.83 0.58 (6, Table 6) Positive and Negative Affect Scale (negative affect) Phobia and anxiety 0.57 0.63 0.72 (6, Table 6) General population 0.60 0.60 0.67 (8, Table 7) Positive and Negative Affect Scale (positive affect) Phobia and anxiety −0.45 −0.18 −0.20 (6, Table 6) General population −0.48 −0.29 −0.31 (8, Table 7) State-Trait Anxiety Inventory (trait scale) Psychiatric outpatients (42-item DASS) 0.65 0.44 0.59 (5, Table 3) Psychiatric outpatients (21-item DASS) 0.71 0.55 0.68 (5, Table 3) Hospital Anxiety and Depression Scale (depression score) General population 0.66 0.49 0.54 (8, Table 7) Employees with mental health problems 0.75 0.46 0.58 (7, Table 3) Hospital Anxiety and Depression Scale (anxiety score) General population 0.59 0.62 0.71 (8, Table 7) Employees with mental health problems 0.53 0.66 0.60 (7, Table 3) The factor structure is clear and stable across not specific to depression. Similarly, the low cor- studies. The inclusion of the nonspecific stress relations with the State-Trait Anxiety Inventory dimension is supported theoretically and may reflect a lack of specificity of that measure matches the tripartite model of Clark and Wat- (10). During the development of the DASS, items son (1); this dimension brings the DASS close to included in the Beck scale such as disturbance of the content of the general psychological well- appetite, weight loss, sleeping difficulties, tired- being scales reviewed in Chapter 5. It extends ness, or lack of energy and irritability, were re- coverage to symptoms (e.g., tension, irritability) jected as not being specific to depression (2, that are not included in instruments such as the p341). The correlation between the DASS anxi- Beck anxiety or depression scales, but it should ety and depression scales (0.48 to 0.70) appear also be borne in mind that the DASS focuses on comparable with those between the Beck anxiety discriminating between anxiety and depression and depression scales, but the omission of non- rather than providing comprehensive coverage of specific items from the DASS may have increased these constructs. Hence, it omits several com- specificity at the expense of sensitivity. There is mon symptoms of anxiety and of depression that little information on the sensitivity of the DASS may be useful in screening or case-detection. to change. Commenting on the coverage of the DASS, Lovi- The response scale represents a mix of perti- bond attributed the relatively low correlation be- nency of the symptom and its frequency. tween the DASS and Beck depression scales to Whether this can be interpreted as indicating in- the inclusion of items on the Beck scale that are tensity of the symptom may be debatable, but  Anxiety 319 given that the DASS has a broad scope, the over- (5) Antony MM, Bieling PJ, Cox BJ, et al. all effect is to produce a scale that records the di- Psychometric properties of the 42-item and versity of symptoms experienced by a person, 21-item versions of the Dperession Anxiety modified by their frequency. This would appear Stress scales in clinical groups and a very relevant in a general screening instrument, community sample. Psychol Assess 1998;10:176–181. but it would tend to give a relatively low score to (6) Brown TA, Chorpita BF, Korotitsch W, et a person who is crippled by intense feelings of al. Psychometric properties of the one particular type. Depression Anxiety Stress Scales (DASS) in The results of factor analyses suggests that clinical samples. Behav Res Ther the DASS does distinguish depression, anxiety, 1997;35:79–89. and stress but, as so often shown before, these (7) Nieuwenhuijsen K, de Boer AEGM, are correlated with each other, especially the Verbeek JHAM, et al. The Depression stress and anxiety scales. Anxiety Stress Scales (DASS): detecting The 21-item version appears, if anything, su- anxiety disorder and depression in perior to the full 42-item version; convergent va- employees absent from work because of lidity correlations are consistently higher and it mental health problems. Occup Environ Med 2003;60 (suppl 1):i77–i82. has a cleaner factor structure, so for general sur- (8) Crawford JR, Henry JD. The Depression vey use or as an outcome measure, there is little Anxiety Stress Scales (DASS): normative point in using the longer version. However, one data and latent structure in a large non- advantage of the longer version is the broader clinical sample. Br J Clin Psychol range of symptoms included, which may provide 2003;42:111–131. clinically useful information (5, p181). Future (9) Lovibond PF. Long-term stability of investigations will indicate whether the DASS depression, anxiety, and stress syndromes. proves capable of serving also as a screening in- J Abnorm Psychol 1998;107:520–526. strument. (10) Bieling PJ, Antony MM, Swinson RP. The State-Trait Anxiety Inventory, trait version: Address structure and content re-examined. Behav Res Ther 1998;36:777–788. The DASS web site is at www.psy.unsw.edu .au/dass/. The State-Trait Anxiety Inventory References (STAI) (1) Clark LA, Watson D. Tripartite model of (C.D. Spielberger, 1968, 1977) anxiety and depression: psychometric Purpose evidence and taxonomic implications. J This self-report measure indicates the intensity Abnorm Psychol 1991;100:316–336. of feelings of anxiety; it distinguishes between (2) Lovibond PF, Lovibond SH. The structure state anxiety (a temporary condition experi- of negative emotional states: comparison of enced in specific situations) and trait anxiety (a the Depression Anxiety Stress Scales (DASS) with the Beck Depression and general tendency to perceive situations as threat- Anxiety Inventories. Behav Res Ther ening). It was originally developed as a research 1995;33:335–343. instrument to study anxiety in normal adult (3) Lovibond SH, Lovibond PF. Manual for population samples, but it can also be used to the Depression Anxiety Stress Scales. 2nd screen for anxiety disorders and can be used ed. Sydney: Psychology Foundation, 1995. with patient samples (1, p3). (4) Clara IP, Cox BJ, Enns MW. Confirmatory factor analysis of the Depression-Anxiety- Stress Scales in depressed and anxious Conceptual Basis patients. J Psychopathol Behav Assess Spielberger developed the STAI in the context of 2001;23:61–67. his experimental work on the effects of anxiety 320 Measuring Health on learning, in which anxiety was used as an in- students. Items for the scale were drawn from dex of drive level in research guided by hullian the TMAS, from Cattell’s scale, and adapted learning theory; existing measures of anxiety from Zuckerman’s Affect Adjective Checklist; were considered inadequate for use in such re- further items were written as the development search (2). Early measures of anxiety such as progressed (1, p19; 6). The initial intent was to Taylor’s 1953 Manifest Anxiety Scale (TMAS) select items that could be used both for state and or Cattell’s Anxiety Scale Questionnaire did not trait anxiety, but using a different time reference distinguish between the intensity of anxiety as (e.g., feelings now versus feelings in general). an emotional state and individual differences in This proved infeasible, however, because some anxiety as a personality trait (3). During the of the most suitable items for each scale could 1960s, Cattell used multivariate analyses to ex- not be phrased to fit the other: “I feel upset” was amine the structure of questionnaire items a good state item, but did not fit on the trait deemed to measure anxiety and empirically dis- scale (1, p9; 3, p11). For the original version tinguished between trait and state components (Form X), Spielberger selected 20 items for each (3, p9). Meanwhile, Spielberger reviewed Tay- scale, of which five were common to both. The lor’s scale and concluded that it tapped a broad trait scale was largely based on the TMAS, construct that should be differentiated into com- whereas state anxiety items were chosen as being ponents such as stress, threat, state, and trait sensitive to stressful situations (6). Copies of the anxiety (2, p139). He proposed an instrument Form X version are included in the 1970 manual with trait and state subscales that would repre- (1, pp20–21). sent Cattell’s perspective and also reflect Freud’s During the late 1970s, Form X was revised to theory of anxiety as a response to danger (3, discriminate more clearly between anxiety and p10). depression and to improve psychometric proper- State anxiety refers to transitory unpleasant ties, including the factor structure. Twelve of the feelings of apprehension, tension, nervousness, original 40 items were replaced, including those and worry, accompanied by activation of the items that did not work well with less-educated autonomic nervous system; it reflects how respondents and other items that appeared to threatening a person perceives his environment measure depression (7–10). This led to Form Y, to be. Spielberger referred to it as “a temporal which, with recent minor modifications, remains cross-section in the emotional stream-of-life of in current use. However, Form X continued to a person” (3, p10). Trait anxiety is a personal- be used in published studies well into the 1990s, ity disposition that describes a person’s ten- and because much of the validity and reliability dency to perceive situations as threatening, and evidence for the STAI was derived from the Form hence to experience state anxiety in stressful X version, the following review combines infor- situations (4). Trait anxiety is not observed di- mation from both versions. Form Y is shown in rectly but is expressed as state anxiety when the 1983 manual (7, pp36ff ). stress is experienced (5, p204). Spielberger Items one through 20 measure situational or drew an analogy with energy: trait anxiety state anxiety (STAI-S), and items 21 to 40 mea- would be equivalent to potential energy and sure underlying or trait anxiety (STAI-T). Both state anxiety to kinetic energy (1, p3). Because scales were intended to form unidimensional emotions play a crucial role in determining how measures. For the state items, respondents are patients react to a diagnosis and anxiety is cen- asked to indicate “How you feel right now, that tral in this process, Spielberger argued that it is is, at this moment.” Responses indicate intensity important to assess anxiety in routine clinical of feeling on a one to four scale, from “not at practice. all” through “somewhat,” “moderately so,” to “very much so.” For the trait items, the question Description concerns “how you generally feel” and the re- Development of the STAI began in the early sponse scale indicates frequency: “almost 1960s with research on high school and college never,” “sometimes,” “often,” and “almost al-  Anxiety 321 ways.” Roughly half the items are worded posi- As would be expected from the conceptual tively (indicating the absence of anxiety). After formulation of the STAI, trait scores are gener- reversing scores for positively worded items, to- ally more stable than state scores. In studies over tal scores for state and trait are calculated, rang- various time-delays, Spielberger et al. reported ing from 20 to 80. If one to three items are not retest figures for Form X that ranged from 0.73 answered, a score can be derived by calculating to 0.86 for trait scores, whereas state results the mean score on items answered, multiplying ranged from 0.16 to 0.54 (1, p9). Other findings this by 20 and rounding upward to the next included one-week retest correlations ranging higher whole number. Where four or more items from 0.78 to 0.83 for the trait scores compared are not answered, the validity of that scale with 0.69 to 0.76 for state scores (21, Table 2); a should be questioned. mean of 0.86 for trait, compared with 0.52 for Copyright restriction prevents reproduction state has been reported (22, p197). The retest of the full scale here, but sample items are shown correlation after 11 days was 0.73 for the trait in Exhibit 6.8 and the content of the items is score (23, p59). Two-week figures were 0.86 for summarized in various publications (e.g., refer- trait and 0.40 for the state scale (24, p277), ences 11; 12, Table 1; 13, Table 1). Copies of the whereas seven-week retest correlations were STAI are available through the Mind Garden 0.85 for the trait and 0.68 for the state scale (25, company (www.mindgarden.com). p480). Eight-month retest correlations for a sample of medical students were 0.54 for trait Reliability scores and 0.26 for state; 11-month figures were Numerous studies have reported alpha internal 0.29 for trait and 0.15 for state scores (26, consistency, which is consistently high; selected p358). Higher figures have also been reported: results are shown in Table 6.6. state scale retest correlations after 11 months Internal consistency for the state anxiety scale were 0.51 for high school students and 0.49 for tends to be higher under conditions of stress, adults (27, p586). A Dutch study suggested that whereas there are only small differences for the although total trait scores may appear stable trait scale, although this is not completely con- over time, this may mask considerable instability sistent. For example, low-stress testing condi- in responses to individual items (28, p381). For tions led to lower split-half correlations, but to Form Y, 30-day retest values ranged from 0.71 higher K-R 20 correlations in the same sample to 0.75 for the trait scale and 0.34 to 0.62 for (17, Tables 2, 3). Item-total correlations ranged the state score. Test-retest values were markedly from 0.49 to 0.64 for the state scale, and from lower for female respondents than male (7, Table 0.38 to 0.69 for the trait scale (14, p38). 11). Exhibit 6.8 Examples of Items from the State-Trait Anxiety Inventory The S-Anxiety scale consists of twenty statements that evaluate how respondents feel “right now, at this moment” 1 = NOT AT ALL 2 = SOMEWHAT 3 = MODERATELY SO 4 = VERY MUCH SO A. I feel at ease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 B. I feel upset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 The T-Anxiety scale consists of twenty statements that evaluate how respondents feel “generally” 1 = ALMOST NEVER 2 = SOMETIMES 3 = OFTEN 4 = ALMOST ALWAYS A. I am a steady person . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 B. I lack self-confidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 Reproduced with permission of Mind Garden Inc, 1690 Woodside Rd # 202, Redwood City, CA 94061 (650) 261-3500. 322 Measuring Health Table 6.6 Internal Consistency Coefficients (alpha or K-R20) for the State-Trait Anxiety Inventory (STAI), Versions X and Y Study Sample State Scale Trait Scale Reference Students (Form X) 0.83–0.94 0.86–0.92 (4, p332) Students (Form Y) 0.86–0.94 0.90–0.91 (7, Table 1) Undergraduate male students 0.89 0.90 (1, Table 3) Undergraduate female students 0.89 0.89 (1, Table 3) Adult psychiatric outpatients 0.92 0.90 (14, p378) General population 0.95 0.91 (15, p283) Working adults (Form Y) 0.93 0.91 (7, Table 1) Anxious older adults 0.94 0.88 (16, Table 2) Students: low stress conditions 0.92 0.87 (17, Table 3) Students: high stress conditions 0.81 0.67 (17, Table 3) Students (two samples); low stress 0.92; 0.90 0.92; 0.88 (6, p533) Pain patients; different ethnic groups 0.93–0.95 0.92–0.95 (18, Table 3) Parents of young children 0.87 0.81 (19, Table 1) Surgical patients, cancer patients, and medical students 0.93 0.91 (20, Table 4) Validity two-factor solution separating state and trait Content validity was assessed by Okun et al., items (12, Table 1). Most studies, however, iden- who noted that the STAI covered five of eight tified several factors. One consistent finding was domains for generalized anxiety disorder in the that the positively and negatively worded items DSM-IV (29). (anxiety present versus anxiety absent) load on separate factors, especially for the state items. factorial validity. Because state and trait are For example, in samples of college and of high considered conceptually distinct but related con- school students under conditions of stress about structs, numerous studies have commented on examinations, Gaudry et al. extracted five and the empirical relationship between the two six factors, including a factor that merely distin- facets of the STAI. Correlations between state guished the positively worded items (4, and trait scales typically fall in the range of 0.7 pp335–339). Positively and negatively worded to 0.8 (6, p532), but appear to vary from sample items again loaded on separate factors in a study to sample (30); coefficients may be higher when of psychiatric outpatients (14, p40), whereas a the scale is administered under conditions of study of undergraduates grouped the trait items stress. In one study, for example, the correlation together but separated positively and negatively was 0.88 for a sample of 132 cancer patients fol- worded state items (11, p407). Conversely, an- lowing completion of chemotherapy, 0.73 for other study found the positively and negatively 121 medical students, and 0.52 for 194 surgical worded trait items to load on separate factors, patients immediately before their operation (20, but the state items did not (32, Table 1). Spiel- Table 5). A lower value of 0.43 was reported in a berger et al. found different factor structures for study of parents (19, Table 3). Trait anxiety may male and female respondents (9, Table 6–1). tend to predict state more strongly in males These inconsistencies contributed to the decision (range of correlations, 0.51–0.67) than in fe- to develop the revised Form Y version of the males (0.44–0.55) (1, p12). STAI, as described in some detail by Spielberger Early factorial studies of Form X investigated et al. (9, pp101–6). whether the state and trait items measured sepa- In 1980, Spielberger and Vagg et al. pub- rate concepts. Some analyses reported a single lished two separate factor analytic studies using factor (21; 31), whereas others produced a clear Form Y (8, Table 2; 9, Table 6–4). The studies  Anxiety 323 produced four factors that represented the dis- formly higher than scores for general population tinction between state and trait, and that be- samples (7, Table 8). Likewise, state anxiety tween positively and negatively worded items scores for military recruits tested under stressful (state anxiety absent and present, and trait anxi- training conditions were higher than those for ety absent and present). This structure was later college students of similar age; there was little upheld during confirmatory analysis (33), and in difference between the trait scores (7, Table 1). a study of geriatric patients (34, p449). As noted In a study of mixed diagnosis psychiatric outpa- by Spielberger et al., the results for Form Y tients, the state scale failed to discriminate sig- clearly support the state-trait distinction, nificantly between those with an anxiety whereas the anxiety present versus absent dis- disorder and those without; the trait scale did tinction reflects severity, whereby anxiety absent discriminate significantly, although at a lower items are sensitive to low levels of anxiety, and level than achieved by the Beck Anxiety Inven- anxiety present items are sensitive to higher lev- tory (14, p38). State and trait scores for patients els (9, p107). diagnosed with agoraphobia with panic attacks A factor analysis that included both the STAI were significantly higher than those for patients and the Beck Anxiety Inventory (BAI) showed with panic disorder alone (12, Table 2). the two to load on separate factors, suggesting The question of discriminating anxiety from that they may represent separate concepts (25, depression has led many studies to correlate Table 4). scores on anxiety and depression scales. A study A Rasch analysis showed that six of the 20 of undergraduate students found that the STAI trait items did not meet the scaling criteria; it correlated more strongly with the Beck Depres- also identified redundancies among several sion Inventory (BDI) than did the BAI. The coef- items, and yet showed that there is inadequate ficients were 0.71 for state and 0.78 for trait, coverage at the low trait-anxiety end of the con- compared with 0.63 for the BAI (25, p482). In a tinuum. Results for the state scale also showed sample of outpatients with anxiety disorder, the six items that did not meet scale criteria, and Trait scale correlated 0.73 with the BDI, whereas that most of the items were grouped around the the State scores correlated 0.59, compared with midpoint of the scale, making for a relatively in- 0.50 for the BAI (23, Table 1). (Note that the efficient measurement scale (35, Tables 1 and 2). correlation for the Trait scale matches its test- retest reliability correlation (23, p60)). Similarly, convergent validity. Many studies have com- in a comparison with the Depression Anxiety pared the STAI with other scales. The original Stress Scales (DASS), the correlation for the studies of Spielberger et al. produced correla- STAI-T was higher with the DASS depression tions between the trait scale and the TMAS score than with the DASS Anxiety score (36, ranging from 0.79 to 0.83 in three samples, Table 3). A comparison with the Center for Epi- whereas correlations with Cattell’s IPAT scale demiologic Studies Depression scale (CES-D) ranged from 0.75 to 0.77 (1, Table 6). (Note again showed a higher correlation (0.71) for that the TMAS and Institute for Personality and trait anxiety than for state (0.44) (19, Table 3). Ability Testing (IPAT) scales record trait, rather The impression is that the STAI scales are not than state, anxiety and that these correlations specific for anxiety and so correlate with depres- approach the reliabilities of the scales). How- sion scores more highly than some other leading ever, as so often happens, subsequent authors anxiety scales. This was confirmed by Bieling et have reported somewhat lower coefficients; a se- al. who used hierarchical confirmatory factor lection of results is included in Table 6.7. analysis to show that all 20 trait items loaded on a single negative affect factor, but that distinct discriminant validity. The Form Y manual lower-order depression and anxiety factors also provides mean scores for diverse neuropsychi- existed (13, Table 2). They derived subscores atric patient groups, and these scores are uni- from the anxiety and depression items and 324 Measuring Health Table 6.7 Convergent Validity Correlations for the State-Trait Anxiety Inventory-Y Comparison S-Anxiety T-Anxiety Reference Beck Anxiety Inventory 0.52 0.44 (14, Table 3) 0.47 0.58 (23, Table 1) 0.56 0.57 (25, p482) 0.64 0.68 (25, p482) Depression Anxiety Stress Scales 0.44 (36, Table 3) (42-item version; Anxiety scale) Depression Anxiety Stress Scales 0.55 (36, Table 3) (21-item version; Anxiety scale) Anxiety diary 0.53 0.34 (23, Table 1) Fear Questionnaire (2 samples) 0.16, 0.38 0.34, 0.43 (16, Tables 4, 8) Worry Scale (2 samples) 0.22, 0.41 0.40, 0.57 (16, Tables 4, 8) Test Anxiety Inventory 0.41 to 0.64 (3, p13) showed that the anxiety items correlated more by the presence of physical illness in a study of highly with the BAI than the depression items HIV patients (37, p363). Responses did not ap- did, whereas depression items correlated more pear to vary according to racial group or gender highly than the anxiety items with the depres- (18). sion scale of the DASS (13, Table 3). Alternative Forms responsiveness. One consistent finding is that Based on their Rasch analysis, which showed under stressful and nonstressful testing condi- considerable redundancy in the STAI, Tenen- tions, trait scores remain relatively stable, baum et al. proposed an abbreviated version whereas state scores show a significant change in with ten trait items and nine state items that the expected direction (11; 17, p278; 24, p276; would have greater discriminal efficiency (35, 31, pp578–579). Another consistent finding is p244). Based on item-total correlations, that the positively worded state items are more Marteau and Bekker selected six items from the sensitive to a range of levels of stress; negatively state scale to form an abbreviated form that had worded items react to higher levels of stress and an alpha of 0.82 and correlated 0.95 with the so produce a ceiling effect (4, pp335–339; 11, 20-item state scale. The six items are shown in p409). their report (38, p306). The question of whether trait scores do, in- For use in studies requiring repeated mea- deed, predict change in state scores under condi- sures, van Knippenberg et al. proposed alterna- tions of stress has not been fully resolved. It is tive forms of both state and trait scales, each clear that state scores change following stress with eight items and each giving equivalent re- and that state and trait scores are correlated, but sults. The internal consistency of the eight-item in most studies the pattern of change in state forms reduced the alpha reliability coefficients scores following stress is very comparable for by .07 for both scales compared with the 20- high and low trait anxiety groups (see, for exam- item versions (20, p998). ple, Figure 1 in reference 11). It may be, how- A children’s version (STAI-C) was developed ever, that trait scores do predict state responses and standardized for children aged eight to 12 to ego threats, but not threats of physical harm years old, but it can be used with children as (11, p410). young as five years old if the items are read to The STAI was found to be uncontaminated them (39). It uses a three-point answer scale,  Anxiety 325 provides more explicit instructions, and includes it has been used with students, with adults in the different items. This version has also been tested community, with military personnel, and even for use with elderly respondents, on the basis of prison inmates (14). The manuals are informa- its being easier for them to understand than the tive and provide clear instructions on adminis- standard version (40). For the state scale, the re- tration. sults gave an alpha coefficient of 0.86 and a cor- Although it is used routinely in clinical set- relation with the Geriatric Depression Scale of tings, the STAI was developed largely with non- 0.51 (40, Table 3). clinical undergraduate college and high school The STAI has been translated into more than students (43, p896). Discriminant validity was 60 languages (CD Spielberger, personal commu- not specifically addressed in its development, so nication, 2005; some translations are listed at it is not clear whether it actually measures anxi- www.qolid.org/). ety alone or a mixture of anxiety and depression. After developing the STAI, Spielberger devel- Thus, the Trait scale correlated 0.73 with the oped several other scales, including the State- Beck Depression Inventory, and the State scale Trait Anger Scale (STAS); the State-Trait correlated 0.60 (43, p896). However, some of Personality Inventory (STPI), which covers state the criticisms of the STAI’s limited ability to dis- and trait forms of anxiety, anger, depression and tinguish anxiety from depression refer to the curiosity; and the State-Trait Anger Expression original version, Form X (25, p478). This lack of Inventory (STAXI). All use a format very similar specificity may be less of an issue with the re- to that of the STAI. vised version, although results such as its higher correlation with the DASS depression scale than Reference Standards the anxiety scale (36, Table 3) suggest that the The Form X manual reports mean scores for STAI may still retain items that assess depression various population samples, including working in addition to anxiety. adults, college students, high school students, The distinction between state and trait scores various patient groups, and even prison in- is complex and a debate arose in the early 1980s mates (1, Tables 1–4 and 7–9; 7). It also re- over their relationship. As postulated, state ports percentile scores and t-scores for college scores change in reaction to threats whereas trait and high school students (Tables 1, 2). Like- scores do not. However, the two scores correlate wise, norms for Form Y are available from the quite closely and may load on a single general 1983 manual (7, Tables 1–8). Version Y mean factor. Furthermore, the extent to which trait state scores for working men were 35.7 (stan- scores (as postulated in the conceptual formula- dard deviation [SD], 10.4), and mean trait tion) predict change in state scores under condi- scores were 34.9 (SD, 9.2). Women had mean tions of threat seems to vary. One interpretation state scores of 35.2 (SD, 10.6), and trait scores is that the trait score primarily measures fear of of 34.8 (SD, 9.2) (7, Table 1). Typical scores failure or loss of self-esteem (11, p406). Ra- for people with diagnosed anxiety fall in the manaiah et al. argued that some items in each range 47 to 61 (41, p68) and psychiatric pa- scale are not specific to that scale, resulting in a tients generally score above the 80th percentile spuriously high correlation between state and of the adult norms. trait scores and complicating the interpretation of factor analyses (6). Spielberger and Vagg Commentary pointed out, however, that they had already ad- The STAI is one of the best-established anxiety dressed that concern in the transition from Form measures, having been used in thousands of X to Form Y of the STAI in 1980 (10). Although studies in many fields of health research; one lit- academic debates over details of the scales will erature review showed that it is cited ten times likely continue, it is clear that the STAI repre- more frequently than its nearest rival (42). In ad- sents one of the best measures of anxiety avail- dition to being used with many types of patients, able. 326 Measuring Health Address systematic evaluation. J Consult Clin Mind Garden Inc, 1690 Woodside Rd #202, Psychol 1976;44:406–412. Redwood City, CA 94061 (650) 261-3500 (12) Oei TPS, Evans L, Crook GM. Utility and (www.mindgarden.com). validity of the STAI with anxiety disorder Charles D. Spielberger, PhD, Center for Re- patients. Br J Clin Psychol 1990;29:429–432. search in Behavioral Medicine and Health Psy- (13) Bieling PJ, Antony MM, Swinson RP. The chology, Psychology Department, University of State-Trait Anxiety Inventory, trait version: South Florida, 4202 East Fowler Ave, Tampa, FL structure and content re-examined. Behav 33620-7200. Res Ther 1998;36:777–788. (14) Kabacoff RI, Segal DL, Hersen M, et al. Psychometric properties and diagnostic utility of the Beck Anxiety Inventory and References the State-trait Anxiety Inventory with older (1) Spielberger CD, Gorsuch RL, Lushene RE. adult psychiatric outpatients. J Anx Disord Test manual for the State-Trait Anxiety 1997;11:33–47. Inventory. Palo Alto, CA: Consulting (15) Lisspers J, Nygren A, Söderman E. Hospital Psychologists Press, 1970. Anxiety and Depression Scale (HAD): some (2) Spielberger CD. Anxiety: state-trait-pro- psychometric data for a Swedish sample. cess. In: Spielberger CD, Sarason IG, eds. Acta Psychiatr Scand 1997;96:281–286. Stress and anxiety. Vol. I. Washington, DC: (16) Stanley MA, Beck JG, Zebb BJ. Hemisphere Publishers, 1975:115–143. Psychometric properties of four anxiety (3) Spielberger CD. Assessment of state and measures in older adults. Behav Res Ther trait anxiety: conceptual and methodological 1996;34:827–838. issues. South Psychol 1985;2:6–16. (17) Metzger RL. A reliability and validity (4) Gaudry E, Vagg P, Spielberger CD. study of the State-Trait Anxiety Inventory. Validation of the state-trait distinction in J Clin Psychol 1976;32:276–278. anxiety research. Multivariate Behav Res (18) Novy DM, Nelson DV, Goodwin J, et al. 1975;10:331–341. Psychometric comparability of the State- (5) Reiss S. Trait anxiety: it’s not what you trait Anxiety Inventory for different ethnic think it is. J Anx Disord 1997;11:201–214. subpopulations. Psychol Assess (6) Ramanaiah NV, Franzen M, Schill T. A 1993;5:343–349. psychometric study of the State-Trait Anxiety (19) Orme JG, Reis J, Herz EJ. Factorial and Inventory. J Pers Assess 1983;47:531–535. discriminant validity of the Center for (7) Spielberger CD. Manual for the State-Trait Epidemiological Studies Depression (CES- Anxiety Inventory (Form Y). Palo Alto, D) scale. J Clin Psychol 1986;42:28–33. CA: Consulting Psychologists Press, 1983. (20) van Knippenberg FCE, Duivenvoorden HJ, (8) Vagg PR, Spielberger CD, O’Hearn TP. Is Bonke B, et al. Shortening the State-Trait the State-Trait Anxiety Inventory Anxiety Inventory. J Clin Epidemiol multidimensional? Pers Individ Dif 1990;43:995–1000. 1980;1:207–214. (21) Barker HR, Jr., Wadsworth AP, Jr., Wilson (9) Spielberger CD, Vagg PR, Barker LR, et al. W. Factor structure of the State-Trait The factor structure of the State-Trait Anxiety Inventory in a nonstressful Anxiety Inventory. In: Sarason IG, situation. J Clin Psychol 1976;32:595–598. Spielberger CD, eds. Stress and anxiety. (22) Newmark CS. Stability of state and trait Vol. 7. Washington, DC: Hemisphere anxiety. Psychol Rep 1972;30:196–198. Publishing, 1980:95–109. (23) Fydrich T, Dowdall D, Chambless DL. (10) Spielberger CD, Vagg PR. Psychometric Reliability and validity of the Beck Anxiety properties of the STAI: a reply to Inventory. J Anx Disord 1992;6:55–61. Ramanaiah, Franzen, and Schill. J Pers (24) Rule WR, Traver MD. Test-retest Assess 1984;48:95–97. reliabilities of State-Trait Anxiety Inventory (11) Kendall PC, Finch AJ, Jr., Auerbach SM, et in a stressful social analogue situation. J al. The State-Trait Anxiety Inventory: a Pers Assess 1983;47:276–277.  Anxiety 327 (25) Creamer M, Foran J, Bell R. The Beck (38) Marteau TM, Bekker H. The development Anxiety Inventory in a non-clinical sample. of a six-item short-form of the state scale of Behav Res Ther 1995;33:477–485. the Spielberger State-Trait Anxiety Inventory (26) Nixon GF, Steffeck JC. Reliability of the (STAI). Br J Clin Psychol 1992;31:301–306. State-Trait Anxiety Inventory. Psychol Rep (39) Spielberger CD, Edwards CD, Lushene RE, 1977;40:357–358. et al. Preliminary manual for the State-Trait (27) Layton C. Test-retest characteristics of Anxiety Inventory for children. Palo Alto, State-Trait Anxiety Inventory, A-state scale. CA: Consulting Psychologists Press, 1973. Percept Mot Skills 1986;62:586. (40) Rankin EJ, Gfeller JD, Gilner FH. Measur- (28) van der Ent CK, Smorenburg JM, Bonke B. ing anxiety states in the elderly using the The stability of the A-trait subscale of the State-Trait Anxiety Inventory for Children. STAI for stress and passage of time. J Clin J Psychiatr Res 1993;27:111–117. Psychol 1987;43:379–385. (41) Antony MM, Orsillo SM, Roemer L. (29) Okun A, Stein RE, Bauman LJ, et al. Practitioner’s guide to empirically based Content validity of the Psychiatric measures of anxiety. New York: Kluwer Symptom Index, CES-Depression Scale, Academic, 2001. and State-Trait Anxiety Inventory from the (42) Piotrowski C. The status of the Beck perspective of DSM-IV. Psychol Rep Anxiety Inventory in contemporary 1996;79:1059–1069. research. Psychol Rep 1999;85:261–262. (30) Kennedy BL, Schwab JJ, Morris RL, et al. (43) Beck AT, Epstein N, Brown G, et al. An Assessment of state and trait anxiety in inventory for measuring clinical anxiety: subjects with anxiety and depressive psychometric properties. J Consult Clin disorders. Psychiatr Q 2001;72:263–276. Psychol 1988;56:893–897. (31) Wadsworth AP, Jr., Barker HR, Barker BM. Factor structure of the State-Trait Anxiety Inventory under conditions of variable Conclusion stress. J Clin Psychol 1976;32:576–579. (32) Barker BM, Barker HR, Jr., Wadsworth Several other anxiety scales were considered for AP, Jr. Factor analysis of the items of the inclusion in this chapter, some closely related to State-Trait Anxiety Inventory. J Clin those we have reviewed. The Covi Anxiety Scale Psychol 1977;33:450–455. is a brief clinical rating that includes verbal re- (33) Bernstein IH, Eveland DC. State vs trait ports on behavioral and somatic symptoms of anxiety: a case study in confirmatory factor anxiety. Each is rated on a five-point scale analysis. Pers Individ Dif 1982;3:361–372. (1; 2). The Hopkins Symptom Checklist (HSCL) (34) Bouchard S, Ivers H, Gauthier JG, et al. Psychometric properties of the French contains an 11-item anxiety subscale with an version of the State-Trait Anxiety Inventory emphasis on somatic symptoms (e.g., heart (form Y) adapted for older adults. Can J pounding, sweating, dizziness) (3). The Ameri- Aging 1998;17:440–453. can College of Neuropsychopharmacology (35) Tenenbaum G, Furst D, Weingarten G. A (ACNP) has developed clinician checklists for statistical reevaluation of the STAI anxiety anxiety and depression, which require at least questionnaire. J Clin Psychol moderate levels of feelings of nervousness and of 1985;41:239–244. fear, and at least three other signs or symptoms (36) Antony MM, Bieling PJ, Cox BJ, et al. of anxiety (2, Table 5). Parkerson and Broad- Psychometric properties of the 42-item and head at Duke University have proposed a 21-item versions of the Depression Anxiety brief, seven-item anxiety and depression scale Stress scales in clinical groups and a community sample. Psychol Assess (4). This is shown in the Duke Scale reviewed in 1998;10:176–181. Chapter 10 (5, Figure 4). Further information is (37) Savard J, Laberge B, Gauthier JG, et al. available from the MAPI web site: www.qolid Evaluating anxiety and depression in HIV- .org/public/DUKE-AD.html. Endler et al. have infected patients. J Pers Assess produced several measures of anxiety. The 1998;71:349–367. Endler Multidimensional Anxiety scales includes 328 Measuring Health anxiety state, trait, and perceptions scales (6–8). (5) Parkerson GR, Jr. Users guide for Duke The trait scale gauges tendencies to experience health measures. Durham, NC: anxiety in different situations: social evaluation, Department of Community and Family physical danger, or ambiguous situations that Medicine, Duke University Medical Center, are unfamiliar to the person. By contrast with 2002. (6) Endler NS, Cox BJ, Parker JDA, et al. Self- Spielberger’s approach, Endler et al. do not rec- reports of depression and state-trait ommend an overall trait score by summing the anxiety: evidence for differential four trait components (6, pp241–242). Anxiety assessment. J Pers Soc Psychol reactions occur as an interaction between the 1992;63:832–838. relevant trait dispositions and the current situa- (7) Endler NS, Parker J, Bagby RM, et al. tion. For example, a person with a trait propen- Multidimensionality of state and trait sity to feel anxious in social situations might anxiety: factor structure of the Endler develop state anxiety reactions when meeting Multidimensional Anxiety Scales. J Pers prospective parents-in-law, but not necessarily Soc Psychol 1991;60:919–926. when doing a math test (6, pp238–239). (8) Endler NS, Edwards JM, Vitelli R. Endler Kellner has described the Symptom Ques- Multidimensional Anxiety Scales: manual. 1st ed. Los Angeles: Western Psychological tionnaire that includes 92 yes/no items to cover Services, 1991. anxiety, depression, hostility, and somatization (9) Kellner R. A Symptom Questionnaire. J (9; 10). There are also measures of particular as- Clin Psychiatry 1987;86:1609–1620. pects of anxiety, such as the Liebowitz Social (10) Bull MJ, Luo D, Maruyama G. Symptom Anxiety Scale (11; 12) and the Brief Social Pho- Questionnaire anxiety and depression bia Scale (13). Bouchard et al. offered a review scales: reliability and validity. J Nurs Meas of measures of panic disorders (14); finally, there 1994;2:25–36. is a very useful compendium of anxiety measures (11) Heimberg RG, Horner KJ, Juster HR, et al. for clinicians that gives brief reviews of about 39 Psychometric properties of the Liebowitz instruments (15). Social Anxiety Scale. Psychol Med 1999;29:199–212. (12) Fresco DM, Coles ME, Heimberg RG, et al. The Liebowitz Social Anxiety Scale: a References comparison of the psychometric properties (1) Lipman RS, Covi L, Downing RW, et al. of self-report and clinician-administered Pharmacotherapy of anxiety and depression. formats. Psychol Med Psychopharmacol Bull 1981;17:91–103. 2001;31:1025–1035. (2) Lipman RS. Differentiating anxiety and (13) Kobak KA, Schaettle SC, Greist JH, et al. depression in anxiety disorders: use of Computer-administered rating scales for rating scales. Psychopharmacol Bull social anxiety in a clinical drug trial. Dep 1982;18:69–77. Anx 1998;7:97–104. (3) Derogatis LR, Lipman RS, Rickels K, et al. (14) Bouchard S, Pelletier M-H, Gauthier JG, et The Hopkins Symptom Checklist (HSCL): al. The self-assessment of panic using self- a self-report symptom inventory. Behav Sci report: a comprehensive survey of validated 1974;19:1–15. instruments. J Anxiety Disord (4) Parkerson GR, Jr., Broadhead WE. 1997;11:89–111. Screening for anxiety and depression in (15) Antony MM, Orsillo SM, Roemer L. primary care with the Duke Anxiety- Practitioner’s guide to empirically based Depression Scale. Fam Med measures of anxiety. New York: Kluwer 1997;29:177–181. Academic, 2001.  7 Depression epression scales are among the best- itatively different states has been debated (3). D established of health measurements; as with the activities of daily living scales, some of the Snaith noted that depression best measures are over 20 years old. Most of the . . . is used to indicate quite different con- scales have received rigorous testing and have cepts . . . for some, clinical depression is an been used in numerous studies spanning many extension of grief, for some it is a set of self- countries. Our selection of depression scales was defeating attitudes, and for others it is the in- relatively simple for there is close agreement evitable result of adversity, while the among review articles over the ten-or-so leading medically oriented psychiatrist considers methods in this field. This chapter describes there is a state based upon malfunctions of purpose-built depression scales. In addition to neurotransmitter systems in the brain. (4, these, most of the psychological well-being scales p293) reviewed in Chapter 5 and the general health mea- sures in Chapter 10 contain items or brief sub- A depressed or sad mood is a normal reaction to scales covering depression. disappointment or loss; indeed, as with anxiety, As with disability and pain, depression is a it may hold survival advantages. Most mammals term whose familiar use in everyday language seem capable of depression in reaction to loss, complicates the acceptance of a precise, clinical and depression may reinforce the development of definition (1). social bonds between infant and mother due to the fear of separation; it may foster social com- Depression as an affect or feeling tone is a munication about negative experiences and ubiquitous and universal condition which as thereby support social learning and group be- a human experience extends on a continuum havior. It may also serve more direct physiologi- from normal mood swings to a pathological cal or intrapsychic functions of protection state. Thus, depression as a word can be used through withdrawal (5). Such situational depres- to describe: (i) an affect which is a subjective sion should not, however, be confused with dis- feeling tone of short duration; or (ii) a mood, orders of mood, in which the person suffers which is a state sustained over a longer period intense mental, emotional, and physical anguish of time; or (iii) an emotion, which is com- and substantial disability. The two probably dif- prised of the feeling tones along with objec- fer in kind, and certainly differ in seriousness; tive indications; or (iv) a disorder which has the worst outcome of major depression is sui- characteristic symptom clusters, complexes, cide. or configurations. (2, p330) Depression is a relatively common, but un- derdiagnosed condition in the general popula- “Depression” covers a wide range of states, from tion. In 1988, the World Health Organization feeling sad, helpless, or demoralized to grief, (WHO) launched a 14-nation study of psycho- poor self-esteem or a major depressive episode. logical problems in general health care settings, Whether these fall on a continuum or form qual- and set the prevalence of current depression at 329 330 Measuring Health 10.4%; almost half these cases had not been rec- types. The latter are characterized by the pres- ognized by primary care physicians (6, ence of mania, a mood that is elated, expansive, pp39–40). Depression is associated with comor- or irritable, associated with symptoms such as bidities, especially anxiety, and the WHO study hyperactivity, racing thoughts, or distractibility estimated that people with depression experience (9, p320). Bipolar disorder includes cyclical six or more disability days per month (6, Table mood swings in which the patient can present in 6). either the manic or depressive phase; cy- clothymia is a mild but chronic form of mood swings. Depression, meanwhile, can be divided Classifications of Depression on etiological lines into endogenous and reactive depression, whereas the distinction between There are many possible ways to classify any neurotic and psychotic depression reflects sever- disease or disorder—for example, in terms of ity (2). Other classifications consider the level of etiology, biology, or pattern of presenting symp- psychomotor effects and whether the depression toms. These may produce different classifica- is accompanied by other behavioral distur- tions, as seen with psychiatric disorders, and the bances. In terms of severity, depressive disorders continuing debate over the best way to classify can be subdivided into major depression and affective disorders such as depression has pro- dysthymia, each with subcategories. Both are duced rival schools of thought. The British characterized by a depressed mood, loss of in- school established diagnostic criteria for depres- terest or pleasure in usual activities anhedonia), sion based largely on the WHO’s International and associated symptoms such as changed ap- Classification of Disease (ICD). The ICD-10 petite or sleep disturbance. The difference is one criteria require that, for at least two weeks, the of degree, and dysthymia describes a milder dis- patient must have reported two or more of: ab- order that does not meet the criteria for major normally depressed mood, or marked loss of in- depression described later in this chapter. terest or pleasure, or decreased energy. In Whether the two conditions form a continuum, addition, at least two further symptoms must be whether dysthymia is a precursor of depression, identified from a list that includes loss of con- or whether they are separate disorders—all are fidence, self-reproach and guilt, thoughts of issues for further investigation. suicide, evidence of poor concentration, psy- Much of the challenge in assessing depres- chomotor agitation or retardation, sleep distur- sion lies in deciding when the boundaries divid- bance, and change in appetite and weight (7, ing normal depressed mood, dysthymia, and Table 1). The ICD criteria have been modified major depression are crossed. Very much the over the years, and have converged somewhat same issue occurs with dementia, in which the with the North American approach represented boundaries between normal cognitive declines of by the Diagnostic and Statistical Manual of the aging, impaired cognition, and full dementia are American Psychological Association (DSM), matters of continuing debate. In the case of de- currently in its fourth edition (DSM-IV) (8). In- pression, the occurrence of features such as delu- deed, the content validity of many of the depres- sions, marked loss of weight, and suicidal sion measurements is judged in terms of their threats are commonly taken as boundary mark- coverage of the DSM criteria. The DSM tradi- ers for major depression; several sets of criteria tion situates depression within the broad cate- have been proposed (10). The criteria in the gory of mood or affective disorders, whose fourth edition of the DSM (DSM-IV) take common feature is the patient’s disturbed mood change in affect as the central feature, accompa- (whether depressed or elated). This is a symp- nied by a depressed mood for most of the day, or tom classification, for mood disorders are bio- a markedly diminished interest in daily activi- logically heterogeneous and need not share a ties. Frequency and duration are also considered: common etiology (9). The DSM subdivides these symptoms should have occurred nearly mood disorders into depressive and bipolar every day for at least two consecutive weeks. In  Depression 331 addition, four or more further symptoms must the anxious patient appears apprehensive, be present to qualify for major depression; the sweats, and complains of nervousness, palpita- DSM-IV provides detailed definitions. These tions, and faintness; somatic signs include rapid further symptoms include significant change in breathing, tachycardia, and erratic blood pres- weight, insomnia or hypersomnia, psychomotor sure. Identifying depression is more challenging agitation or retardation, fatigue, feelings of and may require time-consuming interviews; it is worthlessness or guilt, diminished ability to con- rare that a depressed patient does not also ex- centrate, and suicidal ideation (8, p327). Severe hibit signs of anxiety (14, p1). However, a hier- depression is present when the person shows archical diagnostic approach is often followed in nearly all the symptoms of depression, and it which a clinician will overlook symptoms of keeps her or him from participating in regular anxiety in a depressed patient, so that depression activities; moderate depression is present when tends to overrule anxiety (15, p137). This tends the person has many of the symptoms and the to mean that in patients with a diagnosis of de- depression often keeps her or him from doing pression, anxiety scale scores may be high, the things he needs to do. Mild depression im- whereas the reverse is not as true; it also reduces plies that a person shows some of the symptoms the specificity of depression measures. Further- of depression, and doing regular activities takes more, both anxiety and depression form a com- extra effort. plex interplay of symptoms that may vary from These criteria have been criticized. For exam- patient to patient, making a standardized test ple, Boyd et al. commented on the requirement that distinguishes the two very difficult to de- that decline be present for a diagnosis of depres- sign. Because of the overlap of symptoms, it sion, offering an anecdotal case of an elderly proves much easier to develop more sensitive woman with virtually no social life, living in iso- tests than specific ones, and some of the general lation, who appeared depressed but did not meet psychological assessment scales described in the criteria because her low role performance Chapter 5 may perform as well as some of the could not be shown to have declined (11). Alter- purpose-built depression screening scales. native criteria were proposed in the Research Di- Depressive symptoms may also be con- agnostic Criteria (RDC) for major depressive founded with physical disorders. Clinicians are disorder (12). These require the following crite- encouraged to identify depression in their pa- ria to be present: dysphoric mood or pervasive tients, but this is complicated by reference to loss of interest that lasted at least one week and symptoms such as sleep disturbance, weight loss, for which assistance was sought, plus at least poor concentration or decreased energy that are five from a list of symptoms that include dis- common in patients with physical illness. Be- turbed sleep, poor appetite, loss of energy, loss cause several of the ICD-10 criteria include so- of interest, psychomotor agitation or retarda- matic changes that may be confounded with tion, feelings of self-reproach, diminished ability physical illness, Endicott proposed diagnostic to concentrate, and thoughts of suicide (13, criteria for use with medically ill patients that do Table 2). not refer to any somatic symptoms (16). Differing perspectives on the relationship be- The structure of the syndrome also compli- tween anxiety and depression were reviewed in cates scoring measurement methods. For exam- the introduction to Chapter 6 on anxiety. Al- ple, more symptoms may not imply more severe though the conditions share common symptoms depression and for this reason, clinicians rarely and may result from similar circumstances, in consider depressive symptoms as linear and ad- theory at least, the two are distinguishable and ditive, but classify them into dimensions or cate- anxiety is not generally seen as an aspect of de- gories (17). Hence, a more sophisticated scoring pression. Anxiety suggests arousal, engagement, algorithm may be required in place of an addi- and an attempt to cope with the situation; pure tive summary score. This issue also arises in depression entails withdrawal, disengagement, scoring screening tests for dementia, as seen in and inactivity. Anxiety is relatively observable: Chapter 8. 332 Measuring Health Measurement of Depression cannot distinguish depression from general malaise. Only a clinical rating can record non- The coexistence of many measurement scales verbal indicators of depression, and a clinical reflects the divergence of conceptual ap- rating can handle distorting factors such as de- proaches to depression and the fact that depres- nial or exaggeration. Somatic symptoms are not sion is a syndrome rather than a single entity. accurately recorded by self-assessment, produc- No one symptom is diagnostic for depression, ing false positive results among medically ill and and different people will exhibit widely differ- elderly persons (21, pp92,94), a difficulty that is ent symptoms. These are generally grouped into easier to control in the clinical interview. This affective (crying, sadness, apathy), cognitive echoes the debate over the Health Opinion Sur- (thoughts of hopelessness, helplessness, suicide, vey described in Chapter 5. worthlessness, guilt), and somatic (sleep distur- Because patients (especially those severely de- bance, change in energy level, appetite, sleep, pressed) may lack the clinical perspective re- elimination) (18; 19). Not all are present in quired to rate the severity of their symptoms, every case. Hence, measurements must cover self-rating methods also cannot accurately indi- several dimensions, and it is the choice of cover- cate severity (22, p361). Patients with milder age that distinguishes most rival scales. These neurotic forms of depression may rate them- differences in content in turn reflect underlying selves as more depressed than clinicians do; the differences—for example, in the etiological reverse seems true of patients with severe depres- theory of depression that different measures sion (20, p85). It may be more difficult to distin- represent (e.g., biological versus psychody- guish moderate and severe depression than mild namic)—or differences in the response system and moderate forms because in severely de- monitored (e.g., cognitive behavioral, physio- pressed patients, the symptom patterns become logic, or affective) (20). Depression measure- broader and more complex (23, p385), and in ments are divided into two major groups: such instances a clinical rating scale may prove self-rating methods and clinician-rating scales, superior. Hence, although depression is a subjec- which correspond roughly to their use in clini- tive phenomenon, there are clear limits to assess- cal versus epidemiological studies. The basic ing depression via self-report. Hamilton put it approach is the clinical rating, with self-ratings bluntly: “Self-rating scales are popular because offering a less costly alternative. A formal diag- they are easy to administer. Apart from the noto- nosis of depression requires the exclusion of rious unreliability of self-assessment, such scales other explanations for the symptoms, and this are of little use for semi-literate patients and are requires a clinical examination. The DSM-IV, no use for seriously ill patients who are unable to for example, requires the exclusion of possible deal with them” (24, p56). explanations such as the physiological effects Certain measurement concerns that exist in of drugs or medications and medical conditions other fields seem magnified in depression. The such as hypothyroidism or schizophrenia. Be- effects of the disorder complicate its measure- cause this requires a clinical assessment, it is ment; patients with severe depression may be un- widely accepted that self-assessed measures of able to communicate meaningfully (25) and may depression can identify the syndrome of depres- lack motivation to respond to questions. Assess- sion but, as with dementia, cannot be regarded ing older people holds the added complication of as diagnostic devices. a tendency among the elderly to deny depression Self-ratings naturally emphasize the subjec- (19); questions that ask directly about being de- tive and affective elements: the experience of de- pressed seem to be checked less frequently than pression. The practical advantages of the those phrased in less direct terms such as “. . . self-report are speed and economy because they have you been bothered by emotional problems do not require clinical observation. The disad- such as feeling anxious or depressed . . .” (26, vantages include the lack of specificity of self- p879). The frequent co-occurrence of cognitive report methods: they identify dysfunction but impairments also complicates the assessment of  Depression 333 depression, whereas somatic complaints due to ered for studies requiring a brief depression physical illness or to medication use may falsely screening instrument, and by clinicians who elevate scores on depression scales that have a seek to assess depression in outpatient medical somatic component (27). Characteristic biases in clinics. measuring depression include the apparent dif- ference between men and women in the preva- References lence of depressive symptoms. Some Western cultures sanction greater expression of feeling (1) Kendall PC, Hollon SD, Beck AT, et al. among women than among men, so women Issues and recommendations regarding use of the Beck Depression Inventory. score higher on depression scales (28, p976). Cognitive Ther Res 1987;11:289–299. Whether this indicates measurement bias or a (2) Zung WWK. From art to science: the real difference remains open to debate, and diagnosis and treatment of depression. methods such as the signal detection analyses Arch Gen Psychiatry 1973;29:328–337. used in pain measurement might equally be ap- (3) Cooke DJ. The structure of depression plied to correct for response thresholds. There is found in the general population. Psychol also a confounding effect whereby people in Med 1980;10:455–463. lower socioeconomic groups are likely to suffer (4) Snaith P. What do depression rating scales greater stress and more depression but also to measure? Br J Psychiatry exhibit different response styles and admit to 1993;163:293–298. more problems than the more educated. (5) Klerman GL. The nature of depression: mood, symptom, disorder. In: Marsella AJ, Hirschfeld RMA, Katz MM, editors. The measurement of depression. New York: Scope of the Chapter Guilford Press, 1987:3–19. (6) Sartorius N, Üstün TB, Lecrubier Y, et al. This chapter opens with reviews of five self- Depression comorbid with anxiety: results rating depression scales. The Beck and Zung from the WHO study on psychological scales were developed in the 1960s to measure disorders in primary health care. Br J severity of depression; they are among the most Psychiatry 1996;168(suppl 30):38–43. widely used of all health measures. They have (7) Le Fevre P, Devereux J, Smith S, et al. been widely tested, and we have extensive infor- Screening for psychiatric illness in the mation on their performance. Next, we review palliative care inpatient setting: a comparison between the Hospital Anxiety two depression screening instruments. The Cen- and Depression Scale and the General ter for Epidemiologic Studies Depression Scale Health Questionnaire-12. Palliat Med was designed for survey use, and the Geriatric 1999;13:399–407. Depression Scale was intended for elderly peo- (8) American Psychiatric Association. ple. We then review Lubin’s Depression Adjec- Diagnostic and statistical manual of mental tive Check Lists, which focus on mood and the disorders (DSM-IV). Washington, DC: affective aspects of depression. This is followed American Psychiatric Association, 1994. by reviews of three clinical rating methods: the (9) Weissman MM, Klerman GL. Depression: Hamilton Rating Scale for Depression, the Brief current understanding and changing trends. Assessment Scale, and the Montgomery-Åsberg Annu Rev Public Health 1992;13:319–339. Depression Rating Scale. We also review the (10) Zisook S, Click M, Jr., Jaffe K, et al. Research criteria for the diagnosis of Carroll Rating Scale for Depression, which is a depression. Psychiatry Res 1980;2:13–23. self-administered version of the Hamilton scale. (11) Boyd JH, Weissman MM, Thompson WD, The comparative strengths of these scales are et al. Screening for depression in a summarized in Table 7.1. Note that the Hospital community sample: understanding the Anxiety and Depression Scale (HADS), reviewed discrepancies between depression symptom in Chapter 6, could equally well have been in- and diagnostic scales. Arch Gen Psychiatry cluded in this chapter and should also be consid- 1982;39:1195–1200.  Table 7.1 Comparison of the Quality of Depression Measurements* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurement Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Beck Depression Inventory ordinal 21 clinical, interviewer, self many *** *** *** *** (Beck, 1961) screening (10–15 min) The Self-Rating Depression ordinal 20 clinical self many ** ** *** ** Scale (Zung, 1965) (10–15 min) Center for Epidemiologic ordinal 20 survey self many ** ** *** *** Studies Depression Scale (10–12 min) (Radloff, 1972) Geriatric Depression Scale ordinal 30 clinical, self many *** *** *** *** 334 (Brink and Yesavage, 1982) screening (8–10 min) Depression Adjective ordinal 32 research self many *** ** ** ** Check-Lists (Lubin, 1965) (3 mins per list) Hamilton Rating Scale for ordinal 21 clinical expert many *** ** *** ** Depression (Hamilton, 1960) (30–40 min) Brief Assessment Schedule— ordinal 21 clinical clinician several ** ** * ** Depression (MacDonald, 1982) Montgomery-Åsberg ordinal 10 research expert several ** ** ** *** Depression Rating Scale (20–60 min) (Montgomery, 1979) Carroll Rating Scale for ordinal 52 clinical, self several * * ** ** Depression (Carroll, 1981) research * For an explanation of the categories used, see Chapter 1, pages 6–7.  Depression 335 (12) Feighner JP, Robins E, Guze SB, et al. symptoms of depression among young Diagnostic criteria for use in psychiatric adults, institutionalized aged and research. Arch Gen Psychiatry noninstitutionalized aged. J Gerontol 1972;26:57–63. 1979;34:716–722. (13) Lipman RS. Differentiating anxiety and (28) Blumenthal MD. Measuring depressive depression in anxiety disorders: use of symptomatology in a general population. rating scales. Psychopharmacol Bull Arch Gen Psychiatry 1975;32:971–978. 1982;18:69–77. (14) Wetzler S, Katz MM. Problems with the differentiation of anxiety and depression. J The Beck Depression Inventory Psychiatr Res 1989;23:1–12. (Aaron T. Beck, 1961, Revised 1978) (15) Bramley PN, Easton AME, Morley S, et al. The differentiation of anxiety and Purpose depression by rating scales. Acta Psychiatr The Beck Depression Inventory, second version Scand 1988;77:133–138. (BDI-II) was designed as an indicator of the pres- (16) Endicott J. Measurement of depression in ence and degree of depressive symptoms consis- out-patients with cancer. Cancer tent with the DSM-IV (1–3). It is not intended as 1984;53:2243–2248. a diagnostic instrument. It has subsequently (17) Myers JK, Weissman MM. Use of a self- been used as a community screening instrument report symptom scale to detect depression and for clinical research (4). in a community sample. Am J Psychiatry 1980;137:1081–1084. (18) Brink TL, Yesavage JA, Lum O, et al. Conceptual Basis Screening tests for geriatric depression. The content of the BDI was guided by clinical Clin Gerontol 1982;1:37–43. observations made during psychotherapy rather (19) Thompson LW, Futterman A, Gallagher D. than on the basis of a particular theory of de- Assessment of late-life depression. pression (5). Because Beck stressed the cognitive Psychopharmacol Bull 1988;24:577–586. genesis of depression, the content of the BDI em- (20) Hughes JR, O’Hara MW, Rehm LP. Mea- phasizes the patient’s attitudes toward self. Beck surement of depression in clinical trials: an defined depression as: overview. J Clin Psychiatry 1982;43:85–88. (21) Beck AT, Steer RA, Garbin MG. an abnormal state of the organism manifested Psychometric properties of the Beck by signs and symptoms such as low subjective Depression Inventory: twenty-five years of evaluation. Clin Psychol Rev mood, pessimistic and nihilistic attitudes, loss 1988;8:77–100. of spontaneity and specific vegetative (22) Carroll BJ, Fielding JM, Blashki TG. signs. . . This particular variable may occur Depression rating scales: a critical review. together with any combination of other psy- Arch Gen Psychiatry 1973;28:361–366. chopathological variables such as anxiety, ob- (23) Biggs JT, Wylie LT, Ziegler VE. Validity of sessions, phobias, and hallucinations. (5, the Zung Self-rating Depression Scale. Br J pp201–202) Psychiatry 1978;132:381–385. (24) Hamilton M. A rating scale for depression. Beck argued for a broad content, seeing depres- J Neurol Neurosurg Psychiatry sion “as a complex disorder involving affective, 1960;23:56–62. cognitive, motivational, and behavioral compo- (25) Kearns NP, Cruickshank CA, McGuigan nents.” (5, p187). He recorded attitudes, behav- KJ, et al. A comparison of depression rating scales. Br J Psychiatry 1982;141:45–49. iors, and symptoms that were specific to (26) Paterson C. Seeking the patient’s clinically depressed patients and were consistent perspective: a qualitative assessment of with psychiatric literature at the time. These EuroQol, COOP-WONCA charts and were grouped into the 21 symptom categories MYMOP. Qual Life Res 2004;13:871–881. that make up the BDI. Beck observed that as de- (27) Zemore R, Eames N. Psychic and somatic pression deepens, the number of symptoms in- 336 Measuring Health creases, with a progression in the frequency and pp336–337). If the respondent selects more intensity of each (5, p188; 6, p154). Hence, the than one statement in any group, the higher structure of the BDI includes graded levels of in- value is recorded; if his feelings lie between two tensity of each symptom. alternatives, the one that is closer is scored. Computerized forms of the BDI are available (4, Description p80). There are three versions of the BDI. The original Beck warns against rigid adherence to set 1961 instrument was revised in 1978, and re- cutting-points; these should be chosen according vised again in 1996 to form the BDI-II (1). The to the application. The following guidelines are modifications brought the instrument in line commonly given: scores of less than 10 indicate with DSM-IV diagnostic criteria and responded no or minimal depression, 10 to 18 indicate to other criticisms of the instrument. As with the mild-to-moderate depression, 19 to 29 indicate 1978 revision, many of the 1996 changes related moderate-to-severe depression, and scores of 30 to the response categories, but changes were also or more indicate severe depression (4, p79). made to four items. The revised version is avail- Moran and Lambert suggested that scores of 10 able through the Psychological Corporation, but to 15 indicate mild, and 16 to 19 indicate mod- copyright forbids its reproduction here. How- erate depression (7, p270). For psychiatric pa- ever, copies can be found on the Web, for exam- tients, a screening cutting-point of 12/13 is ple by searching for a phrase such as “I do not suitable, whereas 9/10 is appropriate in screen- feel sad.” ing nonpsychiatric medical patients (6, p163). The BDI evaluates 21 symptoms of depres- Further discussion of cutting-points for the ear- sion, 15 of which cover emotions, four cover be- lier versions of the BDI was given by Kendall et havioral changes, and six somatic symptoms. al. (8). Scores on the BDI-II tend to be about Each symptom is rated on a four-point intensity three points higher than the BDI-I (1, p25). For scale and scores are added together to give a to- the BDI-II, scores of zero to 13 indicate minimal tal ranging from 0 to 63; higher scores represent depression; 14 to 19 indicate mild depression; more severe depression. Of the 21 items, 13 20 to 28 moderate, and scores of 29 to 63 severe form an abbreviated version, as indicated by as- (1, p11). If a single cutting-point is required, terisks in the following list. The 21 items cover 12/13 suggests dysphoria and 19/20 suggests sadness,* pessimism,* past failure,* loss of plea- dysphoria or depression (depending on the diag- sure,* guilty feelings,* punishment feelings, self- nostic criteria being used) (9, p85). dislike,* self-criticism, suicidal thoughts or wishes,* crying, agitation, loss of interest,* inde- Reliability cisiveness,* worthlessness,* loss of energy,* Beck’s original paper reported a split-half relia- changes in sleeping patterns, irritability,* bility of 0.86, or 0.93 with the Spearman-Brown changes in appetite,* difficulty concentrating, correction (2, p565; 5, p194). In their review ar- tiredness or fatigue, and loss of interest in sex (1, ticle, Beck et al. cited ten studies of psychiatric Table 1.1). The reading level of the revised ver- patients that yielded alpha coefficients ranging sion is that of a fifth- or sixth-grade student (4, from 0.76 to 0.95, with a mean of 0.86. In 15 p81). Most of the validation of the BDI was un- nonpsychiatric samples, alpha coefficients (al- dertaken on the earlier versions of the scale, al- phas) ranged from 0.73 to 0.92, with a mean of though some information has become available 0.81. Three of these studies concurrently re- on the 1996 revision. viewed the short form BDI in which the alphas The BDI was originally administered by a fell about 0.02 below those for the long form (4, trained interviewer, while the patient read a Table 1). Subsequent studies confirmed these copy (2, p562); it is more commonly self- findings (10–12). Gallagher reviewed reliability administered now, taking five to ten minutes. studies on older adults and found similar results Administration instructions are in the manual (13, p152). Schaefer et al. reported coefficient (1), and in the Appendix of Beck’s book (5, alphas of 0.94 for the long form for 15 psychi-  Depression 337 atric patients and 0.88 for 15 chemically depend- ples of nonpsychiatric patients, correlations ent patients (14, p416). ranged from 0.55 to 0.73, and the mean was For the 1996 BDI-II revision, alpha values ap- 0.60 (4, p89). Correlations with the Hamilton pear higher. Beck et al. cited alphas of 0.93 for Rating Scale for Depression (HRSD) ranged college students and 0.92 for outpatients (1, from 0.61 to 0.86 (mean 0.73) in five studies of p14), whereas a study by Dozois et al. obtained psychiatric patients; the results were 0.73 and an alpha of 0.89 for the BDI-I and 0.91 for the 0.80 in two studies of nonpsychiatric popula- BDI-II on the same sample (9, p85). Steer et al. tions (4, p89; 20). A correlation of 0.62 was re- reported an alpha of 0.92 for the BDI-II (15, ported with the Hopkins Symptom Checklist p186). Beck et al. in another study reported an depression score (21, Table 2). Oliver and Sim- alpha of 0.89 for the BDI-I and 0.91 for the BDI- mons validated the BDI against the Diagnostic II on the same sample of patients; the two instru- Interview Schedule (DIS) in a sample of 298 ments correlated 0.93 (16, pp593–4). community volunteers. At a cutting-point of Beck reviewed ten reports of retest reliability 9/10, sensitivity was 84.6% and specificity was of the BDI at varying time intervals; Pearson cor- 86.4% (22, Table 1). If respondents with dys- relations between administrations ranged from thymic disorder were excluded from the sample, 0.48 to 0.86 in five studies of psychiatric pa- sensitivity rose to 100% and specificity was tients and from 0.60 to 0.83 in six nonpsychi- 86%. In a study of 102 elderly outpatients, the atric populations (4, Table 2). Among elderly BDI had a sensitivity of 93.3% at a specificity of patients, test-retest reliability at six to 21 days 81.5% compared with the Schedule for Affec- was 0.79 in depressed patients, and 0.86 in nor- tive Disorders (23, p945). Compared with mal respondents (17, Table 1). A seven-week DSM-III diagnoses made by clinical psycholo- retest correlation was 0.79 for undergraduate gists, sensitivity was 83% and specificity was students (18, p480). Pearson retest correlations 82% at a cutting-point of 12/13 (24, Table 1). may, of course, mask systematic changes in the However, Whiffen warned that the BDI may distribution of scores over time, and there is miss cases of postpartum depression: in her some evidence that BDI scores may decline by small study, sensitivity was only 47.6% at a about three points over a period of a month in cutting-point of 9/10 (25, Table 1). Kearns et al. the absence of any form of intervention (19, Fig- found that the BDI discriminated poorly be- ure 1). tween levels of depression (26). In a sample of chronic pain patients, a cutting-point of 26/27 Validity proved optimal, giving a sensitivity of 81.8% Extensive information is available on the validity and a specificity of 72.7%, superior to results of the BDI. In terms of content validity, the BDI obtained by the CES-D (27, Table 4). covers six of the nine DSM-III criteria directly, A challenge lies in differentiating between two partially (it omits increases in appetite and anxiety and depression using self-reports; de- sleep), and omits agitation (7, p294). The 1996 pression and anxiety inventories may share items BDI-II has been found to correlate 0.93 with the in common, and the two conditions often occur earlier version; kappa agreement was 0.70 (9, together. In a study of psychiatric outpatients, p85). Beck et al. compared anxious and depressed pa- Beck reviewed 11 studies that showed the tients; BDI scores were higher for the depressed BDI capable of discriminating between groups patients, but not always significantly so (28, that contrasted in level of depression (4, p90). A Table 2). In the same study, the Beck Anxiety In- further 35 concurrent validation studies com- ventory (BAI) showed a better ability to discrim- pared the BDI with other ratings of depression inate between the two groups of patients than (4, Table 3). Fourteen studies reported correla- the BDI did (29). Steer et al. have evaluated the tions between the BDI and clinical ratings; coef- relationship between the BAI and versions I and ficients for psychiatric patients ranged from II of the BDI. They obtained a correlation of 0.55 to 0.96, with a mean of 0.72. In three sam- 0.58 for the BAI version I, and 0.59 for version 338 Measuring Health II (15, p186). Equivalent figures were 0.65 and 0.85 for GDS) in another (36, p407). Compared 0.66 in another study (16, Table 2), while the with the Center for Epidemiologic Studies De- BDI-II manual quotes 0.60 (1, p27). The BDI pression Scale, a study of primary care patients also shows high correlations with the State-Trait showed the BDI to perform slightly better in de- Anxiety Inventory: 0.74 for the state scale and tecting depression (37, Table 2), while a study in 0.77 for the trait scale in one study (18, p482). a veterans administration medical center showed Likewise, Matsumi and Kameoka found higher it to perform slightly worse (38, Table 2). The correlations between the BDI and Spielberger’s latter study showed the 13-item short form to trait anxiety scale (0.73) and Taylor’s anxiety perform very similarly to the full version: area scale (0.71) than with other measures of depres- under the ROC curve 0.87 for the long form, sion (0.68 with Zung and 0.54 with the Depres- compared with 0.86 for the short form (38, sion Adjective Check-lists) (30, Table 2). Steer et Table 2). A correlation of 0.74 with the depres- al. reported a correlation of 0.71 between the sion subscale of the Depression Anxiety Stress BDI-II and the SCL-90R anxiety scale (29, Table Scales has been reported in an Australian study 1). In a comparative study of the ability of de- (39, Table 2). In terms of divergent validity, the pression and anxiety measures to discriminate BDI correlated 0.14 with clinical ratings of anxi- between patients with major depressive disorders ety, a lower association than that obtained for and others with panic disorder, Somoza et al. the depression scale of the Multiple Affect Ad- found the BDI to be significantly less able to dis- jective Checklist (5, pp200–201). It correlated criminate than the Hamilton Rating Scale for 0.33 with the trait anxiety scale of the State- Depression, but significantly better than the Trait Anxiety Inventory; this figure was lower Beck Anxiety Inventory (31, Table 2). than the 0.47 reported between the STAI and the The BDI has been compared with most other GDS (34, Table 2). The BDI correlated 0.49 with depression measures. The mean correlation with the 12-item General Health Questionnaire, and the Zung Self-rating Depression Scale was 0.76 0.42 with Bradburn’s negative affect score in an in eight studies of psychiatric populations and Australian study (40, Table 1). Note, however, 0.71 in five nonpsychiatric samples (4, p89). that it also correlated -0.43 with a single-item Higher values of 0.89 (32, Table 4) and 0.86 “delighted-terrible” scale measure of satisfaction (24, Table 2) have been reported, as have lower with “your life as a whole” in the same study. values such as 0.60 (33, Table 2). Compared Reviews of the adequacy of the BDI as a mea- with the Minnesota Multiphasic Personality In- surement of change are mixed. Moran and Lam- ventory Depression scale, the mean correlation bert found the BDI to be more sensitive to was 0.76 in seven studies of psychiatric patients change than other instruments (7, p294), and 0.60 for three studies of nonpsychiatric whereas a more sophisticated meta-analysis of samples (4, p89). Correlations with the Multiple 19 studies with 1,150 respondents showed that Affect Adjective Checklist depression scale in the BDI was more conservative than the HRSD three studies ranged from 0.59 to 0.66 (4, p89), in estimating treatment change (20). while in Beck’s original studies the correlations Beck et al. reviewed 13 factor analytic studies ranged from 0.40 to 0.66 in different samples (5, of the BDI and concluded that it contains one p199). The BDI showed correlations of 0.79 underlying factor which may be subdivided into with the Geriatric Depression Scale (GDS) for 51 three correlated components: negative attitudes elderly volunteers (11, p1165), 0.78 in a sample toward self, performance impairment, and so- of patients with anxiety (34, Table 2), and 0.85 matic disturbance (4, p92). This second-order when used with medical outpatients (35, p994). factorial structure was confirmed by Byrne et al. The sensitivity and specificity of the BDI were (41). Welch et al. compared analyses from five slightly better than those of the GDS in one samples and identified a “large general factor study (35, Table 1), but were virtually identical that resulted in high internal reliability for the (area under the ROC curve 0.87 for Beck and measure as a whole” (42, p825). Using a latent  Depression 339 trait analysis, Clark et al. also found that a single whether their patients required more detailed dimension adequately described the data (43, psychological evaluation (51). p709). Their analysis suggested that the vegeta- tive symptoms (e.g., sleep disturbance, weight Alternative Forms loss) did not discriminate well between de- The BDI has been translated into many lan- pressed and nondepressed subjects. In a study guages, including French, Spanish (45; 52–54), comparing the BDI with the BDI-II revision, two German, Polish, Danish (4; 13, p154; 55, p127), factors (i.e., cognitive-affective, somatic- Chinese (56; 57), Lebanese (58), and Turkish vegetative) were found for both instruments, but (59). A study that compared Chinese and En- the revised version showed a clearer factor struc- glish language versions for bilingual respondents ture with fewer cross-loadings (1, Table 3.6; 9, obtained alpha values of 0.87 for the English Table 3). Other studies have reported slightly and 0.85 for the Chinese versions; similar, two- different two-factor solutions for the BDI-II (so- factor structures were obtained for both versions matic-affective and cognitive) that can be sub- (60, pp618–620). sumed under a single, general factor (44; 45). In In 1972, Beck and Beck proposed a 13-item a study of the BDI-II and the Beck Anxiety In- abbreviation of the BDI-I for use by family ventory, items from the two scales fell onto sep- physicians as a rapid screening tool (61). Items arate factors; only the agitation item of the were selected on the basis of their item-total cor- BDI-II and the BAI item on fear of the worst relations and of their agreement with a clini- showed strong cross-loadings (15, Table 1). cian’s ratings. Total scores of 0 to 4 on the short The validity of the short-form BDI has been form indicate no or minimal depression; 5 to 7 tested in several studies. Turner and Romano indicate mild depression; 8 to 15, moderate; and found a sensitivity of 83% and a specificity of 16 or more, severe (6, p164; 61, Table 2). Fol- 89% at a cutting-point of 7/8; at 3/4 sensitivity lowing development of the BDI-II, Beck again was 100% and specificity was 57% (24, Table proposed an abbreviation, this time with seven 1). Scogin et al. found the sensitivity of the short items, the BDI-PC. Items cover sadness, loss of form to be 98% and specificity 65%, using 4/5 pleasure, suicidal thoughts, pessimism, past fail- as a cutting-point. Raising the cutting-point ure, self-dislike and self-criticalness (62, p786; to 5/6 increased specificity to 77% (46, 63, p212). Each item is rated on a 0 to 3 scale, pp855–856). The short form has good internal giving a maximum score of 21. One-week retest consistency (alpha = 0.78 to 0.86) (10, p1168; reliability was 0.82, and alpha was 0.86 (62, 47, p769). Correlations with the long form lie pp786, 788). Another estimate of alpha was between 0.89 and 0.97 (4, p80; 10, p1168; 47, 0.88 and item-total correlations ranged from p769; 48, p1185); correlations with clinical rat- 0.51 to 0.74 (63, p214). A correlation of 0.62 ings of depression lie between 0.55 and 0.67 was obtained with the depression score from (48, p1186). Factor analytic studies suggest that the Hospital Anxiety and Depression scale the short form measures somewhat different as- (HADS), and the area under the ROC curve pects of depression than the long form; its cov- (AUC) was 0.92 in detecting mood disorders erage of somatic components is weaker (47; 49). (62, pp788–789). In a study of stroke patients, A 1993 review by Volk et al. identified weak- the area under the ROC curve for the BDI-I was nesses with items that were subsequently re- 0.79, slightly lower than the value of 0.83 for placed in the 1996 revision; using confirmatory the HADS. Results were better for men (AUC factor analysis, Volk et al. proposed a two- 0.86) than for women (0.69) (64, pp389–90). factor solution separating somatic and non- The AUC in a second study was 0.92 (63, p216). somatic items as being optimal (50, Figure 2). The BDI-PC correlated 0.86 with the abbrevi- Steer and Beck stressed that the short form ated version of the Beck Anxiety Inventory 0.86 should not replace the long form; its intent was (63, p215). A provocative result, however, was only to assist family practitioners in deciding presented by Chochinov et al. in a study of 197 340 Measuring Health terminally ill patients (65, Table 1). They changes (4, p85; 51), but the 1994 develop- showed that the 13-item Beck had a sensitivity ment of DSM-IV (70) formed the occasion for of 79% and specificity of 71%, whereas a single a revision. Most of the psychometric evidence question on depression, based on the Research on the BDI-II suggests that it is an improve- Diagnostic Criteria, had a sensitivity and speci- ment; although the overwhelming majority of ficity of 100%. evidence for validity and reliability has been A 27-item self-report version of the BDI, the collected using earlier versions, one may sup- Child Depression Inventory, is available for pose that the new version will perform at least school-aged children and adolescents (66). Steer as well. et al. review the use of the BDI with children A potential disadvantage of the inclusion of (55, pp133–134). A factor analysis and valida- somatic items in the BDI is that it may lead to tion identified four factors and concluded that false-positive results among patients with physi- the adolescent version has discriminant validity cal problems; an illustration has been given for for adolescents with comorbid conditions (67). pain patients (71). Another aspect of the speci- A version designed for use with deaf people is ficity debate concerns the association between given by Leigh et al. (68). BDI scores and anxiety; the BDI and BAI show positive correlations mostly in the range of 0.56 to 0.65, so the BDI may be falsely identifying Reference Standards anxious patients as depressed. In developing the Scores have been reported from studies of gen- BAI, Beck’s group paid particular attention to eral practice patients. Unfortunately, the propor- this issue and several reports consider the rela- tions of patients scoring above commonly used tionship between the two scales (see the review cutting-points varied widely in three reports (69, of the BAI in Chapter 6). The general finding Table 2). seems to be that the scales are capable of distin- guishing the two conditions, although they share Commentary a major portion of common variance that repre- The BDI is an established and well-researched sents negative affect in general. The BAI appears scale that is routinely included in psychological somewhat more specific to anxiety than the BDI test batteries. It has been as well tested as other is to depression. Several symptoms (e.g., insom- scales, but compared with the Zung and Lubin nia, loss of appetite) are common to both anx- instruments reviewed in this chapter, the BDI has ious and depressed patients, but symptoms such a broader coverage of the somatic aspects of de- as dissatisfaction, sense of failure, self-dislike, pression. This emphasis on somatic symptoms crying, fatigability, or social withdrawal are also characterizes the BAI which is reviewed sep- probably unique symptom markers of depres- arately. The revised BDI-II version appears to be sion (72, p651). In an analysis of the BDI and a refinement of an already successful scale. BAI together, D.A. Clark et al. identified a two- As with all of the self-rating depression factor first-order solution that distinguished anx- scales, the BDI has its critics, but the develop- iety from depression; there was then a ment of the 1996 BDI-II appears to have ad- higher-order general distress factor (72). This re- dressed many of these. The earlier comments sult matches the structure proposed by L.A. included the breadth of coverage of the BDI Clark and Watson (73). A small study has shown and difficulties with some of the somatic items that the BDI can discriminate between depressed (50). Items on increased weight, appetite, and patients and those with generalized anxiety dis- sleep (47), and psychomotor activity (7) were orders. The most discriminating questions were proposed. Vredenburg et al. tested a revised on sadness and loss of libido, which were less version and concluded that it had “a better fac- commonly checked by anxious patients (74, tor structure and somewhat higher reliability p477). than the original scale” (47, p775). Beck et al. Several authors have found a social desirabil- initially disputed some details of the proposed ity response bias in which people respond on the  Depression 341 basis of their judgment of the desirability or un- Address desirability of the item content (4, p92; 75). The Copies of the Beck inventory are available from BDI correlates with other self-report measures of The Psychological Corporation, Order Service maladaptive functioning (29) and so may in part Center, 555 Academic Court, San Antonio, reflect general “negative affect” (8). Langevin Texas, USA 78204-0952 and Stancer argued that this, together with the http://harcourtassessment.com results of factor analytic studies, demonstrate Information on the Beck Institute is available that the BDI reflects a social undesirability re- from http://www.beckinstitute.org/, while Dr. sponse set rather than depression (76). Beck re- Beck’s home page is at http://mail.med.upenn sponded that people who are depressed tend .edu/~abeck/index.html. toward low self-image; this indicates psy- chopathology and so the two are inherently con- founded. He pointed to the agreement between References the BDI and clinical ratings of depression to sug- (1) Beck AT, Steer RA, Brown GK. BDI-II. gest that the validity of the scale is not reduced Beck Depression Inventory. 2nd ed. San by the negative response set among depressed Antonio: Psychological Corporation, people (5, p207). 1996. Some studies of college students have made (2) Beck AT, Ward CH, Mendelson M, et al. much of the apparent instability of scores over An inventory for measuring depression. time (77; 78), but the original BDI measured Arch Gen Psychiatry 1961;4:561–571. (3) Bech P. Rating scales for affective current state, and it is not surprising that a state disorders: their validity and consistency. measurement of depression finds mood changes Acta Psychiatr Scand 1981;64(suppl over short periods. The intent of the revised ver- 295):1–101. sion is to measure more stable characteristics; (4) Beck AT, Steer RA, Garbin MG. unfortunately, its success in this remains un- Psychometric properties of the Beck known. Depression Inventory: twenty-five years of The ease of completing the BDI has been de- evaluation. Clin Psychol Rev bated: some suggest that the four-item response 1988;8:77–100. scales are difficult to administer (26); scales such (5) Beck AT. Depression: clinical, as the GDS use a simpler yes/no response format experimental, and theoretical aspects. New that may require less concentration on the part York: Harper & Row, 1967. (6) Beck AT, Beamesderfer A. Assessment of of the respondent (13, p158; 35, p994). depression: the Depression Inventory. Mod Because Beck never intended the BDI to be Probl Pharmacopsychiatry used for diagnosing depression (1, 2), much of 1974;7:151–169. the extended debate over whether it can be used (7) Moran PW, Lambert MJ. A review of in this way is misplaced. Most commentators current assessment tools for monitoring agree that although the BDI is sensitive, it may changes in depression. In: Lambert MJ, have low specificity, and that high scores on the Christensen ER, DeJulio SS, eds. The mea- BDI cannot be taken as diagnostic for depression surement of psychotherapy outcome in (8). For example, Oliver and Simmons found research and evaluation. New York: Wiley, that “the number of subjects scoring depressed 1983:263–303. on the BDI was about 21⁄2 times the number di- (8) Kendall PC, Hollon SD, Beck AT, et al. Issues and recommendations regarding use agnosed by the DIS” (22, pp893–894). Respon- of the Beck Depression Inventory. dents who receive intermediate scores on the Cognitive Ther Res 1987;11:289–299. Beck should be considered dysphoric, and only (9) Dozois DJA, Dobson KS, Ahnberg JL. A with a fuller clinical interview should they be psychometric evaluation of the Beck termed “depressed” (8, p297). With this warn- Depression Inventory-II. Psychol Assess ing, the BDI is one of the best depression screen- 1998;10:83–89. ing tools available. (10) Gould J. A psychometric investigation of 342 Measuring Health the standard and short form Beck (22) Oliver JM, Simmons ME. Depression as Depression Inventory. Psychol Rep measured by the DSM-III and the Beck 1982;51:1167–1170. Depression Inventory in an unselected (11) Gatewood-Colwell G, Kaczmarek M, adult population. J Consult Clin Psychol Ames MH. Reliability and validity of the 1984;52:892–898. Beck Depression Inventory for a white and (23) Gallagher D, Breckenridge J, Steinmetz J, Mexican-American gerontic population. et al. The Beck Depression Inventory and Psychol Rep 1989;65:1163–1166. research diagnostic criteria: congruence in (12) Lisspers J, Nygren A, Söderman E. an older population. J Consult Clin Psychol Hospital Anxiety and Depression Scale 1983;51:945–946. (HAD): some psychometric data for a (24) Turner JA, Romano JM. Self-report Swedish sample. Acta Psychiatr Scand screening measures for depression in 1997;96:281–286. chronic pain patients. J Clin Psychol (13) Gallagher D. The Beck Depression 1984;40:909–913. Inventory and older adults. Review of its (25) Whiffen VE. Screening for postpartum development and utility. Clin Gerontol depression: a methodological note. J Clin 1986;5:149–163. Psychol 1988;44:367–371. (14) Schaefer A, Brown J, Watson CG, et al. (26) Kearns NP, Cruickshank CA, McGuigan Comparison of the validities of the Beck, KJ, et al. A comparison of depression Zung, and MMPI depression scales. J rating scales. Br J Psychiatry Consult Clin Psychol 1985;53:415–418. 1982;141:45–49. (15) Steer RA, Clark DA, Beck AT, et al. (27) Geisser ME, Roth RS, Robinson ME. Common and specific dimensions of self- Assessing depression among persons with reported anxiety and depression: the BDI-II chronic pain using the Center for versus the BDI-IA. Behav Res Ther Epidemiological Studies-Depression scale 1998;37:183–190. and the Beck Depression Inventory: a (16) Beck AT, Steer RA, Ball R, et al. comparative analysis. Clin J Pain Comparison of Beck Depression Inventories 1997;13:163–170. -IA and -II in psychiatric outpatients. J Pers (28) Beck AT, Epstein N, Brown G, et al. An Assess 1996;67:588–597. inventory for measuring clinical anxiety: (17) Gallagher D, Nies G, Thompson LW. psychometric properties. J Consult Clin Reliability of the Beck Depression Psychol 1988;56:893–897. Inventory with older adults. J Consult Clin (29) Steer RA, Ball R, Ranieri WF, et al. Psychol 1982;50:152–153. Further evidence for the construct validity (18) Creamer M, Foran J, Bell R. The Beck of the Beck Depression Inventory-II with Anxiety Inventory in a non-clinical sample. psychiatric outpatients. Psychol Rep Behav Res Ther 1995;33:477–485. 1997;80:443–446. (19) Ahava GW, Iannone C, Grebstein L, et al. (30) Tanaka-Matsumi J, Kameoka VA. J. Is the Beck Depression Inventory reliable Reliabilities and concurrent validities of over time? An evaluation of multiple test- popular self-report measures of depression, retest reliability in a nonclinical college anxiety, and social desirability. J Consult student sample. J Pers Assess Clin Psychol 1986;54:328–333. 1998;70:222–231. (31) Somoza E, Steer RA, Beck AT, et al. (20) Edwards BC, Lambert MJ, Moran PW, et Differentiating major depression and panic al. A meta-analytic comparison of the Beck disorders by self-report and clinical rating Depression Inventory and the Hamilton scales: ROC analysis and information Rating Scale for Depression as measures of theory. Behav Res Ther 1994;32:771–782. treatment outcome. Br J Clin Psychol (32) Plutchik R, van Praag HM. 1984;23:93–99. Interconvertibility of five self-report mea- (21) Winokur A, Guthrie MB, Rickels K, et al. sures of depression. Psychiatry Res Extent of agreement between patient and 1987;22:243–256. physician ratings of emotional distress. (33) Meites K, Lovallo W, Pishkin V. A Psychosomatics 1982;23:1135–1146. comparison of four scales for anxiety,  Depression 343 depression, and neuroticism. J Clin Psychol (45) Penley JA, Wiebe JS, Nwosu A. 1980;36:427–432. Psychometric properties of the Beck (34) Snyder AG, Stanley MA, Novy DM, et al. Depression Inventory-II in a medical Measures of depression in older adults sample. Psychol Assess 2003;15:569–577. with generalized anxiety disorder: a (46) Scogin F, Beutler L, Corbishley A, et al. psychometric evaluation. Dep Anx Reliability and validity of the short form 2000;11:114–120. Beck Depression Inventory with older (35) Norris JT, Gallagher D, Wilson A, et al. adults. J Clin Psychol 1988;44:853–857. Assessment of depression in geriatric (47) Vredenburg K, Krames L, Flett GL. medical outpatients: the validity of two Reexamining the Beck Depression screening measures. J Am Geriatr Soc Inventory: the long and short of it. Psychol 1987;35:989–995. Rep 1985;56:767–778. (36) Laprise R, Vézina J. Diagnostic perfor- (48) Beck AT, Rial WY, Rickels K. Short form mance of the Geriatric Depression Scale of Depression Inventory: cross-validation. and the Beck Depression Inventory with Psychol Rep 1974;34:1184–1186. nursing home residents. Can J Aging (49) Berndt DJ. Taking items out of context: 1998;17:401–413. dimensional shifts with the short form of (37) Zich JM, Attkisson CC, Greenfield TK. the Beck Depression Inventory. Psychol Screening for depression in primary care Rep 1979;45:569–570. clinics: the CES-D and the BDI. Int J (50) Volk RJ, Pace TM, Parchman ML. Psychiatry Med 1990;20:259–277. Screening for depression in primary care (38) Whooley MA, Avins AL, Miranda J, et al. patients; dimensionality of the short form Case-finding instruments for depression: of the Beck Depression Inventory. Psychol two questions are as good as many. J Gen Assess 1993;5:173–181. Intern Med 1997;12:439–445. (51) Steer RA, Beck AT. Modifying the Beck (39) Lovibond PF, Lovibond SH. The structure Depression Inventory: reply to Vredenburg, of negative emotional states: comparison of Krames, and Flett. Psychol Rep the Depression Anxiety Stress Scales 1985;57:625–626. (DASS) with the Beck Depression and (52) Novy DM, Stanley MA, Averill P, et al. Anxiety Inventories. Behav Res Ther Psychometric comparability of English- and 1995;33:335–343. Spanish-language measures of anxiety and (40) Headey B, Kelley J, Wearing A. Dimensions related affective symptoms. Psychol Assess of mental health: life satisfaction, positive 2001;13:347–355. affect, anxiety and depression. Soc Indicat (53) Bonicatto S, Dew AM, Soria JJ. Analysis of Res 1993;29:63–82. the psychometric properties of the Spanish (41) Byrne BM, Baron P, Larsson B, et al. The version of the Beck Depression Inventory in Beck Depression Inventory: testing and Argentina. Psychiatry Res cross-validating a second-order factorial 1998;79:277–285. structure for Swedish nonclinical (54) Suarez-Mendoza AA, Cardiel MH, adolescents. Behav Res Ther Caballero-Uribe CV, et al. Measurement of 1995;33:345–356. depression in Mexican patients with (42) Welch G, Hall A, Walkey F. The replicable rheumatoid arthritis: validity of the Beck dimensions of the Beck Depression Depression Inventory. Arthritis Care Res Inventory. J Clin Psychol 1990;46:817–827. 1997;10:194–199. (43) Clark DC, Cavanaugh SA, Gibbons RD. (55) Steer RA, Beck AT, Garrison B. The core symptoms of depression in Applications of the Beck Depression medical and psychiatric patients. J Nerv Inventory. In: Sartorius N, Ban T, eds. Ment Dis 1983;171:705–713. Assessment of depression. Berlin: Springer, (44) Arnau RC, Meagher MW, Norris MP, et 1986:123–142. al. Psychometric evaluation of the Beck (56) Zheng Y, Wei L, Lianggue G, et al. Depression Inventory-II with primary care Applicability of the Chinese Beck medical patients. Health Psychol Depression Inventory. Compr Psychiatry 2001;20:112–119. 1988;29:484–489. 344 Measuring Health (57) Shek DTL. Reliability and factorial (70) American Psychiatric Association. structure of the Chinese version of the Beck Diagnostic and statistical manual of mental Depression Inventory. J Clin Psychol disorders (DSM-IV). Washington, DC: 1990;46:35–43. American Psychiatric Association, 1994. (58) Bryce JW, Walker N, Peterson C. (71) Williams AC, Richardson PH. What does Predicting symptoms of depression among the BDI measure in chronic pain? Pain women in Beirut: the importance of daily 1993;55:259–266. life. Int J Mental Health 1989;18:57–70. (72) Clark DA, Steer RA, Beck AT. Common (59) Karanci NA. Patterns of depression in and specific dimensions of self-reported medical patients and their relationship with anxiety and depression: implications for causal attributions for illness. Psychother the cognitive and tripartite models. J Psychosom 1988;50:207–215. Abnorm Psychol 1994;103:645–654. (60) Chan DW. The Beck Depression Inventory: (73) Clark LA, Watson D. Tripartite model of what difference does the Chinese version anxiety and depression: psychometric make? Psychol Assess 1991;3:616–622. evidence and taxonomic implications. J (61) Beck AT, Beck RW. Screening depressed Abnorm Psychol 1991;100:316–336. patients in family practice: a rapid technic. (74) Steer RA, Beck AT, Riskind JH, et al. Postgrad Med 1972;52:81–85. Differentiation of depressive disorders from (62) Beck AT, Guth D, Steer RA, et al. generalized anxiety by the Beck Depression Screening for major depression disorders in Inventory. J Clin Psychol medical inpatients with the Beck 1986;42:475–478. Depression Inventory for Primary Care. (75) Cappeliez P. Social desirability response set Behav Res Ther 1997;35:785–791. and self-report depression inventories in (63) Beck AT, Steer RA, Ball R, et al. Use of the the elderly. Clin Gerontol 1989;9:45–52. Beck Anxiety and Depression Inventories (76) Langevin R, Stancer H. Evidence that for primary care with medical outpatients. depression rating scales primarily measure Assessment 1997;4:211–219. a social undesirability response set. Acta (64) Aben I, Verhey F, Lousberg R, et al. Psychiatr Scand 1979;59:70–79. Validity of the Beck Depression Inventory, (77) Burkhart BR, Rogers K, McDonald WD, et Hospital Anxiety and Depression Scale, al. The measurement of depression: SCL-90, and Hamilton Depression Rating enhancing the predictive validity of the Scale as screening instruments for Beck Depression Inventory. J Clin Psychol depression in stroke patients. 1984;40:1368–1372. Psychosomatics 2002;43:386–393. (78) Hatzenbuehler LC, Parpal M, Matthews L. (65) Chochinov HM, Wilson KG, Enns M, et Classifying college students as depressed or al. “Are you depressed?” Screening for nondepressed using the Beck Depression depression in the terminally ill. Am J Inventory: an empirical analysis. J Consult Psychiatry 1997;154:674–676. Clin Psychol 1983;51:360–366. (66) Kovacs M. The Children’s Depression Inventory (CDI). Psychopharmacol Bull 1985;21:995–998. The Self-Rating Depression Scale (67) Bennett DS, Ambrosini PJ, Bianchi M, et al. (W.W.K. Zung, 1965) Relationship of Beck Depression Inventory Purpose factors to depression among adolescents. J The Self-Rating Depression Scale (SDS) was Affect Disord 1997;45:127–134. originally developed to quantify the severity of (68) Leigh IW, Robins CJ, Welkowitz J. current depression in patients of all ages with a Modification of the Beck Depression Inventory for use with a deaf population. J primary diagnosis of depressive disorder (1). It Clin Psychol 1988;44:728–732. has subsequently been used in clinical studies to (69) Williamson HA, Williamson MT. The monitor changes following treatment, as a Beck Depression Inventory: normative data screening instrument in family practice, and in and problems with generalizability. Fam cross-cultural studies (2; 3, p51). It is brief and Med 1989;21:58–60. simple to administer, yet comprehensive.  Depression 345 Conceptual Basis verted to an index by dividing the total by 0.8, Guided by previous literature and by factor ana- producing a range from 25 to 100 (2, Table 5). lytic studies, Zung classified the symptoms of Most guidelines for interpreting results suggest depression into pervasive affective disturbances that index scores of less than 50 are within the (e.g., sadness, tearfulness), somatic or physio- normal range, scores of 50 to 59 indicate mini- logical disturbances (e.g., sleep or appetite mal or mild depression, 60 to 69 moderate-to- changes), psychomotor disturbances (e.g., agita- marked depression, and scores above 70 indicate tion, retardation), and psychological distur- severe depression (3; 5; 6, p335; 7, p176). A cut- bances (e.g., confusion, emptiness, irritability) off of 54/55 is commonly used to identify de- (4; 5). Within each category, he identified nu- pression in people aged over 60 (E.M. Zung, merous symptoms. For example, the affective or personal communication, 1994). mood disorders include feelings and complaints of being depressed, sad, downhearted, and tear- ful; the physiological and somatic symptoms in- Reliability clude tachycardia, constipation, fatigue, sleep Several studies have estimated the internal con- disturbances, and decreased appetite, weight, sistency of the SDS; sample sizes ranged from and libido. The psychomotor disturbances are 100 to 225. Alphas ranging from 0.75 to 0.95 either retardation or agitation, whereas psycho- have been reported (9–11). Zung and Zung re- logical disturbances include confusion, indeci- viewed studies using the SDS with elderly peo- siveness, irritability and feelings of emptiness, ple; in four such studies alphas ranged from 0.59 hopelessness, personal devaluation, dissatisfac- to 0.87 (12, Table 2). Split-half reliability was tion, and suicidal ruminations (6, p332). These estimated at 0.73 (4, p225) and at 0.81 (13, subcategories guided the content of the SDS, and Table 2). Mean item-total correlations were 0.36 Zung selected items to reflect them from verba- in samples of New York drug addicts and Niger- tim records of patient interviews (2; 4, Table ian students; alphas were 0.75 and 0.79 (11, 21.1; 6). p28). The one-year test-retest correlation was 0.61 in 279 older subjects (14, p179). Description The SDS comprises 20 items; ten are worded positively and ten negatively. For each item, re- Validity spondents indicate the frequency with which The validity of the SDS has been extensively they experience the symptom or feeling, either at studied; review articles (3) and a meta-analysis the time of testing (7) or in the previous week summarize the results (15). The findings are (4). The 1974 version of the scale, slightly mod- mixed, however, giving an overall impression of ified from the original (5), is shown in Exhibit moderate validity. 7.1 (4, Table 21.2; 6, Table 3). Note that there is Moran and Lambert commented on content an error in item 13 in the version printed in ref- validity. They rated the SDS third out of six erence (4); the version shown in Exhibit 7.1 is scales in terms of coverage of the DSM-III crite- correct. The scale takes ten to 15 minutes to ria; it covers five DSM criteria well, four par- complete, according to the patient’s age and con- tially, and overlooks one variable (16, dition (8, p364). Toner et al. found that only pp286–287). 43% of a sample of elderly patients could re- Hedlund and Vieweg reviewed six factor ana- spond to all of the questions independently (9, lytic studies of the SDS, which identified be- p138). tween two and seven factors, with reasonable The exhibit also shows the response options; agreement over three dimensions, covering note the reverse scoring for the positive items. emptiness, agitation, and appetite change (3). It Item scores are added to form a total ranging has been observed, however, that the factors tend from 20 to 80, with higher scores indicating in- to separate positively and negatively worded creasing depression. The raw score is then con- items (17). Blumenthal found that 12 of the 346 Measuring Health Exhibit 7.1 The Zung Self-Rating Depression Scale Note: The scores shown here are not included in the version presented to the respondent; they are replaced by boxes in which the respondent places check marks. The raw scores are summed and converted to a 25–100 scale by dividing the total by 0.8. Please answer these items as they pertain to you now. None or Good Most a little Some of part of or all of of the time the time the time the time 1. I feel down-hearted, blue, and sad. 1 2 3 4 2. Morning is when I feel the best. 4 3 2 1 3. I have crying spells or feel like it. 1 2 3 4 4. I have trouble sleeping through the night. 1 2 3 4 5. I eat as much as I used to. 4 3 2 1 6. I enjoy looking at, talking to, and being with 4 3 2 1 attractive women/men. 7. I notice that I am losing weight. 1 2 3 4 8. I have trouble with constipation. 1 2 3 4 9. My heart beats faster than usual. 1 2 3 4 10. I get tired for no reason. 1 2 3 4 11. My mind is as clear as it used to be. 4 3 2 1 12. I find it easy to do the things I used to. 4 3 2 1 13. I am restless and can’t keep still. 1 2 3 4 14. I feel hopeful about the future. 4 3 2 1 15. I am more irritable than usual. 1 2 3 4 16. I find it easy to make decisions. 4 3 2 1 17. I feel that I am useful and needed. 4 3 2 1 18. My life is pretty full. 4 3 2 1 19. I feel that others would be better off if I were dead. 1 2 3 4 20. I still enjoy the things I used to do. 4 3 2 1 Adapted from Zung WWK. How normal is depression? Kalamazoo, Michigan: The Upjohn Company, 1981:17. Used with per- mission from Pfizer, Inc. All rights reserved. items loaded on four subscales, which she la- tors reflecting cognitive, affective and somatic beled well-being, depressed mood, somatic symptoms (19). symptoms, and optimism. However, the sub- In Zung’s early work the SDS identified sig- scales were not consistent across different sam- nificant differences among patients with depres- ples and she raised concerns that for elderly sion, anxiety, other psychiatric patients, and people the somatic items may falsely identify controls; it was also sensitive to treatment effects physical problems common with aging, thus re- (2, Table 6; 5; 7, Table 3; 20, Fig. 3). Biggs found ducing their validity as indicators of depression the SDS to discriminate significantly among lev- (18). This hypothesis was subsequently tested by els of depression (N = 26) (21, Table III). He Steuer et al., who found that the overall SDS noted, however, that it is easier for a self-rating scores did not correlate with physicians’ ratings scale to distinguish levels of depression than to of overall health, although one somatic item (fa- discriminate between diagnostic groups (21, tigue) did correlate significantly (rho = 0.35) (17, p382). Indeed, other studies have found the SDS p686). A principal components analysis of the to discriminate poorly among diagnostic groups Japanese version of the SDS identified three fac- (22; 23). Carroll et al. found that, unlike the  Depression 347 Hamilton Rating Scale for Depression (HRSD), format of the SDS is a limitation with severely the SDS failed to distinguish between severely depressed patients. Biggs et al. obtained Spear- depressed inpatients, moderately depressed day man correlations with the Hamilton scale rang- patients, and mildly depressed general practice ing from 0.45 to 0.76 in an intervention study patients (8, Table 1). Moran and Lambert re- followed over six weeks; the overall correlation viewed three discriminant validity studies and was 0.80 (21, Table I). They noted that the low- concluded that “the SDS is not a sensitive mea- est correlations were obtained at the start of the sure of depressive symptomatology” (16, p289). study, perhaps due to the narrower spread of de- They also reviewed five drug trials and found pression scores before treatment. In a sample of that the SDS was consistently less sensitive than 110 mentally retarded adults, Kazdin et al. ob- other instruments in detecting treatment effects tained a correlation of only 0.14 with the (16, Table 9.4). Meta-analysis by Lambert et al. Hamilton scale, whereas the Beck and Hamilton confirmed that the SDS showed smaller differ- scales correlated 0.24 (27, p1042). ences between treatment and control subjects Toner et al. compared the SDS with the de- than HRSD did, and somewhat smaller than the pression section of the abbreviated Comprehen- Beck Depression Inventory (BDI) (15, pp54,57). sive Assessment and Referral Evaluation The SDS correlated relatively highly, at 0.53, (Short-CARE) and to physicians’ ratings for a with Zung’s Anxiety Inventory (24, p135). sample of 80 elderly outpatients. Only 65% of The SDS shows moderate-to-high correla- patients completed enough of the SDS to permit tions with other instruments measuring depres- scoring. The Short-CARE agreed more closely sion. For example, Hedlund and Vieweg’s review (kappa = 0.46) with the physician’s rating of de- quotes five correlations between the SDS and the pression than did the SDS (kappa = 0.29) (9, Minnesota Multiphasic Personality Inventory p138). (MMPI) depression scale that ranged from 0.55 Correlations between SDS scores and psychi- to 0.70 (3, Table 1); other studies found correla- atrists’ ratings of depression severity include tions of 0.59 (7) and 0.76 (25). A review of sev- 0.20 and 0.69 (3, Table 1), 0.23 and 0.58 (28, eral studies comparing the SDS and the BDI p16). At a cutoff of 50, the SDS showed a sensi- reported correlations ranging ranged from 0.60 tivity of 83% for detecting depression in chronic to 0.83 (16, Table 9.2); Turner and Romano re- pain patients; specificity was 81%—results al- ported a correlation of 0.86 (25, Table 1). most identical to those obtained for the BDI Tanaka-Matsumi et al. found the SDS correlated (25). Using the same cutoff, Zung had reported more highly with Spielberger’s Trait Anxiety slightly higher figures of 88% for both sensitiv- scale (0.74) than with the BDI (0.68) or the Lu- ity and specificity among people aged 20 to 64 bin Depression Adjective Check Lists (0.54) (10, (5, p177). In four studies of elderly people, sen- Table 2). Snow and Crapo found the SDS to cor- sitivity ranged from 58 to 82%, while specificity relate −0.53 with Bradburn’s Affect Balance ranged from 80 to 87% (12, Table 2). The SDS Scale, and −0.61 with the Life Satisfaction Index had better sensitivity and specificity (76 and A (N = 205) (26, Table 2). 96%, respectively) for detecting clinically diag- Many studies have compared the SDS with nosed depression among 40 geriatric stroke pa- clinical ratings made using the HRSD; the find- tients than the Geriatric Depression Scale (GDS) ings vary widely. Hedlund and Vieweg quoted or the Center for Epidemiologic Studies Depres- 18 such results, with correlations mainly falling sion Scale (29, Table 3). Yet, in another study, in the range 0.38 to 0.80, but with outliers of sensitivity (80%) and specificity (84%) results 0.22 and 0.95 (3, Table 1). Zung and Zung for the SDS were lower than those for the GDS quoted figures of 0.79 and 0.80 in elderly people and Hamilton scales (30). (12, Table 2); Carroll obtained a correlation of 0.41 (8, Table 1). He noted that the agreement Alternative Forms between the Hamilton and Zung scales was The SDS has seen extensive use in North Amer- highest in mild depression as the self-assessed ica, the Far East, Australia, and Europe. It has 348 Measuring Health been translated into at least 30 languages and Commentary has been used in cross-cultural studies (31; 32). Zung’s SDS became one of the more widely used Zung developed a clinician’s rating scale ver- of the depression scales but, as so often happens sion of the SDS, the Depression Status Inventory with older methods, the early tests of validity (DSI) (4, Table 21.7; 5, p179; 33, Table 1). This showed a promise that was not borne out subse- was intended to facilitate combined self-ratings quently. Evidence for the reliability of the scale is and clinician-ratings of depression that, Zung restricted to its internal consistency, which is ad- argued, may be superior to each alone. Dis- equate. Results of validity studies have been parate scores might, for example, help distin- mixed; a few studies show good results; many guish people who overinterpret and ruminate show the SDS to be adequate; a few suggest about their condition (“sensitizers”), or others weaknesses. (“repressors”) who tend to avoid or deny feel- All self-rating depression scales encounter ings (5, p179). For 225 patients, the correlation problems of denial and of poor motivation to re- between the DSI and SDS was 0.87, and the in- spond among more severely depressed, with- ternal consistency of the DSI was 0.81 (4, p230; drawn patients. The 1973 review by Carroll et 33, p543). The DSI was used in the 1976 Na- al. pointed to limitations in self-rating methods tional Institute of Mental Health Early Clinical in general, suggesting that they may be adequate Drug Evaluation (ECDEU) assessment battery, to establish the presence of symptoms but not to which altered item 1 to read “I feel down- quantify their severity, because patients lack the hearted and blue,” and changed item 4 to “. . . clinical perspective required to rate severity (8, sleeping through the night,” and item 6 to “I p361). Specific criticisms of the SDS include its still enjoy sex” (34). focus on assessing the frequency rather than the Snaith et al. described the Wakefield Self- severity of symptoms, even though its intent was assessment Depression Inventory, a modified to assess severity. Most clinical ratings of depres- version of the SDS that retains ten of the 20 sion rate severity, and this may contribute to the items and adds two items covering anxiety (35). low agreement between the Beck or Hamilton They also used response categories that blend and the Zung scales (38, p321). The frequency frequency with severity: “Yes, definitely,” “Yes, response format does not suit items such as “I sometimes,” “No, not much,” and “No, not at notice that I am losing weight.” Some authors all.” have proposed changing to severity ratings to Because of the concern that the somatic items make the SDS more sensitive to the degree of de- could produce false-positive results with patients pression (39, p987); Snaith’s modifications are who have physical conditions, an 11-item abbre- an example (35). The mix of positive and nega- viation omits the somatic items. This has been tive items may confuse some patients (9), and the tested on cancer patients and found to have the negatively worded symptoms may measure a dif- same internal consistency as the full SDS, and to ferent construct than the those worded posi- correlate 0.92 with the full version (36). As an- tively (40). Like the BDI, the SDS correlates ticipated, it identified somewhat fewer patients highly with anxiety scales, suggesting that they with apparent depression. may measure a broader state than depression alone (10). With the exception of Zung’s own studies, Reference Standards most research has shown that the SDS discrimi- Zung and Zung provided age-specific reference nates only moderately well between diagnostic values from 938 normal controls (12, Table 3). groups (3). When used as a screening test, sensi- Zung also summarized results from 11 countries tivity and specificity seem adequate, although that show mean scores for depressed and normal somewhat lower than for other self-rating scales. respondents (37, Tables 3 and 6). The 1976 Recent reviews of the SDS are generally negative. ECDEU assessment battery provides some refer- Boyle and others view it as “a psychometrically ence values (34, p173). crude instrument” that should not be used in re-  Depression 349 search (3, p54; 8; 40, p54), whereas Rabkin and Self-Rating Depression Scale in the elderly. Klein stated that it is not the instrument of Clin Gerontol 1986;5:137–148. choice for assessing depressive severity or change (13) Yesavage JA, Brink TL, Rose TL, et al. after treatment, although it may have a role as an Development and validation of a geriatric adjunctive screening instrument (41). depression screening scale: a preliminary report. J Psychiatr Res 1983;17:37–49. (14) McKegney FP, Aronson MK, Ooi WL. Identifying depression in the old old. References Psychosomatics 1988;29:175–181. (1) Zung WWK. A Self-Rating Depression (15) Lambert MJ, Hatch DR, Kingston MD, et Scale. Arch Gen Psychiatry al. Zung, Beck, and Hamilton rating scales 1965;12:63–70. as measures of treatment outcome: a meta- (2) Schotte CK, Maes M, Cluydts R, et al. analytic comparison. J Consult Clin Effects of affective-semantic mode of item Psychol 1986;54:54–59. presentation in balanced self-report scales: (16) Moran PW, Lambert MJ. A review of biased construct validity of the Zung Self- current assessment tools for monitoring Rating Depression Scale. Psychol Med changes in depression. In: Lambert MJ, 1996;26:1161–1168. Christensen ER, DeJulio SS, eds. The mea- (3) Hedlund JL, Vieweg BW. The Zung Self- surement of psychotherapy outcome in Rating Depression Scale: a comprehensive research and evaluation. New York: Wiley, review. J Operat Psychiatry 1983:263–303. 1979;10:51–64. (17) Steuer J, Bank L, Olsen EJ, et al. (4) Zung WWK. Zung Self-Rating Depression Depression, physical health and somatic Scale and Depression Status Inventory. In: complaints in the elderly: a study of the Sartorius N, Ban T, eds. Assessment of Zung Self-rating Depression Scale. J depression. Berlin: Springer, Gerontol 1980;35:683–688. 1986:221–231. (18) Blumenthal MD. Measuring depressive (5) Zung WWK. The measurement of affects: symptomatology in a general population. depression and anxiety. Mod Probl Arch Gen Psychiatry 1975;32:971–978. Pharmacopsychiatry 1974;7:170–188. (19) Sakamoto S, Kijima N, Tomoda A, et al. (6) Zung WWK. From art to science: the Factor structures of the Zung Self-Rating diagnosis and treatment of depression. Depression Scale (SDS) for undergraduates. Arch Gen Psychiatry 1973;29:328–337. J Clin Psychol 1998;54:477–487. (7) Zung WWK. Factors influencing the Self- (20) Zung WWK, Richards CB, Short MJ. Self- Rating Depression Scale. Arch Gen rating Depression Scale in an outpatient Psychiatry 1967;16:543–547. clinic. Arch Gen Psychiatry (8) Carroll BJ, Fielding JM, Blashki TG. 1965;13:508–515. Depression rating scales: a critical review. (21) Biggs JT, Wylie LT, Ziegler VE. Validity of Arch Gen Psychiatry 1973;28:361–366. the Zung Self-Rating Depression Scale. Br J (9) Toner J, Gurland B, Teresi J. Comparison Psychiatry 1978;132:381–385. of self-administered and rater-administered (22) Schnurr R, Hoaken PCS, Jarrett FJ. methods of assessing levels of severity of Comparison of depression inventories in a depression in elderly patients. J Gerontol clinical population. Can Psychiatr Assoc J 1988;43:P136–P140. 1976;21:473–476. (10) Tanaka-Matsumi J, Kameoka VA. (23) Shanfield S, Tucker GJ, Harrow M, et al. Reliabilities and concurrent validities of The schizophrenic patient and depressive popular self-report measures of depression, symptomatology. J Nerv Ment Dis anxiety, and social desirability. J Consult 1970;151:203–210. Clin Psychol 1986;54:328–333. (24) Bramley PN, Easton AME, Morley S, et al. (11) Jegede RO. Psychometric properties of the The differentiation of anxiety and Self-Rating Depression Scale (SDS). J depression by rating scales. Acta Psychiatr Psychol 1976;93:27–30. Scand 1988;77:133–138. (12) Zung WWK, Zung EM. Use of the Zung (25) Turner JA, Romano JM. Self-report 350 Measuring Health screening measures for depression in reports and clinical assessments of chronic pain patients. J Clin Psychol depressed outpatients. J Psychopathol 1984;40:909–913. Behav Assess 1991;13:313–324. (26) Snow R, Crapo L. Emotional bondedness, (39) Zinkin S, Birtchnell J. Unilateral subjective well-being, and health in elderly electroconvulsive therapy: its effects on medical patients. J Gerontol memory and its therapeutic efficacy. Br J 1982;37:609–615. Psychiatry 1968;114:973–988. (27) Kazdin AE, Matson JL, Senatore V. (40). Boyle GJ. Self-report measures of Assessment of depression in mentally depression: some psychometric retarded adults. Am J Psychiatry considerations. Br J Clin Psychol 1983;140:1040–1043. 1985;24:45–59. (28) Downing RW, Rickels K. Some properties (41) Rabkin JG, Klein DF. The clinical measure- of the Popoff Index. Clin Med ment of depressive disorders. In: Marsella 1972;79:11–18. AJ, Hirschfeld RMA, Katz MM, eds. The (29) Agrell B, Dehlin O. Comparison of six measurement of depression. New York: depression rating scales in geriatric stroke Guilford Press, 1987:30–83. patients. Stroke 1989;20:1190–1194. (30) Brink TL, Yesavage JA, Lum O, et al. Screening tests for geriatric depression. Clin Gerontol 1982;1:37–43. The Center for Epidemiologic (31) Zung WWK. A cross-cultural survey of Studies Depression Scale (CES-D) symptoms in depression. Am J Psychiatry (L.S. Radloff and the National 1969;126:116–121. Institute of Mental Health, (32) Zung WWK. A cross-cultural survey of USA, 1972) depressive symptomatology in normal adults. J Cross-cult Psychol Purpose 1972;3:177–183. The Center for Epidemiologic Studies Depres- (33) Zung WWK. The Depression Status sion Scale (CES-D) is a 20-item, self-report de- Inventory: an adjunct to the Self-Rating pression scale developed to identify depression Depression Scale. J Clin Psychol in the general population (1). 1972;28:539–543. (34) Guy W. Early clinical drug evaluation Conceptual Basis (ECDEU) assessment manual for The CES-D was designed to cover the major psychopharmacology. DHEW Pub No. 76- symptoms of depression identified in the litera- 338 ed. Rockville, Maryland: US Department of Health, Education, and ture, with an emphasis on affective components: Welfare National Institute of Mental depressed mood, feelings of guilt and worthless- Health, 1976. ness, feelings of helplessness and hopelessness, (35) Snaith RP, Ahmed SN, Mehta S, et al. psychomotor retardation, loss of appetite, and Assessment of the severity of primary sleep disorders (1, p386). These symptoms can depressive illness: Wakefield Self- be experienced in the absence of depression, but Assessment Depression Inventory. Psychol Radloff argued that in people without depres- Med 1971;1:143–149. sion they tend to be counterbalanced by positive (36) Dugan W, McDonald MV, Passik SD, et al. affect (1, p391). The CES-D accordingly covers Use of the Zung Self-Rating Depression positive affect in a two-dimensional model of Scale in cancer patients: feasibility as a well-being that reflects Bradburn’s Affect Bal- screening tool. Psycho-Oncology 1998;7:483–493. ance Scale or Watson’s Positive and Negative Af- (37) Zung WWK. How normal is depression? fect Scales. As with these scales, empirical Kalamazoo, Michigan: The Upjohn evidence suggests that the positive items on the Company, 1981. CES-D form a separate dimension, rather than (38) Polanio A, Senra C. Measures of merely being the inverse of the negative items (2, depression: comparison between self- p269).  Depression 351 Radloff recognized that considerable varia- (9; 10) and pain patients (11). Husaini et al. pro- tion exists in the presentation of symptoms of posed 16 of 17 (one standard deviation below depression. For example, lower socioeconomic the mean score for acutely depressed outpa- groups report more physical symptoms whereas tients) to indicate possible cases and 22/23 to in- higher socioeconomic groups express depression dicate probable cases of depression (12). The affectively. The heterogeneity of response im- validity results for the cutting-points of 15/16 plies low intercorrelations among items and that and 16/17 were virtually identical, so 15/16 is cutting-points are difficult to set. probably the most suitable (Exhibit 7.2). Description Reliability Items for the CES-D were selected from existing Radloff reported alpha coefficients of 0.85 for scales, including Beck’s Depression Inventory general population samples and of 0.90 for a pa- (BDI), Zung’s Self-rating Depression Scale tient sample; split-half reliability ranged from (SDS), Raskin’s Depression Scale, and the Min- 0.76 to 0.85 (1, Table 3). Alphas of 0.85 were nesota Multiphasic Personality Inventory (1). also reported by Roberts for black and for white The CES-D is normally self-administered but may respondents (13, Table 3). Himmelfarb and be used in an interview; there is, however, some Murrell reported alpha coefficients of 0.85 for a evidence of systematic differences in response be- community sample (N = 235) and 0.91 for a clin- tween interviewer- and self-administered versions ical sample (N = 88) (7, Table 1). An alpha of (3). Estimated administration times range from 0.86 was obtained for frail elderly people (14, seven to 12 minutes, the higher figures being for p161), and a figure of 0.88 for a sample of par- elderly people (4, p182). Items refer to the fre- ents of young children (15, Table 1). A lower al- quency of symptoms during the past week (see pha of 0.82 was obtained in a small sample of Exhibit 7.1). Each question uses a 0 to 3 re- Native American students (10, p233). A correla- sponse scale; except for the four positive ques- tion of 0.76 (N = 24) was obtained between ver- tions, a higher score indicates greater depression. sions administered by a nurse and a research Questions 4, 8, 12, and 16 were worded posi- assistant (16, p242). tively, in part to discourage a response set, and The concentration of the CES-D on recent their scores are reversed by subtracting the score symptoms may reduce test-retest reliability. from 3. Question scores are then summed to Radloff reported low retest correlations, running provide an overall score ranging from 0 to 60. If from 0.32 to 0.67; most coefficients fell between more than five items on the scale are missing, a 0.50 and 0.60 (1, Table 4). score is generally not calculated (5, p31). If one to five items are missing, scores on the com- Validity pleted items are summed (after reversal of the Okun et al. reported on the content validity of positive items); this total is divided by the num- the CES-D, which covers seven of the nine DSM- ber of items answered and multiplied by 20. IV symptoms of major depressive episode (17). Variants of the basic scoring approach are de- Radloff reported consistent results of factor scribed under “Alternative Forms”. analyses across three samples during the early Scores of 16 or more are commonly taken as development of the CES-D. These analyses iden- indicative of depression (6, p206). This is equiv- tified four factors, interpreted as depressed af- alent to experiencing six symptoms for most of fect, positive affect, somatic symptoms, and the previous week, or a majority of the symp- interpersonal problems (1, p397). The same fac- toms on one or two days. However, several alter- tor structure was obtained on samples of black, native cutting-points have been suggested. white, and Chicano populations by Roberts (13) Studies of elderly patients have found 19/20 to and has been widely replicated with only minor be superior (7, p339; 8, p443), whereas cutting- variations in a range of populations (2; 14; points of 24 or even 28 have been proposed to 18–22). There are remarkably few exceptions; reduce false-positive results among adolescents one study of elderly people suggested that factor 352 Measuring Health Exhibit 7.2 The Center for Epidemiologic Studies Depression Scale (CES-D) Note: Items 4, 8, 12, and 16 have their scores reversed before totalling. Instructions for questions: Below is a list of the ways you might have felt or behaved. Please tell me how often you felt this way during the past week. Rarely or none of Some or a Occasionally Most or the time little of or a moderate all of (less than the time amount of the the time During the past week: 1 day) (1–2 days) time (3–4 days) (5–7 days) 1. I was bothered by things that usually 0 1 2 3 don’t bother me. 2. I did not feel like eating; my appetite 0 1 2 3 was poor. 3. I felt that I could not shake off the 0 1 2 3 blues even with help from my family or friends. 4. I felt that I was just as good as 0 1 2 3 other people. 5. I had trouble keeping my mind on 0 1 2 3 what I was doing. 6. I felt depressed. 0 1 2 3 7. I felt that everything I did was 0 1 2 3 an effort. 8. I felt hopeful about the future. 0 1 2 3 9. I thought my life had been 0 1 2 3 a failure. 10. I felt fearful. 0 1 2 3 11. My sleep was restless. 0 1 2 3 12. I was happy. 0 1 2 3 13. I talked less than usual. 0 1 2 3 14. I felt lonely. 0 1 2 3 15. People were unfriendly. 0 1 2 3 16. I enjoyed life. 0 1 2 3 17. I had crying spells. 0 1 2 3 18. I felt sad. 0 1 2 3 19. I felt that people dislike me. 0 1 2 3 20. I could not get “going.” 0 1 2 3 Adapted from Radloff LS. The CES-D Scale: a self-report depression scale for research in the general population. Appl Psychol Measurement 1977; 1:387, Table 1. Copyright 1977 West Publishing Co./Applied Psychological Measurement Inc. Reproduced by permission. structures may differ by race and gender (23), scores for the CES-D. This is because the scale as while Liang et al. proposed a three-factor model a whole has high internal consistency (4); the that excluded the interpersonal problem factor factors also intercorrelate (14, Table 4) and not (24). Although the four-factor solution is rela- all items load significantly on the factors. A dis- tively consistent, it is generally recommended cordant viewpoint was, however, expressed by that this not be used as a basis for forming sub- Schroevers et al., who questioned the value of  Depression 353 the positively worded items in a depression CES-D scores and used MIMIC structural equa- scale, and presented a two-factor solution that tion modeling to separate variance in response to separated the 4 positively-worded items from the the somatic items due to depression from that remainder. Furthermore, they showed consider- due to physical illness. Items such as “everything ably stronger correlations between the negative an effort” or “could not get going” showed clear items and other instruments than for the positive bias due to physical illness (28, Table 2). The items (e.g., a correlation of 0.64 with the STAI CES-D showed significant differences between state scale, compared with 0.39 for the positive respondents who reported needing help for an affect questions; and 0.74 with the GHQ com- emotional problem and those who did not; mean pared with 0.25 for the positive questions) (25, scores also varied significantly with the presence Table 5). of life events (1, Tables 7, 8). The sensitivity and specificity of the CES-D The issue of the specificity of depression have been frequently reported and generally ap- scales (as, indeed, of anxiety scales) is a peren- pear very good. Using the Raskin scale as a crite- nial one. Both may identify general psychologi- rion, Weissman et al. reported high sensitivity cal distress rather than depression per se. Thus, for detecting depression in various patient in a study of primary care patients, the CES-D groups. Sensitivity was 99% for acute depres- scores correlated as highly with measures of sion, 94% (at a specificity of 84%) for detecting anxiety as the anxiety scales correlated among depression among alcoholic patients, and 93% themselves. The correlation with the Beck Anxi- (specificity 86%) for schizophrenic patients (6, ety Inventory was 0.68 (29, Table 1); this did not Table 7). Shinar et al. reported a sensitivity of appear to be due merely to the co-occurrence of 73% at a specificity of 100% for the cutting- depression and anxiety. Correlations with other point of 16 or over (N = 27) (16, Table 4). Parikh general distress scales are also high: correlations et al. obtained a sensitivity of 86% at a speci- with Bradburn’s Affect Balance Scale ranged ficity of 90% for the cutting-point of 16 for 80 from 0.61 to 0.72; correlations with Langner’s stroke patients (26, Table 3). In a sample of pa- 22-item scale ranged from 0.54 to 0.60, and tients with chronic pain, a cutting-point of those with the Lubin Depression Adjective 20/21 proved optimal but gave a relatively low Check Lists were from 0.51 to 0.70 (1, Table 6). sensitivity of 68.2% and a specificity of 78.4 In primary care settings, however, the low (11, Table 3). In a sample of parents of disabled prevalence of depression may mean that even children, sensitivity was 87.5% and specificity reasonably high specificity does not guarantee 73% at the cutting-point of 15 of 16; however, high predictive values. For example, in a Nether- sensitivity for anxiety was almost as good, at lands study, sensitivity was 100% compared 80% (specificity 73%) (27, Tables 2 and 3). with the Diagnostic Interview Schedule and The question of the specificity of the CES-D specificity was 88%, but with a prevalence of for depression has been examined by several au- only 2%, the positive predictive value was just thors. Weissman et al. compared CES-D scores 13% (30, p233). Zich et al. tested the CES-D in from a community sample, a group of depressed a primary care setting; sensitivity was 100% and patients, and patients with other psychiatric con- specificity 53% at the cutting-point of 16 (kappa ditions. Average scores were 38.1 for 148 0.12); at a cutting-point of 27, specificity rose to acutely depressed patients, 14.9 for 87 recovered 81% (kappa 0.34) (2, Table 2). They suggested depressive patients, 13 for 50 schizophrenic pa- that in a primary care setting with a low preva- tients, and 9.1 for 3,932 adults in the commu- lence of depression, the cutting-point should be nity (6, Table 3). The scale proved capable of raised above the conventional 16. Similarly, a identifying depression among alcoholic and study of adolescents reported 100% sensitivity schizophrenic patients; Weissman concluded that compared with the DSM-III criteria, and a the CES-D was not merely measuring overall specificity of 81% at a cutting-point of 28 (10, psychiatric impairment. However, Grayson et al. p233). In another study of primary care patients, examined the influence of physical disorders on the sensitivity was 79% at a specificity of 77% 354 Measuring Health using the cut-point of 16. However, in predicting been proposed (24, Table 1; 35; 36). Unfortu- psychiatric disorder of any kind, the sensitivity nately, these are not identical and at present was 72% and specificity 80%, suggesting that there seems little agreement over which short- the CES-D is not specific to depression (29, ened version should become standard. Kohout et p25). Whooley et al. reported a sensitivity of al. proposed 10- and 11-item abbreviations to 88% and a specificity of 74% using the Diag- reduce response burden for elderly people (4). nostic Interview Schedule as a criterion; the These cover the four factors in the complete scale equivalent results for the Beck Depression Inven- and both include items 6, 7, 11, 12, 14–16, and tory were 84% and 71% (31, Table 2). A dis- 18–20 in our Exhibit, whereas the 11-item also crepant report is that of Myers and Weissman, version includes item 2. The 10-item version who found a sensitivity of only 63.6% for de- used a Yes/No format, referring to having expe- tecting major depression, at a specificity of rienced the symptom “much of the time during 93.9% in 482 adults living in the community the past week,” while the 11-item version used a (32). Boyd et al. examined the discrepant cases 3-point answer format (“Hardly ever or never”, from this study; the eight false-negative cases in- “Some of the time” and “Much or most of the cluded four patients who denied their symptoms time”) (4, Fig. 2). Alpha internal consistency for and three who had difficulty in completing the the 11-item version was 0.81, compared with questionnaire (33). Lyness et al. found the CES- 0.86 for the full scale and 0.73 for the 10-item D and the Geriatric Depression Scale to perform Yes/No version (37, p186). Andresen et al. pro- equally well in detecting major depression in an posed a 10-item abbreviation for elderly respon- elderly population, but the CES-D proved less dents that included items 1, 5–7, 8, 10–12, 14, able to detect minor depression (34, Table 2). and 20, using a 4-point response scale (37). This Radloff reported a correlation of 0.56 with a has been found to perform almost as well as the clinical rating of severity of depression (1, 20-item version: area under the ROC curve of p393). In a different sample, the CES-D scores 0.88 compared with 0.89 for the full scale (31, correlated 0.44 with the Hamilton Rating Scale Table 2). Santor and Coyne proposed a version for Depression at admission and 0.69 after treat- for the general population that included items 1, ment (N = 35); the equivalent correlations with 3, 5–7, 11, 12, 16, and 18, scored dichoto- the Raskin scale were 0.54 at admission and mously (38, p237). This version correlated 0.93 0.75 after treatment (1, p393). Weissman et al. with the original CES-D, whereas the sensitivity reported correlations between the CES-D and and specificity of the abbreviated version were other depression scales for five different patient actually superior (38, p241). Furukawa et al. groups. Correlations with the Hamilton scale compared the full version with 10- and 5-item ranged from 0.49 to 0.85 whereas those for the abbreviations, and with the use of item 6 (“feel Symptom Check List-90 ranged from 0.72 to depressed”) alone. The area under the ROC 0.87. Correlations with age, social class, and sex curve was 0.75 for the conventional method, were almost all nonsignificant (6, Table 5). Shi- 0.74 for a 10-item version, 0.71 for the 5-item nar et al. reported correlations of 0.57 with the abbreviation, and 0.69 for item 6 alone (36, Hamilton scale, 0.65 with the SDS, and 0.77 Table 2). with a depression categorization based on the Many different approaches have been pro- DSM-III (16, Table 2). Virtually identical results posed to scoring the CES-D, mainly in an at- were obtained by Parikh et al. (26, Table 2). tempt to enhance sensitivity and specificity and better reflect diagnostic criteria (36). The ap- Alternative Forms proaches include scoring the presence of any Shrout and Yager argued that, owing to the high symptom rated as having occurred at any time internal consistency of the CES-D, it could be during the past week (hence the 0 to 3 scale is shortened without substantial loss of reliability scored as 0,1,1,1); accordingly, scores range (19), and numerous abbreviated versions have from 0 to 20. Another option includes counting  Depression 355 only symptoms that have persisted for more than pared favorably with the Beck Depression Inven- half the previous week (coding the responses tory (48). 0,0,1,1). Others have scored the persistence of symptoms, counting only those that have lasted Reference Standards for five to seven days during the past week The CES-D was included in the U.S. Health and (hence, 0,0,0,1). An empirical comparison of Nutrition Examination Survey (HANES) popu- these alternatives suggested that the original 0 to lation survey in 1974 to 1975; norms were pub- 3 scale and the 0,0,1,1 option were equally good lished for those aged 25 to 74 by age, gender, (area under the ROC curve 0.75), whereas the race, marital status, income, and occupation (5). 0,1,1,1 was less good (AUC 0.73), and the The average score for noninstitutionalized adults 0,0,0,1 option was least good (AUC 0.70) (36, was 8.7; women had higher scores than men (5, Table 2). Schoenbach et al. proposed a scoring p2). Eaton and Kessler further analyzed the approach that reflected the Research Diagnostic HANES survey data for those who scored above Criteria (RDC) for depression that specify six the cutting-point (49, Table 2). The results criteria to be met for a diagnosis (33; 39, p800). showed strong associations with employment In place of an overall score, the CES-D responses status, educational level, income, and sex. Mur- indicate whether each criterion has been met; rell et al. reported CES-D scores from a survey in Boyd et al. indicated which CES-D questions Kentucky by gender, race, and educational and correspond to the RDC criteria (33, Table 1). In income levels (50, Table 3). Golding et al. pro- one empirical test, this scoring approach did not vide age-adjusted norms for samples of whites perform significantly better than the conven- and Mexican Americans (51, Tables 1–3). Vera tional approach (36, p74). Another approach et al. provide reference scores for two samples of sought to replicate the DSM-III-R criteria by re- Puerto Ricans by age, sex, and socioeconomic quiring a minimum score of 1 on item 6, plus a status (52, Table 2). score of 16 or above; a variant of this approach required a score of 2 or more on item 6 in addi- Commentary tion to an overall score >16 (8, p440). Neither The CES-D is one of the best-known survey in- approach proved superior to the simple use of struments for identifying symptoms of depres- >16 and, indeed, requiring a score of at least 2 sion. It has been extensively used in large studies on item 6 caused the sensitivity to fall from 0.69 and norms are available; it is applicable across to only 0.23 (8, Table 2). age and sociodemographic groups, and it has There are various translations, including been used in cross-cultural research. It has often Spanish (40), while a Mandarin version showed been used in studying the relationships between an alpha reliability of 0.92 (41); a two-factor so- depressive symptoms and other variables. lution was obtained in a Cantonese version (42). As with other self-assessed depression scales, A 10-item abbreviated Cantonese version had an the CES-D cannot be viewed as a diagnostic alpha coefficient of 0.78, and a kappa agreement tool, but only as a screening test to identify of 0.84 with the full-scale version in classifying members of groups at risk of depression. In clin- cases (43). A Korean version has been validated ical studies, the CES-D has been shown capable (44). In a study of Japanese adolescents, re- of detecting depression in a variety of psychi- sponse to the positive items in the CES-D dif- atric patients, and results agree quite well with fered markedly from that in U.S. samples, other, more detailed, instruments. It has limita- whereas responses to negatively worded items tions, however. It cannot distinguish primary de- was similar; this resulted in poor psychometric pression from that secondary to other diagnoses; properties for this Japanese version (45). this is relevant because the two types of condi- The CES-D has been used with children and tion may have differing courses so that different adolescents (9; 46; 47). A Swedish translation of treatment may be indicated (6, p213). Further- the children’s version has been validated; it com- more, it may fail to distinguish depression from 356 Measuring Health generalized anxiety and may not distinguish ear- especially among those with intercurrent med- lier from current disorders: “many CES-D items ical disorders; in such cases, the GDS may prove are not essential symptoms of major depressive more suitable (8). disorder, and serve equally well as symptoms of other syndromes or even of nonspecific demoral- References ization” (53, p77). Santor and Coyne identified as nonspecific the items concerning the percep- (1) Radloff LS. The CES-D Scale: a self-report tions of others (items 15 and 19), talkativeness depression scale for research in the general population. Appl Psychol Measurement (item 13), or comparisons with others (item 4) 1977;1:385–401. (38, p241). Finally, when applied in primary (2) Zich JM, Attkisson CC, Greenfield TK. care or general population settings, the positive Screening for depression in primary care predictive value can be under 30% (38, Table 2). clinics: the CES-D and the BDI. Int J Analyses of item bias due to physical illness led Psychiatry Med 1990;20:259–277. Grayson et al. to conclude that people with par- (3) Chan KS, Orlando M, Ghosh-Dastidar B, ticular types of physical disorder “are likely to et al. The interview mode effect on the respond to items on the CES-D in a manner de- Center for Epidemiological Studies termined by the disorder rather than solely ac- Depression (CES-D) scale: an item response cording to their underlying depressive state.” theory analysis. Med Care (28, p281). An illustration of differential item 2004;42:281–289. (4) Kohout FJ, Berkman LF, Evans DA, et al. function came from a comparison of responses Two shorter forms of the CES-D to the CES-D by black and white respondents. depression symptoms index. J Aging Blacks were more likely to endorse items such as Health 1993;5:179–193. “people are unfriendly” or “people dislike me,” (5) Sayetta R, Johnson D. Basic data on whereas women were more likely than men to depressive symptomatology, United States, endorse the item on “crying spells” (54, Table 1974–75. Washington, DC DHEW 2). (PHS)80-1666: United States Government Boyd et al. proposed that the CES-D should Printing Office, Public Health Services, be administered before other questions, that the 1980. questions should be read to the respondent, and (6) Weissman MM, Sholomskas D, Pottenger that a way should be found to deal with people M, et al. Assessing depressive symptoms in five psychiatric populations: a validation who deny symptoms (e.g., interviewing their rel- study. Am J Epidemiol 1977;106:203–214. atives) (33, p1200). Compared with the Beck (7) Himmelfarb S, Murrell SA. Reliability and BDI, it may be that the CES-D performs better validity of five mental health scales in older when there is a relatively high prevalence of de- persons. J Gerontol 1983;38:333–339. pression (11, p168). (8) Schein RL, Koenig HG. The Center for The suitability of 16 as a cutting-score has Epidemiological Studies-Depression (CES- been widely discussed. Although undue credence D) scale: assessment of depression in the should never be placed in a single cutting-point, medically ill elderly. Int J Geriatr 16 is most commonly used and seems adequate Psychiatry 1997;12:436–446. in clinical settings, although a higher point may (9) Roberts RE, Lewinsohn PM, Seeley JR. be beneficial in primary care settings to reduce Screening for adolescent depression: a comparison of scales. J Am Acad Child the number of false-positive results. Adolesc Psychiatry 1991;30:58–66. The CES-D performs comparably with other (10) Manson SM, Ackerson LM, Dick RW, et self-report scales, but concern continues about al. Depressive symptoms among American its specificity. As a first screen, it is adequate, but Indian adolescents: psychometric properties caution should be exercised in interpreting the of the Center for Epidemiologic Studies nature of the disorder it detects. In particular, Depression scale (CES-D). Psychol Assess concern remains over its use with older people, 1990;2:231–237.  Depression 357 (11) Geisser ME, Roth RS, Robinson ME. (CES-D) in a sample of women in middle Assessing depression among persons with life. Behav Res Ther 1997;35:373–380. chronic pain using the Center for (23) Callahan CM, Wolinsky FD. The effect of Epidemiological Studies-Depression scale gender and race on the measurement and the Beck Depression Inventory: a properties of the CES-D in older adults. comparative analysis. Clin J Pain Med Care 1994;32:341–356. 1997;13:163–170. (24) Liang J, van Tran T, Krause N, et al. (12) Husaini BA, Neff JA, Harrington JB. Generational differences in the structure of Depression in rural communities: the CES-D Scale in Mexican Americans. J establishing CES-D cutting points. Nashville, Gerontol 1989;44:S110–S120. Tennessee: Tennessee State University, 1979. (25) Schroevers MJ, Sanderman R, van (13) Roberts RE. Reliability of the CES-D Scale Sonderen E, et al. The evaluation of the in different ethnic contexts. Psychiatry Res Center for Epidemiologic Studies 1980;2:125–134. Depression (CES-D) scale: depressed and (14) Davidson H, Feldman PH, Crawford S. positive affect in cancer patients and Measuring depressive symptoms in the frail healthy reference subjects. Qual Life Res elderly. J Gerontol 1994;49:P159–P164. 2000;9:1015–1029. (15) Orme JG, Reis J, Herz EJ. Factorial and (26) Parikh RM, Eden DT, Price TR, et al. The discriminant validity of the Center for sensitivity and specificity of the Center for Epidemiological Studies Depression (CES- Epidemiologic Studies Depression Scale in D) scale. J Clin Psychol 1986;42:28–33. screening for post-stroke depression. Int J (16) Shinar D, Gross CR, Price TR, et al. Psychiatry Med 1988;18:169–181. Screening for depression in stroke patients: (27) Breslau N. Depressive symptoms, major the reliability and validity of the Center for depression, and generalized anxiety: a Epidemiologic Studies Depression Scale. comparison of self-reports on CES-D and Stroke 1986;17:241–245. results from diagnostic interviews. (17) Okun A, Stein RE, Bauman LJ, et al. Psychiatry Res 1985;15:219–229. Content validity of the Psychiatric (28) Grayson DA, Mackinnon A, Jorm AF, et Symptom Index, CES-Depression Scale, al. Item bias in the Center for and State-Trait Anxiety Inventory from the Epidemiologic Studies Depression scale: perspective of DSM-IV. Psychol Rep effects of physical disorders and disability 1996;79:1059–1069. in an elderly community sample. J (18) Clark VA, Aneshensel CS, Fredrichs RR, et Gerontol Psychol Sci al. Analysis of effects of sex and age in 2000;55B:P273–P282. response to items on the CES-D Scale. (29) McQuaid JR, Stein MB, McCahill M, et al. Psychiatry Res 1981;5:171–181. Use of brief psychiatric screening measures (19) Shrout PE, Yager TJ. Reliability and in a primary care sample. Dep Anx validity of screening scales: effect of 2000;12:21–29. reducing scale length. J Clin Epidemiol (30) Beekman ATF, Deeg DJH, van Limbeek J, 1989;42:69–78. et al. Criterion validity of the Center for (20) Joseph S, Lewis CA. Factor analysis of the Epidemiologic Studies Depression scale Center for Epidemiological Studies- (CES-D): results from a community-based Depression Scale. Psychol Rep sample of older subjects in the Netherlands. 1995;76:40–42. Psychol Med 1997;27:231–235. (21) Golding JM, Aneshensel CS. Factor (31) Whooley MA, Avins AL, Miranda J, et al. structure of the Center for Epidemiologic Case-finding instruments for depression: Studies Depression scale among Mexican two questions are as good as many. J Gen Americans and non-Hispanic whites. Intern Med 1997;12:439–445. Psychol Assess 2004;1:163–168. (32) Myers JK, Weissman MM. Use of a self- (22) Knight RG, Williams S, McGee R, et al. report symptom scale to detect depression Psychometric properties of the Center for in a community sample. Am J Psychiatry Epidemiologic Studies Depression scale 1980;137:1081–1084. 358 Measuring Health (33) Boyd JH, Weissman MM, Thompson WD, of the CES-D in Chinese elderly. Int J et al. Screening for depression in a Geriatr Psychiatry 1999;14:608–617. community sample: understanding the (44) Cho MJ, Kim KH. Use of the Center for discrepancies between depression symptom Epidemiologic Studies Depression (CES-D) and diagnostic scales. Arch Gen Psychiatry Scale in Korea. J Nerv Ment Dis 1982;39:1195–1200. 1998;186:304–310. (34) Lyness JM, Noel TK, Cox C, et al. (45) Iwata N, Saito K, Roberts RE. Responses Screening for depression in elderly primary to a self-administered depression scale care patients: a comparison of the Center among younger adolescents in Japan. for Epidemiologic Studies-Depression scale Psychiatry Res 1994;53:275–287. and the Geriatric Depression Scale. Arch (46) Radloff LS. The use of the Center for Intern Med 1997;157:449–454. Epidemiologic Studies Depression Scale in (35) Burnam MA, Wells KB, Leake B, et al. adolescents and young adults. J Youth Development of a brief screening Adolesc 1991;20:149–166. instrument for detecting depressive (47) Roberts RE, Andrews JA, Lewinsohn PM, disorders. Med Care 1988;26:775–789. et al. Assessment of depression in (36) Furukawa T, Anraku K, Hiroe T, et al. adolescents using the Center for Screening for depression among first-visit Epidemiologic Studies-depression scale. psychiatric patients: comparison of Psychol Assess 1990;2:122–128. different scoring methods for the Center (48) Olsson G, von Knorring AL. Depression for Epidemiologic Studies Depression Scale among Swedish adolescents measured by using receiver operating characteristic the self-rating scale Center for analyses. Psychiatry Clin Neurosci Epidemiology Studies-depression Child 1997;51:71–78. (CES-DC). Eur Child Adolesc Psychiatry (37) Andresen EM, Malmgren JA, Carter WB, 1997;6:81–87. et al. Screening for depression in well (49) Eaton WW, Kessler LG. Rates of older adults: evaluation of a short form of symptoms of depression in a national the CES-D. Am J Prev Med sample. Am J Epidemiol 1994;10:77–84. 1981;114:528–538. (38) Santor DA, Coyne JC. Shortening the CES- (50) Murrell SA, Himmelfarb S, Wright K. D to improve its ability to detect cases of Prevalence of depression and its correlates depression. Psychol Assess in older adults. Am J Epidemiol 1997;9:233–243. 1983;117:173–185. (39) Schoenbach VJ, Kaplan BH, Grimson RC, (51) Golding JM, Aneshensel CS, Hough RL. et al. Use of a symptom scale to study the Responses to Depression Scale items prevalence of a depressive syndrome in among Mexican-Americans and non- young adolescents. Am J Epidemiol Hispanic whites. J Clin Psychol 1982;116:791–800. 1991;47:61–75. (40) Gonzalez VM, Stewart A, Ritter PL, et al. (52) Vera M, Alegria M, Freeman D, et al. Translation and validation of arthritis Depressive symptoms among Puerto outcome measures into Spanish. Arthritis Ricans: island poor compared with Rheum 1995;38:1429–1446. residents of the New York City area. Am J (41) Rankin SH, Galbraith ME, Johnson S. Epidemiol 1991;134:502–510. Reliability and validity data for a Chinese (53) Rabkin JG, Klein DF. The clinical measure- translation of the Center for ment of depressive disorders. In: Marsella Epidemiological Studies-Depression. AJ, Hirschfeld RMA, Katz MM, eds. The Psychol Rep 1993;73:1291–1298. measurement of depression. New York: (42) Cheung CK, Bagley C. Validating an Guilford Press, 1987:30–83. American scale in Hong Kong: the Center (54) Cole SR, Kawachi I, Maller SJ, et al. Test for Epidemiological Studies Depression of item-response bias in the CES-D scale: Scale (CES-D). J Psychol experience from the New Haven EPESE 1998;132:169–186. Study. J Clin Epidemiol (43) Boey KW. Cross-validation of a short form 2000;53:285–289.  Depression 359 scale is administered orally, the interviewer may The Geriatric Depression Scale have to repeat the question to get a clear yes-or- (T.L. Brink and Jerome A. no response. The time frame is the past week. Yesavage, 1982) One point is counted for each answer indicating depression, as shown in the exhibit; points are Purpose added to form a total score ranging from 0 to The Geriatric Depression Scale (GDS) was de- 30, with higher values reflecting greater depres- signed as a screening test for depression in elder- sion. Scores of 0 to 10 are considered within the ly people (1). It is a self-rating scale intended normal range, 11 to 20 indicate mild depression, primarily for clinical applications and was de- and 21 to 30 indicate moderate-to-severe depres- signed to be easily administered in an office set- sion. A cutting-point of 10/11 is generally used, ting. although Kafonek, et al. used 13/14 (7) and others have used 14/15 (8), or even 16/17 (9, Conceptual Basis p240). Brink and Yesavage argued that depression Sheikh and Yesavage proposed an abbrevi- scales developed for younger people are inappro- ated version of the GDS to reduce problems of priate for the elderly (2). Symptoms indicative of fatigue, especially among physically ill or de- depression in young people (e.g., sleep distur- mented patients (10). The 15 items are identified bance, weight loss, pessimism about the future), in the exhibit and take five to seven minutes to may also occur in the elderly as normal effects of complete. Scores can be prorated to compensate aging or as the result of a physical illness. In el- for unanswered questions: (score × 15) / number derly people, depression commonly coexists of questions answered. Scores between 0 and 4 with dementia; cognitive problems compromise are considered normal, 5 to 9 indicate mild de- the accuracy of self-reports just as depression pression, and 10 to 15 indicate moderate-to- may mask cognitive abilities (1, p37). The GDS severe depression (3). Several authors have was designed as a simple, clear, and self- reported correlations between scores on the administered scale that does not rely on somatic short and long forms: 0.84 (10, p169), 0.89 symptoms. (11), and 0.66 in a sample of 81 without severely depressed patients (3). Description The GDS was developed from an initial pool of Reliability 100 items chosen by clinicians and researchers Reflecting the manner in which the GDS was for their ability to distinguish elderly depressive constructed, numerous studies show high inter- people from unaffected people. Questions that nal consistency. Yesavage et al. reported an al- might elicit defensiveness were avoided and, for pha coefficient of 0.94 in a mixed sample of simplicity, a yes/no answer format was used for unaffected and depressed elderly people all items. The 30 items in the GDS were empiri- (N = 100) (1, p43). Agrell and Dehlin found an cally selected from the pool, on the basis of high alpha of 0.90 in 40 stroke patients (12, p1192) item-total correlations when administered to a whereas Lyons et al. tested 69 hip fracture pa- mixed sample of 47 community residents and tients and reported alpha coefficients of 0.88 at people hospitalized for depression, all of whom admission and 0.93 at discharge (13, p205). were older than 55 (1, p41). Rule et al. obtained an alpha coefficient of 0.82 The GDS is shown in Exhibit 7.3. The scale is and a split-half reliability of 0.80 for 193 univer- normally self-administered in eight to ten min- sity students, higher than the figure for Zung’s utes; it can also be read in an interview (1; 3) Self-Rating Depression Scale (SDS) (14, Table 1). and a telephone interview format has been used In a further study of 389 community residents, successfully (4; 5). Interviewer administration Cronbach’s alpha ranged from 0.80 to 0.85 may take between four and 12 minutes (6). If the across five age groups of 29- to 99-year-olds 360 Measuring Health Exhibit 7.3 The Geriatric Depression Scale Note: N or Y indicates which answer is scored. The fifteen items marked with an asterisk form the abbreviated version. Choose the best answer for how you felt over the past week 1. Are you basically satisfied with your life? yes / no N* 2. Have you dropped many of your activities and interests? yes / no Y* 3. Do you feel that your life is empty? yes / no Y* 4. Do you often get bored? yes / no Y* 5. Are you hopeful about the future? yes / no N 6. Are you bothered by thoughts that you just cannot get out of your head? yes / no Y 7. Are you in good spirits most of the time? yes / no N* 8. Are you afraid that something bad is going to happen to you? yes / no Y* 9. Do you feel happy most of the time? yes / no N* 10. Do you often feel helpless? yes / no Y* 11. Do you often get restless and fidgety? yes / no Y 12. Do you prefer to stay at home, rather than go out and do new things? yes / no Y* 13. Do you frequently worry about the future? yes / no Y 14. Do you feel that you have more problems with memory than most? yes / no Y* 15. Do you think it is wonderful to be alive now? yes / no N* 16. Do you often feel downhearted and blue? yes / no Y 17. Do you feel pretty worthless the way you are now? yes / no Y* 18. Do you worry a lot about the past? yes / no Y 19. Do you find life very exciting? yes / no N 20. Is it hard for you to get started on new projects? yes / no Y 21. Do you feel full of energy? yes / no N* 22. Do you feel that your situation is hopeless? yes / no Y* 23. Do you think that most people are better off than you are? yes / no Y* 24. Do you frequently get upset over little things? yes / no Y 25. Do you frequently feel like crying? yes / no Y 26. Do you have trouble concentrating? yes / no Y 27. Do you enjoy getting up in the morning? yes / no N 28. Do you prefer to avoid social gatherings? yes / no Y 29. Is it easy for you to make decisions? yes / no N 30. Is your mind as clear as it used to be? yes / no N Reproduced from the Geriatric Depression Scale obtained from Dr. TL Brink. With permission. (14, p40). Lesher reported an alpha coefficient sample of 54 residents of long-term care facili- of 0.99, a median item-total correlation of 0.34, ties (16, Table 1). and a split-half reliability of 0.84 for 52 resi- dents of long-term care facilities (15, pp24–25). Validity The test-retest reliability of the GDS has also Several studies have compared the GDS with the been shown to be high: 0.85 at one week Hamilton Rating Scale for Depression (HRSD), (N = 20) (1, p44), 0.86 at one hour (16, Table 1), providing correlations of 0.82 (2, p42), 0.81 and and 0.98 at ten to 12 days (13, p206). Brink et 0.61 (13, p206), 0.78 (17, Table 2), and 0.58 al. reported an inter-rater reliability of 0.85 in a (18, Table 3). Yesavage et al. found the GDS per-  Depression 361 formed comparably with the Hamilton scale and Status Schedule diagnosis, the 30-item version that it was significantly better than the SDS in of the GDS showed a sensitivity of 0.83 and discriminating among 100 patients with differ- specificity of 0.80 (cutting-point 11/12); the 15- ent levels of depression, as defined by the Re- item version (cutting-point 5/6) showed a sensi- search Diagnostic Criteria. Scores on the GDS tivity and specificity of 0.71 and 0.80, increased significantly with level of depression respectively (24, pp462–3). Compared with the (r= 0.82). The equivalent correlation for the SDS Center for Epidemiologic Studies Depression was 0.69, whereas for the Hamilton scale it was scale, the GDS proved equally sensitive to major 0.83 (1, pp44–45). Koenig et al. found the GDS depression but was superior in identifying mi- correlated 0.82 with the Montgomery-Åsberg nor depression in a sample of people aged 60 or Depression Rating Scale (17, Table 2). For 67 older (25, Table 2). Results for the short and veterans, the GDS correlated 0.85 with the Beck long forms of the GDS were almost identical. Depression Inventory; the Beck instrument Agrell and Dehlin compared the GDS with showed slightly higher sensitivity and specificity five other depression scales in a sample of 40 (19, pp993–994). In assessing whether the GDS stroke patients. At a cut-off of 10/11, the GDS can be used with cognitively impaired patients, had the highest sensitivity (88%), but its speci- Feher et al. found that Mini-Mental State Exam- ficity was low (64%). The GDS correlated 0.88 ination (MMSE) scores correlated −0.15 with with the Zung SDS, 0.82 with the CES-D, and the GDS but did not explain additional variance 0.77 with the HRSD (12, pp1192–1193). in GDS scores beyond that explained by the Lesher reported a sensitivity of 100% for de- Hamilton depression scale (18, Table 2). tecting major depression and 69% for depres- In terms of sensitivity, the GDS may perform sive disorders among 51 residents of long-term slightly better than the Zung and Hamilton care facilities. Specificity was 74% (15, p25). scales; the GDS showed a sensitivity of 84% Snowdon obtained a sensitivity of 93% at a and specificity of 95% at a cutting-point of specificity of 83%, using a cutting-point of 10/11 (2, Table 2). Measured against diagnoses 10/11 (26). based on the Diagnostic Interview Schedule in a The validity of the GDS among cognitively sample of 128 hospitalized veterans, the GDS impaired patients has been reviewed extensively. had a sensitivity of 92%, a specificity of 89%, a Sheikh and Yesavage studied 43 demented pa- positive-predictive value of 53%, and a tients and argued that the GDS distinguished negative-predictive value of 99% at a cut-off of successfully between those who were depressed 10/11. Sensitivity and specificity were identical and those who were not (10, p168). By contrast, to those of the Beck instrument in a small sam- other researchers have found that the cognitive ple of elderly people (20, Table 1), whereas in status of respondents heavily influences the va- another the GDS produced a correct classifica- lidity of the GDS. Kafonek et al. found the GDS tion rate of 75% compared with 68% for the to be 75% sensitive in cognitively normal people Beck (9). A study of patients in long-care facili- but 25% sensitive in a group of cognitively im- ties gave virtually identical results: the area un- paired residents of long-term care facilities (7, der the ROC was curve 0.87 for the Beck and p32). Although Brink suggested that this might 0.85 for GDS (21, p407). The kappa agreement be an artefact of the unconventional cut-off or between the GDS and a structured psychiatric scoring method used (8), Kafonek argued that interview was 0.62 (22, Table 1). Lower validity the results would have been the same using was reported by Koenig et al., who found a sen- Brink’s cut-off. Burke et al. tested the GDS on sitivity of 82% at a specificity of 67%. Among 70 cognitively intact people and 72 others with those with a positive GDS score, the likelihood mild dementia. Using the area under the receiver of having a major depressive disorder was 27%; operating characteristic curve (AUC) as an indi- these results were comparable with those ob- cator of test performance, the GDS was sensitive tained using the Brief Carroll Rating Scale (23, in detecting depression among those cognitively Table 2). Compared with the Geriatric Mental intact (AUC = 0.85) but performed no better 362 Measuring Health than chance among those with dementia and Yesavage’s contention that somatic items are (AUC = 0.66, not significant) (27, p859). A re- not useful indicators of depression among the el- cent study of long-term care facilities identified a derly (39; 40), and so the GDS focuses on the af- MMSE score of 14/15 as the threshold below fective aspects of depression. It is easily and which the GDS performs poorly. At 15 or over, quickly administered by nonphysicians and is sensitivity was 84% and specificity was 91%; comprehensible to older people (22). It has been the results were markedly less good at lower tested in community and patient samples. Its re- MMSE scores (28). Lesher et al. found the sensi- liability is good, and sensitivity and specificity tivity of the standard and short forms of the have generally been high among cognitively in- GDS to be equal (11). tact elderly people. Indeed, as a screening test, it Weiss et al. assessed the suitability of six de- appears to perform as well as the Hamilton pression scales for people aged 75 or older, a scale, although it is briefer and more easily ad- group in which depression is often atypical. ministered. It has been suggested that the GDS None of the scales studied, including the GDS, may be easier for elderly respondents to com- measured two-week persistence of symptoms as plete (20). Another noted that depression may be required by DSM-III. The GDS contains more overidentified among elderly people who are dis- items characteristic of depression in elderly peo- engaging from their social contacts. Kathryn ple than other scales, but it nonetheless omits Adams identified a component of the GDS that several relevant themes (29). appears to measure withdrawal without sadness, Cappeliez found that scores on the GDS were rather than actual depression (41). unrelated to a social desirability response set The GDS is less valid in assessing cognitively (30). impaired patients. This is not surprising, since asking a person how he has felt in the past week Alternative Forms is a recall task (27, p859). In such cases, self- Cwikel and Ritchie translated the short version report is commonly supplemented by evidence of the GDS for use in Israel. Sensitivity was 70% from an informant. The GDS has not been well and specificity was 75% (31, p66); in comment- validated on the very old. Although it may lack ing on this indifferent performance, they sug- some items relevant to identifying depression, gested that some questions did not have the GDS performs better than most self-rating equivalent meanings in different ethnic groups. scales when applied to elderly people. We recom- A German version has been described (32), and mend its use but, as with most self-rating depres- some information is available on the validity of a sion scales, it should be followed by a Chinese translation (33). Ganguli has described psychiatric interview to confirm the classifica- a Hindi version (34), while a Japanese version tion. has been tested (35). A Canadian French version was described by Bourque et al.; it had alpha val- ues of 0.84 and 0.89 in two samples, and four- References week retest reliability of 0.70 and 0.83. A copy (1) Yesavage JA, Brink TL, Rose TL, et al. of the translation is given in the article (36). Development and validation of a geriatric depression screening scale: a preliminary Reference Standards report. J Psychiatr Res 1983;17:37–49. In a small population sample of healthy seniors (2) Brink TL, Yesavage JA, Lum O, et al. Screening tests for geriatric depression. aged between 60 to 95 years, Rankin et al. re- Clin Gerontol 1982;1:37–43. ported a mean score of 5.8 (SD 4.4) (37, Table (3) Alden D, Austin C, Sturgeon R. A 1). correlation between the Geriatric Depression Scale long and short forms. J Commentary Gerontol 1989;44:124–125. Assessing depression among the elderly is chal- (4) Burke WJ, Roccaforte WH, Wengel SP, et lenging (38). Empirical evidence supports Brink al. The reliability and validity of the  Depression 363 Geriatric Depression Rating Scale (17) Koenig HG, Meador KG, Cohen HJ, et al. administered by telephone. J Am Geriatr Depression in elderly hospitalized patients Soc 1995;43:674–679. with medical illness. Arch Intern Med (5) Morishita L, Boult C, Ebbitt B, et al. 1988;148:1929–1936. Concurrent validity of administering the (18) Feher EP, Larrabee GJ, Crook TH. Factors Geriatric Depression Scale and the physical attenuating the validity of the Geriatric functioning dimension of the SIP by Depression Scale in a dementia population. telephone. J Am Geriatr Soc J Am Geriatr Soc 1992;40:906–909. 1995;43:680–683. (19) Norris JT, Gallagher D, Wilson A, et al. (6) Adshead F, Cody DD, Pitt B. BASDEC: a Assessment of depression in geriatric novel screening instrument for depression medical outpatients: the validity of two in elderly medical inpatients. Br Med J screening measures. J Am Geriatr Soc 1992;305:397. 1987;35:989–995. (7) Kafonek S, Ettinger WH, Roca R, et al. (20) Olin JT, Schneider LS, Eaton EM, et al. Instruments for screening for depression The Geriatric Depression Scale and the and dementia in a long-term care facility. J Beck Depression Inventory as screening Am Geriatr Soc 1989;37:29–34. instruments in an older adult outpatient (8) Brink TL. Proper scoring of the Geriatric population. Psychol Assess Depression Scale. J Am Geriatr Soc 1992;4:190–192. 1989;37:819. (21) Laprise R, Vézina J. Diagnostic perfor- (9) Kogan ES, Kabacoff RI, Hersen M, et al. mance of the Geriatric Depression Scale Clinical cutoffs for the Beck Depression and the Beck Depression Inventory with Inventory and the Geriatric Depression nursing home residents. Can J Aging Scale with older adult psychiatric 1998;17:401–413. outpatients. J Psychopathol Behav Assess (22) Koenig HG, Meador KG, Cohen HJ, et al. 1994;16:233–242. Self-rated depression scales and screening (10) Sheikh JI, Yesavage JA. Geriatric for major depression in the older Depression Scale (GDS): recent evidence hospitalized patient with medical illness. J and development of a shorter version. Clin Am Geriatr Soc 1988;36:699–706. Gerontol 1986;5:165–172. (23) Koenig HG, Meador KG, Cohen HJ, et al. (11) Lesher EL, Berryhill JS. Validation of the Screening for depression in hospitalized el- Geriatric Depression Scale—Short Form derly medical patients: taking a closer look. among inpatients. J Clin Psychol J Am Geriatr Soc 1992;40:1013–1017. 1994;50:256–260. (24) Neal RM, Baldwin RC. Screening for (12) Agrell B, Dehlin O. Comparison of six anxiety and depression in elderly medical depression rating scales in geriatric stroke outpatients. Age Ageing 1994;23:461–464. patients. Stroke 1989;20:1190–1194. (25) Lyness JM, Noel TK, Cox C, et al. (13) Lyons JS, Strain JJ, Hammer JS, et al. Screening for depression in elderly primary Reliability, validity, and temporal stability care patients: a comparison of the Center of the Geriatric Depression Scale in for Epidemiologic Studies-Depression scale hospitalized elderly. Int J Psychiatry Med and the Geriatric Depression Scale. Arch 1989;19:203–209. Intern Med 1997;157:449–454. (14) Rule BG, Harvey HZ, Dobbs AR. (26) Snowdon J. Validity of the Geriatric Reliability of the Geriatric Depression Depression Scale. J Am Geriatr Soc Scale for younger adults. Clin Gerontol 1990;38:722–723. 1989;9:37–43. (27) Burke WJ, Houston MJ, Boust SJ, et al. (15) Lesher EL. Validation of the Geriatric Use of the Geriatric Depression Scale in Depression Scale among nursing home dementia of the Alzheimer type. J Am residents. Clin Gerontol 1986;4:21–28. Geriatr Soc 1989;37:856–860. (16) Brink TL, Curran P, Dorr ML, et al. (28) McGivney SA, Mulvihill M, Taylor B. Geriatric Depression Scale reliability: order, Validating the GDS depression screen in examiner and reminiscence effects. Clin the nursing home. J Am Geriatr Soc Gerontol 1983;2:57–60. 1994;42:490–492. 364 Measuring Health (29) Weiss IK, Nagel CL, Aronson MK. Applicability of depression scales to the old The Depression Adjective Check Lists old person. J Am Geriatr Soc (Bernard Lubin, 1965) 1986;34:215–218. (30) Cappeliez P. Social desirability response set Purpose and self-report depression inventories in The Depression Adjective Check Lists (DACL) the elderly. Clin Gerontol 1989;9:45–52. comprise seven scales that were developed as (31) Cwikel J, Ritchie K. The Short GDS: brief research instruments for measuring de- evaluation in a heterogeneous, multilingual pressed mood (1). They are applicable to all ages population. Clin Gerontol 1988;8:63–71. and diagnostic groups. The seven forms of the (32) Bach M, Nikolaus T, Oster P, et al. DACL are designed to be equivalent and are in- Diagnosis of depression in the elderly. The tended for use where repeated assessments are to “Geriatric Depression Scale.” Z Gerontol be made (2). und Geriatr 1995;28:42–46. (33) Mui AC. Geriatric Depression Scale as a community screening instrument for elderly Conceptual Basis Chinese immigrants. Int Psychogeriatr Developed during the early 1960s in the context 1996;8:445–458. of a shift away from behaviorism, the DACL fo- (34) Ganguli M, Dube S, Johnston JM, et al. cused on the affective dimensions of depression. Depressive symptoms, cognitive Lubin bypassed conceptual debates over the impairment and functional impairment in a components of depression and concentrated on rural elderly population in India: a Hindi lowered mood but did not define exactly which version of the Geriatric Depression Scale aspects were to be covered. This may place the (GDS-H). Int J Geriatr Psychiatry 1999;14:807–820. DACL within the European phenomenological (35) Schreiner AS, Morimoto T, Asano H. tradition rather than the North American tradi- Depressive symptoms among poststroke tion. The adjective check list format was chosen patients in Japan: frequency distribution as being suited to assessing mood, easy to ad- and factor structure of the GDS. Int J minister, and acceptable to subjects. Geriatr Psychiatry 2001;16:941–949. (36) Bourque P, Blanchard L, Vézina J. Étude Description psychométrique de l’échelle de dépression The DACL form one of a series of adjective gériatrique. Can J Aging 1990;9:348–355. check list scales. In 1960, Zuckerman published (37) Rankin EJ, Gfeller JD, Gilner FH. Measur- ing anxiety states in the elderly using the the Anxiety Adjective Check List; Lubin began State-trait Anxiety Inventory for Children. developing the Depression Adjective Check Lists J Psychiatr Res 1993;27:111–117. in 1963 and then collaborated with Zuckerman (38) Thompson LW, Futterman A, Gallagher D. to incorporate the most discriminating items Assessment of late-life depression. from the DACL into an expanded version of the Psychopharmacol Bull 1988;24:577–586. anxiety check list, named the Multiple Affect (39) Zemore R, Eames N. Psychic and somatic Adjective Check List (MAACL), which was pub- symptoms of depression among young lished in 1964 (3). As well as anxiety and de- adults, institutionalized aged and pression, this scale covers hostility, positive noninstitutionalized aged. J Gerontol affect, and sensation-seeking. Subsequently, 1979;34:716–722. Zuckerman et al. collaborated to publish a re- (40) Bolla-Wilson K, Bleecker ML. Absence of vised version of the MAACL in 1983 (4). depression in elderly adults. J Gerontol 1989;44:53–55. The DACL were derived from item analyses (41) Adams KB. Depressive symptoms, of 171 adjectives that connoted varying degrees depletion, or developmental change? of depression and elation (1, p5). The item Withdrawal, apathy, and lack of vigor in analyses selected adjectives that distinguished the Geriatric Depression Scale. depressed patients from unaffected people. The Gerontologist 2001;41:768–777. instrument was originally developed for a study  Depression 365 of postpartum depression, so the initial analyses Exhibit 7.4 Depression Adjective Check Lists, used women subjects and identified 128 adjec- Form E tives. The adjectives were grouped into lists A, B, DIRECTIONS: Below you will find words which C, and D, each containing 22 positive (i.e., in- describe different kinds of moods and feelings. Check dicative of depression) and ten negative adjec- the words which describe HOW YOU GENERALLY tives. These lists are known as set 1. Subsequently, FEEL. Some of the words may sound alike, but we the initial 171 adjectives were tested on samples want you to check all the words that describe your of 100 unaffected and 47 depressed males. The feelings. Work rapidly and check all of the words which describe how you generally feel. 102 most discriminating adjectives were as- signed to lists E, F, and G. These lists, known as 1. Unhappy 18. Well set 2, each contain 22 positive and 12 negative 2. Active 19. Apathetic adjectives. Lists A to D have no items in com- mon, nor have lists E to G, but some adjectives 3. Blue 20. Chained are common to both sets. The instrument is 4. Downcast 21. Strong available in state and trait forms. The same ad- 5. Dispirited 22. Dejected jective lists are used in each, but the instruction 6. Composed 23. Awful for the state form reads “How do you feel now today?” whereas that for the trait form reads 7. Distressed 24. Glum “How do you generally feel?” This is intended to 8. Cheerless 25. Great cover “longer-enduring, more pervasive depres- 9. Lonely 26. Finished sive mood” (2). A manual includes complete in- 10. Free 27. Hopeless structions and shows normative data (1). Exhibit 7.4 shows list E as an illustration of the DACL. 11. Lost 28. Lucky The full scale is shown by Robinson et al. (5, 12. Broken 29. Tortured pp215–219). 13. Good 30. Listless The DACL are self-administered in two to 14. Burdened 31. Safe three minutes but may be read to the subject if needed. Unaffected subjects take about two-and- 15. Forlorn 32. Wilted a-half minutes to complete a list; psychiatric pa- 16. Vigorous 33. Criticized tients take longer. Scores for each list count the 17. Peaceful 34. Fit positive adjectives that are checked, plus the number of negative adjectives that are not Reproduced from an original provided by Dr. Lubin. With permission. checked: high scores indicate depressed mood. The maximum score is 32 for lists A to D and 34 for lists E to G. Scoring is simplified by a trans- parent overlay that indicates which responses are counted for each list. A cutting-point of 13 or Internal consistency estimates for individual above (one standard deviation above the popula- lists, based on analyses of variance, have ranged tion mean score obtained by Lubin) has been from 0.79 to 0.90 (1, Table 12); Byerly and used to indicate mild depression, and 16 or Carlson reported Kuder-Richardson 20 reliabili- above (+2 standard deviations) to indicate severe ties for different patient groups ranging from depression (6). 0.59 to 0.72 (7, p801). Beckingham and Lubin reported mean coefficient alpha scores for the Reliability trait form lists of 0.84 with elderly subjects. Lubin reviewed the equivalence of the seven lists. Split-half reliability averaged 0.76 (8, p407). Al- Correlations among them range from 0.80 to phas of 0.85 and 0.87 (9, p419) and 0.91 (10, 0.93 (1, Table 15). Several studies have found no Table 2) have been reported for list E. Split-half significant differences in total scores among the reliability estimates for the state form range lists in set 1 or in set 2 (1, p11). from 0.82 to 0.93 for unaffected subjects and 366 Measuring Health from 0.86 to 0.93 for patients (1, p10). The pos- 0.73 (7, p801); other results include a correla- itive and negative adjectives in the lists correlate tion of 0.60 between list B and the BDI (10, between −0.75 and −0.82 (1, Table 14). Table 2) and of 0.69 in a mixed sample of unaf- Test-retest reliability of the state form lists fected subjects and patients. The correlation ranged from 0.19 to 0.24 after a one-week inter- with the Hamilton Rating Scale for Depression val in a sample of 75 subjects (1, p10). Lomranz was 0.72 in the same study (15, Table 1). Corre- et al. reported retest reliability results of 0.45 for lations with Zung’s Self-Rating Depression Scale the state form after one week, and 0.59 after are lower, perhaps because it covers more physi- three weeks, in two studies using a Hebrew ver- ological symptoms than other instruments. Cor- sion of the DACL (11, p1313). relations ranged from 0.27 to 0.38 in a Lubin et al. examined the relationship be- psychiatric sample (1, Table 22), and from 0.44 tween comprehension difficulty of the adjectives (14, Table 1) to 0.54 in normal samples (10, and reliability of the trait form of the DACL. Re- Table 2). Other results of correlating the DACL sults were more limited for subjects with lower and Zung scales include 0.62 (11, p1313) and a reading ability, but overall reliability was re- range between 0.53 and 0.64 (1, Table 25). In duced when adjectives above the eighth grade two studies of psychiatric inpatients, correla- level were excluded, as the benefit of eliminating tions between lists E, F, and G and the Min- the difficult words was offset by losses due to nesota Multiphasic Personality Inventory abbreviating the scale (12, Table 4). (MMPI) depression scale ranged from 0.54 to 0.60 (1, Table 26). List E of the DACL corre- Validity lated 0.77 with the CES-D in a sample of 48 un- Moran and Lambert rated the DACL fifth out of affected subjects (1, Table 29). Correlations with six depression scales in terms of its coverage of the Bradburn scale included −0.37 for unaf- the DSM-III criteria because the lists tap only fected patients and −0.47 for psychiatric patients three variables directly. They argued that be- (1, Table 28). cause the instrument focuses on the affective as- Correlations between the DACL and global pects of depression, it needs major additions to ratings of depression lie mainly between 0.3 and conform to DSM-III criteria (13, p295). 0.5 (1, Table 22; 16). An exception was a study Analyses of variance showed significant dif- by Fogel, in which the DACL correlated 0.71 ferences between the DACL responses of 625 with a psychiatric rating (1, p13). unaffected subjects, 174 nondepressed patients, Lubin summarized evidence from several stud- and 128 depressed patients (1, Table 20). Byerly ies concerning sensitivity to change. For example, and Carlson found that depressed patients re- among users of a mental health center, there was ceived significantly higher scores on list E of the a significant drop in DACL scores over a three- DACL than nondepressed patients (7). Christen- month treatment period but no change after the feld et al. also found that mean scores on list E end of treatment (1, p19). Other studies have increased as a psychiatrist’s rating of depressed found the DACL sensitive to change in interven- mood increased (14). In Byerly’s discriminant tion studies (17; 18). analysis, the DACL correctly classified 75% of A factor analysis of list E placed the positive depressed, and of nondepressed patients (1, and negative adjectives on separate factors (1, Table 38). In a series of such analyses, the DACL Table 32). This has been replicated for other performed slightly better than the Beck Depres- scales in both state and trait versions (19). How- sion Inventory (BDI) (1, p18). ever, when factoring the DACL items along with Correlations with other instruments are mod- items from the Zung and Beck instruments, Gi- erate to high. For example, correlations with the ambra found that the DACL adjectives all loaded BDI ranged from 0.38 to 0.50 in psychiatric pa- on one factor, which he labeled affective malaise tients (1, Table 22), and 0.41 in unaffected sub- (20). The discussion over whether the positive jects (14, Table 1). Byerly found correlations and negative items fall into one or more than between list E and the BDI ranging from 0.42 to one dimension (21, p391) recalls the discussions  Depression 367 of the dimensions of affect that are reviewed in native response modes and readability (which Chapter 5. lies between the fourth- and eighth-grade levels) (1; 12; 25; 26). Although separate lists were de- veloped for the two sexes, this is commonly ig- Alternative Forms nored. There is some evidence that the lists are Lubin created 14 briefer check lists by treating not equally valid for both genders. Lists A each column of adjectives in lists A through G as through D appear more suitable for females, and a separate instrument. He reported that internal lists E through G seem to better suit males (27). consistency, alternate-form reliability, and con- The lists have adequate internal consistency; current and discriminant validity of the brief the close agreement between the alternative lists are high, although somewhat lower than for forms is a strong feature (13, p273). Evidence the full lists (1, pp23–27). for the stability of the measurement, however, A redesigned version called the State-Trait is sparse and the available retest results are low. Depression Adjective Check Lists has been pro- At present, they are limited to the state form; duced; this is available from www.edits.net/ST- data on the reliability of the trait forms are DACL.html. needed (9). The check lists have been translated into Commentaries on the validity of the DACL many languages, including Hebrew, French, give conflicting impressions. This seems to re- Spanish, Arabic, Chinese (22), Dutch, and Por- flect reviewers’ differing expectations of the tuguese; most studies find their reliability com- method and their different reactions to its com- parable with that of the English version (1, p22; paratively narrow scope. Lubin intentionally fo- 11; 22–24). A version for adolescents has been cused on mood, a major, but not the sole, developed (2). symptom of depression. The DACL do not in- clude DSM criteria such as the inability to think Reference Standards or concentrate, thoughts of suicide, changes in The DACL manual presents norms based on appetite, sleep, or psychomotor activity, and U.S. data grouped by gender and age (1, Tables other behavioral manifestations (13, p286; 28, 2–10). Means and standard deviations are also p421). Reflecting this focus, concurrent validity presented for depressed and nondepressed pa- is moderate: correlations with psychiatric ratings tient samples. A 1993 revision of the manual of depression are somewhat lower than those of presents norms and percentile scores for commu- other self-rating instruments. Correlations be- nity subjects, adolescents, and the elderly, and is tween the DACL and other self-report depres- available through Edits Corporation (www.ed sion measures are low to moderate (21); the low its.net/pdfs/ST-DACL.pdf). correlations between the DACL and the somatic items in the Zung and Beck instruments are to be expected (1, Tables 23–24). Although the Commentary check lists can be used to screen for depression, The DACL are widely used: a bibliography com- they must not be viewed as diagnostic. Lubin piled by Lubin listed more than 500 citations by has, however, developed a brief Depression Ex- 1989, and publications have continued. The tender Check List (DECL), which samples addi- DACL are acceptable to respondents and are tional aspects of depression to form a diagnostic simple to administer. Other advantages include instrument (B. Lubin, personal communication, the existence of normative data and the alterna- 1992). tive versions, which are useful when repeated as- The validation of the DACL took a psycho- sessments have to be made, as in longitudinal metric rather than an epidemiological approach. studies (21). Furthermore, Lubin has extensively We know more about its correlations with other studied the psychometric properties of the instruments than about its sensitivity in identify- scales, examining issues rarely considered in ing depression. This may be weak: data from By- other health indexes, such as response set, alter- erly and Carlson’s study show that the DACL 368 Measuring Health correctly identified only 75% of depressive sub- (7) Byerly FC, Carlson WA. Comparison jects. The manual of the DACL pays little atten- among inpatients, outpatients, and normals tion to cutting-points for defining depression, on three self-report depression inventories. and the discriminant validity of the DACL has J Clin Psychol 1982;38:797–804. not been well-investigated (28). The issue of (8) Beckingham AC, Lubin B. Reliability and validity of the trait form of set 2 of the specificity has also been raised: the original item Depression Adjective Check Lists with analysis selected adjectives that distinguished de- Canadian elderly. J Clin Psychol pressed patients from unaffected patients. This 1991;47:407–414. does not ensure specificity and the DACL may (9) DeSouza ER, Lubin B, van Whitlock R. reflect a range of conditions broader than de- Preliminary report on reliability and pression (29). This may limit their usefulness as validity of the trait form of the Depression a screening test or as an outcome measurement. Adjective Check List in a representative Because of the large standard deviations in community sample. J Clin Psychol DACL scores and the resulting overlaps between 1991;47:418–420. group scores, several commentators suggest that (10) Tanaka-Matsumi J, Kameoka VA. the DACL may be adequate to detect group dif- Reliabilities and concurrent validities of ferences in depression, but not to identify indi- popular self-report measures of depression, anxiety, and social desirability. J Consult vidual depressed patients (21, p389; 30; 31, Clin Psychol 1986;54:328–333. p217). McNair recommended that the DACL be (11) Lomranz J, Lubin B, Eyal N, et al. Measur- restricted to research uses (29) and Moran and ing depressive mood in elderly Israeli: Lambert’s review concluded that there were not reliability and validity of the Depression adequate data to support their use as an outcome Adjective Check Lists. Psychol Rep measure with patient populations (13, p295). 1991;68:1311–1316. (12) Lubin B, Collins JF, Seever M, et al. Relationships among readability, reliability, and validity in a self-report adjective check References list. Psychol Assess 1990;2:256–261. (1) Lubin B. Manual for the Depression (13) Moran PW, Lambert MJ. A review of Adjective Check Lists. Odessa, Florida: current assessment tools for monitoring Psychological Asessment Resources, Inc., changes in depression. In: Lambert MJ, 1981. Christensen ER, DeJulio SS, eds. The mea- (2) Lubin B. Citation Classic. Measuring surement of psychotherapy outcome in depressive affect with an adjective research and evaluation. New York: Wiley, checklist. Curr Cont 1991;6:16. 1983:263–303. (3) Zuckerman M, Lubin B, Vogel L, et al. (14) Christenfeld R, Lubin B, Satin M. Measurement of experimentally induced Concurrent validity of the Depression affects. J Consult Psychol Adjective Check List in a normal 1964;28:418–425. population. Am J Psychiatry (4) Zuckerman M, Lubin B, Rinck CM. 1978;135:582–584. Construction of new scales for the Multiple (15) Fitzgibbon ML, Cella DF, Sweeney JA. Affect Adjective Check List. J Behav Assess Redundancy in measures of depression. J 1983;5:119–120. Clin Psychol 1988;44:372–374. (5) Robinson JP, Shaver PR, Wrightsman LS. (16) Cresswell DL, Lanyon RI. Validation of a Measures of personality and social screening battery for psychogeriatric psychological attitudes. San Diego: assessment. J Gerontol 1981;36:435–440. Academic Press, 1991. (17) Cutler NR, Cohen HB. The effect of one (6) Davis TC, Nathan RG, Crouch MA, et al. night’s sleep loss on mood and memory in Screening depression in primary care: back normal subjects. Compr Psychiatry to the basics with a new tool. Fam Med 1979;20:61–66. 1987;19:200–202. (18) Davis GR, Armstrong HE, Donovan DM,  Depression 369 et al. Cognitive-behavioral treatment of Highland Park, New Jersey: Gryphon depressed affect among epileptics: Press, 1974:133–134. preliminary findings. J Clin Psychol (30) Lewinsohn PM, Hoberman HM. 1984;40:930–935. Depression. In: Bellack AS, Hersen M, (19) van Whitlock R, Lubin B, Noble E. Factor Kazdin AE, editors. International structure of the state and trait versions of handbook of behavior modification and the depression adjective check lists. J Clin therapy. New York: Plenum, 1990. Psychol 1995;51:614–625. (31) Petzel TP. Depression Adjective Check (20) Giambra LM. Independent dimensions of Lists. In: Keyser DJ, Sweetland RC, editors. depression: a factor analysis of three self- Test critiques. Kansas City, Missouri: Test report depression measures. J Clin Psychol Corporation of America, 1985:215–220. 1977;33:928–935. (21) Mayer JM. Assessment of depression. In: McReynolds P, ed. Advances in psychological assessment. Vol. 4. San The Hamilton Rating Scale for Francisco: Jossey-Bass, 1977:358–425. Depression (22) Chu CRL, Lubin B, Sue S. Reliability and (Max Hamilton, 1960) validity of the Chinese Depression Adjective Check Lists. J Clin Psychol Purpose 1984;40:1409–1413. The Hamilton Rating Scale for Depression (23) Lubin B, Collins JF. Depression Adjective (HRSD) is completed by a clinician to indicate Check Lists: Spanish, Hebrew, Chinese and the severity of depression in patients already di- English versions. J Clin Psychol agnosed with a depressive disorder; it is not a di- 1985;41:213–217. agnostic instrument (1; 2). It covers depressive (24) DeSouza ER, Lubin B, Zanelli J. Norms, state rather than trait and is intended primarily reliability, and concurrent validity mea- as a research tool to be used in “quantifying the sures of the Portuguese version of the results of an interview” (1, p56). Depression Adjective Check Lists. J Clin Psychol 1994;50:208–215. (25) Lubin B, Collins JF, Seever M, et al. Conceptual Basis Readability of the Depression Adjective Check Lists (DACL) and the Multiple Hamilton’s involvement in early clinical trials of Affect Adjective Check List-Revised then newly developed antidepressants led to the (MAACL-R). J Clin Psychol development of the HRSD (3). A reliable scale 1991;47:91–93. was needed to measure the severity of depression (26) Caplan M, Lubin B, Collins JF. Response in patients with mental illness; one that could manipulation on the Depression Adjective also be used with semiliterate or severely ill pa- Check List. J Clin Psychol tients and could give information about symp- 1982;38:156–159. toms, and was clearly related to diagnosis (4). (27) Goodstein LD. Review of “Depression The symptoms included were parsimonious Adjective Check Lists.” In: Burros O, ed. rather than exhaustive, and were selected on the Seventh mental measurements yearbook. basis of the medical literature and clinical expe- Highland Park, New Jersey: Gryphon rience of the symptoms most frequently pre- Press, 1974:132–133. (28) Carson TP. Assessment of depression. In: sented by patients (5). Ciminero AR, Calhoun KS, Adams HE, editors. Handbook of behavioural assessment. New York: Wiley, Description 1986:404–445. The HRSD contains 21 ratings measured on (29) McNair DM. Review of “Depression three (0 to 2) or five (0 to 4) point scales (Ex- Adjective Check Lists.” In: Burros O, ed. hibit 7.5). The first 17 items are used in scoring Seventh mental measurements yearbook. the instrument, whereas the final four items Exhibit 7.5 The Hamilton Rating Scale for Depression Range Range Item of Item of No. Scores Symptom No. Scores Symptom 1 0–4 Depressed Mood Obvious restlessness picking Gloomy attitude, pessimism about at hands & clothes . . . . . . . 2 the future Patient has to get up . . . . . . . 3 Feeling of sadness Patient paces, picks at face Tendency to weep and hair—tears at clothes . . 4 Sadness, etc. . . . . . . . . . . . . . . . 1 10 0–4 Anxiety, psychic Occasional weeping . . . . . . . . . 2 Tension and irritability Frequent weeping . . . . . . . . . . . 3 Worrying about minor matters Extreme symptoms . . . . . . . . . . 4 Apprehensive attitude 2 0–4 Guilt Fears Self-reproach, feels he has let 11 0–4 Anxiety, somatic people down Gastrointestinal, wind, indigestion Ideas of guilt Cardiovascular, palpitations, Present illness is a punishment headache Delusions of guilt Respiratory, genito-urinary, etc. Hallucinations of guilt 12 0–2 Somatic Symptoms, Gastrointestinal 3 0–4 Suicide Loss of appetite Feels life is not worth living Heavy feelings in abdomen Wishes he were dead Constipation Suicidal ideas & half-hearted ideas 13 0–2 Somatic Symptoms, General Attempts at suicide Heaviness in limbs, back, or head 4 0–2 Insomnia, initial Diffuse backache Difficulty in falling asleep Loss of energy and fatiguability 5 0–2 Insomnia, middle 14 0–2 Loss of libido Patient restless and disturbed Pathological change during the night 15 0–4 Hypochondriasis Waking during the night Self-absorption (bodily) 6 0–2 Insomnia, delayed Preoccupation with health Waking in early hours of the morn- Querulous attitude ing and unable to fall asleep again Strong conviction of organic 7 0–4 Work and Interests disease Feelings of incapacity Hypochondriacal delusions Listlessness, indecisions and 16 0–2 Loss of Weight vacillation 17 2–0 Insight Loss of interest in hobbies Loss of insight. . . . . . . . . . . . . . 2 Decreased social activities Partial or doubtful loss . . . . . . . 1 Productivity decreased No loss . . . . . . . . . . . . . . . . . . . 0 Unable to work (Insight must be interpreted in Stopped working because of terms of patient’s understand- present illness only . . . . . . . . 4 ing and background.) (Absence from work after treatment or recovery may 18 0–2 Diurnal Variation rate a lower score.) Symptoms worse in morning or evening (Note which it is.) 8 0–4 Retardation Slowness of thought, speech, 19 0–4 Depersonalization and Derealization and activity Feelings of unreality Apathy Nihilistic ideas Stupor 20 0–4 Paranoid Symptoms Slight retardation at interview. . 1 Suspicious Obvious retardation at inter- view . . . . . . . . . . . . . . . . . . . . . 2 Interview difficult . . . . . . . . . . . 3 Ideas of reference } Not with a Delusions of reference depressive and persecution quality Complete stupor . . . . . . . . . . . . 4 Hallucinations, persecutory 9 0–4 Agitation 21 0–2 Obsessional Symptoms Restlessness associated with anxiety Obsessive thoughts and compulsions Figitiness at interview . . . . . . . . 1 against which the patient struggles Reproduced from Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960;23:62. Reproduced with per- mission from the BMJ Publishing Group. 370  Depression 371 provide more detail on the clinical characteris- a threshold for including patients in drug trials tics of the depression; Hamilton did not include (8). these in the score, although others have done Unfortunately, there are many variants of the so. Hamilton scale, several of which are described The HRSD is administered through a semi- in the Alternative Forms section. Of these, the structured clinical interview; Hamilton did not 11-item abbreviation called the Bech-Rafaelsen specify precise questions because he preferred Melancholia Scale (BRMS) is mentioned here not to constrain the freedom of the clinician. because it is used as a comparison in the reliabil- Subsequently, he published a four-page guide to ity and validity studies described in the follow- making ratings (2, pp291–295); these were ing sections. reprinted in the Early Clinical Drug Evaluation (ECDEU) manual (6, pp186–191, and ten years Reliability later Hamilton gave some further guidelines (4, Alpha internal consistency was reported from a pp148–151). Alternatively, Bech et al. present World Health Organization study in five coun- scoring instructions that were developed in col- tries and showed figures of alpha = 0.48 at base- laboration with Hamilton (7, pp23–28). An ade- line, rising to 0.85 after 11 days of treatment (9, quate interview requires at least 30 minutes; Table 2). Figures from a Spanish study ranged interviewers should have good clinical experi- from alpha = 0.70 to 0.76, according to diagnos- ence. Hamilton recommended that two raters be tic group (10, Table 5). Other estimates include used; one interviews the patient and the other 0.76 (11, p35), 0.83, 0.94, and 0.95 (12, Table adds questions where appropriate (2, p291). The 1), and 0.84 (13, Table 3). Carroll et al. obtained two then complete rating forms independently a median item–total correlation of 0.54 (range and later compare scores and discuss differences 0.19 to 0.78) (14, p198). (4, p147). To obtain the most complete informa- Maier et al. applied Rasch analyses to the tion possible, near relatives or significant others HRSD and other depression ratings to evaluate should also be interviewed. The ratings cover de- the internal consistency and consistency of inter- pressive symptoms during the past few days or a relations among items in different samples. The week. Intensity and frequency of symptoms are results showed clear deviations from the as- considered; ratings are based on a synthesis of sumptions of the model (15, pp7–8; 16; 17). both. Themes that are difficult to quantify are Bech et al. confirmed these findings, showing rated grossly on a 0 to 2 scale: 0 = symptom ab- that the hierarchical ranking of items on the 17- sent, 1 = slight or doubtful, and 2 = clearly pres- item version varied from sample to sample (18, ent. Other items are graded more finely on a 0 to p208). Maier commented: 4 scale in terms of increasing intensity: 0 = symp- tom absent, 1 = doubtful or trivial, 2 = mild, The consequence of the insufficient Rasch 3 = moderate, and 4 = severe. Half-points may be model fitting is that the HAM-D scores can- used. not be compared across different samples or A total score sums the item responses and within a sample across different conditions ranges from 0 to 52 points with rising severity of (e.g., pre- and post-treatment). (16, p70) depression. Hamilton did not specify cutting- points, but it is generally agreed that scores These concerns led Bech et al. to propose a six- lower than 7 indicate an absence of depression, item abbreviation of the HRSD that did fit a scores of 7 to 17 represent mild depression, 18 Rasch model (described in the Alternative Forms to 24 moderate, and 25 or above represent severe section). depression. Alternative cutting-points were sug- Inter-rater reliability for the HRSD total gested by Bech et al: 0 to 7 = no depression, 8 to score is generally high. Hamilton originally re- 15 = minor, and 16 or more = major depression ported a correlation of 0.90 between pairs (7, p10). Scores of 17 or more are often used as of ratings for 70 patients (1). Montgomery and 372 Measuring Health Åsberg found an inter-rater correlation of 0.89 cluded other symptoms (1, pp58–59; 2, before treatment (19). Hedlund and Vieweg re- pp288–289; 28). Many subsequent authors have viewed ten reliability studies: in only one small reported factor analyses of the HRSD and re- study did inter-rater correlation fall below view articles have been produced (10; 12; 29). 0.84; most coefficients fell above 0.88 (12, One meta-analysis of 12 studies showed that the Table 1). Even higher values include 0.96 and number of factors extracted ranges from three to 0.98 in two small samples of long-term care pa- seven, with a modal value of six (10, Table 6). tients (20, Table 1). Intraclass coefficients in- Hedlund and Vieweg described two consistent clude 0.70 to 0.72 (15, Table 2), 0.83 (21, factors: a general severity of depression factor Table 3), 0.85 (22, p1137) and 0.91 (23, Table and a bipolar factor that distinguishes reactive 3). Reliability results for individual items are from endogenous depression (12, p157). Several much less adequate. Maier et al. found that analyses identify factors in the HRSD that corre- three items on the Hamilton scale had unac- spond to anxiety (28; 30). ceptably low inter-rater reliability (intraclass As a measure of severity, the Hamilton scale coefficients of 0.19 to 0.40), and that five other has frequently been tested against clinical ratings items had poor, or only fair reliability (15, p6). of severity. The results are generally good, al- Rehm and O’Hara reported inter-rater reliabili- though variable. The total score of the 17-item ties for individual items (across different raters version was significantly related to globally as- and different studies) ranging from -0.07 to sessed severity of depression on the Raskin 0.95, while overall reliabilities ranged from three-item rating scale for depression (r= 0.65); 0.78 to 0.96 (11, Table 1). correlations for the Montgomery-Åsberg De- Intraclass test-retest reliability at 3 weeks was pression Rating Scale (MADRS) (0.71) and 0.72 for the 17-item version, 0.70 for the 21- BRMS (0.70) were slightly higher (26, Tables item version, and 0.69 for an abbreviated five- 1–3). Four items in the Hamilton scale were not item version (16, Table 1). significantly related to the global rating. Other figures include a correlation of 0.81 between the Validity HRDS and the Raskin scale, and -0.86 between The content validity of the HRSD has been criti- the HRDS and overall severity rated by the cized on the apparently incompatible grounds of Global Assessment Scale (GAS); correlations be- its being restricted and of its covering more than tween individual items and the global ratings just the severity of depression. Claims of restric- ranged from 0.10 to 0.86 (11, Table 3). Hedlund tiveness hold that it neglects cognitive and affec- and Vieweg’s review cited correlations of 0.84, tive symptoms in favor of somatic symptoms 0.89, and 0.90 with clinical ratings of severity; that are indicative of more severe depression they also listed a lower correlation of 0.67 for a (24). It includes only four of the eight melancho- study that included depressed and nondepressed lia symptoms in DSM-III (25, p132) and omits patients (12, p152). Other low correlations with nonreactivity of mood, reduction of concentra- a clinical severity rating include 0.68 (16, Table tion, and anhedonia (26). Another criticism 3). The HRSD may be limited in its ability to holds that its content is mixed: some items cover distinguish moderate and severe depression (16; severity, whereas others classify depression 31), and the results given here indicate that not rather than measure its severity (27, p130). This all the items correlate with global assessments. last criticism has stimulated the development of Many studies have compared the HRSD with abbreviated forms of the HRSD (see later discus- self-report scales; a consistent finding is that sion). agreement improves following treatment. The Hamilton had a strong personal interest in pretreatment correlations, for example, between factor analysis and identified four main factors: a the Hamilton and the Beck Depression Inventory general factor, a second that contrasted agitated (BDI) in seven studies ranged from only 0.21 to and retarded depression, a third that covered in- 0.82, with a median of 0.58; posttreatment cor- somnia and fatigability, and a fourth that in- relations ranged from 0.68 to 0.80 (32, p152).  Depression 373 In nine comparable studies, correlations between Eight of the HRSD items were not sensitive to the HRSD and Zung’s Self-rating Depression change. Montgomery and Åsberg found correla- Scale (SDS) ranged from 0.22 to 0.95 (12, Table tions between their MADRS and a clinical judg- 2). The wide range of results depends partly on ment of improvement (0.70) to be superior to the differing types of patients included, but the that of the Hamilton scale (0.59) (19, p385). reasons are not fully clear. Correlations with a range of other depression scales covered in Hed- Alternative Forms lund and Vieweg’s review ranged from 0.63 to So many variants on the HRSD exist that it is 0.87; correlations with the Minnesota Multipha- tempting to issue warnings of impending chaos; sic Personality Inventory (MMPI) depression at the very least, readers must exercise caution in scale, however, were lower, at 0.27 and 0.34 (12, comparing studies that purportedly used the p156). We conclude that, compared with clini- HRSD. Zitman et al. reviewed some of the many cian ratings, the HRSD consistently performs variants, noting that fewer than half of the 37 better than the SDS (33, Table 1; 34; 35), about researchers they interviewed had actually used as well as the BDI (12; 31), but less well than the the version of the HRSD that they cited (40). BRMS (31; 36). In a comparative study of the There follows a brief overview of some of the ability of several scales to discriminate between variants. patients with major depressive disorders and Although Hamilton regarded the 17-item ver- others with panic disorder, Somoza et al. found sion of the HRSD as definitive, the four extra the HRSD to be significantly better able to dis- items are sometimes scored to form a 21-item tinguish than the Beck Depression Inventory or version; the 17- and 21-item versions correlated the Beck Anxiety Inventory; indeed, the informa- 0.94 in one study (26, p15).There is also a 24- tion gain for the HRSD was twice that of the item version that adds items on helplessness, BDI (37, Table 2). hopelessness, and worthlessness (29). Another Hedlund and Vieweg summarized over 30 24-item version was published through the U.S. studies that reviewed the HRSD as an indicator Department of Health, Education, and Welfare of change in clinical trials; they commented: (6) and has been widely used, although it di- “The HRSD is consistently reported to reflect verges from Hamilton’s guidelines, and Hamil- clinically observed treatment changes . . . the ton did not approve (4). Hedlund and Vieweg HRSD and BDI tend to be about equally sensi- have published a variant of the HRSD (12, tive to severity of illness, while the SDS tends to pp163–164). Miller et al. added items related to be somewhat less reliable and sensitive” (12, cognitive and melancholic symptoms and deleted p156). A meta-analysis of 36 studies compared items with low reliability and validity. They effect sizes for the HRSD, the Zung Self-rating modified rating points and used a structured in- Depression Scale, and the BDI (38). A second re- terview guide (25, pp137–142). Other modifica- view compared effect sizes for BDI and HRSD in tions of the HRSD have reduced its overlap with 19 studies (39). In both, the HRSD was found to anxiety measurements, increasing specificity provide the largest consistent index of change. (41). Hammond noted the poor internal consis- Compared with other clinician ratings, however, tency of the HRSD and recommended deleting the HRSD may not be superior. Change scores 11 items, which increased alpha from 0.46 to on the HRSD were compared with a clinician’s 0.60 (42). Questionnaire versions have been de- classification of 35 patients under treatment into veloped (43), including one named the Hamilton improved or not improved; the point biserial Depression Inventory (44), whereas the best correlation was 0.59, comparing unfavorably known self-rating version, the Carroll Rating with a figure of 0.70 for the MADRS (19, p385). Scale for Depression, is described in a separate Maier et al. have compared the discriminatory review. power of the HRSD and MADRS in two studies; Hamilton’s hesitancy to publish precise de- in one the HRSD was superior (15, p8) and in tails of rating procedures stimulated others to at- the other the MADRS proved superior (17). tempt this. A Structured Interview Guide 374 Measuring Health developed by Williams, the SIGH-D, suggests sion; one review reported its use in at least 500 open-ended questions for each item and indi- published studies in 10 years (57). cates how to translate replies into ratings; use of the guide improved reliability of ratings (45, Since its initial publication, the HAM-D has Table 2). The SIGH-D is presented in an appen- emerged as the most widely used scale for pa- dix to Williams’s article (45, pp57–63). Potts et tient selection and follow-up in research al. developed a Structured Interview Guide for studies of treatments for depression. . . . Un- use in the Medical Outcomes Study, the SI- doubtedly, the success of this scale is due to HDRS, which can be used by lay interviewers its comprehensive coverage of depressive (46). The SI-HDRS has an alpha reliability of symptoms and related psychopathology, as 0.82 (46, p344). Rehm and O’Hara also pro- well as its strong psychometric properties. vided guidelines for standardizing the ratings (45, p48) (11). Difficulties with certain items led to revisions The literature on the HRSD has reached a size to the content of the scale. Bech et al. showed such that there are several major reviews (12; that the items did not all measure severity of de- 58). Validity appears adequate: the HRSD gener- pression and, using Rasch analysis, Bech identi- ally correlates well with ratings of depression fied six items that more specifically measure severity, albeit somewhat lower than the correla- severity: items 1 (depressed mood), 2 (guilt), 7 tions obtained for other rating scales. The HRSD (work and interests), 8 (retardation), 10 (psychic is generally held to be sensitive to change in de- anxiety), and 13 (general somatic symptoms) pression following treatment. Inter-rater reliabil- (31, p168). These formed a more adequate sever- ity is generally good, although agreement at the ity scale psychometrically, showing a more mo- item level is less satisfactory. Despite these criti- notonic relationship with clinical judgments of cisms, the HRSD is often used as a criterion severity of depression than the overall scale did against which self-report scales are validated; (47, Table 1). Bech et al. referred to these six the various attempts to improve the Hamilton items as the melancholia subscale of the HRSD scale have not led to its displacement as the (48). Later, Bech and Rafaelsen included this mainstay of depression rating scales. subset in their BRMS, which contains 11 items Although the Hamilton scale appears success- and more closely reflects the content of the ful, it has been criticized on a number of DSM-III-R and ICD-10 criteria; the 11 items grounds. Indeed, Hamilton (who was deeply re- form a Guttman-type cumulative severity scale spected by many of his critics) parried some of (27; 47; 49; 50). The scale properties of the the criticism with his wry and modest humor: BRMS are superior to those of the HRSD; inter- rater reliability is generally higher, and the BRMS seems more sensitive to clinical change I went around with my scale and it created a (15). Subsequently, Bech et al. developed a Ma- tremendous wave of apathy. They all thought jor Depression Rating scale that is based on the I was a bit mad. Eventually it got published in HRSD and BRMS (51). the Journal of Neurology, Neurosurgery and The Hamilton scale has been translated and Psychiatry. It was the only one that would tested in numerous languages, including French take it. And now everyone tells me the scale is (52), Spanish (10; 53), Dutch (40), Cantonese wonderful. I always remember when it had a (54), Arabic (55), Russian, Japanese, and Hindi different reception. This makes sure I don’t (9). A Brazilian version correlated 0.89 with the get a swollen head. (10, p90) MADRS (56). The issues regarding the HRSD include its be- Commentary havioral and somatic content that are greater For 40 years the HRSD has remained the most than those of other scales (33, p361); it has been frequently used clinical rating scale for depres- criticized for omitting affective symptoms (e.g.,  Depression 375 nonreactivity of mood, reduction of concentra- and clarification. Unfortunately, however, no one tion, anhedonia) (26, p18). This criticism should assumed stewardship of the scale by developing be considered carefully, however: the HRSD is a a manual, answering points raised, and issuing severity scale and not a diagnostic instrument, so revisions as appropriate. Development of a good it may not need to cover every aspect of depres- first edition of a scale is only half the task: main- sion (10, p81). But the criticism is important if tenance is equally important. the emphasis on somatic items, which are indica- Hedlund and Vieweg concluded that the tors of more severe depression, means that the HRSD is a valid scale for its purpose, and scale cannot distinguish milder levels of depres- Rabkin and Klein state that “we are confident sion (24). It may also mean that the HRSD exag- that the Hamilton is an accurate index of depres- gerates depression in patients with physical sive severity in patients diagnosed with a depres- illness and intercurrent depression (59). Hamil- sive disorder” (29, p52). It is suitable for ton argued that anxiety is an important symp- determining the severity of depression in de- tom of depression (60), but the inclusion of pressed patients, if proper caution is used. It can- items on anxiety will clearly reduce the ability of not be used to diagnose depression and so is not the HRSD to discriminate between anxiety and suitable for unselected populations (61). Care depression. should be taken in reporting which version of Several authors have criticized its lack of de- the scale is used and whether any changes are scription of item content and item scores (15). made. Thorough training is required on the in- Hamilton himself stated that the value of the strument, as well as inter-rater reliability checks scale “depends entirely on the skill of the inter- before and after training. viewer in eliciting the necessary information” (1, p56). Considerable interpretation is neces- sary for rating items; consistency across settings References is questionable (25). Other criticisms involve its (1) Hamilton M. A rating scale for depression. heterogeneous factor structure and the mixing J Neurol Neurosurg Psychiatry of frequency and severity of symptoms. Because 1960;23:56–62. it covers several facets of depression, patients (2) Hamilton M. Development of a rating scale with low scores on several symptoms can re- for primary depressive illness. Br J Soc Clin ceive the same scores as patients who score Psychol 1967;6:278–296. highly on a few symptoms; it is not clear (3) Roth M. Max Hamilton: a life devoted to whether these are equivalent in terms of severity psychiatric science. In: Bech P, Coppen A, eds. The Hamilton scales. Berlin: Springer- (17). This raises questions about the internal Verlag, 1990:1–9. structure of the HRSD. Having been developed (4) Hamilton M. The Hamilton rating scale empirically from reports made by patients, the for depression. In: Sartorius N, Ban T, eds. scale is heterogeneous in content and does not Assessment of depression. Berlin: Springer- meet the criteria for internal construct validity Verlag, 1986:143–152. laid down by the Rasch approach to test devel- (5) Lyerly SB. Handbook of psychiatric rating opment. Whether this is a limitation depends on scales. 2nd ed. Rockville, Maryland: one’s philosophy; the debate between theoreti- National Institute of Mental Health, 1978. cians and pragmatists is reminiscent of that sur- (6) Guy W. Early clinical drug evaluation rounding the Health Opinion Survey, another (ECDEU) assessment manual for scale from the same era that appears to meet its psychopharmacology. DHEW Pub No. 76- 338 ed. Rockville, Maryland: U.S. objectives despite far from perfect psychometric Department of Health, Education, and properties. Welfare National Institute of Mental The case of HRSD holds an important lesson Health, 1976. for development of the field in general. From the (7) Bech P, Kastrup M, Rafaelsen OJ. Mini- outset agreement was general that the scale was compendium of rating scales for states of basically sound but required some refinement anxiety, depression, mania, schizophrenia, 376 Measuring Health with corresponding DSM-III syndromes. validity study in patients with panic Acta Psychiatr Scand 1986;73(suppl disorders. Br J Psychiatry 326):1–37. 1992;160:206–211. (8) Paykel ES. Use of the Hamilton Depression (19) Montgomery SA, Åsberg M. A new scale in general practice. In: Bech P, depression scale designed to be sensitive to Coppen A, editors. The Hamilton scales. change. Br J Psychiatry Berlin: Springer-Verlag, 1990:40–47. 1979;134:382–389. (9) Gastpar M, Gilsdorf U. The Hamilton (20) Foster JR, Sclan S, Welkowitz J, et al. Depression Rating Scale in a WHO Psychiatric assessment in medical long- collaborative program. In: Bech P, Coppen term care facilities: reliability of commonly A, eds. The Hamilton scales. Berlin: used rating scales. Int J Geriatr Psychiatry Springer-Verlag, 1990:10–19. 1988;3:229–233. (10) Berrios GE, Bulbena-Villarasa A. The (21) Kørner A, Nielsen BM, Eschen F, et al. Hamilton Depression Scale and the Quantifying depressive symptomatology: numerical description of the symptoms of inter-rater reliability and inter-item depression. In: Bech P, Coppen A, eds. The correlations. J Affect Disord Hamilton scales. Berlin: Springer-Verlag, 1990;20:143–149. 1990:80–92. (22) Winokur A, Guthrie MB, Rickels K, et al. (11) Rehm LP, O’Hara MW. Item Extent of agreement between patient and characteristics of the Hamilton Rating physician ratings of emotional distress. Scale for Depression. J Psychiatr Res Psychosomatics 1982;23:1135–1146. 1985;19:31–41. (23) Endicott J, Cohen J, Nee J, et al. Hamilton (12) Hedlund JL, Vieweg BW. The Hamilton Depression Rating Scale extracted from Rating Scale for Depression: a regular and change versions of the Schedule comprehensive review. J Operat Psychiatry for Affective Disorders and Schizophrenia. 1979;10:149–165. Arch Gen Psychiatry 1981;38:98–103. (13) Diefenbach GJ, Stanley MA, Beck JG, et al. (24) Raskin A. Sensitivity to treatment effects of Examination of the Hamilton scales in evaluation instruments completed by assessment of anxious older adults: a psychiatrists, psychologists, nurses, and replication and extension. J Psychopathol patients. In: Sartorius N, Ban T, eds. Behav Assess 2001;23:117–124. Assessment of depression. Berlin: Springer, (14) Carroll BJ, Feinberg M, Smouse PE, et al. 1986:367–376. The Carroll Rating Scale for Depression. I. (25) Miller IW, Bishop S, Norman WH, Development, reliability and validation. Br Maddever H. The Modified Hamilton J Psychiatry 1981;138:194–200. Rating Scale for Depression: reliability and (15) Maier W, Philipp M, Heuser I, Schlegel S, validity. Psychiatry Res 1985;14:131–142. Buller R, Wetzel H. Improving depression (26) Maier W, Heuser I, Philipp M, et al. severity assessment—I. Reliability, internal Improving depression severity assessment— validity and sensitivity to change of three II. Content, concurrent and external observer depression scales. J Psychiatr Res validity of three observer depression scales. 1988;22:3–12. J Psychiatr Res 1988;22:13–19. (16) Maier W. The Hamilton Depression Scale (27) Bech P, Rafaelsen OJ. The use of rating and its alternatives: a comparison of their scales exemplified by a comparison of the reliability and validity. In: Bech P, Coppen Hamilton and the Bech-Rafaelsen A, editors. The Hamilton scales. Berlin: Melancholia Scale. Acta Psychiatr Scand Springer-Verlag, 1990:64–71. 1980;62(suppl 285):128–131. (17) Maier W, Philipp M. Comparative analysis (28) Onega LL, Abraham IL. Factor structure of observer depression scales. Acta of the Hamilton Rating Scale for Psychiatr Scand 1985;72:239–245. Depression in a cohort of community- (18) Bech P, Allerup P, Maier W, et al. The dwelling elderly. Int J Geriatr Psychiatry Hamilton scales and the Hopkins Symptom 1997;12:760–764. Checklist (SCL-90): a cross-national (29) Rabkin JG, Klein DF. The clinical measure-  Depression 377 ment of depressive disorders. In: Marsella Depression Rating Scale. In: Bech P, AJ, Hirschfeld RMA, Katz MM, editors. Coppen A, eds. The Hamilton scales. The measurement of depression. New Berlin: Springer-Verlag, 1990:28–34. York: Guilford Press, 1987:30–83. (41) Riskind JH, Beck AT, Brown G, et al. (30) Gibbons RD, Clark DC, Kupfer DJ. Taking the measure of anxiety and Exactly what does the Hamilton depression: validity of the reconstructed Depression Rating Scale measure? Int J Hamilton scales. J Nerv Ment Dis Psychiatr Res 1993;27:259–273. 1987;175:474–479. (31) Bech P, Gram LF, Dein E, et al. (42) Hammond MF. Rating depression severity Quantitative rating of depressive states: in the elderly physically ill patient; correlation between clinical assessment, reliability and factor structure of the Beck’s self-rating scale and Hamilton’s Hamilton and the Montgomery-Åsberg objective rating scale. Acta Psychiatr Scand depression rating scales. Int J Geriatr 1975;51:161–170. Psychiatry 1998;13:257–261. (32) Shumaker SA, Anderson R, Hays R, et al. (43) Bent-Hansen J, Lauritzen L, Clemmesen L, International use, application and perfor- et al. A definite and a semidefinite mance of health related quality of life questionnaire version of the instruments. Qual Life Res Hamilton/melancholia (HDS/MES) scale. J 1993;2:367–487. Affect Disord 1995;33:143–150. (33) Carroll BJ, Fielding JM, Blashki TG. (44) Reynolds WM, Kobak KA. Reliability and Depression rating scales: a critical review. validity of the Hamilton depression Arch Gen Psychiatry 1973;28:361–366. inventory: a paper-and-pencil version of (34) Schnurr R, Hoaken PCS, Jarrett FJ. the Hamilton depression rating scale Comparison of depression inventories in a clinical interview. Psychol Assess clinical population. Can Psychiatr Assoc J 2004;7:472–483. 1976;21:473–476. (45) Williams JBW. Structured interview guides (35) Biggs JT, Wylie LT, Ziegler VE. Validity of for the Hamilton rating scales. In: Bech P, the Zung Self-rating Depression Scale. Br J Coppen A, eds. The Hamilton scales. Psychiatry 1978;132:381–385. Berlin: Springer-Verlag, 1990:48–63. (36) Kearns NP, Cruickshank CA, McGuigan (46) Potts MK, Daniels M, Burnam MA, et al. KJ, et al. A comparison of depression A structured interview version of the rating scales. Br J Psychiatry Hamilton Depression Rating Scale: 1982;141:45–49. evidence of reliability and versatility of (37) Somoza E, Steer RA, Beck AT, et al. administration. J Psychiatr Res Differentiating major depression and 1990;24:335–350. panic disorders by self-report and clinical (47) Bech P. Psychometric development of the rating scales: ROC analysis and Hamilton scales: the spectrum of information theory. Behav Res Ther depression, dysthymia, and anxiety. In: 1994;32:771–782. Bech P, Coppen A, eds. The Hamilton (38) Lambert MJ, Hatch DR, Kingston MD, et scales. Berlin: Springer-Verlag, al. Zung, Beck, and Hamilton rating scales 1990:72–79. as measures of treatment outcome: a meta- (48) Bech P, Allerup P, Gram LF, et al. The analytic comparison. J Consult Clin Hamilton Depression Scale: evaluation of Psychol 1986;54:54–59. objectivity using logistic models. Acta (39) Edwards BC, Lambert MJ, Moran PW, et Psychiatr Scand 1981;63:290–299. al. A meta-analytic comparison of the Beck (49) Bech P, Rafaelsen OJ. The Melancholia Depression Inventory and the Hamilton Scale: development, consistency, validity, Rating Scale for Depression as measures of and utility. In: Sartorius N, Ban T, eds. treatment outcome. Br J Clin Psychol Assessment of Depression. Berlin: Springer, 1984;23:93–99. 1986:259–269. (40) Zitman FG, Mennen MFG, Griez E, et al. (50) Bech P. Rating scales for affective The different versions of the Hamilton disorders: their validity and consistency. 378 Measuring Health Acta Psychiatr Scand 1981;64(suppl 295):1–101. The Brief Assessment Schedule— (51) Bech P, Stage KB, Nair NP, et al. The Depression Major Depression Rating Scale (MDS). (A.J.D. MacDonald and others, Inter-rater reliability and validity across 1982) different settings in randomized moclobemide trials. Danish University Purpose Antidepressant Group. J Affect Disord The Brief Assessment Schedule—Depression 1997;42:39–48. (BAS-D) is a semistructured clinician rating (52) Pull CB. French experiences with the method intended to screen for depressive symp- Hamilton scales in comparison with other toms in geriatric medical services, whether in the scales for depression and anxiety. In: Bech community or in long-term care facilities. P, Coppen A, eds. The Hamilton scales. Berlin: Springer-Verlag, 1990:35–39. (53) Ramos-Brieva JA, Cordero-Villafafila A. A Conceptual Basis new validation of the Hamilton Rating The items of the BAS-D were derived from the Scale for Depression. J Psychiatr Res depression subscale of Gurland’s Comprehensive 1988;22:21–28. (54) Leung CM, Wing YK, Kwong PK, et al. Assessment and Referral Evaluation (CARE) Validation of the Chinese-Cantonese system (see Chapter 10). This was originally de- version of the Hospital Anxiety and veloped as part of a system to distinguish depres- Depression Scale and comparison with the sion from dementia, which are clinically hard to Hamilton Rating Scale of Depression. Acta separate because the two categories of symptoms Psychiatr Scand 1999;100:456–461. often coexist in the same patient (1). Hence Gur- (55) Stone EJ, Perry CL, Luepker RV. Synthesis land proposed an 18-item depression scale and a of cardiovascular behavioral research for 17-item dementia scale that were concurrently youth health promotion. Health Educ Q tested in the United States and in England (1, Ta- 1989;16:155–169. bles 1 and 2). These items subsequently went (56) Dratcu L, da Costa Ribeiro L, Calil HM. through various refinements to produce the ver- Depression assessment in Brazil: the first application of the Montgomery-Åsberg sion reviewed here (2–5). The BAS-D was in- Depression Rating Scale. Br J Psychiatry tended for use in residential settings (6). 1987;150:797–800. (57) Hooijer C, Zitman FG, Griez E, et al. The Hamilton Depression Rating Scale Description (HDRS): changes in scores as a function of The 21 items of the BAS-D were modified from training and version used. J Affect Disord the depression subscale of Gurland’s CARE and 1991;22:21–29. SHORT-CARE system. The depression items (58) Bech P, Coppen A. The Hamilton scales. 1 are often used along with eight of the items from ed. New York: Springer-Verlag, 1990. the organic brain syndrome (or dementia) scale (59) Linden M, Borchelt M, Barnow S, et al. from the CARE, which is shown in Exhibit The impact of somatic morbidity on the 10.23, but the present review covers only its de- Hamilton Depression Rating Scale in the pression component. very old. Acta Psychiatr Scand As with the CARE, the BAS-D rating method 1995;92:150–154. comprises a question or statement that is read to (60) Hamilton M. Frequency of symptoms in melancholia (depressive illness). Br J the subject, but one that may be supplemented Psychiatry 1989;154:201–206. by additional questions, such as “How long has (61) Mayer JM. Assessment of depression. In: this lasted?”, or “Tell me more about that”, McReynolds P, ed. Advances in which guide the rater in judging whether each psychological assessment. Vol. 4. San criterion shown in Exhibit 7.6 is met. Scores Francisco: Jossey-Bass, 1977:358–425. range from 0 to 24; a cutting-point of 6/7 is gen-  Depression 379 Exhibit 7.6 The Brief Assessment Scale for currently in England, Germany and Australia Depression (BAS-D) (7). The team that developed the BAS-D also pro- 1. Admits to worrying in the last month posed an eight-item abbreviation, called the 2. Worries about almost everything Even Briefer Assessment Scale for Depression 3. Sad or depressed mood in last month (EBAS-DEP) (8; 9). This included items on the 4. Depression lasts longer than a few hours BAS that had the highest item-total correlations; 5. Depression worst at beginning of the day a cut-off of 2/3 is suggested (8, p213). The EBAS 6. Has felt life not worth living in past month scale is shown in Exhibit 7.7. 7. Has not felt happy in last month 8. Bothered and depressed by current loneliness Reliability 9. Almost nothing enjoyed lately Reliability results were reported for the original 10. Less enjoyment than 1 year ago CARE versions of the depression items in several 11. Less enjoyment because depressed/nervous studies. The reports, unfortunately, do not indi- cate the items, making it difficult to know ex- 12. Has had episodes of depression longer than 1 week duration prior to past year actly which version was being used in each 13. Reports headaches in past month study. Typical results include inter-rater reliabil- ity of 0.94 and internal consistency of 0.75 (2, 14. Poor appetite in the absence of medical cause p167); alpha values of 0.87, 0.78 and 0.82 have 15. Slowed in physical movement compared to 1 year ago been reported (10, Table 1). Mann et al. reported intraclass correlations 16. Difficulty sleeping due to altered moods, thoughts or tension (ICC) between raters of 0.98 and 0.99 (7, Fig- 17. Has cried or felt like crying in past month ures 2 and 4). Kappa values for the items all ex- ceeded 0.75. They also reported an ICC of 0.80 18i. Pessimistic or empty expectations of future when bilingual patients were interviewed in En- ii. Thinks future bleak or unbearable glish and then again in German (7, Figure 6). Al- 19i. Has wished to be dead in past month pha internal consistency values were 0.93 for ii. Suicidal thoughts both German and English versions (7, p224). In iii. Serious consideration of methods of suicide or an Italian study, inter-rater agreement (kappa) at actual attempts in past month successive phases of the study ranged from 0.88 20. Obvious self-blame present to 0.70 (5, Table 3). 21. Describes self as not very happy/not at all happy Allen et al. compared performance of the (opposed to fairly or very happy) BAS-D and the EBAS in a sample of 811 elderly From Allen N, Ames D, Ashby D, Bennetts K, Tuckwell V, subjects in Australia and England (8). Because West C. A brief sensitive screening instrument for depression the EBAS retained the most internally consistent in late life. Age and Ageing 1994;23:218. By permission of Oxford University Press. items, alpha coefficient values were only about 0.05 lower than those for the full 21-item scale, and the area under the ROC curve was actually erally used (7, p222). Note that different ver- higher than that for the complete scale (8). Al- sions exist; the one shown in the article by Mann pha internal consistency for the EBAS was 0.80 et al. (7, Table 1) differs slightly from that (8, p213). In three German samples, the EBAS shown in our Exhibit. Owing to the difficulty in alpha values were all 0.85, whereas for an En- rating depression in subjects with cognitive im- glish sample alpha was 0.86 (9, Table 1). pairments, the BAS-D is typically administered only after the subjects have shown that they can Validity correctly state their age or birth date, or name In Gurland’s original studies, a score of 7 or the institution or place where they are being in- more had a sensitivity of 94% and a specificity terviewed (8, p214). The BAS-D was tested con- of 79% in a community sample, whereas in a 380 Measuring Health Exhibit 7.7 The Even Briefer Assessment Scale for Depression (EBAS DEP) The 8 items of this schedule require raters to make a judgment as to whether the proposition in the middle column is satisfied or not. If a proposition is satisfied then a depressive symptom is present and raters should ring ‘1’ in the right-hand column, otherwise ‘0’ should be ringed. Each question in the left-hand column must be asked exactly as printed but the follow-up or subsidiary questions may be used to clarify the initial answer until the rater can make a clear judgement as to whether the proposition is satisfied or not. For items which enquire about symptoms over the past month, note that the symptom need not have been present for the entire month nor at the moment of interview, but it should have been a problem for the patient or troubled him/her for some of the past month. Question Assessment Rating 1. Do you worry? In the past month? Admits to worrying in past month 1 0 2. Have you been sad or depressed in the Has had sad or depressed mood 1 0 past month? during the past month 3. During the past month have you ever Has felt that life was not worth 1 0 felt that life was not worth living? living at some time during the past month 4. How do you feel about your future? Pessimistic about the future or has 1 0 What are your hopes for the future? empty expectations (i.e. nothing to look forward to) 5. During the past month have you at any Has wished to be dead at any time 1 0 time felt you would rather be dead? during past month 6. Do you enjoy things as much as you Less enjoyment in activities than a 1 0 used to—say like you did a year ago? year previously If question 6 rated 0, then rate 0 for question 7 and skip to question 8. If question 6 rated 1, ask question 7. 7. Is it because you are depressed or Loss of enjoyment because of 1 0 nervous that you don’t enjoy things depression/nervousness as much? 8. In general how happy are you? Not very happy or not happy at all 1 0 (Read out) Are you—very happy—fairly happy—not very happy—or—not happy at all? From Weyerer S, Killmann U, Ames D, Allen N. The even briefer assessment scale for depression (EBAS-DEP). Int J Geriatr Psychi- atry 1999;14:473–480. British study of residents of long-term care facil- was 0.92 for the BAS-D, corresponding to a sen- ities the figures were lower, at 76% sensitivity sitivity of 0.89 and a specificity of 0.81. For the and 70% specificity (6, pp259–60). In an Italian EBAS, the area under the curve was 0.89, with a study, the BAS-D showed an area under the sensitivity of 91% and a specificity of 72% (8, ROC curve of 0.97, or a sensitivity of 95% and p216). specificity of 92% against a DSM-III clinical di- Two studies in Singapore compared the agnosis (5, Table 1 and Figure 2) EBAS with the Geriatric Depression Scale and a In the study by Allen et al., the EBAS corre- single-item depression measure. Sensitivity was lated 0.94 with the complete BAS-D (8, p216). 77% and specificity 90% for the EBAS, com- In determining whether subjects were depressed, pared with figures of 84% and 86% for the GDS the scales showed an 89% agreement; the main (11). ROC analysis showed the GDS to be discrepancy arising from the lower specificity of slightly superior. In a study of patients with de- the EBAS. Compared with a DSM-III-R diagno- mentia, the EBAS proved less adequate than the sis of depression, the area under the ROC curve Cornell Scale for Depression in Dementia (12).  Depression 381 Alternative Forms progress. New York: Raven Press, To simplify administration for elderly patients, 1982:119–125. the items may be printed in large print on 3 × 5 (2) Gurland B, Golden RR, Teresi JA, et al. cards; this version includes 19 items and is called The SHORT-CARE: an efficient BASDEC (13). The cards are presented one at instrument for the assessment of depression, dementia and disability. J a time and a “True” or “False” response is Gerontol 1984;39:166–169. recorded. Each true response gains one point, (3) Livingston G, Hawkins A, Graham N, et except for “I’ve given up hope” and “I’ve seri- al. The Gospel Oak Study: prevalence rates ously considered suicide,” which score two of dementia, depression and activity points, giving a maximum score of 21. Adminis- limitation among elderly residents in Inner tration takes between two and eight minutes, London. Psychol Med 1990;20:137–146. and a cut-point of 6/7 is used (13). This version (4) Gurland BJ, Teresi J, Smith WM, et al. performed identically to the Geriatric Depres- Effects of treatment for isolated systolic sion Scale in one study, with sensitivity of 71% hypertension on cognitive status and and specificity of 88% (13). depression in the elderly. J Am Geriatr Soc Translations into Italian (5; 14) and German 1988;36:1015–1022. (15) have been described. (5) Spagnoli A, Foresti G, MacDonald A, et al. Italian version of the organic brain syndrome and the depression scales from Commentary the CARE: evaluation of their performance Being based on the CARE, this scale benefits in geriatric institutions. Psychol Med from the extensive development work that went 1987;17:507–513. into the CARE in England, Australia, and the (6) Mann AH, Graham N, Ashby D. United States. The BAS-D was developed with an Psychiatric illness in residential homes for international perspective and is widely used in a the elderly. A survey in one London number of countries. borough. Age Ageing 1984;13:257–266. The main difficulty is to know exactly which (7) Mann AH, Ames D, Graham N, et al. The version of the BAS-D was used in a particular reliability of the Brief Assessment Scale. Int study. Gurland’s original work tested several ver- J Geriatr Psychiatry 1989;4:221–225. (8) Allen N, Ames D, Ashby D, et al. A brief sions, and their reports tended not to indicate sensitive screening instrument for which, or even how many, items were in the ver- depression in late life. Age Ageing sion used. It is not fully clear whether subse- 1994;23:213–219. quent users have adhered to a single version. It is (9) Weyerer S, Killmann U, Ames D, et al. The therefore difficult to interpret information on re- even briefer assessment scale for depression liability and validity, but the results that we do (EBAS DEP): its suitability for the elderly have seem adequate or good, if not entirely con- in geriatric care in English- and German- sistent. Having been developed by clinicians for speaking countries. Int J Geriatr Psychiatry clinicians, we know a fair amount about its crite- 1999;14:473–480. rion validity but the instrument has not received (10) Teresi JA, Golden RR, Gurland BJ, et al. intensive psychometric scrutiny. Construct validity of indicator-scales developed from the Comprehensive Assessment and Referral Evaluation interview schedule. J Gerontol References 1984;39:147–157. (1) Gurland B, Golden R, Challop J. (11) Lim PPJ, Ng LL, Chiam PC, et al. Unidimensional and multidimensional Validation and comparison of three brief approaches to the differentiation of depression scales in an elderly Chinese depression and dementia in the elderly. In: population. Int J Geriatr Psychiatry Corkin S, Crowdon JH, Davis KL, et al., 2000;15:824–830. eds. Alzheimer’s disease: a report of (12) Lam CK, Lim PPJ, Low BL, et al. 382 Measuring Health Depression in dementia: a comparative and ment. The items selected for the MADRS were validation study of four brief scales in the those most frequently checked and those most elderly Chinese. Int J Geriatr Psychiatry sensitive to change (3; 4). The ten MADRS rat- 2004;19:422–428. ings are completed during a clinical interview, (13) Adshead F, Cody DD, Pitt B. BASDEC: a which moves “from broadly phrased questions novel screening instrument for depression about symptoms to more detailed ones which al- in elderly medical inpatients. Br Med J low a precise rating of severity” (1, p387). Rat- 1992;305:397. (14) Spagnoli A, Foresti G, MacDonald A, et al. ings may blend patient responses with Dementia and depression in Italian information from observation or interviews with geriatric institutions. Int J Geriatr other informants. The MADRS is suitable for Psychiatry 1986;1:15–23. use by psychiatrists, general practitioners, psy- (15) Weyerer S, Platz S, Eichhorn S, et al. Die chologists, or nurses; interviews take 20 to 60 deutsche Version des Brief Assessment minutes. The ten ratings use 0 to 6 severity Interviews (BAI): ein Instrument zur scales, with higher scores reflecting more severe Erfassung von Demenz und Depression. symptoms. The topic of each rating is described, Zeitschr Gerontopsychol Psychiat and definitions are provided for key scale steps 1988;1:147–152. (Exhibit 7.8). Ratings can be added to form an overall score (range 0 to 60); no weights are used. Snaith et al The Montgomery-Åsberg Depression proposed the following cutting-points: scores of Rating Scale zero to six indicate an absence of symptoms; (Stuart A. Montgomery and Marie seven to 19 represent mild depression; 20 to 34 Åsberg, 1979) moderate; and 35 to 60 indicate severe depres- Purpose sion (5). Other cut-points proposed include 12 The Montgomery-Åsberg Depression Rating for mild depression, 24 for moderate, and 35 for Scale (MADRS) is used by clinicians to assess severe (6, p88). the severity of depression among patients in whom a diagnosis of depressive illness has been Reliability made. It is designed to be sensitive to change re- An alpha of 0.86 was found in a study of 151 sulting from treatment (1; 2). depressed patients (7), but a much lower alpha, 0.61, has also been reported (8). The consistency Conceptual Basis of fit of the MADRS scores to a Rasch measure- Psychiatric rating scales are commonly used to ment model across different subgroups of pa- evaluate drug treatment effects. Although most tients was assessed by Maier et al. and showed are able to distinguish between active drugs and clear deviations from the assumptions of the placebos, Montgomery and Åsberg noted that model (9, pp7–8). few have been able to differentiate between vari- As with any rating scale, inter-rater reliability ous active drugs. Standard rating scales such as is an important concern and has been frequently the Hamilton Rating Scale for Depression examined. Montgomery and Åsberg reported (HRSD) do not seem sensitive enough to detect inter-rater reliability ranged from 0.89 to 0.97 these differences (3). The major goal in develop- for various combinations of raters in small sam- ing the MADRS was to provide an instrument ples of 12 to 30 patients (1, Table III). Davidson sensitive to minor changes that was also practi- et al. reported Spearman inter-rater correlations cal to apply in a clinical setting. of 0.76 for the total score, whereas correlations for individual items ranged from 0.57 to 0.76 Description (2, Table 2). Intraclass coefficients for the The MADRS was developed from Åsberg’s Com- MADRS fell between 0.66 and 0.82 (9, Table 2). prehensive Psychopathological Rating Scale, An equivalent figure of 0.86 was reported by which was designed to evaluate psychiatric treat- Kørner et al. (10, Table 3). A Pearson correlation  Depression 383 Exhibit 7.8 The Montgomery-Åsberg Depression Rating Scale (MADRS) The rating should be based on a clinical interview moving from broadly phrased questions about symptoms to more detailed ones which allow a precise rating of severity. The rater must decide whether the rating lies on the defined scale steps (0, 2, 4, 6) or between them (1, 3, 5). It is important to remember that it is only on rare occasions that a depressed patient is encountered who cannot be rated on the items in the scale. If definite answers cannot be elicited from the patient all relevant clues as well as information from other sources should be used as a basis for the rating in line with customary clinical practice. The scale may be used for any time interval between ratings, be it weekly or otherwise but this must be recorded. 1. Apparent sadness Representing despondency, gloom and despair, (more than just ordinary transient low spirits) reflected in speech, facial expression, and posture. Rate by depth and inability to brighten up. 0 No sadness. 1 2 Looks dispirited but does brighten up without difficulty. 3 4 Appears sad and unhappy most of the time. 5 6 Looks miserable all the time. Extremely despondent. 2. Reported sadness Representing reports of depressed mood, regardless of whether it is reflected in appearance or not. Includes low spirits, despondency or the feeling of being beyond help and without hope. Rate according to intensity, duration and the extent to which the mood is reported to be influenced by events. 0 Occasional sadness in keeping with the circumstances. 1 2 Sad or low but brightens up without difficulty. 3 4 Pervasive feelings of sadness or gloominess. The mood is still influenced by external circumstances. 5 6 Continuous or unvarying sadness, misery or despondency. 3. Inner tension Representing feelings of ill-defined discomfort, edginess, inner turmoil, mental tension mounting to either panic, dread or anguish. Rate according to intensity, frequency, duration and the extent of reassurance called for. 0 Placid. Only fleeting inner tension. 1 2 Occasional feelings of edginess and ill-defined discomfort. 3 4 Continuous feelings of inner tension or intermittent panic which the patient can only master with some difficulty. 5 6 Unrelenting dread or anguish. Overwhelming panic. 4. Reduced sleep Representing the experience of reduced duration or depth of sleep compared to the subject’s own normal pattern when well. 0 Sleeps as usual. 1 2 Slight difficulty dropping off to sleep or slightly reduced, light or fitful sleep. 3 4 Sleep reduced or broken by at least two hours. 5 6 Less than two or three hours sleep. (continued) 383 384 Measuring Health Exhibit 7.8 (continued) 5. Reduced appetite Representing the feeling of a loss of appetite compared with when well. Rate by loss of desire for food or the need to force oneself to eat. 0 Normal or increased appetite. 1 2 Slightly reduced appetite. 3 4 No appetite. Food is tasteless. 5 6 Needs persuasion to eat at all. 6. Concentration difficulties Representing difficulties in collecting one’s thoughts mounting to incapacitating lack of concentration. Rate according to intensity, frequency, and degree of incapacity produced. 0 No difficulties in concentrating. 1 2 Occasional difficulties in collecting one’s thoughts. 3 4 Difficulties in concentrating and sustaining thought which reduces ability to read or hold a conversation. 5 6 Unable to read or converse without great difficulty. 7. Lassitude Representing a difficulty getting started or slowness initiating and performing everyday activities. 0 Hardly any difficulties in getting started. No sluggishness. 1 2 Difficulties in starting activities. 3 4 Difficulties in starting simple routine activities which are carried out with effort. 5 6 Complete lassitude. Unable to do anything without help. 8. Inability to feel Representing the subjective experience of reduced interest in the surroundings, or activities that normally give pleasure. The ability to react with adequate emotion to circumstances or people is reduced. 0 Normal interest in the surroundings and in other people. 1 2 Reduced ability to enjoy usual interests. 3 4 Loss of interest in the surroundings. Loss of feelings for friends and acquaintances. 5 6 The experience of being emotionally paralysed, inability to feel anger, grief or pleasure and a complete or even painful failure to feel for close relatives and friends. 9. Pessimistic thoughts Representing thoughts of guilt, inferiority, self-reproach, sinfulness, remorse and ruin. 0 No pessimistic thoughts. 1 2 Fluctuating ideas of failure, self-reproach or self-depreciation. 3 4 Persistent self-accusations, or definite but still rational ideas of guilt or sin. Increasingly pessimistic about the future. 5 6 Delusions of ruin, remorse or unredeemable sin. Self-accusations which are absurd and unshakable.  Depression 385 Exhibit 7.8 10. Suicidal thoughts Representing the feeling that life is not worth living, that a natural death would be welcome, suicidal thoughts, and preparations for suicide. Suicidal attempts should not in themselves influence the rating. 0 Enjoys life or takes it as it comes. 1 2 Weary of life. Only fleeting suicidal thoughts. 3 4 Probably better off dead. Suicidal thoughts are common, and suicide is considered as a possible solution, but without specific plans or intention. 5 6 Explicit plans for suicide when there is an opportunity. Active preparations for suicide. Adapted from Montgomery SA, Åsberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry 1979;134:387–389. With permission. of 0.98 between raters has been reported (11, study, MADRS scores correlated rho = 0.83 with p135). a clinician’s global severity rating (5, p600). The MADRS has also been shown to be more suc- Validity cessful at distinguishing endogenous and nonen- In terms of content validity, the MADRS covers dogenous depression than the HRSD (2, Table the core symptoms of depression with the excep- 4). The four MADRS factor scores were not tion of motor retardation (1; 12). One analysis found to distinguish between unipolar and bipo- identified four factors: suicidal thoughts, pes- lar depression (16). simism and sadness; inability to feel; concentra- Snaith and Taylor reported convergent and tion difficulties; and reduced sleep and appetite discriminant validity: MADRS scores correlated (13). Another analysis identified two factors la- 0.81 with the depression score of the Hospital beled anhedonia and dysphoria (8, Table 2). Anxiety and Depression Scale (HADS) and 0.37 Compared with a clinical rating of depres- with its anxiety score (17, Table 1). Two items, sion severity, the MADRS discriminated signifi- psychic anxiety and insomnia, showed correla- cantly between five levels of depression; its tions in excess of 0.5 with the anxiety score, and performance was about equal to that of the omitting these items from the MADRS reduced HRSD and clearly superior to that of the Beck the discriminant correlation from 0.37 to 0.31 Depression Inventory (BDI) (14, p47). Maier et although leaving the correlation with the depres- al. compared the MADRS and HRSD with clini- sion score unchanged (17, Table 1). Davidson et cal assessments of severity of depression; the al. reported Spearman correlations averaging correlation for the MADRS was 0.71, slightly only 0.46 between HRSD ratings and the higher than that for the HRSD (0.65) (12, Tables MADRS (2, Table 3). However, this seems the 1, 2). Equivalent figures from a second study exception. Other coefficients are higher, includ- were 0.75 for the MADRS and 0.68 for the ing Spearman correlations with the HRSD of HRSD (15, Table 3). All items on the MADRS 0.82 and with the Bech-Rafaelsen Melancholia were significantly associated with severity rat- Scale (BRMS) of 0.92 (10, Table 4), and Pearson ings. In addition, the MADRS correlated more correlations of 0.85 with the HRSD, 0.89 with highly with the global assessment of depression the BRMS, and 0.71 with the Raskin Depression severity than with that of anxiety, was able to Scale (12, pp15–16). discriminate major depression with and without Because a goal of the MADRS was to evalu- melancholia, and was significantly associated ate change in depression, sensitivity to change with psychosocial impairment (12). In Snaith’s has been reported in several studies. Change 386 Measuring Health scores on the MADRS were compared with a terms of general validity and reliability the clinician’s classification of 35 patients under MADRS performs comparably with, or slightly treatment into improved or not improved; the better than, the HRSD. Studies give varying im- point biserial correlation was 0.70, although the pressions of whether the MADRS is, in fact, equivalent correlation for the HRSD was lower, more sensitive to change; overall, little difference at 0.59 (1, p385). Maier and Phillipp found that exists between the MADRS and the Hamilton. MADRS had higher mean discriminatory power The main contrast between the two scales is that (coefficient = 0.60) than the Hamilton (coeffi- the MADRS omits psychomotor symptoms. Fo- cient = 0.39) (7). By contrast, in two other stud- cusing only on psychic symptoms of depression, ies the MADRS had slightly lower correlations the MADRS is valuable for assessing depression with global assessments of change (0.61 and in physically ill people (5), whereas if the appli- 0.63) than did the HRSD or the BRMS (9, p8; cation calls for a broad assessment of the psy- 15, Table 3). chic, behavioral, and somatic features of depression, the HRSD is more suitable (14, Alternative Forms p48). English and Swedish versions of the MADRS were developed concurrently. A Brazilian version showed a correlation of 0.89 with the Hamilton References scale (18). A French version has been validated (19), as has a Spanish, giving an internal consis- (1) Montgomery SA, Åsberg M. A new depression scale designed to be sensitive to tency of 0.88, retest reliability of 0.94 (18–20) change. Br J Psychiatry and inter-rater agreement of 0.98 (20). 1979;134:382–389. Given that items 3 (inner tension) and 4 (re- (2) Davidson J, Turnbull CD, Strickland R, et duced sleep) have been found to correlate highly al. The Montgomery Åsberg Depression with an anxiety scale (17), some users have Scale: reliability and validity. Acta omitted both (11), or only item 3 (21) to form Psychiatr Scand 1986;73:544–548. an abbreviated version of the MADRS designed (3) Åsberg M, Montgomery SA, Perris C, et al. to give a purer measure of depression. This gave A comprehensive psychopathological rating a correlation of 0.70 with the Hospital Anxiety scale. Acta Psychiatr Scand 1978;58(suppl and Depression Scale (HADS), and 0.64 with the 271):5–27. Zung depression scale (11, Table 1). Indicating (4) Cronholm B. The Comprehensive Psychopathological Rating Scale (CPRS). discriminant validity, correlations with anxiety In: Helgason T, ed. Methodology in scales were lower, at 0.47 for the HADS anxiety evaluation of psychiatric treatment. scale, 0.35 with Zung’s anxiety scale, and 0.18 Cambridge: Cambridge University Press, with the Clinical Anxiety Scale (11, p135). 1983:163–169. (5) Snaith RP, Harrop FM, Newby DA, et al. Commentary Grade scores of the Montgomery Åsberg The MADRS was developed largely in response Depression and the Clinical Anxiety Scales. to criticism that established modalities such as Br J Psychiatry 1986;148:599–601. the HRSD were insensitive to clinically impor- (6) Wilkin D, Hallam L, Doggett MA. Mea- tant changes in level of depression. It has been sures of need and outcome for primary used widely in Britain and Europe. Most evalua- health care. Oxford: Oxford University Press, 1992. tions of the MADRS compare it with the Hamil- (7) Maier W, Philipp M. Comparative analysis ton scale, and it makes a generally favorable of observer depression scales. Acta impression. Compared with the HRSD, the Psychiatr Scand 1985;72:239–245. MADRS is easily administered and provides (8) Hammond MF. Rating depression severity clearer guidelines to the rater. In turn, however, in the elderly physically ill patient; the Bech-Rafaelstan modification of the Hamil- reliability and factor structure of the ton scale seems to offer further improvement. In Hamilton and the Montgomery-Åsberg  Depression 387 depression rating scales. Int J Geriatr Montgomery-Åsberg depression and Psychiatry 1998;13:257–261. Hamilton anxiety rating scales]. Med Clin (9) Maier W, Philipp M, Heuser I, et al. (Barc) 2002;118:493–499. Improving depression severity assessment— (21) Aylard PR, Gooding JH, McKenna PJ, et I. Reliability, internal validity and al. A validation study of three anxiety and sensitivity to change of three observer depression self-assessment scales. J depression scales. J Psychiatr Res Psychosom Res 1987;31:261–268. 1988;22:3–12. (10) Kørner A, Nielsen BM, Eschen F, et al. Quantifying depressive symptomatology: The Carroll Rating Scale For inter-rater reliability and inter-item Depression correlations. J Affect Disord (Bernard J. Carroll, 1981, 1998) 1990;20:143–149. (11) Bramley PN, Easton AME, Morley S, et al. Purpose The differentiation of anxiety and The Carroll Rating Scale (CRS) is a self- depression by rating scales. Acta Psychiatr administered version of the Hamilton Rating Scand 1988;77:133–138. Scale for Depression (HRSD) (1). It is intended (12) Maier W, Heuser I, Philipp M, et al. to measure the severity of diagnosed depression. Improving depression severity assessment— It can also be used as a screening instrument and II. Content, concurrent and external in research and clinical applications. validity of three observer depression scales. J Psychiatr Res 1988;22:13–19. (13) Craighead WE, Evans DD. Factor analysis Conceptual Basis of the Montgomery-Åsberg Depression The CRS closely follows the item content of the Rating Scale. Depression 1996;4:31–33. HRSD (1). The major difference between the (14) Kearns NP, Cruickshank CA, McGuigan HRSD and most of the self-rating scales lies in KJ. A comparison of depression rating its coverage of somatic and behavioral symp- scales. Br J Psychiatry 1982;141:45–49. toms of depression. To avoid falsely classifying (15) Maier W. The Hamilton Depression Scale physical illness as symptomatic of depression, and its alternatives: a comparison of their most self-rating scales focus on subjective reliability and validity. In: Bech P, Coppen thoughts and feelings, leaving physical symp- A, editors. The Hamilton scales. Berlin: toms to a clinician to elicit through clinical ex- Springer-Verlag, 1990:64–71. amination. However, because feelings may (16) Benazzi F. Comparison of MADRS factors indicate general malaise rather than specifically in bipolar II depression versus unipolar depression. Dep Anx 1998;8:49. depression, a self-rating scale oriented toward (17) Snaith RP, Taylor CM. Rating scales for the behavioral and somatic features of depres- depression and anxiety: a current sion may be advantageous (2). perspective. Br J Clin Pharmacol 1985;19:17S–20S. Description (18) Dratcu L, da Costa Ribeiro L, Calil HM. The CRS as originally published in 1981 con- Depression assessment in Brazil: the first tained 52 items; an expanded version was devel- application of the Montgomery-Åsberg oped in 1998. The latter includes 61 items that Depression Rating Scale. Br J Psychiatry parallel the DSM-IV diagnostic criteria. This 1987;150:797–800. version is copyrighted by Multi-Health Systems (19) Peyre F, Martinez R, Calache M, et al. Inc (www.mhs.com), who also produce scoring New validation of the Montgomery and Åsberg Depression Scale (MADRS) on a software and diagnostic guides. Because most of sample of 147 hospitalized depressed the reported validity and reliability evidence patients. Ann Med Psychol (Paris) refers to the original version of the scale, it will 1989;147:762–767. be described here. Interested users can then con- (20) Lobo A, Chamorro L, Luque A, et al. tact Multi-Health Systems to obtain the revised [Validation of the Spanish versions of the version. 388 Measuring Health Like the HRSD, the 1981 Carroll Rating lower for adolescents: Robbins et al. obtained a Scale covered 17 symptoms of depression. figure of 0.46 for 81 adolescents aged 13 to 18 Where Hamilton used a 0 to 2 rating for a symp- (5, p124). Feinberg et al. provided some evi- tom, the CRS includes two items and where dence that the CRS may prove more sensitive to Hamilton used a 0 to 4 rating, Carroll uses four a global clinical rating of depression than the items, giving a total of 52 yes/no statements and HRSD (3, Figure 1). The CRS correlated 0.67 a maximum severity score of 52 (1, p194). The with a global rating of depression applied to a items are presented in a randomized order, as sample of 232 outpatients (3, Table Ib). It corre- shown in Exhibit 7.9, which indicates the re- lated −0.68 with a visual analogue mood scale sponse that is scored for each item. Each depres- phrased “How are you feeling today?” (3, Table sive response counts one point toward the total Ib) and 0.86 with the Beck Depression Inventory score. The 17 HRSD symptoms of depression (BDI) (1, p197). Using partial correlations, Fein- and the corresponding Carroll items are as fol- berg and Carroll showed that the BDI did not lows (1, pp198–199): contain information in the HRSD beyond that also included in the CRS; the CRS did, however, 1. Depression 16, 32, 34, 48 contain information in the HRSD that was not 2. Guilt 14, 20, 24, 44 covered by the BDI. Furthermore, the CRS and 3. Suicide 12, 17, 29, 46 the BDI were significantly correlated (0.77) after 4. Initial insomnia 9, 22 their intercorrelations with the HRSD were par- 5. Middle insomnia 19, 27 tialed out. This suggests that the self-rating 6. Delayed insomnia 11, 35 scales have access to a subjective dimension of 7. Work and interests 3, 7, 25, 42 depression not covered by the clinical rating 8. Retardation 21, 28, 30, 47 method (6, p197). 9. Agitation 6, 10, 37, 43 Smouse et al. compared the factor structure 10. Psychic anxiety 8, 23, 31, 38 of the CRS and HRSD (N = 278). A strong first 11. Somatic anxiety 13, 18, 33, 41 factor underlay both instruments, and they con- 12. Gastrointestinal somatic symptoms 36, cluded that use of subscores (e.g., symptom 50 scores) for either instrument holds no advantage 13. General somatic symptoms 1, 51 over using the total scores (7). 14. Libido 4, 15 15. Hypochrondriasis 5, 39, 45, 49 Alternative Forms 16. Loss of weight 2, 52 French, Italian, and Chinese translations of the 17. Loss of insight 26, 40. CRS exist (6, p189). A 12-item abbreviation of the CRS, the Brief Carroll proposed that a cutoff of 10 should be Carroll Depression Rating Scale (BCDRS) has used when the CRS is applied as a screening in- been used as a depression screening tool for strument (1, p195). older adults (8). It requires about two minutes to complete. Koenig et al. tested the BCDRS as a Reliability screen for depression on 64 older medically ill Split-half reliability was 0.87 based on multiple hospital patients; a cutoff of 6 was optimal, giv- ratings of a mixed sample of patients (1, p197). ing a 100% sensitivity for detecting clinically di- Item-total correlations ranged from 0.05 to 0.78 agnosed DSM-III depression at a specificity of with a median of 0.55 (1, p198), similar to those 93%. Kappa agreement with the clinician’s rat- for the HRSD. ing was 0.76 (8, Table 2). In a separate study, however, they reported less satisfactory results: Validity sensitivity and specificity were 73% and 79%, Correlations of the CRS and the HRSD include respectively; among those whose results were 0.71 and 0.80 (1, p197), 0.75 (3, Table Ib), and positive, the likelihood of a diagnosis of major 0.79 (4, p167). The correlation may, however, be depressive disorder was 28% (9, Table 2). Exhibit 7.9 The Carroll Rating Scale for Depression Note: An asterisk indicates the response that is scored. These are not shown on the version presented to the respondent. Complete ALL the following statements by CIRCLING YES or NO, based on how you have felt during the past few days. 1. I feel just as energetic as always Yes No* 2. I am losing weight Yes* No 3. I have dropped many of my interests and activities Yes* No 4. Since my illness I have completely lost interest in sex Yes* No 5. I am especially concerned about how my body is functioning Yes* No 6. It must be obvious that I am disturbed and agitated Yes* No 7. I am still able to carry on doing the work I am supposed to do Yes No* 8. I can concentrate easily when reading the papers Yes No* 9. Getting to sleep takes me more than half an hour Yes* No 10. I am restless and fidgety Yes* No 11. I wake up much earlier than I need to in the morning Yes* No 12. Dying is the best solution for me Yes* No 13. I have a lot of trouble with dizzy and faint feelings Yes* No 14. I am being punished for something bad in my past Yes* No 15. My sexual interest is the same as before I got sick Yes No* 16. I am miserable or often feel like crying Yes* No 17. I often wish I were dead Yes* No 18. I am having trouble with indigestion Yes* No 19. I wake up often in the middle of the night Yes* No 20. I feel worthless and ashamed about myself Yes* No 21. I am so slowed down that I need help with bathing and dressing Yes* No 22. I take longer than usual to fall asleep at night Yes* No 23. Much of the time I am very afraid but don’t know the reason Yes* No 24. Things which I regret about my life are bothering me Yes* No 25. I get pleasure and satisfaction from what I do Yes No* 26. All I need is a good rest to be perfectly well again Yes* No 27. My sleep is restless and disturbed Yes* No 28. My mind is as fast and alert as always Yes No* 29. I feel that life is still worth living Yes No* 30. My voice is dull and lifeless Yes* No 31. I feel irritable or jittery Yes* No 32. I feel in good spirits Yes No* 33. My heart sometimes beats faster than usual Yes* No 34. I think my case is hopeless Yes* No 35. I wake up before my usual time in the morning Yes* No 36. I still enjoy my meals as much as usual Yes No* 37. I have to keep pacing around most of the time Yes* No 38. I am terrified and near panic Yes* No 39. My body is bad and rotten inside Yes* No 40. I got sick because of the bad weather we have been having Yes* No 41. My hands shake so much that people can easily notice Yes* No 42. I still like to go out and meet people Yes No* 43. I think I appear calm on the outside Yes No* 44. I think I am as good a person as anybody else Yes No* 45. My trouble is the result of some serious internal disease Yes* No 46. I have been thinking about trying to kill myself Yes* No 47. I get hardly anything done lately Yes* No 48. There is only misery in the future for me Yes* No 49. I worry a lot about my bodily symptoms Yes* No 50. I have to force myself to eat even a little Yes* No 51. I am exhausted much of the time Yes* No 52. I can tell that I have lost a lot of weight Yes* No Adapted from Carroll BJ, Feinberg M, Smouse PE, Rawson SG, Greden JF. The Carroll Rating Scale for Depression. I. Develop- ment, reliability and validation. Br J Psychiatry 1981;138;198–200. With permission. 389 390 Measuring Health Commentary Development, reliability and validation. Br The CRS accurately reflects information contained J Psychiatry 1981;138:194–200. in the HRSD and, with the possible exception of (2) Carroll BJ, Fielding JM, Blashki TG. assessing adolescents, intra-scale agreement is Depression rating scales: a critical review. high. Available evidence suggests that the CRS is Arch Gen Psychiatry 1973;28:361–366. (3) Feinberg M, Carroll BJ, Smouse PE, et al. reliable and the available validity results are prom- The Carroll Rating Scale for Depression. ising. By saving the clinician’s time, the CRS offers III. Comparison with other rating an economical alternative in regular monitoring of instruments. Br J Psychiatry patients or in routine practice to identify patients 1981;138:205–209. who need more extensive follow-up (10). Its ade- (4) Nasr SJ, Altman EG, Rodin MB, et al. quacy as a screen for depression needs more study. Correlation of the Hamilton and Carroll As with any self-rating scale, some patients may be Depression Rating Scales: a replication unable to complete the CRS due to illiteracy or se- study among psychiatric outpatients. J Clin vere illness (4). Psychiatry 1984;45:167–168. The scoring system of the CRS assumes that (5) Robbins DR, Alessi NE, Colfer MV, et al. patients who endorse the more severe items in a Use of the Hamilton Rating Scale for Depression and the Carroll Self-rating Scale symptom category will also endorse those less in adolescents. Psychiatry Res severe, thus giving a high score to that area in a 1985;14:123–129. manner equivalent to the higher weight assigned (6) Feinberg M, Carroll BJ. The Carroll Rating by a clinician using the HRSD (10, p70). This Scale for Depression. In: Sartorius N, Ban seems valid for most categories, with the possi- T, editors. Assessment of depression. ble exception of the symptoms of somatic anxi- Berlin: Springer, 1986:188–199. ety and loss of insight. The random ordering of (7) Smouse PE, Feinberg M, Carroll BJ, et al. items in the form used by patients may also gen- The Carroll Rating Scale for Depression. II. erate some inconsistencies. Evidence from two Factor analyses of the feature profiles. Br J studies suggests that the lowest agreement be- Psychiatry 1981;138:201–204. tween self-ratings and clinicians’ judgments lies (8) Koenig HG, Meador KG, Cohen HJ, et al. Self-rated depression scales and screening in the scales on weight loss, retardation, and ag- for major depression in the older itation (1, p196; 4, p167). hospitalized patient with medical illness. J The HRSD has many strong points and has Am Geriatr Soc 1988;36:699–706. been extremely widely used; a good self- (9) Koenig HG, Meador KG, Cohen HJ, et al. administered version therefore holds consider- Screening for depression in hospitalized el- able inherent value. Carson commented: “it derly medical patients: taking a closer look. appears to be a promising measure, potentially J Am Geriatr Soc 1992;40:1013–1017. superior to other self-report depression scales” (10) Rabkin JG, Klein DF. The clinical measure- (11, p420). ment of depressive disorders. In: Marsella AJ, Hirschfeld RMA, Katz MM, editors. Address The measurement of depression. New York: Guilford Press, 1987:30–83. USA: Multi-Health Systems Inc. 908 Niagara (11) Carson TP. Assessment of depression. In: Falls Blvd, North Tonawanda, NY 14120-2060 Ciminero AR, Calhoun KS, Adams HE, (1-800-456-3003) eds. Handbook of behavioural assessment. Canada: Multi-Health Systems Inc. 3770 Victo- New York: Wiley, 1986:404–445. ria Park Ave, Toronto ON M2H 3M6 (1-800- 268-6011) www.mhs.com. Conclusion References The scope of any review must be restricted, and (1) Carroll BJ, Feinberg M, Smouse PE, et al. many other depression measures could have The Carroll Rating Scale for Depression. I. been included in this chapter. Scales in other  Depression 391 chapters are relevant: for example, the depres- pression. The depression checklist replicates the sion component of the Hospital Anxiety and De- Research Diagnostic Criteria (14, Table 5). pression scale, or of the Depression, Anxiety and Finally, evidence suggests that two simple Stress Scales that are both reviewed in Chapter questions can perform remarkably well in case- 6. Several of the general health measures re- finding in primary care settings (15). The ques- viewed in Chapter 10 contain small depression tions are “During the past month, have you components such as the DUKE Anxiety- often been bothered by feeling down, depressed, Depression scale, as does the General Health or hopeless?” and “During the past month, have Questionnaire reviewed in Chapter 5. you often been bothered by little interest or plea- There are many reviews of depression scales sure in doing things?” This recalls the successful to which the reader can turn for descriptions of use of visual analogue scales (VAS) for rating de- methods we did not review (1–6). Several scales pression; we review the VAS as a pain measure in that are not reviewed here are closely related to Chapter 9 and for other topics in Chapter 10. those that are included. The Wakefield Depres- Correlations between VAS measures and full de- sion Inventory (7), for example, is a variant of pression scales can be high, suggesting that a Zung’s scale and Max Hamilton was involved in simple VAS may be useful in many applications. its development; the Cronholm-Ottosson De- One review of VAS in measuring depression pression Rating Scale (8) was developed by a quoted correlations between the VAS and the team that also involved Marie Åsberg. Other Beck Depression Inventory in four studies rang- measures are more broad-ranging; the scales in- ing from 0.53 to 0.76; correlations with the cluded in this chapter measure depression alone. Hamilton Rating Scale for Depression ranged For example, we did not review more general as- from 0.63 to 0.79 (three studies) and correla- sessments that include a depression component, tions with the Zung Self-rating Depression Scale such as the Middlesex Hospital Questionnaire ranged from 0.51 to 0.83 (16, Table 1). (9), Åsberg et al.’s Comprehensive Psychopatho- Those who have wrestled with the choice of a logical Rating Scale (10), or the Schedule for Af- depression measure will have discovered that no fective Disorders and Schizophrenia (SADS) instrument covers all symptoms of depression. (11). The Hopkins Symptom Checklist (HSCL) For an outcome measure or a measure of sever- contains an 11-item depression subscale (12). ity, this may not matter; for screening instru- The omission of combined instruments does not ments, however, there may be stronger imply any limitations in them; on the contrary, arguments for comprehensive coverage. The they prove ideal for many applications and read- choice between self-report and clinician-rating ers who wish to measure general psychopathol- methods is complex but some guidelines can be ogy may find them useful. given. Self-rating methods are economical and Many other purpose-built depression scales do not require a skilled administrator. They can- might also be considered. The Popoff Index of not provide diagnostic information but serve Depression, for example, is a 15-item screening well as screening instruments. For this, the Beck questionnaire for use in the family physician’s Depression Inventory (BDI), Geriatric Depres- office (13). Several clinical rating scales for de- sion Scale (GDS), or Zung instruments are suit- pression deserve mention. The Raskin depres- able. The main difficulty with self-ratings lies in sion scale is a brief clinical rating that covers the respondent’s lack of skill and experience: three symptom areas: the patient’s verbal report having a limited perspective on the range of (e.g., reports feeling helpless); the patient’s be- severity of depression, she can only rely on her havior (e.g., cries, lacks energy), and secondary own experience in judging severity of symptoms, symptoms (e.g., insomnia, lack of appetite). so self-rating methods are limited as measures of Each is rated on a five-point intensity scale (14, intensity of depression (4, p408). Self-ratings as- Table 4). As an alternative, the American College sume, of course, that accurate information will of Neuropsychopharmacology (ACNP) has pro- be reported; this may be problematic because a posed clinician checklists for anxiety and for de- depressed patient may not be aware of feelings 392 Measuring Health or may be unable or unwilling to reveal them. not seem a priority for expansion within the Where the depressed patient cannot judge and field of health measurement. rate her symptoms accurately, the cost of clinician-rating methods may be justified. Self- References ratings and clinician-ratings probably reflect dif- ferent aspects of depression and cannot be (1) Shaw BF, Vallis TM, McCabe SB. The expected to have identical results: assessment of the severity and symptom patterns in depression. In: Beckham EE, The BDI, for example, emphasizes the subjec- Leber WR, eds. Handbook of depression: treatment, assessment and research. tive experience of depression including pes- Homewood, Illinois: Dorsey Press, simism and self-punitive wishes. On the other 1985:372–407. hand, the HRS-D seems to emphasize symp- (2) Carson TP. Assessment of depression. In: toms reflecting the intensity of depression Ciminero AR, Calhoun KS, Adams HE, and its behavioral manifestations. Whereas eds. Handbook of behavioural assessment. 29% of a BDI score may be attributable to a New York: Wiley, 1986:404–445. physiological factor, 50% to 80% of the (3) Rabkin JG, Klein DF. The clinical measure- total score on the HRS-D is made up of be- ment of depressive disorders. In: Marsella havioral and physiological (somatic) compo- AJ, Hirschfeld RMA, Katz MM, eds. The nents. (17, p57) measurement of depression. New York: Guilford Press, 1987:30–83. (4) Mayer JM. Assessment of depression. In: This leads to different screening applications for McReynolds P, ed. Advances in each type of measure: “Since cognitive distur- psychological assessment. Vol. 4. San bances are prominent in mild or ‘neurotic’ de- Francisco: Jossey-Bass, 1977:358–425. pressions and physiologic disturbances are (5) Attkisson CC, Zich JM. Depression in prominent in severe or ‘psychotic’ disturbances, primary care: screening and detection. New a self-report scale like the BDI may be a more ac- York: Routledge, 1990. curate measure in the neurotically depressed (6) Moran PW, Lambert MJ. A review of while the HRS may be more accurate in the psy- current assessment tools for monitoring chotically depressed” (18, p85). However, if the changes in depression. In: Lambert MJ, purpose is to rate severity, it remains possible Christensen ER, DeJulio SS, eds. The mea- that patients with neurotic depression may exag- surement of psychotherapy outcome in research and evaluation. New York: Wiley, gerate their symptoms compared with findings in 1983:263–303. a clinician’s rating (4, p408). Aside from cost, (7) Snaith RP, Ahmed SN, Mehta S, et al. the most commonly cited limitation to clinical Assessment of the severity of primary ratings is the possibility of bias in the rater; depressive illness: Wakefield Self- Hamilton proposed that two independent raters assessment Depression Inventory. Psychol should compare their impressions, an approach Med 1971;1:143–149. that must rarely be feasible in practice. As an (8) Cronholm B, Schalling D, Åsberg M. outcome indicator, the Hamilton scale seems to Development of a rating scale for be more sensitive to change than either the Beck depressive illness. Mod Probl or Zung scale, but certain instruments may be Pharmacopsychiatry 1974;7:139–150. suited to certain types of treatment: there was a (9) Crown S, Crisp AH. A short clinical diagnostic self-rating scale for larger contrast in effect sizes between the Zung psychoneurotic patients. The Middlesex and Hamilton scales for evaluating psychother- Hospital Questionnaire (M.H.Q.). Br J apy than pharmacotherapy (17, p58). Psychiatry 1966;112:917–923. Although many will continue to search for (10) Åsberg M, Montgomery SA, Perris C, et al. the more perfect mousetrap, the range of depres- A comprehensive psychopathological rating sion measures we have seems adequate for most scale. Acta Psychiatr Scand 1978;58(suppl purposes. The measurement of depression does 271):5–27.  Depression 393 (11) Endicott J, Spitzer RL. A diagnostic Case-finding instruments for depression: interview for affective disorders and schizo- two questions are as good as many. J Gen phrenia. Arch Gen Psychiatry Intern Med 1997;12:439–445. 1978;35:837–844. (16) McCormack HM, Horne DJ, Sheather S. (12) Derogatis LR, Lipman RS, Rickels K, et al. Clinical applications of visual analogue The Hopkins Symptom Checklist (HSCL): scales: a critical review. Psychol Med a self-report symptom inventory. Behav Sci 1988;18:1007–1019. 1974;19:1–15. (17) Lambert MJ, Hatch DR, Kingston MD, et (13) Downing RW, Rickels K. Some properties al. Zung, Beck, and Hamilton rating scales of the Popoff Index. Clin Med as measures of treatment outcome: a meta- 1972;79:11–18. analytic comparison. J Consult Clin (14) Lipman RS. Differentiating anxiety and Psychol 1986;54:54–59. depression in anxiety disorders: use of (18) Hughes JR, O’Hara MW, Rehm LP. Mea- rating scales. Psychopharmacol Bull surement of depression in clinical trials: an 1982;18:69–77. overview. J Clin Psychiatry (15) Whooley MA, Avins AL, Miranda J, et al. 1982;43:85–88.  8 Mental Status Testing easures of mental status and cognitive func- and cognitive faculties in a state of unimpaired M tioning have long formed part of clinical practice, especially in geriatrics. They have also consciousness” (4). Three elements in this defi- nition of dementia hold implications for its become an established part of epidemiological measurement. First, the condition involves a de- studies and health surveys to assess the natural cline from a previously higher level of function history of cognitive decline, to study need for and so excludes, for example, those who have care and capacity for independent living, and to suffered cognitive deficits from childhood. A evaluate treatment. This growing interest reflects measurement should therefore record alter- the impact of increased longevity combined with ations in state, rather than merely current state. the sharp rise in prevalence of cognitive prob- Second, functional losses of several types are lems in old age and their impact on reducing in- implied in “global deterioration.” Although loss dependence, set against the growing interest in of memory is the central feature, memory loss is the theme of successful aging. Central to mental not unique to dementia and most operational status testing in the elderly is the assessment of definitions agree that dementia also implies lim- cognition; a useful overview is provided by the itations in other cognitive functions (5). These Merck Manual, which can be found at include aphasia (disorders of language, gener- www.merck.com/mrkshared/mm_geriatrics/sec5 ally due to left hemisphere lesions), apraxia /ch38.jsp. (disorders in performing purposeful movements, The definition of cognition remains elusive; of which constructional apraxia reflects a disor- phrases such as “the use or handling of knowl- der of visual and motor integration), and ag- edge” and the “overall functioning of mental nosia (disorders of recognition). Hence, abilities” seem unsatisfactory, and so cognitive dementia is a complex of symptoms rather than function is often defined operationally, in terms a single condition and a screening test for de- of success on cognitive tests (1). Problems of mentia should have a broad content. Third, cognition form a spectrum, beginning with mild Roth referred to “unimpaired consciousness” declines—in recall and memory, or in other ar- because symptoms of dementia may be mim- eas of functioning, such as concentration or rea- icked by reversible conditions such as intoxica- soning, or in finding the appropriate word—that tion, depression, delirium, or an acute may be a normal part of the aging process. At confusional state (4). These must be excluded the other end of the spectrum lies dementia, the before a true dementia can be diagnosed and at- most extreme form of cognitive deterioration. tributed to an irreversible cause such as cerebral Dementia afflicts 5 to 8% of all people aged infarctions or the neurodegeneration of over 65, but it is concentrated in the very old: Alzheimer’s disease. Delirium is associated with one third of those aged 85 or over have demen- an alteration in the level of consciousness that tia (2; 3). Huppert and Tym reviewed defini- leads to confusion and an inability to focus or tions of dementia (2), but Roth defined it in sustain attention. Because this affects perfor- nontechnical language as “the global deteriora- mance on cognitive tests, it must be excluded tion of an individual’s intellectual, emotional before dementia can be diagnosed. As is the case 394  Mental Status Testing 395 with assessing depression, excluding these rival ten share items in common. Most mental status explanations implies the need for a clinical as- tests include an assessment of orientation to time sessment; thus, self-administered tests may suf- and place, tests of concentration and attention, fice for screening, but they are not adequate for and memory tests for short- and long-term re- diagnosing dementia. call. Alzheimer’s disease accounts for up to two The purpose-built mental status tests typi- thirds of cases of dementia and is characterized cally draw elements from clinical neuropsycho- by abnormal brain cells containing tangles of logical instruments used to assess specific fibers (neurofibrillary tangles) and clusters of de- aspects of cognitive functioning. Most of these generating nerve endings (neuritic plaques) in the tests originated as bedside ratings intended to as- cerebral cortex (6). Alzheimer’s disease can only sess cognitive function objectively; they were be definitively diagnosed at autopsy, although it subsequently adapted for use as screening tests increasingly accepted that the correlation be- for cognitive impairment and dementia. The em- tween brain pathology and the severity of a per- phasis was on practicality, and many instru- son’s symptoms of Alzheimer’s disease is far ments we review were developed by physicians from perfect (7). Hence, clinical assessment and in reaction to difficulties experienced in adminis- neuropsychological testing are not merely seen tering full neuropsychological test batteries to el- as crude ways to assess Alzheimer’s pathology, derly patients. The focus on simplicity in but as relevant end-points in themselves. Evi- designing tests evoked criticisms, and several dence for the clinical diagnosis of Alzheimer’s themes recur in the reviews that follow. First, the disease require formal mental status testing; a tests are often narrow in scope. This may make history; physical, neurological, and psychiatric them insensitive to the early stages of cognitive examinations; laboratory tests and a computer- decline and thus unable to distinguish normal ized tomography scan (6; 8; 9). senescent decline in cognitive function from the early stages of pathological decline. Second, they may not distinguish among the more severe lev- Measurements of Cognition, els of dementia. Third, it proves exceedingly dif- Cognitive Impairment, and ficult to design a structured cognitive test that Dementia allows for fair comparison of people from differ- ent educational and cultural backgrounds. Most Tests of cognitive function may be divided into cognitive tests tend to confuse lack of education three main categories: intelligence tests, clinical with presence of cognitive impairment, so they neuropsychological tests, and laboratory tests have to be interpreted in light of the respon- (1). The present review focuses on the middle dent’s educational level. The problem of cross- category, which includes simple mental status cultural testing may be illustrated by the screening examinations and detailed tests of spe- challenge of testing long-term memory: there are cific cognitive functions. Of these, clinical neu- few historical events that can be used to test ropsychological tests provide in-depth appraisals long-term memory for people from diverse back- of particular functions such as orientation, exec- grounds, because most historical dates apply to utive function, or praxis. Formal neuropsycho- certain parts of the world more than to others. logical tests fall outside the scope of the present Recognition of famous names and faces has been review; the mental status tests that we do briefly tried but with limited success (10). review evaluate a range of cognitive functions to Because of these limitations of cognitive tests, provide a clinical overview. Mental status tests several alternatives have been tried. These in- are used to broadly assess cognitive perfor- clude self-reports of cognitive function, which mance, to screen for cognitive impairment in can be reliable, but tend not to be valid because general, or to screen for dementia. The distinc- people with cognitive deficits often cannot accu- tion between these overlapping goals is often not rately evaluate their own performance. Scores precise, and tests designed for each purpose of- seem to be more related to depression than to 396 Measuring Health the results of cognitive tests (11). Other ap- Storandt Battery (26), and the Eslinger Battery proaches include observations by clinic staff, as (27) are examples. used in the London Psychogeriatric Rating Scale (12). These may be useful in inpatient settings, but are less relevant to people living in the com- Scope of the Chapter munity. For those in the community, reports of cognitive changes made by an informant offer an The profusion of cognitive screening question- alternative to cognitive testing. Jorm reviewed a naires, diagnostic instruments, and neuropsy- selection of these methods, including the Rela- chological tests made selection for this chapter tives’ Questionnaire (13), the Geriatric Evalua- both necessary and difficult. To narrow our re- tion by Relatives Rating Instrument (GERRI) view we excluded neuropsychological tests (14), the Cognitive Behavior Rating Scale (15), such as the Benton tests or the Wechsler Adult the Present Functioning Questionnaire (PFQ) Intelligence Tests, because these instruments (16), the Memory Observation Questionnaire generally require clinical training to administer (17), the Psychogeriatric Assessment Scales and interpret. The test marketing companies (PAS) (18), and others (19). Of the various infor- (e.g., Psychological Corporation) that distrib- mant rating scales, we review the Camdex, the ute them can provide extensive information on IQCODE, and the Blessed Dementia Scale in this validity, reliability, and scoring procedures. chapter. Nor did we review the large clinical diagnostic A final option in cognitive status testing is to systems such as the Diagnostic Interview employ combinations of tests. Given that demen- Schedule (DIS), the Present State Examination tia is a syndrome with several characteristic fea- (PSE), the Geriatric Mental State Examination tures, all assessment instruments include (GMS) that was derived from the PSE, or the separate components, but few are capable of dis- Canberra Interview for the Elderly. The only criminating across all levels and types of demen- exception in this category is the Cambridge tia. Thus, tests that are useful in distinguishing Mental Disorders of the Elderly Examination mild impairment from normal cognitive func- (CAMDEX), which is included because its tioning generally are not suited to differentiating components are used as stand-alone instru- among more advanced stages of dementia. Ac- ments in epidemiological surveys and in evalu- cordingly, tests may be applied in combination. ating outcomes of care. For example, Katzman suggested using the The chapter opens with a review of several as- Blessed or short Blessed test supplemented by sessments that were designed for clinical use. the Mini-Mental State Examination (MMSE), The Mattis Dementia Rating Scale was designed which covers a broader range of functions (20). for seriously ill patients who could not complete The Consortium to Establish a Registry for standard neuropsychological test batteries. The Alzheimer’s Disease (CERAD) at Duke Univer- Cognitive Capacity Screening Examination was sity used the short Blessed Test, along with as- intended as a cognitive screening test for use sessments of insight, depression, calculation with general medical patients. The Clock Draw- ability, and language; clinical history; and the ing Test originated as a brief bedside screen but Blessed Dementia Scale (21). Shore et al., in the has seen recent application as a community National Institutes of Mental Health longitudi- screening instrument. The Alzheimer’s Disease nal study of Alzheimer’s disease, used the Assessment Scale was designed to measure be- MMSE and the MSQ, the Extended Rating Scale havioral and affective deficits in patients with for Dementia, the London Psychogeriatric Rat- Alzheimer’s disease. We then review two instru- ing Scale, and the Hachinski Scale (22). Several ments developed by Gary Blessed: the formal batteries of tests have been developed for Information-Memory-Concentration Test and assessing dementia: those proposed by Brancon- the Dementia Scale. Both were developed for use nier (23), Pfeffer’s Mental Function Index (24), with patients but can also be used in community the approach of Ferris and Crook (25), the screening. Next, we review the Mental Status  Mental Status Testing 397 Questionnaire and its revised version the Short the elderly. J Gen Intern Med Portable Mental Status Questionnaire. The for- 1990;5:55–64. mer is similar to the Blessed tests, whereas the (9) McKhann G, Drachman D, Folstein M, et latter is intended for community screening. The al. Clinical diagnosis of Alzheimer’s Mini-Mental State Examination, reviewed next, disease: report of the NINCDS-ADRDA Work Group under the auspices of is the screening test most widely used in North Department of Health and Human Services American studies; we also describe Teng’s modi- Task Force on Alzheimer’s disease. fied version, the 3MS, that appears to improve Neurology 1984;34:939–944. on the original. This is followed by a review of (10) Greene JD, Hodges JR. Identification of Jorm’s IQCODE, an informant rating of a per- famous faces and famous names in early son’s cognitive ability. Finally, we review two Alzheimer’s disease. Relationship to broader-ranging clinical assessment tools—the anterograde episodic and general semantic Clifton Assessment Procedures for the Elderly memory. Brain 1996;119:111–128. and the Cambridge Mental Disorders of the El- (11) Jorm AF. Assessment of cognitive derly Examination. A tabular comparison of impairment and dementia using informant these instruments is shown in Table 8.1. reports. Clin Psychol Rev 1996;16:51–73. (12) Hersch EL, Kral VA, Palmer RB. Clinical value of the London Psychogeriatric Rating Scale. J Am Geriatr Soc References 1978;26:348–354. (1) Colsher PL, Wallace RB. Epidemiologic (13) Sunderland A, Harris JE, Baddeley AD. Do considerations in studies of cognitive laboratory tests predict everyday memory? function in the elderly: methodology and A neuropsychological study. J Verbal Learn nondementing acquired dysfunction. Verbal Behav 1983;22:341–357. Epidemiol Rev 1991;13:1–27. (14) Schwartz GE. Development and validation (2) Huppert FA, Tym E. Clinical and of the Geriatric Evaluation by Relatives neuropsychological assessment of Rating Instrument (GERRI). Psychol Rep dementia. Br Med Bull 1986;42:11–18. 1983;53:479–488. (3) Canadian Study of Health and Aging (15) Williams JM, Klein K, Little M, Family Working Group. The Canadian Study of observations of everyday cognitive Health and Aging: study methods and impairment in dementia. Arch Clin prevalence of dementia. Can Med Assoc J Neuropsychol 1986;1:103–109. 1994;150:899–913. (16) Crockett D, Tuokko H, Koch W, et al. The (4) Roth M. The diagnosis of dementia in late assessment of everyday functioning using and middle life. In: Mortimer JA, Schuman the Present Functioning Questionnaire and LM, eds. The epidemiology of dementia. the Functional Rating Scale in elderly New York: Oxford University Press, samples. Clin Gerontol 1989;24:3–25. 1981:24–61. (17) McGlone J, Gupta S, Humphrey D, et al. (5) American Psychiatric Association. Screening for early dementia using memory Diagnostic and statistical manual of mental complaints from patients and relatives. disorders (DSM-IV). 4th ed. Washington, Arch Neurol 1990;47:1189–1193. DC: American Psychiatric Association, (18) Jorm AF, Mackinnon AJ, Henderson AS, et 1994. al. The Psychogeriatric Assessment Scales: (6) Katzman R. Alzheimer’s disease. N Engl J a multi-dimensional alternative to Med 1986;314:964–973. categorical diagnoses of dementia and (7) Snowdon DA, Kemper SJ, Mortimer JA, et depression in the elderly. Psychol Med al. Linguistic ability in early life and 1995;25:447–460. cognitive function and Alzheimer’s disease (19) Chang W-C, Chan C, Slaughter SE, et al. in late life. Findings from the Nun Study. Evaluating the FONE FIM: Part II. JAMA 1996;275:528–532. Concurrent validity & influencing factors. J (8) Ramsdell JW, Rothrock JF, Ward HW, et Outcome Meas 1997;1:259–285. al. Evaluation of cognitive impairment in (20) Katzman R. Differential diagnosis of  Table 8.1 Comparison of the Quality of Mental Status Tests* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurements Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Dementia Rating Scale ordinal 22 research expert many ** ** *** *** (Mattis, 1973) Cognitive Capacity Screening ordinal 30 clinical expert several * ** ** ** Exam (Jacobs, 1977) (5–15 min) Clock Drawing Test ordinal 1† clinical, self many ** *** ** ** (Various authors, 1986) screening Alzheimer’s Disease Assessment ordinal 21 screening expert several ** ** *** ** Scale (Rosen and Mohs, 1983) Information-Memory- ordinal 29 clinical, staff many * ** ** *** Concentration Test survey (Blessed, 1968) Dementia Scale (Blessed, 1968) ordinal 22 research interviewer several * * ** ** Mental Status Questionnaire ordinal 10 clinical, interviewer several * ** ** ** (Kahn, 1960) survey 398 Short Portable Mental Status ordinal 10 screening, interviewer many * * ** ** Questionnaire (Pfeiffer, 1975) survey (2 min) Mini-Mental Status ordinal 30 clinical, interviewer many *** ** *** *** Examination (Folstein, 1975) screening (5–15 min) Modified Mini-Mental State ordinal 34 screening, interviewer several ** *** ** *** Test (3MS) clinical (10–20 min) (Teng, 1987) Informant Questionnaire on ordinal 26 clinical, interviewer several ** *** *** *** Cognitive Decline in the Elderly screening (IQCODE) (Jorm, 1989) Clifton Assessment Procedures ordinal 12 clinical staff several ** ** ** ** for the Elderly (Pattie and Gilleard, 1975) Cambridge Mental Disorders of ordinal 57 research, expert several ** *** ** *** the Elderly Examination clinical (80 min) (CAMDEX) (Roth, 1986) * For an explanation of the categories used, see Chapter 1, pages 6–7. † The Clock test may include clock drawing, clock setting and clock reading.  Mental Status Testing 399 dementing illnesses. Neurol Clin dementia patients from those of normal cogni- 1986;4:329–340. tion but such scales are often too demanding for (21) Welsh K, Butters N, Hughes J, et al. use in discriminating among patients with vari- Detection of abnormal memory decline in ous types of dementia. He therefore developed mild cases of Alzheimer’s disease using the DRS to identify presenile and senile demen- CERAD neuropsychological measures. tias by recording behaviors that correspond to Arch Neurol 1991;48:278–281. stages in preschool-age development (2, p99). (22) Shore D, Overman CA, Wyatt RJ. Improving accuracy in the diagnosis of Alzheimer’s disease. J Clin Psychiatry Description 1983;44:207–212. The DRS covers attention, perseveration (verbal (23) Branconnier RJ. A computerized battery and motor), construction, conceptualization, for behavioral assessment in Alzheimer’s and memory (verbal and nonverbal). It is a rat- disease. In: Poon LW, ed. Handbook for clinical memory assessment of older adults. ing scale administered by a clinician; there are Washington, DC: American Psychological 36 items and items in each section are presented Association, 1986:189–196. in a hierarchical order starting with the most dif- (24) Pfeffer RI, Kurosaki TT, Harrah CH, et al. ficult so that after the patient passes an item it is A survey diagnostic tool for senile assumed that he will pass simpler items on that dementia. Am J Epidemiol section (an approach also used in Cole’s Hierar- 1981;114:515–527. chic Dementia Scale). Applying the test to unaf- (25) Ferris SH, Crook T. Cognitive assessment fected elderly patients takes 10 to 15 minutes; in mild to moderately severe dementia. In: for Alzheimer’s patients, 30 to 45 minutes are Crook T, Ferris S, Bartus R, eds. required (2, p99). The complete scale is too long Assessment in geriatric to present here; its content is summarized in Ex- psychopharmacology. New Canaan, Connecticut: Mark Powley, 1983:177–186. hibit 8.1 and the complete scale is shown in the (26) Storandt M, Botwinick J, Danziger WL, et Appendix to Mattis’s book chapter (2, al. Psychometric differentiation of mild pp108–121). Administration instructions are senile dementia of the Alzheimer type. available (3). Arch Neurol 1984;41:497–499. Scores range from 0 to 144. Unaffected indi- (27) Eslinger PJ, Damasio AR, Benton AL, et al. viduals over 65 years of age are expected to Neuropsychologic detection of abnormal score 140 or above; a score of 100 or below “is mental decline in older persons. JAMA often not consonant with survival over the next 1985;253:670–674. 20 months” (2, p99). Subsequent authors have used scores of 130 or more as indicating normal- ity (4); the DRS manual recommends that scores The Dementia Rating Scale below 123 be interpreted as representing impair- (Steven Mattis, 1973) ment (3). A lower threshold of 110 may, how- Purpose ever, be optimal for elderly people (5). The Dementia Rating Scale (DRS) was intended to identify cognitive deficits caused by neurolog- Reliability ical disease and was originally developed for re- A one-week test-retest reliability of 0.97 was re- search on cerebral blood flow in dementia (1). It ported in the original study of 30 Alzheimer’s was intended for use with severely affected insti- patients (1, Table 1). Reliability for subsections tutionalized patients who would not be able to ranged from 0.61 on the attention section to complete standard neuropsychological tests. 0.94 on the conceptualization section. Retest re- liability at a mean of 16 months was reported as Conceptual Basis r= 0.75 (6, Table 3). Mattis noted that instruments such as the Wech- Split-half reliability was 0.90 in a sample of sler Adult Intelligence Scale (WAIS) distinguish 25 residents of long-term care facilities (7, 400 Measuring Health Exhibit 8.1 Content of the Dementia Rating Scale Scales and items Examples of test items I. Attention Digit span Repeat three random digit strings forwards and backwards Respond to successive commands Open your mouth and close your eyes Respond to single command Stick out tongue Imitate movement Raise your right hand Counting Count A’s in matrix of letters, and in scrambled letter pattern Reading Lists of words II. Initiation and perseveration Verbal Name things a person can buy at the supermarket; naming articles of clothing; Say “Bee, Key, Gee” four times. Motor—double alternating movement Left palm up, right palm down, then switch simultaneously several times Graphomotor functions Copy four geometric figures III. Construction Geometric figures Copy five figures IV. Conceptualization Multiple choice: Similarities—verbal In what way are an apple and a banana alike? Primary inductive Name three things people eat. Differences Which does not belong: dog, cat, car? Similarities Apple, banana. Are they both animals, fruit, or green? Identities and oddities Which two figures are the same? Which one is different? Create sentence Make up a sentence using the words man and car. V. Memory Verbal recall Recall a simple sentence after distraction. Orientation Awareness of time, place, president VI. Sentence recall Repeat the simple sentence and the sentence involving man and car Verbal recognition Forced choice format: select which word in a pair you have read on previous page. Design recognition Forced choice format: select which design in a pair you saw on the previous page. Adapted from Mattis S. Mental status examination for organic mental syndrome in the elderly patient. In: Bellak L, Karasu TB, eds. Geriatric psychiatry: a handbook for psychiatrists and primary care physicians. New York: Grune & Stratton, 1976:108–21. p273). Alpha internal consistency coefficients 0.81 to 0.87 across a number of U.S. long-term were 0.95 for attention, 0.87 for perseveration, care facility samples, with one outlying value of 0.95 for conceptualization, 0.75 for memory (8, 0.74. KR-20 was 0.92 (10, p1655). Overall p748), and 0.87 for the overall score (9, p212). score alphas were 0.82 for patients with demen- Alpha for the attention subscale ranged from tia; coefficients for the scales ranged from 0.44  Mental Status Testing 401 (initiation and perseveration) to 0.74 (construc- rising to 0.96 when adjusted for age and educa- tion, and conceptualization scores) (11, Table 3). tion (12, p49). The equivalent figure for the Alpha values in the Chinese version were 0.70 Mini-Mental State Exam (MMSE) was 0.92 (not for unaffected controls and 0.91 for Alzheimer’s significantly different). The optimal cutting- patients (12, p48). point was 141/142, giving a sensitivity of 85% and a specificity of 92% (12, Table 4). Differ- Validity ences in scores between three severity levels of Hofer et al. reported a five-factor solution that Alzheimer’s disease were significant at p< 0.05 partially corresponded to the structure of the (p51). test shown in Exhibit 8.1. The factors included: DRS scores correlated 0.75 with full-scale long-term memory and verbal fluency; construc- WAIS IQ scores (1, p300). Correlations with the tion; short-term memory; and initiation and sim- Mayo IQ scales included 0.75 with the Mayo ple commands (13, p403). verbal IQ score, 0.73 with the full scale IQ, and Several studies have validated the DRS 0.57 with the performance IQ (11, Table 4). against physiological measures of brain func- Correlations with Boston Naming Test scores tion. The correlation between DRS scores and were 0.35 and 0.49 in two samples differing in cerebral blood flow was 0.86 (1, Figure 3; 2, severity of dementia (8, Table 8). In a study of p99; 7, p272). A positron emission tomography patients with Parkinson’s disease, convergent (PET) study of 17 patients with Alzheimer’s dis- correlations with neuropsychological tests were ease and five controls gave correlations ranging high, including 0.54 for Attention with Digits from 0.50 to 0.69 between DRS scores and me- forward; 0.62 for Conceptualization with tabolism at various sites in the left hemisphere WAIS-R Block design; 0.85 for Conceptualiza- (14, Table 3). The correlations were similar to tion with WAIS-R Similarities, and 0.58 for those obtained for the WAIS performance IQ Memory with WMS memory (17, Table 4). The test. validity of each subscale was also tested by The DRS has been validated against clinical Marson et al., who likewise found that conver- diagnoses in several studies. At a cutting-point gent validity correlations for each scale with cri- of 123, the DRS correctly identified 62% of pa- terion measures matched to the content of that tients with clinically diagnosed dementia (15, scale exceeded correlations with measures cov- Table 1). Scores discriminated significantly be- ering the content of other subscales (18, Table tween patients with dementia, others with mild 2). Correlations with the MMSE include 0.78 impairment, and controls (11, p126). DRS (9, p211), 0.82 (19), 0.83 and 0.91 (20, p304), scores showed a clear association with a clinical 0.85 (12, Table 2), and a surprisingly low value rating of severity of Alzheimer’s disease (4, Table of 0.29 (21, p319). Nadler et al. showed that 3). Everyone in a cognitively normal control DRS total scores predicted six categories of group scored 136 or higher; 83% of the group ADL limitation (correlations 0.32 to 0.64) (22, with “mild” Alzheimer’s disease fell in the range Table 3). of 103 to 130, and 71% of the “moderate” As with most cognitive screening instru- group scored 102 or below (4, p20); DRS scores ments, the DRS shows consistent associations provided a better discrimination of severity levels with age and educational level in several studies. than did an activities of daily living scale, al- Correlations of 0.27 with education, −0.31 with though in combination the two instruments ap- age, and 0.18 with sex (women had higher proximated the accuracy of the clinical staging scores) have been reported; the variance ex- of dementia. In a study of mild cognitive impair- plained by these variables was 17% (23, Tables 2 ment, the DRS achieved amazingly good detec- and 3). A correlation of −0.32 with age and 0.31 tion rates: sensitivity of 95% at a specificity of with education were reported (11, p126). Lucas 100% (16). Using the Chinese version of the et al. reported a correlation of −0.32 with age, DRS, the area under the ROC curve was 0.93, and 0.33 with education (24, Table 2). In the 402 Measuring Health Bobholz and Brandt study, the correlation with by age and sex, and to report additional validity age was −0.31, and that with education was 0.24 information (31). To assess change over time, (9, p212). Paolo et al. reviewed the issue further. an alternative form of the DRS-2 has been devel- They reported a correlation of −0.44 with age oped and shown to produce equivalent results and 0.17 with education. The age correlation (32). was highest for the attention and conceptualiza- A Cantonese adaptation was described by tion subscales and intermediate for memory and Chan et al., validity results were already noted initiation perseveration (25, Table 2). But given earlier in this chapter (12). A Brazilian version that cognitive impairments are expected to in- showed sensitivity of 92% and specificity of crease with age, they compared sensitivity and 88% (33). specificity for a single cut-off score with an age- adjusted cut-off, showing that the adjustment produced no significant improvement in test per- Reference Standards formance (25, Table 5). Mean scores (and standard deviations) by age and education are available for a small sample of cognitively normal rehabilitation patients in the Alternative Forms United States (23, Tables 4 and 5). Change A revised version of the DRS was developed by scores in unaffected subjects over an average of the U.S. National Institute on Aging (NIA) and 1.4 years were reported by Smith et al. (11, is known as the NIA Research Mattis DRS (RM- Table 2). Norms for healthy people aged 50 to DRS). This uses the same items, but standardizes 80 by educational level were reported from Aus- the administration instructions more fully and tria (34, Table 2) whereas norms for healthy reorders the items from easiest to most difficult American adults aged 55 and older were pro- to facilitate abbreviating the interview (10). It vided by Lucas et al. (24, Tables 3 to 11). Norms was intended for use with institutionalized elder- for different age-groups of cognitively normal ly. Reliability and validity information is con- African-Americans were provided by Rilling et tained in a chapter of a book by Teresi and al. (35). Evans (26, p13). The Extended Scale for Dementia uses 14 items from the DRS, supplemented with a fur- Commentary ther nine items to assess the severity of dementia The DRS can achieve very good correct classifi- in patients with advanced disease (27). A result cation rates, even on patients with mild cognitive of 0.94 was obtained for both test-retest reliabil- impairment that is notoriously difficult to clas- ity and internal consistency (27, p350). Sensitiv- sify. It takes considerably longer than other ity was 89% at a specificity of 96%, rising to a screening scales such as the 3MS and may not be sensitivity of 97% and a specificity of 100% for practical for use in population-based studies for patients aged more than 80 years old (28, p851). this reason. It is principally a clinical or research In the group of patients younger than 65 years, instrument that may offer a useful compromise sensitivity was only 75% and specificity was between extensive methods such as the WAIS or 95% (28, Table 8). Scores correlated signifi- the Halstead-Reitan Battery and the briefer cantly with total EEG change (r= −0.77) screening instruments. The Mattis scale has seen and with CT ratings of ventricular diameter (r= moderate but consistent use over the years, and −0.58) (29, Table 2). evidence for its psychometric quality has been A slightly revised version, the DRS-2, is pub- accumulating steadily. Although many validity lished and copyrighted by Psychological Assess- studies are small, results are good and unusually ment Resources (30). Information is available diverse in that they include studies linking the from www.parinc.com. This version is reported DRS with neurological and physiological find- to be more user-friendly, to have extended norms ings. Reference standards are available, so that  Mental Status Testing 403 this scale deserves serious consideration as a mid application to cognitive assessment mea- length cognitive assessment instrument. sures. Stat Med 2000;19:1651–1683. (11) Smith GE, Ivnik RJ, Malec JF, et al. Psychometric properties of the Mattis Dementia Rating Scale. Assessment References 1994;1:123–131. (1) Coblentz M, Mattis S, Zingesser LH, et al. (12) Chan AS, Choi A, Chiu H, et al. Clinical Presenile dementia: clinical aspects and validity of the Chinese version of the evaluation of cerebrospinal fluid dynamics. Mattis Dementia Rating Scale in Arch Neurol 1973;29:299–308. differentiating dementia of Alzheimer’s (2) Mattis S. Mental status examination for type in Hong Kong. J Int Neuropsychol organic mental syndrome in the elderly Soc 2003;9:45–55. patient. In: Bellak L, Karasu TB, eds. (13) Hofer SM, Piccinin AM, Hershey D. Geriatric psychiatry: a handbook for Analysis of structure and discriminative psychiatrists and primary care physicians. power of the Mattis Dementia Rating New York: Grune and Stratton, Scale. J Clin Psychol 1996;52:395–409. 1976:77–121. (14) Chase TN, Foster NL, Fedio P, et al. (3) Mattis S. Dementia Rating Scale: Regional cortical dysfunction in professional manual. Odessa, Florida: Alzheimer’s disease as determined by Psychological Assessment Resources, 1988. positron emission tomography. Ann (4) Shay KA, Duke LW, Conboy T, et al. The Neurol 1984;15(suppl):S170–S174. clinical validity of the Mattis Dementia (15) Nelson A, Fogel BS, Faust D. Bedside Rating Scale in staging Alzheimer’s cognitive screening instruments: a critical dementia. J Geriatr Psychiatry Neurol assessment. J Nerv Ment Dis 1991;4:18–25. 1986;174:73–83. (5) Bennett A, Nadler J, Spigler M, et al. The (16) Green RC, Woodard JL, Green J. Validity Mattis Dementia Rating Scale in nursing of the Mattis Dementia Rating Scale for home octogenarians and nonagenarians: detection of cognitive impairment in the el- effects of age and education. J Geriatr derly. J Neuropsychiatry Clin Neurosci Psychiatry Neurol 1997;10:114–118. 1995;7:357–360. (6) Uhlmann RF, Larson EB, Buchner DM. (17) Brown GG, Rahill AA, Gorell JM, et al. Correlations of Mini Mental State and Validity of the Dementia Rating Scale in modified Dementia Rating Scale to mea- assessing cognitive function in Parkinson’s sures of transitional health status in disease. J Geriatr Psychiatry Neurol dementia. J Gerontol 1987;42:33–36. 1999;12:180–188. (7) Gardner R, Jr., Oliver-Muñoz S, Fisher L, (18) Marson DC, Dymek MP, Duke LW, et al. et al. Mattis Dementia Rating Scale: Subscale validity of the Mattis Dementia internal reliability study using a diffusely Rating Scale. Arch Clin Neuropsychol impaired population. J Clin Neuropsychol 1997;12:269–275. 1981;3:271–275. (19) Salmon DP, Thal LJ, Butters N, et al. (8) Vitaliano PP, Breen AR, Russo J, et al. The Longitudinal evaluation of dementia of the clinical utility of the Dementia Rating Scale Alzheimer type: a comparison of 3 for assessing Alzheimer patients. J Chronic standardized mental status examinations. Dis 1984;37:743–753. Neurology 1990;40:1225–1230. (9) Bobholz JH, Brandt J. Assessment of (20) Hohl U, Grundman M, Salmon DP, et al. cognitive impairment: relationship of the Mini-Mental State Examination and Mattis Dementia Rating Scale to the Mini-Mental Dementia Raging Scale performance differs State Examination. J Geriatr Psychiatry in Hispanic and non-Hispanic Alzheimer’s Neurol 1993;6:210–213. disease patients. J Int Neuropsychol Soc (10) Teresi JA, Kleinman M, Ocepek-Welikson 1999;5:301–307. K. Modern psychometric methods for (21) Freidl W, Schmidt R, Stronegger WJ, et al. detection of differential item functioning: Sociodemographic predictors and 404 Measuring Health concurrent validity of the Mini Mental Rating Scale: diagnosis of mild dementia in State Examination and the Mattis Alzheimer’s disease. Arq Neuropsiquiatr Dementia Rating Scale. Eur Arch 2003;61:339–345. Psychiatry Clin Neurosci (34) Schmidt R, Freidl W, Fazekas F, et al. The 1996;246:317–319. Mattis Dementia Rating Scale: normative (22) Nadler JD, Richardson ED, Malloy PF, et data for 1,001 healthy volunteers. al. The ability of the Dementia Rating Scale Neurology 1994;44:964–966. to predict everyday functioning. Arch Clin (35) Rilling LM, Lucas JA, Ivnik RJ, et al. Neuropsychol 1993;8:449–460. Mayo’s older African American normative (23) Bank AL, Yochim BP, MacNeil SE, et al. studies: norms for the Mattis Dementia Expanded normative data for the Mattis Rating Scale. Clin Neuropsychologist Dementia Rating Scale for use with urban, 2005;19:229–242. elderly medical patients. Clin Neuropsychol 2000;14:149–156. (24) Lucas JA, Ivnik RJ, Smith GE, et al. The Cognitive Capacity Screening Normative data for the Mattis Dementia Examination Rating Scale. J Clin Exp Neuropsychol (John W. Jacobs, 1977) 1998;20:536–547. (25) Paolo AM, Tröster AI, Glatt SL, et al. Purpose Influence of demographic variables on the The Cognitive Capacity Screening Examination Dementia Rating Scale. J Geriatr Psychiatry (CCSE) is a 30-item test designed to assist the Neurol 1995;8:38–41. clinician in identifying organic mental syn- (26) Teresi JA, Evans DA. Cognitive assessment dromes, particularly delirium, among medical measures for chronic care populations. In: patients (1). Teresi JA, Lawton MP, Holmes D, Ory M, eds. Measurement in elderly chronic care populations. New York: Springer, Conceptual Basis 1997:1–23. Jacobs et al. noted that busy clinicians often re- (27) Hersch EL. Development and application of strict their cognitive testing to orientation to the Extended Scale for Dementia. J Am time, place, and person; these, he argued, are the Geriatr Soc 1979;27:348–354. least sensitive indicators of organic mental syn- (28) Lau C, Wands K, Merskey H, et al. dromes. They may not distinguish between func- Sensitivity and specificity of the Extended tional and organic syndromes and they may Scale for Dementia. Arch Neurol falsely classify depressed or anxious patients. Ja- 1988;45:849–852. cobs et al. also noted that several cognitive tests (29) Merskey H, Ball MJ, Blume WT, et al. Relationships between psychological mea- discriminate against less educated patients, and surements and cerebral organic changes in that patients may resent questions that imply Alzheimer’s disease. Can J Neurol Sci that they are ignorant (1, pp40–41). The CCSE 1980;7:45–49. was based on the concept that organic mental (30) Jurica PJ, Leitten CL, Mattis S. Dementia syndromes could be identified by requiring the Rating Scale-2: professional manual. Lutz, subject to shift rapidly from one task to another, Florida: Psychological Assessment often with interposed distracting tasks. Resources, 2001. (31) Johnson-Greene D. Test review: Dementia Description Rating Scale-2 (DRS-2). Arch Clin Addressing these limitations, the CCSE assesses Neuropsychol 2004;19:145–147. judgment, mental speed, and sustained effort. (32) Schmidt KS, Mattis PJ, Adams J, Nestor P. Alternate-form reliability of the Dementia Jacobs et al. argued that rapid shifts in thinking Rating Scale-2. Arch Clin Neuropsychol are relevant to diagnosing delirium, so the order 2005;20:435–441. of items was mixed. The 30 items, shown in Ex- (33) Porto CS, Fichman HC, Caramelli P, et al. hibit 8.2, are based on previous questionnaires Brazilian version of the Mattis Dementia and are administered by a physician, psycholo-  Mental Status Testing 405 Exhibit 8.2 The Cognitive Capacity Screening Examination Instructions: Check items answered correctly. Write incorrect or unusual answers in space provided. If necessary, urge patient once to complete task. Introduction to patient: “I would like to ask you a few questions. Some you will find very easy and others may be very hard. Just do your best.” 1. What day of the week is this? ______ 15. The opposite of fast is slow. 2. What month? ______ The opposite of up is ______ 3. What day of month? ______ 16. The opposite of large is ______ 4. What year? ______ 17. The opposite of hard is ______ 5. What place is this? ______ 18. An orange and a banana are both fruits. Red and blue are both ______ 6. Repeat the numbers 8 7 2. ______ 19. A penny and a dime are both ______ 7. Say them backwards. ______ 20. What were those words I asked you 8. Repeat these numbers 6 3 7 1. ______ to remember? (HAT) ______ 9. Listen to these numbers 6 9 4. Count 21. (CAR) ______ 1 through 10 out loud, then repeat 6 9 4. (Help if needed. Then use 22. (TREE) ______ numbers 5 7 3.) ______ 23. (TWENTY-SIX) ______ 10. Listen to these numbers: 8 1 4 3. 24. Take away 7 from 100, then take Count l through 10 out loud, then away 7 from what is left and keep repeat 8 1 4 3. ______ going: 100 – 7 is ______ 11. Beginning with Sunday, say the days 25. Minus 7 ______ of the week backwards. ______ 26. Minus 7 (write down answers; check 12. 9 + 3 is ______ correct subtraction of 7) ______ 13. Add 6 (to the previous answer or 27. Minus 7 ______ “to 12”). ______ 28. Minus 7 ______ 14. Take away 5 (“from 18”). Repeat 29. Minus 7 ______ these words after me and remember them. I will ask for them later: HAT, 30. Minus 7 ______ CAR, TREE, TWENTY-SIX. ______ Total Correct (maximum score = 30) ______ Adapted from Jacobs JW, Bernhard MR, Delgado A, Strain JJ. Screening for organic mental syndromes in the medically ill. Ann Intern Med 1977;86:45. gist, or other health professional in 5 to 15 min- variables. Internal consistency alpha was 0.97 utes. An overall score counts the number of for 63 patients (3, p218). questions answered correctly; 19/20 was recom- mended as the cutting-point (1, p45). Validity Several studies have examined the sensitivity and Reliability specificity of the CCSE. For 24 patients undergo- Two psychiatrists applied the CCSE twice ing psychiatric consultation, sensitivity was within 24 hours to 50 consecutive patients. The 94% and specificity was 71% (1, Table 1). As a Pearson correlation was 0.92 (kappa 0.62) (2, demonstration of specificity, mean scores for pa- p322). In another case, there was complete tients with schizophrenia and others with de- agreement among three examiners for six sub- pression fell above the cutting-point, suggesting jects, three of whom were demented (1, p41). that these patients had not been falsely classified However, correction was not made for length of as having cognitive disorders (1, p42). As a com- time between examinations or other intervening parison, the questions were administered to the 406 Measuring Health ward doctors and nurses; reassuringly, they also that the test does identify patients with dimin- scored above 20 (1, p43). Compared with re- ished cognitive capacity arising from organic sults of a full neurological examination, sensi- syndromes or mental retardation but it also erro- tivity was 62% for all cognitive deficits at a neously includes people with low levels of intelli- specificity of 90% with a cutting-point of 19/20 gence or education or with a background of (4, Tables 1,2). Beresford et al. reported a sensi- cultural deprivation. Kaufman et al. noted that tivity of 84% and a specificity of 94% using a the CCSE detected cognitive deficits due to cutting-point of 19/20 (2, p324). In a study of Alzheimer’s disease and metabolic problems but 65 elderly hospitalized patients, the CCSE cor- did not identify deficits due to structural prob- rectly identified 100% of patients diagnosed lems such as tumors or the aftermath of stroke, with organic mental syndrome and 78% of and it often failed to detect mild dementia. stroke patients; specificity was 82% (5, Table Schwamm et al. commented that the items in the 1). In a study of 62 patients from teaching hos- CCSE (like those in the Mini-Mental State pitals, Webster et al. found a sensitivity of 49% Exam) are comparatively undemanding, so mild for detecting brain impairment and a specificity impairments may be missed; they also noted that of 90% (6, Table 1). Schwamm et al. found a patients may fail all the memory items and yet false negative rate of 53% for the CCSE com- the overall score may remain within the unim- pared with 43% for the Mini-Mental State Ex- paired range (7, pp489–490). amination (7, p486). For 66 general medical The CCSE is a broad-spectrum screening in- and surgical patients, the CCSE had a sensitivity strument that can identify cognitive impairments and specificity of 100% (3, Table 2). In screen- due to a range of conditions reasonably well. We ing for vascular dementia, sensitivity was 85% know little about its performance as a commu- and specificity 87%; equivalent results for the nity screening test; if the purpose is to screen for Functional Activities Questionnaire (FAQ) were dementia, another instrument will prove supe- 92% sensitivity and 87% specificity (8, Tables 2 rior. and 3). Foreman reported correlations of 0.63 with the Short Portable Mental Status Questionnaire, References 0.83 with the Dementia Rating Scale, and 0.88 with the Mini-Mental State Exam (3, Table 3). (1) Jacobs JW, Bernhard MR, Delgado A, et Scores on the CCSE varied by educational level al. Screening for organic mental syndromes in the medically ill. Ann Intern Med (1, pp42–43). The correlation with the FAQ was 1977;86:40–46. −0.60, whereas the correlation with education (2) Beresford TP, Holt RE, Hall RCW, et al. was 0.56 (8, Table 4). Cognitive screening at the bedside: usefulness of a structured examination. Commentary Psychosomatics 1985;26:319–324. The CCSE has been quite widely tested for a (3) Foreman MD. Reliability and validity of number of years, showing variable and some- mental status questionnaires in elderly what indifferent validity results. The variation hospitalized patients. Nurs Res may reflect the diverse populations on which it 1987;36:216–220. has been tested. Jacobs et al. were “forced to (4) Kaufman DM, Weinberger M, Strain JJ, et conclude from these findings that our stated al. Detection of cognitive deficits by a brief mental status examination: the Cognitive goals had not been fully achieved” (1, p43). The Capacity Screening Examination, a CCSE was shown not to distinguish successfully reappraisal and a review. Gen Hosp between mental retardation and organic mental Psychiatry 1979;1:247–255. syndromes. Responses were influenced by edu- (5) Omer H, Foldes J, Toby M, et al. Screening cational level and by language comprehension; for cognitive deficits in a sample of conversely, mental deficits may be obscured in hospitalized geriatric patients: a re- people of higher intelligence. Jacobs concluded evaluation of a brief mental status  Mental Status Testing 407 questionnaire. J Am Geriatr Soc whereas abnormal clock setting may reflect the 1983;31:266–268. conceptual difficulties seen in demented people (6) Webster JS, Scott RR, Nunn B, et al. A (4). Royall et al. argued that clock drawing brief neuropsychological screening could also reveal problems in executive function procedure that assesses left and right (which guides complex tasks in the presence of hemispheric function. J Clin Psychol novel or distracting cues) (5). They accordingly 1984;40:237–240. developed a clock drawing task that included (7) Schwamm LH, van Dyke C, Kiernan RJ, et al. The Neurobehavioral Cognitive Status ambiguity in the instructions. Rouleau et al. il- Examination: comparison with the lustrated a number of clocks that revealed differ- Cognitive Capacity Screening Examination ent types of cognitive deficit (6, Figures 1–4). and the Mini-Mental State Examination in a neurosurgical population. Ann Intern Description Med 1987;107:486–491. The clock task may take the form of clock draw- (8) Hershey LA, Jaffe DF, Greenough PG, et ing, clock setting, or clock reading. For clock al. Validation of cognitive and functional drawing, a predrawn circle is generally supplied assessment instruments in vascular and the respondent is asked to place the numbers dementia. Int J Psychiatry Med in the appropriate places; alternatively, a blank 1987;17:183–192. sheet is used, and the respondent is asked to draw the circle and add the numbers. The clock setting task involves drawing hour and minute The Clock Drawing Test hands to specified times, such as ten minutes to (Various authors, 1986 onward) two or quarter to eight. In clock reading, the Purpose subject is typically shown a circle without num- The ability to draw a clock face and to insert the bers, but with marks indicating the locations of numbers correctly has long been used as a test of numbers and with hands drawn in to depict a cognitive function and has recently been pro- particular time (4). posed as a screening test for dementia. Although the distorted clock drawn by some- one with severe dementia can be immediately Conceptual Basis recognized, the creation of a reliable numerical Clock drawing was introduced in the early score for the distortions is challenging. There are 1900s as an indicator of constructional apraxia. several scoring schemes, most of which consider Disorder in clock drawing ability was originally omissions of numbers, errors in their placement, attributed to focal lesions of the occipital or repetitions, and irregular spacing between the parietal lobes, as demonstrated by studies of numbers. Examples of scoring approaches in- World War I soldiers with head wounds (1, clude: p1238). More recently, disturbance of visuospa- tial skills has been shown to be an early sign of 1. Wolf-Klein et al. used a partially ordered dementia. Distortions in placing the numbers on classification of ten categories of errors, a clock face and in drawing the hands to indicate of which six appeared to correspond to specified times reveal and characterize deficits in dementia and four to cognitive normality visuospatial abilities and abstract thinking. In (7). patients with advanced dementia, the clock 2. Sunderland et al. asked six raters to place drawing task may prove easier to administer 150 completed clocks into ten ordered than verbally mediated memory tasks (2). Clock categories running from “the best repre- drawing may also identify particular types of sentation of a clock” to “the worst repre- deficit; patients with left unilateral spatial neg- sentation of a clock.” From this they lect, for example, may place all 12 numbers on derived descriptive criteria with which to the right half of the clock (3). Errors in clock rank clocks on a 1 to 10 scale; examples drawing may reflect constructional apraxia, of criteria are: “5. Crowding of numbers 408 Measuring Health at one end of the clock or reversal of num- fied ten categories of errors and counted bers” and “7. Placement of hands is sig- scores within each of these categories in- nificantly off course” (2, Table 1). They cluding, for example, omissions, rota- also provided drawings to represent the tions, and reversals of digits. They also ten scale points (2, Figure 1; 8, Figure 2). counted the recognizability of the digits, 3. Rouleau et al. addressed limitations in the number of marks not representing Sunderland’s scoring system and pro- digits, and inconsistent sizes of digits (1, posed a modification in which a maxi- p1236). They provided succinct scoring mum of two points are awarded for rules and mused about computerized drawing the clock face circle, four points scoring (1, Table 4 and p1239; 8, Figure are given for correct placement of the 1). Interestingly, they found that weight- numbers, and a maximum of four points ing errors in the quadrant of the clock for setting the hands correctly. Scoring between 9 and 12 by a factor of four im- criteria were specified (6, Table 2). proved sensitivity and specificity (1, 4. Shua-Hakim et al. suggested a six-point p1237). approach they termed “a simple scoring 10. Manos and Wu described the “Ten Point system.” This awards one point to each Clock Test” in which the circle is pre- of: approximate drawing of a clock face; drawn. Scoring involves rotating the presence of numbers in sequence; correct clock until 12 is at the top, then drawing spatial arrangement of numbers; pres- a vertical line through the center and the ence of clock hands; hands approxi- 12, followed by a horizontal line and two mately showing the correct time (2:45), further lines to divide the clock face into and hands depicting the exact time (9, eighths. Then, a point is given for each of Table 1). the numbers 1, 2, 4, 5, 7, 8, 10, and 11 5. Shulman et al. rank ordered the severity that falls in its correct quadrant. A fur- of errors in clock drawing along a five- ther point is given to a short hand point- point scale (10). ing to 11, and a point for a longer hand 6. Royall et al. scored the form of the circle, pointing to the 2 (no points if the relative the placement and sequence of numbers, lengths of the hands are wrong or if they the quality and accuracy of the clock point to the wrong numbers). This sys- hands, and the absence of a number of tem provides a maximum of 10 points other irrelevant details (5, p592). and is simple to remember, but Manos 7. Ishiai et al. awarded a maximum of four and Wu agree that it does not score vari- points, giving a point each for the correct ous types of potentially significant errors placement of the 3, the 6 and the 9 rela- (12, p231). tive to the 12, and a point for correctly 11. Tuokko et al. developed a combined pro- placing the remaining numbers (3). cedure that provides quantitative and 8. A score that placed greater emphasis on qualitative ratings (13). Separate scores clock setting than on drawing was pro- are derived for omissions, perseverations, posed by Ganguli et al. The respondent rotations, misplacements, distortions, receives two points for drawing a reason- substitutions, and additions; within each able circle, two for putting the numbers of these, subtypes of error are defined (4, in correctly (deleting one point for any Table 1). A transparent overlay with lines numbers in the wrong place), and four indicating zones of acceptable positions points for drawing the appropriate length for the numbers is used to enhance accu- and position of the hands (11, pp50–51). racy in judging the misplacement of 9. Watson et al. developed a scoring ap- numbers. For clock setting (maximum 3 proach that considers only the placement points), one point is awarded for the cor- of the numbers on the clock. They speci- rect placement of each hand and a fur-  Mental Status Testing 409 ther point for the correct relative lengths 85% for clock reading (4, p579). Watson et al. of the hands (4, p580). reported sensitivity at 87% and specificity at 82% (1, Table 3). Sensitivity appears low for early stage Alzheimer’s disease, but increases as Reliability the disease progresses (17). Clock drawing has Ainslie and Murden tested the inter-rater relia- been found to identify deficits in executive func- bility of various scoring methods. Reliability for tion that were missed by the Mini-Mental State the Shulman and Wolf-Klein methods was good Exam (MMSE); in this, the Watson clock scor- (kappa = 0.74 and 0.73, respectively) but was ing method proved superior to that of Sunder- only fair for the Sunderland method land (8, p861). (kappa = 0.48) (14, p250). Sunderland et al. re- Clock drawing scores have shown significant ported intraclass correlations for various sub- differences between patients with Alzheimer’s groups of patients and normals ranging from disease and controls, and correlated 0.56 with 0.62 to 0.97; the agreement for the complete the Global Deterioration Score, 0.51 with the sample was 0.98 (2, pp727–728). Tuokko et al. Blessed Dementia Rating Scale, and 0.59 with reported inter-rater reliability coefficients rang- the Short Portable Mental Status Questionnaire ing from 0.90 to 0.95 for their scoring system (2, pp727–728). A correlation of 0.30 with (4, p581); Royall et al. also reported an inter- Katzman’s Orientation-Memory-Concentration rater correlation of 0.94 (5, p591). Other inter- Test was reported (18, p143). Clock drawing did rater values for scoring clocks ranged from not correlate with other tests of unilateral spatial rho = 0.85 to 0.95 according to the identity of neglect, but scores correlated 0.75 with verbal the rater (1, Table 5), 0.81 (15) and 0.88, 0.95, Wechsler Adult Intelligence Scale scores (3, and 0.97 (12, p236). Table 1). Tuokko et al. reported correlations Test-retest reliability reported by Watson et with other neuropsychological tests, showing al. was rho = 0.76 (1, p1238). Four-day test- higher coefficients for clock setting than for retest reliability was 0.70 in Tuokko’s study (4, clock reading or drawing (13, Table 5.3). The p581). Other values include 0.89 (15) and 0.94 Shua-Haim et al. simple scoring method corre- for a one-day retest in a study of 39 stable med- lated 0.57 with MMSE scores (9). Royall et al. ical or surgical inpatients (12, p236). compared six different scoring approaches and found that they correlated 0.56 to 0.78 with the Executive Interview (EXIT) (19). In another Validity study, Royall et al. reported a correlation of Ainslie and Murden compared the sensitivity 0.82 with the MMSE, and −0.83 with the EXIT and specificity of three scoring methods for a interview (5, Table 2). Manos and Wu reported sample of 187 elderly people; sensitivity varied correlations ranging from rho 0.22 to 0.56 be- from 48 to 87% whereas specificity varied from tween clock scores and a range of neuropsycho- 93 to 54% (14, Table 2). Sensitivity for logical tests (12, Table 1); clock scores were Alzheimer’s disease was 75% at a specificity of significantly different for Alzheimer’s patients 94% (7, p733). At an arbitrary cutting-point of and controls (Table 2) and sensitivity was 76% 6, Sunderland et al. reported a sensitivity of at a specificity of 78% (12, p240). However, 78% at a specificity of 96%. Adjusting the correlations with MMSE scores were not signifi- cutting-point would not greatly improve these cant, at 0.40 and 0.30 (20, p443). figures, because some patients with Alzheimer’s disease received nearly perfect scores on the clock drawing task (2, p728). Mendez reported Reference Standards a sensitivity of 91% and a specificity of 100% Tuokko et al. provide reference standards and (16). Tuokko et al. reported sensitivity of 92% show typical clock drawing profiles for patients and specificity of 86% for clock drawing; 87 and with Alzheimer’s disease and those with depres- 97%, respectively, for clock setting; and 92 and sion (13, Chapters 3, 4). 410 Measuring Health Alternative Forms assessment distinguishing normal from abnor- The clock test has been used in various coun- mal, to establishing a numerical rating. Differ- tries; a review from Israel summarized local ent approaches give differing results; reviews of studies (21). scoring issues are contained in reports by Wat- son et al. (1) and Mendez et al. (16). Ainslie and Commentary Murden concluded that “if clock-drawing is to The clock drawing test offers a rapid screening become a screening test that can be used by pri- method that respondents find more interesting mary care physicians in a busy office practice, than the memory and arithmetic tasks reminis- future study would need to focus on devising a cent of grade-school found in other instruments. standard set of instructions for the patient and The clock test is attractive; Salvador Dali’s fa- on devising an educationally neutral standard mous 1931 painting of distorted clocks entitled scoring system” (14, p252). And yet, it may not “The Persistence of Memory,” gave a surrealist be all that necessary to debate fine details of nu- representation of distorted time, memory, and merical scoring. Even though quantitative scores perception that may depict more than simply an on the clock test may not distinguish between artist’s view of the world. Clock drawing should different diagnoses, qualitative review of clock be applicable across most cultures and language drawing may succeed. Rouleau et al., for exam- groups, although it is evidently unsuitable for ple, found that their qualitative analysis of people with visual impairments. The strong sen- clocks distinguished patients with Alzheimer’s sitivity and specificity results reinforce the argu- disease from those with and Huntington’s ment that visuospatial deficits offer a useful chorea even though their numerical scoring did screening approach for dementia, although ar- not (6). guably as a supplement to other methods rather Clock drawing offers an adjunct to other than a stand-alone test. verbal tests of memory and orientation; on its There has, however, been some discussion own, it is unlikely to suffice as a screen for de- over what the clock drawing task measures. Al- mentia. Luckily, most people whom we assess though it has traditionally been seen as a visu- for cognitive impairments are still familiar with ospatial task that may reflect disease in the right analogue clocks; as children raised with digital hemisphere, it may also reflect frontal system ex- watches grow old, the clock test may have to be ecutive control functions (22). It also appears replaced by something more appropriate to the that clock drawing can suffer in the presence of times, such as troubleshooting a computer net- depression without cognitive impairments (23). work or programming phone numbers into a The correlations between clock results and cell phone. those of other screening instruments are only moderate, and the sensitivity and specificity re- sults range from modest to levels comparable References with other screening instruments. Clock draw- (1) Watson YI, Arfken CL, Birge SJ. Clock ing, like other tests, appears to be affected by ed- completion: an objective screening test for ucational level, although the extent seems to vary dementia. J Am Geriatr Soc according to the scoring system used (14, p251). 1993;41:1235–1240. Intelligence may compensate for spatial deficits (2) Sunderland T, Hill JL, Mellow AM, et al. Clock drawing in Alzheimer’s disease: a in the clock-drawing task, because with fore- novel measure of dementia severity. J Am thought, the person may first insert the 3, 6, 9, Geriatr Soc 1989;37:725–729. and 12 and then fill the other numbers around (3) Ishiai S, Sugishita M, Ichikawa T, et al. these (3). Clock-drawing test and unilateral spatial Grossly abnormal clock drawings readily neglect. Neurology 1993;43:106–110. identify dementia, but a major challenge lies in (4) Tuokko H, Hadjistavropoulos T, Miller moving from clock drawing as a simple bedside JA, et al. The clock test: a sensitive mea-  Mental Status Testing 411 sure to differentiate normal elderly from Development of scoring criteria for the those with Alzheimer disease. J Am Geriatr clock drawing task in Alzheimer’s disease. J Soc 1992;40:579–584. Am Geriatr Soc 1992;40:1095–1099. (5) Royall DR, Cordes JA, Polk M. CLOX: an (17) Lee H, Swanwick GR, Coen RF, et al. Use executive clock drawing task. J Neurol of the clock drawing task in the diagnosis Neurosurg Psychiatry 1998;64:588–594. of mild and very mild Alzheimer’s disease. (6) Rouleau I, Salmon DP, Butters N, et al. Int Psychogeriatr 1996;8:469–476. Quantitative and qualitative analyses of (18) Huntzinger JA, Rosse RB, Schwartz BL, et clock drawings in Alzheimer’s and al. Clock drawing in the screening Huntington’s disease. Brain Cogn assessment of cognitive impairment in an 1992;18:70–87. ambulatory care setting: a preliminary (7) Wolf-Klein GP, Silverstone FA, Levy AP, et report. Gen Hosp Psychiatry al. Screening for Alzheimer’s disease by 1992;14:142–144. clock drawing. J Am Geriatr Soc (19) Royall DR, Mulroy AR, Chiodo LK, et al. 1989;37:730–734. Clock drawing is sensitive to executive (8) Juby A, Tench S, Baker V. The value of control: a comparison of six methods. J clock drawing in identifying executive Gerontol 1999;54B:328–333. cognitive dysfunction in people with a (20) Manos PJ. The utility of the ten-point normal Mini-Mental State Examination. clock test as a screen for cognitive Can Med Assoc J 2002;167:859–864. impairment in general hospital patients. (9) Shua-Haim J, Koppuzha G, Gross J. A Gen Hosp Psychiatry 1997;19:439–444. simple scoring system for clock drawing in (21) Heinik J. [Clock-drawing test in dementia patients with Alzheimer’s disease. J Am due to Alzheimer’s disease]. Harefuah Geriatr Soc 1996;44:335. 1998;134:143–148. (10) Shulman K, Shedletsky R, Silver IL. The (22) Royall DR. Comments on the executive challenge of time: clock drawing and control of clock-drawing. J Am Geriatr Soc cognitive functioning in the elderly. Int J 1996;44:218–219. Geriatr Psychiatry 1986;1:135–140. (23) Lee H, Lawlor BA. State-dependent nature (11) Ganguli M, Ratcliff G, Huff FJ, et al. of the clock drawing task in geriatric Effects of age, gender, and education on depression. J Am Geriatr Soc cognitive tests in a rural elderly community 1995;43:796–798. sample: norms from the Monongahela Valley Independent Elders Survey. Neuroepidemiology 1991;10:42–52. (12) Manos PJ, Wu R. The ten point clock test: The Alzheimer’s Disease Assessment a quick screen and grading method for Scale cognitive impairment in medical and (Wilma G. Rosen and Richard C. surgical patients. Int J Psychiatry Med Mohs, 1983) 1994;24:229–244. (13) Tuokko H, Hadjistavropoulos T, Miller Purpose JA. The Clock Test: administration and The Alzheimer’s Disease Assessment Scale scoring manual. 1st ed. Toronto: Multi- (ADAS) is a clinical rating scale that evaluates Health Systems, Inc., 1995. the severity of cognitive, affective, and behav- (14) Ainslie NK, Murden RA. Effect of ioral deficits in patients with Alzheimer’s disease education on the clock-drawing dementia (AD) and gives an index of the overall severity of screen in non-demented elderly persons. J dementia. Am Geriatr Soc 1993;41:249–252. (15) Ploenes C, Sharp S, Martin M. The Clock Test: drawing a clock for detection of cognitive disorders in geriatric patients. Z Conceptual Basis Gerontol 1994;27:246–252. Rosen et al. classified the major clinical manifes- (16) Mendez MF, Ala T, Underwood KL. tations of AD into cognitive and noncognitive 412 Measuring Health dysfunctions. The former include memory, lan- Exhibit 8.3 Alzheimer’s Disease Assessment guage, and praxis, whereas the latter include Scale: Summary of Item Content mood state and behavioral changes (1, p1357). Scale Item range Description The ADAS assesses the clinical symptoms most Cognitive Behavior frequently reported in patients with neuropatho- 1. Spoken language ability (0–5) logically confirmed AD (2, Table 1). The items 2. Comprehension of spoken language (0–5) were drawn from existing scales or written by 3. Recall of test instructions (0–5) the authors. Initial testing shortened the ADAS 4. Word-finding difficulty (0–5) from 40 to 21 items by selecting those with the 5. Following commands (0–5) highest inter-rater and test-retest reliabilities. 6. Naming objects and fingers (0–5) Nine items assess cognitive performance through 7. Constructions—copying forms (0–5) tests done by the patient; there are two memory tasks, and ten items cover noncognitive function- 8. Ideational praxis (0–5) ing obtained from observational ratings of be- 9. Orientation to time, place, and person (0–8) havior during testing or from an interview with 10. Word recall memory test (0–10) an informant. The complete scale is too long to 11. Word recognition memory test (0–12) reproduce here, so we present a summary of the Noncognitive behavior scale content in Exhibit 8.3. The scale and scor- 12. Tearful appearance (0–5) ing procedures are shown in the Appendices to 13. Appears or reports depressed mood (0–5) the article by Rosen et al. (1, pp1361–1364), or 14. Concentration, distractibility (0–5) alternatively in the book by Sajatovic and 15. Lack of cooperation in testing (0–5) Ramirez (3, pp317–323). They did not present 16. Delusions (0–5) the actual words used in tasks related to the word recall and recognition (items 10 and 11) 17. Hallucinations (0–5) but these were listed by Zec et al., who also pro- 18. Pacing (0–5) vided further information on administration (4, 19. Increased motor activity (0–5) pp168–169). 20. Tremors (0–5) As originally described, the ADAS is a rating 21. Increased or decreased appetite (0–5) scale in which the general content of each item Adapted from Rosen WG, Mohs RC, Davis KL. A new rating (but not the precise wording) is specified; it is ad- scale for Alzheimer’s disease. Am J Psychiatry 1984;141: ministered by a neuropsychologist or psychome- 1361–1362, Appendix 2. trician who rates each item on a severity scale using the guidelines. Subsequent efforts have been made to standardize administration of the ADAS (see Alternative Forms). For each severity the normal range (5, p97). A score between 16.5 scale, 0 signifies no impairment on a task or the and 16.8 on the ADAS-Cog (based on two sepa- absence of a particular behavior. A rating of 5 rate analyses) has been shown to be equivalent to reflects severe impairment or high frequency of a a Mini-Mental State Exam (MMSE) score of 23 behavior (1, p1362). The ADAS takes about 30 (6). minutes to administer (5, p97). Scores are gener- ally calculated separately for the cognitive sec- Reliability tion (ADAS-Cog) (range, 0–70) and for the Inter-rater reliability of the original, 40-item ver- noncognitive section (range, 0–50); these may be sion was high: intraclass correlations were 0.99 combined to give a total score ranging from 0 to for patients with dementia and 0.89 for unaf- 120. Scores on the two memory items (range, fected elderly respondents (1, Table 1). One- 0–22) are occasionally presented separately. For month test-retest reliability was rho = 0.97 for the ADAS-Cog, scores ≤ 10 may be considered in patients with AD, and 0.52 for the unaffected re-  Mental Status Testing 413 spondents among whom there was a very nar- cell loss in AD) (2, p43). In a study of 1648 row spread of scores (2, p42). Alzheimer’s patients, the correlation with the For the 21-item version, one-month test- MMSE was −0.76, and that with the Geriatric retest reliability was 0.92 for the cognitive score Evaluation by Relatives Rating Instrument (omitting the memory items), but only 0.59 for (GERRI) was 0.40 (6, Figures 1, 2). Zec et al. the noncognitive score and 0.84 for the overall also reported a correlation of −0.76 with the score. Inter-rater reliability was 0.99 for the cog- MMSE (5, Figure 3). In a small study, 6 ADAS nitive and total scores, and 0.95 for the noncog- items (e.g., remembering instructions, depres- nitive score (7, p297). Test-retest reliability was sion, following commands, pacing, restlessness, 0.93 for the ADAS-Cog, 0.98 for the Noncog, and word finding difficulties) discriminated be- and 0.96 for the total score in a clinical trial; al- tween Alzheimer’s patients and healthy controls pha values were 0.80 for both the cognitive and with 100% sensitivity and specificity (12). In a the noncognitive sections (8). study in Hong Kong, the Spearman correlation For a Cantonese version of the ADAS-Cog, with the MMSE was −0.91; that with the Clini- alpha was 0.88 for patients with AD, but only cal Dementia Rating was 0.89 (9). A correlation 0.65 for cognitively normal people; test-retest re- of −0.90 with the MMSE has been reported in a liability was 0.86 for both groups, whereas inter- study of psychiatric patients (13), whereas lower rater agreement was (rho) 0.91 and 0.65 for the values of −0.64 (ADAS total score), −0.66 (Cog) two groups (9). and −0.49 (Noncog) were obtained in an thera- peutic trial for AD (8, Table 5). Validity Compared with unaffected controls, patients A factor analysis identified three factors in the with AD were significantly more impaired on all Cog scale: mental status, verbal fluency, and of the cognitive items and the memory tasks, but praxis; there was also an important common on only three of the noncognitive items (1, Table factor. The three factors had test-retest reliabili- 2). Comparing baseline scores with results one ties of 0.83, 0.78, and 0.87, respectively; the year later, total error scores increased signifi- overall ADAS Cog had a reliability of 0.90; in- cantly for the AD group but not for controls (1, ternal consistency was 0.75 (10, Table 3). The Table 3). These analyses were replicated in an- cognitive and noncognitive scales correlated other sample (7, Table 30-2). Eight patients with 0.59 in one study (7, p297) but only 0.20 in an- AD were retested at 18 months and demon- other (11, Table 2). strated continued decline (7, p299). Zec et al. Rosen et al. assessed the criterion validity of showed that each of the cognitive items differen- the ADAS using a group of 15 patients with AD tiated among AD patients of different levels of and matched controls. The Sandoz Clinical severity (as classified by the MMSE) (4, Assessment-Geriatric score correlated 0.52 with pp170–172; 5, Figures 1 & 2); not all the the ADAS total score; the correlation with the noncognitive items, however, reflected level of nine-item cognitive scale was 0.67 and the corre- dementia (4, Figure 2). Sensitivity (again, as lation with the noncognitive scale was 0.25 (1, compared to the MMSE) was 90 to 100%, and p1359). Equivalent correlations for the Blessed specificity 98% (5, Table 3). Memory-Information Test were higher, at −0.67 Zec et al. reported no significant association for the total score, and −0.78 and −0.42, respec- between ADAS scores and educational level for a tively (1, p1359). In a study of 49 patients with sample of cognitively normal subjects (4, p173), AD, the ADAS total score correlated rho = 0.77 whereas Doraiswamy et al. found that people with the Brief Cognitive Rating Scale and −0.71 who had not graduated from high school scored with the MMSE. The equivalent correlations for significantly lower on the ADAS-Cog than those the cognitive section were 0.80 and −0.81 (11, with more education, although there was no dif- Table 2). The ADAS total score correlated −0.63 ference between high school graduates and those with the concentration of acetylcholine in cere- with higher levels of education (14, Figure 1). brospinal fluid (used as a marker of cholinergic Comparing the ADAS-Cog with the MMSE, 414 Measuring Health Burch and Andrews found the ADAS scores to Evans concluded that there is no clear evidence be slightly less influenced by education than the that the ADAS-L is superior to the MMSE, al- MMSE (correlations of −0.29, versus 0.42); the though a combination of both may be valuable two scales were correlated −0.74 (15, (20, p10). Mohs et al. suggested adding a num- pp196–197). ber of additional tasks to extend the range of The ADAS-Cog is capable of detecting symptoms, and of severity, covered by the ADAS changes in function following treatment and (21). shows a dose-response (16). Reliable change The ADAS has been translated into most Eu- scores were reported by Weyer et al., who found ropean languages, and particular attention has that a change in ADAS-Cog scores of 7 or higher been paid to Spanish versions (22–24). In addi- represented a reliable change. The threshold for tion, Greek (25), Korean (26), Slovak (27), Ice- the ADAS-Noncog was 3, whereas the threshold landic (28), Brazilian Portuguese (29), and for the total score was 8 (8, Table 4). A discus- Chinese (30) versions of the ADAS have been sion of effect size and of the sample sizes re- tested. quired for using the ADADS in clinical trials is given by Kim et al. (10, pp79–81). Reference Standards Doraiswamy et al. presented mean ADAS-Cog Alternative Forms scores for patients with AD by Global Deteriora- Because the ADAS is used as an outcome mea- tion Scale stage and also by age and educational sure in international drug trials and the original level (6, Tables 3 and 4; 31, Tables 1 and 5). scale did not provide explicit instructions, sev- eral efforts have been made to standardize it. Commentary This is well illustrated by the European Harmo- The ADAS has gained widespread use as the nization Project for Instruments in Dementia clinical rating of choice in drug trials for AD, (EURO-HARPID) (17). This involved eight Eu- where the cognitive subscale is commonly used ropean countries, and focused on tasks involving alone (4; 32). The ADAS is broad in scope; exist- word recall, word recognition, and object nam- ing validity and reliability results suggest that it ing. Validity correlations for the harmonized is valid as an indicator of severity of AD. It ap- versions with the Cambridge Mental Disorders pears to produce fewer floor effects (i.e., is more of the Elderly Examination (CAMDEX) cogni- capable of detecting further deterioration) than tion score ranged from −0.82 to −0.90; those the Blessed IMC test (33, p198). A Spanish with the MMSE ranged from −0.70 to −0.88 group under Peña-Casanova has led a systematic (17, Table 5). In Canada, Standish et al. pro- review of measures for AD (NORMACODEM) posed (but did not completely describe) a stan- (34). Their review indicates that the ADAS-Cog dardized version of the ADAS, which showed is practical to administer and to score, but that markedly reduced standard deviations for scores training is needed to administer the test. The re- in inter-rater reliability trials. The disadvantage view judged the ADAS-Cog to be more compre- is that the standardized version takes five min- hensive than other methods, although not a utes longer to administer, but the increased relia- substitute for neuropsychological testing. It is bility would greatly reduce the sample size not suitable for primary care physicians, al- required (18, Table 2). though it is a valuable screening test useful in Kincaid et al. have proposed a “Late” ver- early detection and staging of AD in the severity sion, the ADAS-L, intended for advanced demen- range of Global Dementia Stages 1 to 6. Relia- tia (19). Six items test praxis, language, and bility is adequate, but (as with most cognitive memory; six assessments by the rater cover psy- measures) age and education are associated with chiatric and cognitive impairment; four ratings the ADAS-Cog score. of cognitive impairment and interpersonal skills At the same time, the NORMACODEM re- are provided by an informant, and four ADL rat- view also concluded that the lack of standardiza- ings cover ambulation and self-care. Teresi and tion limits the comparability of study results  Mental Status Testing 415 (34). Doraiswamy et al., for example, showed Alzheimer Disease Assessment Scale: a how the variability of scores makes the Cog subtest analysis. Alzheimer Dis Assoc score unsuitable for use in short-term studies, es- Disord 1992;6:164–181. pecially for people with high cognitive function (5) Zec RF, Landreth ES, Vicari SK, et al. The (35, p180). As Standish et al. wrote: “In our ex- Alzheimer’s Disease Assessment Scale: useful for both early detection and staging perience, guidelines for administration and scor- of dementia of the Alzheimer type. ing are inadequate; they are brief, vague, and Alzheimer Dis Assoc Disord leave too much to interpretation.” (18, p712). 1992;6:89–102. This has led other groups to propose closer stan- (6) Doraiswamy PM, Bieber F, Kaiser L, et al. dardization of administration (17; 18). The Alzheimer’s Disease Assessment Scale: The original validation samples were small patterns and predictors of baseline and did not show whether the scale discriminates cognitive performance in multicenter between subtypes of dementia. Zec et al. raised a Alzheimer’s disease trials. Neurology number of concerns. They found the naming and 1997;48:1511–1517. constructional praxis items to be less sensitive to (7) Rosen WG, Mohs RC, Davis KL. mild dementia than the equivalent measures in Longitudinal changes: cognitive, other tests; this may limit the sensitivity of the behavioral, and affective patterns in Alzheimer’s disease. In: Poon LW, Crook T, ADAS to improvement (4, p177). They also Davis KL, et al., eds. Handbook for clinical noted that the cube drawing in item 7 produced memory assessment of older adults. false-positive ratings from the control group, Washington, DC: American Psychological and they suggested various improvements to the Association, 1986:294–301. test (4, pp178–180). Because the noncognitive (8) Weyer G, Erzigkeit H, Kanowski S, et al. items correlate neither with the ADAS cognitive Alzheimer’s Disease Assessment Scale: score nor with other cognitive tests, use of an reliability and validity in a multicenter overall summary score has been criticized (11, clinical trial. Int Psychogeriatr p103). Mattes has suggested that validity can be 1997;9:123–138. improved by reducing the number of words in (9) Chu LW, Chiu KC, Hui SL, et al. The the word recall task (36). reliability and validity of the Alzheimer’s The ADAS has nonetheless seen widespread Disease Assessment Scale Cognitive Subscale (ADAS-Cog) among the elderly use and overcomes some of the shortcomings of Chinese in Hong Kong. Ann Acad Med earlier mental status tests and should be consid- Singapore 2000;29:474–485. ered for use as a measure of severity and as a (10) Kim YS, Nibbelink DW, Overall JE. Factor general clinical assessment. structure and reliability of the Alzheimer’s Disease Assessment Scale in a multicenter trial with linopirdine. J Geriatr Psychiatry Neurol 1994;7:74–83. References (11) Ihl R, Frölich L, Dierks T, et al. (1) Rosen WG, Mohs RC, Davis KL. A new Differential validity of psychometric tests rating scale for Alzheimer’s disease. Am J in dementia of the Alzheimer type. Psychiatry 1984;141:1356–1364. Psychiatry Res 1992;44:93–106. (2) Mohs RC, Rosen WG, Greenwald BS, et (12) Ihl R, Brinkmeyer J, Janner M, et al. A al. Neuropathologically validated scales for comarison of ADAS and EEG in the Alzheimer’s disease. In: Crook T, Ferris S, discrimination of patients with dementia of Bartus R, eds. Assessment in geriatric the Alzheimer type from healthy controls. psychopharmacology. New Canaan, Neuropsychobiology 2000;10241:107. Connecticut: Mark Powley, 1983:37–45. (13) Burch EA, Jr., Andrews SR. Cognitive (3) Sajatovic M, Ramirez LF. Rating scales in dysfunction in psychiatric consultation mental health. Hudson, Ohio: Lexi-Comp, subgroups: use of two screening tests. Inc., 2001. South Med J 1987;80:1079–1082. (4) Zec RF, Landreth ES, Vicari SK, et al. (14) Doraiswamy PM, Krishen A, Stallone F, et 416 Measuring Health al. Cognitive performance on the (24) Manzano JM, Llorca G, Ledesma A, et al. Alzheimer’s Disease Assessment Scale: [Spanish adaptation of the Alzheimer’s effect of education. Neurology disease assessment scale (ADAS)]. Actas 1995;45:1980–1984. Luso Esp Neurol Psiquiatr Cienc Afines (15) Burch EA, Jr., Andrews SR. Comparison of 1994;22:64–70. two cognitive rating scales in medically ill (25) Tsolaki M, Fountoulakis K, Nakopoulou E, patients. Int J Psychiatry Med et al. Alzheimer’s Disease Assessment Scale: 1987;17:193–200. the validation of the scale in Greece in el- (16) Rockwood K. Size of the treatment effect derly demented patients and normal on cognition of cholinesterase inhibition in subjects. Dement Geriatr Cogn Disord Alzheimer’s disease. J Neurol Neurosurg 1997;8:273–280. Psychiatry 2004;75:1086. (26) Youn JC, Lee DY, Kim KW, et al. (17) Verhey FR, Houx P, van Lang N, et al. Development of the Korean version of Cross-national comparison and validation Alzheimer’s Disease Assessment Scale of the Alzheimer’s Disease Assessment (ADAS-K). Int J Geriatr Psychiatry Scale: results from the European 2002;17:797–803. Harmonization Project for Instruments in (27) Kolibas E, Kornikova V, Novotny V, et al. Dementia (EURO-HARPID). Int J Geriatr ADAS-cog (Alzheimer’s Disease Assessment Psychiatry 2004;19:41–50. Scale-cognitive subscale)–validation of the (18) Standish TIM, Molloy DW, Bédard M, et Slovak version. Bratisl Lek Listy al. Improved reliability of the Standardized 2000;101:598–602. Alzheimer’s Disease Assessment Scale (28) Hannesdottir K, Snaedal J. A study of the (SADAS) compared with the Alzheimer’s Alzheimer’s Disease Assessment Scale- Disease Assessment Scale (ADAS). J Am cognitive (ADAS-Cog) in an Icelandic el- Geriatr Soc 1996;44:712–716. derly population. Nord J Psychiatry (19) Kincaid MM, Harvey PD, Parrella M, et al. 2002;56:201–206. Validity and utility of the ADAS-L for mea- (29) Reynolds CR, Richmond BO. What I surement of cognitive and functional Think and Feel: a revised measure of impairment in geriatric schizophrenic Children’s Manifest Anxiety. J Abnorm inpatients. J Neuropsychiatry Clin Child Psychol 1997;25:15–20. Neurosci 1995;7:76–81. (30) Liu HC, Teng EL, Chuang YY, et al. The (20) Teresi JA, Evans DA. Cognitive assessment Alzheimer’s DIsease Assessment Scale: measures for chronic care populations. In: findings from a low education population. Teresi JA, Lawton MP, Holmes D, et al., Dement Geriatr Cogn Disord eds. Measurement in elderly chronic care 2002;13:21–26. populations. New York: Springer, (31) Doraiswamy PM, Bieber F, Kaiser L, et al. 1997:1–23. Memory, language, and praxis in (21) Mohs RC, Knopman D, Petersen RC, et al. Alzheimer’s disease: norms for outpatient Development of cognitive instruments for clinical trial populations. Psychopharmacol use in clinical trials of antidementia drugs: Bull 1997;33:123–128. additions to the Alzheimer’s Disease (32) Kluger A, Ferris SH. Scales for the Assessment Scale that broaden its scope. assessment of Alzheimer’s disease. Alzheimer Dis Assoc Disord 1997;11(suppl Psychiatr Clin North Am 2):S13–S21. 1991;14:309–326. (22) Peña-Casanova J, Aguilar J, Santacruz P, et (33) Mohs RC. The Alzheimer’s Disease al. Adaptación y normalización españolas Assessment Scale. Int Psychogeriatr de la Alzheimer’s Disease Assessment Scale 1996;8:195–203. (ADAS) (NORMACODEM). Neurologia (34) Peña-Casanova J. Alzheimer’s Disease 1997;12:69–77. Assessment Scale-Cognitive in clinical (23) Pascual LF, Saz P, Larumbe R, et al. practice. Int Psychogeriatr 1997;9(suppl [Standardization of the Alzheimer’s Disease 1):105–114. Assessment Scale in a Spanish population]. (35) Doraiswamy PM, Kaiser L, Bieber F, et al. Neurologia 1997;12:238–244. The Alzheimer’s Disease Assessment Scale:  Mental Status Testing 417 evaluation of psychometric properties and Memory Scale; and last, the Concentration Scale patterns of cognitive decline in multicenter (1). Scores on the last three are often combined to clinical trials of mild to moderate form the Blessed IMC, which we review here; Alzheimer’s disease. Alzheimer Dis Assoc sometimes only two of the three are used, as in Disord 2001;15:174–183. the Information-Concentration Test. The IMC is (36) Mattes JA. Can the sensitivity of the often referred to simply as “The Blessed Test”; it Alzheimer’s Disease Assessment Scale be is occasionally named the “Newcastle Memory, increased? Am J Geriatr Psychiatry 1997;5:258–260. Information and Concentration Test.” Finally (and as if the above were inadequately confusing), as the Information subtest covers awareness of time and place, it is sometimes called the “Orien- The Information-Memory- tation Test,” so the overall scale can also be Concentration Test called the “Orientation-Memory-Concentration (G. Blessed, 1968) test” (OMC). This review follows Blessed’s origi- Purpose nal phrase, and refers to the IMC. This also helps The Information-Memory-Concentration Test to distinguish it from Katzman’s modification (IMC) provides a quantitative estimate of the de- which is known as the OMC (see Alternative gree of intellectual and personality deterioration Forms, below). in senile dementia (1). It can be used in clinical Blessed developed the scales in the context of practice and the community. clinicopathological investigations of the links among cerebral pathology (notably the forma- Conceptual Basis tion of plaques), intellectual deterioration, and A brief historical introduction may help clarify the clinical syndrome of senile dementia. The the relationship among several tests developed by scales were designed to quantify degrees of intel- Blessed, each of which has several different lectual and personality deterioration. names. As early as 1953 Roth (see the review later in Description this chapter on the Cambridge Mental Disorders The IMC uses simple tests of orientation, remote of the Elderly Examination [CAMDEX]) and and recent memory, and concentration to iden- Hopkins published a test of orientation and infor- tify dementia and to estimate its severity. It is ap- mation, which they showed was capable of dis- plied in a clinical interview with the patient, but criminating between organic mental illness personal memory information must be obtained (including the dementias) and functional illness from a collateral source to check the accuracy of such as affective disorders and schizophrenia (2). the patient’s report. Administration takes about This was then expanded by Blessed, Tomlinson, 10 minutes; the items are shown in Exhibit 8.4. and Roth to form a broad-ranging patient assess- In this scale, a positive score is given to each ment, covering changes in activities of daily living item answered correctly, and overall scores range (ADL) and performance on a range of cognitive from 0 (complete failure) to 37 (full marks on and behavioral assessments suitable for people the battery) (1, p799). After Blessed’s original with dementia (1). The complete assessment, or work, many authors now count errors, thus re- “Blessed Test” comprises four scales. First, a set versing the direction of scoring. Using this ap- of behavioral ratings that form the Dementia proach, Katzman et al. took scores of 0 to 8 Scale, which we review next and which has sev- errors as indicating normality or minimal im- eral aliases, including the Blessed-Roth Dementia pairment; 9 to 19 errors as moderate impair- Scale, the Newcastle Dementia Scale (Blessed was ment; and 20 or more errors as indicating severe working in Newcastle-upon-Tyne in northern impairment (4, p735; see also 5, p131). Fuld England), and the Blessed Dementia Rating Scale found that people making between 0 and 10 er- (3). Second, the “Information” Scale, which mea- rors usually had good memory storage and re- sures orientation to time and place. Third, the tention, but might have some difficulty with 418 Measuring Health Exhibit 8.4 The Information-Memory- School attended 1 Concentration Test Occupation 1 Note: Scores for correct responses shown; incorrect Name of sibs responses score 0. or Name of wife 1 Name of any town where Information Test patient had worked 1 Name 1 Name of employers 1 Age 1 2. Non-personal Time (hour) 1 Date of World War I1 1 Time of day 1 Date of World War II1 1 Day of week 1 Monarch2 1 Date 1 Prime Minister3 1 Month 1 3. Name and address (5-minute recall) Season 1 Mr. John Brown, 42 West Street, Year 1 Gateshead 5 Place—Name 1 Concentration Test Street 1 Months of year backwards 2 1 0 Town 1 Counting 1–20 2 1 0 Type of place (e.g., home, hospital, etc.) 1 Counting 20–1 2 1 0 Recognition of persons (cleaner, doctor, 11 nurse, patient, relative; any two available) 2 /2 point for approximation within 3 years 2 President in U.S. version Memory Test 3 Vice President in U.S. version 1. Personal Reproduced from Blessed G, Tomlinson BE, Roth M. The as- Date of birth 1 sociation between quantitative measures of dementia and of senile change in the cerebral grey matter of elderly subjects. Place of birth 1 Br J Psychiatry 1968;144:809. With permission. retrieval that could represent “benign senescent tained a correlation of −0.59 with the count of forgetfulness”; 17 or more errors may represent senile plaques found in the brain cortex at au- “malignant memory disorder.” topsy (1, p803). Katzman et al. reported a corre- lation of 0.60 with brain plaque counts (4, Reliability p737). A study by Blessed’s group also showed Two- to four-week test-retest reliability for 36 associations with levels of choline acetyl trans- residents of long-term care facilities was 0.88, ferase in the cerebral cortex at autopsy (10). and alpha was 0.93 (6, Table 1). Two-week Katzman et al. also provide data on the pre- retest reliability for 17 patients was rho = 0.96 (7, dictive validity of IMC error scores from a five- p189) and 0.82 (8, Table 4). Thal et al. compared year follow-up of 434 subjects in New York. For the reliabilities of the IMC and Mini-Mental those with an initial error-free score, the rate of State Exam (MMSE) at intervals ranging from developing dementia was 0.63 per 100 person- one to six weeks. Results for the IMC ranged years; at four errors the risk had risen to four from 0.89 to 0.82, slightly higher than the equiv- cases of dementia per 100 person-years, whereas alent figures for the MMSE (9, Table 2). at eight or more errors, the rate was 26.5 demen- tias per 100 person-years (11, Table 1). Those Validity making five to eight errors were 2.8 times more The IMC was originally developed for use in likely to develop dementia than those with zero studies linking clinical symptoms of dementia to two errors (11, Table 3). with pathological findings. Blessed et al. ob- The IMC correlated 0.81 with the Clinical  Mental Status Testing 419 Dementia Rating (5, Table 3). It correlated 0.94 Commentary with the Mental Status Questionnaire (4, p735). Blessed’s IMC scales are among the oldest and Correlations with the MMSE include −0.83, most frequently used in assessing the severity of −0.80, −0.73, 0.71 (a version of the IMC that dementia, and they are also frequently used as used the reverse scoring) (8, Table 4), and −0.88 screening tests. They provide a good overall esti- (12, Table 2). The IMC scores correlated 0.82 mate of intellectual functioning that correlates with scores on the Dementia Rating Scale (12, well with clinical ratings of the severity of de- Table 2). mentia. A major reason for the popularity of the The sensitivity of the IMC to change was re- IMC lies in its predictive validity as measured ported by Stern et al., who showed tables esti- against subsequent autopsy findings; it can iden- mating the required sample size for studies using tify cognitive and behavioral symptoms typical the instrument (13, Table 4). of patients with Alzheimer’s disease confirmed on autopsy. The content of the scale is simple Alternative Versions and focuses on the practical tasks of daily life. For use in the United States, “Monarch” is re- The IMC was recommended by the NINCDS- placed by “President,” and “Prime Minister” by ADRDA Work Group (16, Table 1) and has been “Vice President,” presumably without claim of used in several large studies of dementia. political equivalence. “Gateshead” is normally re- Whereas other measurements we review have placed with “Chicago,” which someone (perhaps undergone many changes in content, the Blessed from the north of England: Blessed was in New- scales have the advantage that their original con- castle) presumably considered of equal salience. tent has been retained (despite the various There are various abbreviations of the IMC. changes in title). Three items (recognition of persons, name of em- The validity findings for the six-item abbrevi- ployers, and town where patient worked) have ation suggest that it may serve as a good screen- sometimes been deleted, yielding an overall score ing instrument that is briefer to administer and ranging from 0 to 33 rather than 37 (13; 7). Katz- score; all items are applicable to patients man et al. developed a six-item version called the whether in the home or in an institution, and it “Orientation-Memory-Concentration test” or the can be applied to patients with physical handi- “Short Blessed Test” (SBT). The items are: caps (e.g., blindness) (8, p926). What year is it now? References What month is it now? Repeat this phrase after me: John Brown, 42 (1) Blessed G, Tomlinson BE, Roth M. The Market Street, Chicago. association between quantitative measures of dementia and of senile change in the ce- About what time is it? rebral grey matter of elderly subjects. Br J Count backwards 20 to 1. Psychiatry 1968;114:797–811. Say the months in reverse order. (2) Roth M, Hopkins B. Psychological test per- Repeat the memory phrase. (4, p739). formance of patients over 60: senile psychosis and affective disorders. J Ment The SBT correlated 0.94 with the complete test Sci 1953;101:281–301. (4, p735) and 0.79 with the Clinical Dementia (3) Blessed G, Tomlinson BE, Roth M. Blessed-Roth Dementia Scale (DS). Rating (5, Table 3). Correlations with the Psychopharmacol Bull 1988;24:705–708. MMSE include 0.71 and −0.73 (due to reversed (4) Katzman R, Brown T, Fuld P, et al. scoring in different versions of the OMC) (8, Validation of a Short Orientation-Memory- Table 4). Sensitivity was 88% at a specificity of Concentration Test of cognitive 94% (14, Table 3). impairment. Am J Psychiatry A Chinese version of the IMC has been de- 1983;140:734–739. scribed (15). (5) Davis PB, Morris JC, Grant E. Brief 420 Measuring Health screening tests versus clinical staging in a Shanghai dementia survey. Psychol Aging senile dementia of the Alzheimer type. J 1989;4:471–479. Am Geriatr Soc 1990;38:129–135. (16) McKhann G, Drachman D, Folstein M, et (6) Lesher EL, Whelihan WM. Reliability of al. Clinical diagnosis of Alzheimer’s mental status instruments administered to disease: report of the NINCDS-ADRDA nursing home residents. J Consult Clin Work Group under the auspices of Psychol 1986;54:726–727. Department of Health and Human Services (7) Fuld PA. Psychological testing in the Task Force on Alzheimer’s disease. differential diagnosis of the dementias. In: Neurology 1984;34:939–944. Katzman R, Terry RD, Bick KL, eds. Alzheimer’s disease: senile dementia and related disorders. New York: Raven Press, The Dementia Scale 1978:185–192. (G. Blessed, 1968) (8) Fillenbaum GG, Heyman A, Wilkinson WE, et al. Comparison of two screening Purpose tests in Alzheimer’s disease: the correlation Blessed’s Dementia Scale (BLS-D) was developed and reliability of the Mini-Mental State as a research instrument to quantify the cogni- Examination and the modified Blessed tive and behavioral symptoms typically seen in Test. Arch Neurol 1987;44:924–927. patients with dementia. It was originally used in (9) Thal LJ, Grundman M, Golden R. studies linking manifestations of dementia with Alzheimer’s disease: a correlational analysis neuropathological findings in the brain. of the Blessed Information-Memory- Concentration Test and the Mini-Mental State Exam. Neurology 1986;36:262–264. Conceptual Basis (10) Perry EK, Tomlinson BE, Blessed G, et al. The Dementia Scale forms one of four tests de- Correlation of cholinergic abnormalities veloped by Blessed; the others are the Informa- with senile plaques and mental test scores tion Scale, the Memory Scale, and the in senile dementia. Br Med J Concentration Scale, which are often combined 1978;2:1457–1459. to form the Blessed Information-Memory- (11) Katzman R, Aronson M, Fuld P, et al. Concentration Test (IMC), which we review Development of dementing illness in an 80- separately (1). The Dementia Scale is also known year-old volunteer cohort. Ann Neurol as the “Blessed-Roth Dementia Scale,” the 1989;25:317–324. “Newcastle Dementia Scale,” and the “Blessed (12) Salmon DP, Thal LJ, Butters N, et al. Dementia Rating Scale” (2). We prefer to term it Longitudinal evaluation of dementia of the the “Dementia Scale,” which distinguishes it Alzheimer type: a comparison of 3 standardized mental status examinations. from the Dementia Rating Scale developed by Neurology 1990;40:1225–1230. Mattis and which is reviewed in this chapter. (13) Stern RG, Mohs RC, Bierer LM, et al. Blessed’s scales were developed in the context Deterioration on the Blessed Test in of investigations of the links among cerebral Alzheimer’s disease: longitudinal data and pathology (notably the formation of neuritic their implications for clinical trials and plaques), intellectual deterioration, normal identification of subtypes. Psychiatry Res senescence, and the clinical syndrome of senile 1992;42:101–110. dementia. The scales were designed to quantify (14) Davous P, Lamour Y, Debrand E, et al. A the extent of intellectual and personality deterio- comparative evaluation of the short ration. Orientation Memory Concentration Test of cognitive impairment. J Neurol Neurosurg Psychiatry 1987;50:1312–1317. Description (15) Jin H, Zhang MY, Qu OY, et al. Cross- The BLS-D is a clinical rating scale with 22 items cultural studies of dementia: use of a that measure changes in performance of every- Chinese version of the Blessed-Roth day activities (eight items), self-care habits (three Information-Memory-Concentration test in items), and changes in personality, interests, and  Mental Status Testing 421 drives (11 items). Ratings are based on informa- and 76%, respectively; those for the 3MS tion from relatives or friends and concern behav- were 80% and 96%, respectively (8, Table V). ior over the preceding six months (2). The Compared with diagnoses in the American Psy- content of each rating is indicated, but precise chiatric Association’s DSM-IIIR, 3rd revised question phrasing is not indicated. edition in a study in Germany, the optimal Scores for each item are shown in Exhibit cutting-point was one or more points and the 8.5: total incapacity in an activity is rated 1 and area under the ROC curve was 0.92. The area partial, variable or intermittent incapacity is under the curve result for the Mini-Mental State awarded a half-point. Overall scores range from Examination was 0.988 (9, Table 2). 0 (normal) to 28 (extreme incapacity); a cogni- Blessed also found that the scores for deterio- tive subscale omits the personality questions ration of personality were neither sensitive nor (12–22) and has a range from 0 (normal) to 17 specific for dementia (10, p24). (severely demented). Commentary Reliability The Blessed Dementia Scale is of historical in- The inter-rater reliability for two raters was terest as an early form of an informant-based r= 0.59 (intraclass correlation = 0.30) (3, p329). instrument. It is also distinctive in focusing on functional changes accompanying dementia; Validity this may be relevant in understanding the care- The Dementia Scale covers a wide range of top- giver’s task; it covers handicaps rather than im- ics, and a factor analysis identified four factors, pairments or disabilities. The Dementia covering cognitive problems, personality change, Behavior Disturbance scale has a similar pur- apathy or withdrawal, and performance in basic pose and coverage (11). Because of its diversity self-care (4, Table 2). Stern et al. showed that the of coverage, care is needed in selecting an ap- pattern of decline over time in patients with de- propriate scoring method. The overall score mentia varied for the four factor scores in a plau- probably masks more subtle contrasts within sible fashion. For example, deficits on the the three components of the scale, and the fac- cognitive factor began early in the disease and tor scores proposed by Stern et al. appear to of- continued to accumulate, whereas deficits on the fer an improvement on the overall score (4). self-care factor appeared only 4 or 5 years after Likewise, the emotional score cannot be used as the onset of dementia (4, p11). an indicator of the severity of dementia, because Dementia Scale scores correlated 0.77 with a the emotional changes it records do not neces- count of plaques in the brains of 60 elderly pa- sarily accompany dementia, and more severely tients (1, p802). The correlation with duration impaired patients may demonstrate fewer emo- of survival was −0.40 (1, p805). The Dementia tional changes (12, p311). The reliability figures Scale correlated 0.80 with the Clinical Dementia are low, which may be due to the lack of guid- Rating; the equivalent correlation for the cog- ance over how the items calling for complex nitive subscale was 0.84 (5, Table 3). Scores judgments (e.g., impairment of emotional con- differed significantly between patients with trol) should be phrased (3, pp329–330). The Alzheimer’s disease (AD) and other forms of se- Dementia Scale may have clinical utility in esti- nile dementia and between different severities of mating the level of care required, but users AD defined pathologically (6, Tables 3, 4). Erk- should bear in mind the rather weak evidence injuntti et al. tested the BLS-D as a screening test for its reliability and validity. for dementia and concluded that it performed better if the personality items were left out (7). Sensitivity was 90% and specificity 85% in a small study of dementia. The equivalent results References for the Cambridge Mental Disorders of the El- (1) Blessed G, Tomlinson BE, Roth M. The derly Examination CAMCOG test were 90% association between quantitative measures Exhibit 8.5 The Dementia Scale (Blessed) Change in Performance of Everyday Activities 1. Inability to perform household tasks 1 1/2 0 2. Inability to cope with small sums of money 1 1/2 0 3. Inability to remember short list of items, e.g., in shopping 1 1/2 0 4. Inability to find way about indoors 1 1/2 0 5. Inability to find way about familiar streets 1 1/2 0 6. Inability to interpret surroundings (e.g. to recognize whether in hospital, or at home, to 1 1/2 0 discriminate between patients, doctors and nurses, relatives and hospital staff, etc.) 7. Inability to recall recent events (e.g. recent outings, visits of relatives or friends to hospital, etc.) 1 1/2 0 8. Tendency to dwell in the past 1 1/2 0 Changes in Habits 9. Eating: Cleanly with proper utensils 0 Messily with spoon only 2 Simple solids, e.g. biscuits 2 Has to be fed 3 10. Dressing: Unaided 0 Occasionally misplaced buttons, etc. 1 Wrong sequence, commonly forgetting items 2 Unable to dress 3 11. Complete sphincter control: 0 Occasional wet beds 1 Frequent wet beds 2 Doubly incontinent 3 Changes in Personality, Interests, Drive No change 0 12. Increased rigidity 1 13. Increased egocentricity 1 14. Impairment of regard for feelings of others 1 15. Coarsening of affect 1 16. Impairment of emotional control, e.g. increased petulance and irritability 1 17. Hilarity in inappropriate situations 1 18. Diminished emotional responsiveness 1 19. Sexual misdemeanour (appearing de novo in old age) 1 Interests retained 0 20. Hobbies relinquished 1 21. Diminished initiative or growing apathy 1 22. Purposeless hyperactivity 1 Reproduced from Blessed G, Tomlinson BE, Roth M. The association between quantitative measures of dementia and of senile change in the cerebral grey matter of elderly subjects. Br J Psychiatry 1968;144:808–809. With permission. 422  Mental Status Testing 423 of dementia and of senile change in the ce- rebral grey matter of elderly subjects. Br J The Mental Status Questionnaire Psychiatry 1968;114:797–811. (Robert L. Kahn, 1960) (2) Blessed G, Tomlinson BE, Roth M. Blessed-Roth Dementia Scale (DS). Purpose Psychopharmacol Bull 1988;24:705– The Mental Status Questionnaire (MSQ) pro- 708. vides a brief, objective, and quantitative mea- (3) Cole MG. Interrater reliability of the surement of cognitive functioning of elderly Blessed Dementia Scale. Can J Psychiatry people (1). It was intended for use with patient 1990;35:328–330. or community samples but results show it to be (4) Stern Y, Hesdorffer D, Sano M, et al. Mea- less appropriate in community studies. surement and prediction of functional capacity in Alzheimer’s disease. Neurology 1990;40:8–14. (5) Davis PB, Morris JC, Grant E. Brief Conceptual Basis screening tests versus clinical staging in No information is available. senile dementia of the Alzheimer type. J Am Geriatr Soc 1990;38:129–135. (6) Blessed G. Clinical features and Description neuropathological correlations of The ten items in the MSQ were selected as the Alzheimer type disease. In: Kay DWK, most discriminating of 31 questions taken from Burrows GD, eds. Handbook of studies on existing mental status examinations and from psychiatry and old age. Amsterdam: clinical experience. The questions cover orienta- Elsevier, 1984:133–143. tion in time and place, remote memory, and gen- (7) Erkinjuntti T, Hokkanen L, Sulkava R, et eral knowledge (see Exhibit 8.6). The MSQ is al. The Blessed dementia scale as a screening test for dementia. Int J Geriatr generally administered by an interviewer and re- Psychiatry 1988;146:267–273. quires a few minutes. Some latitude is commonly (8) Cappeliez P, Quintal M, Blouin M, et al. allowed in the precision of responses, although Les propriétés psychométriques de la this is rarely specified in reports on the scale. For version française du Modified Mini-Mental example, a latitude of three days in the item on State (3MS) avec des patients âgés suivis en today’s date is generally permitted. psychiatrie gériatrique. Can J Psychiatry 1996;41:114–121. (9) Heun R, Papassotiropoulos A, Jennssen F. The validity of psychometric instruments Exhibit 8.6 The Mental Status Questionnaire for detection of dementia in the elderly general population. Int J Geriatr Psychiatry 1. What is the name of this place? 1998;13:368–380. 2. Where is it located (address)? (10) Blessed G. A clinical diagnosis of senile 3. What is today’s date? dementia of the Alzheimer type in its early 4. What is the month now? and established forms. In: Bès A, ed. Senile dementias: early detection. London: John 5. What is the year? Libbey Eurotext, 1986. 6. How old are you? (11) Baumgarten M, Becker R, Gauthier S. 7. When were you born (month)? Validity and reliability of the Dementia 8. When were you born (year)? Behavior Disturbance scale. J Am Geriatr 9. Who is the president of the United States? Soc 1990;38:221–226. (12) Kluger A, Ferris SH. Scales for the 10. Who was the president before him? assessment of Alzheimer’s disease. Adapted from Kahn RL, Goldfrab AI, Pollack M, Peck A. Psychiatr Clin North Am Brief objective measures for the determination of mental status 1991;14:309–326. in the aged. Am J Psychiatry 1960;117:326. With permission. 424 Measuring Health The score counts the number of errors, so a sion analysis, the MSQ items explained 46% of score of 0 is ideal; omissions are counted as er- the variance in the diagnosis (7, Table 4). rors. Milne et al. suggested a cutting-point of Zarit et al. showed highly significant associ- three or more errors for identifying cases of ations, adjusting for education, with a range of chronic brain syndrome (2, Table 2). Zarit et al. standard memory tests (3, Table 2). De Leon et judged one or two errors as indicating a non- al. compared MSQ scores to computed tomog- significant problem; three to five indicating raphy (CT) scan results for 43 outpatients with mild-to-moderate impairment, and six to ten as suspected senile dementia of the Alzheimer indicating moderate to severe impairment (3, type. The MSQ correlated 0.46 with a CT scan p60). Others have altered the cutting-point to rating of ventricular dilatation and 0.47 with allow one extra error for poorly educated peo- cortical sulcal prominence; the equivalent cor- ple and for people from certain ethnic groups relations for the Global Deterioration Scale (4). were higher, at 0.62 and 0.53 (8, p860). In a sample of psychogeriatric patients, the MSQ scores correlated 0.87 with a psychiatrist’s rat- Reliability ing of organicity (9, Table 1). Using a sample of Lesher and Whelihan obtained test-retest relia- 230 patients, Wilson and Brass obtained a cor- bility of 0.87 at two to four weeks; split-half re- relation of 0.82 between the MSQ and a 4- liability was 0.82, and alpha was 0.81 (5, Table point clinical rating of severity of dementia (6, 1). Wilson and Brass reported an unpublished p95). They found that the questions on aware- study in which, over four administrations of the ness of personal details and location (i.e., town, MSQ at 3-week intervals, scores changed one place, age, month born, year born) served to point or less in 75% of repeat administrations differentiate moderate from severe dementia (6, p99). whereas the other five questions separated mild from moderate (6, p98). Fillenbaum reported Validity correlations of 0.88 and 0.97 between the MSQ Kahn et al. compared the MSQ to the Face- and the Short Portable MSQ in two studies (7, Hand Test and to psychiatric evaluations of p381). A correlation of 0.60 was reported be- 1,077 institutionalized patients. The results tween the MSQ and the Dementia Rating Scale showed a strong association between MSQ (10, p344). The MSQ correlated 0.92 with the scores and diagnoses of chronic brain syndrome Isaacs and Walkey Mental Impairment Mea- (CBS). For example, of those with ten MSQ er- sure (2, Table 1). MSQ scores correlated 0.80 rors, 95% had moderate or severe CBS; 75% with the Mini-Mental State Examination, 0.45 were rated as having moderate or severe CBS with the Storandt battery, and 0.57 with with psychosis. Conversely, of those scoring no Blessed’s Information-Memory-Concentration errors, 94% were rated as having mild or no test (11, Table 4). management problems (1, pp327–328). In a community survey of 487 elderly people, Alternative Forms the MSQ was 64% sensitive and 99% specific in Readers should be cautioned that there are many detecting CBS, at a cutting-point of three (2, minor variants of the MSQ and different Table 3). Using a sample of 83 community resi- cutting-points are often used (see, for example, dents, Fillenbaum assessed the capability of the reference 4). The Orientation Scale, for example, MSQ to discriminate between those subjects is intended for patients in long-term care facili- with psychiatric diagnoses of organic brain syn- ties and replaces items difficult for people in re- drome and normal subjects. At a cutting-point stricted environments (e.g., name of the of 3/4 errors, sensitivity of the MSQ was 45% president) with people more familiar (e.g., name and specificity was 98%. At a cutting-point of of head nurse) (10, p343). When used in En- 2/3 errors, the sensitivity was 55% and speci- gland or Canada, questions 9 and 10 are altered ficity was 96% (7, Table 2). In a multiple regres- to “Prime Minister.”  Mental Status Testing 425 Whelihan et al. developed the Extended Men- their overlap, the reader should also consider the tal Status Questionnaire (EMSQ) with an addi- next review, which describes Pfeiffer’s modifica- tional 14 items (12). These cover recall of family tion of the MSQ. names, orientation to immediate environment, recall of important cultural events, and arith- References metical operations. The EMSQ was better at classifying dementia than the Philadelphia Geri- (1) Kahn RL, Goldfarb AI, Pollack M, et al. atric Center Delayed Memory Test was, cor- Brief objective measures for the rectly classifying all 38 of the patients without determination of mental status in the aged. Am J Psychiatry 1960;117:326–328. dementia and 18 of 20 cases of moderate or se- (2) Milne JS, Maule MM, Cormack S, et al. vere dementia. It did, however, falsely classify all The design and testing of a questionnaire 12 mildly demented patients as nondemented and examination to assess physical and (12, p575). Test-retest reliability of the EMSQ mental health in older people using a staff was 0.88, and alpha was 0.85 (5, Table 1). nurse as the observer. J Chronic Dis Pfeiffer developed the Short Portable MSQ 1972;25:385–405. (SPMSQ) in 1975 (13); like the MSQ it has ten (3) Zarit SH, Miller NE, Kahn RL. Brain items, five of which are identical to those in the function, intellectual impairment and MSQ. The scoring procedures are, however, dif- education in the aged. J Am Geriatr Soc ferent. The SPMSQ is described in the following 1978;26:58–67. review. (4) Brink TL, Capri D, De Neeve V, et al. Senile confusion: limitations of assessment by the Face-Hand Test, Mental Status Commentary Questionnaire, and staff ratings. J Am The MSQ is one of the first generation of brief Geriatr Soc 1978;16:380–382. rating scales for dementia. It has been widely (5) Lesher EL, Whelihan WM. Reliability of used and was influential in the design of other mental status instruments administered to scales, such as the SPMSQ dementia scale that nursing home residents. J Consult Clin was then included in the Comprehensive Assess- Psychol 1986;54:726–727. ment and Referral Evaluation (CARE) reviewed (6) Wilson LA, Brass W. Brief assessment of in Chapter 10 (14). The MSQ is chiefly suited to the mental state in geriatric domiciliary institutionalized respondent populations and practice: the usefulness of the Mental appears to work less well in office or outpatient Status Questionnaire. Age Ageing 1973;2:92–101. assessments: compared with other question- (7) Fillenbaum GG. Comparison of two brief naires, its sensitivity in community screening for tests of organic brain impairment, the dementia is low. One strength of the MSQ is the MSQ and the Short Portable MSQ. J Am low association of scores with educational at- Geriatr Soc 1980;28:381–384. tainment. Zarit showed only a modest differ- (8) de Leon MJ, Ferris SH, Blau I, et al. ence in performance by educational level for Correlations between computerised most of the MSQ items with the exception of tomographic changes and behavioural spelling, for which the differences were marked deficits in senile dementia. Lancet (3, Figure 1). Educational level does not appear 1979;2:859–860. to obscure the ability of the MSQ to identify (9) Cresswell DL, Lanyon RI. Validation of a impairment. screening battery for psychogeriatric assessment. J Gerontol 1981;36:435–440. Among the shortcomings of the MSQ are its (10) Eastwood MR, Lautenschlaeger E, Corbin lack of clear justification for item content. It has S. A comparison of clinical methods for been criticized for its omission of recent mem- assessing dementia. J Am Geriatr Soc ory, which forms a critical indicator of the early 1983;31:342–347. dementing process (12). It offers a brief assess- (11) Fillenbaum G, Heyman A, Williams K, et ment that is easy to score, suited to assessing se- al. Sensitivity and specificity of vere levels of cognitive impairment. Because of standardized screens of cognitive 426 Measuring Health impairment and dementia among elderly Exhibit 8.7 Short Portable Mental Status black and white community residents. J Questionnaire Clin Epidemiol 1990;43:651–660. (12) Whelihan WM, Lesher EL, Kleban MH, et Instructions: Ask questions 1–10 in this list and al. Mental status and memory assessment record all answers. Ask question 4A only if patient does not have a telephone. Record total number of as predictors of dementia. J Gerontol errors based on ten questions. 1984;39:572–576. (13) Pfeiffer E. A Short Portable Mental Status 1. What is the date today? Questionnaire for the assessment of (Month ______ Day ______ Year ______ ) organic brain deficit in elderly persons. J 2. What day of the week is it? Am Geriatr Soc 1975;23:433–441. 3. What is the name of this place? (14) Gurland B, Golden R, Challop J. Unidimensional and multidimensional 4. What is your telephone number? approaches to the differentiation of 4A. What is your street address? (Ask only if depression and dementia in the elderly. In: patient does not have a telephone) Corkin S, Crowdon JH, Davis KL, et al., 5. How old are you? eds. Alzheimer’s disease: a report of 6. When were you born? progress. New York: Raven Press, 7. Who is the President of the U.S. now? 1982:119–125. 8. Who was President just before him? 9. What was your mother’s maiden name? The Short Portable Mental Status 10. Subtract 3 from 20 and keep subtracting 3 Questionnaire from each new number, all the way down. (Eric Pfeiffer, 1975) Adapted from Pfeiffer E. A Short Portable Mental Status Ques- tionnaire for the assessment of organic brain deficit in elderly Purpose patients. J Am Geriatr Soc 1975;23:440. With permission. The Short Portable Mental Status Questionnaire (SPMSQ) is intended to offer a rapid screen for cognitive deficit in institutions and for community-dwelling elderly people (1, p120). It memory, orientation to surroundings, knowl- detects organic intellectual impairment and de- edge of current events, and ability to perform termines its degree (2). mathematical tasks (see Exhibit 8.7). Some MSQ questions were combined (e.g., month, Conceptual Basis day, and year today) and, to make the test The SPMSQ addressed shortcomings in the more challenging, an answer is accepted as Mental Status Questionnaire (MSQ), which was correct only if all parts of the answer are cor- not successful in identifying mild-to-moderate rect (1, p119). Administration instructions are cases and not ideal for use in community set- given by Pfeiffer (2, p441). The SPMSQ is ad- tings. Accordingly, it was designed “as a some- ministered by a clinician in approximately two what more difficult but still brief test of minutes. cognitive functioning [that] could better discrim- As with the MSQ, errors are counted, and inate among the more competent people living in unanswered items are treated as errors (3, the open setting” (1, p119). Pfeiffer sought a p382). As capacity to answer the questions brief test that covered several aspects of intellec- varies with race and education, the practice in tual functioning, was simple to administer and the United States has been to correct raw scores score, and was applicable to both community to relate results for the respondent to others of and institutionalized populations (2, p435). the same race and educational level. For this, education is classified into three levels: no more Description than grade school, no more than high school, The ten questions were drawn from the MSQ and beyond high school. Race is divided into and other tests and cover short- and long-term two categories: black and all others. An error  Mental Status Testing 427 score greater than that made by about 90% of Validity those in the same race and educational combi- The SPMSQ has been compared with clinical nation is considered to represent an impaired ratings of dementia in numerous studies. Pfeiffer state (3, p382). For white respondents with compared the SPMSQ with a clinical diagnosis some high school education, the following cri- of organic brain syndrome in institutional and teria were established: zero to two errors = in- community samples. The sensitivity differed tact functioning, three to four errors = mild markedly in the two settings: 68% at a speci- impairment, five to seven errors = moderate in- ficity of 96% in the clinic sample versus 26% tellectual impairment, and eight to ten er- (specificity 98%) in the survey sample (2, Tables rors = severe impairment. Pfeiffer took more 6,7). In a further community study, with a cut- than four errors as indicative of “significant im- point of two or more errors, sensitivity was 55% pairment” (4, p435). One more error is allowed at a specificity of 96% for identifying cases of if the respondent has only a grade school edu- organic brain syndrome (3, p383). In an East cation, and one less error is allowed for those Boston community study, sensitivity was 34% with education beyond high school. In each ed- and specificity was 94% (9, p173). This com- ucational group, one more error is allowed for pared adversely with the performance of the East blacks (2, p441). A similar approach, which Boston Memory Test (sensitivity 48%, speci- produces slightly different results, was given by ficity 95%). The sensitivity of the SPMSQ was Fillenbaum (3), who listed these error scores as particularly poor in people aged between 65 to indicative of at least moderate cognitive impair- 74 years (16%) and with those with more edu- ment (see Table 8.2). In other countries scores cation (18%) (9, Table IV). In a Finnish study, may or may not be adjusted for educational the sensitivity and specificity for detecting mod- level (5). erate to severe dementia in a community sample were 67% and 100%; for a sample of medical inpatients, the figures were 86% and 99% (5, Reliability Table 5). From a mixed institutional and com- Four-week test-retest reliabilities were 0.82 and munity sample, Smyer et al. presented data from 0.83 in two samples (2, p439); a four-week test- which a sensitivity of 85% and a specificity of retest intraclass correlations was 0.81 (6, Table 84% can be calculated (10, Table 5); compared 4). Lesher and Whelihan examined reliability of with an activities of daily living assessment of the SPMSQ on 36 residents of long-term care fa- functioning, sensitivity was 82% and specificity cilities (mean age 85 years). Test-retest reliability 83% (10, Table 3). In a mixed clinical sample of (at 2 to 4 weeks) was 0.85, and alpha was 0.83 40 neurologic and psychiatric inpatients, the (7, Table 1). SPMSQ failed to identify a large number of A low alpha coefficient of 0.37 was obtained cases of organic brain syndrome, whether de- in a study of elderly people with mental retarda- fined in terms of clinical or neuropsychological tion (8, p68). diagnosis (sensitivity 27% in both analyses) (11, Table 8.2 Cutting Points for the Short Portable Mental Status Questionnaire Educational Level Grade School High School Beyond High School White 5 4 3 Black 6 5 4 Derived from Fillenbaum GG. Multidimensional functional assessment of older adults: the Duke Older Americans Resources and Services procedures. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1988, p120. With permission. 428 Measuring Health Table 1). The positive predictive value of the and inability to maintain and alternate between SPMSQ for detecting organic brain syndrome two sequences of thought. It is not clear whether among 95 consecutive admissions to a geriatric the SPMSQ can distinguish between mild, mod- unit was 88%; the negative predictive value was erate, and severe dementia, because most valida- 78% (12, p713). tion studies combined all levels of severity (16); SPMSQ scores correlated 0.63 with a psychi- Pfeiffer’s testing did not examine his stated ob- atric diagnosis of chronic brain syndrome (13, jective of differentiating four levels of cognitive Table 3); the correlation was 0.79 with a clinical impairment. Somewhat at odds with these dementia rating, comparable with the 0.81 ob- claims of low sensitivity to mild impairment, tained for the Blessed Information-Memory- Davis et al. found a ceiling effect in which even Concentration test (14, Table 3). at moderate levels of dementia, scores on the Correlations reported between the SPMSQ SPMSQ approached their maximum (14, p133). and MSQ include 0.88 (3, p381), 0.76 (15, Even though it does not seem realistic to de- Table 4) and 0.84 (13, Table 2). The SPMSQ has mand that a brief instrument cover a wide range been compared with standard tests of cognitive of areas (Dalton et al. listed 16 or more areas performance; it correlated 0.66 with the Digit that should be assessed), the omission of a learn- Span test, 0.60 with the Bender test, and 0.57 ing task may have been critical in leading to with the Basic Life Skills Assessment (12, p713). false-negative classifications (11). There may Scores correlated 0.69 with results of the Mini- also be practical difficulties in scoring the ques- Mental State Exam (15, Table 4). tion on mother’s maiden name (difficult to ver- ify), and the question on knowing the address Reference Standards proved difficult for long-term institutionalized Pfeiffer applied the SPMSQ to a community patients (11). sample of 997 people aged 65 and older and re- As with many other scales, scores vary by ed- ported error rates for each item by educational ucational level, although this effect was less level and by racial group (2, Tables 3–5). marked in the Finnish study (5, p414). Unlike other instruments, the developers of the SPMSQ Commentary tackled this problem from the outset, although The SPMSQ has been extensively used and has the system of allowing an extra error here or been incorporated, for example, into the OARS denying one there suggests an air of improvisa- assessment (see Chapter 10). It has been tested tion. in a number of studies and shows adequate relia- The SPMSQ can serve well as a brief, perhaps bility and validity in clinical samples that include narrow, instrument for use in assessing relatively people with more advanced dementia. However, severe levels of cognitive impairment. This dif- the validity results for community samples are fers from its original intended use as a commu- markedly lower, suggesting that the SPMSQ may nity screen, for which other tests such as the not be successful in classifying people with mild Modified Mini-Mental State Examination seem levels of impairment correctly. It may also fail to superior. identify organic disorders in psychiatrically mixed populations. This may, of course, be the drawback inherent in an instrument with only References ten items. Dalton et al. argued that the test is (1) Fillenbaum GG. Multidimensional “too simple in its task requirements, and, there- functional assessment of older adults: the fore, many patients with diagnosed brain dam- Duke Older Americans Resources and Ser- age were able to perform adequately on the vices procedures. Hillsdale, New Jersey: SPMSQ” (11, p514). In their study, the cogni- Lawrence Erlbaum Associates, 1988. tive deficits missed by the SPMSQ included im- (2) Pfeiffer E. A Short Portable Mental Status paired short-term memory, impaired learning, Questionnaire for the assessment of  Mental Status Testing 429 organic brain deficit in elderly patients. J senile dementia of the Alzheimer type. J Am Geriatr Soc 1975;23:433–441. Am Geriatr Soc 1990;38:129–135. (3) Fillenbaum GG. Comparison of two brief (15) Fillenbaum G, Heyman A, Williams K, et tests of organic brain impairment, the al. Sensitivity and specificity of MSQ and the Short Portable MSQ. J Am standardized screens of cognitive Geriatr Soc 1980;28:381–384. impairment and dementia among elderly (4) Pfeiffer E, Johnson TM, Chiofolo RC. black and white community residents. J Functional assessment of elderly subjects in Clin Epidemiol 1990;43:651–660. four service settings. J Am Geriatr Soc (16) Ritchie K. The screening of cognitive 1981;29:433–437. impairment in the elderly: a critical review (5) Erkinjuntti T, Sulkava R, Wikström J, et of current methods. J Clin Epidemiol al. Short Portable Mental Status 1988;41:635–643. Questionnaire as a screening test for dementia and delirium among the elderly. J Am Geriatr Soc 1987;35:412–416. (6) Cairl RE, Pfeiffer E, Keller DM, et al. An The Mini-Mental State Examination evaluation of the reliability and validity of (Marshal Folstein, 1975) the Functional Assessment Inventory. J Am Purpose Geriatr Soc 1983;31:607–612. The Mini-Mental State Examination (MMSE) (7) Lesher EL, Whelihan WM. Reliability of was designed to give a practical clinical assess- mental status instruments administered to nursing home residents. J Consult Clin ment of change in cognitive status in geriatric in- Psychol 1986;54:726–727. patients (1). It covers the person’s orientation to (8) Shultz JM, Aman MG, Rojahn J. time and place, recall ability, short-term mem- Psychometric evaluation of a measure of ory, and arithmetic ability. It may be used as a cognitive decline in elderly people with screening test for cognitive loss or as a brief bed- mental retardation. Res Devel Disabil side cognitive assessment. It cannot be used to 1998;19:63–71. diagnose dementia (2). (9) Albert M, Smith LA, Scherr PA, et al. Use of brief cognitive tests to identify Conceptual Basis individuals in the community with Evaluating the mental state of elderly psychiatric clinically diagnosed Alzheimer’s disease. patients using formal psychological tests has be- Int J Neurosci 1991;57:167–178. come a routine part of the clinical examination. (10) Smyer MA, Hofland BF, Jonas EA. Validity study of the Short Portable Mental Status The tests are often too lengthy for such elderly Questionnaire for the elderly. J Am Geriatr subjects and are based on theories rather than on Soc 1979;27:263–269. the types of cognitive impairment that lead to (11) Dalton JE, Pederson SL, Blom BE, et al. practical difficulties in daily living (2; 3). Fol- Diagnostic errors using the Short Portable stein designed the MMSE as a clinical aid in the Mental Status Questionnaire with a mixed cognitive assessment of elderly patients. It had to clinical population. J Gerontol be practical: brief enough for routine use and 1987;42:512–514. easy to remember; “Of the many possible items I (12) Wolber G, Romaniuk M, Eastman E, et al. selected those which I could remember and ap- Validity of the Short Portable Mental Status ply without any additional equipment at the bed- Questionnaire with elderly psychiatric side. This feature it turns out, made it useful for patients. J Consult Clin Psychol 1984;52:712–713. field studies as well.” (1, p290). Folstein in- (13) Haglund RMJ, Schuckit MA. A clinical cluded orientation (“a traditional but poorly de- comparison of tests of organicity in elderly fined domain”), registration, and recall as “the patients. J Gerontol 1976;31:654–659. essential screener items.” Language comprehen- (14) Davis PB, Morris JC, Grant E. Brief sion and expression were included as being rele- screening tests versus clinical staging in vant in diagnosis and patient management, 430 Measuring Health including explaining the patient’s situation to questions not answered as errors is recommended families (1). (5). The issue of how to handle nonresponses due to illiteracy or blindness has been handled either Description by treating these as errors or by prorating the Except for the language and motor skills items, overall score, deleting such items from the nu- the content of the MMSE was derived from ex- merator and denominator. Folstein has com- isting instruments (3). The MMSE was termed mented that he administers the items without “mini” because it concentrates only on the cog- regard to the reason for failure (e.g., deafness), nitive aspects of mental function and excludes and then, after scoring, comments on possible mood and abnormal mental functions that are reasons for failure. “A basic rule of clinical med- covered, for example, in Blessed’s Dementia icine is to collect the facts or observations before Scale. It is administered by clinical or lay person- making interpretations” (1). nel after brief training and requires five to ten The cutting-point most commonly used to in- minutes for completion. dicate cognitive impairment deserving further in- The MMSE includes 11 items, covering orien- vestigation is 23/24. Some authors recommend tation to time and place; registration (repeating 24/25 to enhance sensitivity for mild dementia three objects); calculation or attention (serial sev- (6). The cutting-point is commonly modulated ens or spelling “world” backward); recall of the according to educational level because a single three objects; naming two items shown; repeti- cutting-point may miss cases among more edu- tion of a phrase; following a verbal and a written cated people and generate false-positive results command; writing a sentence; and construction among those with less education. Murden et al., (copying a diagram) (2). In a surprising move, for example, suggested that 23/24 was optimal and fully thirty years after its publication, the for people with ninth grade (or higher) education MMSE has been copyrighted by Psychological level whereas 17/18 was optimal for those with Assessment Resources (PAR) (www.parinc less (7, Table 4). Uhlmann and Larson refined .com) and permission and payment now are re- this by proposing 20/21 for those with eight or quired for its use. The content of the scale is re- nine years of schooling, 22/23 for those with ten produced in Folstein’s articles (2; 3) and to 12 years of schooling and 23/24 for those elsewhere. Two versions are available: the origi- with more education (8). A Finnish study also nal, hospital version and a Field Survey Form, proposed a sloping cutting-point across age- which is described under Alternative Forms. The groups (9). However, see the Commentary sec- version provided by PAR offers a standard word- tion for a further discussion on this issue. ing for the items, and selected items from this version are shown in Exhibit 8.8. The questions Reliability can be scored immediately by summing the Foreman reported an internal consistency alpha points assigned to each successfully completed of 0.96 in a sample of 66 elderly hospitalized pa- task; the maximum score is 30. Details of scoring tients (10, p218). Kay et al. reported an alpha of have occasioned considerable discussion. For ex- 0.68 (6, p774). ample, it was originally proposed that counting Test-retest reliability has been examined in backward by sevens could be replaced by spelling many studies; in a review of his own studies, “world” backward. Folstein has clarified that he Folstein reported that for samples of psychiatric uses the serial sevens if at all possible; it is more and neurologic patients, the test-retest reliability difficult than the spelling alternative (1, p291). “has not fallen below 0.89; inter-rater reliability The practical issues in scoring “world” back- has not fallen below 0.82” (11, p47). Results ward are considerable, as described in our review from other studies are summarized in Table 8.3, of the Modified Mini-Mental State Examination and more are cited in the review by Tombaugh test (3MS). The challenge of scoring the overlap- and McIntyre (12, Table 1). With the exception ping pentagon diagram has even been addressed of the study by Uhlmann, reliability declines as by computer digitizing and analysis (4). Treating the time lapse increases. Thal’s study compared  Mental Status Testing 431 Exhibit 8.8 Sample Items from The Mini-Mental State Examination Orientation to Time “What is the date?” Registration “Listen carefully. I am going to say three words. You say them back after I stop. Ready Here they are. . . HOUSE (pause), CAR (pause), LAKE (pause). Now repeat those words back to me.” [Repeat up to 5 times, but score only the first trial.] Naming “What is this?” [Point to a pencil or pen.] Reading “Please read this and do what it says.” [Show examinee the words on the stimulus form.] CLOSE YOUR EYES Reproduced by special permission of the Publisher, Psychological Assessment Resources, Inc., 16204 North Florida Avenue, Lutz, Florida 33549, from the Mini Mental State Examination by Marshal Folstein and Susan Folstein, Copyright 1975, 1998, 2001 by Mini Mental LLC, Inc. Published 2001 by Psychological Assessment Resources, Inc. Further reproduction is prohibited without permission of PAR, Inc. The MMSE can be purchased from PAR, Inc. by calling (813) 968-3003. the reliabilities of the Blessed IMC test and the 1.00 with a mean kappa of 0.97 (16, p90). A MMSE, showing that at all testing intervals, re- comparison between self-administration and ad- sults for the Blessed test (average 0.86) were ministration by a nurse (after a median delay of slightly higher than those for the MMSE (aver- 49 days) gave an ICC of 0.78 (17, Table 2). age 0.80) (13, Table 2). Inter-rater reliability has also been widely Validity studied. Molloy et al. reported inter-rater relia- In terms of content validity, the MMSE mea- bility of 0.69 and 0.78 (14, Table 1). In a sample sures eight of the 11 main aspects of cognitive of 15 neurological inpatients, inter-rater reliabil- status; it omits abstraction, judgment, and ap- ity gave a Pearson correlation of 0.95 and a pearance (10). Although factor analyses have Kendall coefficient of 0.63 (15, p497). In a used different types of sample and differing ver- study by O’Connor et al., five coders rated taped sions of the MMSE, they commonly identify fac- interviews with 54 general practice patients. tors relating to orientation, memory, and Kappas for individual items ranged from 0.88 to attention. A study by Jones and Gallo (22) came Table 8.3 Test-Retest Reliability for Mini-Mental State Examination Study Time Lapse Statistic* Reliability Anthony et al. (18) 24 hours r 0.85–0.90 Folstein et al. (2) 24 hours r 0.89 Dick et al. (15) 24 hours r 0.92 Molloy et al. (14) 2 weeks ICC 0.69 Fillenbaum et al. (19) 1 month r 0.89 Thal et al. (13) 6 weeks r 0.80 Mitrushina et al. (20) 1 year r 0.45 Mitrushina et al. (20) 2 years r 0.38 Uhlmann et al. (21) 16 months r 0.86 *r = Pearson correlation, ICC = intraclass correlation 432 Measuring Health close to replicating the original structure of the 0.41 with WAIS Verbal IQ and of 0.42 with Per- MMSE, identifying five factors that have been formance IQ (30, p129). Similar values were re- further replicated subsequently (23). The five ported by Jorm et al. in a community sample factors are: orientation, attention, and verbal re- (31, Table 1). It correlated 0.83 with the Wech- call (which can be subsumed under a higher- sler Adult Intelligence Scale-Revised version order factor of executive functioning), and scores on a sample of 105 Alzheimer’s disease comprehension and naming (subsumed under patients (29, p510). Jorm et al. reported correla- language-praxis) (23, Table 3). tions with CT brain scan results; MMSE scores correlated 0.23 with brain diameter (perhaps re- group differences. Folstein compared the flecting premorbid findings), −0.07 with the mean MMSE scores for different groups of pa- width of the third ventricle (correlation not sig- tients, obtaining 9.7 for patients with dementia, nificant). Correlations were −0.22 with the num- 19.0 for patients with depression and cognitive ber of infarcts in the left hemisphere and −0.13 impairment, 25.1 for those with uncomplicated (not significant) with infarcts in the right hemi- affective disorder or depression, and 27.6 for sphere; there was no association with markers of normals (2, p192). There was little variation in atrophy (31, Table 4). scores by age. Folstein also presented results in- In a small sample of 40 subjects, the MMSE dicating sensitivity to treatment (2, pp193–194). and Reisberg’s Global Deterioration Scale scores Jorm reported an effect size of 1.53 for the correlated −0.92 (32, Table 7). Correlations with MMSE when used as a screening test, based on Blessed’s Dementia Rating Scale include 0.67 to eight studies (24, p159). Christensen et al. re- 0.79 (21, Table 3). A correlation of 0.87 was re- ported a higher effect size value of 2.91 for the ported with a separate Dementia Rating Scale MMSE, but this analysis was based on studies developed by Lawson (10, Table 3). The MMSE with well-defined diagnostic groups, making dis- showed modest correlations (ranging from 0.24 crimination between them easier (25). In longi- to 0.39) with verbal tests (20, Table 2). Correla- tudinal studies there is a slow and steady decline tions with other cognitive screening tests are in scores, but there can be wide variation be- higher: the MMSE scores correlated −0.88 with tween patients in the annual change in scores. the Blessed Information-Memory-Concentration This may mean that individual change scores test and 0.82 with the Dementia Rating Scale based on a small number of observations (e.g., (33, Table 2). Correlations with the Orientation- annual examinations for people observed for less Memory-Concentration test include −0.77 (34, than three years) cannot be reliably interpreted pP71) and −0.83 (19, p925). Correcting the lat- (26). ter figure for unreliability of both tests gives an MMSE scores varied by age, education, and estimated correlation of −0.93 (19, p926). gender in a Canadian study, and there were also Tombaugh and McIntyre conclude that correla- interaction effects of age and education (27, tions with the Blessed test range from −0.66 to Table 2). −0.93 (12, p927). Several studies have compared MMSE scores with the Informant Questionnaire concurrent validity. Tierney et al. obtained on Cognitive Decline in the Elderly (IQCODE). convergent correlations between 0.5 and 0.6 be- Coefficients range from −0.37 to −0.78, with a tween MMSE subscores and corresponding neu- mean of −0.59 (35, p58). The results appear to ropsychological test results. Discriminant vary according to the balance between cogni- correlations, however, were also high, suggesting tively normal and impaired people in the sam- that the components of the MMSE lack speci- ples. ficity (28, p715). On small samples of elderly pa- tients, the MMSE correlated 0.78 with the WAIS predictive validity. Mitrushina and Satz re- Verbal IQ scale and 0.66 with the WAIS Perfor- ported that all five respondents whose score de- mance scale (29, p509). In 90 psychiatric inpa- creased by more than seven points in three years tients, the MMSE had Spearman correlations of were diagnosed with neurological deficits (20,  Mental Status Testing 433 Table 8.4 Sensitivity and Specificity of the Mini-Mental State Examination in Detecting Dementia Study Sample Cut-point Sensitivity % Specificity % Anthony et al. (18) 97 hospital inpatients 23/24 87 82 Foreman (10) 66 hospital patients 23/24 82 80 Dick et al. (15) 143 neurological inpatients 23/24 76 96 van der Cammen et al. (42) 138 geriatric outpatients 24/25 88 82 Kafonek et al. (43) 70 chronic care patients 23/24 81 83 Kay et al. (6) 274 community residents 24/25 86 81 O’Connor et al. (16) 2,302 general practice patients 23/24 86 92 O’Connor et al. (16) 2,302 general practice patients 24/25 98 89 Weston (44) 98 general practice patients 23/24 83 100 Fillenbaum et al. (37) 164 community residents 23/24 100 78 Roth et al. (45) 92 hospital patients and 23/24 94 85 community residents Gagnon et al. (46) 2,792 community residents 23/24 100 78 p540). Faustman et al., however, found that the lidity of the scale, because educational level is MMSE was of limited use in predicting the psy- also related to risk of cognitive impairment chological functioning of 90 psychiatric inpa- among elderly people. Murden et al. found that tients (30). education (but not race) was significantly re- The major focus in validating the MMSE lated to MMSE scores. At a cut-point of 23/24, has been on its sensitivity and specificity com- the MMSE had 93% sensitivity and 100% pared with clinical diagnoses. Representative specificity in the high education group, and results are shown in Table 8.4. Note that for 98% sensitivity but only 75% specificity in the most studies, the criterion was a clinical assess- low education group. In the low-education ment using DSM-III diagnoses of dementia. A group, using 17/18 as a cutoff resulted in a sen- fuller list of results is contained in the review sitivity of 81% and specificity of 100% (7, by Tombaugh and McIntyre (12, Table 2). Few p152). Fillenbaum et al. showed that specificity studies have presented validity findings in the was much lower in blacks than in whites, per- form of ROC analyses; an exception is that of haps reflecting an education bias (37). A Span- Kay et al., who showed that the MMSE per- ish study likewise showed significant differences formed very well in identifying moderate and in validity across educational levels (38). How- severe cases of dementia, but less well in identi- ever, by adjusting cutting-points for different fying mild cases (6, p779). A Chilean study educational levels it may be possible to maintain compared the MMSE and Pfeffer’s Functional validity. Uhlmann and Larson, for example, Activities Questionnaire (FAQ). Sensitivity for showed that if the MMSE cutting-point was the MMSE (at a cutting-point of 21/22) was varied by educational level, the area under the 93.6%, but specificity was only 46%; figures ROC curve remained constant at 0.95 or 0.96 for the FAQ were 89% and 71%. The combina- for all educational groups. At the optimal tion of MMSE plus FAQ improved specificity: cutting-points, sensitivity was comparable sensitivity was 94% and specificity rose to 83% (82%, 79%, and 83%) in three educational (36, Table 3). groups. The corresponding specificity figures Several studies have commented on the effect were 94%, 97%, and 100% (8, Table 1). Com- of educational level on validity, but the crucial menting on these findings, a study comparing question concerns whether this indicates inva- identical and fraternal twins suggested that 434 Measuring Health most of the association between MMSE scores ternative proved easiest for the respondent, but it and education may reflect genetically-mediated turns out that subtracting sevens is the more dif- differences in cognitive capacity rather than ficult task (48; 49), although perhaps not equally educational biases in test-taking (39). A more so for males and females (4), and maybe not sophisticated analysis of differential item func- equally so across cultures. In Spanish, for exam- tioning (DIF) by Jones and Gallo confirmed that ple, counting and spelling backward is not as there is some educational DIF in certain items familiar a children’s game as it is in English- (e.g., serial subtractions, spelling backwards, speaking cultures, so “World” backward seems writing a sentence), but that less than 2% of the to elicit poorer results in Spanish versions (50; difference in cognitive function between high 51, Figure 1). The issue has been formally ad- and low educational groups was due to this item dressed in Spanish versions using differential bias (40; 41, p554). Jones and Gallo concluded item functioning analyses (52). There is also de- “Detected DIF is not sufficient to account for bate over how the “World” backward item education group differences in overall level of should be scored (this is not so simple as may estimated cognitive ability” (p555). first appear) (53–55). Given that the MMSE is not sensitive to mild impairment, it may be ad- Alternative Forms vantageous to use the more difficult item; how- Teng and Chui’s Modified Mini-Mental State ever, some studies select the higher-scoring Examination (3MS) is described in a separate response (56, p79). Other applications have entry in this chapter. Teng et al. have subse- used both items on the basis that the combina- quently developed the Cognitive Abilities tion appears to predict the overall MMSE score Screening Instrument (CASI), which is an exten- better than either item alone, giving a 12-item sion to the 3MS (47). The modifications made test and a maximum score of 35 points (6). by Roth et al. for the Cambridge Mental Disor- Reacting to all this uncertainty, Molloy et al. ders of the Elderly Examination (CAMDEX) are proposed a Standardized MMSE, with expanded described in our review of the CAMDEX later guidelines for administration, scoring, and time in this chapter. limits. They report that the Standardized MMSE There are several variations in wording, ad- was easier to administer, and intraclass inter- ministration, and scoring of questions for the rater reliability improved from 0.69 for the MMSE. Because the instrument was developed MMSE to 0.92 for the standardized version (14, for hospital patients, the original orientation Table 1). questions referred to the name and floor of the Various authors have taken scissors to the hospital. For community residents, Folstein de- MMSE. A four-item abbreviation included ori- scribed a Field Survey Form in which the hospi- entation to time, orientation to place, recall, and tal questions are altered to “What are two main spelling “World” backward; this version had a streets nearby?”, “What floor of the building are sensitivity of 98% and a specificity of 87% com- we on?”, and “What is this address?” or “What pared with the overall MMSE (57). In a replica- is the name of this place?” (3, Appendix 4). The tion study, the four items had a sensitivity of choice of words for the recall question was orig- 95.5% and a specificity of 90.5% (58). A five- inally left up to the examiner, but subsequently item abbreviation included the items on orienta- “apple,” “penny,” and “table” have been used tion to day, month, and year, ‘World’ backward, (3), or else “shirt,” “brown,” and “honesty” and recall; adding the patient’s age in a discrimi- (12, pp922–923). Finally, the MMSE included nant function equation yielded a sensitivity of spelling the word “World” backward as an alter- 95.8% at a specificity of 83.2% (59). A seven- native to counting down from 100 by sevens, a item version used a scoring system based on re- move that gave rise to one of those occasionally gression weights and showed a sensitivity of fascinating backwaters of academic investiga- 100% and specificity ranging from 93% to tion: whole articles have addressed the choice. 100% in different age and educational groups The intention had been to accept whichever al- compared with the complete MMSE (60). A 12-  Mental Status Testing 435 item abbreviation had a sensitivity of 98% and certainly seem to reverse the normal trend of specificity of 91% compared with those of the questions being simpler for educated white peo- full scale (61). The impression is that the four- ple to answer! (78; 79) item abbreviation has the virtue of brevity and simplicity of scoring and shows good results. Reference Standards Tierney et al. found that an abbreviated version Folstein et al. presented scores from a popula- that was supplemented by ratings made by an in- tion sample in Baltimore; 4.2% of those be- formant provided higher sensitivity and speci- tween 18 and 64 years old scored 23 or less ficity than the overall MMSE (62). compared with 20.8% of those older than 65 (3, Uhlmann et al. tested a written version Table 1). Crum et al. present norms based on a among people with mild-to-moderate hearing total of 18,056 participants in U.S. community loss. They found that verbal administration did surveys; mean and median scores and percentile not bias scores for people with mild or moderate distributions are given by age and educational hearing impairments; it also provided prelimi- levels (80). Bleecker et al. presented median and nary evidence that the written and interview quartile scores by age for a small sample of MMSE are comparable (63). A version for peo- healthy people (81). Heeran et al. presented ple with visual impairments (the “MMSE- norms from 532 healthy respondents over the blind”) has been described and norms calculated age of 85; they suggested that in this age group, (64, Table 3). scores lower than 25 warrant further testing (82, A telephone-administered version is called the p1096). Holzer et al. provided norms for each Telephone Interview for Cognitive Status (TICS) item on the MMSE by age group (83, Table 3), (65). A comparison of telephone administration and for the total scores by age, sex, and educa- versus face-to-face gave a correlation of 0.85 tion (83, Tables 7 and 8). Brayne and Calloway (66, Figure 1) or 0.81 (67). provided some norms for British samples by age There are innumerable translations of the and socioeconomic status (84). Reference stan- MMSE, some including abbreviated versions. dards by age-group and educational level have Translations include French (46), Dutch (42), been derived from a nondemented Canadian Spanish (50; 68–70), Italian (71), Swedish (72), population (27, Table 3; 85, Table 3), and from Chinese (73–75), Finnish (9; 74), Korean (76), an Italian sample (86). and Icelandic (4). The Chinese version showed a sensitivity of 77% and specificity of 70%, but Commentary varying the cutting-points for different educa- In a testimonial to the MMSE, Carol Brayne tional groups improved validity (73, Table 5). A wrote “At the time it was published the authors Brazilian version showed a sensitivity of 84% at cannot have conceived how widespread its use a specificity of 60% using the 23/24 cut-point would become, indicated by the speed with (77). A Spanish adaptation called the Mini Exa- which large numbers of papers incorporating it men Cognitivo (MEC) had a sensitivity of were published . . . I think it can be seen that the 93.5% and a specificity of 82% (70, p167). MMSE has been unexpectedly and wildly suc- Several issues have been noted in developing cessful, whatever the reservations.” (87, pp286, cross-culturally equivalent versions of the 288). The MMSE is brief enough for routine MMSE. Certain items appear to show particular clinical use and can be administered in survey variability across ethnic groups and should be settings by nonprofessionals (88). Validity re- modified (68). A trivial example is the orienta- sults appear as good as, or slightly better than, tion item “What county are we in?” which dif- those of other scales. Essentially, the MMSE is fers from country to country: a county in Britain an aid to the clinician and not too much should is a very much more significant political entity be expected of it; diagnosis requires a full men- than in the United States. The MMSE was modi- tal status examination, history, and physical ex- fied for use in a West African setting; questions amination. In his delightful 1998 retrospective, such as “How long does it take maize to ripen?” Folstein wrote “the MMS is now 22 years old 436 Measuring Health and can speak for itself. It travels around the but, more important in a screening instrument, world, never sends home any money, never wins several studies (6; 44; 88) suggest that the prizes and depends on me to answer the mail and MMSE may miss many mild cases of dementia. the phone and to speak to its friends” (1, p290). This may be especially true among psychiatric Because of its centrality in the literature, the patients (30). Schwamm also found that mild MMSE has been closely examined. It is a brief and moderate impairments are missed by the and practical scale, so cannot be expected to per- MMSE, as they are by most other cognitive form perfectly in every situation. Various limita- screening instruments (94; 95). He and others tions have been identified. It may miss propose that this is because the MMSE involves impairments resulting from right hemisphere le- relatively simple tasks, has a limited number of sions and may miss mild impairments (89). It test items within each cognitive domain, and has both floor and ceiling effects. Instructions combines results of performance in different do- for administration and scoring lacked detail. mains into one overall score (88; 94; 95). Fillen- Many authors have reported that people with baum reminded us that extreme caution should low education tend to give false-positive re- be used when applying most instruments, in- sponses (12, p928). The major issue concerns cluding the MMSE, to minority groups, or whether this should be viewed as a bias in the groups with low education (37). test or as a risk factor. Kittner et al. provided a Perhaps reflecting its wide use and its appar- general discussion of adjusting cutting-scores to ent potential, various improvements have been remove the effect of education (90), whereas proposed for the MMSE; these include altered Berkman raised the challenge that, if low educa- cutting-points by age and education, differential tion is of etiological significance in dementia, weighting of items, altering the content of the then one should not adjust scores for educational MMSE, and supplementing it with other tests. attainment for fear of over adjusting (91). Both The first approach does not work well, because factors may be at work. As Tombaugh and it usually trades gains in sensitivity for losses in McIntyre noted, “the prevalent view is that edu- specificity. The differential weighting has shown cation introduces a psychometric bias leading to greater promise; for example, the scores assigned a misclassification of individuals from different to each item appear somewhat arbitrary and are educational backgrounds, and this bias should not based on evidence of their relative contribu- be corrected by employing norms stratified for tions in detecting dementia. Kay et al. showed education” (12, p928). Empirical studies that that, when item scores were standardized, alpha have compared age- and education-adjusted internal consistency rose from 0.68 to 0.80 (6, scores with unadjusted scores have not shown an p774). Magaziner et al. showed that regression advantage in terms of validity to the adjustment analyses could be used to reweight the individual (92; 93). Most norms are presented by educa- items and thereby reduce the number of items tional level. Anthony et al. investigated an alter- required to achieve equally good discrimination native approach, attempting to remove (rather (60). Both differential weighting and extending than compensate for) the bias by deleting items the MMSE were tested by the 3MS approach, that caused the false-positive responses. Unfor- which appears to improve validity compared tunately, each approach that improved speci- with the original; supplementing the MMSE ficity did so at the expense of sensitivity, and so with other tests to enhance its sensitivity to mild no simple modification significantly improved dementia remains a common recommendation, the performance of the MMSE with less edu- although there is little consensus over what such cated people (18, pp405–406). This was conso- a supplement should contain. nant with Katzman’s results in the Chinese study The MMSE forms the leading screening in- (73). strument in North America but is somewhat less The MMSE may not discriminate between popular in Europe. Although it has known moderate and severe cases of Alzheimer’s disease weaknesses, it has the great virtue of being well-  Mental Status Testing 437 understood. The diversity of efforts to improve it R, eds. Assessment in geriatric illustrate the difficulty of developing the ideal de- psychopharmacology. New Canaan, mentia screening instrument. Connecticut: Mark Powley, 1983:47–51. (12) Tombaugh TN, McIntyre NJ. The Mini- Mental State Examination: a comprehensive review. J Am Geriatr Soc References 1992;40:922–935. (1) Folstein M. Mini-Mental and son. Int J (13) Thal LJ, Grundman M, Golden R. Geriatr Psychiatry 1998;13:290–294. Alzheimer’s disease: a correlational analysis (2) Folstein MF, Folstein SE, McHugh PR. of the Blessed Information-Memory- “Mini-Mental State”: a practical method Concentration Test and the Mini-Mental for grading the cognitive state of patients State Exam. Neurology 1986;36:262–264. for the clinician. J Psychiatr Res (14) Molloy DW, Alemayehu E, Roberts R. 1975;12:189–198. Reliability of a Standardized Mini-Mental (3) Folstein M, Anthony JC, Parhad I, et al. State Examination compared with the The meaning of cognitive impairment in traditional Mini-Mental State Examination. the elderly. J Am Geriatr Soc Am J Psychiatry 1991;148:102–105. 1985;33:228–235. (15) Dick JPR, Guiloff RJ, Stewart A, et al. (4) Lindal E, Stefansson JG. Mini-Mental State Mini-Mental State Examination in Examination scores: gender and lifetime neurological patients. J Neurol Neurosurg psychiatric disorders. Psychol Rep Psychiatry 1984;47:496–499. 1993;72:631–641. (16) O’Connor DW, Pollitt PA, Hyde JB, et al. (5) Fillenbaum GG, George LK, Blazer DG. The reliability and validity of the Mini- Scoring nonresponse on the Mini-Mental Mental State in a British community State Examination. Psychol Med survey. J Psychiatr Res 1989;23:87–96. 1988;18:1021–1025. (17) Bravo G, Hébert R. Reliability of the (6) Kay DWK, Henderson AS, Scott R, et al. Modified Mini-Mental State Examination Dementia and depression among the elder- in the context of a two-phase community ly living in the Hobart community: the prevalence study. Neuroepidemiology effect of the diagnostic criteria on the 1997;16:141–148. prevalence rates. Psychol Med (18) Anthony JC, LeResche L, Niaz U, et al. 1985;15:771–788. Limits of the “Mini-Mental State” as a (7) Murden RA, McRae TD, Kaner S, et al. screening test for dementia and delirium Mini-Mental State Exam scores vary with among hospital patients. Psychol Med education in blacks and whites. J Am 1982;12:397–408. Geriatr Soc 1991;39:149–155. (19) Fillenbaum GG, Heyman A, Wilkinson (8) Uhlmann RF, Larson EB. Effect of WE, et al. Comparison of two screening education on the Mini-Mental State tests in Alzheimer’s disease: the correlation Examination as a screening test for and reliability of the Mini-Mental State dementia. J Am Geriatr Soc Examination and the modified Blessed 1991;39:876–880. Test. Arch Neurol 1987;44:924–927. (9) Ylikoski R, Erkinjuntti T, Sulkava R, et al. (20) Mitrushina M, Satz P. Reliability and Correction for age, education and other validity of the Mini-Mental State Exam in demographic variables in the use of the neurologically intact elderly. J Clin Psychol Mini Mental State Examination in Finland. 1991;47:537–543. Acta Neurol Scand 1992;85:391–396. (21) Uhlmann RF, Larson EB, Buchner DM. (10) Foreman MD. Reliability and validity of Correlations of Mini Mental State and mental status questionnaires in elderly modified Dementia Rating Scale to mea- hospitalized patients. Nurs Res sures of transitional health status in 1987;36:216–220. dementia. J Gerontol 1987;42:33–36. (11) Folstein MF. The Mini-Mental State (22) Jones RN, Gallo JJ. Dimensions of the Examination. In: Crook T, Ferris S, Bartus Mini-Mental State Examination among 438 Measuring Health community dwelling older adults. Psychol Alzheimer type: a comparison of 3 Med 2000;30:605–618. standardized mental status examinations. (23) Baños JH, Franklin LM. Factor structure Neurology 1990;40:1225–1230. of the Mini-Mental State Examination in (34) Zillmer EA, Fowler PC, Gutnick HN, adult psychiatric inpatients. Psychol Assess et al. Comparison of two cognitive 2002;14:397–400. bedside screening instruments in nursing (24) Jorm AF. methods of screening for home residents: a factor analytic study. dementia: a meta-analysis of studies J Gerontol 1990;45:P69–P74. comparing an informant questionnaire (35) Jorm AF. Assessment of cognitive with a brief cognitive test. Alzheimer Dis impairment and dementia using Assoc Disord 1997;11:158–162. informant reports. Clin Psychol Rev (25) Christensen H, Hadzi-Pavlovic D, Jacomb 1996;16:51–73. PA. The psychometric differentiation of (36) Quiroga P, Albala C, Klaasen G. dementia from normal aging: a meta- [Validation of a screening test for age analysis. Psychol Assess 1991;3:147–155. associated cognitive impairment, in Chile.] (26) Clark CM, Sheppard L, Fillenbaum GG, et Rev Med Chil 2004;132:467–478. al. Variability in annual Mini-Mental State (37) Fillenbaum G, Heyman A, Williams K, et Examination scores in patients with al. Sensitivity and specificity of probable Alzheimer disease. Arch Neurol standardized screens of cognitive 1999;56:857–862. impairment and dementia among elderly (27) Bravo G, Hébert R. Age- and education- black and white community residents. J specific reference values for the Mini- Clin Epidemiol 1990;43:651–660. Mental and Modified Mini-Mental State (38) Manubens JM, Martinez-Lage P, Examinations derived from a non- Martinex-Lage JM, et al. [Variation of demented elderly population. Int J Geriatr Mini-Mental-State examination scores due Psychiatry 1997;12:1008–1018. to age and educational level. Normalized (28) Tierney MC, Szalai JP, Snow WG, et al. data in the population over 70 years of age Domain specificity of the subtests of the in Pamplona.] Neurologia Mini-Mental State Examination. Arch 1998;13:111–119. Neurol 1997;54:713–716. (39) Pedersen NL, Reynolds CA, Gatz M. (29) Farber JF, Schmitt FA, Logue PE. Sources of covariation among Mini-Mental Predicting intellectual level from the Mini- State Examination scores, education, and Mental State Examination. J Am Geriatr cognitive abilities. J Gerontol Soc 1988;36:509–510. 1996;51:P55–P63. (30) Faustman WO, Moses JAJr, Csernansky (40) Jones RN, Gallo JJ. Education bias in the JG. Limitations of the Mini-Mental State Mini-Mental State Examination. Int Examination in predicting Psychogeriatr 2001;13:299–310. neuropsychological functioning in a (41) Jones RN, Gallo JJ. Education and sex psychiatric sample. Acta Psychiatr Scand differences in the Mini-Mental State 1990;81:126–131. Examination: effects of differential item (31) Jorm AF, Broe GA, Creasey H, et al. functioning. J Gerontol Psychol Sci Further data on the validity of the Infor- 2002;57B:P548–P558. mant Questionnaire on Cognitive Decline (42) van der Cammen TJ, van Harskamp F, in the Elderly (IQCODE). Int J Geriatr Stronks DL, et al. Value of the Mini- Psychiatry 1996;11:131–139. Mental State Examination and informants’ (32) Reisberg B, Ferris S, Anand R, et al. data for the detection of dementia in Clinical assessments of cognition in the geriatric outpatients. Psychol Rep aged. In: Shamoian LA, ed. Biology and 1992;71:1003–1009. treatment of dementia in the elderly. (43) Kafonek S, Ettinger WH, Roca R, et al. Washington, DC: American Psychiatric Instruments for screening for depression Press, 1984:16–37. and dementia in a long-term care facility. J (33) Salmon DP, Thal LJ, Butters N, et al. Am Geriatr Soc 1989;37:29–34. Longitudinal evaluation of dementia of the (44) Weston WW. Screening for dementia in a  Mental Status Testing 439 family practice. Can Fam Physician (55) Schulzer M, Calne DB, Snow B, et al. Re: a 1987;33:2495–2500. scoring error in the Mini-Mental State test. (45) Roth M, Tym E, Mountjoy CQ, et al. The authors respond. Can J Psychiatry CAMDEX: a standardised instrument for 1994;39:384–385. the diagnosis of mental disorder in the el- (56) O’Connor DW, Pollitt PA, Hyde JB, et al. derly with special reference to the early A follow-up study of dementia diagnosed detection of dementia. Br J Psychiatry in the community using the Cambridge 1986;149:698–709. Mental Disorders of the Elderly (46) Gagnon M, Letenneur L, Dartigues J-F, et Examination. Acta Psychiatr Scand al. Validity of the Mini-Mental State 1990;81:78–82. Examination as a screening instrument for (57) Paveza GJ, Cohen D, Blaser CJ, et al. A cognitive impairment and dementia in brief form of the Mini-Mental State French elderly community residents. Examination for use in community care Neuroepidemiology 1990;9:143–150. settings. Behav Health Aging (47) Teng EL, Hasegawa K, Homma A, et al. 1990;1:133–139. The Cognitive Abilities Screening (58) Koenig HG. An abbreviated Mini-mental Instrument (CASI): a practical test for State Exam for medically ill older adults. J cross-cultural epidemiological studies of Am Geriatr Soc 1996;44:235–236. dementia. Int Psychogeriatr (59) Klein LE, Roca RP, McArthur J, et al. 1994;6:45–58. Diagnosing dementia: univariate and (48) Watkins P, Gouvier WD, Callon E, et al. multivariate analyses of the Mental Status Equivalence of items on the Mini-Mental Examination. J Am Geriatr Soc State. Arch Clin Neuropsychol 1985;33:483–488. 1989;4:381–384. (60) Magaziner J, Bassett SS, Hebel JR. (49) Ganguli M, Ratcliff G, Huff FJ, et al. Serial Predicting performance on the Mini- sevens versus world backwards: a Mental State Examination: use of age- and comparison of the two measures of education-specific equations. J Am Geriatr attention from the MMSE. J Geriatr Soc 1987;35:996–1000. Psychiatry Neurol 1990;3:203–207. (61) Braekhus A, Laake K, Engedal K. The (50) Bermejo F, Morales J-M, Valerga C, et al. Mini-Mental State Examination: A comparison between two abbreviated identifying the most efficient variables for Spanish versions of mental status detecting cognitive impairment in the elder- assessments in the diagnosis of dementia. ly. J Am Geriatr Soc 1992;40:1139–1145. Study data from elderly community (62) Tierney MC, Herrmann N, Geslani DM, et residents. Med Clin (Barc) al. Contributions of informant and patient 1999;112:330–334. ratings to the accuracy of the Mini-Mental (51) Hohl U, Grundman M, Salmon DP, et al. State Examination in predicting probably Mini-mental State Examination and Mattis Alzheimer’s disease. J Am Geriatr Soc Dementia Raging Scale performance differs 2003;51:813–818. in Hispanic and non-Hispanic Alzheimer’s (63) Uhlmann RF, Teri L, Rees TS, et al. Impact disease patients. J Int Neuropsychol Soc of mild to moderate hearing loss on mental 1999;5:301–307. status testing: comparability of standard (52) Marshall SC, Mungas D, Weldon M, et al. and written Mini-Mental State Differential item functioning in the Mini- Examinations. J Am Geriatr Soc Mental State Examination in English- and 1989;37:223–228. Spanish-speaking older adults. Psychol (64) Busse A, Sonntag A, Bischkopf J, et al. Aging 1997;12:718–725. Adaptation of dementia screening for (53) Schulzer M, Calne DB, Snow B, et al. A vision-impaired older persons: scoring error in the Mini-Mental State test. administration of the Mini-Mental State Can J Psychiatry 1993;38:603–605. Examination (MMSE). J Clin Epidemiol (54) Gallo JJ. Re: a scoring error in the Mini- 2002;55:909–915. Mental State test. Can J Psychiatry (65) Brandt J, Spencer M, Folstein M. The 1994;39:382. Telephone Instrument for Cognitive Status. 440 Measuring Health Neuropsychiatry Neuropsychol Behav (77) Almeida OP. Mini Mental State Neurol 1988;1:11–17. Examination and the diagnosis of dementia (66) Roccaforte WH, Burke WJ, Bayer BL, et in Brazil. Arq Neuropsiquiatr al. Validation of a telephone version of the 1998;56:605–612. Mini-Mental State Examination. J Am (78) Ogunniyi A, Osuntokun BO, Lekwauwa Geriatr Soc 1992;40:697–702. UG. Screening for dementia in elderly (67) Lanska DJ, Schmitt FA, Stewart JM, et al. Nigerians: results of the pilot test of a new Telephone-Assessed Mental State. instrument. East Afr Med J Dementia 1993;4:117–119. 1991;68:448–454. (68) Escobar JI, Burnam A, Karno M, et al. Use (79) Ogunniyi A, Lekwauwa UG, Osuntokun of the Mini-Mental State Examination BO. Influence of education on aspects of (MMSE) in a community population of cognitive functions in non-demented elder- mixed ethnicity: cultural and linguistic ly Nigerians. Neuroepidemiology artifacts. J Nerv Ment Dis 1991;10:246–250. 1986;174:607–614. (80) Crum RM, Anthony JC, Bassett SS, et al. (69) Mungas D, Marshall SC, Weldon M, et al. Population-based norms for the Mini- Age and education correction of Mini- Mental State Examination by age and Mental State Examination for English and educational level. JAMA Spanish-speaking elderly. Neurology 1993;269:2386–2391. 1996;46:700–706. (81) Bleecker ML, Bolla-Wilson K, Kawas C, et (70) Vilalta-Franch J, Llinás-Regla J, López- al. Age-specific norms for the Mini-Mental Pousa S. The Mini Cognitive Examination State Exam. Neurology for screening in epidemiologic studies of 1988;38:1565–1568. dementia. Neurologia 1996;11:166–169. (82) Heeren TJ, Lagaay AM, van Beek WCA, et (71) Mazzoni M, Ferroni L, Lombardi L, et al. al. Reference values for the Mini-Mental Mini-Mental State Examination (MMSE): State Examination (MMSE) in octo- and sensitivity in an Italian sample of patients nonagenarians. J Am Geriatr Soc with dementia. Ital J Neurol Sci 1990;38:1093–1096. 1992;13:323–329. (83) Holzer CE, Tischler GL, Leaf PJ, et al. An (72) Grut M, Fratiglioni L, Viitanen M, et al. epidemiologic assessment of cognitive Accuracy of the Mini-Mental Status impairment in a community population. Examination as a screening test for Res Community Ment Health dementia in a Swedish elderly population. 1984;4:3–32. Acta Neurol Scand 1993;87:312–317. (84) Brayne C, Calloway P. The association of (73) Katzman R, Zhang M, Qu OY, et al. A education and socioeconomic status with Chinese version of the Mini-Mental State the Mini Mental State Examination and the Examination: impact of illiteracy in a clinical diagnosis of dementia in elderly Shanghai dementia survey. J Clin people. Age Ageing 1990;19:91–96. Epidemiol 1988;41:971–978. (85) Tombaugh TN, McDowell I, Kristjansson (74) Salmon DP, Riekkinen PJ, Katzman R, et B, et al. Mini-Mental State Examination al. Cross-cultural studies of dementia: a (MMSE) and the Modified MMSE (3MS): comparison of Mini-Mental State a psychometric comparison and normative Examination performance in Finland and data. Psychol Assess 1996;8:48–59. China. Arch Neurol 1989;46:769–772. (86) Grigoletto F, Zappalà G, Anderson DW, et (75) Chiu HFK, Lee HC, Chung WS, et al. al. Norms for the Mini-Mental State Reliability and validity of the Cantonese Examination in a healthy population. version of Mini-Mental State Neurology 1999;53:315–320. Examination—a preliminary study. J Hong (87) Brayne C. The Mini-Mental State Kong Coll Psychiatry 1994;4:25–28. Examination, will we be using it in 2001? (76) Park JH, Kwon YC. Modification of the Int J Geriatr Psychiatry 1998;13:285–290. Mini-Mental State Examination for use in (88) Huppert FA, Tym E. Clinical and the elderly in a non-western society. Int J neuropsychological assessment of Geriatr Psychiatry 1990;5:381. dementia. Br Med Bull 1986;42:11–18.  Mental Status Testing 441 (89) Naugle RI, Kawczak K. Limitations of the Conceptual Basis Mini-Mental State Examination. Cleve Clin Teng and Chui intended the 3MS to improve J Med 1989;56:277–281. sensitivity and specificity of the MMSE by (90) Kittner SJ, White LR, Farmer ME, et al. adding items and extending the scoring preci- Methodological issues in screening for sion; these changes were also intended to reduce dementia: the problem of education floor and ceiling effects in the MMSE scores. adjustment. J Chronic Dis 1986;39:163–170. (91) Berkman LF. The association between Description educational attainment and mental status The 3MS includes the same items as the MMSE examinations: of etiologic significance for from which it was derived, but includes four ad- senile dementias or not? J Chronic Dis ditional items, and extends the scoring range 1986;39:171–174. from a 30-point range for the MMSE to a 100- (92) Kraemer HC, Moritz DJ, Yesavage J. point range (see Exhibit 8.9). The four new Adjusting Mini-Mental State Examination items cover long-term memory (recall of date scores for age and educational level to and place of birth), verbal fluency (naming ani- screen for dementia: correcting bias or mals), abstract thinking, and the recall of the reducing validity? Int Psychogeriatr three words an additional time (1). The 3MS is 1998;10:43–51. administered during an interview, and a correla- (93) O’Connell ME, Tuokko H, Graves RE, et al. Correcting the 3MS for bias does not tion of 0.82 has been reported between tele- improve accuracy when screening for phone and in-person administrations (2, p34). cognitive impairment or dementia. J Clin Compared with the MMSE, Teng and Chui Exp Neuropsychol 2004;26:970–980. also provided more detailed instructions for ap- (94) Schwamm LH, van Dyke C, Kiernan RJ, et plying and scoring the 3MS, addressing, for ex- al. The Neurobehavioral Cognitive Status ample, the surprisingly complex question of how Examination: comparison with the to score the ‘World’ item, which has frequently Cognitive Capacity Screening Examination been scored inconsistently. A considerable de- and the Mini-Mental State Examination in bate arose over this issue in a series of letters to a neurosurgical population. Ann Intern the Canadian Journal of Psychiatry and it ap- Med 1987;107:486–491. pears that there is no easy solution (3–5). Teng (95) Nelson A, Fogel BS, Faust D. Bedside cognitive screening instruments: a critical and Chui’s approach offers a clear, but conser- assessment. J Nerv Ment Dis vative approach, based on relative order of the 1986;174:73–83. letters (6). Gallo offered a guide to scoring based on the idea of “what is the minimum number of moves or changes required to make the reverse spelling accurate?” (4). Teng developed detailed The Modified Mini-Mental State Test interviewer training materials that even included (Evelyn Teng, 1987) review questions for testing the interviewers’ un- Purpose derstanding of the scale (E. Teng, personal com- The Modified Mini-Mental State (MMMS, or munication, 1993). A scoring method that 3MS) test extends the scope of the Mini-Mental compensates for sensory impairments and ad- State Examination (MMSE). The 3MS was in- justs for educational level has been proposed by tended to improve discrimination among differ- Khachaturian et al. (7, p533). Various cutting- ent levels of dementia. It offers a brief points have been used (generally somewhere be- assessment of attention, concentration, orienta- tween 76 and 80) and it is not clear that a tion to time and place, long- and short-term consensus has yet arisen. The large Canadian memory, language ability, constructional praxis, Study of Health and Aging used 77/78, which and abstract thinking. It may be used as a was chosen to ensure high sensitivity (8). Other screening test for cognitive loss or as a brief bed- studies have found higher values to be optimal side cognitive assessment. (9, Table 4). Exhibit 8.9 The Modified Mini-Mental State Test Note: alternatives printed in parentheses after items 6 to 10 may be used for people in institutional care settings THE 3MS Now I am going to ask some questions of a different kind. Some of the questions that I ask you will be easy; others may be more difficult. They are all routine questions that we ask of everyone. I may also ask you the same question twice. Just answer all of them as best you can. 1. ___ WHEN AND WHERE BORN? Date _____/ _____/ ________ Place ________________/ ________ 5 dd mm yyyy city/town province Day 1 0 Month 1 0 Town 1 0 Year 1 0 Province 1 0 2. ___ THREE WORDS (Number of presentations _______) 3 Shoes 1 0 Blue 1 0 Modesty 1 0 3. ___ COUNTING and WORLD BACKWARDS 7 COUNTING FORWARDS Can do Can’t 5 to 1 (write their answer) ____ ____ ____ ____ ____ 5 4 3 2 1 Score 0 1 2 SPELL “WORLD” Can do Can’t “World” backwards (print letters) ____ ____ ____ ____ ____ D L R O W Score 0 1 2 3 4 5 67 Not completed: Subject can’t read 4. ___ FIRST RECALL 9 Spontaneous recall: Shoes 3 Cue: Something to wear 2 Multiple: Shirt, shoes, socks 1 Missed completely 0 Spontaneous recall: Blue 3 Cue: A colour 2 Multiple: Black, brown, blue 1 Missed completely 0 Spontaneous recall: Modesty 3 Cue: A good personal quality 2 Multiple: Modesty, charity, honesty 1 Missed completely 0 442 Exhibit 8.9 5. ___ TODAY’S DATE 15 Today’s date _________________ Month _________________ Accurate 3 Accurate or within 5 days 2 Missed by 1 or 2 days 2 Missed by 1 month 1 Missed by 3–5 days 1 Missed by more than a month 0 Missed by more than 5 days 0 Year _________________ Day of week _________________ Accurate 8 Accurate 1 Missed by 1 year 4 Missed 0 Missed by 2–5 years 2 Missed by more than 5 years 0 Season _________________ Accurate or within a month 1 Missed 0 6. ___ SPATIAL ORIENTATION 5 Province 2 0 Country 1 0 City or town 1 0 Hosp., store, home 1 0 *MMSE: Number Y N Street Y N (Place) (Floor) 7. ___ NAMING 5 Forehead 1 0 Elbow 1 0 Chin 1 0 Knuckle 1 0 Shoulder 1 0 *MMSE: Pencil Y N Watch Y N Not completed: Subject blind 66 8. ___ FOUR-LEGGED ANIMALS (Write animals named) ?? (Timed item) (30 seconds) 10 ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ , ___________________ . (continued) 443 Exhibit 8.9 (continued) 9. ___ SIMILARITIES (Write answer) 6 Arm-leg Limbs, extremities 2 Body, parts, bend, move, joint 1 Very weak similarity or no similarity 0 Laughing-crying Feeling, emotion 2 Expressions, sounds, relieve tension 1 Very weak similarity or no similarity 0 Eating-sleeping Necessary bodily functions 2 Bodily functions, relaxing, good for you 1 Very weak similarity or no similarity 0 10. ___ REPETITION 5 I would like to go home (out) Correct 2 1 or 2 missed/wrong words 1 More than 2 missed/wrong words 0 No ifs 1 0 ands 1 0 or buts 1 0 11. ___ READ AND OBEY “CLOSE YOUR EYES” (Use Cue Card) ▲ 3 Obeys without prompting 3 Obeys after prompting 2 Read aloud only 1 None of the above 0 Not completed: subject blind 66 subject illiterate 67 12. ___ WRITING (Timed item) (1 minute) 5 (I) would like to go home (out) 0 1 2 3 4 5 *MMSE: Sentence Y N Not completed: subject physically unable 66 subject illiterate 67 Note handedness L2 R 1 (This is used in Item 14, below) 444 Exhibit 8.9 13. ___ COPYING TWO PENTAGONS (Timed item) (1 minute) 10 Pentagon 1 Pentagon 2 5 approx equal sides 4 4 5 unequal (2:1) sides 3 3 Other enclosed figures 2 2 2 or more lines 1 1 Less than 2 lines 0 0 Intersection 4 corners 2 Not 4 corner enclosure 1 No intersection or no enclosure 0 Not completed: Physically unable 66 14. ___ THREE STAGE COMMAND 3 Take this paper with your . . . Left/right hand 1 0 fold it in half 1 0 and hand it back to me 1 0 66 Physically unable 15. ___ SECOND RECALL 9 Spontaneous recall: Shoes 3 Cue: Something to Wear 2 Multiple: Shirt, shoes, socks 1 Missed completely 0 Spontaneous recall: Blue 3 Cue: A colour 2 Multiple: Black, brown, blue 1 Missed completely 0 Spontaneous recall: Modesty 3 Cue: A good personal quality 2 Multiple: Modesty, charity, honesty 1 Missed completely 0 ________ 3MS TOTAL SCORE The 3MS test as administered in the Canadian Study of Health and Aging. Adapted from an original provided by Dr. E. Teng. With permission. 445 446 Measuring Health Reliability the 3MS and the MMSE. Coefficients were con- Teng et al. reported retest correlations over de- sistently higher for the 3MS. Correlations with lays between 52 and 98 days ranging from 0.91 the Boston Naming Test were 0.61 for the 3MS to 0.93. Equivalent figures for the MMSE were and 0.55 for the MMSE; with the Controlled 0.79 to 0.89 (10, Table 3). One-month stability Word Association Test, the results were 0.81 and coefficients were 0.80 for the 3MS and 0.71 for 0.59; with the Logical Memory test, the coeffi- the MMSE in a study of stroke patients (9, cients were 0.62 and 0.55. Finally, the 3MS cor- p479). related 0.44 with the Functional Independence In the Canadian study, alpha was 0.87 for the Measure; the equivalent correlation for the 3MS, compared with 0.78 for the MMSE; split- MMSE was 0.36 (9, Table 6). half reliability was 0.82 (0.76 for the MMSE) Teng et al. reported a range of sensitivity and (11, p380). An alpha of 0.80 has been reported, specificity results for the 3MS and the MMSE, along with a 14-day retest correlation of 0.87 for people with different educational levels. For (12, p116). people with 7 to 12 years of education and at a In a study of patients in long-term care facili- specificity of 0.95, sensitivity was 0.94 for the ties, inter-rater reliability was equal for the 3MS 3MS and 0.88 for the MMSE. For people with and MMSE (r= 0.99). Retest reliability was 0.92 13 or more years of education, again at a speci- for the 3MS, compared with 0.85 for the ficity of 0.95, sensitivity was 0.91 for the 3MS MMSE, while alpha was 0.90 for the 3MS com- and 0.86 for the MMSE (10, Table 4). In the pared with 0.84 for the MMSE (13, p179). Canadian Study of Health and Aging A comparison between self-administration (N = 8,900), sensitivity was 87% and specificity and administration by a nurse (after a median 89%. The area under the ROC curve was 0.94, delay of 49 days) gave an intraclass correlation compared with 0.89 for the MMSE (11, p380). (ICC) of 0.87, compared with a value of 0.78 In a subset of the same study participants, sensi- for the MMSE (14, Table 2). A different phase tivity was 88% and specificity 90% at a cutting- of the same study gave an ICC figure of 0.85 point of 77/78 (19, p508). Further analyses (15, p76). A much higher inter-rater ICC value compared the 3MS and MMSE, giving different of 0.98 has been reported, along with an alpha weights to sensitivity and specificity (false- of 0.91 and a one-year retest reliability of 0.78 negative and false-positive errors). The 3MS (16, pp624–5). Likewise, a dichotomous classifi- proved slightly superior at all levels but per- cation by the 3MS into impaired versus not im- formed best when sensitivity was weighted more paired remained stable over time (kappa = 1.0) highly than specificity (11, Table 3). An analysis (17). of a combination of the 3MS and the IQCODE as a screening test produced an area under the Validity ROC curve of 0.96 (7, p535). In a study of pa- A factor analytic study identified five factors, la- tients at long-term care facilities, areas under the beled psychomotor skills, memory, identification ROC curve were identical for 3MS and MMSE, and association, orientation, and concentration at 0.84 and 83 (13, p180). In that sample, speci- (18, Table 2). A four factor solution has been re- ficity for both instruments was low, perhaps be- ported (12, Table IV). cause many participants at long-term care Correlations with other measures include facilities who were not diagnosed with dementia 0.90 with the MMSE, −0.80 with the Blessed had milder forms of cognitive impairment that Dementia Scale, and 0.85 with the Cambridge are difficult to distinguish from dementia. Mental Disorders of the Elderly Examination Scores on the 3MS typically vary by age, edu- (CAMDEX) Cognitive scale CAMCOG (12, cation, and perhaps also with interaction effects Table II). Other estimates of correlations with of age and education (20, Table 2; 21, Table 2). the MMSE include 0.84 and 0.85 (9, p479). Gender-education interactions were identified in Grace et al. presented a range of convergent cor- a study in Utah (2, Figure 2). Accordingly, “cor- relations with neuropsychological tests for both rected” norms have been proposed that adjust  Mental Status Testing 447 for the effects of age, sex, and education. How- to 4). However, note the findings of O’Connell et ever, it has proven difficult to demonstrate that al., summarized above, that show that use of the association of scores with age and education corrected norms may actually reduce validity if actually imply a reduction in validity of the in- the purpose is to set cutting-scores in screening strument. Some studies that have used a regres- for dementia (21). sion approach to correct scores for the effect of age and education actually reduce the validity of the 3MS (21; 22). In O’Connell’s study, the AUC Commentary for detecting dementia was 0.91 for the uncor- The 3MS appears to offer increased validity over rected 3MS scores, falling to 0.88 when cor- the MMSE (9; 11; 25), but at the cost of more rected for age and education (21, p975). time required for administration and somewhat O’Connell also used cutting-scores based on greater complexity in scoring. The conclusion of norms that correct for age and education, again the Canadian analyses was that the superiority showing that this actually reduced validity (AUC of the 3MS was attributable to both its addi- 0.91 vs. 0.86) (21, p977). tional items and the extended scoring system (11, p381). Alternative Forms An interesting debate has surrounded the use A Canadian French version has been described of adjustments for age and education in estab- by Hébert et al, who show a copy of the scale in lishing norms for the MMSE and the 3MS (see a format that permits scoring both 3MS and an extended discussion in the review of the MMSE (23, p445). The correlation between MMSE). Although scores unquestionably vary MMSE and 3MS scores was 0.97. The alpha in- by education and age, so does the incidence of ternal consistency was 0.89, and the one-week dementia. Hence, this source of variance in test-retest intraclass correlation was 0.94 (23, scores should perhaps not be removed if the pur- p447). Hébert et al. also compared three raters pose is to screen for cognitive impairment or de- who administered the 3MS twice, one week mentia. Ultimately, formal analyses of apart; the intraclass correlation between the two differential item functioning will be necessary to raters with the highest agreement was 0.95. identify items that show educational or age bias Tschanz et al. have modified the remote (as opposed to true differences in cognitive func- memory items in the 3MS, replacing recall of tion that correspond to differences in age or edu- date and place of birth by recall of current and cation), but such analyses are only just past politicians, and altering some of the scoring beginning to be undertaken. (2, pp35–37). They provided normative data Teng and her colleagues have subsequently (means, SDs and percentiles) from a sample in developed the Cognitive Abilities Screening In- Utah (2, Tables 2 and 3). The 3MS has also been strument (CASI), which is an extension to the used with children (24). 3MS (28). Reference Standards Reference standards by age-group and educa- References tional level have been derived from a nonde- mented Canadian population (20, Table 4; 25, (1) Teng EL, Chui HC. The Modified Mini- Table 4). These gave similar results to the norms Mental State (3MS) Examination. J Clin Psychiatry 1987;48:314–318. derived in Utah (2). Percentile scores from a (2) Tschanz JT, Welsh-Bohmer KA, Plassman small study of relatively highly educated white BL, et al. An adaptation of the Modified people in Florida have been presented (26, Table Mini-Mental State Examination: analysis 2); this study also reported adjustments for age- of demographic influences and normative group and education (Table 3). Similarly, norms data. Neuropsychiatry Neuropsychol and adjustments for age and education are avail- Behav Neurol 2002;15:28–38. able from a small sample of blacks (27, Tables 1 (3) Schulzer M, Calne DB, Snow B, et al. A 448 Measuring Health scoring error in the Mini-Mental State test. individual change scores on the 3MS in Can J Psychiatry 1993;38:603–605. older persons with dementia: results from (4) Gallo JJ. Re: a scoring error in the Mini- the Canadian Study of Health and Aging. Mental State test. Can J Psychiatry Int Psychogeriatr 2000;12(suppl. 2):73–80. 1994;39:382. (16) Bassuk SS, Murphy JM. Characteristics of (5) Schulzer M, Calne DB, Snow B, et al. Re: a the Modified Mini-Mental State Exam scoring error in the Mini-Mental State test. among elderly persons. J Clin Epidemiol The authors respond. Can J Psychiatry 2003;56:622–628. 1994;39:384–385. (17) Lamarre CJ, Patten SB. Evaluation of the (6) Teng EL, Chui HC. A scoring error in the Modified Mini-Mental State Examination Mini-Mental State test: a comment. Can J in a general psychiatric population. Can J Psychiatry 1994;39:383–384. Psychiatry 1991;36:507–511. (7) Khachaturian AS, Gallo JJ, Breitner JCS. (18) Abraham IL, Manning CA, Boyd MR, et Performance characteristics of a two-stage al. Cognitive screening of nursing home dementia screen in a population sample. J residents: factor structure of the Modified Clin Epidemiol 2000;53:531–540. Mini-Mental State (3MS) examination. Int (8) Canadian Study of Health and Aging J Geriatr Psychiatry 1993;8:133–138. Working Group. The Canadian Study of (19) Bland RC, Newman SC. Mild dementia or Health and Aging: study methods and cognitive impairment: the Modified Mini- prevalence of dementia. Can Med Assoc J Mental State Examination (3MS) as a 1994;150:899–913. screen for dementia. Can J Psychiatry (9) Grace J, Nadler JD, White DA, et al. 2001;46:506–510. Folstein vs modified Mini-Mental State (20) Bravo G, Hébert R. Age- and education- Examination in geriatric stroke. Stability, specific reference values for the Mini- validity, and screening utility. Arch Neurol Mental and Modified Mini-Mental State 1995;52:477–484. Examinations derived from a non- (10) Teng EL, Chui HC, Gong A. Comparisons demented elderly population. Int J Geriatr between the Mini-Mental State Psychiatry 1997;12:1008–1018. Examination (MMSE) and its modified (21) O’Connell ME, Tuokko H, Graves RE, et version - the 3MS test. Excerpta Med al. Correcting the 3MS for bias does not 1990;189–192. improve accuracy when screening for (11) McDowell I, Kristjansson B, Hill GB, et al. cognitive impairment or dementia. J Clin Community screening for dementia: the Exp Neuropsychol 2004;26:970–980. Mini-Mental State Exam (MMSE) and (22) Kraemer HC, Moritz DJ, Yesavage J. Modified Mini-Mental State Exam (3MS) Adjusting Mini-Mental State Examination compared. J Clin Epidemiol scores for age and educational level to 1997;50:377–383. screen for dementia: correcting bias or (12) Cappeliez P, Quintal M, Blouin M, et al. reducing validity? Int Psychogeriatr Les propriétés psychométriques de la 1998;10:43–51. version française du Modified Mini-Mental (23) Hébert R, Bravo G, Girouard D. Validation State (3MS) avec des patients âgés suivis en de l’adaptation française du modified mini- psychiatrie gériatrique. Can J Psychiatry mental state (3MS). Rev Gér 1996;41:114–121. 1992;17:443–450. (13) Nadler JD, Relkin NR, Cohen MS, et al. (24) Besson PS, Labbe EE. Use of the Modified Mental status testing in the elderly nursing Mini-Mental State Examination with home population. J Geriatr Psychiatry children. J Child Neurol Neurol 1995;8:177–183. 1997;12:455–460. (14) Bravo G, Hébert R. Reliability of the (25) Tombaugh TN, McDowell I, Kristjansson Modified Mini-Mental State Examination B, et al. Mini-Mental State Examination in the context of a two-phase community (MMSE) and the Modified MMSE (3MS): prevalence study. Neuroepidemiol a psychometric comparision and normative 1997;16:141–148. data. Psychol Assess 1996;8:48–59. (15) Correa JA, Perrault A, Wolfson C. Reliable (26) Jones TG, Schinka JA, Vanderploeg RD, et  Mental Status Testing 449 al. 3MS normative data for the elderly. to obtain a history of decline from an informant Arch Clin Neuropsychol who knows the person well, and Jorm set out to 2002;17:171–177. standardize this approach in developing the (27) Brown LM, Schinka JA, Mortimer JA, et IQCODE (3). al. 3MS normative data for elderly African The IQCODE contrasts with other assess- Americans. J Clin Exp Neuropsychol ments of cognition in several ways. As shown in 2003;25:234–241. Exhibit 8.10, it focuses on performance of com- (28) Teng EL, Hasegawa K, Homma A, et al. The Cognitive Abilities Screening mon cognitive activities, rather than on artificial Instrument (CASI): a practical test for tasks such as counting backward (1, p1015); in cross-cultural epidemiological studies of effect, it records disability or handicap rather dementia. Int Psychogeriatr 1994;6:45–58. than impairment. Next, by focusing on change in cognitive performance and the impact of this on daily life, the IQCODE was designed to re- The Informant Questionnaire on flect two of the criteria for dementia, and the Cognitive Decline in the Elderly focus on change was expected to enhance sensi- (IQCODE) tivity to the early stages of cognitive decline. Fi- (Anthony Jorm, 1989) nally, by relying on an informant’s judgment, the IQCODE is not influenced by a person’s test- Purpose taking ability, so may avoid the bias toward The IQCODE provides an estimate of change in those with higher education that occurs with cognitive abilities, based on the report of an in- many cognitive tests. formant who knows the person being assessed well. It may be used as part of a clinical assess- Description ment of cognition or in screening for dementia. The IQCODE originated as an interview in which the informant was asked to rate the sub- Conceptual Basis ject’s improvement or decline in 26 aspects of Diagnostic criteria for dementia require docu- memory and intelligence over a ten-year period mentation of a decline from previous levels of (3). The self-administered questionnaire de- cognitive function, which demands an assess- scribed here was derived from that interview ment of premorbid ability. Without assessing schedule (2). The questions cover changes in change, a person who has always had limited learning and recall abilities, in recognition and cognitive abilities might be falsely classified as comprehension, and other aspects of intelli- demented; alternatively, dementia might be gence. The ten-year period was chosen as being missed in a formerly very high-functioning per- an easy reference for respondents, and because son who has experienced a significant decline this would encompass the entire natural history but has not yet reached a typical threshold for of a typical dementia (2). diagnosing dementia. Neuropsychological test Each item is rated on a 1 to 5 scale in which evidence of premorbid ability is seldom avail- 1 represents considerable improvement, 3 indi- able, but several approaches may be taken to es- cates no change, and 5 represents considerable timate it (1). These include extrapolating from deterioration (see exhibit 8.10). An overall current performance on tests of cognition that change score is calculated by averaging the are not affected by dementia, such as word read- scores on each item, so that the overall score can ing ability. Alternatively, premorbid ability may be interpreted on the same scale as each item. be crudely estimated from a person’s education Scores are presented to two decimal places, and and occupation, but this does not provide an ac- up to one quarter of the items can be missing curate indication in individual cases (1; 2). The before the overall score is regarded as missing approach of asking the person to rate his or her (4). own change may be inaccurate among people Scores below 3 indicate improvement and 3.0 with cognitive impairment. A fourth strategy is indicates no change. Slight decline is indicated Exhibit 8.10 The IQCODE. Items retained in the 16-item abbreviated version are indicated with an asterisk Now we want you to remember what your friend or relative was like 10 years ago and to compare this with what he / she is like now. 10 years ago was in 19__. Below are situations where this person has to use his / her memory or in- telligence and we want you to indicate whether this has improved, stayed the same, or got worse in that situation over the past 10 years. Note the importance of comparing his / her present performance with 10 years ago. So if 10 years ago this person always forgot where he / she had left things, and he / she still does, then this would be consid- ered ‘Hasn’t changed much’. Please indicate the changes you have observed by circling the appropriate answer. Compared with 10 years ago how is this person at: 1 2 3 4 5 1. Recognizing the faces of family and friends Much A bit Not much A bit Much improved improved change worse worse 2. Remembering the names of family and Much A bit Not much A bit Much friends improved improved change worse worse 3. *Remembering things about family and Much A bit Not much A bit Much friends, e.g., occupations, birthdays, improved improved change worse worse addresses 4. *Remembering things that have happened Much A bit Not much A bit Much recently improved improved change worse worse 5. *Recalling conversations a few days later Much A bit Not much A bit Much improved improved change worse worse 6. Forgetting what he/she wanted to say in Much A bit Not much A bit Much the middle of a conversation improved improved change worse worse 7. *Remembering his/her address and Much A bit Not much A bit Much telephone number improved improved change worse worse 8. *Remembering what day and month Much A bit Not much A bit Much it was improved improved change worse worse 9. *Remembering where things are usually Much A bit Not much A bit Much kept improved improved change worse worse 10. *Remembering where to find things Much A bit Not much A bit Much which have been put in a different place improved improved change worse worse than usual 11. Adjusting to any change in his/her Much A bit Not much A bit Much day-to-day routine improved improved change worse worse 12. *Knowing how to work familiar Much A bit Not much A bit Much machines around the house improved improved change worse worse 13. *Learning to use a new gadget or Much A bit Not much A bit Much machine around the house improved improved change worse worse 14. *Learning new things in general Much A bit Not much A bit Much improved improved change worse worse 15. Remembering things that happened to Much A bit Not much A bit Much him/her when he/she was young improved improved change worse worse 16. Remembering things he/she learned Much A bit Not much A bit Much when he/she was young improved improved change worse worse 17. Understanding the meaning of Much A bit Not much A bit Much unusual words improved improved change worse worse 18. Understanding magazine or newspaper Much A bit Not much A bit Much articles improved improved change worse worse 19. *Following a story in a book or on TV Much A bit Not much A bit Much improved improved change worse worse 450  Mental Status Testing 451 Exhibit 8.10 20. Composing a letter to friends Much A bit Not much A bit Much or for business purposes improved improved change worse worse 21. Knowing about important historical Much A bit Not much A bit Much events of the past improved improved change worse worse 22. *Making decisions on everyday matters Much A bit Not much A bit Much improved improved change worse worse 23. *Handling money for shopping Much A bit Not much A bit Much improved improved change worse worse 24. *Handling financial matters, e.g. the Much A bit Not much A bit Much pension, dealing with the bank improved improved change worse worse 25. *Handling other everyday arithmetic Much A bit Not much A bit Much problems, e.g., knowing how much improved improved change worse worse food to buy, knowing how long between visits from family or friends 26. *Using his/her intelligence to understand Much A bit Not much A bit Much what’s going on and to reason things improved improved change worse worse through. From an original provided by Dr. A. Jorm. Reproduced with permission. by scores from 3.01 to 3.5; moderate decline in- The lowest reported value was an alpha of 0.86 cludes scores from 3.51 to 4; scores of 4.01 and for the short version of the IQCODE in a small higher indicate severe decline (5, p982). Used as sample of people with mental retardation (9, a screening test for dementia, different studies p67). have proposed different cutting-points. These in- In a one-year follow-up of 260 people with clude 3.6 and higher to designate a case (opti- cognitive impairment or dementia, the IQCODE mized on the assumption that sensitivity and showed a significant decline in scores, giving a specificity are equally important) (4, p788). A retest correlation of 0.75 (1, p1017). A retest lower threshold of 3.27/3.3 was found optimal correlation over an average delay of only 2.8 in a combined analysis of four studies; the equiv- days yielded a correlation of 0.96 (compared to alent point for the 16-item abbreviated form was 0.79 for the Mini-Mental State Examination 3.31/3.38 (6, p150). For patients without de- [MMSE]) (4, p789). mentia, IQCODE scores higher than 3.3 pre- dicted the subsequent development of dementia Validity and also mortality compared with those with factorial validity. The IQCODE was designed lower scores (7). to cover several aspects of cognition, but its high internal consistency suggests that the informant Reliability ratings do not distinguish these different facets. Although the questions were originally con- Jorm et al. reported a factor analysis in which ceived as measuring several different aspects of the first factor accounted for 42% of the vari- memory and intelligence, all 26 items have high ance, followed by 10% for the second factor, item-total correlations. Cronbach’s alpha was and 7% for the third. All items loaded highly on 0.93 in a sample of Alzheimer’s patients (2, p the first factor, ranging from 0.47 to 0.81 in dif- 36). Values of 0.95 have been reported for a ferent samples (2, p36). Jorm concluded that a general population sample (N = 613) (1, general factor underlies the IQCODE and that a p1017), and 0.97 in a Chinese sample (8, p95). single score is optimal. Subsequent studies have 452 Measuring Health also identified a general factor, accounting for neuropathological findings at necropsy, sensitiv- 48% of the variance (1, Table 2) and 61% in a ity was 73% and specificity 75% (12, p235). Chinese sample (8, p95). Jorm has estimated the effect size for the IQ- CODE when used as a screening test, using data criterion validity. Studies in several countries pooled from eight studies. The mean effect size have shown that the IQCODE distinguishes be- was 1.82, compared with 1.53 for the MMSE tween patients with dementia and controls. In (13, p159). This translates into a sensitivity of Jorm and Jacomb’s original work, every item dis- 86% for the IQCODE compared with 79% for tinguished significantly and, using a cutting- the MMSE, at the same specificity of 80%. point of 4+, the IQCODE would correctly There was, however, considerable variation be- classify 92.7% of those with dementia, whereas tween studies in the effect size values for the IQ- specificity was 88%. The latter is a conservative CODE, whereas the MMSE performed equally estimate because some people in the general pop- well across studies. ulation comparison sample may in reality have Jorm et al. reported a correlation of 0.44 been impaired (1, p1018). Compared with clini- with an indicator of the level of care being re- cal diagnoses of dementia based on the Ameri- ceived (2, p37), and IQCODE scores were signif- can Psychiatric Association’s Diagnostic and icantly worse in people who were subsequently Statistical Manual, 3rd edition, revised (DSM- admitted to an institution than for those who re- IIIR) criteria, the IQCODE had a sensitivity of mained in the community one year later (1, 69% and specificity of 80%; corresponding fig- p1018). ures for tenth edition of the International Classi- fication of Diseases (ICD-10) diagnoses were concurrent validity. Many studies have com- 80% and 82% (4, p788). The closer correspon- pared the IQCODE and MMSE. Jorm calculated dence to ICD-10 diagnoses may reflect the inclu- a mean correlation of −0.59 across seven studies sion of cognitive decline in the ICD, but not in (14, p58), and subsequent results are compara- the DSM criteria (4, p789). Jorm et al. com- ble. The highest correlations with the MMSE in- pared the IQCODE against four diagnostic ap- clude −0.74 (3), −0.75 (15), and −0.78 (2, p36); proaches to dementia, and found that the lower values include −0.44 (16, p542) and −0.41 IQCODE consistently provided a larger area un- (17, p133). der the ROC curve than the MMSE did, al- A study of elderly general practice patients though the difference was significant in only one reported a correlation of −0.52 between the IQ- of these four instances (4, p787). By contrast, a CODE and the information/orientation subscale study of hospital patients obtained an area under of the Clifton Assessment Procedures for the El- the ROC curve of 0.86 for the IQCODE, but derly, and a correlation of −0.53 with the 0.92 for the MMSE (10, p534). A Chinese study MMSE (15, Table 2). A correlation of −0.58 has calculated an area under the ROC of 0.91 for been reported with the Abbreviated Mental Test the IQCODE, compared with 0.84 for the (12, p235). A value of only 0.29 was obtained MMSE (8, p94). Harwood et al. reported sensi- between the abbreviated IQCODE and the Short tivity of 100% and specificity of 86% for the Portable Mental Status Questionnaire in a study IQCODE, compared to 96% and 73% for the of people with mental retardation (9, p68). Abbreviated Mental Test (11). Correlations between the IQCODE and neu- The IQCODE has been used retrospectively ropsychological test results have been compared to assess ante mortem cognitive status in people with equivalent coefficients for the MMSE. The who had died between 10 months and 3.5 years MMSE agreed more closely than the IQCODE previously (12). Compared with a clinical diag- with four of the Wechsler Adult Intelligence nosis of Alzheimer’s disease made post mortem Scale WAIS-R subtests, the Boston Naming Test, from case notes, sensitivity was 76% and speci- and the Benton Visual Retention Test. The coef- ficity 86%. Compared with diagnoses based on ficients were very similar for the WMS-R logical  Mental Status Testing 453 memory test, the Trail Making test, and most by a few of the IQCODE items could, presum- components of the Rey Auditory Verbal Learn- ably, reflect physical disabilities, leading to false- ing test (17, Table 1). In a study of stroke pa- positive results. Scores have been found to tients, IQCODE scores correlated −0.54 with correlate 0.37 with loss in ability to use hands scores on the Minnesota Paper Form Board test, and fingers, and 0.29 with loss in ability to use which examines spatial ability; the IQCODE legs (2, p37). In a study of stroke patients the correlated −0.28 with Raven’s Progressive Ma- IQCODE scores correlated −0.60 with Barthel trices, −0.34 with the Auditory Verbal Learning scores (18, p218). Several studies have found Test and −0.42 with a general cognitive factor moderate correlations between IQCODE scores based on a range of tests (18, Table 2). A corre- and age (a range of 0.30 to 0.35 is typical), and lation of 0.35 with depression was obtained in there was a slight tendency for female respon- the same study (18, p218). dents to show greater declines (1, p1017). IQCODE scores have also been compared Because it measures change in cognitive sta- with CT scan results. IQCODE scores correlated tus, it was hoped that IQCODE scores would be significantly with the width of the third ventricle independent of education. Virtually all studies and with the number of infarcts (r= 0.37 for left have confirmed this (8, p94; 10; 14, p58; 16). In hemisphere, and 0.27 for right). These coeffi- one population study, a slight correlation with cients exceeded those for the Mini-Mental Status education was significant, due to a large sample Exam (MMSE) (17, Table 2). Neither scale cor- size (1). To assess whether this represented test related with measures of bain atrophy but, inter- bias or a real tendency for people with lower ed- estingly, MMSE scores correlated significantly ucation to experience greater decline, validity with brain diameter (which may indicate pre- and reliability analyses were repeated when morbid intelligence and not cognitive decline) stratified for educational level. The results were whereas the IQCODE did not. Thomas has re- closely comparable, suggesting that the IQ- ported a positive correlation between IQCODE CODE was not biased against less educated peo- scores and levels of amyloid precursor protein ple (1, p1019). Scores on the IQCODE have also (19). been shown to have almost zero correlations The retrospective assessments of the IQ- with other socioeconomic indicators (2, p37). CODE have been compared with actual changes in MMSE scores recorded over time in an Aus- Alternative Forms tralian longitudinal study. A correlation of 0.39 The abbreviated version of the IQCODE in- was obtained with changes in MMSE scores over cludes the 16 items indicated in Exhibit 8.10. a 3.6-year period (20, p127); this rose to 0.47 The short and long versions correlated 0.98, and for changes after 7.7 years (5, Table 1). the short form correlated −0.58 with the MMSE, In principle, the contrasting approaches of compared to −0.61 for the long version (6, the IQCODE and MMSE should provide com- p149). Long and short forms showed similar plementary information in identifying dementia. correlations with a range of neuropsychological Indeed, Mackinnon and Mulligan found that a tests; if anything, the coefficients for the short combination of the two tests was superior to ei- form were fractionally higher (17, Table 1). The ther given alone, giving an area under the ROC area under the ROC curve was 0.85 for both curve of 0.96 (21). By contrast, Jorm et al. found versions, compared with 0.90 for the MMSE (6, no benefit to combining the two tests (4, p788), p150). Analyses on different samples have con- and the results obtained by Flicker et al. likewise firmed that the short version performs almost as seem to show no advantage to the combination, well as the full IQCODE (8). although the Abbreviated Mental Test worked Translations include Italian (23) and Spanish, better than the MMSE in combination with the including an abbreviated version (24). Of three IQCODE (22, Table 3). Spanish studies comparing the sensitivity and The changes in normal functioning recorded specificity of the IQCODE with that of the 454 Measuring Health MMSE, two found the IQCODE superior; all information that can complement the results of three studies found the IQCODE did not corre- cognitive testing, and that is directly relevant to late with educational status or pre morbid intel- the diagnosis of dementia. As a screening tool, it ligence, unlike the MMSE (24–26). A Dutch appears to perform better than established in- version showed an internal alpha consistency of struments such as the MMSE. It also appears to 0.95 and significant correlations with behavioral have overcome the common bias in such tests to- disturbances (27, p463); the correlation between ward people with higher education. Dutch short and long versions was 0.97 (28). A Jorm has listed several potential advantages Canadian French version has also been com- in assessing cognition using informant reports pared with the MMSE. At a cutting-point of 3.6 (14). These include greater acceptability com- or over, the IQCODE(F) produced a sensitivity pared with direct cognitive testing, which draws of 76% and a specificity of 96% in screening for the patient’s attention to their deficits. Informant dementia of all types, compared with 71% and assessments can be used with people who are too 82% for the MMSE (16, p543). The IQCODE sick to undergo formal testing; they can be used and MMSE correlated −0.44, and IQCODE post mortem, and they can be obtained by tele- scores were unrelated to educational level, com- phone or mail. It is simpler to use a single infor- pared with a correlation of 0.28 for the MMSE mant report to record change in performance (16). A Chinese version is available, along with over time than to undertake repeated cognitive an abbreviated version that is similar, but not testing. Further, diagnostic criteria for dementia identical, to Jorm’s abbreviation shown in Ex- require that deficits be sufficient to hinder daily hibit 8.10 (8, Table 2). A Thai version has been function; this can be better assessed by an infor- compared with clinical diagnoses of dementia mant than by standardized cognitive testing. The for 160 patients. The area under the ROC curve IQCODE capitalizes on the sensitivity and for the IQCODE was 0.93, exceeding that for awareness of family members to early signs of the MMSE, at 0.81. Three of the questions cognitive loss, and the use of informant reports yielded a sensitivity of 84.9% and a specificity eliminates bias due to test-taking skills among of 92% (29). people with higher levels of education. The focus on daily function may offer greater cross- cultural portability compared with artificial Reference Standards tasks such as counting backward, which vary in Population norms are available based on a sam- familiarity in different cultures (14, p53). ple of 613 people from Canberra, Australia (1, Although the IQCODE has high validity, p1022). A simple graph, the DemeGraph, com- there are potential limitations to the informant bines MMSE and IQCODE scores to indicate approach. These include the occasional lack of a which combinations of scores may indicate suitable informant, possible informant bias, and pathology (www.mhri.edu.au/biostats/deme the potential influence of informant characteris- graph/). tics such as age, mental or cognitive state, and how well they actually know the person being Commentary judged. Ratings do appear to be influenced by Although it is standard procedure for clinicians the subject’s depression, and probably by the to incorporate information from family members personalities of both subject and informant. Age into a cognitive assessment, systematic protocols itself does not appear to influence the accuracy for doing this are relatively recent and Jorm’s IQ- of an informant’s rating (14, pp67–68), but Jorm CODE has been the leader in this field. Its valid- has shown small but significant correlations be- ity testing has been extensive: few other scales tween the informant’s depression or anxiety and have been compared against such a wide variety their judgments (14, Table 5). This does not nec- of neuropsychological tests and even brain imag- essarily mean that such judgment is inaccurate ing. Validation results in several countries sug- because the informant’s depression could have gest that the IQCODE provides accurate resulted from the subject’s cognitive problems,  Mental Status Testing 455 but MMSE ratings in the same study showed no able to use both types of cognitive assessment” correlation with the informant’s affective state, (17, p138). suggesting that their perception may have been distorted (14, p68). In the same study, Jorm also provided some evidence that the quality of the References relationship between informant and the person being rated may influence the ratings (p69). It (1) Jorm AF, Jacomb PA. The Informant may also be that the subject’s physical or affec- Questionnaire on Cognitive Decline in the tive state influences the informant’s ratings, so Elderly (IQCODE): socio-demographic correlates, reliability, validity and some scores could be confounded by depression; cor- norms. Psychol Med 1989;19:1015–1022. relations between IQCODE scores and the sub- (2) Jorm AF, Scott R, Jacomb PA. Assessment ject’s depression range from 0.15 to 0.45. An of cognitive decline in dementia by infor- association with depression might occur, for ex- mant questionnaire. Int J Geriatr ample, if the depressed person tended to com- Psychiatry 1989;4:35–39. plain more to the informant about their impaired (3) Jorm AF, Korten AE. Assessment of cognition, but we do not yet know which is cognitive decline in the elderly by infor- cause and which effect (14, p66). mant interview. Br J Psychiatry The existence of a general factor underlying 1988;152:209–213. the IQCODE scores suggests that informants (4) Jorm AF, Scott R, Cullen JS, et al. Perfor- provide global assessments, rather than testing mance of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) particular cognitive functions. Informants ap- as a screening test for dementia. Psychol pear to be responding to their perception of the Med 1991;21:785–790. subject’s everyday behavior and cognitive com- (5) Jorm AF, Christensen H, Korten AE, et al. plaints, as modified by their own reactions to Informant ratings of cognitive decline in this. The focus is on episodic memory plus reac- old age: validation against change on tion speed (17, p137), and “spatial rather than cognitive tests over 7 to 8 years. Psychol verbal abilities” (18, p219). Even though ques- Med 2000;30:981–985. tions were intended to tap different aspects of (6) Jorm AF. A short form of the Informant cognition, these are not revealed empirically. Questionnaire on Cognitive Decline in the Perhaps the tasks recorded in the IQCODE are Elderly (IQCODE): development and cross- complex and involve several different abilities, validation. Psychol Med 1994;24:145–153. (7) Lewis B, Harwood D, Hope T, et al. Can or perhaps untrained informants are unable to an informant questionnaire be used to distinguish different facets of decline. predict the development of dementia in Although it remains unclear precisely which medical patients? Int J Geriatr Psychiatry underlying general factor the IQCODE is mea- 1999;14:941–945. suring, the instrument nonetheless performs re- (8) Fuh JL, Teng EL, Lin KN, et al. The Infor- markably well as a general cognitive screening mant Questionnaire on Cognitive Decline test. Its brevity, ease of administration by tele- in the Elderly (IQCODE) as a screening phone or mail, and that the patient need not be tool for dementia for a predominantly present are strong points. But no test is perfect illiterate Chinese population. Neurology and, as Jorm noted: “The IQCODE has the ad- 1995;45:92–96. vantage over the MMSE of not being affected by (9) Shultz JM, Aman MG, Rojahn J. Psychometric evaluation of a measure of education and premorbid ability. However, it is cognitive decline in elderly people with contaminated in ways the MMSE is not, for ex- mental retardation. Res Devel Disabil ample by the affective state and personality of 1998;19:63–71. the subject and by the affective state of the infor- (10) Mulligan R, Mackinnon A, Jorm AF, et al. mant and the quality of the relationship between A comparison of alternative methods of the subject and the informant. Because of their screening for dementia in clinical settings. contrasting strengths and weaknesses, it is desir- Arch Neurol 1996;53:532–536. 456 Measuring Health (11) Harwood DM, Hope T, Jacoby R. (23) Isella V, Villa ML, Frattola L, et al. Cognitive impairment in medical Screening cognitive decline in dementia: inpatients. I: Screening for dementia—is preliminary data on the Italian version of history better than mental state? Age the IQCODE. Neurol Sci 2002;23(suppl Ageing 1997;26:31–35. 2):S79–S80. (12) Thomas LD, Gonzales MF, Chamberlain (24) Morales JM, Gonzalez-Montalvo JI, A, et al. Comparison of clinical state, Bermejo F, et al. The screening of mild retrospective informant interview and the dementia with a shortened Spanish version neuropathologic diagnosis of Alzheimer’s of the “Informant Questionnaire on disease. Int J Geriatr Psychiatry Cognitive Decline in the Elderly.” 1994;9:233–236. Alzheimer Dis Assoc Disord (13) Jorm AF. Methods of screening for 1995;9:105–111. dementia: a meta-analysis of studies (25) Morales JM, Bermejo F, Romero M, et al. comparing an informant questionnaire Screening of dementia in community- with a brief cognitive test. Alzheimer Dis dwelling elderly through informant report. Assoc Disord 1997;11:158–162. Int J Geriatr Psychiatry 1997;12:808–816. (14) Jorm AF. Assessment of cognitive (26) Del-Ser T, Morales J-M, Barquero M-S, et impairment and dementia using informant al. Application of a Spanish version of the reports. Clin Psychol Rev 1996;16:51–73. “Informant Questionnaire on Cognitive (15) Bowers J, Jorm AF, Henderson S, et al. Decline in the Elderly” in the clinical General practitioners’ detection of assessment of dementia. Alzheimer Dis depression and dementia in elderly Assoc Disord 1997;11:3–8. patients. Med J Aust 1990;153:192–196. (27) de Jonghe JFM. Differentiating between (16) Law S, Wolfson C. Validation of a French demented and psychiatric patients with the version of an informant-based questionnaire Dutch version of the IQCODE. Int J as a screening test for Alzheimer’s disease. Geriatr Psychiatry 1997;12:462–465. Br J Psychiatry 1995;167:541–544. (28) de Jonghe JFM, Schmand B, Ooms ME, et (17) Jorm AF, Broe GA, Creasey H, et al. al. [Abbreviated form of the Informant Further data on the validity of the Infor- Questionnaire on cognitive decline in the mant Questionnaire on Cognitive Decline elderly.] [Dutch.] Tijdschr Gerontol Geriatr in the Elderly (IQCODE). Int J Geriatr 1997;28:224–229. Psychiatry 1996;11:131–139. (29) Senanarong V, Assavisaraporn S, (18) Starr JM, Nicolson C, Anderson K, et al. Sivasiriyanonds N, et al. The IQCODE: an Correlates of informant-rated cognitive alternative screening test for dementia for decline after stroke. Cerebrovasc Dis low educated Thai elderly. J Med Assoc 2000;10:214–220. Thai 2001;84:648–655. (19) Thomas LD. Neuropsychological correlates of amyloid precursor protein in Alzheimer’s disease. Int J Nurs Pract 1996;2:29–32. The Clifton Assessment Procedures (20) Jorm AF, Christensen H, Henderson AS, et for the Elderly al. Informant ratings of cognitive decline of elderly people: relationship to longitudinal (A.H. Pattie and C.J. Gilleard, 1975) change on cognitive tests. Age Ageing Purpose 1996;25:125–129. The Clifton Assessment Procedures for the El- (21) Mackinnon A, Mulligan R. Combining derly (CAPE) evaluates the presence and severity cognitive testing and informant report to of impairment in mental and behavioral func- increase accuracy in screening for dementia. Am J Psychiatry tioning. It was intended for elderly long-term 1998;155:1529–1535. psychiatric patients (1). (22) Flicker L, Logiudice D, Carlin JB, et al. The predictive value of dementia screening Conceptual Basis instruments in clinical populations. Int J This scale indicates the patient’s likely level of Geriatr Psychiatry 1997;12:203–209. dependency, suggesting the “degree of support  Mental Status Testing 457 typically provided by society in association with Exhibit 8.11 Summary Content of the Clifton given levels of disability” (1, p10). It was de- Assessment Procedures for the Elderly signed to be brief enough for practical use with COGNITIVE ASSESSMENT SCALE elderly patients and applicable across a wide range of abilities. The CAPE was named after Information and Orientation the Clifton Hospital in York, England. Correct recall of: Name: Hospital/address: Colour of British flag: Description Age: City: Day: The CAPE consists of two components: the Cog- D.o.B.: P.M.: Month: nitive Assessment Scale (CAS) and the Behavior Ward/place: U.S. President: Year: Rating Scale (BRS). The former was first pub- Mental Ability lished as the Clifton Assessment Schedule (2), Ability to count 1 to 20 (timed) whereas the latter was derived from the Stockton Reciting alphabet (timed) Geriatric Rating Scale (3). Write name The CAPE is generally administered by Reading a list of pre-selected words nurses treating a patient. The complete schedule is shown in the manual of the CAPE (1, pp3–7); Psychomotor a summary of its content is given in Exhibit The patient is asked to draw around a pre-printed 8.11. It includes a 12-item information and ori- spiral maze, avoiding “obstacles” drawn in the pathway; performance is timed. entation subtest (taking the form of questions such as “What is your date of birth?”), a brief BEHAVIOR RATING SCALE mental abilities test (e.g., “Will you count up Physical Disability from 1 to 20 for me—as quickly as you can?”), Bathing and dressing and a psychomotor performance test that in- volves tracing a line through a maze. The time Walking ability taken for the psychomotor maze test and the Continence number of errors on the other tests are con- Staying in bed during the day verted into a CAS score out of 12. The conver- Confusion (unable to find way around, loses things) sion is explained in the manual (1, p12). Pattie Appearance and Gilleard recommend a cutting-point of 8, Apathy commenting that scores of 7 and lower generally Would need supervision if allowed outside indicate dementia or acute organic brain syn- Helps out in the home or ward drome. Keeps occupied The BRS contains 18 items and is completed Socializes by relatives or staff familiar with the patient’s behavior. It covers physical disability including Accepts suggestions to do things performance of activities of daily living, apathy, Communication Difficulties communication difficulties, and social distur- Understands what is communicated to him/her bance. BRS scores range from 0 to 36, with Able to communicate higher scores indicating greater disability (1, Social Disturbance pp7–9). Objectionable to others during day Scores on the two components are trans- Objectionable to others during night ferred onto a report form that summarizes raw Accuses others of wrongdoing scores in a five-category grading of the patient’s level of dependency and hence the support the Hoards apparently meaningless items patient is likely to require (1, pp11–12). Grade A Sleep pattern represents no mental impairment and no signifi- Adapted from Pattie AH, Gilleard CJ. Manual of the Clifton cant behavioral disability; grade B represents Assessment Procedures for the Elderly (CAPE). Sevenoaks, mild impairment in both areas that will require Kent, UK: Hodder and Stoughton, 1979:3–9. 458 Measuring Health some support for people living in the commu- ated with subsequent mortality (6); in another nity. Grade C includes medium levels of impair- sample, both cognitive and behavior scores pre- ment requiring considerable support for dicted mortality (7). community living. Grade D includes those with The validity of the I/O subtest of the CAS marked impairment and dependency; people in has been studied separately. Comparing 33 de- this category are usually institutionalized. Grade mentia patients and 67 psychiatric patients, the E includes those with maximal impairment, typ- I/O had a sensitivity of 91% and a specificity of ical of psychogeriatric patients requiring a great 93% (2, Table II). Figures of 87% and 97% deal of nursing attention and care (1, p10). were reported at a cutting-point of 7/8 for a general practice study (8, Table V). The mental Reliability ability subtest had lower sensitivity and speci- Pattie and Gilleard reported test-retest reliability ficity: 52% and 89%, respectively. A subse- results for the CAS scales from a several studies. quent study replicated these results (9). A Results for the information/orientation (I/O) cutting-point of 7/8 resulted in a sensitivity of scale ranged from 0.79 to 0.90; for the mental 80% for discriminating patients with dementia ability scale results ranged from 0.61 to 0.89, from those with functional psychiatric disorders and for the psychomotor test, the results ranged such as affective illness or schizophrenia (10, from 0.56 to 0.79 (1, p21). p458). Specificity was 85%. In the same study, Inter-rater reliability for the BRS subscales scores on the CAS successfully predicted poor were reported from five studies. Correlations for outcome at the end of the two-year study. Black physical disability ranged from 0.70 to 0.91; ap- et al. reported sensitivity and specificity of the athy results ranged from 0.81 to 0.87; communi- I/O subtest as somewhat lower than those for cation from 0.45 to 0.72, whereas social the Mini-Mental State Exam (MMSE) or the disturbance ranged from 0.69 to 0.88 (1, Table CAMCOG (11, Table II). Further examination 7). Inter-rater agreement for the CAS was 0.99 of this was undertaken by Jagger et al., who in one study (4, Table 1). proposed altering the cutting-point from 7/8 to Internal consistency of the CAPE has been re- 8/9 (12, p208). viewed in several ways. The intercorrelations The information/orientation score of the among the subscale scores range from 0.30 to CAS correlated 0.90 with the Wechsler Mem- 0.78 (1, Table 14). Lesher and Whelihan tested ory Scale (N = 33) and gave an 84.5% correct reliability of the information/orientation section classification into organic or functional disor- on 36 residents of long-term care facilities (mean ders as rated by the Wechsler scale (1, p23). In age, 85 years). Test-retest reliability at two to three studies, correlations between the I/O test four weeks was 0.84, and alpha was 0.77 (5, and the Wechsler Adult Intelligence Scale Table 1). These results were lower than those (WAIS) ranged from 0.22 to 0.37 with the obtained for the other scales included in Lesher WAIS Verbal scale and 0.51 to 0.74 with its and Whelihan’s review. Performance scale. Correlations for the psy- chomotor score were similar, whereas those be- Validity tween the mental ability scale and the WAIS All subscores showed clear distinctions between Performance scale were lower, at 0.35 to 0.54 healthy elderly people, people under care in the (1, p24). The I/O scale correlated 0.90 with the community by the social services agencies, and MMSE and −0.52 with the Informant Ques- psychiatric patients (1, Tables 3 to 5; 2, Table 1). tionnaire on Cognitive Decline (IQCODE) in a Each subscore of the CAS predicted subsequent study of elderly general practice patients (13, likelihood of discharge (1, Table 1). Predictive Table 2). validity has also been tested by comparing CAPE Factor analyses of the BRS identified three scores with subsequent mortality. Total CAS main factors. The first combined the physical scores and the mental ability score were associ- disability and communication scales; the second  Mental Status Testing 459 included the apathy items, and the third con- greatly from reanalyses of the data using ROC tained the social disturbance items (1, p27). analyses to summarize the performance of the tests across all cutting-points. Furthermore, the Alternative Forms debate over the CAPE versus the MMSE appears An abbreviated version of the CAPE is designed to concern adequacy in estimating the prevalence for use in large-scale surveys. This includes the of dementia; this does not seem an appropriate I/O scale from the CAS and the physical disabil- application for a screening test. ity (PD) scale from the BRS. Scores are formed The CAPE provide reliable estimates of cog- by subtracting the PD scores from the I/O (range nitive and behavioral impairment for institution- +12 to −12). These scores are converted into five alized elderly populations. As a screening test for categories (14). Test-retest reliability ranged community use, the CAPE would probably be from 0.82 to 0.89 in three studies (14, Table 2) less adequate than other available instruments and was 0.91 in another (15). Concurrent and such as the MMSE. predictive validity have also been demonstrated in several studies (14–16), although McPherson et al. found that the PD score added little beyond References that of the I/O and that classification of patients by severity of dementia showed considerable (1) Pattie AH, Gilleard CJ. Manual of the overlap between groups in the middle of the Clifton Assessment Procedures for the El- derly (CAPE). Sevenoaks, Kent, UK: severity spectrum (16, p90). Hodder and Stoughton, 1979. (2) Pattie AH, Gilleard CJ. A brief Reference Standards psychogeriatric assessment schedule: Normative data were reported in the CAPE man- validation against psychiatric diagnosis and ual, which shows mean scores, medians, and in- discharge from hospital. Br J Psychiatry terquartile ranges on each subscale for well 1975;127:489–493. people in the community and for a range of pa- (3) Meer B, Baker JA. The Stockton Geriatric tient types (1, Tables 3–5 and p29). Rating Scale. J Gerontol 1966;21:392–403. Commentary (4) Smith AHW, Ballinger BR, Presly AS. The The CAPE has been tested in several studies us- reliability and validity of two assessment scales in the elderly mentally handicapped. ing large samples of patients. The results show Br J Psychiatry 1981;138:15–16. good reliability and high sensitivity and speci- (5) Lesher EL, Whelihan WM. Reliability of ficity when used with psychiatric inpatients. It mental status instruments administered to has mainly been tested on hospital populations nursing home residents. J Consult Clin but its performance on community samples re- Psychol 1986;54:726–727. mains unknown. (6) Gamsu CV, McLaren SM, McPherson FM. The issue has been raised of how to score the Prediction of survival in dementia by CAPE when a patient, because of blindness or cognitive tests: a six-year follow-up. Br J hands disabled by arthritis, cannot complete the Clin Psychol 1990;29:99–104. maze test. The original approach of awarding (7) Moran SM, Cockram LL, Walker B, et al. zero may lead to falsely classifying physical diffi- Prediction of survival by the Clifton Assessment Procedures for the Elderly culties as a cognitive problem; prorating the (CAPE). Br J Clin Psychol score based on scores in other parts of the CAPE 1990;29:225–226. does not work well (17). There have been exten- (8) Clarke M, Jagger C, Anderson J, et al. The sive but rather fruitless discussions over the ade- prevalence of dementia in a total quacy of the I/O subtest as a screening test population: a comparison of two screening compared with the MMSE as well as the best instruments. Age Ageing cutting-points. These discussions would benefit 1991;20:396–403. 460 Measuring Health (9) Pattie AH, Gilleard CJ. The Clifton the identification of specific types of dementia” Assessment Schedule-further validation of a (1, p700). psychogeriatric assessment schedule. Br J Psychiatry 1976;129:68–72. Conceptual Basis (10) Pattie AH, Gilleard CJ. The two-year Roth argued that a complete picture of dementia predictive validity of the Clifton requires three types of instrument: cognitive Assessment Schedule and the Shortened screening tests, dementia severity tests, and be- Stockton Geriatric Rating Scale. Br J Psychiatry 1978;133:457–460. havior rating scales that test problems in activi- (11) Black SE, Blessed G, Edwardson JA, et al. ties of daily living (ADL). The CAMDEX Prevalence rates of dementia in an ageing incorporates these in a single instrument (1, population: are low rates due to the use of p698). The procedure was developed from a re- insensitive instruments? Age Ageing view of weaknesses in existing scales and aims 1990;19:84–90. “to remedy gaps in the existing standardised in- (12) Jagger C, Clarke M, Anderson J, et al. terviews and scales of measurement” (1, p700). Dementia in Melton Mowbray—a validation of earlier findings. Age Ageing Description 1992;21:205–210. The CAMDEX includes three components: a (13) Bowers J, Jorm AF, Henderson S, et al. clinical interview, a set of cognitive tests, and an General practitioners’ detection of depression and dementia in elderly interview with a relative to obtain independent patients. Med J Aust 1990;153:192–196. information on the patient’s past and present (14) Pattie AH. A survey version of the Clifton state. The cognitive tests include existing instru- Assessment Procedures for the Elderly ments such as the Mini-Mental State Examina- (CAPE). Br J Clin Psychol tion (MMSE), the Blessed Dementia Score, and 1981;20:173–178. Hachinski’s Ischemia Scale. The three compo- (15) McPherson FM, Tregaskis D. The short- nents may be divided into eight sections (1, term stability of the survey version of pp701–702): CAPE. Br J Clin Psychol 1985;24:205–206. 1. Section A covers symptoms of severe men- (16) McPherson FM, Gamsu CV, Kiemle G, et tal conditions, past history and family his- al. The concurrent validity of the survey version of the Clifton Assessment tory, of mental disorders. Procedures for the Elderly (CAPE). Br J 2. Section B (67 items) is a cognitive exami- Clin Psychol 1985;24:83–91. nation, the “CAMCOG.” This incorpo- (17) McPherson FM, Gamsu CV, Cockram LL, rates an extended version of the MMSE et al. Use of the CAPE Pm test with (19 items), supplemented by additional disabled patients. Br J Clin Psychol items, because the MMSE does not sample 1986;25:145–146. certain functions such as abstract thinking or perception that are relevant to diagno- sis and are included in DSM-III criteria. Further, “many functions are assessed by The Cambridge Mental Disorders of the MMSE in insufficient detail. For ex- the Elderly Examination ample, memory is assessed only by the (Sir Martin Roth, 1986; 1998) repetition and recall of three words; we Purpose have added items covering remote and re- The Cambridge Mental Disorders of the Elderly cent memory and the recall of new infor- Examination (CAMDEX) is a clinical interview mation . . . [Thus the CAMCOG] assesses schedule that standardizes the information col- orientation, language, memory, praxis, at- lected in a routine diagnostic interview for de- tention, abstract thinking, perception and mentia. It “focuses on the diagnosis of dementia, calculation.” (1, p701). with particular reference to its mild forms and to 3. Section C includes the interviewer’s obser-  Mental Status Testing 461 vations on the patient’s appearance, be- of 79/80 (1, p703). Scores for the Mental Status havior, mood, and level of consciousness. Questionnaire (3) can also be derived from the It is completed at the end of the interview. CAMCOG items. 4. In section D, a physical examination pro- The CAMDEX is generally administered by a vides information to differentiate between psychiatrist; the interview with the patient takes primary and secondary dementias. It cov- about 60 minutes, the informant interview takes ers blood pressure, reflexes, gait, defects about 20 minutes (2, p4). At the end of the inter- of hearing or sight, tremor, and parkin- view, the psychiatrist makes a diagnosis based sonism. on operational criteria that are described in de- 5. Section E covers laboratory tests: blood tail in the manual (2, pp59–77). Eleven diagnos- analyses and radiological investigations. tic categories are specified: four categories of 6. Section F lists medications currently taken dementia, two of clouding or delirium, and a by the subject. range of other psychiatric states (e.g., depres- 7. Section G records further information that sion, paranoia). The severity of dementia and de- arises spontaneously during the interview; pression is graded on five-point scales (1, p702). it is likely to be of particular use in classi- Aside from the CAMCOG, scores can be derived fying atypical and difficult cases. to indicate organicity, composed of 18 items 8. Section H includes a structured interview from the patient interviews, 14 from the relative with a relative or caregiver who knows the interview, and two based on the interviewer’s person well. It incorporates the Blessed observations. A multi-infarct scale, designed to Dementia Score. The Hachinski Ischemia distinguish multi-infarct dementias, includes Score is also incorporated into the eight items from the relative’s interview, two CAMDEX. from the patient interview, and two from the in- 9. Section I includes additional information terviewer’s observations. Finally, a depression from the informant (2, p3). scale includes nine items from the patient inter- view and four from the relative’s interview (4, The original version was revised in 1998 to p404). incorporate new diagnostic criteria for dementia (e.g., Lewy body and frontal dementias), fuller Reliability coverage of remote memory and executive func- An inter-rater reliability among four psychia- tion in the CAMCOG, and enhanced data entry trists, who each interviewed 40 patients, showed capabilities (2). The complete CAMDEX is too median agreement phi coefficients of 0.94 for long to reproduce here; it is shown in Roth et al. section A, 0.90 for section B, 0.83 for section C, (2, pp10–55) and can be purchased from the and 0.91 for section H (1, p703). There was Cambridge Department of Psychiatry (see Ad- complete agreement among raters on the classifi- dress section). Exhibit 8.12 summarizes the con- cation of patients as demented or as normal, but tent of the CAMCOG section from the the phi coefficient of agreement was reduced to CAMDEX-R. The CAMCOG was designed to 0.63 when dementia was divided into four sub- identify dementia in its early stages and to categories. Hendrie et al. calculated inter-rater record cognitive decline. Content was guided by reliabilities across 40 psychogeriatric patients the DSM-IV and ICD-10 criteria and incorpo- and 15 controls who were interviewed by clini- rated items from other widely used scales, in- cians of varying backgrounds (4). Phi coeffi- cluding the MMSE and Hodkinson’s cients for the items of the CAMCOG ranged Abbreviated Mental Test (2, p81). The CAM- from 0.50 to 1.00; the median was 0.86. The COG provides scores for each function shown in Pearson correlation between raters for total the exhibit, and an overall score. This runs from score on the CAMCOG was 0.88. Phi coeffi- 0 to 105, as the executive function section over- cients for the patient interview, interviewer’s ob- laps with other sections and is not counted in the servations, and informant interview ranged from total. Roth et al. recommended a cutting-point 0.50 to 1.00, and Pearson correlations between 462 Measuring Health Exhibit 8.12 Summary of the Content of the CAMCOG Section of the CAMDEX Cognitive function Summary of content Maximum score Orientation Date; season; place 10 Language: comprehension Following instructions; reading comprehension; 30 and expression logical reasoning; naming; writing Memory: remote, recent, Historical people; current politicians; immediate 27 and learning and delayed recall recall Attention and calculation Counting backward; serial sevens; calculation 9 Praxis Copying designs (pentagons, spirals, clock 12 drawing); following instructions Abstract thinking Similarities 8 Perception Recognizing famous people and objects 9 Executive function Animal naming; similarities; developing 28 ideas; visual reasoning. Derived from: Roth M, Huppert FA, Mountjoy CQ, Tym E. CAMDEX-R. Cambridge, Cambridge University Press, 1998. raters for total score on these sections ranged had a sensitivity of 94% and a specificity of from 0.81 to 0.99 (4, Table 2). In a Spanish 91%. However, the CAMCOG missed 21% of study, CAMDEX interviews were observed and cases meeting the DSM-III criteria for dementia: the agreement between interviewers and ob- these were mostly mild cases. With a 79/80 cut- servers was almost perfect (W index of concor- off, sensitivity and specificity were both 100% dance = 0.99) (5, Table 4). in one study of 28 psychogeriatric patients and Data on test-retest reliability are limited be- 15 community controls (4, p405). The MMSE cause of the difficulty of administering such a showed an 82% sensitivity and 100% specificity lengthy test twice to elderly people. (with a 23/24 cutoff ) in the same study. The or- ganicity, multi-infarct dementia, and depression Validity scales succeeded in distinguishing Alzheimer’s Roth et al. reported 92% sensitivity and 96% disease, multi-infarct dementia, and depression, specificity for the CAMCOG, compared with a respectively, from all other conditions (4, Table clinical diagnosis, although it appears from the 5). The MMSE and CAMCOG scores correlate text of the report that these figures may actually highly (0.95 in one study: 7, p984), but ROC refer to positive and negative predictive values analyses showed the CAMCOG superior to the rather than sensitivity and specificity (1, p703). MMSE (area under the ROC curve = 0.95 and The revised version had a sensitivity of 90% at a 0.90, respectively) (8, p2083). specificity of 88% (2, Table 6). Mean scores for A Dutch study found that the DSM-III-R cri- depressed patients did not differ significantly teria both identified more people as having de- from those of unaffected people; mean scores for mentia than the CAMDEX (5.1% versus 2.6%) both groups were clearly different than those for and also rated their dementia as more severe the demented patients (1, Table III). Compared than the CAMDEX did. Kappa agreement on with Copeland’s Geriatric Mental State Exami- severity was only 0.36 (9, p192). A kappa coeffi- nation, the CAMCOG showed a sensitivity of cient of 0.83 has been reported between the 97% and a specificity of 91% (6, pp194–195). CAMDEX and a clinical diagnosis (5). Sensitiv- These results were considerably better than ity was 90% and specificity 76% in a small those obtained by the MMSE but were compara- study of dementia. The equivalent results for the ble with those obtained by the Clifton Assess- Blessed Dementia Scale were 90% and 85%; ment Procedures for the Elderly (CAPE), which those for the Modified Mini-Mental State test  Mental Status Testing 463 were 80% and 96% (10, Table V). In a Spanish used and tested (15); an Italian version has been study, CAMDEX diagnoses agreed closely with described (14). neurologists’ diagnoses of dementia, depression or normal based on the DSM-III-R determina- Reference Standards tions: kappa = 0.95 (5, p135). Agreement for Roth et al. reported mean values and standard subtypes of dementia was 0.73. Sensitivity for deviations for the CAMCOG scales by age, edu- the CAMCOG was 91% at a specificity of 75% cation, and dementia status (2, Tables 2 to 5). (5, Table 3). O’Connor et al. reviewed predictive validity Commentary by reassessing 137 cases one year after an initial The CAMDEX was designed to respond to limi- CAMDEX assessment. Very few errors in the tations of the MMSE and other widely used original assessment were noted: 97% of the di- scales. “It replicates a full clinical assessment . . . agnoses were confirmed (11, p79). Depression in a structured, orderly fashion, but this breadth had been missed in several minimally demented and depth of coverage make it an expensive in- subjects and four of 67 cases of mild dementia strument to use” (16, p218). The CAMCOG were altered on review (11, p81). A subsequent cognitive capacity component succeeds in this two-year follow-up assessment by the same team and has been shown superior to the MMSE. This confirmed the appropriateness of the initial diag- is not surprising in that it is longer, has a wider noses; O’Connor et al. provided case histories to range of scores, and covers a wider range of cog- illustrate the few apparently anomalous changes nitive functions; it detects mild levels of impair- (12). Tierney et al. found that information from ment and avoids ceiling effects (1; 8). the CAMDEX History section contributed to The value of including assessments by rela- identifying people who would develop dementia tives in addition to self-assessments was indi- over a 2-year follow-up period (13). cated by the finding that relatives were better The CAMCOG correlated −0.70 with the able to assess cognitive ability although they Blessed Dementia Score and −0.78 with a clini- were often unaware of the patient’s depression. cal rating of dementia severity (1, p704). The The information provided by informants gener- correlation with the Blessed score in another ally agreed closely with the observations made study was 0.75 (6, p195). CAMCOG scores cor- by clinicians (17). related more highly with an organicity score Because of its length and the need for a psychi- (−0.63) than with a depression score (0.22) (1, atrist to lead the interview, the CAMDEX is gen- Table V). erally applied following an initial screening test. However, it has been used in epidemiological Alternative Forms studies in England, and the CAMCOG section is Owing to the length of the CAMDEX, Neri et often used as a separate assessment instrument. al. proposed an abbreviated version (“Short- CAMDEX”) that includes 106 of the 340 items Address and requires about 30 minutes to complete Dr. Felicia A. Huppert, Department of Psychia- (14). In a comparison study, the cognition score try, Box 189, Addenbrooke’s Hospital, Hills Rd, of the full CAMDEX has a sensitivity of 97% Cambridge CB2 2QQ, England. and a specificity of 79%; figures for the shorter version were 89% and 93% (unfortunately, fig- ures were not provided for cutting-points References equivalent to those for the longer form). For the (1) Roth M, Tym E, Mountjoy CQ, et al. organicity score, the full CAMDEX has a sensi- CAMDEX: a standardised instrument for tivity of 81% and specificity of 100%; the the diagnosis of mental disorder in the el- short form was 85% sensitive and 93% specific derly with special reference to the early (14, Table 3). detection of dementia. Br J Psychiatry The Spanish version has been extensively 1986;149:698–709. 464 Measuring Health (2) Roth M, Huppert FA, Mountjoy CQ, et al. diagnosed in the community using the CAMDEX-R: the Cambridge examination Cambridge Mental Disorder of the Elderly for mental disorders of the elderly— Examination. Acta Psychiatr Scand revised. Cambridge: Cambridge University 1991;83:41–45. Press, 1998. (13) Tierney MC, Szalai JP, Snow WG, et al. (3) Kahn RL, Goldfarb AI, Pollack M, et al. The prediction of Alzheimer disease. The Brief objective measures for the role of patient and informant perceptions determination of mental status in the aged. of cognitive deficits. Arch Neurol Am J Psychiatry 1960;117:326–328. 1996;53:423–427. (4) Hendrie HC, Hall KS, Brittain HM, et al. (14) Neri M, Roth M, Mountjoy CQ, et al. The CAMDEX: a standardized instrument Validation of the full and short forms of for the diagnosis of mental disorder in the the CAMDEX interview for diagnosing elderly: a replication with a US sample. J dementia. Dementia 1994;5:257–265. Am Geriatr Soc 1988;36:402–408. (15) Vilalta J, Llinás J, López Pousa S, et al. The (5) Llinás Regla J, López Pousa S, Vilalta Cambridge Mental Disorders of the Elderly Franch J. The efficiency of CAMDEX in Examination. Validation of the Spanish the diagnosis of dementia and its sub-types. adaptation. Neurologia 1990;5:117–120. Neurologia 1995;10:133–138. (16) O’Connor DW. The contribution of (6) Blessed G, Black SE, Butler T, et al. The CAMDEX to the diagnosis of mild diagnosis of dementia in the elderly. A dementia in community surveys. Psychiatr J comparison of CAMCOG (the cognitive Univ Ott 1990;15:216–220. section of CAMDEX), the AGECAT (17) O’Connor DW, Pollitt PA, Brook CPB, et program, DSM-III, the Mini-Mental State al. The validity of informant histories in a Examination, and some short rating scales. community study of dementia. Int J Geriatr Br J Psychiatry 1991;159:193–198. Psychiatry 1989;4:203–208. (7) Johnston B, Scott N, Po ALW, et al. Psychometric profiling of the elderly using the Cambridge Cognitive Examination. Conclusion Ann Pharmacother 1995;29:982–987. (8) de Koning I, van Kooten F, Dippel DWJ, et The field of cognitive assessment has been ac- al. The CAMCOG: a useful screening tively studied in recent years. Not only is there a instrument for dementia in stroke patients. proliferation of scales, but continuing attention Stroke 1998;29:2080–2086. has been paid to the conceptual and theoretical (9) Boersma F, Eefsting JA, van den Brink W, issues underlying them. Validation studies are et al. Prevalence of dementia in a rural Netherlands population and the influence generally well done; many of the instruments of DSM-III-R and CAMDEX criteria for have received extensive psychometric testing. It the prevalence of mild and more severe will be useful, however, to see the application of forms. J Clin Epidemiol 1998;51:189–197. some of the more recent statistical procedures to (10) Cappeliez P, Quintal M, Blouin M, et al. summarizing the results of testing; validity is Les propriétés psychométriques de la commonly described in terms of sensitivity and version française du Modified Mini-Mental specificity, for example, rather than in terms of State (3MS) avec des patients âgés suivis en ROC analyses, which simplify comparisons be- psychiatrie gériatrique. Can J Psychiatry tween scales. Item response analyses are only be- 1996;41:114–121. ginning to be applied in this field. (11) O’Connor DW, Pollitt PA, Hyde JB, et al. In addition to the reviews in this chapter, A follow-up study of dementia diagnosed scales in other chapters contain cognitive com- in the community using the Cambridge Mental Disorders of the Elderly ponents that may be suitable for particular appli- Examination. Acta Psychiatr Scand cations. These include the Comprehensive 1990;81:78–82. Assessment and Referral Evaluation (see Chap- (12) O’Connor DW, Pollitt PA, Jones BJ, et al. ter 10), the Functional Activities Questionnaire Continued clinical validation of dementia (FAQ), and Functional Status Rating System  Mental Status Testing 465 (both reviewed in Chapter 3). The FAQ is often chological tests are worthy of mention. First, used as a cognitive screening test, although it is Martha Storandt et al. developed a brief battery expressed in the format of an instrumental activ- of four tests to differentiate persons in the early ities of daily living questionnaire. This feature stages of Alzheimer’s disease from mentally may make it more acceptable to respondents healthy older adults. The tests were selected than the quiz format of memory questions and through discriminant analysis in a case-control arithmetic tests that characterize most of the in- study and included the logical memory and men- struments described in the present chapter. tal control subtests of the Wechsler Memory Scale, form A of the Trailmaking Test, and word fluency for letters S and P. This battery took ap- Additional Instruments proximately 10 minutes to administer and could There are numerous other cognitive function be handled by research assistants. The initial va- tests that we have not included as full reviews, lidity figures presented by Storandt were out- for reasons of space or of their early stages of standing: 98% correct classification in the development. They are mentioned here to guide challenging task of identifying patients in the the reader who has not found a suitable method early stages of Alzheimer’s disease (7). Replica- among the instruments that we review in this tion confirmed that the battery did distinguish chapter. mentally healthy people from those with early dementia, but that it did not distinguish different Screening Tests types of dementia (8). In a later report, Storandt The first category of additional instruments in- and Hill retested the battery on a larger sample cludes several screening tests. The Cross- and again showed good discrimination ability, Cultural Cognitive Examination (CCCE) is a but the tests selected by the new discriminant brief screening test designed to identify all forms analysis changed so that Storandt recommended of dementia; it can be applied in nonliterate pop- changing to the newly identified tests (9, p385). ulations by lay interviewers (1–3). As an innova- The Storandt Battery is worth consideration in tive way to achieve validity in a brief instrument, screening for the early stages of dementia. Sec- the CCCE contains a five-minute screening test ond, Eslinger et al. also proposed a brief battery plus a 20-minute mental status examination ap- of tests to distinguish dementia from normal ag- plied to those who fail the screening component. ing (10). The Eslinger battery includes the Ben- The screen is intended to be sensitive; the exami- ton Visual Retention Test (a test of short-term nation, specific. Early results seemed good and memory for designs), the Benton Controlled interested readers should monitor the further de- Oral Word Association Test (which requires the velopment of this instrument. Another instru- person to name as many words that begin with a ment designed for use in nonliterate groups is the selected letter as possible in a set time), and the ten-item Elderly Cognitive Assessment Ques- Benton Temporal Orientation Test, which as- tionnaire (ECAQ), which drew items from the sesses the patient’s accuracy of recall of the MMSE and GMS. Sensitivity was reported as month, day, year, and time of day. The battery 85% at a specificity of 92% (4, p119). An eight- takes 10 to 15 minutes to complete. Validity is item screening test described by Kokmen et al. is comparable with that of the Storandt battery (8). called the Short Test of Mental Status (STMS) Finally, Pfeffer et al. assembled three tests into a (5). It contains one item on each of orientation, brief battery to serve as a population screen for attention, immediate recall, calculation, abstrac- mild dementia in community surveys (11; 12). tion, construction (using the clock test), infor- This was named the Mental Function Index mation, and delayed recall. The test takes about (MFI) and is a weighted combination of scores five minutes to administer and preliminary valid- from the Mini-Mental State Exam, the Smith ity results appear good (sensitivity 86%, speci- Symbol Digit Modalities test, and the Raven ma- ficity 94%) (6). trices subtest B. In effect, it is an extension of the Three brief batteries that combine neuropsy- MMSE designed to overcome the MMSE’s 466 Measuring Health weaknesses. The instrument is applied by a Mental State (3MS) test and the Hasegawa De- nurse clinician in 15 to 25 minutes. Reliability mentia Screening Scale and were tested simulta- results were high: a test-retest reliability of 0.97; neously in the United States and in Japan. These sensitivity was 0.93 at a specificity of 0.80 (12, early trials suggest that both sensitivity and Table 4). Further validation results include a sen- specificity lie in the mid 90s, higher than the sitivity of 71% and a specificity of 91% (13)— level achieved by the 3MS (17, Table 4). A little different from the MMSE alone. Some feature of this scale and of the 3MS is the avail- commentators have been critical. Mowry and ability of high-quality materials, including ex- Burvill, for example, found that the MFI did not amination questions, for training interviewers. differentiate mild dementia from other diag- Careful attention has been paid to detail in these noses, and they concluded that “in its present instruments, in a manner reminiscent of scales form, the MFI cannot be recommended as a such as the Short-Form-36 Health Survey re- screening instrument for mild dementia” (14, viewed in Chapter 10. It will be interesting to p332). see further information on reliability and valid- The East Boston Memory Test has been ity over the coming years. shown to have good validity (15). It consists of a brief story containing three sentences, each of Instruments for Clinical Application which includes two ideas. The subject retells the Several measures have been developed for use story immediately after hearing it, and errors are with patient samples. These include the Stroke scored. This test has been shown superior to the Unit Mental Status Examination (SUMSE), Short Portable Mental Status Questionnaire as a which is a bedside evaluation of cognitive func- screen for dementia (16). tion following a stroke. It focuses on the types The Delayed Word Recall Test is a memory of deficit caused by strokes (e.g., language disor- task designed for large-scale screening to dis- ders, visuoperceptual deficits, and memory dis- criminate patients with dementia from mentally orders). Evidence for validity and reliability are normal elderly subjects (13). Disturbance of re- available (18). The Neurobehavioral Cognitive cent memory is the most common deficit seen in Status Examination (NCSE) assesses multiple early Alzheimer’s disease, mainly learning new domains of cognitive functioning in patient pop- material and retrieving it with minimal cuing. ulations and presents these in profile form rather The test requires that the subject makes up sen- than as a summary score (19). The Executive In- tences that incorporate the words to be remem- terview (EXIT) specifically assesses those cogni- bered, followed by a delay period, then a free tive processes that “orchestrate relatively simple recall phase. Early results suggest that the instru- ideas, movements, or actions into complex, goal- ment is highly sensitive in discriminating sub- directed behaviors” (20, p10). Disorders of jects with mild Alzheimer’s disease from those executive functioning are associated with frontal mentally normal, although it may not be suitable lobe lesions and commonly occur in Alzheimer’s for patients with more severe dementia, because disease. the requirement of making up a sentence may Reisberg has developed two measures of the prove too difficult. severity of cognitive decline: the Brief Cognitive A more recent test that has attracted consid- Rating Scale (BCRS) and the Global Deteriora- erable attention is the Cognitive Abilities tion Scale (21–23). The BCRS is a clinical rating Screening Instrument (CASI) developed by Eve- scale that summarizes the extent of cognitive im- lyn Teng et al. This may be the Swiss army knife pairment on five clinical axes: concentration, re- of the dementia field; it is designed to be cross- cent memory, past memory, orientation and culturally applicable and serves as a screening functioning, and self-care. Each is rated on a test for dementia and as an indicator of disease seven-point severity scale (21, pp29–31). The progression; it also provides a profile of impair- Global Deterioration Scale provides a seven- ment among various cognitive domains (17). category overall rating of the progression of de- Items were drawn from the Modified Mini- mentia, ranging from no cognitive decline to  Mental Status Testing 467 very severe decline. Clinical markers of each cognitive performance test, and a series of diag- stage are provided (21, pp32–35). nostic algorithms that guide differential diag- The Hachinski et al. Ischemic Score is a list noses. Early results have shown satisfactory of clinical signs that distinguish between the de- reliability and concordance with independent di- generative dementias such as Alzheimer’s dis- agnoses (33). Other diagnostic instruments in- ease and vascular or multi-infarct dementias clude the British Present State Examination (24). It has been extensively tested, criticized, (PSE), the American Diagnostic Interview and revised over the years (25–27), but remains Schedule (DIS), and the Geriatric Mental State one of the most widely quoted instruments of its Examination (GMS), which was derived from kind. the PSE. In turn, the Canberra Interview for the Shader et al. developed the Sandoz Clinical Elderly was based on the GMS and includes ma- Assessment–Geriatric (SCAG) to assess change terial from ADL scales and from the Compre- following treatment (28). This is an 18-item clin- hensive Assessment and Referral Evaluation ical rating scale that covers agitation, cognitive (CARE) instrument described in Chapter 10 dysfunction, depressed mood, and withdrawal. (34–36). These instruments provide complete It is widely used as an outcome measure in drug diagnostic algorithms and sometimes use com- research (29), although it is not without its crit- puter scoring. They have often been examined ics (30). for reliability and validity and extend the field Cole et al. developed the Hierarchic Demen- of cognitive testing far beyond the scope of tia Scale to measure the severity of cognitive and those instruments we have selected to review in behavioral symptoms of dementia. In a manner this chapter. similar to that used by Katz in his Index of Ac- tivities of Daily Living (ADL), a hierarchy of cognitive development during childhood was References postulated, which is reversed during senile de- (1) Glosser G, Wolfe N, Albert ML, et al. Cross- cline (31, p298; 32). For example, in declining Cultural Cognitive Examination: validation language skills, nominal aphasia usually occurs of a dementia screening instrument for before paraphasia, which precedes the appear- neuroepidemiological research. J Am Geriatr ance of word substitutions, and then the use of Soc 1993;41:931–939. deformed words. Within each of the 20 cogni- (2) Glosser G, Wolfe N, Kliner-Krenzel L, et tive functions assessed, there is also a hierarchy al. Cross-Cultural Cognitive Examination of severity. If the response to the initial item in a performance in patients with Parkinson’s disease and Alzheimer’s disease. J Nerv list is incorrect, the interviewer keeps testing on Ment Dis 1994;182:432–436. subsequently lower items in the subscale until (3) Wolfe N, Imai Y, Otani C, et al. Criterion two consecutive items are answered correctly. validity of the Cross-Cultural Cognitive For more severely demented patients, the inter- Examination in Japan. J Gerontol viewer starts with easier items and proceeds up 1992;47:P289–P296. until two consecutive items are missed. Reliabil- (4) Kua EH, Ko SM. A questionnaire to screen ity appears good and there is some evidence for for cognitive impairment among elderly concurrent validity. people in developing countries. Acta Psychiatr Scand 1992;85:119–122. Diagnostic Instruments (5) Kokmen E, Naessens JM, Offord KP. A Finally, several major clinical assessments are short test of mental status: description and preliminary results. Mayo Clin Proc used in diagnosing dementia. The Structured In- 1987;62:281–288. terview for the Diagnosis of Dementia of the (6) Kokmen E, Smith GE, Petersen RC, et al. Alzheimer type (SIDAM) provides a differential The short test of mental status: correlations diagnosis of dementias according to DSM-III with standardized psychometric testing. and ICD10 criteria. It comprises a brief clinical Arch Neurol 1991;48:725–728. interview covering medical history, a standard (7) Storandt M, Botwinick J, Danziger WL, et 468 Measuring Health al. Psychometric differentiation of mild al. The Neurobehavioral Cognitive Status senile dementia of the Alzheimer type. Examination: comparison with the Arch Neurol 1984;41:497–499. Cognitive Capacity Screening Examination (8) Tierney MC, Snow G, Reid DW, et al. and the Mini-Mental State Examination in Psychometric differentiation of dementia: a neurosurgical population. Ann Intern replication and extension the findings of Med 1987;107:486–491. Storandt and coworkers. Arch Neurol (20) Teresi JA, Evans DA. Cognitive assessment 1987;44:720–722. measures for chronic care populations. In: (9) Storandt M, Hill RD. Very mild senile Teresi JA, Lawton MP, Holmes D, et al., dementia of the Alzheimer type: II. eds. Measurement in elderly chronic care Psychometric test performance. Arch populations. New York: Springer, Neurol 1989;46:383–386. 1997:1–23. (10) Eslinger PJ, Damasio AR, Benton AL, et al. (21) Reisberg B. The Brief Cognitive Rating Neuropsychologic detection of abnormal Scale and Global Deterioration Scale. In: mental decline in older persons. JAMA Crook T, Ferris S, Bartus R, eds. 1985;253:670–674. Assessment in geriatric (11) Pfeffer RI, Kurosaki TT, Harrah CH, et al. psychopharmacology. New Canaan, A survey diagnostic tool for senile Connecticut: Mark Powley, 1983:19–35. dementia. Am J Epidemiol (22) Reisberg B, Ferris S, Anand R, et al. 1981;114:515–527. Clinical assessments of cognition in the (12) Pfeffer RI, Kurosaki TT, Chance JM, et al. aged. In: Shamoian LA, ed. Biology and Use of the Mental Function Index in older treatment of dementia in the elderly. adults: reliability, validity and measurement Washington, DC: American Psychiatric of change over time. Am J Epidemiol Press, 1984:16–37. 1984;120:922–935. (23) Reisberg B, Schneck MK, Ferris SH, et al. (13) Knopman DS, Ryberg S. A verbal memory The Brief Cognitive Rating Scale (BCRS): test with high predictive accuracy for findings in primary degenerative dementia dementia of the Alzheimer type. Arch (PDD). Psychopharmacol Bull Neurol 1989;46:141–145. 1983;19:47–51. (14) Mowry BJ, Burvill PW. A study of mild (24) Hachinski VC, Iliff LD, Zilhka E. Cerebral dementia in the community using a wide blood flow in dementia. Arch Neurol range of diagnostic criteria. Br J Psychiatry 1975;32:632–637. 1988;153:328–334. (25) Huppert FA, Tym E. Clinical and (15) Scherr P, Albert M, Funkenstein HH, et al. neuropsychological assessment of Correlates of cognitive function in an elder- dementia. Br Med Bull 1986;42:11–18. ly community population. Am J Epidemiol (26) Small GW. Revised ischemic score for 1988;128:1084–1101. diagnosing multi-infarct dementia. J Clin (16) Albert M, Smith LA, Scherr PA, et al. Use Psychiatry 1985;46:514–517. of brief cognitive tests to identify (27) Dening TR, Berrios GE. The Hachinski individuals in the community with ischaemic score: a reevaluation. Int J clinically diagnosed Alzheimer’s disease. Geriatr Psychiatry 1992;7:585–590. Int J Neurosci 1991;57:167–178. (28) Shader RI, Harmatz JS, Salzman C. A new (17) Teng EL, Hasegawa K, Homma A, et al. scale for clinical assessment in geriatric The Cognitive Abilities Screening populations: Sandoz Clinical Assessment- Instrument (CASI): a practical test for Geriatric (SCAG). J Am Geriatr Soc cross-cultural epidemiological studies of 1974;22:107–113. dementia. Int Psychogeriatr 1994;6:45–58. (29) Venn RD. The Sandoz Clinical Assessment- (18) Hajek VE, Rutman DL, Scher H. Brief Geriatric (SCAG) scale: a general purpose assessment of cognitive impairment in psychogeriatric rating scale. Gerontology patients with stroke. Arch Phys Med 1983;29:185–198. Rehabil 1989;70:114–117. (30) Salzman C. The Sandoz Clinical Assessment- (19) Schwamm LH, van Dyke C, Kiernan RJ, et Geriatric scale. In: Crook T, Ferris S, Bartus  Mental Status Testing 469 R, eds. Assessment in geriatric (34) Kay DWK, Henderson AS, Scott R, et al. psychopharmacology. New Canaan, Dementia and depression among the elder- Connecticut: Mark Powley, 1983:53–58. ly living in the Hobart community: the (31) Cole MG, Dastoor DP. A new hierarchic effect of the diagnostic criteria on the approach to the measurement of dementia. prevalence rates. Psychol Med Psychosomatics 1987;28:298–304. 1985;15:771–788. (32) Cole MG, Dastoor DP, Koszycki D. The (35) Henderson AS. The Canberra Interview for Hierarchic Dementia Scale. J Clin Exp the Elderly: a new field instrument for the Gerontol 1983;5:219–234. diagnosis of dementia and depression by (33) Zaudig M, Mittelhammer J, Hiller W, et ICD-10 and DSM-III-R. Acta Psychiatr al. SIDAM—a structured interview for the Scand 1992;85:105–113. diagnosis of dementia of the Alzheimer (36) Mackinnon A, Christensen H, Cullen JS, et type, multi-infarct dementia and dementias al. The Canberra Interview for the Elderly: of other aetiology according to ICD-10 and assessment of its validity in the diagnosis of DSM-III-R. Psychol Med dementia and depression. Acta Psychiatr 1991;21:225–236. Scand 1993;87:146–151.  9 Pain Measurements here are special challenges in measuring pain. ports of pain reflect the combined influence of T Like depression and anxiety, pain is a private and internal sensation that cannot be directly the pain stimulus, environmental circumstances, and the characteristics of the individual experi- observed or measured; its measurement depends encing it. on the subjective response of the person experi- encing it. By contrast, physical disability can be measured more directly: we both define and Theoretical Approaches to Pain measure it in terms of observable behaviors (e.g., walking a given distance, climbing stairs). It is Most historical discussions of pain begin with also clear that pain is multidimensional; a single biomedical models, often tracing these back to assessment of intensity will not adequately re- Descartes who, around 1644, proposed an flect the contrast between, say, a toothache and a alarm bell model of pain. This posited that the burn. Finally, pain measurement is above all the nerve functioned like a tiny rope; the end at- area in which subjective reports represent a tached to the damaged part of the skin pulled blend of the strength of the underlying stimulus the other end located in a pain center in the and of the patient’s emotional response to it. brain (4, p14). The biomedical model treats This distinguishes pain from depression, for ex- pain as a sensory experience resulting from the ample. transmission of signals, originating from noci- Measuring depression also differs from mea- ceptors, indicating that tissue damage has oc- suring pain in that we consider depression as a curred. When the signal reaches the response, whereas we normally talk of pain as a somatosensory cortex in the brain it is inter- stimulus. This may be the nub of the problem in preted as painful, and this has obvious survival pain measurement; we try to infer pain (as a benefits of immediate recoil and subsequent fear stimulus) from the sufferer’s subjective response and avoidance. Biomedical models focus on the to it. But the way the pain is reported is influ- sensory processes of pain: nociception and the enced by many factors—biological, social, and neurological transmission of stimuli to recep- psychological. Biologically, there may not be a tors in the brain. In keeping with cartesian dual- linear relationship between pain and the extent ism, this model originally paid little attention to of tissue damage; minor damage can give rise to psychological, social, or behavioral aspects of intense pain and vice versa. Numerous individ- the pain response. Modern biomedical models ual and cultural factors, including gender, up- of pain have, of course, broadened this perspec- bringing, personality, and age, have been shown tive to acknowledge the importance of cogni- to influence a person’s response to pain (1–3). tion and emotions as modifiers of the pain Psychological factors also modify pain reac- response, but they still give primacy to sensa- tions, and these may vary independently of the tion: mainstream pharmacological treatments strength of the pain stimulus so that they cannot for pain ignore the emotional aspects and tackle be predicted from it. Thus, more than is the case the transmission of pain stimuli. Although suc- with other areas of subjective measurement, re- cessful in many cases, the growth in comple- 470  Pain Measurements 471 mentary and alternative medicine suggests the concerning cognitive and emotional factors orig- limitations inherent in this approach. Pain and inating in the brain: it was influenced by higher nociception are different: pain is conscious level brain processes. Following three decades of whereas nociception is not and each can occur investigation into this model, Melzack subse- without the other. Nociception is a transmission quently proposed a more general “neuromatrix mechanism whereas pain is a complex experi- model” that focused attention away from one ence of external and internal stimuli; it is an particular part of the spinal cord. The new emergent phenomenon in which the whole can- model arose from a discussion of how the body not adequately be reduced to its parts (1). experiences itself as a whole, and of the brain By contrast, psychodynamic theories of pain mechanisms that underlie this recognition of shifted the focus to the other extreme, away from self. Instead of a single center, Melzack pro- physical pathology to focus on the emotional di- posed a widespread network of neurons consist- mensions and functions of pain. These theories ing of feedback loops among the cortex, have been applied chiefly in discussions of persis- thalamus, and limbic system. The inputs to this tent pain that appears not to be explained by a system include sensory signals from the body physical stimulus. Freud, for example, argued (e.g., pain), stress response inputs from emotion- that persistent pain that seemed not to have a related areas of the brain such as the limbic sys- physical source could be explained by neurotic tem and cognitive inputs (e.g., memories of past reactions to an emotional loss or conflict. This experiences). Reacting to a painful stimulus, out- might occur through a conversion process in puts include pain perception, voluntary and in- which some unresolved conflict came to be ex- voluntary actions, and other functions such as pressed as physical pain (5). The theme of psy- endocrine responses to regulate stress. The re- chogenic pain draws attention to the range of peated cyclical processing and synthesis of im- possible functions that pain may serve for the pulses through the neuromatrix results in sufferer: perhaps an unconscious protection characteristic patterns of reactions for each indi- against having to do something they do not wish vidual, or their “neurosignature” (4). to do, or possibly as a way to gain attention. Thus, the neuromatrix model incorporates Although these theories may not command wide- the notion of learned responses to pain, influ- spread attention, they did highlight the impor- enced by environment and social factors. This tance of emotional factors in the pain experience, theme has been picked up in biopsychosocial challenging the adequacy of a purely biomedical models of pain, in which the pain experience is basis for determining approaches to treatment (5, seen as resulting from an interaction among p39). physical, psychological, and social influences. Melzack’s gate control theory of pain was the Predispositional and biological factors mold the first systematic attempt to integrate the physical person’s pain sensations, whereas psychological and psychological components of the pain re- factors influence their appraisal and perceptions sponse: something of a synthesis between the of pain, and social factors shape the behavioral biomedical and psychodynamic models (see the responses to pain (5, p42). This approach fo- review of the McGill Pain Questionnaire later in cuses less on disease than on illness, that is, the this chapter). In 1965, Melzack and Wall pro- subjective experience of disease. A connection posed the metaphor of a gate, located in the dor- was made here to the notion of illness behavior sal horns of the spinal column, which is opened (see the review of the Illness Behavior Question- or shut to permit or stop the transfer of a pain naire later in this chapter). As emphasized in the stimulus. The gate operates according to the bal- psychodynamic theories, pain behaviors are sub- ance between the excitation of two types of cells: ject to reinforcement and come to be linked to large diameter nerve fibers open the gate, anxiety and other emotions. The notion of ill- whereas small ones close it (5; 6, p40). The cru- ness behavior refers to the various ways in which cial innovation in the theory was that the gate symptoms may be perceived and acted on (or mechanism is also influenced by transmissions not) by different people. 472 Measuring Health Treatments for pain reflect the multiple influ- emotional, and social factors (8). Most authors ences on the pain experience: therapy may tackle concur that measurements can capture two di- either the pain stimulus or the patient’s reaction mensions of the pain experience: the sensory as- to it. Psychological or behavioral approaches to pects and the person’s emotional reaction (9, therapy may be used to modify the response to p296). Some measurements also attempt to dis- intractable pain where the stimulus itself cannot tinguish the stimulus strength from the subjec- be alleviated. Clinical experience also shows that tive response, although these techniques are not the required dose of analgesics vary according to yet widely used. They are mainly based on sen- the individual’s pain threshold, with larger doses sory decision theory, which was described in being needed for those with lower thresholds. Chapter 2. In its application to pain, this in- These alternative conceptual approaches to volves the repeated presentation of two stimuli: pain remain under active debate; one source of noise (e.g., a low level of electrical current) or reports and discussions is the Web site for the In- experimentally induced pain, in random order. ternational Association for the Study of Pain The respondent classifies each as painful or not, (www.iasp-pain.org/). They also offer defini- and from the resulting pattern of true- and false- tions of terms used in pain studies (www.iasp positive responses, indexes of discriminability -pain.org/terms-p.html). A review of theories and of response bias can be derived (10). Dis- concerning pain mechanisms is contained in criminability shows how accurately the respon- www.coventrypainclinic.org.uk/aboutpain-pain dent can distinguish between various levels of a mechanisms.htm. Information of a more clinical pain stimulus; it is an indicator of perceptual nature can be found on the site of the American performance that reflects the functioning of the Pain Society (www.ampainsoc .org). neurosensory system. Response bias shows the threshold at which the respondent applies the term “pain” in describing these stimuli; it is Approaches to Pain related to his attitudes toward pain. A high re- Measurement sponse threshold suggests stoicism; a low one in- dicates an individual who readily reports pain. Various pain measurement methods have been This analytic technique has been used in experi- proposed. Historically, these have evolved from mental studies to determine whether analgesics a straightforward approach in which pain is de- work by influencing discriminability (by making fined and measured in terms of the person’s sub- the stimulus feel less noxious) or by shifting the jective response, to more complex methods that response bias (making the respondent less will- attempt to disentangle the subjective element in ing to call the stimulus “painful”). Clark and the response from an objective estimate of the Yang showed that a placebo worked as an anal- underlying pain. As an example of the simple ap- gesic principally by reducing the respondent’s proach, Merskey’s definition of pain speaks in tendency to label an experimental stimulus as terms of a single-dimensional subjective re- “painful” rather than by altering the person’s sponse: “Pain is an unpleasant experience which ability to feel it (10). we primarily associate with tissue damage, or Although they offer exciting research possi- describe in terms of tissue damage . . . the pres- bilities, measurement methods based on sensory ence of which is signalled by some form of visi- decision theory are seldom used in clinical appli- ble or audible behavior” (7, p195). cations, which employ far simpler scales that However, as early as the 1950s, there were record intensity. This chapter describes methods multidimensional measures of pain. In 1957, of both types, partly in the hope of encouraging Beecher distinguished two components of pain: a shift away from the current heavy reliance on the original sensation, which he described in simple intensity measurements. terms of its intensity and its temporal and spatial The data used to measure pain fall into three distribution, and the reactive component, which main categories: verbal or written descriptions he considered to be a function of personality, of pain, changes in observable behavior of the  Pain Measurements 473 person experiencing the pain, and analogue tech- later in this chapter, have sought to improve on niques, commonly used in laboratory studies, in its scale characteristics. which the respondent compares her pain to an These methods concentrate on the sufferer’s experimentally induced painful stimulus of own pain level. To indicate a person’s affective known intensity. Unfortunately, few studies have response to pain, questionnaires have been de- yet compared these different approaches, and veloped to assess responses to pain other than the comparisons that have been made suggest a the respondent’s own; this is presumed to influ- rather low agreement: they appear to measure ence their reporting of pain severity. An early different aspects of pain (11). General reviews of method consisted of a color film depicting in- the early development of pain measurement tech- creasingly severe levels of pain applied to a hu- niques have been given by Frederiksen et al. (2), man hand. The viewer’s emotional response to Chapman (12), Reading (13), and Over (14). the scenes is graded (20). Questionnaire Techniques Behavioral Measurements of Pain Most pain questionnaires concentrate on inten- sity and use adjectives such as mild, moderate, or Because pain causes changes in behavior (often severe, or else a numerical scale to represent the involuntarily), such changes can be used as indi- intensity continuum. A variant has been to use cators of pain levels. This is analogous to the be- visual analogue scales that represent the inten- havioral rating scales used to measure functional sity dimension by a plain line without verbal or disability (see, for example, the Sister Kenny numerical guides (see the review of Huskisson’s scale in Chapter 3). Fordyce et al. discussed the measurement technique later in this chapter). In- correspondence between verbal responses and tensity ratings may be extended to add a time di- overt pain behavior (21; 22). Whether recording mension by using diaries or pain charts that behavior offers a more objective approach to record variations in pain over the course of a measuring pain is open to debate; behavior could day, while also recording the medications taken. reflect the subjective responses to pain as much Cumulative scores representing the duration of as verbal reports, but it may also be argued that pain at each intensity level may be derived from involuntary pain behavior such as grimacing or the chart (15). In clinical settings, verbal ratings, wincing is less influenced by attitudes than is an- numerical scales, and visual analogue scale seem swering questions. A critical review of recording to be interchangeable (16; 17). All three meth- behavior as a basis for inferring pain was given ods appear highly intercorrelated, although the by Turk and Flor (23). numerical and visual analogue scales correlate There are several types of behavioral mea- the most strongly. With less-educated patients, surement of pain. Clinical studies often record numerical ratings may be more easily under- reductions in functional performance due to stood than the visual analogue scale (18), and pain, with the pain graded according to how se- the numerical scales can be easily administered riously it limits physical function. Examples of to sicker patients. Hence, numerical scales have such scales are given by Kessler (24) and by been endorsed for use in cancer clinical trials as Fields et al. (25) and are illustrated by Jette’s in- easier to understand and score (19). dex, which is described in Chapter 3. Alterna- Extending the questionnaire approach to tively, pain may be inferred from level of use of cover more than the intensity and duration of health care, including taking medications. This pain, Melzack’s McGill Pain Questionnaire gives is a common approach in research on headaches, a qualitative description of the pain and of the although medication use reflects both the pain patient’s affective response to pain. Subsequent and the attitudes of the sufferer to taking drugs. refinements of the McGill method, such as Facial expressions of pain have been studied in Tursky’s Pain Perception Profile, both reviewed infants and adults; systematic coding schemes 474 Measuring Health have been proposed (26; 27). An experimental clinical pain (11). Many of the affective features procedure may be used in which the patient is of the pain response normally seen in clinical asked to make a series of painful body move- settings may be modified in laboratory experi- ments and the response is observed (28; 29). Ob- ments. Much of the fear and anxiety that accom- servation should evidently be as unobtrusive as panies normal pain are absent: the sufferer possible, although this may be hard to standard- knows the experimental pain is of a fixed dura- ize and may prove costly. Craig and Prkachin tion and under the control of the experimenter. listed a number of behavioral measurements of Although laboratory experiments studying pain this type (30), and examples include the Univer- responses may constitute good psychophysics, sity of Alabama in Birmingham (UAB) Pain Be- they may have limited relevance to pain outside havior Scale (31) and an observational method the laboratory (14). used in patients with chronic low back pain (32). Analogue Methods Scope of the Chapter The analogue approach requires the person to The measures reviewed in this chapter illustrate match his clinical pain with various levels of ex- several of these themes. We begin with visual perimentally induced pain, typically radiant heat analogue scales that measure pain intensity and or an electric shock. After a match is found, the follow this with a description of Melzack’s clinical pain is described in terms of the strength McGill Pain Questionnaire as a multidimen- of the stimulus used to induce the experimental sional pain rating. The Brief Pain Inventory is an pain. Alternatively, the respondent may be asked abbreviation of the McGill instrument for use in to apply a physical effort (such as squeezing a clinical settings. Stewart’s Medical Outcomes pressure bulb) at an intensity that matches his Study pain measures are intended for survey use. pain level; Peck had patients match the intensity Returning to the clinical orientation, Fairbank’s of their pain with the intensity of a sound pro- Oswestry questionnaire as an example of a brief duced by an audiometer (33). Smith et al. de- measurement of the disabling effect of low back scribed the submaximal effort tourniquet pain. The next two scales in the chapter distin- method, in which the blood circulation in the guish between pain that reflects psychological arm is stopped by an inflated blood pressure disturbance and that based on organic causes: cuff; the respondent is then asked to repeatedly Leavitt’s Back Pain Classification Scale and squeeze a hand exerciser. This produces an in- Zung’s Pain and Distress Scale. The final two creasingly intense pain, and the time between methods, the Illness Behavior Questionnaire de- squeezing the exerciser and the point at which veloped by Pilowsky and the Pain Perception the patient judges the experimental pain to Profile by Tursky, both distinguish between the match his clinical pain gives a numerical indica- affective response to pain and its underlying tion of his pain. The time until the pain gener- physical intensity. Table 9.1 summarizes the ated by the exerciser becomes unbearable characteristics of the scales reviewed. indicates the maximum pain tolerance level (12; 34). Clinical pain may be expressed as a percent- age of the pain tolerance level. Limitations of the References tourniquet test include variability in strength be- (1) Chapman CR. Pain perception, affective tween individuals and problems of scaling pain mechanisms, and conscious experience. when arguably pain does not rise as a linear In: Hadjistavropoulos T, Craig KD, eds. function of time in the test; there may also be a Pain: psychological perspectives. practice effect (35). Mahwah, NJ: Lawrence Erlbaum, There are more general concerns over the va- 2004:59–85. lidity of the analogue methods as indicators of (2) Frederiksen LW, Lynd RS, Ross J.  Table 9.1 Comparison of the Quality of Pain Scales* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurements Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Visual Analogue Pain Rating ratio 1 clinical self many ** *** ** ** Scale (Various authors, 1974) (30 sec) McGill Pain Questionnaire ordinal, 20 clinical self many * * ** ** (Melzack, 1975) interval (15 min) Brief Pain Inventory ordinal 11 clinical self several ** ** ** * (Cleeland, 1982) (10–15 min) Medical Outcomes Study Pain ordinal 12 survey self few * ** * ** Measures (Sherbourne, 1992) 475 Oswestry Low Back Pain ordinal 60 clinical self several * ** ** ** Disability Questionnaire (<5 min) (Fairbank, 1980) Back Pain Classification Scale interval 13 clinical self several * ** ** ** (Leavitt, 1978) (5–10 min) Self-Rating Pain and Distress ordinal 20 clinical self few * ** * ** Scale (Zung, 1983) Illness Behavior Questionnaire ordinal 52 clinical self many ** ** *** ** (Pilowky and Spence, 1975) Pain Perception Profile ratio 37 research expert few * 0 * 0 (Tursky, 1976) * For an explanation of the categories used, see Chapter 1, pages 6–7 476 Measuring Health Methodology in the measurement of pain. (15) Elton D, Burrows GD, Stanley GV. Clinical Behav Ther 1978;9:486–488. measurement of pain. Med J Aust (3) Tursky B, Jamner LD, Friedman R. The 1979;1:109–111. Pain Perception Profile: a psychophysical (16) Jensen MP, Karloy P, Braver S. The mea- approach to the assessment of pain report. surement of clinical pain intensity: a Behav Ther 1982;13:376–394. comparison of six methods. Pain (4) Melzack R, Katz J. The gate control 1986;27:117–126. theory: reaching for the brain. In: (17) Dalton JA, McNaull F. A call for Hadjistavropoulos T, Craig KD, editors. standardizing the clinical rating of pain Pain: psychological perspectives. intensity using a 0 to 10 rating scale. Mahwah, NJ: Lawrence Erlbaum, Cancer Nurs 1998;21:46–49. 2004:13–34. (18) Ferraz MB, Quaresma MR, Aquino LRL, (5) Asmundson GJG, Wright KD. et al. Reliability of pain scales in the Biopsychosocial approaches to pain. In: assessment of literate and illiterate patients Hadjistavropoulos T, Craig KD, eds. Pain: with rheumatoid arthritis. J Rheumatol psychological perspectives. Mahwah, NJ: 1990;17:1022–1024. Lawrence Erlbaum, 2004:35–57. (19) Moinpour CM, Feigl P, Metch B, et al. (6) Melzack R, Wall PD. Pain mechanisms. Quality of life end points in cancer clinical Science 1965;150:971–979. trials: review and recommendations. J Natl (7) Bond MR. New approaches to pain. Cancer Inst 1989;81:485–495. Psychol Med 1980;10:195–199. (20) Elton D, Burrows GD, Stanley GV. (8) Beecher HK. The measurement of pain: Apperception of pain. In: Peck D, Wallace prototype for the quantitative study of M, eds. Problems in pain. Sydney, subjective responses. Pharmacol Rev Australia: Pergamon Press, 1980:117–120. 1957;9:159–209. (21) Fordyce WE. The validity of pain behavior (9) Cleeland CS. Pain assessment in cancer. In: measurement. In: Melzack R, ed. Pain mea- Osoba D, ed. Effect of cancer on quality of surement and assessment. New York: life. Boca Raton, FL: CRC Press, Raven Press, 1983:145–153. 1991:293–305. (22) Fordyce WE, Lansky D, Calsyn DA, et al. (10) Clark WC, Yang JC. Applications of Pain measurement and pain behavior. Pain sensory decision theory to problems in 1984;18:53–69. laboratory and clinical pain. In: Melzack R, (23) Turk DC, Flor H. Pain > pain behaviors: the ed. Pain measurement and assessment. New utility and limitations of the pain behavior York: Raven Press, 1983:15–25. construct. Pain 1987;31:277–295. (11) Postlethwaite R, Grieve N, Santacroce T, et (24) Kessler HH. Disability—determination and al. An analysis of pain produced by the evaluation. Philadelphia: Lea & Febiger, submaximum effort tourniquet test. In: 1970. Peck C, Wallace M, eds. Problems in pain. (25) Fields HL, Florence D, Keefe F, et al. Mea- Sydney, Australia: Pergamon Press, suring pain and dysfunction. In: Osterweis 1980:128–135. M, Kleinman A, Mechanic D, eds. Pain and (12) Chapman CR. Measurement of pain: disability: clinical, behavioral, and public problems and issues. In: Bonica JJ, Albe- policy perspectives. Washington, DC: Fessard DG, eds. Advances in pain research National Academy Press, 1987:211–231. and therapy. Vol. I. New York: Raven (26) Lilley CM, Craig KD, Grunau RE. The Press, 1976:345–353. expression of pain in infants and toddlers: (13) Reading AE. Testing pain mechanisms in developmental changes in facial action. persons in pain. In: Wall PD, Melzack R, Pain 1997;72:161–170. eds. Textbook of pain. Edinburgh: (27) Solomon PE, Prkachin KM, Farewell V. Churchill-Livingstone, 1989:269–280. Enhancing sensitivity to facial expression (14) Over R. Clinical and experimental pain. In: of pain. Pain 1997;71:279–284. Peck C, Wallace M, eds. Problems in pain. (28) Waddell G, McCulloch J, Kummel E, et al. Sydney, Australia: Pergamon Press, Nonorganic physical signs in low back 1980:94–100. pain. Spine 1980;5:117–125.  Pain Measurements 477 (29) Follick MJ, Ahern DK, Aberger EW. measuring pain during the 1970s (2). A review Development of an audiovisual taxonomy of VAS ratings in other domains of health mea- of pain behavior: reliability and surement is contained in the entry on single-item discriminant validity. Health Psychol scales in Chapter 10. 1985;4:555–568. Visual analogue measurements are normally (30) Craig KD, Prkachin KM. Nonverbal mea- used to rate the overall severity of pain, although sures of pain. In: Melzack R, ed. Pain mea- there is no reason why a VAS could not be ap- surement and assessment. New York: Raven Press, 1983:173–179. plied in measuring other dimensions, such as lev- (31) Richards JS, Nepomuceno C, Riles M, et els of anxiety or the emotional responses al. Assessing pain behavior: the UAB Pain associated with pain. Color analogue scales have Behavior Scale. Pain 1982;14:393–398. also been used to judge pain severity (3). (32) Keefe FJ, Block AR. Development of an observation method for assessing pain Description behavior in chronic low back pain patients. A VAS is a line that represents the continuum of Behav Ther 1982;13:363–375. the symptom to be rated. The scale, convention- (33) Peck RR. A precise technique for the mea- ally a straight line 10 cm long, is marked at each surement of pain. Headache end with labels that indicate the range being con- 1967;7:189–194. sidered: Huskisson and his collaborators, used (34) Smith GM, Lowenstein E, Hubbard JH, et al. Experimental pain produced by the the phrases “pain as bad as it could be” and “no submaximal effort tourniquet technique: pain” (4; 5). Alternative phrases that have been further evidence of validity. J Pharmacol recommended for the extreme end include “ago- Exp Ther 1968;163:468–474. nizing pain” (6, Table 4) and “worst pain imagi- (35) Sternbach RA. The tourniquet pain test. In: nable” (7). Patients are asked to place a mark on Melzack R, ed. Pain measurement and the line at a point representing the severity of assessment. New York: Raven Press, their pain. The scale requires only about 30 sec- 1983:27–31. onds to complete. Typical formats are shown in Exhibit 9.1, but many variants have been tested. These range Visual Analogue Pain Rating Scales from the pure VAS to numerical rating scales. (Various authors, from 1974 For example, dividing marks may be printed on onward) the line, although a plain line is normally used; Purpose this may be oriented vertically or horizontally. Visual analogue rating scales (VAS) provide a Curved lines have also been tried, with or with- simple way to record subjective estimates of pain out intermediate marks. Descriptive terms may intensity. be placed along the plain line, such as “severe,” “moderate,” or “mild.” When numbers are Conceptual Basis added to the line, but without vertical marks to The overview of psychophysics in Chapter 2 de- indicate the boundaries, the result has been scribed how accurately people can rate stimuli in called the Graphic Rating Scale (8); numbers terms of the length of lines; these experiments may be placed in squares in the Box scale (9, Fig- lent support to the idea of using plain lines to ures 1 and 2). Some suggestions concerning represent the intensity of subjective phenomena. these alternative formats are provided in the VAS involve the respondent’s placing a mark a Commentary section. certain distance along a line to represent the in- There are several ways to score the VAS (1; tensity of a stimulus. They have long been used 10). The distance of the respondent’s mark from in many areas of psychology (1); their applica- the lower end of the scale, measured in millime- tion in health measurement began with pain but ters, forms the basic score, ranging from 0 to has extended to general health assessments. 100. Alternatively, a 20-point grid may be super- Huskisson popularized the application of VAS to imposed over the line to give an integer rating. 478 Measuring Health Exhibit 9.1 Alternative Formats for Visual Analogue Rating Scales, as Tested by Huskisson PAIN AS BAD AS IT COULD BE NO PAIN PAIN AS BAD AS IT NO PAIN COULD BE PAIN AS BAD AS IT NO PAIN COULD BE S E V E R E M O D E R AT E S L I G H T Adapted from Scott J, Huskisson EC. Graphic representation of pain. Pain 1976;2:176. Scott and Huskisson justified this approach by form of a descriptive scale (such as none, slight, noting that it represents the maximal level of dis- moderate, or complete relief ) or a VAS running crimination people can use in recording pain (2). from no relief to complete relief of pain. Com- The distribution of results is often not normal, paring the two methods, Huskisson showed that and transformations may be applied to normal- the simple descriptive scale gave better results if ize the data (2, p1128; 11). Although nonpara- patients had to complete it without assistance metric statistical analyses are generally (2). The advantage of a pain relief scale is that, considered appropriate (4), one study showed whatever their initial pain level, each respondent that VAS measures produced a measurement has the same range of potential response (6, with ratio scale properties (12). p239). In estimating pain relief, it may not be appro- priate simply to compare scores before and after Reliability treatment, because the magnitude of this differ- Scott and Huskisson studied the repeatability of ence is limited by the placement of the initial visual analogue pain scales and also compared mark (2). Indeed, initial and subsequent pain rat- scales printed vertically and horizontally. One ings tend to be correlated; Huskisson reported hundred patients with rheumatological disorders coefficients of 0.62 and 0.63 (2, Figures 5, 6). were given vertical and horizontal scales in ran- Rather than comparing ratings before and after dom order. The correlation was 0.99 between treatment, therefore, Huskisson recommended the scores, although scores on the horizontal using a rating of pain relief. This can take the scale were slightly, but not significantly, lower  Pain Measurements 479 than on the vertical (mean 10.85 versus 11.05) (affective score) to 0.57 (evaluative) (20, Table (13, Table 1). A comparison of retest reliability 1); a third study gave correlations that varied with literate and nonliterate patients found the widely according to the type of patient being as- VAS more reliable (0.94) in literate than in non- sessed: the correlation between the VAS and the literate groups (0.71); equivalent figures for a Present Pain Index ranged from 0.21 to 0.76 numerical rating scale were 0.96 and 0.95, (21, Tables 2, 3). McCormack et al. reviewed 17 whereas results for a verbal rating were 0.90 and correlations between VAS pain measures and 0.82, suggesting that the numerical rating was other pain assessments. Only two correlations least affected by literacy (13, Table 2). were not significant and the coefficients ranged In a study of geriatric patients, the Spearman from 0.29 to 0.92 (1, Table 4). In a study of low retest correlation for the VAS was 0.78, whereas back pain during pregnancy, VAS pain scores that for a numerical rating scale was 0.82 (14, correlated 0.70 with back muscle activity during Table 1). In a study of low back pain during flexion (15, p1211). Correlations with the Os- pregnancy, first and last trimester VAS pain westry low back pain questionnaire were 0.36 scores correlated 0.82 (15, p1211). during the first trimester, rising to 0.67 during the last (15, Table 2). Validity VAS appear to be more sensitive to change Huskisson reported a correlation of 0.75 be- than verbal rating scales are and so require tween a VAS printed vertically and a four-point smaller sample sizes in evaluative studies (6, descriptive scale rating pain as slight, moderate, p238; 16). The comparison with NRS is less severe, or agonizing (5). This figure was presum- clear, however. Guyatt et al. showed that both ably attenuated by the restricted number of cate- methods showed similar responsiveness when gories on the descriptive scale. Similar analyses standardized to the same scale range and so, in patients with rheumatic disorders gave corre- given its greater simplicity, they recommended lations ranging from 0.71 to 0.78 between de- the numerical rating (22). Beurskens et al. com- scriptive scales and VAS printed vertically or pared effect sizes for VAS to two other back pain horizontally. A correlation of 0.81 was reported questionnaires. The effect size for the VAS was between a VAS and a five-point verbal rating 1.58, better than that for the Oswestry question- scale using repeated measurements of six pa- naire (0.80) but less good than that for the tients (16). In an experimental study, Price et al. Roland Disability Questionnaire (2.02) (23, found that a VAS provided a better ratio scale in- Table IV). Jensen et al. compared VAS with a dication of facial pain than a numerical rating box format and reported very similar sensitivity scale did; a mechanically administered VAS pro- to change, although the box scale is simpler to vided similar results to a paper-and-pencil ver- administer (9, Table 2). sion (17). Another comparison obtained a correlation of 0.85 between VAS and a numeric Alternative Forms rating scale (NRS) (18). In a sample of geriatric Exhibits 9.1 and 9.2 illustrate some formats for patients, Spearman correlations between a VAS the VAS, but others exist (24, Figure 2). Other and a 10-point numerical rating ranged from methods that have been tested include a com- 0.81 to 0.92; equivalent correlations with a puter animation, in which the respondent stops a seven-point adjectival rating scale were fraction- marker that moves along a scale at the appropri- ally lower, at 0.79 to 0.91. The correlations were ate place; the computer calculates the score (25). lower for older patients (14, Table 2). A simpler mechanical device uses a 15-cm plastic Elton et al. obtained correlations from 0.60 slider that is moved by the respondent, revealing to 0.63 between a visual analogue scale and a bar of deepening red to indicate perceived in- Melzack’s McGill Pain Questionnaire (MPQ), tensity; a numerical score is indicated on the although the latter was scored in an unconven- back (17). Scales have also been presented as tional manner (19, Table 1). In a second study, thermometers to try and clarify the metaphor correlations with the MPQ ranged from 0.42 and to underscore the links with health (26). 480 Measuring Health The correlation between vertical and horizon- Most studies that compare VAS with numerical tal scales is generally high. In measuring short- and verbal ratings conclude that the VAS or the ness of breath, for example, the correlation NRS is statistically preferable to the verbal rat- between vertical and horizontal scales was 0.97 ing scales (6). Emotional reactions to pain seem (27). Correlations between the vertical and hori- to cloud verbal ratings more than they do the zontal scales ranged from 0.89 to 0.91 (28, Ta- VAS (33). bles 1, 2). Other studies have found lower values Some patients have difficulty with the VAS, (e.g., 0.88), which was lower than the correla- however; these seem mainly to be elderly or less- tion between a verbal scale and a numerical rat- educated respondents who may not grasp the ing (24, p42). As discussed later in this chapter, metaphor of the continuum represented by the there may be reasons to favor the horizontal for- line (9; 34). One study found response errors in mat. 7% of elderly people using VAS ratings, com- In their study of children aged 3 to 15 years, pared with 2% for verbal ratings (24, p41). Of McGrath et al. used a VAS to establish numeri- 100 patients assessed by Huskisson, seven were cal scores for a pain measurement that used pic- unable to understand the VAS when it was first tures of faces expressing pain (29, p391). explained, whereas all could use a descriptive The VAS has been administered over the tele- scale (2). Accordingly, Scott and Huskisson rec- phone by translating it into a numerical rating ommended that patients complete a first VAS un- scale. Here, respondents are asked to describe der supervision before using it on their own their pain on a scale from 0 to 100, with the an- (13). A pain scale showing faces that vary in chor of 0 representing no pain and 100 repre- their expression of pain may prove superior for senting the worst imaginable pain. When applied nonliterate groups (35) and for the elderly (36). to a characteristic such as self-perceived health, However, in a study of patients with Alzheimer’s the anchor definitions would typically be re- disease, a colored VAS scale was better under- versed, so that 0 might refer to the “least desir- stood than either a “faces” pain scale or a able state of health,” and 100 to “perfect “faces” scale depicting affect (37). Downie et al. health” (30). compared a VAS with a ten-point numerical rat- ing scale of the type shown in Exhibit 9.2 and Reference Standards On a small sample of patients, Lee et al. esti- mated that a 30 mm change represents patients’ Exhibit 9.2 Formats of the Numerical Rating perceptions of a clinically important reduction (NRS) and Visual Analogue Scales (VAS) as in pain (31). Used by Downie et al. Commentary VAS pain ratings have become standard; a review was given by Ho et al. (32). Huskisson summa- rized the advantages of visual analogue scales as follows: Visual analogue scales provide the patients with a robust, sensitive, reproducible method of expressing pain severity. Results correlate well with other methods of measuring pain. The method is applicable to all patients re- Reproduced from Downie WW, Leatham PA, Rhind VM, Wright V, Branco JA, Anderson JA. Studies with pain rating gardless of language and can be used by chil- scales. Ann Rheum Dis 1978;37:378, Figure 1. With permis- dren aged five or more years. (4, p768) sion.  Pain Measurements 481 argued that the latter “was to be preferred on the assessing pain on follow-up. Scott and grounds of measurement error,” perhaps be- Huskisson argued that, because increasing time cause the VAS offers a confusingly wide range of between assessments may make it difficult to re- choice (28). Some patients cluster their scores at call the initial pain, patients should see their pre- the midpoint of the scale or at either end, sug- vious scores when making subsequent judgments gesting that they do not comprehend the idea of of pain severity (42). Other authors concur, ar- the continuum (1, p1014). Finally, a Swedish guing that recall of the initial ratings is inaccu- study compared the ability of cognitively normal rate and perhaps biased (43; 44). Dixon and Bird elderly patients to complete a VAS, a numerical disagreed: although patients may find it easier to scale, and an adjectival scale. Success in complet- rate their current pain when shown previous ing the rating declined with age; by age 95, scores, such ratings may be biased by seeing the fewer than 70% of patients could complete the earlier scores and so may not agree with other VAS, compared with 85% for the adjective scale indexes of disease progress (38). They also and 75% for the numerical scale (14, Figure 1). showed that the accuracy with which a person Several conclusions may be drawn from the could replicate earlier VAS marks varied accord- studies that have compared formats for the VAS. ing to the placement of the original mark: not First, horizontal lines are generally preferred to surprisingly, marks close to the ends or the cen- vertical. Dixon and Bird warned that a distortion ter of the line were more accurately reproduced of perspective may arise in using a vertical scale than marks in other positions (38). when the page lies flat on a table (38). Vertical VAS have been successfully used in a wide va- scales appear to give less normally distributed riety of health (and other) assessments. In addi- data (6; 39, p238) and create more difficulty in tion to assessing pain intensity or frequency, administration (1, p1008). An interesting excep- they have been used in assessing many aspects of tion was found with Chinese patients with post- quality of life (see the review of single-item operative pain, who made fewer errors using the health measures in Chapter 10). They give more vertical format (which, of course, resembles the sensitive and precise measurements than verbal orientation of Chinese writing) (40). Second, a scales. They avoid the clustering of scores due to comparison of 5-, 10-, 15-, and 25-cm lines sug- digit preference that can occur with numerical gested that the 5-cm line provided less reliable rating scales. The choice between verbal pain results than other lengths (41). The conventional rating scales and the VAS often depends on the 10-cm line seems the most appropriate (7). balance between simplicity of administration Third, there seems little to gain by adding inter- and the degree of sensitivity required: a categor- mediate dividing marks along the line (6, p238). ical rating is easy to administer but may be too Fourth, a straight line seems adequate, although crude for research purposes. As a compromise in one study some respondents found that a solution, the ten-point NRS may avoid the limi- curved line had the benefit of familiarity by re- tations of both the four-point rating and the vi- minding them of an analogue dial (e.g., sual analogue methods. speedometer, bathroom scale) (6, p238). Fifth, it is generally recommended that neither numbers nor verbal labels be placed along the length of the References line (1, p1008). Scott and Huskisson tested many formats and concluded that scales with adjectives (1) McCormack HM, Horne DJ, Sheather S. Clinical applications of visual analogue tended to produce a clustering of responses be- scales: a critical review. Psychol Med side the adjectives (2, Figures 2, 3). If adjectives 1988;18:1007–1019. are to be used, they should be printed so that (2) Huskisson EC. Measurement of pain. they spread along the entire length of the line, as Lancet 1974;2:1127–1131. illustrated in the last example of Exhibit 9.1. (3) Bromley J, Emerson E, Caine A. The A debate has arisen over whether patients development of a self-report measure to should be shown their initial ratings when re- assess the location and intensity of pain in 482 Measuring Health people with intellectual disabilities. J analogue and simple numerical rating Intellect Disabil Res 1998;42:72–80. scales. Pain 1994;56:217–226. (4) Huskisson EC. Measurement of pain. J (18) Paice JA, Cohen FL. Validity of a verbally Rheumatol 1982;9:768–769. administered numeric rating scale to mea- (5) Scott J, Huskisson EC. Graphic representa- sure cancer pain intensity. Cancer Nurs tion of pain. Pain 1976;2:175–184. 1997;20:88–93. (6) Sriwatanakul K, Kelvie W, Lasagna L, et (19) Elton D, Burrows GD, Stanley GV. Clinical al. Studies with different types of visual measurement of pain. Med J Aust analog scales for measurement of pain. 1979;1:109–111. Clin Pharmacol Ther 1983;34:234–239. (20) Ahles TA, Ruckdeschel JC, Blanchard EB. (7) Seymour RA, Simpson JM, Charlton JE, et Cancer-related pain—II. Assessment with al. An evaluation of length and end-phrase visual analogue scales. J Psychosom Res of visual analogue scales in dental pain. 1984;28:121–124. Pain 1985;21:177–185. (21) Perry F, Heller PH, Levine JD. Differing (8) Dalton JA, McNaull F. A call for correlations between pain measures in standardizing the clinical rating of pain syndromes with or without explicable intensity using a 0 to 10 rating scale. organic pathology. Pain 1988;34:185–189. Cancer Nurs 1998;21:46–49. (22) Guyatt GH, Townsend M, Berman LB, et (9) Jensen MP, Miller L, Fisher LD. al. A comparison of Likert and visual Assessment of pain during medical analogue scales for measuring change in procedures: a comparison of three scales. function. J Chronic Dis Clin J Pain 1998;14:343–349. 1987;40:1129–1133. (10) Maxwell C. Sensitivity and accuracy of the (23) Beurskens AJHM, de Vet HCW, Köke visual analogue scale: a psycho-physical AJA. Responsiveness of functional status in classroom experiment. Br J Clin Pharmacol low back pain: a comparison of different 1978;6:15–24. instruments. Pain 1996;65:71–76. (11) Stubbs DF. Visual analogue scales. Br J (24) Herr KA, Mobily PR. Comparison of Clin Pharmacol 1979;7:124. selected pain assessment tools for use with (12) Price DD, McGrath PA, Rafii A, et al. The the elderly. Appl Nurs Res 1993;6:39–46. validation of visual analogue scales as ratio (25) Swanston M, Abraham C, Macrae WA, et scale measures for chronic and al. Pain assessment with interactive computer experimental pain. Pain 1983;17:45–56. animation. Pain 1993;53:347–351. (13) Scott J, Huskisson EC. Vertical or (26) Choiniere M, Auger FA, Latarjet J. Visual horizontal visual analogue scales. Ann analogue thermometer: a valid and useful Rheum Dis 1979;38:560. instrument for measuring pain in burned (14) Bergh I, Sjöström B, Odén A, et al. An patients. Burns 1994;20:229–235. application of pain rating scales in geriatric (27) Gift AG. Validation of a vertical visual patients. Aging Clin Exp Res analogue scale as a measure of clinical 2000;12:380–387. dyspnoea. Rehabil Nurs 1989;14:323–325. (15) Sihvonen T, Huttunen M, Makkonen M, et (28) Downie WW, Leatham PA, Rhind VM, et al. Functional changes in back muscle al. Studies with pain rating scales. Ann activity correlate with pain intensity and Rheum Dis 1978;37:378–381. prediction of low back pain during (29) McGrath PA, de Veber LL, Hearn MT. pregnancy. Arch Phys Med Rehabil Multidimensional pain assessment in 1998;79:1210–1212. children. In: Fields HL, Dubner R, Cervero (16) Ohnhaus EE, Adler R. Methodological F, eds. Advances in pain research and problems in the measurement of pain: a therapy. Vol. 9. New York: Raven Press, comparison between the verbal rating scale 1985:387–393. and the visual analogue scale. Pain (30) Houle C, Berthelot J-M. A head-to-head 1975;1:379–384. comparison of the Health Utilities Index (17) Price DD, Bush FM, Long S, et al. A Mark 3 and the EQ-5D for the population comparison of pain measurement living in private households in Canada. characteristics of mechanical visual Quality of Life Newsletter 2000;24:5–6.  Pain Measurements 483 (31) Lee JS, Hobden E, Stiell IG, et al. Clinically I. Aspects of the reliability and validity of important change in the visual analog scale the visual analogue scale. Pain after adequate pain control. Acad Emerg 1983;16:87–101. Med 2003;10:1128–1130. (32) Ho K, Spence J, Murphy MF. Review of pain-measurement tools. Ann Emerg Med The McGill Pain Questionnaire 1996;27:427–432. (Ronald Melzack, 1975) (33) Tammaro S, Berggren U, Bergenholtz G. Representation of verbal pain descriptors Purpose on a visual analogue scale by dental The McGill Pain Questionnaire (MPQ) was de- patients and dental students. Eur J Oral Sci signed to provide a quantitative profile of three 1997;105:207–212. aspects of pain (1). The method was originally (34) Ferraz MB, Quaresma MR, Aquino LRL, used in evaluating pain therapies, but it has also et al. Reliability of pain scales in the been used as a diagnostic aid (2). assessment of literate and illiterate patients with rheumatoid arthritis. J Rheumatol Conceptual Basis 1990;17:1022–1024. Melzack’s major contribution to pain measure- (35) Frank AJM, Moll SMH, Hart JF. A ment was to emphasize that the pain experience comparison of three ways of measuring comprises several distinct aspects. In 1988, he pain. Rheumatol Rehabil wrote: 1982;21:211–217. (36) Herr KA, Mobily PR, Kohout FJ, et al. Evaluation of the faces pain scale for use The problem of pain, since the beginning of with the elderly. Clin J Pain this century, has been dominated by the con- 1998;14:29–38. cept that pain is purely a sensory experience. (37) Scherder EJA, Bouma A. Visual analogue Yet it has a unique, distinctively unpleasant, scales for pain assessment in Alzheimer’s affective quality that differentiates it from disease. Gerontology 2000;46:47–53. sensory experiences such as sight, hearing or (38) Dixon JS, Bird HA. Reproducibility along touch. . . . a 10 cm vertical visual analogue scale. Ann The motivational-affective dimension of Rheum Dis 1981;40:87–89. pain is brought clearly into focus by clinical (39) Ogon M, Krismer M, Sollner W, et al. studies on frontal lobotomy . . . Typically, Chronic low back pain measurement with these patients report after the operation that visual analogue scales in different settings. Pain 1996;64:425–428. they still have pain but it does not bother (40) Aun C, Lam YM, Collett B. Evaluation of them . . . the suffering, the anguish are the use of visual analogue scale in Chinese gone. . . . Similarly, patients exhibiting “pain patients. Pain 1986;25:215–221. asymbolia” . . . after lesions of portions of (41) Revill SI, Robinson JO, Rosen M, et al. the parietal lobe or the frontal cortex are able The reliability of a linear analogue for to appreciate the spatial and temporal proper- evaluating pain. Anaesthesia ties of noxious stimuli (for example, they rec- 1976;31:1191–1198. ognize pin pricks as sharp) but fail to (42) Scott J, Huskisson EC. Accuracy of withdraw or complain about them. . . . subjective measurements made with or These considerations suggest that there without previous scores: an important are three major psychological dimensions of source of error in serial measurement of pain: sensory-discriminative, motivational- subjective states. Ann Rheum Dis 1979;38:558–559. affective, and cognitive-evaluative. (3, (43) Linton SJ, Gotestam KG. A clinical pp93–95) comparison of two pain scales: correlation, remembering chronic pain, and a measure Melzack argued that these three aspects of the of compliance. Pain 1983;17:57–65. pain experience are subserved by distinct sys- (44) Carlsson AM. Assessment of chronic pain. tems in the brain (3) and he sought to measure 484 Measuring Health these dimensions with the MPQ. This link be- ings made by patients. These values are shown in tween emotions and neurological structures be- Exhibit 9.4. came a central feature of his neuromatrix model The words were originally read to the patient of pain, developed long after the MPQ. This by an interviewer so that unfamiliar words could newer model builds on his gate control theory be explained. This took 15 to 20 minutes, reduc- (see the introduction to this chapter) (4). ing to five to ten minutes for patients familiar Melzack stressed that the questionnaire repre- with the method. Hospitalized patients may re- sented a first attempt at developing a measure- quire longer: a mean of 24 minutes has been ment reflecting his theory of pain, and he reported (5, p36). Subsequent users have em- suggested that other investigators might ulti- ployed the checklist in a written format. The re- mately refine it (1). Nonetheless, the method spondent is asked to select the one word in each continues to be used in its original form. subclass that most accurately describes her pain at that time (1). If none of the words applies, Description none is chosen. Other instructions have been The complete MPQ comprises sections record- used: patients may be asked to describe their ing the patient’s diagnosis, drug regimen, pain “average pain,” their “most intense” pain (7), or history, present pain pattern, accompanying how their pain “typically feels” (8). symptoms and modifying factors, the effects of Four scoring methods were proposed by pain, and the list of words describing pain, Melzack: which is the part of the instrument most com- monly used (5, Table 1). The present discussion 1. The sum of the scale values for all the concerns only this last section of the overall words chosen in a given category (e.g., questionnaire. sensory), or across all categories. This was Melzack and Torgerson selected 102 words called a Pain Rating Intensity Score using describing pain from the literature and from ex- scale values: PRI(S). The scale weights are isting questionnaires (6). They sorted these shown in Exhibit 9.4; note that no scale words into the three major classes proposed in weights are available for category 19. Sep- Melzack’s theory of pain: words concerned with arate PRI scores can be calculated for the the sensory qualities of pain (e.g., temporal, sensory scales (range, 0–42), for the affec- thermal), those covering affective qualities of tive scales (range, 0–14), for the evaluative pain (e.g., fear, tension), and “evaluative words scales (range, 0–5), and for the miscella- that describe the subjective overall intensity of neous scales (range, 0–17) (5, p37). The the total experience of pain” (6). Within the category scores and the PRI(S) can also be three major classes, Melzack and Torgerson placed on a common scale by dividing the grouped words that were qualitatively similar, as summed value by the maximum possible shown in Exhibit 9.3. The suitability of this a score, giving a range from 0 to 1.00 (9, priori grouping was checked by 20 reviewers. At p45). first, there were 16 such subclasses, but four oth- 2. As in 1, but replacing the scale values by a ers were added to give the final 20 subclasses (1). code indicating the rank placement of An equal-appearing interval scaling proce- each word selected within its subclass: dure was used to estimate the intensity of pain PRI(R). This score can be modified by represented by the words in each subclass. Three multiplying the rank placement by a groups of judges (140 students, 20 physicians, weight for each of the 20 groups of words and 20 patients) rated each word on a seven- in order to reflect the differing severity point scale (6). Where disagreement existed represented by the groups. The 20 weights among the three groups of judges on the rank were shown by Melzack et al. (10, Table ordering of a word within a subclass, the word 1). was deleted from the questionnaire; scale values 3. The Number of Words Chosen: NWC. for the remaining words were based on the rat- 4. The Present Pain Intensity (PPI) is a simple  Pain Measurements 485 Exhibit 9.3 The McGill Pain Questionnaire Reproduced from Melzack R. Psychologic aspects of pain. Pain 1980;8:145. Reprinted with permission from the International As- sociation for the Study of Pain. numerical zero-to-five scale as shown in three times at intervals ranging from three to Exhibit 9.3 (although a one-to-five scale seven days; there was an average consistency of has also been used). response of 70.3% (1, p287). The extent of cor- relation between sensory and affective scores Reliability varies according to the type of pain being con- Melzack reported a small test-retest study in sidered: in one study the range was from 0.51 to which ten patients completed the questionnaire 0.85 (11, Tables 2, 3). 486 Measuring Health Exhibit 9.4 Scale Weights for Scoring the McGill Pain Questionnaire 1 Flickering 1.89 2 Jumping 2.60 3 Pricking 1.94 Quivering 2.50 Flashing 2.75 Boring 2.05 Pulsing 2.56 *Shooting 3.42 Drilling 2.75 *Throbbing 2.68 *Stabbing 3.45 Beating 2.70 Lancinating 3.50 Pounding 2.85 4 *Sharp 2.95 5 Pinching 1.95 6 Tugging 2.16 Cutting 3.20 Pressing 2.42 Pulling 2.35 Lacerating 3.64 *Gnawing 2.53 Wrenching 3.47 *Cramping 2.75 Crushing 3.58 7 *Hot 2.47 8 Tingling 1.60 9 Dull 1.60 *Burning 2.95 Itchy 1.70 Sore 1.90 Scalding 3.50 Smarting 2.00 Hurting 2.45 Searing 3.88 Stinging 2.25 *Aching 2.50 *Heavy 2.95 10 *Tender 1.35 11 *Tiring 2.42 12 *Sickening 2.75 Taut 2.36 *Exhausting 2.63 Suffocating 3.45 Rasping 2.61 *Splitting 3.10 13 *Fearful 3.30 14 *Punishing 3.50 15 Wretched 3.16 Frightful 3.53 Gruelling 3.73 Blinding 3.45 Terrifying 3.95 *Cruel 3.95 Vicious 4.26 Killing 4.50 16 Annoying 1.89 17 Spreading 3.30 18 Tight 2.25 Troublesome 2.42 Radiating 3.38 Numb 2.10 Miserable 2.85 Penetrating 3.72 Drawing 2.53 Intense 3.75 Piercing 3.78 Squeezing 2.35 Unbearable 4.42 Tearing 3.68 19 Cool 20 Nagging 2.25 Cold Nauseating 2.74 Freezing Agonizing 3.20 Dreadful 4.11 Torturing 4.53 *Asterisks identify the words included in the short-form MPQ. Adapted from Melzack R, Torgerson WS. On the language of pain. Anesthesiology 1971;34:54–55, Table 1. Validity Correlations between MPQ scores and visual Melzack reviewed the agreement among the four analogue scales for 40 patients ranged from 0.50 scoring methods (1). Correlations are presented for the affective score to 0.65 for the PPI and for in Table 9.2 and show that the PPI score does not the total score (13, Table 2). Equivalent figures agree closely with the other scores. from another study were 0.42 for the affective Dubuisson and Melzack compared responses score and 0.57 for the evaluative (14, Table 1). to the MPQ given by 95 patients suffering from Several reviewers have addressed the question various pain syndromes. Discriminant function of whether Melzack’s selection and grouping of analyses showed that 77% of patients could be words do indeed reflect the three dimensions he correctly classified into diagnostic groups on the proposed. Studies in Canada (7), the United basis of their verbal description of pain (2; 12). States (15), and Britain (16), each using different  Pain Measurements 487 Table 9.2 Pearson Correlations Among Four Scoring Procedures for the McGill Pain Questionnaire (N = 248) PRI(S) PRI(R) NWC PPI PRI(S) ... 0.95 0.97 0.42* PRI(R) ... 0.89 0.42 NWC ... 0.32 PPI ... * Melzack did not give a precise figure, but stated it was “virtually identi- cal” to the 0.42 obtained between PPI and PRI(R). Adapted from Melzack R. The McGill Pain Questionnaire: major proper- ties and scoring methods. Pain 1975;1:285. types of pain patient, have reviewed the factor debate over the structure of the MPQ, the fac- structure of the MPQ. Critical reviews of factor tors tend to intercorrelate highly (0.64–0.81), al- analytic investigations were given by Prieto and though Melzack argued that this does not Geisinger (17) and by Holroyd et al. (18). contradict their use as separate scores (22). Analyses have extracted as few as three factors Torgerson and BenDebba applied multidimen- and as many as seven. For example, two studies sional scaling to a subset of 17 descriptors and extracted four factors (13; 15), two found five basically confirmed their placement in the (7; 19), and one found six (20). Holroyd et al. groups shown in the Exhibit (23, Table 1). discussed methodological reasons for the dis- The PPI scale correlated −0.68 with the pain crepancies and then proceeded to analyze data questions from the World Health Organization’s from a large sample of patients with low back quality of life scale, the WHOQOL. Regression pain. They concluded that the sensory category analyses showed that it was principally the eval- of adjectives fell into two groups, separate from uative component of the MPQ that contributed the affective and evaluative scales, making a to predicting quality of life (24, p401). four-factor solution. However, a single, higher- order factor accounted for two thirds of the vari- Alternative Forms ance, suggesting that forming separate factor Wilkie et al. reviewed several variants on the scores would have little utility (18, Table III and overall McGill instrument that may cause confu- p307). Not surprisingly, differing conclusions sion: the McGill-Melzack Pain Questionnaire, have been drawn concerning the structure of the the McGill Pain Assessment Questionnaire, and MPQ. Reading concluded that “Factorial inves- the MPQ, all of which differ, although all con- tigations of the questionnaire provide support tain the pain descriptors reviewed here (5, Table for the distinction between affective and sensory 1). dimensions, but not for a distinctive evaluative A short-form MPQ has been developed, component” (16). Crockett et al. concluded that which includes 15 words (11 sensory, four affec- “the repeated demonstration that the MPQ as- tive) (25). These words are identified with an as- sesses substantially more than the three compo- terisk in Exhibit 9.4; note that Melzack nents originally proposed suggests that presented hot and burning together, as “hot- considerable caution is warranted with respect burning,” and likewise combined “tiring- to using the ‘sensory-evaluative-affective’ exhausting” and “punishing-cruel.” Each word method of scoring this test” (19, p461). Turk et or phrase is rated on a four-point intensity scale: al., however, concluded that the factor analyses zero = none, one = mild, two = moderate, and confirmed the three-factor structure (sensory, af- three = severe. The six-point PPI rating is also in- fective, evaluative), although these factors were cluded, as is a visual analogue scale (25, Figure intercorrelated (21). Although there is room for 1). The short form was found to perform almost 488 Measuring Health as well as the full MPQ in a study of cancer pain PRI-sensory score of 16.3, compared with pa- (26). Other abbreviations of the MPQ can be tients with cancer (12.1) or patients with dental found on the web; it is frequently not clear pain (10.7), but overall, the differences between whether these have been validated. conditions were small (5, p39). Wilkie et al. also Because of the linguistic nuances in the reported the frequency of reporting particular MPQ, translation is challenging. Nonetheless, adjectives by each type of pain (5, Table 3). there are versions in most European languages There were no clear patterns in the words chosen including French (27), Italian (28), German (29), by patients with particular types of pain. Dutch (30; 31), Danish (32), Norwegian (33), and Finnish (34). There is an Arabic version Commentary (35); a Mexican Spanish version has been tested Melzack’s MPQ is the leading instrument for de- (36). Many of these adapt, rather than translate, scribing the diverse dimensions of pain. It was the original and include widely varying numbers based on an explicit theory of pain. It is used ex- of adjectives: results cannot be compared across tensively in many countries and has taken on the them. A Swedish abbreviated version, for exam- status of a gold standard against which newer ple, contains 15 items and appears to have ade- instruments are compared. Sections of the MPQ quate reliability and validity (37). At times, the have been incorporated into other scales, such as linguistic challenge of translating the MPQ the Brief Pain Inventory, the MOS Pain Mea- seems too strong to resist: there are at least three sures, the Back pain Classification Scale, and the Canadian French versions, as well as a French- Pain Perception Profile reviewed in this chapter. French version. To the surprise of few, a study Nonetheless, Melzack originally considered the from Paris declared the version from France to MPQ as a preliminary version of a measurement be superior (27). method rather than a definitive scale, and vari- There have been various extensions to the ants have been used in several studies. A limita- MPQ. One version, called the McGill Compre- tion of using verbal descriptors is that people’s hensive Pain Questionnaire, covers details of the use of pain language is probably not sufficiently patient’s illness, personality, milieu, and coping consistent for the MPQ to be used to diagnose resources (38). The Dartmouth Pain Question- types of pain. In analyses of groups of patients, naire adds questions on pain complaints, so- responses for different types of pain are distin- matic pain symptoms, and reductions in guishable, but it is unlikely that a patient’s re- function and in self-esteem; some validity evi- sponses could be used to diagnose the cause of dence is available (39). Another version added pain (20). This is especially true in patients with computer animation to illustrate categories such psychological disturbance (41). as pressure (e.g., picture of a clamp squeezing a Discussions of the MPQ have raised four ball); throbbing (e.g., a hammer falling from main issues: the constitution of subcategories various heights); piercing (e.g., a needle), and within the questionnaire; the suitability of select- burning (40). ing only one word from each category; the choice of summary scores, and the question of Reference Standards whether the MPQ does, in fact, reflect Melzack’s Wilkie et al. analyzed pain scores from studies theory of pain. The approach of selecting only of patients with different types of pain: cancer, one word in each category assumes that the chronic back pain, postoperative, labor, dental, words in each category are homogeneous and among others. They presented mean pooled differ only in intensity. Some of the early studies scores for seven pain conditions as follows: PRI- by Reading et al. suggest that this approach is total = 23; PRI-sensory = 13.9; PRI-affective = 3.3; valid (42, p381), although this would appear to PRI-evaluative = 2.5, and PRI-miscellaneous = 4.9; be a suitable topic for IRT. In terms of scoring NWC = 9.2 and PPI = 2.3 (5, Table 2). They then the MPQ, the 20 category scores proved supe- presented profiles for each condition. For exam- rior to the three subscale scores in one study ple, patients with low back pain had a high mean (42). Hand and Reading found no advantage in  Pain Measurements 489 an alternative scoring approach that used 16 in- named McGill who could apparently be located stead of 20 categories (43). in the United Kingdom (to judge from the name, We may enquire whether the MPQ does, in perhaps Scotland?) Lest anyone else may be con- fact, succeed in reflecting Melzack’s theory of fused, Ronald Melzack is an emeritus professor pain. Crockett et al. argued that the results of of psychology at McGill University in Montreal, their factor analyses offered empirical validation Quebec, Canada. Aside from his work on pain, of Melzack’s a priori classification of pain de- he is also the author of books on Inuit folklore. scriptors and concurred with Melzack’s empha- sis on the need to describe pain in terms of several dimensions rather than as a single inten- References sity score. However, debate continues over the (1) Melzack R. The McGill Pain structure of the MPQ, and technical problems in Questionnaire: major properties and examining how closely it reflects Melzack’s the- scoring methods. Pain 1975;1:277–299. ory have enraptured methodologists. For exam- (2) Melzack R. Psychologic aspects of pain. ple, there are several problems in using factor Pain 1980;8:143–154. analysis to assess the validity of the question- (3) Melzack R, Wall PD. The challenge of naire. Evidently, the conventional response pro- pain. 2nd ed. New York: Penguin Books, cedure cannot be used whereby the respondent 1988. selects only one word in each subclass, because (4) Melzack R, Katz J. The gate control the correlations among the words will then be theory: reaching for the brain. In: Hadjistavropoulos T, Craig KD, eds. Pain: zero, and so the grouping of words into sub- psychological perspectives. Mahwah, New classes cannot be tested empirically. Melzack Jersey: Lawrence Erlbaum, 2004:13–34. and Torgerson recognized that words from dif- (5) Wilkie DJ, Savedra MC, Holzemer WL, et ferent components (e.g., affective, evaluative) al. Use of the McGill Pain Questionnaire to may correlate with one another, whereas differ- measure pain: a meta-analysis. Nurs Res ent words in each group, or different subclasses 1990;39:36–41. in each component, will not necessarily intercor- (6) Melzack R, Torgerson WS. On the relate (6). If this is the case, one would not ex- language of pain. Anesthesiology pect words in, say, the sensory component to 1971;34:50–59. load on a single factor. Furthermore, because (7) Crockett DJ, Prkachin KM, Craig KD. each word reflects both a type and an intensity Factors of the language of pain in patient and volunteer groups. Pain of pain, a factor analysis may extract type of 1977;4:175–182. pain or intensity of pain factors, or both. This (8) Leavitt F, Garron DC, Whisler WW, et al. was illustrated by a study that departed from Affective and sensory dimensions of back normal usage and presented the MPQ words in pain. Pain 1978;4:273–281. random order, asking subjects to check every (9) Melzack R. The McGill Pain word that described their pain (8). The seven fac- Questionnaire. In: Melzack R, ed. Pain tors that were interpretable cut across Melzack’s measurement and assessment. New York: groupings and included words at similar levels of Raven Press, 1983:41–47. intensity from a wide range of subclasses. (10) Melzack R, Katz J, Jeans ME. The role of Although it is hard to show that Melzack’s compensation in chronic pain: analysis questionnaire does reflect his conceptual defini- using a new method of scoring the McGill Pain Questionnaire. Pain tion of pain, the MPQ is still the leading pain 1985;23:101–112. measurement scale, and we recommend its con- (11) Perry F, Heller PH, Levine JD. Differing tinued use. On a lighter note, readers should al- correlations between pain measures in ways be wary of the information contained in syndromes with or without explicable apparently authoritative Web sites. One (ac- organic pathology. Pain 1988;34:185–189. cessed in the fall of 2004) that may best remain (12) Dubuisson D, Melzack R. Classification of nameless attributes the MPQ to an author clinical pain descriptions by multiple group 490 Measuring Health discriminant analysis. Exp Neurol (26) Dudgeon D, Raubertas RF, Rosenthal SN. 1976;51:480–487. The short-form McGill Pain Questionnaire (13) Taenzer P. Postoperative pain: relationships in chronic cancer pain. J Pain Symptom among measures of pain, mood, and Manage 1993;8:191–195. narcotic requirements. In: Melzack R, ed. (27) Boureau F, Luu M, Doubrère JF. Pain measurement and assessment. New Comparative study of the validity of four York: Raven Press, 1983:111–118. French McGill Pain Questionnaire (MPQ) (14) Ahles TA, Ruckdeschel JC, Blanchard EB. versions. Pain 1992;50:59–65. Cancer-related pain—II. Assessment with (28) DeBenedittis G, Massei R, Nobili R, et al. visual analogue scales. J Psychosom Res The Italian Pain Questionnaire. Pain 1984;28:121–124. 1988;33:53–62. (15) Prieto EJ, Hopson L, Bradley LA, et al. The (29) Radvila A, Adler RH, Galeazzi RL, et al. language of low back pain: factor structure The development of a German language of the McGill Pain Questionnaire. Pain (Berne) pain questionnaire and its 1980;8:11–19. application in a situation causing acute (16) Reading AE. The internal structure of the pain. Pain 1987;28:185–195. McGill Pain Questionnaire in (30) Vanderiet K, Adriaensen H, Carton H, et dysmenorrhoea patients. Pain al. The McGill Pain Questionnaire 1979;7:353–358. constructed for the Dutch language (MPQ- (17) Prieto EJ, Geisinger KF. Factor-analytic DV). Preliminary data concerning studies of the McGill Pain Questionnaire. reliability and validity. Pain In: Melzack R, ed. Pain measurement and 1987;30:395–408. assessment. New York: Raven Press, (31) Vermote R, Ketelaer P, Carton H. Pain in 1983:63–70. multiple sclerosis patients. A prospective (18) Holroyd KA, Holm JE, Keefe FJ, et al. A study using the McGill Pain Questionnaire. multi-center evaluation of the McGill Pain Clin Neurol Neurosurg 1986;88:87–93. Questionnaire: results from more than (32) Drewes AM, Helweg-Larsen S, Petersen P, 1700 chronic pain patients. Pain et al. McGill Pain Questionnaire translated 1992;48:301–311. into Danish: experimental and clinical (19) Crockett DJ, Prkachin KM, Craig KD, et findings. Clin J Pain 1993;9:80–87. al. Social influences on factored dimensions (33) Strand LI, Wisnes AR. The development of of the McGill Pain Questionnaire. J a Norwegian pain questionnaire. Pain Psychosom Res 1986;30:461–469. 1991;46:61–66. (20) Kremer EF, Atkinson JH. Pain language: (34) Ketovuori H, Pontinen PJ. A pain affect. J Psychosom Res 1984;28:125–132. vocabulary in Finnish—the Finnish Pain (21) Turk DC, Rudy TE, Salovey P. The McGill Questionnaire. Pain 1981;11:247–253. Pain Questionnnaire reconsidered: (35) Harrison A. Arabic pain words. Pain confirming the factor structure and 1988;32:239–250. examining appropriate uses. Pain (36) Escalante A, Lichtenstein MJ, Rios N, et 1985;21:385–397. al. Measuring chronic rheumatic pain in (22) Melzack R. Discriminative capacity of the Mexican Americans: cross-cultural McGill Pain Questionnaire. Pain adaptation of the McGill Pain 1985;23:201–203. Questionnaire. J Clin Epidemiol (23) Torgerson WS, BenDebba M. The structure 1996;49:1389–1399. of pain descriptors. In: Melzack R, ed. Pain (37) Burckhardt CS, Bjelle A. A Swedish version measurement and assessment. New York: of the short-form McGill Pain Raven, 1983:49–54. Questionnaire. Scand J Rheumatol (24) Skevington SM. Investigating the 1994;23:77–81. relationship between pain and discomfort (38) Monks R, Taenzer P. A comprehensive and quality of life, using the WHOQOL. pain questionnaire. In: Melzack R, ed. Pain Pain 1998;76:395–406. measurement and assessment. New York: (25) Melzack R. The short-form McGill Pain Raven Press, 1983:233–237. Questionnaire. Pain 1987;30:191–197. (39) Corson JA, Schneider MJ. The Dartmouth  Pain Measurements 491 Pain Questionnaire: an adjunct to the “sensory” aspects of pain, functional limita- McGill Pain Questionnaire. Pain tions, and distress that result from pain (the “re- 1984;19:59–69. active” dimension) (1; 3). These two dimensions (40) Swanston M, Abraham C, Macrae WA, et represented a simplification of Melzack’s al. Pain assessment with interactive sensory-affective-evaluative model of pain re- computer animation. Pain sponse (1, p130). Although the two dimensions 1993;53:347–351. are conceptually distinct, they are expected to (41) Atkinson JH, Kremer EF, Ignelzi RJ. Diffusion of pain language with affective intercorrelate. disturbance confounds differential diagnosis. Pain 1982;12:375–384. Description (42) Reading AE, Hand DJ, Sledmore CM. A Daut et al. originally developed a scale called the comparison of response profiles obtained Wisconsin Brief Pain Questionnaire (4); this re- on the McGill Pain Questionnaire and an tained some of the pain descriptors from the adjective checklist. Pain 1983;16:375–383. McGill questionnaire, with the diagram showing (43) Hand DJ, Reading AE. Discriminant the location of the pain. Tests of validity led to function analysis of the McGill Pain the deletion of redundant items and the selection Questionnaire. Psychol Rep of others to form the BPI. 1986;59:763–770. The BPI includes four ratings of the intensity of pain (items 3–7), and seven that cover the impact of pain (Exhibit 9.5). Intensity is The Brief Pain Inventory recorded on numerical scales running from zero (Formerly the Wisconsin (no pain) to ten (“pain as bad as you can imag- Brief Pain Questionnaire) ine”). The numerical rating format was chosen (Charles S. Cleeland, 1982) over the visual analogue scale as offering rea- Purpose sonable precision, being simple to administer, The Brief Pain Inventory (BPI) is designed as a and allowing for oral administration if required practical measure for use in clinical settings to (1, p130). Given pain may vary in the course of record pain severity and its impact on the pa- a day, intensity is rated at the time of complet- tient’s functioning. Although designed for pa- ing the questionnaire (pain now), and also at its tients with cancer-related pain, the BPI can also worst, least, and average over the past day or be used for other diseases; it can be self- week. Alternatively, ratings can cover the last administered or used in a clinical interview (1). 24 hours; the study design will indicate the most appropriate period (1, p131). Final inten- Conceptual Basis sity scores may represent the worst pain or the Cleeland and Ryan argued that effective inter- average of the four ratings. Cleeland denoted vention for cancer must control the intensity of “significant pain” as pain that is rated higher pain and restore function that has been limited than the midpoint on the pain intensity scales by pain. They noted that communication be- (3, p298). tween patient and doctor about pain is often un- The impact of the pain is recorded in terms of structured, which “maximizes the chances that how much it interferes with seven fields of nor- personality, cultural, linguistic and situational mal activity (items 14 A–G). These ratings are variables will bias the communication” (1, made on zero-to-ten numeric scales running p130). They argued the need for a practical mea- from “no interference” to “interferes com- sure of the severity of pain. The McGill Pain pletely.” The mean of these scores indicates the Questionnaire was rejected as being too long level of pain interference; the mean score corre- and difficult, and limited by its omission of pain lated highly with a more complex approach history and interference with activities (2). Ac- based on factor analysis (1, p131). cordingly, the Brief Pain Inventory was devel- The BPI also records the location of the pain oped for use in clinical settings to record on a diagram of a human figure; this can offer Exhibit 9.5 The Brief Pain Inventory 1. Throughout our lives, most of us have had pain from time to time (such as minor headaches, sprains, and toothaches). Have you had pain other than these everyday kinds of pain during the last week? 1. Yes 2. No IF YOU ANSWERED YES TO THIS QUESTION, PLEASE GO ON TO QUESTION 2 AND FINISH THIS QUESTIONNAIRE. IF NO, YOU ARE FINISHED WITH THE QUESTIONNAIRE. THANK YOU. 2. On the diagram, shade in the areas where you feel pain. Put an X on the area that hurts the most. 3. Please rate your pain by circling the one number that best describes your pain at its worst in the last week. 0 1 2 3 4 5 6 7 8 9 10 No Pain as bad as Pain you can imagine 4. Please rate your pain by circling the one number that best describes your pain at its least in the last week. 0 1 2 3 4 5 6 7 8 9 10 No Pain as bad as Pain you can imagine 5. Please rate your pain by circling the one number that best describes your pain on the average. 0 1 2 3 4 5 6 7 8 9 10 No Pain as bad as Pain you can imagine 6. Please rate your pain by circling the one number that tells how much pain you have right now. 0 1 2 3 4 5 6 7 8 9 10 No Pain as bad as Pain you can imagine 492 Exhibit 9.5 7. What kinds of things make your pain feel better (for example, heat, medicine, rest)? 8. What kinds of things make your pain worse (for example, walking, standing, lifting)? 9. What treatments or medications are you receiving for your pain? 10. In the last week, how much relief have pain treatments or medications provided? Please circle the one percentage that most shows how much relief you have received. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% No Complete Relief Relief 11. If you take pain medication, how many hours does it take before the pain returns? 1. Pain medication doesn’t help at all. 5. Four hours. 2. One hour. 6. Five to twelve hours. 3. Two hours. 7. More than twelve hours. 4. Three hours. 8. I do not take pain medication. 12. Circle the appropriate answer for each item. I believe my pain is due to: Yes No 1. The effects of treatment (for example, medication, surgery, radiation, prosthetic device). Yes No 2. My primary disease (meaning the disease currently being treated and evaluated). Yes No 3. A medial condition unrelated to primary disease (for example, arthritis). 13. For each of the following words, check yes or no if that adjective applies to your pain. Aching Yes No Throbbing Yes No Shooting Yes No Stabbing Yes No Gnawing Yes No Sharp Yes No Tender Yes No Burning Yes No Exhausting Yes No Tiring Yes No Penetrating Yes No Nagging Yes No Numb Yes No Miserable Yes No Unbearable Yes No (continued) 494 Measuring Health Exhibit 9.5 (continued) 14. Circle the one number that describes how, during the past week, pain has interfered with your: A. General Activity 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes B. Mood 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes C. Walking ability 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes D. Normal work (includes both work outside the home and housework) 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes E. Relations with other people 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes F. Sleep 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes G. Enjoyment of life 0 1 2 3 4 5 6 7 8 9 10 Does not Completely interfere interferes Adapted from an original provided by Dr. CS Cleeland. With permission. clues to the physical mechanisms leading to the cient alpha values for the four pain intensity pain, such as when a tumor impinges on a nerve items were 0.87 for the English version (N = (1, p132). Finally, patients are asked to select 1106), 0.85 for the French version (N = 324), words (item 13, taken from the McGill Pain 0.86 for the Mandarin Chinese version Questionnaire) that best describe their pain, and (N = 200), and 0.80 for the Filipino version they indicate the extent and duration of pain re- (N = 267). Alphas for the interference scale in- lief obtained from analgesics (5). The instru- clude 0.91 (English), 0.90 (French), 0.91 (Chi- ment takes ten to 15 minutes to complete. nese), and 0.86 (Filipino) (C.S. Cleeland, Comparable results have been obtained from personal communication, 1995). Figures of 0.85 self- and interviewer-administered versions (3) for interference and 0.89 for severity were re- (Ex 9.5). ported in a Greek study (6). In a manner similar to Guttman analysis, Cleeland and Ryan have described a hierarchical Reliability structure in the interference dimension whereby Reliability data for four versions of the BPI in more activities become affected as pain level in- different languages have been collected. Coeffi- creases (1, Table 1).  Pain Measurements 495 Validity nese have been validated (1). The scale has also Factor analyses have consistently supported the been translated into other languages, including division into severity and interference factors in Arabic, German, Norwegian, and Russian. studies in the United States, France, China, and These are listed on the Qolid web site, at the Philippines (1, Table 2); congruent results www.qolid.org/public/BPI.html. came from a Greek study (6). The BPI detected expected differences in Commentary severity of pain between groups of patients who Like the Medical Outcomes Study Pain Mea- differed in the site of their disease, in their re- sures reviewed next in this chapter, the BPI was quirements for analgesics, and in the presence of based on the Wisconsin Brief Pain Question- metastases (3, p299). Increased pain severity cor- naire. The BPI offers a brief rating method that responded to reports of disturbed sleep, activity lies between simple, numerical, or visual ana- limitations, and reliance on medications (7). In a logue ratings and more complex methods such as discriminant analysis, the BPI items distin- McGill Pain Questionnaire. Cleeland and Ryan guished between groups of patients receiving dif- have described applications of the BPI in clinical ferent types of analgesia; the intensity scales practice, epidemiological studies, quality control were more discriminative than the interference studies of pain management strategies, and clini- items (5, Table 3). cal trials (1, p132). The correlations among the pain intensity Although the intensity and interference rat- ratings (e.g., now, least, worst, average) fell in ings intercorrelate, they are rated somewhat in- the range 0.57 to 0.80, whereas correlations dependently and so are not redundant. The among the interference scales ranged from 0.44 consistency of findings across several studies to 0.83 (5, Table 1). Correlations between the suggests that the scales are robust. Cross- intensity and interference ratings ran from 0.27 cultural applications of the BPI have led to some to 0.63. Factor analysis confirmed the separation interesting comparisons of pain experience in of severity from interference factors, and this has different cultures and indicates that the scale been replicated in studies in Vietnam, Mexico, would be suitable for international trials of anal- and the Philippines (3, Table 2; 5, Table 2). gesia and cancer treatment. The BPI has been The internal structure of the BPI appears log- used in a multicenter study of pain in the United ically consistent: as ratings of pain intensity in- States (10). It was carefully formulated on the crease, the interference items are endorsed in a basis of a critical review of previous measures; it sequence running from work, to mood, sleep, incorporates components that have been tested activity, walking, and, finally, relations with oth- and is founded on a conceptual approach. Hope- ers (3, Table 1). fully, users of the scale will present further vali- dation results. Alternative Forms A short form, the BPI-SF, was developed for clin- Address ical trials where multiple assessments of pain The BPI is reproduced on the web, at www.city were required. Alpha coefficients were 0.85 or ofhope.org/prc/pdf/BPI%20Long%20Version.pdf. higher for the severity and pain interference sub- scales; retest coefficients ranged from 0.58 to 0.95 according to the retest delay (8). Copies are References available from various web sites, for example (1) Cleeland CS, Ryan KM. Pain assessment: at www.stat.washington.edu/TALARIA/attachb1 global use of the Brief Pain Inventory. Ann .html or at www.ndhcri.org/pain/Tools/Brief_ Acad Med Singapore 1994;23:129–138. Pain_Inventory.pdf, or www.ama-cmeonline.com/ (2) Cleeland CS. Measurement and prevalence pain_mgmt/module08/pop_up/pop_bpi.htm. of pain in cancer. Semin Oncol Nurs Translations into Italian (9), French, Spanish, 1985;1:87–92. Greek (6), Vietnamese (5), and Mandarin Chi- (3) Cleeland CS. Pain assessment in cancer. In: 496 Measuring Health vant outcome indicators for evaluating pain Osoba D, ed. Effect of cancer on quality of treatment (1). “For those concerned with the re- life. Boca Raton, Florida: CRC Press, 1991:293–305. turn of the patient to normal functioning and the (4) Daut RL, Cleeland CS, Flanery RC. assessment of pain relief, it seems important to Development of the Wisconsin Brief Pain measure not only the intensity of the pain expe- Questionnaire to assess pain in cancer and rience but the effects of pain on normal func- other diseases. Pain 1983;17:197–210. tional abilities” (1, p221). Sherbourne termed (5) Cleeland CS, Ladinsky JL, Serlin RC, et al. these the sensory and the performance aspects of Multidimensional measurement of cancer the pain experience (1, p223). Pain measures pain: comparisons of U.S. and Vietnamese have not typically been included in health sur- patients. J Pain Symptom Management veys and Sherbourne wished to compare pain 1988;3:23–27. and other aspects of health. The intention was to (6) Mystakidou K, Mendoza TR, Tsilika E, et develop pain measures that were not specific to a al. Greek Brief Pain Inventory: validation and utility in cancer pain. Oncology particular disease. 2001;60:35–42. (7) Daut RL, Cleeland CS. The prevalence and Description severity of pain in cancer. Cancer The MOS pain measurement was based largely 1982;50:1913–1918. on the Wisconsin Brief Pain Questionnaire (see (8) Mendoza TR, Chen C, Brugger A, et al. review of the Brief Pain Inventory). Twelve self- The utility and validity of the modified report items cover the severity of pain over the brief pain inventory in a multiple-dose past four weeks and its effect on mood and be- postoperative analgesic trial. Clin J Pain haviors (Exhibit 9.6). 2004;20:357–362. When the questions are used in a survey, a fil- (9) Caraceni A, Mendoza TR, Mencaglia E, et ter question “Did you experience any bodily al. A validation study of an Italian version of the Brief Pain Inventory (Breve pain during the past 4 weeks?” may be added to questionario per la valutazione del dolore). identify those for whom the pain questions need Pain 1996;65:87–92. not be asked (1, p224–225). Those who skip the (10) Cleeland CS, Gonin R, Hatfield AK, et al. questions are given a score of one (no interfer- Pain and its treatment in outpatients with ence) on item four, the effects of pain measure. metastatic cancer. N Engl J Med People who experienced pain more than once 1994;330:592–596. during this time are asked to describe their feel- ings of pain “in general” (1, p225). A principal components analysis identified two factors, corresponding to severity and pain The Medical Outcomes Study Pain effects. The question on the number of days that Measures pain interfered with activities loaded on both (Cathy D. Sherbourne, 1992) factors, so Sherbourne suggested three scores: ef- Purpose fects of pain (items 4a–4f ), pain severity (items The Medical Outcomes Study (MOS) Pain Mea- 1, 2, 3, 6, and 7), and days pain interfered (item sures cover severity in terms of the intensity, fre- 5). An overall score can also be calculated (1, quency, and duration of pain and record the p228). Because she was working with large sam- impact of pain on behavior and moods (1, ples in a survey setting, Sherbourne used stan- p223). They are intended for use as outcome dard scoring approaches (see Glossary) in measures and in population surveys. calculating some of the scores. The pain effects score is calculated by averaging across all six Conceptual Basis items, giving a score from one to five; this is Among the many ways of classifying pain, Sher- transformed to a zero-to-100 scale. For the pain bourne selected intensity, frequency, duration, severity scale, each item is first standardized to a and impact of pain on behavior as the most rele- mean of zero and standard deviation of one; then Exhibit 9.6 The Medical Outcomes Study Pain Measures The following questions are about the pain or pains you experienced in the past 4 weeks. If you had more than one pain, answer the questions by describing your feelings of pain in general. 1. How much bodily pain have you generally had during the past 4 weeks? (Circle One) None . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Very mild . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Mild . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Moderate . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Severe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Very severe . . . . . . . . . . . . . . . . . . . . . . . . . . .6 2. During the past 4 weeks, how often have you had pain or discomfort? (Circle One) Once or twice . . . . . . . . . . . . . . . . . . . . . . . . .1 A few times . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Fairly often . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Very often . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Every day or almost every day . . . . . . . . . . . . .5 3. When you had pain during the past 4 weeks, how long did it usually last? (Circle One) A few minutes . . . . . . . . . . . . . . . . . . . . . . . . .1 Several minutes to an hour . . . . . . . . . . . . . . . .2 Several hours . . . . . . . . . . . . . . . . . . . . . . . . . .3 A day or two . . . . . . . . . . . . . . . . . . . . . . . . . .4 More than two days . . . . . . . . . . . . . . . . . . . .5 4. During the past 4 weeks, how much did pain interfere with the following things? (Circle One Number on Each Line) Not A Little Quite At All Bit Moderately A Bit Extremely a. Your mood . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 b. Your ability to walk or move about . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 c. Your sleep . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 d. Your normal work (including both work outside the home and housework) . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 e. Your recreational activities . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 f. Your enjoyment of life . . . . . . . . . . . . . . . 1 2 3 4 5 5. During the past 4 weeks, how many days did pain interfere with the things you usually do? Your answer may range from 0 to 28 days.) WRITE IN # OF DAYS: ____________ 6. Please circle the one number that best describes your pain on the average over the past 4 weeks. Pain As Bad As No You Can Pain Imagine 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 7. Please circle the one number that best describes your pain at its worst over the past 4 weeks. Pain As Bad As No You Can Pain Imagine 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 From Stewart AL, Ware JE Jr. Measuring functioning and well-being: the Medical Outcomes Study approach. Durham, North Car- olina: Duke University Press, 1992:374,378–379. With permission. 497 498 Measuring Health the items are averaged. High scores indicate present form so that evidence for the adequacy more pain, and the range of pain severity scores of the instrument can accumulate. in the validation study ran from −1.17 to +2.26 (1, Table 13–5). The overall score is calculated by first standardizing each item to a mean of References zero and a standard deviation of one, and then (1) Sherbourne CD. Pain measures. In: Stewart averaging all items (1, p231). AL, Ware JE, Jr., eds. Measuring functioning and well-being: the Medical Reliability Outcomes Study approach. Durham, North The severity and effects of pain items all corre- Carolina: Duke University Press, lated at 0.57 or higher with their scale scores (1, 1992:220–234. Table 13–3). Internal consistency for the overall (2) Gonzalez VM, Stewart A, Ritter PL, et al. score was 0.93; for the effects score it was 0.91 Translation and validation of arthritis and for the severity score 0.86 (1, Table 13–5). outcome measures into Spanish. Arthritis Rheum 1995;38:1429–1446. Validity Sherbourne presented correlations of the four The Oswestry Low Back Pain pain measures with 15 criterion scores drawn Disability Questionnaire from the MOS (N = 3,053). Correlations with a (Jeremy Fairbank, 1980, physical symptoms score ranged from 0.52 (for Revised 1986) days pain interfered) to 0.68 (for the overall pain score); correlations with a health distress score Purpose ranged from 0.43 to 0.57 (1, Table 13–7). No The Oswestry questionnaire indicates the extent correlations with other pain measurement scales to which a person’s functional level is restricted are available. by back or leg pain. It was intended for clinical use and is completed by the patient. Alternative Forms A Spanish translation has been described (2). Conceptual Basis No information is available. Commentary The MOS pain instrument offers a brief measure Description suitable for surveys or for clinical settings when The Oswestry Low Back Pain Disability Ques- the goal is to assess the impact of pain on daily tionnaire includes ten six-point scales. The first living, rather than to provide a detailed assess- rates the intensity of pain and the remaining nine ment of the nature of the pain. Although little in- cover the disabling effect of pain on typical daily formation on validity is available, the method activities: personal care, lifting, walking, sitting, seems potentially valuable as a midsized mea- standing, sleeping, sex life, social life, and travel- sure, falling between the visual analogue scales ing (Exhibit 9.7). The questions concentrate on and the longer scales described elsewhere in this the effects, rather than the nature, of pain. chapter. If the MOS instrument is to gain accep- The patient selects the one statement in each tance, it will have to be compared with other section that most accurately describes the effect pain assessments and be tested for sensitivity to of her pain; if two items are marked, the more change. The 20-point rating scales should have severe is scored. Each section is scored on a zero adequate sensitivity to record changes in pain to five scale, with higher values representing levels, although Sherbourne did suggest that ask- greater disability. The sum of the ten scores is ing directly about pain relief may be valuable (1, expressed as a percentage of the maximum p234). Finally, we suggest that the development score; this is termed the Oswestry Disability In- of alternative versions of the MOS scales be re- dex (ODI). If the patient fails to complete a sec- sisted and that the questions be used in their tion, the percentage score is adjusted (i.e., the Exhibit 9.7 The Oswestry Low Back Pain Disability Questionnaire This questionnaire has been designed to give us information as to how your back or leg pain has affected your ability to manage in everyday life. Please answer every section and mark in each section ONLY THE ONE BOX which applies to you. We realise you may consider that two of the statements in any one section relate to you, but PLEASE JUST MARK THE BOX WHICH MOST CLEARLY DESCRIBES YOUR PROBLEM. Section 1—Pain Intensity Section 6—Standing I have no pain at the moment. I can stand as long as I want without extra pain. The pain is very mild at the moment. I can stand as long as I want but it gives me extra pain. The pain is moderate at the moment. Pain prevents me from standing for more than 1 hour. The pain is fairly severe at the moment. Pain prevents me from standing for more than 1/2 hour. The pain is very severe at the moment. Pain prevents me from standing for more than 10 minutes. The pain is the worst imaginable at the moment. Pain prevents me from standing at all. Section 2—Personal Care (washing, dressing, etc.) Section 7—Sleeping I can look after myself normally without causing extra My sleep is never disturbed by pain. pain. My sleep is occasionally disturbed by pain. I can look after myself normally but it is very painful. Because of pain I have less than 6 hours sleep. It is painful to look after myself and I am slow and Because of pain I have less than 4 hours sleep. careful. Because of pain I have less than 2 hours sleep. I need some help but manage most of my personal care. Pain prevents me from sleeping at all. I need help every day in most aspects of self care. Section 8—Sex Life (if applicable) I do not get dressed, wash with difficulty and stay in bed. My sex life is normal and causes no extra pain. Section 3—Lifting My sex life is normal but causes some extra pain. I can lift heavy weights without extra pain. My sex life is nearly normal but is very painful. I can lift heavy weights but it gives extra pain. My sex life is severely restricted by pain. Pain prevents me from lifting heavy weights off the floor My sex life is nearly absent because of pain. but I can manage if they are conveniently positioned, e.g. on a table. Pain prevents any sex life at all. Pain prevents me from lifting heavy weights but I can Section 9—Social Life manage light to medium weights if they are conveniently My social life is normal and causes me no extra pain. positioned. My social life is normal but increases the degree of I can lift only very light weights. pain. I cannot lift or carry anything at all. Pain has no significant effect on my social life apart Section 4—Walking from limiting my more energetic interests, e.g. sport, etc. Pain does not prevent me walking any distance. Pain has restricted my social life and I do not go out as Pain prevents me walking more than 1 mile. often. Pain prevents me walking more than 1/4 mile. Pain has restricted social life to my home. Pain prevents me walking more than 100 yards. I have no social life because of pain. I can only walk using a stick or crutches. Section 10—Traveling I am in bed most of the time and have to crawl to the toilet. I can travel anywhere without pain. Section 5—Sitting I can travel anywhere but it gives extra pain. I can sit in any chair as long as I like. Pain is bad but I manage journeys over two hours. I can sit in my favorite chair as long as I like. Pain restricts me to journeys of less than one hour. Pain prevents me from sitting for more than 1 hour. Pain restricts me to short necessary journeys under 30 Pain prevents me from sitting for more than 1/2 hour. minutes. Pain prevents me from sitting for more than 10 minutes. Pain prevents me from traveling except to receive Pain prevents me from sitting at all. treatment. From an original provided by Dr. J. Fairbank. With permission. 499 500 Measuring Health score would be expressed as a percentage of 45 Morris Pain Questionnaire (6). An effect size of rather than 50). ODI percentage scores from 0.80 has been reported for the Oswestry ques- zero to 20 represent minimal disability, 20 to 40 tionnaire, compared with 1.58 for a single VAS moderate disability, 40 to 60 severe disability, rating of pain during the last week, and 2.02 for while scores exceeding 60 indicate that the pa- the Roland Morris questionnaire (7, Table IV). tient is severely disabled by pain in several areas The same authors also expressed sensitivity to of life (1). change in terms of the area under the ROC When self-administered, the questionnaire curve; the results were lowest for the Oswestry, takes less than five minutes to complete and one at 0.76, compared with 0.93 for the Roland and minute to score; if the questions are read to a pa- 0.91 for the VAS (7, Table V). In the study of tient, it requires about ten minutes. chiropractic patients, responsiveness of the Os- westry instrument ranged from 0.70 to 0.83, Reliability fractionally better than equivalent figures for the Twenty-two patients with low back pain com- Dallas Pain Questionnaire (5, Table 5). pleted the questionnaire twice, on consecutive days, producing a test-retest correlation of 0.99 Alternative Forms (1, p273). Fairbank also reported “a good inter- A computerized version has been described (2, nal consistency,” but offered no summary statis- Appendix 12.2); this correlated 0.89 with the tic (1). conventional version (2, p176). A version that was developed for studies of Validity chiropractic care omitted the section on sex life, In a study of 25 patients suffering from a first at- altered the remaining sections, and added a sec- tack of low back pain, which was expected to re- tion on changes in level of pain (8). Fairbank, mit spontaneously, scores on the questionnaire however, does not recommend this version (Per- showed a significant improvement over a three- sonal communication, 1989). A more recent week period (1). modification was proposed by Fritz and Irrgang The Oswestry questionnaire has been com- (9). This brings the language into a North Amer- pared with the St. Thomas’s Pain Questionnaire; ican idiom, removes the section on sex life, and unfortunately, no statistical summary of the as- adds one on employment or homemaking. sociation was given (2, p177). The Oswestry scale has been used in a study of the predictors of outcomes five years after surgery. Baseline Os- Commentary westry scores, and the number of previous oper- This questionnaire is reviewed here because it il- ations predicted outcomes; the Oswestry scores lustrates a measurement of disability and handi- correlated −0.23 with the Sense of Coherence cap due to pain, rather than of pain as scale that was used to indicate ability to cope impairment. As such, it provides potentially with pain and disability (3). In a study of low valuable information in addition to VAS mea- back pain during pregnancy, Oswestry scores sures and the McGill questionnaire. Indeed, the correlated −0.57 with a measure of back muscle Oswestry scale is often used in combination activity during flexion. Correlations with a vi- with VAS as outcome measures in clinical trials sual analogue scale (VAS) pain score were 0.36 for low back pain management. The British during the first trimester, rising to 0.67 during Medical Research Council and the journal Spine the last (4, Table 2). A series of correlations with have recommended that the Oswestry question- a VAS pain scale ranging from 0.54 to 0.67 were naire be used as a standard measurement for as- reported in a study of chiropractic patients (5, sessing back pain, and it is reasonably Table 7). commonly used in clinical trials of back pain In identifying clinically significant pain, the management, especially in Europe (10). How- area under the ROC curve was 0.78, a result ever, there remain few reports of the validity and very similar to that obtained for the Roland reliability of this instrument, and the results are  Pain Measurements 501 adequate, but not impressive. Further analyses Back Pain Disability Scale. Phys Ther need to be carried out to assess its quality. 2001;81:776–788. (10) Torenbeek M, Caulfield B, Garrett M, et al. Current use of outcome measures for stroke and low back pain rehabilitation in References five European countries: first results of the (1) Fairbank JCT, Couper J, Davies JB, et al. ACROSS project. Int J Rehabil Res The Oswestry Low Back Pain Disability 2001;24:95–101. Questionnaire. Physiotherapy 1980;66:271–273. (2) Baker DJ, Pynsent PB, Fairbank JCT. The The Back Pain Classification Scale Oswestry Disability Index revisited: its (Frank Leavitt and David reliability, repeatability and validity, and a C. Garron, 1978) comparison with the St. Thomas’s Disability Index. In: Roland MO, Jenner Purpose JR, eds. Back pain: new approaches to The Back Pain Classification Scale (BPCS) is a rehabilitation and education. Manchester: screening device that distinguishes low back pain Manchester University Press, due to psychological disturbance from that due 1989:174–186. to organic disease (1). It was principally intended (3) Santavirta N, Bjorvell H, Konttinen YT, et as a clinical tool to identify patients with low al. Sense of coherence and outcome of low- back pain who would merit more thorough psy- back surgery: 5-year follow-up of 80 chological evaluation. patients. Eur Spine J 1996;5:229–235. (4) Sihvonen T, Huttunen M, Makkonen M, et al. Functional changes in back muscle Conceptual Basis activity correlate with pain intensity and Leavitt and Garron noted that for many patients prediction of low back pain during with low back pain, an organic disorder cannot pregnancy. Arch Phys Med Rehabil be demonstrated, so the pain may reflect psycho- 1998;79:1210–1212. logical distress: “People in psychological distress (5) Haas M, Jacobs GE, Raphael R, et al. Low are assumed to develop physical symptoms as a back pain outcome measurement means of communicating and/or managing emo- assessment in chiropractic teaching clinics: tional or interpersonal difficulties” (1, p149). responsiveness and applicability of two Diagnosing such problems is hard: “Patients are functional disability questionnaires. J habitually silent about psychological problems, Manipulative Physiol Ther 1995;18:79–87. and even the most astute of physicians are by (6) Stratford PW, Binkley J, Solomon P, et al. Assessing change over time in patients with their training poorly equipped to evaluate these low back pain. Phys Ther highly complex processes with any degree of so- 1994;74:528–533. phistication” (2, p79). Because of the difficulty (7) Beurskens AJHM, de Vet HCW, Köke in diagnosis, the existence of a psychological ba- AJA. Responsiveness of functional status in sis for pain is commonly inferred from the ab- low back pain: a comparison of different sence of organic pathology, rather than being instruments. Pain 1996;65:71–76. positively demonstrated in its own right (1). Ad- (8) Hudson-Cook N, Tomes-Nicholson K, dressing this problem, several studies have used Breen A. A revised Oswestry Disability Minnesota Multiphasic Personality Inventory Questionnaire. In: Roland MO, Jenner JR, (MMPI) scales to identify an emotional basis for eds. Back pain: new approaches to pain complaints; the MMPI Low Back Scale rehabilitation and education. Manchester: Manchester University Press, identifies consistent response patterns apparently 1989:187–204. reflecting pain of a psychosomatic origin (1). (9) Fritz JM, Irrgang JJ. A comparison of a The MMPI, however, has a high misclassifica- modified Oswestry Low Back Pain tion rate, and its length is not readily acceptable Disability Questionnaire and the Quebec to patients in pain. Leavitt therefore developed 502 Measuring Health the BPCS as an alternative means to distinguish the respondent are added, and a positive total between organic and psychological bases for low implies pain of psychological origin, whereas a back pain. negative score reflects pain of organic origin. The higher the score (in either direction), the more Description confirmed the diagnosis. The BPCS was derived from the observation that patients whose pain reflected psychological dis- Reliability turbance used verbal pain descriptors differently A 24-hour test-retest reliability of 0.86 was ob- than patients whose pain had an organic basis tained from a hospitalized sample (N = 114). A (3). Patients with pain attributable to a psycho- split-half reliability of 0.89 was obtained from logical cause (i.e., “functional pain”) described 158 patients hospitalized with low back pain (2, their pain as more variable, diffuse, and intense. p83). A five-day retest gave a reliability of only They used a wider variety of words to describe 0.44 (5, p274). their pain, typically endorsing more of the pain descriptors referring to affective response or skin Validity pressure (2; 3). Most validation studies have compared classifi- The BPCS forms one component of the cations by the BPCS with independent clinical longer Low Back Pain Symptom Checklist assessments. To check the discriminant analysis shown in Exhibit 9.8. This comprises 103 adjec- used to select the 13 adjectives, Leavitt and Gar- tives taken from the McGill Pain Questionnaire ron carried out a cross-validation study on a sep- and other sources. The 103 words include 71 arate sample. The scale correctly classified 132 that may be scored to provide seven pain scales out of 159 cases: a rate of 83% (1, p152). and the 13 words that form the BPCS. These are Sanders compared the classification made by the randomly distributed through the questionnaire, BPCS with one made on the basis of a medical and all 103 items are normally asked even if only record review. The BPCS correctly classified the 13 words comprising the BPCS are to be an- 80% of the patients with low back pain, but alyzed. The psychometric results reported below only 60% of 50 headache patients: no better refer only to the Classification Scale, which is than would have been achieved by classifying all shown in Exhibit 9.9. The seven scales included patients’ symptoms as psychopathological (6, in the Symptom Checklist were identified using Table 2). A more exacting validation studied factor analyses and describe various aspects of 174 patients who had clear organic pain; of the pain experience. The first factor describes these, 124 had no clinical evidence for psycho- emotional discomfort, whereas the second is a logical involvement, whereas 50 had organic dis- mix of emotional and sensory aspects of pain. ease plus psychological involvement (7). All but The remaining five cover sensory aspects of pain one of the patients with symptoms determined to (4). The Low Back Pain Symptom Checklist is be organic (99.2%) were correctly classified, self-administered and takes between five and ten whereas 86% of those with psychological pain minutes to complete. were classified correctly (7, Table 2). However, The BPCS was developed empirically from a to achieve this classification required 43 pain comparison of the responses of 62 patients words, rather than the 13 in the BPCS. Indeed, whose back pain was explained organically and seven of the 13 BPCS words were not among the 32 patients whose pain was judged (using a bat- 43 words (7, Table 5). A replication study used tery of mental and psychological tests) to be of the same 43 words and showed a small decline in psychological origin (1). A discriminant analysis rates of correct classification (7, Table 4). identified the 13 words that distinguished be- Leavitt studied 91 patients with low back pain tween the two types of patient (1, p152). The that could not be attributed to organic disease. BPCS is scored using weights derived from the This group was divided into 59 patients with clin- discriminant function analyses, shown in Ex- ically manifest symptoms of psychological disor- hibit 9.9. The weights for the items selected by ders and 32 patients without such symptoms. The Exhibit 9.8 The Low Back Pain Symptom Checklist What does your pain usually feel like? Directions: The words below describe different qualities of pain. Place an X in the boxes beside the words that best describe how your pain typically feels. You may check as many boxes as you wish that describe your typical pain this last week. squeezing splitting continuous aching torturing transient gruelling pricking pulling periodic troublesome tender nagging throbbing intermittent quivering numb suffocating radiating nauseating taut heavy drilling frightful boring jumping crushing miserable dreadful pinching cutting drawing flashing cruel rasping killing penetrating blinding fearful annoying spreading beating exhausting tearing cramping wrenching rhythmic lacerating pounding shooting wretched momentary hurting intense dull hot pins and needles pulsing punishing superficial stinging burning deep brief sharp localized cold tiring unlocalized flickering scalding spasms unbearable gnawing diffuse tugging stabbing surface agonizing tingling stiff piercing freezing skin pain smarting tight muscle pain steady itchy bone pain constant pressing joint pain lancinating sore moving pain terrifying sickening electrical vicious searing shock-like cool Reproduced from the Low Back Pain Symptom Checklist, obtained from Dr. Frank Leavitt. With permission. 503 504 Measuring Health Exhibit 9.9 The Back Pain Classification Scale, those for whom there was none (10). All were Showing Discriminant Function Coefficients treated medically and were observed during Note: Words with positive coefficients reflect pain of psycho- follow-up for 14 weeks after discharge. The logical origin, those with negative coefficients indicate organic BPCS scores at admission were used to classify pain. the group without organic signs into two sub- Pain variables Unstandardized groups: those with psychological symptoms and coefficients those without. Over a 14-week follow-up period, the progress of the two latter groups differed, Squeezing 0.67 with the group exhibiting psychological symp- Nagging −0.67 toms showing less improvement (10). Patients Exhausting 0.50 with no organic basis for their pain reported as Dull −0.49 much pain on retesting 14 weeks after treatment Sickening 0.69 as the organic group did before treatment (2). Troublesome 0.47 In an experimental study, Leavitt compared Throbbing −0.33 pain patients with healthy controls who were asked to reply to the Symptom Checklist as Tender 0.66 though trying to convince their doctor that they Intermittent −0.51 had a serious back problem (11). Using up to 17 Numb 0.66 words from the Symptom Checklist achieved an Shooting −0.30 66% correct identification of those simulating Punishing −1.64 pain and a 93% correct classification of those Tiring 0.49 with pain. Using up to 54 words achieved 87% Reproduced from Leavitt F, Garron DC. The detection of and 93% correct classifications, respectively (11, psychological disturbance in patients with low back pain. J Table 4). In a replication study, the percentages Psychosom Res 1979;23;152, Table 2. With permission. classified correctly shrank by about 8% (Table 5). These studies indicate that the pain words can be used to distinguish people in pain from BPCS achieved a 78% correct classification of those who are simulating pain. these two types of pain patient, compared with Concurrent validity was evaluated by com- 64.5% for the MMPI scales measuring hypo- paring the BPCS to the Pain Drawing Test, chondriasis and hysteria, and only 37.4% for the which includes a line drawing of a body on MMPI Low Back Scale (8, pp302–303). which respondents indicate the position and na- In a further examination of the discriminal ture of their pain. The two scales agreed on clas- ability of the BPCS, a sample of 120 patients sifying 76% of respondents into psychologically with low back pain was divided into two groups disturbed or nondisturbed (12, Table 2). The on the basis of their BPCS scores: 79 patients BPCS results also predicted subsequent return to whose pain apparently had an organic basis and work: 72% of those classified as nondisturbed 41 whose pain was classified as nonorganic (9). resumed work, compared with 46% of those All patients were examined by a clinical psychol- with psychological disturbance (12, p956). ogist who also administered the MMPI. The Demographic variables including age, gender, group identified by the BPCS as exhibiting psy- education, religion, and race do not predict chogenic pain gave higher scores on all ten clini- scores on the BPCS, so it can be used in compar- cal scales of the MMPI, eight of these differences ing patients from different sociodemographic being statistically significant (9, Table 2). The backgrounds (2). In a multiple regression study, clearest differences were on the MMPI hypo- Garron and Leavitt showed that the MMPI chondriasis and hysteria scales. hypochondriasis scale explained 15.5% of the In a prospective study, 108 patients were di- variance in the BPCS (13, p62). Adding a battery vided into those for whom an organic basis for of other tests and MMPI scales to the regression their pain could be identified clinically versus analysis raised the variance explained to 34%.  Pain Measurements 505 Alternative Forms These comments echo the reviews of psychologi- The reliability of a Spanish version, the Hispanic cal screening scales such as the Health Opinion Low Back Pain Symptom Checklist, has been Survey or Langner’s 22-item scale covered in tested (14). Chapter 5: they represent indicators of nonspe- cific distress and do not guide the user to any Commentary specific clinical interpretation of the nature of The BPCS performs a diagnostic and screening the psychological problem involved. task similar to that of Zung’s Pain and Distress It is apparent that the 13 words in the BPCS Scale and Pilowsky’s Illness Behavior Question- are not sufficient for some discrimination tasks, naire. In early studies, the scale seemed to suc- and more of those words are required for some ceed in positively identifying patients whose pain applications (7; 11). Of greater concern is the was due to psychological distress. It can also finding of the very modest overlap between the show that not all patients without an apparent original 13 words and the discriminating words organic basis for their pain do, in fact, suffer identified in subsequent studies. Leavitt has from psychological disorders (3). The discrimi- shown good post hoc discrimination in several nal ability of the BPCS is high and appears to ex- studies, but because the identity of the most dis- ceed that of the MMPI. Leavitt and Garron criminative words is not constant, it will be diffi- concluded that the BPCS provides a viable clini- cult to derive a consistently efficient screening cal alternative to the more cumbersome MMPI instrument to distinguish organic from func- for distinguishing between pain of organic and tional reports of pain. The best compromise is psychosomatic origins. probably to use the 13 words from the BPCS, al- Nonetheless, there are hazards in using dis- though users will always retain the uneasy feel- criminant analysis to identify subsets of ques- ing that other words might have worked as well tions. The results are often unstable, and testing or better. on further samples commonly identifies new sets of discriminating items. This was found by Storandt in the dementia field; she, like Leavitt, References achieved remarkably good classification in two separate studies, but using different sets of items (1) Leavitt F, Garron DC. The detection of each time (15). Furthermore, the items chosen in psychological disturbance in patients with this empirical manner often do not make con- low back pain. J Psychosom Res 1979;23:149–154. ceptual sense; Leavitt was candid about this: (2) Leavitt F. Detecting psychological disturbance using verbal pain measure- Why this particular set of verbal pain descrip- ment: the Back Pain Classification Scale. In: tors works as discriminators and others do Melzack R, ed. Pain measurement and not is unclear from research to date. The assessment. New York: Raven Press, shared variance of pain words with MMPI 1983:79–84. items is only 21%, and does not seem to fit (3) Leavitt F, Garron DC. Psychological any particular pattern in terms of the sensory disturbance and pain report differences in and affective divisions of pain experience. both organic and non-organic low back Much research is still needed to understand pain patients. Pain 1979;7:187–195. the apparently heterogeneous content of the (4) Leavitt F, Garron DC, Whisler WW, et al. Affective and sensory dimensions of back scale as it reflects some pain experience pain. Pain 1978;4:273–281. and/or personal characteristics that are as of (5) Biedermann HJ, Monga TN, Shanks GL, et yet not apparent. . . . Although the BPCS in- al. The classification of back pain patients: dicates with a high degree of probability that functional versus organic. J Psychosom Res a psychological disturbance exists, it does not 1986;30:273–276. identify the specific nature of the emotional (6) Sanders SH. Cross-validation of the Back problems. (2, pp83–84) Pain Classification Scale with chronic, 506 Measuring Health intractable pain patients. Pain Conceptual Basis 1985;22:271–277. No information is available. (7) Leavitt F. Use of verbal pain measurement in the detection of hidden psychological Description morbidity among low back pain patients with objective organic findings. Psychol The PAD describes physical and emotional reac- Health 1987;1:315–326. tions that may be attributed to pain. These in- (8) Leavitt F. Comparison of three measures clude limitations in activities of daily living and for detecting psychological disturbance in psychological responses such as agitation, de- patients with low back pain. Pain pression, and decreased alertness. The PAD con- 1982;13:299–305. tains 20 items that were selected on clinical (9) Leavitt F, Garron DC. Validity of a Back grounds to reflect problems (particularly psycho- Pain Classification Scale for detecting logical) that commonly accompany pain. Zung psychological disturbance as measured by grouped the items on a conceptual basis: only the MMPI. J Clin Psychol one item mentions pain (item 18); six reflect gen- 1980;36:186–189. eral mood changes (items 1, 2, 15–17, and 19); (10) Leavitt F, Garron DC. Validity of a Back Pain Classification Scale among patients the remaining 13 items cover behavioral changes with low back pain not associated with (Exhibit 9.10). demonstrable organic disease. J Psychosom The PAD scale is self-administered, and the Res 1979;23:301–306. questions use four-point frequency response (11) Leavitt F. Pain and deception: use of verbal scales; higher scores denote more frequent symp- pain measurement as a diagnostic aid in toms. The time reference is not fixed; any appro- differentiating between clinical and priate period, such as the past week, may be simulated low-back pain. J Psychosom Res used. The scores for each question are summed 1985;29:495–505. and are expressed as a percentage of the maxi- (12) McNeill TW, Sinkora G, Leavitt F. mum attainable score. Psychologic classification of low-back pain patients: a prognostic tool. Spine 1986;11:955–959. Reliability (13) Garron DC, Leavitt F. Psychological and Zung reported an alpha internal consistency of social correlates of the Back Pain 0.89 based on data from 122 pain patients and Classification Scale. J Pers Assess 195 controls (1, p892). 1983;47:60–65. (14) Leavitt F, Gilbert NS, Mooney V. Development of the Hispanic Low Back Validity Pain Symptom Check List. Spine A comparison of pain patients and controls 1994;19:1048–1052. showed that each item discriminated signifi- (15) Storandt M, Hill RD. Very mild senile cantly (p< 0.01) between them (1). A discrimi- dementia of the Alzheimer type: II. nant function analysis identified 11 items that Psychometric test performance. Arch discriminated with a sensitivity of 84.4% and a Neurol 1989;46:383–386. specificity of 99.5% (1, p893). These 11 items included both mood and behavior changes. A factor analysis identified six factors that cut The Pain and Distress Scale across the conceptual assignment of items, in (William W.K. Zung, 1983) that most of the factors included both behavioral Purpose and mood-change items. The self-rating Pain and Distress Scale (PAD) of Zung was intended as a brief measurement of Commentary mood and behavior changes that may be associ- This pain rating scale differs from others dis- ated with acute pain. It does not directly assess cussed in this chapter in that it concentrates on the severity of the pain itself. the distress associated with pain rather than on  Pain Measurements 507 Exhibit 9.10 The Pain and Distress Scale None or a little of the Some of the Good part Most or all time time of the time of the time 1. I feel miserable, low and down 2. I feel nervous, tense, and keyed up 3. I get tired for no reason 4. I can work for as long as I usually do 5. I am as efficient in my work as usual 6. I have trouble falling asleep 7. I have trouble sleeping through the night 8. I wake up earlier than I want to 9. I feel rested when I get out of bed 10. I am restless and can’t keep still 11. I find it hard to do the things I usually do 12. I find it hard to think and remember things 13. My mind is foggy and I can’t concentrate 14. I am as alert as I could be 15. I still enjoy the things I used to 16. I enjoy listening to the radio or watching TV 17. I enjoy visiting friends and relatives 18. I have aches and pains that bother me 19. I am more irritable than usual 20. Everything I do is an effort Adapted from Zung WWK. A self-rating Pain and Distress Scale. Psychosomatics 1983;24:889, Table 2. Copyright W. Zung, 1982. All rights reserved. With permission. the pain itself. Most of these items reflect general Pain Questionnaire, and also with depression anxiety and depressive responses and are not scores. The validity testing used patients with phrased to refer specifically to pain as the source acute, traumatic pain; the method has not been of the symptoms. Zung’s method may be com- tested on patients in chronic pain. Zung’s method pared with that of Pilowsky’s Illness Behavior deliberately identifies separate aspects of pain, so Questionnaire, which was designed to diagnose the maneuver of combining these into a single the psychological causes for an exaggerated pain overall score would appear to contradict his response, and Leavitt’s Back Pain Classification original purpose. It may be valuable to develop a Scale, which distinguishes pain of organic origin more adequate scoring system that provides sep- from that due to psychological causes. Zung’s arate scores for different dimensions of the pain method, by contrast, covers psychological prob- experience. lems associated with pain, commonly as re- sponses to pain rather than as causes. Very little information is available on the reli- ability and validity of the PAD scale. It is impor- Reference tant to show how well the PAD scores correlate (1) Zung WWK. A self-rating Pain and with those obtained from established pain mea- Distress Scale. Psychosomatics surement methods such as Melzack’s McGill 1983;24:887–894. 508 Measuring Health of abnormal illness behavior. In each case, there The Illness Behavior Questionnaire is a discrepancy between the objective somatic (I. Pilowsky and N.D. Spence, 1975, pathology and the patient’s response to it. Al- Revised 1983) though this may occur in many conditions, it is well illustrated by reactions to pain: Purpose The Illness Behavior Questionnaire (IBQ) as- sesses maladaptive responses to illness, including patients with intractable pain may be de- hypochondriacal responses, denial, and changes scribed as displaying “abnormal” or “mal- in affect. It was designed to indicate the extent to adaptive” illness behaviour in so far as their which these states could explain apparently ex- behaviour deviates from that regarded as ap- aggerated responses to illness. Although it is ap- propriate to the degree of somatic pathology plicable to any illness, it has been most widely observed, and is not modified by suitable ex- used in studies of pain. planation and reassurance provided by a doc- tor. (5, p62) Conceptual Basis People are normally aware of their body’s state Pilowsky argued that experiencing an abnormal of health and respond to ill health in characteris- pain response may serve as a form of psycholog- tic ways. In some cases, however, a person’s psy- ical atonement in response to guilt, or it may chological response to illness is exaggerated and permit denial or avoidance of conflicts. Pain may this influences the way he or she reports symp- also arise in conjunction with extended muscu- toms (1). The IBQ is designed to identify psycho- lar activity, perhaps brought on for psychologi- logical syndromes that may account for a cal reasons (5). The questions in the IBQ are discrepancy between the objective level of concerned with the attitudinal and emotional pathology and the patient’s response to it, components of illness behavior rather than with termed abnormal illness behavior. Typical re- overt behavior. sponses include hypochondriasis, neurotic re- Pilowsky has contributed to the further devel- sponses, irritability, dysphoria, and other opment of the concept of abnormal illness be- findings. The IBQ was developed from a havior, adopting the term dysnosognosia (7). A hypochondriasis questionnaire developed for this classification scheme subdivides illness behaviors purpose, the Whiteley Index of Hypochondria- into somatically focused and psychologically fo- sis, which is incorporated within the IBQ (2–4). cused, each divided into illness affirming and ill- The IBQ is a more general assessment tool ness denying (7, Figure 1). that covers other syndromes in addition to hypochondriasis and is based on Mechanic’s Description concept of illness behavior. The term illness be- The IBQ is a self-administered questionnaire havior refers both to overt actions, such as con- that uses a yes/no response format. The original sulting a physician, and to a patient’s emotional version contained 52 questions, later expanded and psychological reactions to illness (4). These to the 62-item version shown in Exhibit 9.11. A cover “the ways in which symptoms may be dif- 30-item abbreviation has also been used (8). Ten ferentially perceived, evaluated and acted (or not of the questions were taken from the Whiteley acted) upon by different kinds of persons” (5, Index of Hypochondriasis (3; 4). The IBQ is in- p62). Illness behavior refers to “the patient’s troduced to the patient as a survey containing “a psychological transactions with his/her physical number of questions about your illness and how symptoms” (6, p222) and can take normal or it affects you” (8). abnormal forms. Pilowsky proposed that a num- The 62 questions are grouped into seven di- ber of common psychiatric syndromes such as mensions identified empirically via factor analy- hypochondriasis, conversion reaction, neuras- sis. Pilowsky describes these dimensions as thenia, and malingering may be viewed as forms follows: Exhibit 9.11 The Illness Behavior Questionnaire Note: The asterisks indicate the response that is scored; the last column shows the factor placement of 30 items that loaded on one of seven factors. We have added these to the questionnaire presented by Pilowsky and Spence. Here are some questions about you and your illness. Circle either YES or NO to indicate your answer to each question. Dimension 1. Do you worry a lot about your health? YES NO __ 2. Do you think there is something seriously wrong with your body? YES* NO 2 3. Does your illness interfere with your life a great deal? YES* NO 2 4. Are you easy to get on with when you are ill? YES NO* 7 5. Does your family have a history of illness? YES NO __ 6. Do you think you are more liable to illness than other people? YES NO __ 7. If the doctor told you that he could find nothing wrong with you, YES NO* 2 would you believe him? 8. Is it easy for you to forget about yourself and think about all sorts of other things? YES NO __ 9. If you feel ill and someone tells you that you are looking better, do you become annoyed? YES* NO 1 10. Do you find that you are often aware of various things happening in your body? YES* NO 2 11. Do you ever think of your illness as a punishment for something you have done wrong in the past? YES* NO 3 12. Do you have trouble with your nerves? YES* NO 5 13. If you feel ill or worried, can you be easily cheered up by the doctor? YES NO __ 14. Do you think that other people realise what it’s like to be sick? YES NO __ 15. Does it upset you to talk to the doctor about your illness? YES NO __ 16. Are you bothered by many pains and aches? YES NO* 3 17. Does your illness affect the way you get on with your family or friends a great deal? YES* NO 7 18. Do you find that you get anxious easily? YES* NO 5 19. Do you know anybody who has had the same illness as you? YES NO __ 20. Are you more sensitive to pain than other people? YES* NO 1 21. Are you afraid of illness? YES* NO 1 22. Can you express your personal feelings easily to other people? YES NO* 4 23. Do people feel sorry for you when you are ill? YES NO __ 24. Do you think that you worry about your health more than most people? YES* NO 1 25. Do you find that your illness affects your sexual relations? YES NO __ 26. Do you experience a lot of pain with your illness? YES NO __ 27. Except for your illness, do you have any problems in your life? YES NO* 6 28. Do you care whether or not people realise you are sick? YES NO __ 29. Do you find that you get jealous of other people’s good health? YES* NO 1 30. Do you ever have silly thoughts about your health which you can’t get out YES* NO 1 of your mind, no matter how hard you try? 31. Do you have any financial problems? YES NO* 6 32. Are you upset by the way people take your illness? YES* NO 1 33. Is it hard for you to believe the doctor when he tells you there is nothing for you to worry about? YES NO __ (continued) 509 510 Measuring Health Exhibit 9.11 (continued) 34. Do you often worry about the possibility that you have got a serious illness? YES NO __ 35. Are you sleeping well? YES NO* 2 36. When you are angry, do you tend to bottle up your feelings? YES* NO 4 37. Do you often think that you might suddenly fall ill? YES* NO 1 38. If a disease is brought to your attention (through the radio, television, newspapers or someone you know) do you worry about getting it yourself? YES* NO 1 39. Do you get the feeling that people are not taking your illness seriously enough? YES NO __ 40. Are you upset by the appearance of your face or body? YES NO __ 41. Do you find that you are bothered by many different symptoms? YES* NO 2 42. Do you frequently try to explain to others how you are feeling? YES NO __ 43. Do you have any family problems? YES NO* 6 44. Do you think there is something the matter with your mind? YES* NO 3 45. Are you eating well? YES NO __ 46. Is your bad health the biggest difficulty of your life? YES NO* 3 47. Do you find that you get sad easily? YES* NO 5 48. Do you worry or fuss over small details that seem unimportant to others? YES NO __ 49. Are you always a co-operative patient? YES NO __ 50. Do you often have the symptoms of a very serious disease? YES NO __ 51. Do you find that you get angry easily? YES* NO 7 52. Do you have any work problems? YES NO __ 53. Do you prefer to keep your feelings to yourself? YES* NO 4 54. Do you often find that you get depressed? YES* NO 5 55. Would all your worries be over if you were physically healthy? YES* NO 6 56. Are you more irritable towards other people? YES* NO 7 57. Do you think that your symptoms may be caused by worry? YES* NO 3 58. Is it easy for you to let people know when you are cross with them? YES NO* 4 59. Is it hard for you to relax? YES* NO 5 60. Do you have personal worries which are not caused by physical illness? YES NO* 6 61. Do you often find that you lose patience with other people? YES* NO 7 62. Is it hard for you to show people your personal feelings? YES* NO 4 Derived from Pilowsky I, Spence ND. Manual for the Illness Behaviour Questionnaire (IBQ). 2nd ed. Adelaide, Australia: Univer- sity of Adelaide, 1983, Appendix A. With permission. 1. General Hypochondriasis. A general factor mation that physical disease exists, symptom marked by phobic concern about one’s state preoccupation, and rejection of the doctor’s of health. Associated with a high level of reassurance; arousal or anxiety and with some insight into 3. Psychological vs Somatic Perception of Ill- inappropriateness of attitudes. A high score ness. A high score indicates that the patient also suggests an element of interpersonal feels somehow responsible for (and in fact de- alienation, but one that is secondary to the serves) his illness, and perceives himself to be patient’s phobic concern; in need of psychiatric rather than medical 2. Disease Conviction. Characterised by affir- treatment. A low score indicates a rejection of  Pain Measurements 511 such attitudes and a tendency to somatise pared with their spouses’ perceptions of what concerns; the patients’ responses would have been. A cor- 4. Affective Inhibition. A high score indicates relation of 0.59 was obtained, rising to 0.65 difficulty in expressing personal feelings, es- when corrected for attenuation (4, p90). A fac- pecially negative ones, to others; tor analysis of the ten items distinguished three 5. Affective Disturbance. Characterised by factors that reflected clinically relevant aspects of feelings of anxiety and/or sadness; hypochondriasis: bodily preoccupation, disease 6. Denial. A high score indicates a tendency phobia, and conviction of the presence of dis- to deny life stresses, and to attribute all prob- ease with refusal to be reassured (4). lems to the effects of illness; 7. Irritability. Assesses the presence of angry fifty-two item version. Much of Pilowsky and feelings, and interpersonal friction. (2, p3) Spence’s development work on the IBQ con- cerned factor analyses of the original 52 items A total score for the questionnaire may be using a sample of only 100 patients with chronic obtained by counting the responses that repre- pain of various types (2; 5; 11). Seven factors sent problems—these are indicated by asterisks were identified, which accounted for 63% of the in the exhibit. Alternatively, scores may be pro- variance (6). Forty items loaded on these factors, vided for the seven dimensions; the numbers in as shown in Exhibit 9.11. The analysis has been the last column in Exhibit 9.11 show the dimen- repeated on other groups of pain patients with sion on which the question is scored. High comparable results. Most applications of the scores “suggest maladaptive ways of perceiving, IBQ report scale scores based on these seven fac- evaluating, or acting in relation to one’s state of tors. health” (5). Modest associations were reported between the IBQ factors and behavioral responses to pain Reliability (e.g., guarding, bracing, rubbing, grimacing). Test-retest correlations for the seven scales of the Using multiple regression, the overall variance 62-item version were reported for 42 cases. Af- explained by IBQ scores and other variables ter a delay of one to 12 weeks, correlations ranged from 6 to 22% (12, Table III). ranged from 0.67 to 0.87, with only three coeffi- cients below 0.84 (2, Appendix E). The ten thirty item version. Speculand et al. used the hypochondriasis items previously tested showed 30-item IBQ to compare 24 patients suffering a test-retest correlation of 0.81 (N = 71) (2; 4). from intractable facial pain with 24 age- and Kappa coefficients for all 62 items were reported gender-matched dental patients. Two of the for a test-retest study; ten items yielded non- seven factor scores showed significant differ- significant kappas (9, Table 1). ences: disease conviction and psychological ver- Coefficient theta internal consistency values sus somatic perception of disease (8). Other for the seven scales ranged from 0.36 to 0.72 (9, studies have shown that some factor scores dis- Table 3). An overall alpha of 0.62 has been re- criminate between different types of pain pa- ported (10, p783). tients (6; 13–16). Speculand et al. used a Comparisons of patient scores and ratings discriminant analysis to compare responses of made by their spouses on the seven scales pro- dental patients in pain with patients with vided correlations ranging between 0.50 and chronic facial pain, and from their results a sen- 0.78 for 42 patients (2, p37). sitivity of 87.5% at a specificity of 62.5% can be calculated (8, Table 3). Validity Pilowsky reported evidence for the validity of the sixty-two item version. Pilowsky et al. applied ten Whiteley Index questions on hypochondria- the 62-item version of the IBQ to patients who sis (4). The scores of 118 patients were com- underwent coronary artery bypass surgery and 512 Measuring Health used a discriminant function analysis to contrast related significantly with the Spielberger state those who showed relief of angina with those anxiety (0.59) and trait anxiety (0.76) scales (2, who did not (3). Six factor scores provided an p10). The affective disturbance scale also corre- 82% correct classification. In a cross-cultural lated 0.28 with the subsequent number of visits study, Pilowsky et al. developed a discriminant to general practitioners. Other scales that pre- function equation from interviews with 100 pain dicted use of care were disease conviction patients and 78 general practice patients in Seat- (0.20), affective state (0.19), and disease affirma- tle, Washington, and applied the equation to tion (0.18) (21, Table III). equivalent groups of patients in Australia to evaluate its discriminal ability. The results Alternative Forms showed a sensitivity of 97% at a specificity of Pilowsky has described the Illness Behavior As- 74% (17, p206). The IBQ has been used in dis- sessment Schedule, which is a standardized clini- tinguishing between organic and functional cal interview tool covering the same fields as the symptoms; it showed a clear distinction between IBQ (22). psychiatric patients and patients with multiple An abbreviated form of the IBQ was designed nonspecific symptoms (18, Table II). In a com- for screening purposes. This takes six items from parison between outpatients seen in psycholo- the Disease Conviction scale and five from the gists’ offices and general population samples, the Psychological vs Somatic Perception of Illness disease conviction, affective disturbance, and ir- subscale (23). Taken together, these scales indi- ritability scales showed significant contrasts in cate disease affirmation; scores distinguished sig- response; the overall scores also differed signifi- nificantly between groups judged on clinical cantly (10, Table 1). grounds to differ in levels of abnormal illness re- The 62 items were analyzed using a numeri- sponse; sensitivity was 86% and specificity 83% cal taxonomy procedure that groups items into (23, p328). classes that are internally consistent and distinct Twenty-one of the IBQ items distinguished from other classes. Two classes comprised more between people told in an experiment to delib- than five items; the first described “patients who erately exaggerate their reports of pain and reject the possibility that their condition might pain patients whose pain was of neurotic ori- be linked to psychological problems,” and the gin. These items were formed into a Conscious second included those who “view their pain Exaggeration (CE) scale (24). This correlated problem within a psychological and emotional 0.64 with psychiatric ratings of patients’ exag- framework” (19, p93). The authors noted that geration (25, p297). The CE scale scores corre- patients whose responses place them in the first lated 0.77 with the hypochondriasis scale and category may find somatic and behavioral treat- 0.71 with the disease conviction scale (26, ment appropriate, whereas patients in the second Table 5). category may be accessible to therapies involving A critical review of the IBQ was provided by cognitive strategies (p94). Main and Waddell. Their item analyses led to a Evidence for convergent and divergent valid- reduction to 37 items that formed three scales; ity may be obtained from correlations with the these predicted disability, distress, and pain re- Center for Epidemiologic Studies depression ports more adequately than the original seven scale (CES-D) (20, Table VIII). Pearson correla- scales did (9). tions were high for affective disturbance (0.55), The IBQ has been widely translated, includ- disease conviction (0.50), and hypochondriasis ing Italian (20), Spanish (27), Japanese (28), (0.47); they were low for affective inhibition Hindi (29), Gujarati, and Urdu (30). (0.10) and denial (−0.19). Correlations between the affective disturbance scale (factor 5) and the Zung Self-rating Depression Scale and the Reference Standards Levine-Pilowsky Depression Questionnaire were The manual of the IBQ reports reference stan- 0.54 and 0.56, respectively. This scale also cor- dards from samples of patients from pain clinics,  Pain Measurements 513 general practice, and psychiatric and general manual provides good documentation and read- hospitals (2, Appendix F). ers interested in using the method should consult recent literature to see whether additional valid- Commentary ity data have been published. The IBQ has been widely used in Australia, North America, Europe, and India, mostly in Address studies analyzing adaptive and maladaptive re- Professor I. Pilowsky, Department of Psychiatry, sponses to illness and not only in pain patients. Royal Adelaide Hospital, Adelaide, South Aus- Abnormal illness behavior, for example, is asso- tralia 5001. ciated with higher use of health care and has been found a predictor of long-term functional disability (31). The widespread use of the References method suggests that it fills a gap in the range of (1) Pilowsky I. Abnormal illness behaviour. measurements currently available. One strength Sydney, New York: J. Wiley, 1997. of the IBQ is its foundation on an extensive con- (2) Pilowsky I, Spence ND. Manual for the ceptual analysis of pain responses (32). The IBQ Illness Behaviour Questionnaire (IBQ). 2nd has been used in many psychosomatic studies of ed. Adelaide, Australia: University of the links among anxiety, depression, and the Adelaide, 1983. presentation of illness (20; 24; 33). (3) Pilowsky I, Spence ND, Waddy JL. Illness Although the IBQ has been extensively used, behaviour and coronary artery by-pass surgery. J Psychosom Res 1979;23:39–44. few studies have reported on its validity. Those (4) Pilowsky I. Dimensions of studies using the IBQ are important and interest- hypochondriasis. Br J Psychiatry ing, but the resulting reports often cannot be 1967;113:89–93. used for our secondary purpose of drawing con- (5) Pilowsky I, Spence ND. Illness behaviour clusions on the performance of the instrument it- syndromes associated with intractable pain. self. Although the IBQ has been used with Pain 1976;2:61–71. depression and anxiety scales, for example, cor- (6) Demjen S, Bakal D. Illness behavior and relations are generally not reported (33; 34). chronic headache. Pain 1981;10:221–229. Other articles present multivariate analyses that (7) Pilowsky I. Abnormal illness behaviour incorporate the IBQ but do not allow us to iden- (dysnosognosia). Psychother Psychosom tify its unique association with the criterion vari- 1986;46:76–84. (8) Speculand B, Goss AN, Spence ND, et al. ables (12; 33). Some evidence for validity of the Intractable facial pain and illness IBQ is methodologically weak. The factor ana- behaviour. Pain 1981;11:213–219. lytic studies used unacceptably small samples, (9) Main CJ, Waddell G. Psychometric often analyzing 52 questions with as few as 100 construction and validity of the Pilowsky cases. Main and Waddell, in particular, were Illness Behaviour Questionnaire in British critical of the poor psychometric properties of patients with chronic low back pain. Pain the IBQ. 1987;28:13–25. The IBQ covers a field similar to that of (10) Boyle GJ, Le Dean L. Discriminant validity Zung’s Pain and Distress Scale. They differ in of the Illness Behavior Questionnaire and that the IBQ is intended principally for explain- Millon Clinical Multiaxial Inventory-III in ing chronic, intractable pain that is not readily a heterogeneous sample of psychiatric outpatients. J Clin Psychol explicable in terms of the level of tissue damage 2000;56:779–791. observed, whereas Zung’s method provides a (11) Pilowsky I, Spence ND. Is illness behaviour simple description of the psychological sequelae related to chronicity in patients with of pain. Pilowsky’s method is similar in intent to intractable pain? Pain 1976;2:167–173. Leavitt’s Back Pain Classification Scale but dif- (12) Keefe FJ, Crisson JE, Maltbie A, et al. fers in being applicable to a broader range of ail- Illness behavior as a predictor of pain and ments. The IBQ has considerable potential; the overt behavior patterns in chronic low back 514 Measuring Health pain patients. J Psychosom Res symptom exaggeration by questionnaire: a 1986;30:543–551. clinical study. J Psychosom Res (13) Chapman CR, Sola AE, Bonica JJ. Illness 1987;31:703–711. behavior and depression compared in pain (27) Llor EB, Nieto MJ, Godoy FC, et al. center and private practice patients. Pain [Adjustment scale for the I. B. Q. 1979;6:1–7. questionnaire for the Spanish clinical (14) Pilowsky I, Spence ND. Pain and illness population]. Actas Luso Esp Neurol behaviour: a comparative study. J Psiquiatr Cienc Afines 1991;19:263–267. Psychosom Res 1976;20:131–134. (28) Guo Y, Kuroki T, Yamashiro S, et al. (15) Pilowsky I, Chapman CR, Bonica JJ. Pain, Illness behaviour and patient satisfaction as depression and illness behaviour in a pain correlates of self-referral in Japan. Fam clinic population. Pain 1977;4:183–192. Pract 2002;19:326–332. (16) Pilowsky I, Spence ND. Pain, anger and (29) Chadda RK. Dhat syndrome: is it a distinct illness behaviour. J Psychosom Res clinical entity? A study of illness behaviour 1976;20:411–416. characteristics. Acta Psychiatr Scand (17) Pilowsky I, Murrell TGC, Gordon A. The 1995;91:136–139. development of a screening method for (30) Bhatt A, Tomenson B, Benjamin S. abnormal illness behaviour. J Psychosom Transcultural patterns of somatization in Res 1979;23:203–207. primary care: a preliminary report. J (18) Singer A, Thompson S, Kraiuhin C, et al. Psychosom Res 1989;33:671–680. An investigation of patients presenting with (31) Clark MS, Smith DS. Psychological multiple physical complaints using the correlates of outcome following Illness Behaviour Questionnaire. rehabilitation from stroke. Clin Rehabil Psychother Psychosom 1987;47:181–189. 1999;13:129–140. (19) Pilowsky I, Katsikitis M. A classification of (32) Pilowsky I. A general classification of illness behaviour in pain clinic patients. abnormal illness behaviours. Br J Med Pain 1994;57:91–94. Psychol 1978;51:131–137. (20) Fava GA, Pilowsky I, Pierfederici A, et al. (33) Joyce PR, Bushnell JA, Walshe JWB, et al. Depression and illness behavior in a Abnormal illness behaviour and anxiety in general hospital: a prevalence study. acute non-organic abdominal pain. Br J Psychother Psychosom 1982;38:141–153. Psychiatry 1986;149:57–62. (21) Pilowsky I, Smith QP, Katsikitis M. Illness (34) Pilowsky I, Bassett DL. Pain and behaviour and general practice utilisation: depression. Br J Psychiatry a prospective study. J Psychosom Res 1982;141:30–36. 1987;31:177–183. (22) Pilowsky I, Bassett DL, Barrett R, et al. The Illness Behavior Assessment Schedule: The Pain Perception Profile reliability and validity. Int J Psychiatry (Bernard Tursky, 1976) Med 1983;13:11–28. (23) Chaturvedi SK, Bhandari S, Beena MB, et Purpose al. Screening for abnormal illness The Pain Perception Profile (PPP) offers quanti- behaviour. Psychopathology tative estimates of the intensity, unpleasantness, 1996;29:325–330. and type of pain a person experiences. It was in- (24) Clayer JR, Bookless C, Ross MW. tended for clinical use by behavior therapists Neurosis and conscious symptom treating pain patients (1). exaggeration: its differentiation by the Illness Behaviour Questionnaire. J Psychosom Res 1984;28:237–241. Conceptual Basis (25) Clayer JR, Bookless-Pratz CL, Ross MW. Tursky’s PPP was developed at about the same The evaluation of illness behaviour and time as Melzack’s McGill Pain Questionnaire exaggeration of disability. Br J Psychiatry (MPQ) and, like the MPQ, it tackles the prob- 1986;148:296–299. lem of how best to devise numerical ratings for (26) Mendelson G. Measurement of conscious the words that people typically use to describe  Pain Measurements 515 their pain. Unlike the category scaling used in estimation methods to examine the respondent’s the MPQ, however, Tursky used magnitude esti- rating of the painfulness of a series of set electri- mation procedures (see page 20) to scale the cal stimuli and identifies the mathematical power characteristic pain response of each individual. function (see page 17), which describes the rela- Tursky argued that the MPQ is limited in its tionship between the intensity of the experimen- ability to provide quantitative pain information tal stimulus and this person’s judgment of pain. and that the use of categorical scales constrains These judgments and the power exponent can be the patient’s ability to adequately rate her pain compared to standards to evaluate the patient’s (1). To provide more adequate numerical esti- ability to make normal judgments of pain stimu- mates of pain, considerably more information lation. must be collected from the respondent. A greater exponent may be indicative of hy- Description persensitivity, a lesser exponent indicative of The patient is given a pain diary that includes hyposensitivity, and a significantly non-linear pain descriptors based on those used in the relationship may indicate a possible neuro- MPQ. But before the diary is completed, a set of logic malfunction or an attempt on the part optional experimental procedures may be ap- of the patient to manipulate his self-report. plied to provide precise estimates of the way in Changes in the exponent or the intercept may which each respondent uses the adjectives to de- reflect alterations in the patient’s pain respon- scribe pain. These procedures identify the re- sivity as a function of treatment intervention. spondent’s pain sensation and tolerance (1, p383) thresholds, her ability to judge reliably between differing levels of pain stimulation, and her char- In the third part of the assessment, the patient acteristic use of the pain descriptors. The prelim- uses a cross-modality matching procedure (see inary stages can be omitted; if used, the full pages 17–18) to rate the intensity of pain repre- profile comprises the following four measure- sented by the descriptors that are part of the pain ment stages. The first part of the PPP establishes diary. The adjectives cover three dimensions of the respondent’s pain sensation threshold. An pain: 14 describe pain intensity, 11 cover the experiment is performed in which gradually in- emotional reaction to pain (unpleasantness), and creasing levels of electrical stimulation are ap- 13 describe the sensation of pain (Exhibit 9.12). plied and the respondent is asked to indicate at The words were selected primarily from the what level she experiences sensation, discomfort, MPQ and were tested in preliminary scaling pain, and her limit of tolerance for them. The studies (1). Reference scale values are available stimulation is applied through an electrode on for the words (1, Table 2). the forearm; the equipment is portable and does The fourth and final phase in administering not have to be used in a laboratory setting. Its the PPP involves the use of the daily pain diary major purpose is to provide the clinician with a shown in Exhibit 9.12. Using the three cate- better understanding of the patient’s characteris- gories of pain descriptors scaled previously, the tic pain response and possible bias in rating patient records her pain at specified times of the pain; the level of electrical stimulation required day. The diary also records the source of the dis- to produce each response may be compared with comfort, its time of onset, its duration, and the reference standards to identify abnormal pain re- medication taken for each reported pain period. sponses. The difference between pain threshold Instructions for using the diary are given by and pain tolerance indicates the patient’s pain Tursky et al. (1, p390). sensitivity range, a predictor of her ability to en- dure pain (1). This may be useful in prescribing Reliability treatment and may be used to study the effect of Some data on the reliability of the four parts of treatment, including psychological intervention. the PPP were provided by Tursky et al. The test- The second part of the PPP uses magnitude retest reliability for 20 pain patients in making Exhibit 9.12 The Pain Perception Profile: Lists of Pain Descriptors and Sample Page from a Pain Diary INTENSITY UNPLEASANTNESS FEELING Moderate Distressing Stinging Just Noticeable Tolerable Grinding Mild Awful Squeezing Excruciating Unpleasant Burning Very Strong Unbearable Shooting Very Intense Uncomfortable Numbing Severe Intolerable Throbbing Intense Bearable Stabbing Very Weak Agonizing Itching Strong Miserable Aching Weak Distracting Cramping Not Noticeable Not Unpleasant None Pressure Reproduced from Tursky B, Jamner LD, Friedman R. The Pain Perception Profile: a psychophysical approach to the assessment of pain report. Behav Ther 1982;13:389. Copyright 1982 by the Association for Advancement of Behavior Therapy. Reprinted by permission of the publisher. 516  Pain Measurements 517 judgments of pain thresholds and discomfort approach to the assessment of pain report. levels were reported as showing close agreement, Behav Ther 1982;13:376–394. although Tursky did not summarize the data sta- (2) Jamner LD, Tursky B. Discrimination tistically (1). between intensity and affective pain descriptors: a psychophysiological evaluation. Pain 1987;30:271–283. Validity Jamner and Tursky examined the validity of the classification into intensity and affective pain de- scriptors using a classical conditioning experi- Conclusion ment. A skin conductance response was conditioned to the concept of pain intensity for The development of pain measurements has in one experimental group and to pain unpleasant- many ways been successful and illustrates several ness for another group (2). Following condition- themes in this book. There have been clear links ing, there was a highly significant (p<0.001) between conceptual work on the definition of difference between the two groups in their re- pain and the development of measurement sponse to affective and intensity adjectives, mea- scales, so that the results obtained using these sured with skin conductance (2, p280). “These methods have led to refinements in the concep- results provide psychophysiological evidence for tual definitions of pain. The leading measure- the distinctness of the concept of pain intensity ment methods have been widely used and in from its associated affective features” (2, p281). consistent ways that permit direct comparisons among results obtained in different studies. Commentary Close attention has been paid to reliability and The PPP differs from most other measurements validity in the development of the measurement discussed in this book in that it lacks published methods and these have often used advanced sta- validity and reliability data, and yet we have in- tistical and analytic techniques. Furthermore, cluded it as an example of a sophisticated rating the link between clinical interests in pain man- technique that could illustrate the types of agement and the measurement techniques has methodological development to be anticipated in been closer in this field than in, say, functional future measurement instruments. Our descrip- disability. Pressure from clinicians eager to eval- tion of the method does not provide sufficient in- uate their interventions led to the development of formation for the reader to apply the PPP; fuller methods that distinguish between the objective details on its administration are contained in experience of pain and the subjective response to Tursky’s report. it, and pain is one of the few areas of measure- Although this method is the most detailed and ment in which this has been attempted. There mathematically complex of the pain measurement was even an attempt to coin a term, analgesime- methods reviewed, Tursky argued that it is suffi- try, to refer to pain measurement (1, p237). Fi- ciently simple to be used in a clinician’s office by nally, the reader gains the impression that the personnel with basic training in its administra- many researchers working on the problem of tion. Nevertheless, the method has not achieved pain measurement, who come from widely dif- widespread use. It is also necessary that validity fering disciplines, benefit from the existence of a data be collected to indicate how far the results of dedicated journal (Pain) that has published many the magnitude estimation procedures produce re- of the leading articles on pain measurement. sults that differ from the simpler estimates ob- New and often innovative pain measures con- tained from the MPQ, which it incorporates. tinue to be developed. The 41-item Biobehav- ioral Pain Profile records cognitive, behavioral, and physiological reactions to pain (2). The 7- References item Pain Disability Index describes pain-related (1) Tursky B, Jamner LD, Friedman R. The disability in areas such as family life, occupa- Pain Perception Profile: a psychophysical tion, recreation, self-care, and social activity (3). 518 Measuring Health In the field of back pain, in which numerous chronic pain believe they cannot function nor- measurements have been proposed, the Low mally because of their pain; the scale uses items Back Pain Rating scale has built on the design of such as “Most people expect too much of me, previous scales and has undergone preliminary given my chronic pain” to assess the strength of validity testing (4). Ho et al. reviewed clinical this belief (14). Roland and Morris developed observations of behavioral and physiologic re- the Roland Disability Questionnaire by taking sponses to pain as ways of measuring pain (5). 24 items that were relevant to low back pain The first edition of this book also described the from the Sickness Impact Profile (15). This scale SAD Index for the Clinical Assessment of Pain. was subsequently modified by Patrick et al. (16). This provides a numerical score summarizing In the field of measuring emotional reactions the Somatic aspects of clinical pain and accom- to pain, one scale deserves mention if only for its panying levels of Anxiety and Depression; each macabre content. As a measure of pain reactiv- of these dimensions of pain is rated on a zero-to- ity, Elton et al. showed subjects a color film por- ten scale of intensity, the axes forming a three- traying increasing levels of insult to a human dimensional system of coordinates. Black and hand; the reactions were recorded. The afflic- Chapman noted that each aspect of pain formed tions included pinching, pricking, hammering, a distinct clinical entity with an accepted, effec- burning, cutting, and, as a finale, severing the tive therapy (6). For research purposes, they pro- hand (17). Despite reassurances that said hand posed that an overall pain score could be belonged to a cadaver and even without actually represented by the vector sum of the dimensions seeing the color film, our reactivity was plainly represented on the three axes, seemingly an ele- visible. gant approach to the problem of summarizing There remain, of course, gaps in the reper- pain on several dimensions. Black and Chapman toire of pain measures. It will be desirable to see did not pursue the idea further, although it re- greater application of high-quality pain measure- mains an interesting notion. ment methods in clinical studies than is now the The names given to pain scales are sometimes case: frequently clinical studies rely on four- quaint. There is the West Haven-Yale Multidi- point verbal pain scales. More cross-validation is mensional Pain Inventory (WHYMPI), which desirable among the pain measures, especially offers a broad-ranging but brief measure of exploring the equivalence of verbal, behavioral, chronic pain. It contains three sections, the first and analogue methods. The few studies that covering five aspects of the pain, the second de- have compared the various approaches do not scribing the reactions of significant others to the suggest close equivalence; this suggests an area patient’s pain, and the third covering the level of for further investigation that would be of inter- pain-induced disability (7). There is also the est to health measurement theory in general. Oucher Index, a scale for children based on faces depicting increasingly severe pain (8; 9), References whereas the CRIES is used for neonatal pain (10). A useful review of pain measurement in (1) Sriwatanakul K, Kelvie W, Lasagna L, et children was given by McGrath (11). A faces al. Studies with different types of visual pain scale designed for elderly people has also analog scales for measurement of pain. been tested (12). Clin Pharmacol Ther 1983;34:234–239. (2) Dalton JA, Feuerstein M, Carlson J, et al. As with the approach of the Oswestry scale, Biobehavioral pain profile: development the theme of pain-related disability was also cov- and psychometric properties. Pain ered by the Pain Disability Index developed by 1994;57:95–107. Tait et al. This uses zero-to-ten ratings of disabil- (3) Jerome A, Gross RT. Pain Disability Index: ity in seven areas of activity (13). A related construct and discriminant validity. Arch theme is covered by the Pain and Impairment Re- Phys Med Rehabil 1991;72:920–922. lationship Scale (PAIRS). Many patients in (4) Manniche C, Asmussen K, Lauritsen B, et  Pain Measurements 519 al. Low Back Pan Rating scale: validation pain: a review of behavioral, physiological of a tool for assessment of low back pain. and direct scaling techniques. Pain Pain 1994;57:317–326. 1987;31:147–176. (5) Ho K, Spence J, Murphy MF. Review of (12) Herr KA, Mobily PR, Kohout FJ, et al. pain-measurement tools. Ann Emerg Med Evaluation of the faces pain scale for use 1996;27:427–432. with the elderly. Clin J Pain (6) Black RG, Chapman CR. SAD Index for 1998;14:29–38. clinical assessment of pain. In: Bonica JJ, (13) Tait RC, Pollard CA, Margolis RB, et al. Albe-Fessard DG, eds. Advances in pain The Pain Disability Index: psychometric research and therapy. Vol. 1. New York: and validity data. Arch Phys Med Rehabil Raven Press, 1976:301–305. 1987;68:438–441. (7) Kerns RD, Turk DC, Rudy TE. The West (14) Riley JF, Ahern DK, Follick MJ. Chronic Haven-Yale Multidimensional Pain pain and functional impairment: assessing Inventory (WHYMPI). Pain beliefs about their relationship. Arch Phys 1985;23:345–356. Med Rehabil 1988;69:579–582. (8) Beyer JE, Knapp TR. Methodological (15) Roland M, Morris R. A study of the issues in the measurement of children’s natural history of back pain. Part I: pain. Child Health Care 1986;14:233–241. Development of a reliable and sensitive (9) Beyer JE, Aradine CR. Convergent and measure of disability in low-back pain. discriminant validity of a self-report mea- Spine 1983;8:141–144. sure of pain intensity for children. Child (16) Patrick DL, Deyo RA, Atlas SJ, et al. Health Care 1988;16:274–282. Assessing health-related quality of life in (10) Krechel SW, Bildner J. CRIES: a new patients with sciatica. Spine neonatal postoperative pain measurement 1995;20:1899–1909. score. Pediatric Anaesthesia (17) Elton D, Stanley GV, Burrows GD. A new 1995;5:53–61. test of pain reactivity. Percept Mot Skills (11) McGrath PA. An assessment of children’s 1978;47:125–126.  10 General Health Status and Quality of Life hapter 3 reviewed measurements of physical status is level of conceptualization. Quality of C functioning; subsequent chapters covered the social and emotional aspects of health. A grow- life as it is used in clinical research is a vague term without conceptual clarity. It is what in- ing number of health measurements combine vestigators mean it to be. . . . Conceptual these themes in one instrument; these form the frameworks for health status, on the other topic of the present chapter. The measures we re- hand, have appeared in the literature, have view cover at least the physical, emotional, and been discussed and debated, and have pro- social dimensions of health; many cover much vided the underpinnings of several measures. more. (3, ppS149–S150) These are variously termed “general health status measures” or “measures of health-related Things have since improved, however, and con- quality of life.” Although the trend has been to- sensus appears to be growing over the scope re- ward the latter term, there is no clear distinction quired for an instrument to be called a “QoL” between quality of life (QoL) measures and or a “health-related QoL” measure. There are methods, such as the Sickness Impact Profile, advantages and disadvantages in considering which were described by their authors as general QoL as an outcome in clinical medicine and health measures. If there is a distinction, it health care, and a brief digression into this topic would be that quality of life is broader, including will introduce some of the measurement issues. the dimensions covered by general health mea- sures, but extending to other topics. In addition to physical, mental, and social well-being, the Measuring Quality of Life EuroQol, for example, covers “usual activities;” the SF-36 includes work and role performance; Medical interest in quality of life was stimulated and the WHOQOL covers spiritual well-being, by success in prolonging life and by the realiza- transportation, and environmental factors under tion that this may be a mixed blessing: patients the rubric of health-related QoL (1). Although want to live, not merely survive. The question- the QoL concept is broad, we still lack a clear able quality of survival of people who in the past consensus over which dimensions should be in- could not have been saved has fueled debate over cluded. The previous editions of this book com- topics such as artificial life support, euthanasia, plained that during the 1980s the concept had and the definition of death itself (4; 5). The been misused in the medical literature, often ob- theme, of course, is not new; Jonathan Swift scuring progress that had painfully been made noted that every man desires to live long, but no previously, a thought that was echoed by Prutkin man wishes to be old. Isaac Stern expressed a and Feinstein (2). For example, in 1989, Marilyn similar sentiment when he advised “Everyone Bergner commented: should die young. But they should delay it as long as possible.” What is new is the develop- One of the striking differences between the ment of formal ways to measure QoL and their notion of quality of life and that of health routine application in outcome evaluation. As 520  General Health Status and Quality of Life 521 with measuring happiness, we became intrepid current ability to function, as compared with to the point that the idea of measuring so ab- their own internalised standards of what is pos- stract and complex a theme as quality of life no sible or ideal” (8). A thoughtful review of the longer seemed presumptuous. components to include in measures of QoL, and Although comparatively new in health re- of the balance between health issues and non search, social scientists have long discussed health issues in such assessments was given by quality of life. They distinguish it from the con- Spilker and Revicki. The relative balance will cepts of life satisfaction, morale, happiness, and vary by health status: for healthy people, non anomie largely in terms of level of subjectivity health matters will chiefly influence perceived (6). In the social sciences, QoL commonly refers quality of life, whereas for people with a chronic to the adequacy of people’s material circum- illness, health factors may outweigh non health stances and to their feelings about these circum- issues (9). stances. Indicators include personal wealth and The new subjective indicators of quality of possessions, level of safety, level of freedom, and life bear a strong resemblance to familiar in- opportunity; health forms but one of many com- dexes of emotional well-being and life satisfac- ponents in this broad concept. Life satisfaction tion used with the aged and the chronically sick. generally refers to people’s subjective assessment The benefits of incorporating quality of life mea- of their circumstances, compared with an exter- sures in health research lie first in broadening nal reference standard or to their own aspira- the scope of outcome measures and second in tions. The term morale is more subjective still, providing a formal means for the patient’s judg- and refers to the sense of optimism, confidence, ment to influence treatment. Quality of life mea- sadness, or depression that may result from life surement is valuable in comparing treatments satisfaction. Happiness commonly refers to that are equivalent in other terms: lumpectomy short-term transient feelings of well-being in re- versus radiation for stage I and II breast cancers sponse to day-to-day events. These distinctions (10), for example. It may also be used in involv- are not rigid; definitions shift with changing so- ing the patient in choosing whether to undergo cial circumstances. Indeed, as wealth increases, therapy at all: antihypertensive therapy may be indicators of QoL have expanded from the mate- more troublesome than a disease without symp- rial terms of income or possessions to include toms (11). Finally, it helps to weigh advantages also more spiritual rewards such as satisfaction, in terms of survival against adverse effects of personal development, and participation in the treatment, as with cancer chemotherapy (10). community. Recognizing that wealth does not Hence, by 1985, the U.S. Food and Drug Ad- necessarily create happiness, QoL indicators fo- ministration included quality of life as one of its cused on people’s feelings about their circum- primary criteria for the approval of new anti- stances, whereas economic indicators covered cancer therapies (12). the objective side. This raises the political ques- The disadvantages of invoking quality of life tion of how to balance needs against people’s measures include apparently deliberate attempts subjective demands as driving forces in planning to direct attention away from the limited success social and health programs. Herein lies the cen- of some therapy when measured by more objec- tral purpose of QoL scales: they provide insight tive indicators. The early trials of bypass graft into the perceived discrepancy between actual surgery, for example, showed no mortality bene- and ideal states. Calman noted that quality of fit compared with medical treatment (13; 14). life “can be said to be present when the hopes of Nonetheless, patients appeared satisfied with the the individual are matched and fulfilled by expe- operation, so it was suggested that the operation rience. The opposite is also true: a poor QoL oc- be evaluated not in terms of morbidity or mor- curs when the hopes do not meet with the tality but in terms of the “satisfaction it provides experience” (7). Similarly, Cella and Tulsky ar- the individual” or the patient’s “achievement of gued that quality of life represents “the impor- a satisfactory social situation within the limits of tance of people’s subjective perceptions of their perceived physical capacity” (15, p457). This 522 Measuring Health satisfaction was termed “quality of life,” for ex- gators to include patient-specific items in mea- ample, by the Coronary Artery Surgery Study surement tools: the Asthma Quality of Life that reported in 1983 (16), but it remained ill- Questionnaire asks the patient to identify five ac- defined. Part of the difficulty was that different tivities that are relevant to their QoL (20), indicators showed different results: if outcomes whereas the Schedule for the Evaluation of were defined in terms of treadmill exercise tests, Quality of Life (SEIQOL) (21) and the Patient chest pain, and activity limitations, the surgical Generated Index (PGI) (22) are essentially idio- group performed better than the medical group; graphic instruments. The more the researcher in terms of return to employment or recreational recognizes that subjectivity is influenced by gen- activity, there was no advantage to surgery (16; der, age, social class, and culture, the less possi- 17). As a global concept, “QoL” had intuitive ble it becomes to draw comparisons between appeal and it formed a convenient rallying call: different people’s QoL ratings (19). A related no one could dispute the ideal of improving a pa- complicating factor in interpreting subjective tient’s QoL. However, leaving it undefined gave changes in QoL lies in the patient’s ability to investigators freedom to select whichever indica- adapt to their illness and to alter their perspec- tors they wished; QoL was invoked as much for tive on severity as their experience of disease ac- political goals as out of a scientific interest in cumulates. This “response shift” can form a evaluating care. valuable strategy for coping with the reality of a Espousing the idea of QoL measures did not chronic disease by recalibrating one’s expecta- show how the term should be defined. Because it tions for health and functioning and the relative is intuitively familiar, it appears undeserving of valuation of health states. The result can also be close definition: everyone believes he knows that changes in patient-rated QoL over time do when he is better or worse off. Although this not track objective measures of loss of function; represents the central theme, definitions of “well the complications this poses for measurement off ” seem more closely to reflect the personal have been discussed by Schwartz and Sprangers values or academic orientation of the researcher (23). than an objective attempt to define the nature of the concept. As a result, a wide variety of mea- surements began to be called QoL indicators, in- Scope of the Chapter cluding scales that bear a striking resemblance to the functional disability measures described in This chapter reviews health profiles, which de- Chapter 3. scribe health status in a set of scores, and health Nonetheless, great progress has been made indexes that summarize health in a single num- and various thoughtful reviews have indicated ber. The profiles emphasize the diverse aspects of the core concepts to be covered in a health- health or QoL; proponents of this measurement related QoL measure. Patrick and Erickson, for school hold that the dimensions of health should example, identify opportunity (e.g., resilience, be kept separate and that measurement is only stigma due to illness); health perceptions (e.g., meaningful within each domain. Supporters of self-rating of health, worry about health); physi- the health index school agree that health has sev- cal, psychological, and social functional status; eral dimensions but argue that real-life decisions and impairment (18, Table 4.1). demand that we combine the impressions from A challenge in using subjective quality of life each dimension into an overall score. Numerical assessments lies in their very subjectivity: there is indexes of health are generally intended for eco- no guarantee that each dimension of QoL con- nomic analyses of program output and for com- sidered by the researcher to be important will paring results across different programs. By have the same salience to the respondent. Indeed, contrast, profiles reflect a clinical perspective, the same measure of QoL can take on different and many scales were derived from symptom meanings for people with different subjective checklists and measures intended to monitor the views of reality (19). This has led some investi- progress and problems of individual patients.  General Health Status and Quality of Life 523 This chapter reviews 22 general health profiles In the final part of the chapter, we review in the first part of the chapter and four health in- four health indexes that provide numerical sum- dexes at the end. mary scores of health. As financial pressure In the first part of the chapter, the scales are grows, measures of this type are increasingly grouped by their intended application: clinical finding application in economic analyses of the scales; QoL for cancer patients; measures in- performance of the health care system. We re- tended for use in primary care; general evaluative view the Disability and Distress Scale, the Qual- methods; and comprehensive indicators of the ity of Well-Being Scale, the Health Utilities well-being of elderly people in the community. Index, and the EuroQol Quality of Life scale. The chapter opens with reviews of three clinical Table 10.1 compares the quality of the scales scales. The Arthritis Impact Measurement Scales in the chapter. They have in general been more represent a disease-specific instrument; this is fol- thoroughly tested for reliability and validity than lowed by the Physical and Mental Impairment- the physical disability scales reviewed in Chapter of-Function Evaluation, which is appropriate for 3. They represent some of the most successful patients living in institutions and covers more se- applications of test development procedures to vere levels of disability, and the Functional As- health measurement and, as can be seen from sessment Inventory, which focuses on a patient’s Table 10.1, are often of high quality. potential for vocational rehabilitation. There fol- low four reviews of QoL questionnaires devel- oped for cancer patients: the Functional Living References Index—Cancer, the Functional Assessment of Cancer Therapy scale, the European Quality of (1) WHOQOL Group. The World Health Or- Life Questionnaire, and Spitzer’s Quality of Life ganization quality of life assessment Index, which can also be applied to patients with (WHOQOL): position paper from the a wide variety of conditions. The QTWiST, a World Health Organization. Soc Sci Med method for evaluating the trade-off between 1995;41:1403–1409. quality and quantity of survival is also included. (2) Prutkin JM, Feinstein AR. Quality-of-life The next group of reviews covers measures measurements: origin and pathogenesis. designed for use in primary care: the Dartmouth Yale J Biol Med 2002;75:79–93. COOP Charts, a selection of single-item mea- (3) Bergner M. Quality of life, health status, sures of health, the Functional Status Question- and clinical research. Med Care 1989;27(suppl):S148–S156. naire, and the Duke Health Profile. We next (4) Jones MB. Health status indexes: the trade- describe four scales that provide more compre- off between quantity and quality of life. hensive appraisals of the well-being of elderly Socioecon Plann Sci 1977;11:301-305. people living in the community: the Multilevel (5) Kottke FJ. Philosophic considerations of Assessment Instrument, the Multidimensional quality of life for the disabled. Arch Phys Functional Activities Questionnaire developed Med Rehabil 1982;63:60–62. by the OARS group, the CARE instruments, and (6) Horley J. Life satisfaction, happiness, and the Self-Evaluation of Life Function Scale. In morale: two problems with the use of their full versions, these require lengthy inter- subjective well-being indicators. views, but they provide results that show good Gerontologist 1984;24:124–127. validity and reliability. (7) Calman KC. Quality of life of cancer patients - a hypothesis. J Med Ethics The next six scales are intended as research 1984;10:124–127. instruments for evaluating health care: the Not- (8) Cella DF, Tulsky DS. Measuring quality of tingham Health Profile, the Sickness Impact Pro- life today: methodological aspects. file, the McMaster Health Index Questionnaire, Oncology 1990;4:29–38. the Multilevel Assessment Instrument, and the (9) Spilker B, Revicki DA. Taxonomy of two Medical Outcomes Study short-form sur- quality of life. In: Spilker B, ed. Quality of veys: the SF-36 and SF-12. life and pharmacoeconomics in clinical  Table 10.1 Comparison of the Quality of General Health Measurements and Quality of Life Scales* Number Administered Studies of by Using Reliability: Reliability: Validity: Validity: Measurements Scale Items Application (Duration) Method Thoroughness Results Thoroughness Results Arthritis Impact Measurement ordinal 57 clinical self many *** ** *** ** Scale-2 (Meenan, 1992) (20 min) Physical and Mental- ordinal 77 clinical staff few * * * * Impairment-of-Function Evalua- (10–15 min) tion (PAMIE) (Gurel, 1972) Functional Assessment ordinal 40 clinical staff few * ** ** ** Inventory (Crewe and Athelstan, 1981) Functional Living Index- ordinal 22 clinical, self several * * ** *** Cancer (Schipper, 1984) research (<10 min) Functional Assessment of ordinal 27 clinical self several ** *** ** *** 524 Cancer Therapy (Cella, 1993) trials (5 min) EORTC Quality of Life ordinal 30 clinical, self several ** *** *** *** Questionnaire QLQ-C30 research (12 min) (EORTC, 1993) Quality of Life Index (Spitzer, ordinal 5 clinical self many ** ** ** ** 1980) (2 min) COOP Charts for Primary ordinal 9 clinical self many ** ** *** *** Care Practice (Nelson, 1987) (<5 min) Single-Item Health Indicators interval 1 survey self several ** ** *** ** (Andrews, 1976) Functional Status Questionnaire ordinal 34 clinical, self several ** ** ** ** (Jette, 1986) screening (15 min) Duke Health Profile ordinal 17 clinical self several ** * ** ** (Parkerson, 1990) OARS Multidimensional ordinal 144 clinical interviewer many ** ** ** ** Functional Assessment (45 min) Questionnaire (OARS, 1975)  Comprehensive Assessment and ordinal 329 clinical interviewer many ** ** *** ** Referral Evaluation (Gurland, (CORE- 1977) CARE) Multilevel Assessment ordinal 147 survey interviewer few ** ** ** ** Instrument (Lawton, 1982) (50 min) Self-Evaluation of Life Function ordinal 54 clinical self few * ** ** * Scale (Linn and Linn, 1984) McMaster Health Index ordinal 59 clinical, self several * * ** * Questionnaire (Chambers, 1976) survey (20 min) WHO Quality of Life ordinal 100 clinical, self several ** *** ** *** Scale (WHOQOL survey (10–20 min) Group, 1994) Sickness Impact Profile interval 136 research, self many *** *** *** *** (Bergner, 1976) survey (20–30 min) Nottingham Health Profile interval 45 clinical, self many ** ** ** ** (Hunt, 1981) survey (10–15 min) Short-Form-36 Health Survey ordinal 36 survey self many *** *** *** *** 525 (Ware, 1990) (5–10 min) Short-Form-12 Health Survey ordinal 12 survey self many ** *** *** *** (Stewart, 1987) (3–4min) Disability and Distress Scale ratio 2 research expert few * ** * ** (Rosser, 1978) Quality of Well-Being Scale ratio 18 research interviewer many ** *** *** *** (Bush and Kaplan, 1973) (7 min) Health Utilities Index ratio 31 survey, self, interviewer several *** ** *** *** (Torrance and Feeny, 1990) clinical (2–5 min) EuroQol EQ-5D ratio 5 research self many ** ** *** *** Quality of Life Scale (EuroQol Group, 1990) * For an explanation of the categories used, see Chapter 1, pages 6–7 526 Measuring Health trials. Philadelphia: Lippincott-Raven, (22) Ruta DA, Garratt AM, Leng M, et al. A 1996:25–31. new approach to the measurement of (10) de Haes JCJM, van Knippenberg FCE. The quality of life: The Patient-Generated quality of life of cancer patients: a review of Index. Med Care 1994;32:1109–1126. the literature. Soc Sci Med 1985;20:809–817. (23) Schwartz CE, Sprangers MAG. Adaptation (11) Williams GH. Quality of life and its impact to changing health: response shifts in on hypertensive patients. Am J Med quality of life research. Washington: 1987;82:98–105. American Psychological Association, 2002. (12) Johnson JR, Temple R. Food and Drug Administration requirements for approval of new anticancer drugs. Cancer Treat Rep The Arthritis Impact 1985;69:1155–1157. Measurement Scales (13) CASS Principal Investigators and their (Robert F. Meenan, 1980, Associates. Coronary Artery Surgery Study revised 1991) (CASS): a randomized trial of coronary artery bypass surgery. Circulation Purpose 1983;68:939–950. The self-administered Arthritis Impact Measure- (14) Veterans Administration Coronary Artery ment Scales (AIMS) cover physical, social, and Bypass Surgery Cooperative Study Group. emotional well-being and were designed as an Eleven-year survival in the Veterans indicator of the outcome of care for arthritic pa- Administration randomized trial of tients. coronary bypass surgery for stable angina. N Engl J Med 1984;311:1333–1339. (15) LaMendola WF, Pellegrini RV. Quality of Conceptual Basis life and coronary artery bypass surgery Meenan criticized the measurements tradition- patients. Soc Sci Med 1979;13A:457–461. ally used with patients with arthritis for their fo- (16) CASS Principal Investigators and their cus on disease activity and functional abilities to Associates. Coronary Artery Surgery Study the exclusion of other components identified in (CASS): a randomized trial of coronary the World Health Organization (WHO) defini- artery bypass surgery. Circulation tion of health (1). “The major argument in favor 1983;68:951–960. of the new questionnaire-based approaches is (17) Scheidt S. Ischemic heart disease: a patient- that they focus on the components of outcome specific therapeutic approach with that are the most relevant to the physician and emphasis on quality of life considerations. the patient, both of whom are primarily inter- Am Heart J 1987;114:251–257. (18) Patrick DL, Erickson P. Health status and ested in how the patient feels and how he or she health policy: quality of life in health care functions” (2, p168). The AIMS were intended evaluation and resource allocation. New to be comprehensive and practical, with an em- York: Oxford University Press, 1993. phasis on proven reliability and validity (3). (19) Stennar PHD, Cooper D, Skevington SM. Putting the Q into quality of life: the Description identification of subjective constructions of The AIMS was revised in 1991, but given that health-related quality of life using the Q much of the evidence for validity and reliability methodology. Soc Sci Med was collected using the original version, both are 2003;57:2161–2172. described here. The original AIMS instrument (20) Juniper EF, Buist AS, Cox FM, et al. included 45 items grouped into nine scales that Validation of a standardized version of the assess mobility, physical activity (e.g., walking, Asthma Quality of Life Questionnaire. Chest 1999;115:1265–1270. bending, lifting), dexterity, household activity (21) Joyce CRB, Hickey A, McGee HM, et al. (e.g., managing money, medications, house- A theory-based method for the evaluation keeping), social activities, activities of daily liv- of individual quality of life: the SEIQoL. ing (ADL), pain, depression, and anxiety. This Qual Life Res 2003;12:275–280. was shown in the second edition of Measuring  General Health Status and Quality of Life 527 Health. The dexterity and pain scales were de- Table 1 of the revised manual. The dimension veloped by Meenan, whereas other items were scores can then be added to form the five compo- adapted from Katz’s Index of ADL, the RAND nents already indicated; occasionally only three instruments, and the Quality of Well-Being Scale (Physical, Affect and Symptoms) are used. (1). Items were selected for inclusion on the ba- sis of Guttman analyses and internal consistency Reliability correlations (3). A total health score was formed original aims. For the original AIMS, the by adding the values for six of the scales: mobil- Guttman coefficients of scalability and repro- ity, physical and household activities, dexterity, ducibility exceeded 0.60 and 0.90, respectively, pain, and depression. The instrument is self- for all but the household activity scale, which administered and takes about 15 minutes to had a coefficient of reproducibility of 0.88 (3, complete. Table 2). In a study of 625 patients with arthri- The revised version, the AIMS2, was first pre- tis, Guttman reproducibility coefficients for all sented in 1991 as a more comprehensive and scales exceeded 0.90 (6, Table 2). sensitive instrument (4). It contains 78 items, of The alpha internal consistencies of all nine which the first 57 are grouped into 12 scales that scales exceeded 0.60; six exceeded 0.80 (2, extend the coverage of the original AIMS (Ex- Table 2). In another study, alpha coefficients ex- hibit 10.1). The 12 scales can be further grouped ceeded 0.70 for all scales except physical activity into five components: Physical (e.g., mobility, (0.63) and social activity (0.69) (6, Table 2). walking and bending, hand and finger function, Test-retest correlations for the nine scales ex- arm function, self-care and household tasks); ceeded 0.80 after a two-week delay; the mean Symptoms (arthritis pain); Role (work); Social test-retest correlation was 0.87 for 100 patients Interaction (social activities and social support), (1; 2; 6). These results were replicated in several and Affect (tension and mood). A further 44 diagnostic groups (6). Test-retest reliability questions cover satisfaction with health, the im- ranged from 0.63 to 0.89 for the scales in three pact of the patient’s arthritis on their function, groups of patients with chronic disease (7, and the patient’s priorities for improvement. The p352). 57 core items shown in the exhibit were derived from the original AIMS, but all were reworded. aims2. For the AIMS2, 10-day retest intraclass The nine topic areas of the AIMS were renamed correlation coefficients for each item ranged and sections were added on arm function, social from 0.34 to 0.90 with a median of 0.65 (8, support, and work. The answer format was al- Table 1). In Meenan’s original article, test-retest tered to make responses more closely standard reliability for the 12 scales ranged from 0.78 to across the sections. The questionnaire is in the 0.94. Alpha coefficients were quoted for two pa- public domain and is available online from Dr. tient groups and ranged from 0.72 to 0.96 (4, Meenan’s web site: http://dcc2.bumc.bu.edu Table 2). .rmeenan/pdfs/aims2.pdf. The user’s manual is available from www.qolid.org/public/aims/cadre/ Validity guide.pdf. The AIMS2 takes about 20 minutes original aims. Correlations between the scales for respondents to complete. and a number of criterion variables were exam- Scoring is straightforward and is explained in ined. These included age (on the expectation of a the user’s manual. Certain questions use a re- reduction in function with age), the patient’s versed phrasing, so they are reversed during perception of general health and of recent dis- scoring so that high scores indicate poor health. ease activity, and a physician’s report of func- Scores for items in each of the 12 scales are then tional activity, joint count, and disease activity added and converted to a range of 0 to 10, using (3). Meenan et al. commented: simple standardization formulae for each sec- tion. These are shown for the original AIMS (5), The performance-oriented scales generally whereas formulas for the AIMS2 are given in correlated closely with age, and all 9 scales Exhibit 10.1 The Arthritis Impact Measurement Scales, Version 2 (AIMS2) Note: The answer scales are shown at the end of the exhibit. Please answer the following questions about your health. Most questions ask about your health during the past month. There are no right or wrong answers to the questions and most can be answered with a simple check (X). Please answer every question. Please check (X) the most appropriate answer for each question. These questions refer to MOBILITY LEVEL DURING THE PAST MONTH. . . . (answer scale A) 1. How often were you physically able to drive a car or use public transportation? 2. How often were you out of the house for at least part of the day? 3. How often were you able to do errands in the neighborhood? 4. How often did someone have to assist you to get around outside your home? 5. How often were you in a bed or chair for most or all of the day? These questions refer to WALKING AND BENDING DURING THE PAST MONTH. . . . (answer scale A) 6. Did you have trouble doing vigorous activities such as running, lifting heavy objects, or participating in strenuous sports? 7. Did you have trouble either walking several blocks or climbing a few flights of stairs? 8. Did you have trouble bending, lifting or stooping? 9. Did you have trouble either walking one block or climbing one flight of stairs? 10. Were you unable to walk unless assisted by another person or by a cane, crutches, or walker? These questions refer to HAND AND FINGER FUNCTION DURING THE PAST MONTH. . . . (answer scale A) 11. Could you easily write with a pen or pencil? 12. Could you easily button a shirt or blouse? 13. Could you easily turn a key in a lock? 14. Could you easily tie a knot or a bow? 15. Could you easily open a new jar of food? These questions refer to ARM FUNCTION DURING THE PAST MONTH. . . . (answer scale A) 16. Could you easily wipe your mouth with a napkin? 17. Could you easily put on a pullover sweater? 18. Could you easily comb or brush your hair? 19. Could you easily scratch your low back with your hand? 20. Could you easily reach shelves that were above your head? These questions refer to SELF-CARE TASKS DURING THE PAST MONTH. . . . (answer scale B) 21. Did you need help to take a bath or shower? 22. Did you need help to get dressed? 23. Did you need help to use the toilet? 24. Did you need help to get in or out of bed? 528 Exhibit 10.1 These questions refer to HOUSEHOLD TASKS DURING THE PAST MONTH. . . . (answer scale B) 25. If you had the necessary transportation, could you go shopping for groceries without help? 26. If you had kitchen facilities, could you prepare your own meals without help? 27. If you had household tools and appliances, could you do your own housework without help? 28. If you had laundry facilities, could you do your own laundry without help? These questions refer to SOCIAL ACTIVITY DURING THE PAST MONTH. . . . (answer scale A) 29. How often did you get together with friends or relatives? 30. How often did you have friends or relatives over to your home? 31. How often did you visit friends or relatives in their homes? 32. How often were you on the telephone with close friends or relatives? 33. How often did you go to a meeting of a church, club, team or other group? These questions refer to SUPPORT FROM FAMILY AND FRIENDS DURING THE PAST MONTH. . . . (answer scale B) 34. Did you feel that your family or friends would be around you if you needed assistance? 35. Did you feel that your family or friends were sensitive to your personal needs? 36. Did you feel that your family or friends were interested in helping you solve problems? 37. Did you feel that your family or friends understood the effects of your arthritis? These questions refer to ARTHRITIS PAIN DURING THE PAST MONTH. . . . (answer scale C) 38. How would you describe the arthritis pain you usually had? DURING THE PAST MONTH. . . . (answer scale A) 39. How often did you have severe pain from your arthritis? 40. How often did you have pain in two or more joints at the same time? 41. How often did your morning stiffness last more than one hour from the time you woke up? 42. How often did your pain make it difficult for you to sleep? These questions refer to WORK DURING THE PAST MONTH. . . . (answer scale D) 43. What has been your main form of work? If you answered unemployed, disabled or retired, please skip the next four questions and go to the next page. DURING THE PAST MONTH. . . . (answer scale A) 44. How often were you unable to do any paid work, housework, or school work? 45. On the days that you did work, how often did you have to work a shorter day? 46. On the days that you did work, how often were you unable to do your work as carefully and accurately as you would like? 47. On the days that you did work, how often did you have to change the way your paid work, housework or school work is usually done? (continued) 529 530 Measuring Health Exhibit 10.1 (continued) These questions refer to LEVEL OF TENSION DURING THE PAST MONTH. . . . (answer scale B) 48. How often have you felt tense or high strung? 49. How often have you been bothered by nervousness or your nerves? 50. How often were you able to relax without difficulty? 51. How often have you felt relaxed and free of tension? 52. How often have you felt calm and peaceful? These questions refer to MOODDURING THE PAST MONTH. . . . (answer scale B) 53. How often have you enjoyed the things you do? 54. How often have you been in low or very low spirits? 55. How often did you feel that nothing turned out the way you wanted it to? 56. How often did you feel that others would be better off if you were dead? 57. How often did you feel so down in the dumps that nothing would cheer you up? Answer scales: A. (1) All days, (2) Most Days, (3) Some Days, (4) Few Days, (5) No Days B. (1) Always, (2) Very Often, (3) Sometimes, (4) Almost Never, (5) Never C. (1) Severe, (2) Moderate, (3) Mild, (4) Very Mild, (5) None D. (1) Paid Work, (2) Housework, (3) School work, (4) Unemployed, (5) Disabled, (6) Retired Adapted from an original provided by Dr. Meenan. were significantly correlated with the pa- hip replacement a lower correlation of 0.48 was tient’s estimates of general health and disease obtained between the short AIMS and the modi- activity. Finally, when the psychological fied HAQ, whereas the AIMS correlated 0.76 scales are excluded, agreement between the with the Functional Status Questionnaire (10, scale scores and the doctor’s report was sig- Table 3). Kazis et al. found that the mobility and nificant in 16 of 21 pairs (76%). (3, p150) general health perceptions scales predicted mor- tality outcomes (11). Similar analyses were carried out using the A factor analysis of the nine scales provided data from the study of 625 arthritic patients. For three factors: physical function, psychological, 444 of the patients, the nine scales were corre- and pain (6). These results were subsequently lated with disease activity (r between 0.14 and precisely replicated (12). Reflecting this three- 0.52) and with the American Rheumatism Asso- component factor structure, results of a multi- ciation (ARA) functional class (r between 0.24 variate approach to criterion validation showed and 0.52) (2, Table 4). Scales measuring physical stronger associations than the single-variate functioning showed higher associations with analyses reported here did. Using multiple re- these disease indicators than the psychological gression analyses, the AIMS scores achieved and social scales did. Several studies have com- multiple correlation coefficients of 0.61 with dis- pared the AIMS and the Health Assessment ease activity and 0.66 with the ARA functional Questionnaire (HAQ). The correlation of the class index (6, p1050). Multiple correlations overall scores was 0.75 in a French study of 70 with a three-item measure of global health status patients (9); the physical scales correlated 0.91, and a visual analogue measure of arthritis im- whereas the pain scales in the two instruments pact were 0.84 and 0.75, respectively (6, correlated 0.64 in a study of 48 arthritic patients p1050). Mason et al. subsequently reconsidered (6). In a study of 106 patients who underwent the three-factor solution, arguing that it was the-  General Health Status and Quality of Life 531 oretically restrictive. They proposed a five-factor (symptom scores) −0.84 (affect scores) solution that presented scores for lower extrem- (19, Table 2), and 0.78 for the overall score ity function (mobility, ADL, and physical activ- (20, Table 1). Correlations with the SF-36 include ity), upper extremity function (dexterity), affects −0.73 (physical scales) and −0.78 (symptoms) (anxiety and depression), pain, and social inter- (19, Table 2). Correlations with measures of dis- action (13). ease activity (e.g., grip strength, morning stiff- Sensitivity of the AIMS to change has been ness, and functional class) were reported for studied by several authors. Changes in the scores patients with psoriatic arthritis. Values were low were correlated with changes in a rating of to moderate, typically 0.3 to 0.5. The AIMS2 health following treatment for 120 patients; cor- scores correlated much less highly, however, relations fell between 0.24 and 0.67 (1; 6). Kazis with measures of disease severity (e.g., number et al. studied the responsiveness of the AIMS in a of affected joints, ARA anatomic stage) (21, Ta- study of injectable gold. Seven of the AIMS bles 3 and 4). Similar figures were reported from scales achieved effect sizes that were small; phys- a study of patients with ankylosing spondylitis; ical activity gave a moderate effect size; and the here, correlations with a range of clinical assess- pain scale gave a large effect size (14, Table 2). ments of disease severity were almost identical to In a second trial, effect sizes for pain and anxiety those of the AIMS2 and the HAQ (20, Table 2). scores were large; those for physical activity and In terms of sensitivity to change, in a study of depression were moderate and the rest were patients with arthritis, the AIMS2 physical func- small (14, Table 3). The AIMS scales provide tion score provided slightly greater sensitivity to similar effect sizes for rheumatoid arthritis and change than the modified HAQ; the pain score osteoarthritis; again, the pain scale showed the was more sensitive than a visual analogue scale largest effect of treatment in two clinical trials (VAS) for pain (22, Table 2). The sensitivity of (15). The AIMS total score showed a more sig- each item to change was reported by Guillemin nificant change in patients after hip replacement et al. The mean was 0.28, with a range from surgery than did the McMaster Health Index 0.18 to 0.73 (8, Table 1). For patients who re- Questionnaire (16, Table IV). In a more compre- ported improvement in their condition, stan- hensive comparison of five scales by Liang et al., dardized response means were 0.77 (physical the pain, mobility, and overall scores of the score), 1.21 (symptoms), and 1.21 (role). For pa- AIMS were clearly superior in terms of effect tients who felt much worse, response means size to those of the Sickness Impact Profile (SIP), were −1.47 for physical scores, −1.83 for symp- the Quality of Well-Being Scale, the Health As- toms, and −0.75 for role (19, Table 3). In a study sessment Questionnaire, and the Functional Sta- of rheumatoid arthritis, standardized response tus Index (17, Table 2). For the global and means were comparable for the physical scales mobility scores, the sample size required by the from the Modified HAQ, the AIMS2, and the AIMS to demonstrate a significant difference SF-36. The SF-36 and AIMS2 pain scales also would be less than half that required for most of provided similar responsiveness, but a VAS was the other measures (17, Table 3). superior (SRM 2.0 compared with 1.2 for the AIMS2) (23, Table 5). Husted et al. compared aims2. Meenan et al. compared AIMS2 scores responsiveness of the SF-36, the HAQ and the according to the patients’ own identification of AIMS2. The results differed depending on the each area as being a problem for them; scores criterion used to identify change, but the overall showed clear contrasts in response pattern (4, impression was of a slight advantage for the SF- Table 3). Comparisons with the Health Assess- 36, and comparable results for the HAQ and the ment Questionnaire include a correlation of AIMS2 (24, Table 7). 0.75 for the overall scores in a French study of 70 patients (9); in another study, the AIMS and Alternative Forms HAQ overall scores correlated 0.89 (18). Other An abbreviated form of the original AIMS in- figures include 0.86 (physical scores), 0.76 cluded 18 items divided into nine scales. The 532 Measuring Health two items with the highest internal consistency related with the Swedish version of the HAQ (36). and correlation with the total AIMS score were Information on the availability of translations selected from the original AIMS scales. The in- may be obtained from www.qolid.org/public/ strument takes six to eight minutes to complete AIMS.html (25). Alpha reliability was only slightly lower than that of the full AIMS, and test-retest relia- Reference Standards bility was virtually identical (25, Table 2). Like- Meenan et al. recorded mean scores and stan- wise, concurrent validity coefficients were dard deviations for samples of rheumatoid and similar for long and short forms (Table 3). In a osteoarthritis patients (4, Table 1). study of patients with hip pain, the abbreviated version of the AIMS proved more sensitive to Commentary change than the SIP, SF-36, or Functional Status The AIMS is one of the most widely used out- Questionnaire (10, Table 4). come measures in arthritis research. It is well A 26-item abbreviation of the AIMS2 has documented and clearly described, and there is been described (8; 19; 26). This provides Physi- strong evidence for the reliability and validity of cal, Symptom, Role, Social Interaction, and Af- the original version, while evidence for the fect components; alpha coefficients ranged from AIMS2 is accumulating, especially in Europe. Al- 0.32 (social interaction) to 0.87 (physical) (8, though the AIMS is intended mainly for research, Table 4). Haarvardsholm et al. reported correla- Kazis et al. described a clinical report format that tions of 0.96 and 0.97 between the AIMS2 and summarized patient profiles on one page; this the short form; intraclass correlations for the 12 may make the instrument suitable for routine scales ranged from 0.85 to 0.97, with all but one clinical use (46). Although relatively long, the being 0.95 or above (19, Table 1). They pro- AIMS deserves serious consideration as an out- posed that the Role scale be slightly modified, by come indicator for use in patients with arthritis. replacing item 42 with item 38. Other suggested changes include replacing item 33 with item 31 Address on the Social Interaction scale (27). The AIMS2 User’s Manual is available online from The AIMS has been used with children aged www.qolid.org/public/aims/cadre/guide.pdf Gen- under 10 years with mixed success. The pain eral information on the AIMS2, and a copy of the scale appeared the most reliable; limited variabil- scale, is provided on Dr. Meenan’s web site: ity on the mobility and ADL scales suggested that http://dcc2.bumc.bu.edu/rmeenan/pdfs/aims2.pdf. they may require modification for use with chil- dren (28, p823). A version suited to geriatric pa- References tients (the GERI-AIMS) has been described (29). The AIMS and AIMS2 have been translated (1) Meenan RF. The AIMS approach to health into many languages, including Spanish (30–34), status measurement: conceptual Swedish (35; 36), Dutch (37–39), Italian (40), background and measurement properties. J Japanese (41), Canadian French (42), and (of Rheumatol 1982;9:785–788. course) British English (43). A French version of (2) Meenan RF. New approaches to outcome the AIMS2 has been tested for validity and relia- assessment: the AIMS questionnaire for bility (44; 45). A Dutch version showed internal arthritis. Adv Intern Med consistency alphas in excess of 0.80 for six 1986;31:167–185. (3) Meenan RF, Gertman PM, Mason JH. scales and an alpha between 0.6 and 0.7 for the Measuring health status in arthritis: the rest (39, Table II). Validity correlations with an Arthritis Impact Measurement Scales. independent assessment of functional status Arthritis Rheum 1980;23:146–152. ranged from 0.7 to 0.8 for the physical function- (4) Meenan RF, Mason JH, Anderson JJ, et al. ing scales (39, Table IV). Internal consistency for AIMS2: the content and properties of a the subscales ranged from 0.57 to 0.90 in the revised and expanded Arthritis Impact Swedish version; the physical function scales cor- Measurement Scales health status  General Health Status and Quality of Life 533 questionnaire. Arthritis Rheum (17) Liang MH, Fossel AH, Larson MG. 1992;35:1–10. Comparisons of five health status (5) Boston University Multipurpose Arthritis instruments for orthopedic evaluation. Med Center. AIMS user’s guide. [Manuscript]. Care 1990;28:632–642. Boston: Boston University, 2000. (18) Hakala M, Nieminen P, Manelius J. Joint (6) Meenan RF, Gertman PM, Mason JH, et impairment is strongly correlated with al. The Arthritis Impact Measurement disability measured by self-report Scales: further investigations of a health questionnaires. Functional status status measure. Arthritis Rheum assessment of individuals with rheumatoid 1982;25:1048–1053. arthritis in a population based series. J (7) Burckhardt CS, Woods SL, Schultz AA, et Rheumatol 1994;21:64–69. al. Quality of life of adults with chronic (19) Haarvardsholm EA, Kvien TK, Uhlig T, et illness: a psychometric study. Res Nurs al. A comparison of agreement and Health 1989;12:347–354. sensitivity to change between AIMS2 and a (8) Guillemin F, Coste J, Pouchot J, et al. The short form of AIMS2 (AIMS2-SF) in more AIMS2-SF: a short form of the Arthritis than 1000 rheumatoid arthritis patients. J Impact Measurement Scales 2. Arthritis Rheumatol 2000;27:2810–2816. Rheum 1997;40:1267–1274. (20) Ward MM, Kuzis S. Validity and sensitivity (9) Taccari E, Spadaro A, Rinaldi T, et al. to change of spondylitis-specific measures Comparison of the Health Assessment of functional disability. J Rheumatol Questionnaire and Arthritis Impact Mea- 1999;26:121–127. surement Scale in patients with psoriatic (21) Husted JA, Gladman DD, Farewell VT, et arthritis. Rev Rhum Eng Ed al. Validation of the revised and expanded 1998;65:751–758. version of the Arthritis Impact Measure- (10) Katz JN, Larson MG, Phillips CB, et al. ment Scales for patients with psoriatic Comparative measurement sensitivity of arthritis. J Rheumatol short and longer health status instruments. 1996;23:1015–1019. Med Care 1992;30:917–925. (22) Taal E, Rasker JJ, Riemsma RP. Sensitivity (11) Kazis LE, Anderson JJ, Meenan RF. Health to change of AIMS2 and AIMS2-SF status as a predictor of mortality in components in comparison to M-HAQ and rheumatoid arthritis: a five-year study. J VAS-pain. Ann Rheum Dis Rheumatol 1990;17:609–613. 2004;63:1655–1658. (12) Brown JH, Kazis LE, Spitz PW, et al. The (23) Hagen KB, Smedstad LM, Uhlig T, et al. dimensions of health outcomes: a cross- The responsiveness of health status mea- validated examination of health status mea- sures in patients with rheumatoid arthritis: surement. Am J Public Health comparison of disease-specific and generic 1984;74:159–161. instruments. J Rheumatol (13) Mason JH, Anderson JJ, Meenan RF. A 1999;26:1474–1480. model of health status for rheumatoid (24) Husted JA, Gladman DD, Cook RJ, et al. arthritis: a factor analysis of the Arthritis Responsiveness of health status Impact Measurement Scales. Arthritis instruments to changes in articular status Rheum 1988;31:714–720. and perceived health in patients with (14) Kazis LE, Anderson JJ, Meenan RF. Effect psoriatic arthritis. J Rheumatol sizes for interpreting changes in health status. 1998;25:2146–2155. Med Care 1989;27(suppl):S178–S189. (25) Wallston KA, Brown GK, Stein MJ, et al. (15) Anderson JJ, Firschein HE, Meenan RF. Comparing the short and long versions of Sensitivity of a health status measure to the Arthritis Impact Measurement Scales. J short-term clinical changes in arthritis. Rheumatol 1989;16:1105–1109. Arthritis Rheum 1989;32:844–850. (26) Ren XS, Kazis L, Meenan RF. Short-form (16) O’Boyle CA, McGee H, Hickey A, et al. Arthritis Impact Measurement Scales: test Individual quality of life in patients of reliability and validity among patients undergoing hip replacement. Lancet with osteoarthritis. Arthritis Care Res 1992;339:1088–1091. 1999;12:163–171. 534 Measuring Health (27) Taal E, Rasker JJ, Riemsma RP. Evaluation of a Dutch version of the Psychometric properties of a Dutch short AIMS2 for patients with rheumatoid form of the Arthritis Impact Measurement arthritis. Br J Rheumatol Scales 2 (Dutch-AIMS2-SF). Rheumatology 1996;35:755–760. 2003;42:427–434. (38) Evers AW, Taal E, Kraaimaat FW, et al. A (28) Coulton CJ, Zborowsky E, Lipton J, et al. comparison of two recently developed Assessment of the reliability and validity of health status instruments for patients with the Arthritis Impact Measurement Scales arthritis: Dutch-AIMS2 and IRGL. for children with juvenile arthritis. Arthritis Impact Measurement Scales. Arthritis Rheum 1987;30:819–824. Impact of rheumatic diseases on general (29) Hughes SL, Edelman P, Chang RW, et al. health and lifestyle. Br J Rheumatol The GERI-AIMS. Reliability and validity 1998;37:157–164. of the arthritis impact measurement scales (39) Taal E, Jacobs JW, Seydel ER, et al. adapted for elderly respondents. Arthritis Evaluation of the Dutch Arthritis Impact Rheum 1991;34:856–865. Measurement Scales (Dutch-AIMS) in (30) Hendricson WD, Russell IJ, Prihoda TJ, et patients with rheumatoid arthritis. Br J al. Development and initial validation of a Rheumatol 1989;28:487–491. dual-language English-Spanish format for (40) Cavalieri F, Salaffi F, Ferracioli GF. the Arthritis Impact Measurement Scales. Relationship between physical impairment, Arthritis Rheum 1989;32:1153–1159. psychosocial variables and pain in (31) Abello-Banfi M, Cardiel MH, Ruiz- rheumatoid disability. An analysis of their Mercado R, et al. Quality of life in relative impact. Clin Exp Rheumatol rheumatoid arthritis: validation of a 1991;9:47–50. Spanish version of the Arthritis Impact (41) Sato H, Araki S, Hashimoto A, et al. The Measurement Scales. Arthritis Rheum validity and reliability of a Japanese version 1991;34(suppl 9):S183. of Arthritis Impact Measurement Scales in (32) Cardiel MH, Abello-Banfi M, Ruiz- patients with rheumatoid arthritis. Mercado R, et al. How to measure health Ryumachi 1995;35:566–574. status in rheumatoid arthritis in non-En- (42) Sampalis JS, Pouchot J, Beaudet F, et al. glish speaking patients: validation of a Arthritis Impact Measurement Scales: Spanish version of the Health Assessment reliability of a French version and validity Questionnaire Disability Index (Spanish in adult Still’s disease. J Rheumatol HAQ-DI). Clin Exp Rheumatol 1990;17:1657–1661. 1993;11:117–121. (43) Hill J, Bird HA, Lawton CW, et al. The (33) Abello-Banfi M, Cardiel MH, Ruiz- Arthritis Impact Measurement Scales: an Mercado R, et al. Quality of life in anglicized version to assess the outcome of rheumatoid arthritis: validation of a British patients with rheumatoid arthritis. Spanish version of the Arthritis Impact Br J Rheumatol 1990;29:193–196. Measurement Scales (Spanish-AIMS). J (44) Pouchot J, Guillemin F, Coste J, et al. Rheumatol 1994;21:1250–1255. Validity, reliability, and sensitivity to (34) Danao LL, Padilla GV, Johnson DA. An change of a French version of the arthritis English and Spanish quality of life measure impact measurement scales 2 (AIMS2) in for rheumatoid arthritis. Arthritis Care Res patients with rheumatoid arthritis treated 2001;45:167–173. with methotrexate. J Rheumatol (35) Archenholtz B, Bjelle A. Reliability and 1996;23:52–60. validity of a Swedish version of the (45) Poiraudeau S, Dougados M, Ait-Hadad H, Arthritis Impact Measurement Scales. et al. Evaluation of a quality of life scale Scand J Rheumatol 1990;20:219. (AIMS2) in rheumatology. Rev Rhum Mal (36) Archenholtz B, Bjelle A. Evaluation of a Osteoartic 1993;60:561–567. Swedish version of the Arthritis Impact (46) Kazis LE, Anderson JJ, Meenan RF. Health Measurement Scales (AIMS). Scand J status information in clinical practice: the Rheumatol 1995;24:64–68. development and testing of patient profile (37) Riemsma RP, Taal E, Rasker JJ, et al. reports. J Rheumatol 1988;15:338–344.  General Health Status and Quality of Life 535 The Physical and Mental IV. Anxiety, depression V. Bedfast, moribund Impairment-of-Function Evaluation VI. Behavioral deterioration (Lee Gurel, 1972) VII. Paranoia, suspicion Purpose VIII. Sensory and motor function The Physical and Mental Impairment-of- IX. Withdrawn, apathetic Function Evaluation (PAMIE) is a clinical rating X. Ambulation scale that records physical, psychological, and social disability in chronically ill, institutional- Where no roman numeral is given, the question ized elderly patients (1). did not load on a factor in the analyses. Al- though conceptually distinct, these factors may Conceptual Basis commonly be found together in the same patient The PAMIE was based on two previous instru- and so are not independent of one another. Fac- ments, the Self-Care Inventory, a rating scale of tor scores may be added to give three more gen- activities of daily living (ADL) used in U.S. Vet- eral scores representing physical infirmity eran’s Administration hospitals with severely (factors I, V, VIII, X), psychological deteriora- disabled geriatric patients, and its refinement, tion (factors III, VI, IX), and psychological agi- the 43-item Patient Evaluation Scale, which as- tation (factors II, IV, VII) (1). Weighted and sessed the potential for hospital discharge (1). unweighted factor scores provided essentially Factor analyses of the latter scale guided the identical results, so the unweighted scores are content of the PAMIE, which was intended to generally used. cover 12 topics. Subsequent empirical testing of the PAMIE indicated that scores are best pre- Reliability sented for ten factors, rather than the 12 first Alpha internal consistency coefficients for the hypothesized (1). factor scores ranged from 0.67 to 0.91 (1, Table 2). Description The PAMIE is a rating scale completed by a Validity caregiver or clinician familiar with the patient The PAMIE scale was tested on 845 male veter- (1). The 77 items are mainly concerned with ob- ans in long-term care facilities. Their mean age servable behaviors during the preceding week; was 66 years; 47% were general medical and all but the first three use a yes/no answer format. surgical patients, whereas the remainder had Exhibit 10.2 shows a slightly revised version of predominantly psychiatric problems, often ac- the PAMIE from that shown in reference 1, companied by additional medical complaints Table 1. The instrument takes 10 to 15 minutes (1). The factor structure of the PAMIE scale was to complete. examined, and nine factors were derived. How- The scoring system for the first three ques- ever, Gurel et al. chose to separate questions on tions is shown in the exhibit; for the remaining ambulation from the self-care factor, thus form- questions, the roman numeral indicates the scale ing the tenth factor scored (1). To assess the sta- on which it is counted, and its position shows bility of the factor solution, the analysis was the response that receives one point. Empirical repeated on medical/surgical and psychiatric pa- analyses did not confirm the existence of sepa- tients separately, with considerable agreement rate factors for irritability and cooperation, and between them, as indicated by a Harman’s coef- so scores may be provided for the following ten ficient of congruence of 0.86 (1, p85). Several of factors (2, p236): the factors were substantially correlated, and a second-order factor analysis provided three fac- I. Self-care tors, which the authors named “physical infir- II. Belligerence, irritability mity” (including ambulation, sensory and motor III. Mental confusion functions, bedfastness), “psychological deterio- Exhibit 10.2 The Physical and Mental Impairment-of-Function Evaluation On the basis of your knowledge of the patient at the present time will you please rate the following items. Answer items 1, 2 and 3 on this page by circling the number beside the most correct statement. For all other items check either Yes or No. Please do not leave any item unanswered. 1. Which of the following best fits the patient? (Circle one) 5 Has no problem in walking 4 Slight difficulty in walking, but manages; may use cane 3 Great difficulty in walking, but manages; may use crutches or stroller 2 Uses wheelchair to get around by himself 1 Uses wheelchair pushed by others 0 Doesn’t get around much; mostly or completely bedfast, or restricted to chair (Factor X) 2. As far as you know, has the patient had one or more strokes (CVA)? (Circle one) 0 No stroke 1 Mild stroke(s) 2 Serious stroke(s) (Factor VIII) 3. Which of the following best fits the patient? (Circle one) 4 In bed all or almost all day 3 More of the waking day in bed than out of bed 2 About half the waking day in bed, about half out of bed 1 More of the waking day out of bed than in bed 0 Out of bed all or almost all day (Factor V) Yes No (Check either Yes or No) 4. _____ _____ Eats a regular diet 5. _____ V _____ Is given bed baths 6. _____ II _____ Gives sarcastic answers 7. _____ _____ I Takes a bath/shower without help or supervision 8. _____ VI _____ Leaves his clothes unbuttoned 9. _____ VI _____ Is messy in eating 10. _____ II _____ Is irritable and grouchy 11. _____ IX _____ Keeps to himself 12. _____ VII _____ Says he’s not getting good care and treatment 13. _____ II _____ Resists when asked to do things 14. _____ IV _____ Seems unhappy 15. _____ III _____ Doesn’t make much sense when he talks to you 16. _____ II _____ Acts as though he has a chip on his shoulder 17. _____ V _____ Is IV or tube fed once a week or more 18. _____ VIII _____ Has one or both hands/arms missing or paralyzed 19. _____ _____ II Is cooperative 20. _____ V _____ Is toileted in bed by catheter and/or enema 21. _____ _____ Is deaf or practically deaf, even with hearing aid 22. _____ IX _____ Ignores what goes on around him 23. _____ _____ V Knows who he is and where he is 24. _____ II _____ Gives the staff a “hard time” 25. _____ VII _____ Blames other people for his difficulties 26. _____ VII _____ Says, without good reason, that he’s being mistreated or getting a raw deal 27. _____ II _____ Gripes and complains a lot 536 Exhibit 10.2 Yes No (Check either Yes or No) 28. _____ VII _____ Says other people dislike him, or even hate him 29. _____ _____ Says he has special or superior abilities 30. _____ _____ Has hit someone or been in a fight in the last six months 31. _____ _____ I Eats without being closely supervised or encouraged 32. _____ IV _____ Says he’s blue and depressed 33. _____ IX _____ Isn’t interested in much of anything 34. _____ III _____ Has taken his clothes off at the wrong time or place during the last six months 35. _____ _____ Makes sexually suggestive remarks or gestures 36. _____ II _____ Objects or gives you an argument before doing what he’s told 37. _____ VII _____ Is distrustful and suspicious 38. _____ _____ VI Looks especially neat and clean 39. _____ IV _____ Seems unusually restless 40. _____ II _____ Says he’s going to hit people 41. _____ III _____ Receives almost constant safety supervision (for careless smoking, objects in mouth, self-injury, pulling catheter, etc.) 42. _____ VI _____ Looks sloppy 43. _____ III _____ Keeps wandering off the subject when you talk with him 44. _____ VI _____ Is noisy; talks very loudly 45. _____ _____ I Does things like brush teeth, comb hair, and clean nails without help or urging 46. _____ _____ Has shown up drunk or brought a bottle on the ward 47. _____ IV _____ Cries for no obvious reason 48. _____ _____ Says he would like to leave the hospital 49. _____ I _____ Wets or soils once a week or more 50. _____ III _____ Has trouble remembering things 51. _____ VIII _____ Has one or both feet/legs missing or paralyzed 52. _____ X _____ Walks flight of steps without help 53. _____ _____ V When needed, takes medication by mouth 54. _____ IV _____ Is easily upset when little things go wrong 55. _____ _____ I Uses the toilet without help or supervision 56. _____ _____ V Conforms to hospital routine and treatment program 57. _____ VIII _____ Has much difficulty in speaking 58. _____ III _____ Sometimes talks out loud to himself 59. _____ _____ IX Chats with other patients 60. _____ I _____ Is shaved by someone else 61. _____ II _____ Seems to resent it when asked to do things 62 _____ _____ I Dresses without any help or supervision 63. _____ II _____ Is often demanding 64. _____ IX _____ When left alone, sits and does nothing 65. _____ VII _____ Says others are jealous of him 66. _____ III _____ Is confused 67. _____ _____ Is blind or practically blind, even with glasses 68. _____ _____ I Decides things for himself, like what to wear, items from canteen (or canteen cart), etc. (continued) 537 538 Measuring Health Exhibit 10.2 (continued) Yes No (Check either Yes or No) 69. _____ II _____ Swears; uses vulgar or obscene words 70. _____ III _____ When you try to get his attention, acts as though lost in a dream world 71. _____ IV _____ Looks worried and sad 72. _____ III _____ Most people would think him a mental patient 73. _____ _____ I Shaves without any help or supervision, other than being given supplies 74. _____ II _____ Yells at people when he’s angry or upset 75. _____ I _____ Is dressed or has his clothes changed by someone 76. _____ X _____ Gets own tray and takes it to eating place 77. _____ III _____ Is watched closely so he doesn’t wander Reproduced from the Physical and Mental Impairment-of-Function Evaluation form obtained from Dr. Lee Gurel. With permission. ration” (including mental confusion, with- The Functional Assessment drawal/apathy, deterioration in behavior and ap- Inventory pearance) and “psychological agitation” (Nancy M. Crewe and Gary T. (including paranoia and suspicion, irritability Athelstan, 1981, Revised 1984) and belligerence, anxiety/depression) (1). The sample was grouped into those scoring Purpose high or low on each of 19 measures reflecting di- The Functional Assessment Inventory (FAI) was agnostic and severity ratings. All the PAMIE developed for clinical use to describe a patient’s scores discriminated between contrasting sub- potential for vocational rehabilitation. It sum- groups in at least three cases. Scores reflecting marizes functional limitations and the personal physical abilities (especially bedfastness, self- and environmental resources that a patient can care, ambulation) provided significant discrimi- use to help cope with problems (1). It was in- nations between almost all of the criterion tended to be applicable to all types of disability. dichotomies (1). Conceptual Basis The FAI is the first component of a two-part Commentary Functional Assessment System. The second part The PAMIE is a relatively old scale that has not is a goal attainment scaling instrument called the been widely used, but it has the advantage of an “Rehabilitation Goals Identification Form” that identifiable internal structure, broad scope, and measures treatment outcomes (1). relevance for institutionalized elderly patients—a The FAI identifies a person’s strengths and group for whom all too often only ADL ques- limitations (whether modifiable or not) that pre- tions are used. dict ability to return to work and so should be taken into account in developing a vocational re- References habilitation plan. “Pinpointing the obstacles to rehabilitation can be helpful in determining (1) Gurel L, Linn MW, Linn BS. Physical and what services are needed even if no attempt will Mental Impairment-of-Function Evaluation be made to directly modify the limitations” (1, in the aged: the PAMIE scale. J Gerontol 1972;27:83–90. p304). Crewe and Turner noted that existing (2) Goga JA, Hambacher WO. Psychologic physical assessment methods such as the Barthel and behavioral assessment of geriatric or PULSES scales are not broad enough to assess patients: a review. J Am Geriatr Soc the potential for vocational rehabilitation: “The 1977;25:232–237. vocational counselor . . . requires general infor-  General Health Status and Quality of Life 539 mation about capacities in a wide variety of Exhibit 10.3. Note that the exhibit refers to a re- physical, emotional, intellectual and social areas vised version, which differs from that published that may be relevant to work” (2, p1). by Crewe and Athelstan (1, p301), two items having been substituted and the order of the Description items changed. Printed copies of the FAI, an in- The FAI is a rating scale of 30 items describing struction sheet, and an interviewer manual are functional limitations and a ten-item checklist of available (see Address section). Full definitions assets or unusual strengths. The strength items of each level are given in the questionnaire and are rated as present or absent and are meant to the accompanying instruction sheet. As an ex- accommodate the instances when a particular as- ample, item 25 reads as follows: set may compensate for a patient’s limitations: considering a patient’s strengths may improve 25. Skills (See instructions) the prediction of success in vocational rehabili- 0. No significant impairment. tation. The functional limitations questions are 1. No available skills that are job-specific. rated on four-point scales representing current However, possesses general skills (i.e., levels of impairment (with aids when used): educational or interpersonal) that could none, mild, moderate, and severe. The FAI iden- be used in a number of jobs. tifies problems that may improve following reha- 2. Has few general skills. Job-specific skills bilitation. Space does not permit showing the are largely unusable due to disability or complete inventory; the item topics are shown in other factors. Exhibit 10.3 The Functional Assessment Inventory: Summary of Ratings Cognition 22. Access to job opportunities 1. Learning ability 23. Need for specialized placement or 2. Ability to read and write accommodations 3. Memory Adaptive behavior 4. Spatial and form perception 24. Work habits Vision 25. Social support system 5. Vision 26. Accurate perception of capabilities Communication & limitations 6. Hearing 27. Effective interaction with 7. Speech employers and co-workers 8. Language functioning 28. Judgment 29. Congruence of behavior with Motor function rehabilitation goals 9. Upper extremity function 30. Initiative and problem-solving 10. Hand function ability 11. Motor speed The strength items include exceptional assets in the Physical condition following areas: 12. Mobility 31. Physical appearance 13. Capacity for exertion 32. Personality 14. Endurance 33. Intelligent or has verbal fluency 15. Loss of time from work 34. Vocational skill in demand 16. Stability of condition 35. Suitable educational qualifications Vocational qualifications 36. Supportive family 17. Work history 37. Financial resources 18. Acceptability to employers 38. Vocational motivation 19. Personal attractiveness 39. Job available with employer 20. Skills 40. Initiative and problem-solving 21. Economic disincentives ability Adapted from the Functional Assessment Inventory obtained from Dr. Nancy M Crewe. Copyright University of Minnesota. With permission. 540 Measuring Health 3. Has no job-specific skills and has very relatively constant when responses from sub- few general or personal skills transfer- groups of the 351 patients with different types able to a job situation. of disability were analyzed separately, and analy- ses of other samples produced similar results (2). The instruction sheet adds: Concurrent validity was assessed by compar- ing the FAI limitation and assets scores with This item refers to skills which the individual judgments made by rehabilitation counselors possesses after onset of disability. concerning employability and severity of disabil- ity. (Note that these do not represent indepen- The FAI is completed by a rehabilitation dent ratings from the FAI responses, because counselor using information available from in- they were recorded by the same rater.) For the terviews, observations of the patient, and mate- 351 patients, the correlations with the FAI limi- rial drawn from medical records to rate the tation score were −0.61 for employability and patient. The ratings concern observable behav- 0.60 for severity of disability (1, p303). For the ior; problems that can only be inferred (e.g., sample of 1,716 cases, equivalent correlations pain, low self-esteem) are excluded (2). The FAI were −0.57 and 0.55 (3). As might be expected, takes five minutes to complete. A total, un- the correlation between the FAI strength scores weighted functional limitation score is provided and the counselor rating of employability (0.53) by adding the raw scores for each item. Alterna- was higher than that with disability (−0.21) (1, tively, scores may be provided for the seven sec- p303). In a multiple regression analysis, the total tions indicated in Exhibit 10.3. functional limitation score plus the total strength The original version of the FAI (as shown in score gave a multiple correlation coefficient of reference 1) was field tested in three studies, first 0.70 (R2 = 0.49) with the employability rating. on 351 physically or mentally disabled patients The R2 value for predicting severity of disability who were assessed by one of 30 vocational reha- was 0.41 (1, p304). bilitation counselors (1). Later it was tested on The third sample of 1,488 patients was di- 1,716 vocational rehabilitation patients, and vided into those who were admitted for rehabili- subsequently on 1,488 patients representing six tation services, those excluded as being too types of disability: visual, hearing, orthopedic, severely disabled, and those excluded as having mental illness, mental retardation, and addiction no impairment. Analyses of variance showed (2). that all FAI scores except for vision and commu- nication distinguished these groups significantly Reliability (2). Strong and logical contrasts were also ob- Alpha internal consistency coefficients were cal- tained between different diagnostic groups (2). culated for five subscales of the questionnaire for 351 patients; the resulting values ranged from Alternative Forms 0.70 to 0.85 (1). To assess inter-rater reliability, A self-administered version called the “Personal a series of 51 interviews was observed and rated Capabilities Questionnaire” has been proposed. by pairs of psychologists. Seventy-five percent of This provides the counselor with information on the ratings made by the pairs of observers were how the client perceives her own limitations. identical; only 3% of all ratings differed by more than one point on the four-point scales (1, Commentary 2). Note that the FAI developed by Crewe and Athelstan is completely distinct from the instru- Validity ment with the same name developed by Pfeiffer The replies of the 351 patients were factor ana- as a variant of the OARS Multidimensional lyzed, providing eight factors that agree quite Functional Activities Questionnaire. Unfortu- closely with the item placements shown in the nately, the two instruments were developed at exhibit (1, Table 4). The factor structure held the same time and were published in different  General Health Status and Quality of Life 541 journals, which prevented the duplication of (3) Crewe NM, Athelstan GT, Meadows GK. names from being detected until too late. Vocational diagnosis through assessment of The Crewe and Athelstan FAI includes clear functional limitations. Arch Phys Med documentation and instructions. It is innovative Rehabil 1975;56:513–516. in incorporating both limitations and assets, a welcome and unusual approach in assessing dis- The Functional Living Index—Cancer ability that would seem to have considerable po- (H. Schipper, 1984) tential benefit. This approach was subsequently used in Hébert’s Functional Measurement Sys- Purpose tem (SMAF), reviewed in Chapter 3. Like Schipper et al. developed the Functional Living Rosser’s Disability and Distress Scale, the FAI can Index—Cancer (FLIC) to determine the response be completed from existing information, so the of cancer patients to their illness and treatment, patient does not necessarily have to be present. and they proposed it as an adjunct to clinical as- The promising validity and reliability evi- sessments of progress and toxicity in clinical tri- dence was based on large samples of patients. als (1; 2). Because the FAI is intended to assess the poten- tial for vocational rehabilitation, the results of Conceptual Basis predictive validity testing are of particular im- Schipper argued for the relevance of quality of portance, although the preliminary evidence for life assessments in the absence of a prospect for predictive validity is not strong. Crewe and cure. Palliation implies the preservation of qual- Turner reported results from 255 of the 351 pa- ity of life, but the hospice movement has focused tients: the data suggest that only about one half on pain, which forms only one part of the qual- of the FAI items correlated with rehabilitation ity of life concept (2, p1120). Schipper notes: outcome scores, and the total FAI score did not significantly predict outcome (2). Nonetheless, Clinical trials analysts measure tumor size, the FAI seems worth considering as a clinical rat- disappearance, reappearance, and survival. ing of rehabilitation potential on the basis of the Patients measure quality of life. What is im- quality of its documentation and its validity and portant to the scientist, the very numbers by reliability results. which he judges the success of his therapy and plans the next steps, may not be relevant Address to patients. . . . patients frequently prefer Manual and scoring information are available function-preserving treatments to more radi- from Materials Development Center of the Stout cal curative attempts, even at the expense of Vocational Rehabilitation Institute, University of survival duration. (2, pp1116–1117) Wisconsin—Stout, Menomonie, Wisconsin, U.S.A., 54751 www.svri.uwstout.edu/. Schipper identified four components of qual- ity of life: physical/occupational function, psy- References chological state, sociability, and somatic (1) Crewe NM, Athelstan GT. Functional discomfort (2, p1117). These form the basis of assessment in vocational rehabilitation: a the FLIC. systematic approach to diagnosis and goal setting. Arch Phys Med Rehabil Description 1981;62:299–305. Items in the FLIC were selected from previous (2) Crewe NM, Turner RR. A functional assessment system for vocational instruments by a panel of patients and health rehabilitation. In: Halpern AS, Fuhrer MJ, professionals in a series of stages (1, p473). The eds. Functional assessment in final 22 items are shown in Exhibit 10.4. rehabilitation. Baltimore: Paul H Brookes, The FLIC is intended for inpatients and out- 1984:223–238. patients with diagnosed malignant cancer. It can Exhibit 10.4 The Functional Living Index—Cancer 1. How well do you appear today? 1 2 3 4 5 6 7 Extremely Extremely Poor Well 2. Rate your confidence in your prescribed course of treatment. 1 2 3 4 5 6 7 No Very Confidence Confident 3. How much of your pain or discomfort over the past 2 weeks was related to your cancer? 1 2 3 4 5 6 7 None All 4. Rate how willing you were to see and spend time with friends, in the past 2 weeks. 1 2 3 4 5 6 7 Unwilling Very Willing 5. Rate the degree to which you are frightened of the future. 1 2 3 4 5 6 7 Constantly Not Terrified Afraid 6. How much nausea have you had in the past 2 weeks? 1 2 3 4 5 6 7 None A Great Deal 7. Rate how willing you were to see and spend time with those closest to you, in the past 2 weeks. 1 2 3 4 5 6 7 Unwilling Very willing 8. How much of your usual household tasks are you able to complete? 1 2 3 4 5 6 7 All None 9. Rate the degree to which your cancer has imposed a hardship on you (personally) in the past 2 weeks. 1 2 3 4 5 6 7 Tremendous No Hardship Hardship 10. How much is pain or discomfort interfering with your daily activities? 1 2 3 4 5 6 7 Not At All A Great Deal 11. Rate in your opinion, how disruptive your cancer has been to those closest to you in the past 2 weeks. 1 2 3 4 5 6 7 Totally No Disruptive Disruption 542 Exhibit 10.4 12. How uncomfortable do you feel today? 1 2 3 4 5 6 7 Not At All Very Uncomfortable 13. Rate your satisfaction with your work and your jobs around the house in the past month. 1 2 3 4 5 6 7 Very Very Dissatisfied Satisfied 14. Rate how often you feel discouraged about your life. 1 2 3 4 5 6 7 Always Never 15. Rate the degree to which your cancer has imposed a hardship on those closest to you in the past 2 weeks. 1 2 3 4 5 6 7 No Tremendous Hardship Hardship 16. Do you feel well enough to make a meal or do minor household repairs today? 1 2 3 4 5 6 7 Very Able Not Able 17. How well do you feel today? 1 2 3 4 5 6 7 Extremely Extremely Poor Well 18. Has nausea affected your daily functioning? 1 2 3 4 5 6 7 Not At All A Great Deal 19. Rate your ability to maintain your usual recreation or leisure activities. 1 2 3 4 5 6 7 Able Unable 20. How much time do you spend thinking about your illness? 1 2 3 4 5 6 7 Constantly Never 21. How well are you coping with your everyday stress? 1 2 3 4 5 6 7 Not well Very Well 22. Most people experience some feelings of depression at times. Rate how often you feel these feelings. 1 2 3 4 5 6 7 Never Continually Reproduced from Schipper H, Levitt M. Measuring quality of life: risks and benefits. Cancer Treatment Reports 1985;69:1121. With permission. 543 544 Measuring Health be self-administered in less than ten minutes. Health Questionnaire ranged between 0.72 and The questions refer to the past two to four 0.77; correlations with the Karnofsky Scale were weeks. The response scales were intended as vi- 0.62 and 0.69 in the two samples; correlations sual analogue scales (VAS) but were divided into with the Spielberger anxiety scales ranged from six categories; patients are instructed to mark 0.55 to 0.60; and correlations with the McGill the line at the point that best reflects their re- Pain Questionnaire were 0.55 and 0.59. Corre- sponse (1, p474). Because of the format, how- lations with the Katz Index of ADL were lower: ever, patients often merely circle one of the 0.17 and 0.31 in the two samples. Ganz et al. numbers (3). For scoring, each interval is divided found a correlation of only 0.33 between the in half and responses are scored to the nearest FLIC total score and the Karnofsky Scale (3, Fig- whole integer. Scores for questions 3, 6, 8, 10, ure 3). The FLIC has been compared with the 12, 15, 16, 18, 19, and 22 are reversed so that Functional Assessment of Cancer Therapy higher scores indicates better health. Schipper et (FACT) scale in several studies. Correlations in- al. recommend using the total score on the FLIC, clude 0.58 (7, Table 5), 0.75 (8, Table 3), and rather than the factor scores (1, p482). 0.74 (9, p23). Gonin et al. provide a chart that establishes the equivalence between scores on the Reliability FACT and the FLIC (9, Table 2). The FLIC has Morrow et al. reported alpha internal consis- been compared to the Short-Form-36 health sur- tency figures for factor scores they derived; these vey (5; 10). Convergent correlations were mod- ranged from 0.64 to 0.87 (4, p293). Figures erate, ranging from 0.26 to 0.54 (5, Table 5), ranged from 0.59 (pain) to 0.90 (overall health) and from 0.50 to 0.62 (10, Table 3). in a study of patients undergoing chemotherapy Schipper and Levitt reported correlations be- (5, Table 5). tween the factor scores and criterion measures in two studies. Correlations of the physical factor Validity score and the Katz Index of ADL were 0.21 and Schipper reported factor analyses from four 0.23; correlations with the McGill Pain Ques- samples, using slightly different versions of the tionnaire scores were 0.56 and 0.66; correlations questionnaire. Four factors were consistently with the Karnofsky score were 0.76 and 0.55 (2, identified, three of which corresponded to the Table 1). As expected, the FLIC physical score components postulated: physical well-being, correlated more highly with the somatic symp- emotional state, and sociability. The final factor toms and social dysfunction scores from the reflected hardship and disruption due to cancer General Health Questionnaire than did the FLIC (1, Tables 1–4). A five-factor solution was re- emotional factor, whereas the emotional factor ported by Morrow et al., including physical, results correlated more highly than the physical emotional, and social functioning; current well- with the anxiety and depression scores. Curi- being; and a factor including the symptoms of ously, however, the physical score correlated pain and nausea (4, Table 2). more highly (0.57 to 0.64) with the Beck De- A preliminary version of the questionnaire pression Inventory than the emotional factor discriminated significantly between patients in scores did (0.52 to 0.53) (2, Table 1). Factors the hospital, getting active treatment, getting ad- three and four correlated “only weakly with the juvant therapy, off treatment, and receiving validation tests” (1, p478). follow-up observation (1, p475). Average scores In a study of chemotherapy for patients with fell significantly with the extent of the disease. breast cancer, effect sizes for the FLIC (0.85 and The FLIC has been shown sensitive to the ad- 0.91) were comparable with those obtained us- verse effects of chemotherapy (6). ing the EuroQol EQ-5D. However, a single VAS The overall scores on the FLIC were corre- score showed an effect size that was as high, or lated with other health measurements for two higher (1.31 and 0.85) (11, Table 2). Evidence samples (1, Table 6). Correlations with the Beck for the FLIC’s responsiveness to change is sum- Depression Inventory and with the General marized by Clinch (12, pp221–223).  General Health Status and Quality of Life 545 Alternative Forms References The FLIC has been translated into most Europe- (1) Schipper H, Clinch J, McMurray A, et al. an languages as well as some Asian (13) and Measuring the quality of life of cancer African ones (12, p218). A French version of the patients: the Functional Living Index- FLIC has been tested for reliability and validity Cancer: development and validation. J Clin (14). The alpha coefficient was 0.90; a principal Oncol 1984;2:472–483. components analysis identified five factors. (2) Schipper H, Levitt M. Measuring quality of A measure known as the Functional Living In- life: risks and benefits. Cancer Treat Rep dex—Emesis (FLIE) was based on the FLIC and 1985;69:1115–1123. assesses the impact of chemotherapy on inducing (3) Ganz PA, Haskell CM, Figlin RA, et al. nausea (12, p221). An abbreviated version, called Estimating the quality of life in a clinical trial of patients with metastatic lung cancer the Quick–FLIC, has been developed in English using the Karnofsky performance status and in Chinese in Singapore (15; 16). Alpha was and the Functional Living Index-Cancer. 0.86 and it was responsive to change (16). Cancer 1988;61:849–856. (4) Morrow GR, Lindke J, Black P. Measure- Commentary ment of quality of life in patients: In the mid-1980s, the FLIC caught the attention psychometric analyses of the Functional of cancer researchers searching for a quality of Living Index-Cancer (FLIC). Qual Life Res life instrument; it was frequently mentioned in 1992;1:287–296. the literature. Although the coverage of the scale (5) Kuenstner S, Langelotz C, Budach V, et al. is broad and care was taken in its initial develop- The comparability of quality of life scores: ment, the FLIC has seen only limited use. a multitrait multimethod analysis of the EORTC QLQ-C30, SF-36 and FLIC There may be several reasons for this. Since questionnaires. Eur J Cancer its initial testing, the scale has received little fur- 2002;38:339–348. ther development; we still lack extensive infor- (6) Lindley CM, Hirsch JD, O’Neill CV, et al. mation on reliability. The validity findings are Quality of life consequences of varied; Schipper reported higher validity than chemotherapy-induced emesis. Qual Life Ganz, for example. Several validity results are Res 1992;1:331–340. based on earlier versions of the scale. The use of (7) Cella DF, Bonomi AE, Lloyd SR, et al. a single overall score has been criticized as being Reliability and validity of the Functional too coarse to measure change (17, p187), and Assessment of Cancer Therapy-Lung this formed one stimulus for the development of (FACT-L) quality of life instrument. Lung Cancer 1995;12:199–220. the European Organization for Research and (8) Ward WL, Hahn EA, Mo F, et al. Treatment of Cancer (EORTC) QLQ-C30 (18, Reliability and validity of the Functional p80). The interpretation of some questions in Assessment of Cancer Therapy-Colorectal the FLIC is unclear. For example, in the light of (FACT-C) quality of life instrument. Qual using mental imagery as an adjunct in cancer Life Res 1999;8:181–195. therapy, what is the correct answer to “How (9) Gonin R, Lloyd S, Cella D. Establishing much time do you spend thinking about your ill- equivalence between scaled measures of ness?” The emphasis on function in the title of quality of life. Qual Life Res the scale also does not seem to accurately repre- 1996;5:20–26. sent its content. Ganz et al. concluded that “The (10) Wilson RW, Hutson LM, VanStry D. FLIC should be evaluated more extensively in Comparison of 2 quality-of-life questionnaires in women treated for breast the clinical trial setting to assess its utility in ac- cancer: the RAND 36-item health survey curately assessing QOL” (3, p855). More recent and the Functional Living Index–Cancer. scales, such as the Functional Assessment of PhysTher 2005;85:851-860. Cancer Therapy or the European EORTC Qual- (11) Conner-Spady B, Cumming C, Nabholtz J- ity of Life Questionnaire, should probably be M, et al. Responsiveness of the EuroQol in considered carefully before using the FLIC. breast cancer patients undergoing high dose 546 Measuring Health chemotherapy. Qual Life Res Conceptual Basis 2001;10:479–486. During the later 1980s, increasing emphasis was (12) Clinch JJ. The Functional Living Index— placed on recording quality of life outcomes in Cancer: ten years later. In: Spilker B, ed. oncology clinical trials (1; 2). Cella argued that Quality of life and pharmacoeconomics in QoL is fundamentally subjective and can only be clinical trials. Philadelphia: Lippincott- understood from the patient’s perspective; it can- Raven, 1996:215–225. not be measured by observing a patient’s behav- (13) Goh CR, Lee KS, Tan TC, et al. Measuring quality of life in different cultures: ior. QoL “refers to patients’ appraisal of and translation of the Functional Living Index satisfaction with their current level of function- for Cancer (FLIC) into Chinese and Malay ing compared to what they perceive to be possi- in Singapore. Ann Acad Med Singapore ble or ideal” (1, p9). QoL considers not only 1996;25:323–334. people’s level of function, but also their value (14) Mercier M, Bonneterre J, Schraub S, et al. system, which influences how they react to a loss The development of a French version of a of function. QoL is multidimensional and in- questionnaire on the quality of life in cludes at least physical, functional, emotional, cancerology (Functional Living Index- and social well-being. In addition to these tradi- Cancer: FLIC). Bull Cancer tional dimensions of QoL, Cella includes areas 1998;85:180–186. such as sexuality, treatment satisfaction, and (15) Cheung YB, Khoo KS, Wong ZW, et al. Quick-FLIC—a short questionnaire for spirituality (1, p11). assessing quality of life of cancer patients. The FACT was intended to be broadly appli- Acta Oncol 2003;42:36–42. cable yet sensitive to change following treat- (16) Cheung YB, Khoo KS, Thumboo J, et al. ment; to achieve this, the FACT is conceived as a Validation of the English and Chinese measurement system comprising a core set of versions of the Quick-FLIC quality of life items applicable to all types of cancer plus questionnaire. Br J Cancer disease-specific supplements. The core items are 2005;92:668–672. presented as a profile of scores, rather than as an (17) Aaronson NK, Ahmedzai S, Bullinger M, index, because different therapies might be ex- et al. The EORTC core quality-of-life pected to have a differential impact on different questionnaire: interim results of an dimensions (3, p201). Although developed for international field study. In: Osoba D, ed. Effect of cancer on quality of life. Boca use in patients with cancer, the FACT has been Raton, Florida: CRC Press, 1991:185–203. tested on those with HIV and rheumatoid arthri- (18) Aaronson NK, Cull A, Kaasa S, et al. The tis, as well as in the general population (D. Cella, EORTC modular approach to quality of personal communication, 2004). life assessment in oncology. Int J Mental Health 1994;23:75–96. Description The FACT-G forms the core component in a set of instruments originally developed for assessing The Functional Assessment of QoL outcomes for patients with cancer, but it Cancer Therapy (FACT) has broadened into an assessment system for (David Cella, 1993) Functional Assessment in Chronic Illness Ther- Purpose apy, or FACIT (see www.facit.org). The FACIT The FACT is a patient-assessed measurement organization develops and distributes question- system comprising a core component (the FACT- naires, ensures their standardization, coordinates G) that covers general aspects of quality of life translations, and provides information on ad- (QoL), plus a range of optional condition- ministration, scoring, and interpretation. Since specific subscales. The system is intended for the original description of the FACT-G in 1993, recording outcomes in clinical trials; it focuses FACIT has developed numerous questionnaires mainly on cancer therapy, but is potentially ap- and coordinated their translation. Besides can- plicable to other conditions. cer, subscales have been developed for fatigue,  General Health Status and Quality of Life 547 treatment satisfaction, spiritual well-being, and Reliability for HIV disease, multiple sclerosis, and other Alpha values taken from a range of language chronic conditions. versions and patient samples are shown in Table The FACT-G, described here, contains generic 10.2. assessments to which supplementary disease- Retest reliability was 0.92 for the FACT-G specific modules may be added. Items were gen- total score after three to seven days; coefficients erated through interviews with patients and for the subscales included 0.88 (PWB), 0.84 oncologists, and the process of test development (FWB), 0.82 (EWB and SWB), and 0.82 for is described by Cella et al. (2; 4). Based on the re- RWD (4, p208). Very similar results were re- sults of factor analyses, the items were grouped ported in another study by Cella et al. (2, p575). into four subscales early in the development pro- Retest figures for a Japanese version (two-week cess: physical well-being (PWB, seven items, delay) were lower, at 0.72 for the overall score, score range, 0–28); social and family well-being with subscale values ranging from 0.63 (SWB (SWB, seven items, 0–28); emotional well-being and RWD) to 0.81 (EWB) (8, Table 3). (EWB, six items; 0–24) and functional well-being (FWB, seven items, 0–28). The answer categories Validity use a Likert format (0 = “not at all” to 4 = “very During the initial test development, Cella et al. much”) and allow for administration by tele- reported results of a factor analysis that pro- phone (2, p577). A total score sums the subscale duced six dimensions, of which two were scores. The FACT-G can be completed in five merged (2, Table 1). minutes (2, p575), although a questionnaire ver- Kemmler et al. applied multidimensional sion took an average of 13.5 minutes to adminis- scaling analysis to the FACT-G and showed that ter in a population with low educational levels the items ‘enjoy life,’ ‘enjoy things,’ and ‘general (5). A comparison of several ways to adjust for quality of life’ were central; most items measur- missing data suggested that the best method is to ing other domains clustered around each other, replace a missing item by the mean score on its although the emotional and physical domains subscale, as long as at least 50% of items were overlapped. Three items (‘feeling distant from answered on that subscale (6). friends,’ ‘coping with the illness,’ and ‘being able Version 4 of the FACT is shown in Exhibit to work’) did not fit their respective clusters (13, 10.5, although much of the reliability and valid- pp228–229). ity evidence for the instrument was collected us- Convergent validity correlations for the ing Version 3. A description of the evolution of FACT are often high. Cella’s original validation the FACT, and a copy of Version 3, are available compared the FACT with a range of other scales; in Cella and Bonomi (4); Version 2 is shown in coefficients included 0.79 with the Functional reference (2). Version 3 contained 34 items, Living Index-Cancer (FLIC); −0.68 with the compared with 27 in the current version. The brief version of the Profile of Mood States additional items included a two-item scale on (POMS), and −0.58 with the Taylor Manifest Relationship With the Doctor (RWD: “I have Anxiety Scale (TMAS) (2, Table 3). The PWB confidence in my doctor(s)” and “My doctor is scale correlated −0.45 with the Eastern Coopera- available to answer my questions”); this was ex- tive Oncology Group performance rating panded into the subscale covering treatment sat- (ECOG) (5, Table 3). High convergent validity isfaction. Version 3 also contained a summary correlations with subscales of the POMS have item following each section that asked respon- been reported: −0.73 between the EWB and the dents to rate (on a 10-point scale) how much POMS depression score and −0.62 with the fa- that area of function affected their overall qual- tigue score (7, Table 4). The correlation between ity of life. These ratings did not prove useful in FWB and the POMS vigor scale was 0.71, the context of clinical trials and so were dis- whereas the correlation between POMS and carded. Finally, several items on the social and FACT total scores was 0.62. Ward et al. reported emotional scales were reworded in Version 4. correlations with the POMS from three samples, Exhibit 10.5 The Functional Assessment of Cancer Therapy (FACT-G), Version 4 Below is a list of statements that other people with your illness have said are important. By circling one (1) number per line, please indicate how true each statement has been for you during the past 7 days. Physical Well-being Not A little Some- Quite Very at all bit what a bit much I have a lack of energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I have nausea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 Because of my physical condition, I have trouble meeting the needs of my family . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I have pain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am bothered by side effects of treatment . . . . . . . . . . . . . . . . 0 1 2 3 4 I feel ill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am forced to spend time in bed . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 Social/Family Well-being Not A little Some- Quite Very at all bit what a bit much I feel close to my friends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I get emotional support from my family . . . . . . . . . . . . . . . . . 0 1 2 3 4 I get support from my friends . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 My family has accepted my illness . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am satisfied with family communication about my illness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I feel close to my partner (or the person who is my main support) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 Regardless of your current level of sexual activity, please answer the following question. If you prefer not to answer it, please check this box and go to the next section. I am satisfied with my sex life . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 By circling one (1) number per line, please indicate how true each statement has been for you during the past 7 days. Emotional Well-being Not A little Some- Quite Very at all bit what a bit much I feel sad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am satisfied with how I am coping with my illness . . . . . . . . 0 1 2 3 4 I am losing hope in the fight against my illness . . . . . . . . . . . . 0 1 2 3 4 I feel nervous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I worry about dying . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I worry that my condition will get worse . . . . . . . . . . . . . . . . . 0 1 2 3 4 Functional Well-being Not A little Some- Quite Very at all bit what a bit much I am able to work (include work at home) . . . . . . . . . . . . . . . . 0 1 2 3 4 My work (include work at home) is fulfilling . . . . . . . . . . . . . . 0 1 2 3 4 I am able to enjoy life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I have accepted my illness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am sleeping well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 3 4 I am enjoying the things I usually do for fun . . . . . . . . . . . . . . 0 1 2 3 4 I am content with the quality of my life right now . . . . . . . . . . 0 1 2 3 4 From Dr. D. Cella, with permission. 548  General Health Status and Quality of Life 549 Table 10.2 Cronbach Alpha Coefficients for the FACT-G Scales from a Range of Language Versions and Patient Samples Physical Social Relationship Emotional Functional Well-Being Well-Being With Doctor Well-Being Well-Being Total Score (PWB) (SWB) (RWD) (EWB) (FWB) (FACT-G) Reference 0.82 0.74 0.75 0.66 0.83 0.89 (7, Table 3) 0.81 0.55 0.76 0.74 0.79 0.85 (8, Table 3) 0.82 0.69 0.65 0.74 0.80 0.89 (2, Table 2) 0.75 0.56 0.40 0.69 0.80 0.87 (3, Table 2) 0.80 0.68 0.62 0.73 0.79 — (6, Table 1) 0.81 0.69 — 0.69 0.86 0.90 (9) 0.75 0.63 — 0.84 0.84 0.90 (10, Table 1) 0.80 0.71 — 0.75 0.79 0.88 (11, Table 2) 0.65 0.75 — 0.70 0.75 0.82 (5, Table 2) 0.79 0.53 — 0.67 0.79 0.84 (12) Note: The first five entries use the third version of the FACT which contained the RWD scale; the remainder use version 4. including −0.72 (with the brief POMS), −0.54 53 language versions (and counting . . . ). An and −0.66 (with the short-form of the POMS) impressive systematic translation project has (11, Table 3). Similarly, the correlation between been described (3; 16), often taking the form of PWB and the ECOG score was −0.64 (7, Table thorough cultural adaptations (7). Aside from 4). Other associations with the ECOG include European languages, versions include three sepa- −0.43 (3, Table 5). Correlations with the FLIC rate South African languages (17), Malayalam include 0.58 (3, Table 5), 0.80 (4, p207), 0.75 (10), and a couple more I had never heard of. (11, Table 3), and 0.74 (14, p23). Gonin et al. Supplementary modules include a nine-item provided a chart showing the equivalence be- breast cancer module (FACT-B) (9), prostate (13 tween raw scores on the FACT and the FLIC (14, items), colorectal (FACT-C, 9 items) (11), blad- Table 2). der (12 items); head and neck (FACT-H&N, 9 FACT-G subscores distinguished significantly items), lung cancer (FACL-L, 10 items) (3; 18), between patients in different stages of their ill- fatigue and anemia (Fact-F, 13 items and FACT- ness, as defined by the NCI criteria (2, Table 4). An, 7 items) (19). There is also a subscale that The physical, functional, and emotional sub- identifies side-effects of biological response scales distinguished significantly between pa- modifiers: the FACT-BRM (20). Updates to tients classified by the ECOG performance these subscales are provided on the FACIT rating in terms of whether they changed over website http://www.facit.org/validity/validation_ time (2, p575). The FACT’s physical, functional, articles.aspx and emotional scales, and the overall score, cor- The Functional Assessment of Multiple Scle- responded to independent patient-ratings of rosis (FAMS) was based on the FACT (21). meaningful change (4, Table 3; 15). Effect sizes have been reported (15, Tables 5–9). In three Reference Standards samples of patients, the PWB, FWB and overall Information on population norms for the FACT scores were consistently related to changes in pa- can be obtained (for a fee) from the validity sec- tient condition and severity of illness (11, Tables tion of the FACIT website. 4–7). Commentary Alternative Forms The FACT provides a good example of a care- The FACT-G may win the Measuring Health fully developed instrument, supported by a translation sweepstakes: the FACIT web site lists strong team that takes an organizational ap- 550 Measuring Health proach to promoting the FACT. The FACIT or- Philadelphia: Lippincott-Raven, ganization has an informative Web site, appears 1996:203–214. welcoming (unlike some agencies mentioned in (5) Pratheepawanit N, Phunmanee A, Chapters 6 and 7), and provides ready access to Sookprasert A, et al. Quality of life in Thai the instrument. The FACT follows the core-plus- cancer patients; validation of an interview- administered FACT-G. Quality of Life disease-specific-module approach also used by Newsletter 2002;29:17–18. the European Organisation for Research and (6) Fairclough DL, Cella DF. Functional Treatment of Cancer in their Quality of Life Assessment of Cancer Therapy (FACT-G): Questionnaire, reviewed later in this chapter. non-response to individual questions. Qual The range of condition-specific modules (and es- Life Res 1996;5:321–329. pecially the effort devoted to translation) is im- (7) Cella D, Hernandez L, Bonomi AE, et al. pressive and should make the FACT a leading Spanish language translation and initial choice as an outcome measure in clinical trials. validation of the Functional Assessment of During its first ten years, the FACT-G has Cancer Therapy quality-of-life instrument. seen a rapid evolution through four published Med Care 1998;36:1407–1418. versions, and most published psychometric evi- (8) Fumimoto H, Kobayashi K, Chang C-H, et dence comes from the original development al. Cross-cultural validation of an international questionnaire, the general team. As it enters its second decade, it is hoped measure of the Functional Assessment of the format of the instrument will stabilize and Cancer Therapy scale (FACT-G), for we will begin to see reliability and validity re- Japanese. Qual Life Res 2002;11:701–709. ports by independent researchers (who always (9) Brady MJ, Cella DF, Mo F, et al. seem to obtain less glowing results than the orig- Reliability and validity of the Functional inal team). It will also be useful to add popula- Assessment of Cancer Therapy-Breast tion norms and more evidence on effect sizes and quality-of-life instrument. J Clin Oncol sensitivity to change. 1997;15:974–986. (10) Pandey M, Thomas BC, Ramdas K, et al. Quality of life in breast cancer patients: Address validation of a FACT-B Malayalam The FACIT system is described at version. Qual Life Res 2002;11:87–90. http://www.facit.org/. (11) Ward WL, Hahn EA, Mo F, et al. Reliability and validity of the Functional Assessment of Cancer Therapy-Colorectal References (FACT-C) quality of life instrument. Qual (1) Cella DF. Quality of life: the concept. J Life Res 1999;8:181–195. Palliat Care 1992;8:8–13. (12) Novik AA, Ionova TI, Fedorenko DA, et al. (2) Cella DF, Tulsky DS, Gray G, et al. The Sensitivity of FACT-G in quality of life functional assessment of cancer therapy assessment of lung cancer patients after scale: development and validation of the radical surgery. Quality of Life Newsletter general measure. J Clin Oncol 2000;24:12. 1993;11:570–579. (13) Kemmler G, Holzner B, Kopp M, et al. (3) Cella DF, Bonomi AE, Lloyd SR, et al. Multidimensional scaling as a tool for Reliability and validity of the Functional analysing quality of life data. Qual Life Assessment of Cancer Therapy-Lung Res 2002;11:223–233. (FACT-L) quality of life instrument. Lung (14) Gonin R, Lloyd S, Cella D. Establishing Cancer 1995;12:199–220. equivalence between scaled measures of (4) Cella DF, Bonomi AE. The Functional quality of life. Qual Life Res 1996;5:20–26. Assessment of Cancer Therapy (FACT) and (15) Cella D, Hahn EA, Dineen K. Meaningful Functional Assessment of HIV Infection change in cancer-specific quality of life (FAHI) quality of life measurement system. scores: differences between improvement In: Spilker B, ed. Quality of life and and worsening. Qual Life Res pharmacoeconomics in clinical trials. 2002;11:207–221.  General Health Status and Quality of Life 551 (16) Bonomi AE, Cella DF, Hahn EA, et al. group initiated a program to develop an outcome Multilingual translation of the Functional measurement system for cancer trials; the con- Assessment of Cancer Therapy (FACT) sortium that developed the QLQ has expanded quality of life measurement system. Qual to involve 16 European countries, plus Aus- Life Res 1996;5:309–320. tralia, Canada, and the United States. A QoL (17) Mullin V, Cella D, Chang C-H, et al. unit and data center located in Brussels were Development of three African language funded by the European Commission in 1993. translations of the FACT-G. Qual Life Res 2000;9:139–149. The design requirements specified that the (18) Hollen PJ, Gralla RJ. Comparison of QLQ be relevant to cancer patients, cross- instruments for measuring quality of life in culturally applicable, and sensitive to change af- patients with lung cancer. Semin Oncol ter treatment. To combine the advantages of 1996;23:31–40. generic and disease-specific instruments, the (19) Cella D. The Functional Assessment of measurement system comprises a generic core Cancer Therapy-Anemia (FACT-An) Scale: questionnaire (the QLQ-C30 described here), a new tool for the assessment of outcomes plus disease- and treatment-specific modules. It in cancer anemia and fatigue. Semin was agreed that the core instrument should be Hematol 1997;34(3, suppl 2):13–19. specific to cancer; that it should cover several di- (20) Bacik J, Mazumdar M, Murphy BA, et al. mensions of function, symptoms, and the possi- The functional assessment of cancer therapy-BRM (FACT-BRM): a new tool for ble side-effects of treatment, and that it should the assessment of quality of life in patients be self-administered and brief. The inclusion of treated with biologic response modifiers. a relatively large number of items covering Qual Life Res 2004;13:137–154. symptoms and the side effects of treatment re- (21) Cella DF, Dineen K, Arnason B, et al. flected the clinical focus of the scale (1). Reports Validation of the functional assessment of from the EORTC group discuss the design of multiple sclerosis quality of life instrument. quality of life instruments (4) and give concep- Neurology 1996;47:129–139. tual justification for the dimensions covered (5); guidelines for the development of disease-specific The EORTC Quality of Life modules have also been published (6). Questionnaire (European Description Organization for Research and A first-generation core questionnaire contained Treatment of Cancer (EORTC), 1993) 36 items (1–3); tests in 15 countries led to the Purpose 30-item instrument, which was described in the The Quality of Life Questionnaire (QLQ) is a 1996 edition of this book. The QLQ-C30 core modular system for evaluating the quality of life component includes multiitem scales covering (QoL) outcomes of cancer patients participating five aspects of function that are common to all in international clinical trials. The system in- cancer patients, plus three categories of symp- cludes the core questionnaire described here, toms, and an overall judgment of health status plus diagnosis-specific additional modules. and quality of life. In addition, single items cover common side-effects of cancer and its treatment Conceptual Basis (7). Minor changes were made in creating ver- The European Organization for Research and sion 2, which was described in 1997 (8), Treatment of Cancer (EORTC), one of the whereas the version 3 presented here was pub- largest clinical trials groups in Europe, was lished in 2000. This retained the same items but founded in 1962 and began to work on QoL broadened the role activities questions and measures in 1980 (1; 2). Clinicians and behav- placed less weight on physical function in the ioral scientists jointly worked on trial design and overall QoL rating. It also modified some of on how best to define and measure QoL (3). Af- the answer scales (9). The current version of the ter reviewing existing instruments, the study questionnaire can be downloaded from the Exhibit 10.6 The EORTC Quality of Life Questionnaire (QLQ-C30), Version 3.0 We are interested in some things about you and your health. Please answer all of the questions yourself by circling the number that best applies to you. There are no “right” or “wrong” answers. The information that you provide will remain strictly confidential. Not at A Quite Very All Little a Bit Much 1. Do you have any trouble doing strenuous activities, like 1 2 3 4 carrying a heavy shopping bag or a suitcase? 2. Do you have any trouble taking a long walk? 1 2 3 4 3. Do you have any trouble taking a short walk outside of the house? 1 2 3 4 4. Do you need to stay in bed or a chair during the day? 1 2 3 4 5. Do you need help with eating, dressing, washing yourself 1 2 3 4 or using the toilet? During the past week: Not at A Quite Very All Little a Bit Much 6. Were you limited in doing either your work or other daily activities? 1 2 3 4 7. Were you limited in pursuing your hobbies or other leisure 1 2 3 4 time activities? 8. Were you short of breath? 1 2 3 4 9. Have you had pain? 1 2 3 4 10. Did you need to rest? 1 2 3 4 11. Have you had trouble sleeping? 1 2 3 4 12. Have you felt weak? 1 2 3 4 13. Have you lacked appetite? 1 2 3 4 14. Have you felt nauseated? 1 2 3 4 15. Have you vomited? 1 2 3 4 16. Have you been constipated? 1 2 3 4 17. Have you had diarrhea? 1 2 3 4 18. Were you tired? 1 2 3 4 19. Did pain interfere with your daily activities? 1 2 3 4 20. Have you had difficulty in concentrating on things, 1 2 3 4 like reading a newspaper or watching television? 21. Did you feel tense? 1 2 3 4 22. Did you worry? 1 2 3 4 23. Did you feel irritable? 1 2 3 4 24. Did you feel depressed? 1 2 3 4 25. Have you had difficulty remembering things? 1 2 3 4 26. Has your physical condition or medical treatment 1 2 3 4 interfered with your family life? 27. Has your physical condition or medical treatment 1 2 3 4 interfered with your social activities? 28. Has your physical condition or medical treatment 1 2 3 4 caused you financial difficulties? 552  General Health Status and Quality of Life 553 Exhibit 10.6 For the following questions please circle the number between 1 and 7 that best applies to you 29. How would you rate your overall health during the past week? 1 2 3 4 5 6 7 Very poor Excellent 30. How would you rate your overall quality of life during the past week? 1 2 3 4 5 6 7 Very poor Excellent Copyright 1995 EORTC Quality of Life Group. Reproduction or use of the questionnaire requires prior written approval of the EORTC. Reproduced with permission. EORTC Web site after registering as a user (see with all rights reserved. A user’s agreement must Address section). Version 3.0 is shown in Ex- be signed for its use (see www.eortc.be/home/ hibit 10.6. qol/downloads/); the intention is to standardize The function subscales and items composing the administration of the instrument and to en- each are as follows: physical function, items one able the EORTC to assemble data that can be to five; role function, items six and seven; cogni- used in validation studies. tive function, items 20 and 25; emotional func- tion, items 21 to 24; social function, items 26 Reliability and 27. The symptom scales include fatigue Alpha internal consistency coefficients for the (items 10, 12, and 18), nausea (items 14 and nine scales are reported in Table 10.3. Results 15), and pain (items 9 and 19). The global QoL appear similar across cultural and language scale includes items 29 and 30. The single items groups with the possible exception of the nausea cover dyspnea (item 8), sleep disturbance (item subscale (7). 11), appetite loss (item 13), constipation (item In a sample of patients after completion of 16), diarrhea (item 17), and the financial impact treatment, four-day retest correlations ranged of cancer (item 28). Separate forms are used for from 0.82 (cognitive function scale) to 0.91 men and for women in some language versions. (physical function) (11, p1251). Five-week retest The questionnaire can be self- or interviewer- intraclass correlations (ICC) were 0.33 (social), administered in 11 to 12 minutes (1, p85; 7, 0.54 (emotional), 0.56 (role), 0.75 (physical), p368); a few patients require assistance to com- 0.77 (cognitive), and 0.82 (overall) (12, Table 2). plete it (3, p191). A touch-screen computer ad- Multitrait analyses have confirmed that the ministration has also been developed and appears items fall on the nine hypothesized scales. Mean acceptable to patients (10). An informative 70- item-to-scale correlations were 0.53 and 0.59 in page scoring manual is available on the Web site; two administrations of the questionnaire (7, this includes computer codes to perform the scor- p368). This was replicated in a Norwegian ing (9). A profile of separate scores is produced study; although correlations between the scales for the nine subscales; the raw scores for items were significant (0.36 to 0.67), in only 2% of in- are added, then divided by the number of items stances did an item correlate more highly with in the scale and the result linearly transformed to another scale than with its own (13, p2262). In a 0-to-100 scale, with a higher score representing another multitrait analysis of the original 36- a higher level of function. The single items de- item version and the C30, 96% of comparisons scribing symptoms are scored separately. The showed the convergent coefficients to exceed the QLQ-C30 does not provide an overall score, as a discriminant. The only errors found were in the profile more accurately reflects the multidimen- role-functioning scale (14, p385). sional character of quality of life (7, p373). Modifications have been proposed to the The QLQ-C30 is a copyrighted instrument, composition of certain subscales; Ringdal and 554 Measuring Health Table 10.3 Alpha Internal Consistency Figures for the QLQ-C30 Function Scales Symptom Scales Physical Role Emotional Cognitive Social Global Fatigue Nausea Pain Reference 0.68 0.54 0.73 0.56 0.68 0.86 0.80 0.65 0.82 (7, Table 3) 0.71 0.52 0.80 0.73 0.77 0.89 0.85 0.73 0.76 (7, Table 3) 0.77 0.88 0.84 0.58 0.80 0.84 0.81 0.50 0.83 (15, Table 3) 0.77 0.68 0.80 0.62 0.78 0.88 0.87 0.81 0.89 (13, Table 1) 0.76 — 0.82 0.67 0.79 0.88 0.89 — 0.89 (16, p29) 0.80 0.61 0.78 — 0.81 0.84 0.89 0.67 0.81 (17, Table 2)* 0.75 0.85 0.90 0.51 0.86 0.82 0.77 0.86 0.76 (12, Table 2) * Note: This study used the 36-item version. Ringdal presented alpha values for the original fatigue scales were the highest, ranging from and the modified scales. Values ranged from 0.54 to 0.63 (7, p369). A factor analysis merged 0.55 to 0.86, with eight of the 14 coefficients several of the subscales, distinguishing a strong exceeding 0.80 (18, Table 3). Similarly, Osoba et physical factor and a weaker psychological fac- al. showed that the alpha values shown in the tor, which correlated 0.47 with each other (18, top line of Table 10.3 rose after shifting some of p137). Osoba et al. identified factors that were the items onto different subscales (19, Table 4). broadly similar to the original scale structure, but with some items shifted to different sub- Validity scales (19, Table 4). Using a multidimensional In a study of the clarity of the first version of the scaling analysis in a German sample, Kemmler et QLQ-C30, patients responded to each question al. identified a two-dimensional structure that and were then interviewed to discuss and elabo- separated physical and role function from emo- rate on their answers. An observer based scores tional, cognitive, social, and global domains. on this discussion, blinded to the initial re- Within the latter grouping, items covering social sponse. The median kappa agreement for the and overall QoL tended to cluster together, as items was 0.85 and values for nine items were did those items covering emotional and cognitive 0.9 or higher; three items (i.e., short walk, stay domains (22, p230). The item on help with eat- in bed, financial difficulties) had kappas below ing or dressing was an outlier, largely because of 0.6 (20, Table 2). The poor results for the first its skewed response distribution. A factor analy- two of these items was attributed to the use of sis of the emotional, role, and social items alone two-point response scales, so that these have identified two factors, separating emotional subsequently been changed to four-point scales. from functional items. These showed different Patient- and proxy-reports have been compared, patterns of correlations with the Profile of Mood giving ICCs in the moderate-to-good range States (POMS) subscales (23, Tables 2 and 5). (0.46–0.73) for the various subscales, with the Comparisons of QLQ-C30 scores across di- lowest value for emotional function (21, p622). agnostic groups and by characteristics (e.g., ex- There was a small bias toward proxy respon- tent of weight loss, performance scores, toxicity dents rating patients as more impaired than the ratings), have demonstrated that many, but not patients themselves did, but the proxy ratings all, of these subscales discriminate between appeared to be as responsive to change in the pa- groups (7, Figures 1, 2). Responsiveness to tient’s condition as were the self-ratings (21, change was evaluated by grouping 262 lung can- pp628–629). cer patients into those whose condition im- In the original validation study, correlations proved, deteriorated, or did not change during among the subscales appeared moderate; those treatment on the basis of the Eastern Coopera- among the physical function, role function, and tive Oncology Group performance status scale.  General Health Status and Quality of Life 555 The QLQ-C30 physical, role, fatigue, and nau- ment is capable of demonstrating) were 0.73 for sea scales showed significant contrasts between global QoL, 1.5 for physical function, 0.86 for these groupings (7, Table 5). A study of radio- role function, 1.3 for nausea, 0.91 for pain, and therapy has shown the scales of the QLQ to be 1.2 for fatigue. Effect sizes for the psychosocial capable of demonstrating significant changes be- dimensions were smaller, at 0.52 for emotional, fore and after treatment, but no effect size statis- 0.52 for social, and 0.39 for cognitive (26, Table tics were reported (13, p2262). 8). Osoba et al. reported large differences in Convergent validity correlations with other mean scores before and eight days after starting scales have been reported in several studies. The antiemetic treatment while receiving chemother- QLQ emotional function score correlated 0.71 apy. From the figures reported, effect sizes range with the total score on the Hospital Anxiety and from 16.5 for nausea and 11.5 for fatigue to 3.0 Depression Scale (HADS) (18, p135). A second for emotion and 1.4 for pain (19, Table 8). study reported lower associations: 0.58 with the HADS anxiety scale and 0.41 with the depres- Alternative Forms sion scale (16, Table 4). Niezgoda and Pater re- Although the initial questionnaire was developed ported a wide range of Spearman convergent in English, this was done with the direct input of validity correlations. Correlations of the QLQ collaborators from numerous countries, and the subscales with equivalent scales on the Sickness QLQ-C30 was tested simultaneously in 13 coun- Impact Profile included 0.73 for the physical tries, including most of Western Europe, scale, 0.58 for the cognitive and the fatigue Canada, Australia, and Japan. The instrument is scales, 0.55 for the role scale and 0.48 for the now available in over 60 languages and the emotional and the social scales (24, Table 3). EORTC group offers guidelines for standardiz- Correlations with subscales of the Cancer Reha- ing further translations. Information on the bilitation Evaluation System (CARES) were sim- available versions can be obtained from ilar, including 0.71 between the physical scales, www.qolid.org/public/qlq-c30.html. 0.56 for the emotional, 0.46 for the social, and Since the mid-1990s, the EORTC group has 0.69 for the pain scales (24, Table 4). The QLQ been actively developing and validating disease- emotional scale correlated 0.61 with the General specific modules (1) so that many are now avail- Health Questionnaire overall score (24, Table able. These include modules for brain cancer 5). Finally, the pain subscale correlated 0.57 and (27), head and neck cancers (H&N35) (28; 29), 0.53 with the sensory and present pain scores cancers of the eye (30), esophagus (the OES24 from the McGill Pain Questionnaire (24, Table module) (31–33), lung (LC13) (34), breast 6). In a study of patients with breast cancer, con- (BR23) (35), pancreas (36), liver with metastases vergent validity correlations with the Psychoso- (37), stomach (the QLQ-STO22) (38–40), cial Adjustment to Illness Scale (PAIS) included ovaries (OV28) (41), prostate (42), colon and 0.63 for the QLQ-C30 global QoL score, 0.57 rectum (the CR-38) (43), as well as multiple for the role score, 0.57 for the social score, and myeloma (44), melanoma (45), and hepatocellu- 0.68 for the emotional score (25, Table 5). A lar cancer (46) . There is also a module to evalu- similar comparison with the POMS gave conver- ate patients’ needs for information (47). These gent correlations of 0.76 between the QLQ emo- modules are described in the QLQ manual (9), tional scale and the POMS tension scale, and and on the Web site. They are available in vari- 0.74 with the POMS depression scale; the QLQ ous languages (see www.eortc.be/home/qol/ cognition scale correlated 0.54 with the POMS downloads/). The modules remain the property confusion scale (25, Table 6). of their developers, from whom permission King calculated effect sizes from a range of should be obtained for their use. published clinical trial results (26). Because the A computer-administered version of the effectiveness of the various studies differed, there QLQ-C30 is available (48); a touch-screen sys- was a wide range in the effect size estimates, but tem proved acceptable to patients and give re- the highest values (suggestive of what the instru- sults as valid as the paper version (10). 556 Measuring Health Reference Standards would also appear to be relevant to other groups Reference values have been derived from studies of patients, not only those with cancer. Evidence in several countries. They are available from the for validity and reliability is accumulating, with EORTC group on paper or as a CD-ROM (49). reports from several countries, and the QLQ- The manual discusses the interpretation of C30 has been compared with various other scores (9, p14). By comparing repeated adminis- scales. The results appear strong: correlations trations of the QLQ-C30 with a separate rating with other, established scales are high and follow of perceived changes in health, it has been a logical pattern. Because the scale is intended as shown that patients who reported “a little” an outcome measure, more evidence for its sensi- change in a given aspect of health in fact had tivity to change and effect size would be desir- changes of five to ten points on the correspon- able. Kind (from the perspective of utilities ding QLQ-C30 scale. “Moderate” changes cor- measurement) has commented adversely on the responded to QLQ score changes of ten to 20 adequacy of the scoring approach (51). It would points, whereas those whose health changed also be good to see more validation studies from “very much” had QLQ scores that differed by independent researchers who are not part of the 20 or more points (9, p14). King reported typi- EORTC group that developed the instrument. cal QLQ score changes over time observed in The reliability scores are moderate to high (as clinical trials for different categories of cancer might be expected in a broad-ranging but brief patients (26). Mean item and scale scores (with instrument), and adequate for studies of groups standard deviations) are given for small samples rather than individual patients. With the of black and white respondents in the United EORTC’s strong collaborative network of re- States (15, Tables 4 and 5). searchers, it seems evident that we will see more testing of this highly promising instrument. Commentary The QLQ exemplifies the trend toward simulta- Address neous development of a measure in many lan- The European Organization for Research and guage and cultural settings that is also seen in Treatment of Cancer, Quality of Life Unit, Av- instruments such as the World Health Organiz- enue E. Mounier 83, Bte 11, B-1200 Brussels, ation Quality of Life measure. This careful at- Belgium. Fax +32 2 779 45 68 (www.eortc tention to cross-cultural applicability is .be/home/qol/ExplQLQ-C30_1.htm) especially relevant given the growing number of Neil K. Aaronson, PhD, Head, Division of international clinical trials. The quality of the Psychosocial Research and Epidemiology, The documentation, the ease of accessing informa- Netherlands Cancer Institute, Plesmanlaan 121, tion on the EORTC web site, and the attention 1066 CX Amsterdam, The Netherlands. to detail in the development of this instrument are outstanding. Work on the EQ-5D is by no means restricted to European languages; it has, References for example, been used to explore the concep- (1) Aaronson NK, Cull A, Kaasa S, et al. The tion of health among the Maori people in New EORTC modular approach to quality of Zealand (50). Based on his experience in cross- life assessment in oncology. Int J Mental national development work on this instrument, Health 1994;23:75–96. Aaronson has offered valuable insights into the (2) Aaronson NK, Cull AM, Kaasa S, et al. The European Organization for Research human dimensions of cross-cultural collabora- and Treatment of Cancer (EORTC) tion (1, pp91–92). modular approach to quality of life The QLQ-C30 is broad in scope, and its de- assessment in oncology: an update. In: sign reflects a workable compromise between the Spilker B, ed. Quality of life and coverage of a generic instrument and the sensi- pharmacoeconomics in clinical trials. tivity of disease-specific questions in evaluating Philadelphia: Lippincott-Raven, change. The items are clearly phrased and most 1996:179–189.  General Health Status and Quality of Life 557 (3) Aaronson NK, Ahmedzai S, Bullinger M, EORTC core quality of life questionnaire et al. The EORTC core quality-of-life (QLQ-C30): validity and reliability when questionnaire: interim results of an analyzed with patients treated with international field study. In: Osoba D, ed. palliative radiotherapy. Eur J Cancer Effect of cancer on quality of life. Boca 1995;31A:2260–2263. Raton, FL: CRC Press, 1991:185–203. (14) Anderson RT, Aaronson NK, Wilkin D. (4) Aaronson NK. Quantitative issues in Critical review of the international health-related quality of life assessment. assessments of health-related quality of life. Health Policy 1988;10:217–230. Qual Life Res 1993;2:369–395. (5) Aaronson NK, Bullinger M, Ahmedzai S. A (15) Ford ME, Havstad SL, Kart CS. Assessing modular approach to quality-of-life the reliability of the EORTC QLQ-C30 in assessment in cancer clinical trials. Recent a sample of older African American and Results Cancer Res 1988;111:231–249. Caucasian adults. Qual Life Res (6) Sprangers MAG, Cull A, Bjordal K, et al. 2001;10:533–541. The European Organization for Research (16) Skarstein J, Aass N, Fosså SD, et al. and Treatment of Cancer approach to Anxiety and depression in cancer patients: quality of life assessment: guidelines for relation between the Hospital Anxiety and developing questionnaire modules. Qual Depression Scale and the European Organ- Life Res 1993;2:287–295. ization for Research and Treatment of (7) Aaronson NK, Ahmedzai S, Bergman B, et Cancer Core Quality of Life al. The European Organization for Questionnaire. J Psychosom Res Research and Treatment of Cancer QLQ- 2000;49:27–34. C30: a quality-of-life instrument for use in (17) Sigurdardóttir V, Bolund C, Brandberg Y, international clinical trials in oncology. J et al. The impact of generalized malignant Natl Cancer Inst 1993;85:365–376. melanoma on quality of life evaluated by (8) Osoba D, Aaronson N, Zee B, et al. the EORTC questionnaire technique. Qual Modification of the EORTC QLQ-C30 Life Res 1993;2:193–203. (version 2.0) based on content validity and (18) Ringdal GI, Ringdal K. Testing the reliability testing in large samples of EORTC Quality of Life Questionnaire on patients with cancer. Qual Life Res cancer patients with heterogeneous 1997;6:103–108. diagnoses. Qual Life Res 1993;2:129–140. (9) Fayers PM, Aaronson NK, Bjordal K, et al. (19) Osoba D, Zee B, Pater JL, et al. The EORTC QLQ-C30 scoring manual. Psychometric properties and responsiveness 3rd ed. Brussels: European Organisation of the EORTC Quality of Life for Research and Treatment of Cancer; Questionnaire (QLQ-C30) in patients with 2001. breast, ovarian and lung cancer. Qual Life (10) Velikova G, Wright EP, Smith AB, et al. Res 1994;3:353–364. Automated collection of quality-of-life (20) Groenvold M, Klee MC, Sprangers MAG, data: a comparison of paper and computer et al. Validation of the EORTC QLQ-C30 touch-screen questionnaires. J Clin Oncol quality of life questionnaire through 1999;17:998–1007. combined qualitative and quantitative (11) Hjermstad MJ, Fossa SD, Bjordal K, et al. assessments of patient-observer agreement. Test/retest study of the European Organiz- J Clin Epidemiol 1997;50:441–450. ation for Research and Treatment of (21) Sneeuw KCA, Aaronson NK, Sprangers Cancer core quality-of-life questionnaire. J MAG, et al. Comparison of patient and Clin Oncol 1995;13:1249–1254. proxy EORTC QLQ-C30 ratings in (12) Chie W-C, Hong R-L, Lai C-C, et al. assessing the quality of life of cancer Quality of life in patients of patients. J Clin Epidemiol nasopharyngeal carcinoma: validation of 1998;51:617–631. the Taiwan Chinese version of the EORTC (22) Kemmler G, Holzner B, Kopp M, et al. QLQ-C30 and the EORTC QLQ-H&N35. Multidimensional scaling as a tool for Qual Life Res 2003;12:93–98. analysing quality of life data. Qual Life (13) Kaasa S, Bjordal K, Aaronson N, et al. The Res 2002;11:223–233. 558 Measuring Health (23) McLachlan S-A, Devins GM, Goodwin PJ. of life assessment for patients with Factor analysis of the psychosocial items of oesophageal cancer. The EORTC Quality the EORTC QLQ-C30 in metastatic breast of Life Study Group. Eur J Cancer cancer patients participating in a 1996;32A:1912–1917. psychosocial intervention study. Qual Life (33) Blazeby JM, Brookes ST, Alderson D. Res 1999;8:311–317. Prognostic value of quality of life scores in (24) Niezgoda HE, Pater JL. A validation study patients with oesophageal cancer. Br J Surg of the domains of the core EORTC Quality 2000;87:362–373. of Life Questionnaire. Qual Life Res (34) Bergman B, Aaronson NK, Ahmedzai S, et 1993;2:319–325. al. The EORTC QLQ-LC13: a modular (25) McLachlan S-A, Devins GM, Goodwin PJ. supplement to the EORTC core Quality of Validation of the European Organization Life Questionnaire (QLQ-C30) for use in for Research and Treatment of Cancer lung cancer clinical trials. Eur J Cancer Quality of Life Questionnaire (QLQ-C30) 1994;30A:635–642. as a measure of psychosocial function in (35) Sprangers MAG, Groenvold M, Arraras JI, breast cancer patients. Eur J Cancer et al. The European Organization for 1998;34:510–517. Research and Treatment of Cancer breast (26) King MT. The interpretation of scores cancer-specific quality-of-life questionnaire from the EORTC quality of life module: first results from a three-country questionnaire QLQ-C30. Qual Life Res field study. J Clin Oncol 1996;5:555–567. 1996;14:2756–2768. (27) Osoba D, Aaronson NK, Muller M, et al. (36) Fitzsimmons D, Johnson CD, George S, et The development and psychometric al. Development of a disease specific validation of a brain cancer quality of life quality of life questionnaire module to questionnaire for use in combination with supplement the EORTC core cancer quality general cancer-specific questionnaires. Qual of life questionnaire, the QLQ-C30, in Life Res 1996;5:139–150. patients with pancreatic cancer. Eur J (28) Bjordal K, Ahlner Elmqvist M. Cancer 1999;35:939–941. Development of an EORTC questionnaire (37) Kavadas V, Blazeby JM, Conroy T, et al. module to be used in quality of life Development of an EORTC disease-specific assessments in head and neck cancer quality of life questionnaire for use in patients. Acta Oncol 1994;33:879–885. patients with liver metastases from (29) Bjordal K, Hammerlid E, Ahlner-Elmqvist colorectal cancer. Eur J Cancer M, et al. Quality of life in head and neck 2003;39:1259–1263. cancer patients: validation of the European (38) Blazeby JM, Conroy T, Bottomley A, et al. Organization for Research and Treatment Clinical and psychometric validation of a of Cancer Quality of Life Questionnaire- questionnaire module, the EORTC QLQ- H&N35. J Clin Oncol STO 22, to assess quality of life in patients 1999;17:1008–1019. with gastric cancer. Eur J Cancer (30) Brandberg Y, Damato B, Kivela T, et al. 2004;40:2260–2268. The EORTC ophthalmic oncology quality (39) Vickery CW, Blazeby JM, Conroy T, et al. of life questionnaire module (EORTC Development of an EORTC disease specific QLQ-OPT30). Development and pre- quality of life module for use in patients testing (Phase I-III). Eye 2004;18:283–289. with gastric cancer. Eur J Cancer (31) Blazeby JM, Conroy T, Hammerlid E, et 2001;37:966–971. al. Clinical and psychometric validation of (40) Blazeby JM, Conroy T, Bottomley A, et al. an EORTC questionnaire module, the Gastrointestinal and Quality of Life EORTC QLQ-OES18, to assess quality of Groups. Clinical and psychometric life in patients with oesophageal cancer. validation of a questionnaire module, the Eur J Cancer 2003;39:1384–1394. EORTC QLQ-STO 22, to assess quality of (32) Blazeby JM, Alderson D, Winstone K, et life in patients with gastric cancer. Eur J al. Development of an EORTC Cancer 2004;40:2260–2268. questionnaire module to be used in quality (41) Greimel E, Bottomley A, Cull A, et al. An  General Health Status and Quality of Life 559 international field study of the reliability for users and developers. Qual Life Res and validity of a disease-specific 2003;12:519–521. questionnaire module (the QLQ-OV28) in assessing the quality of life of patients with ovarian cancer. Eur J Cancer The Quality-Adjusted Time Without 2003;39:1402–1408. Symptoms and Toxicity Method (42) Borghede G, Sullivan M. Measurement of (Q-TWiST) quality of life in localized prostatic cancer (R.D. Gelber, 1989) patients treated with radiotherapy. Development of a prostate cancer-specific Purpose module supplementing the EORTC QLQ- The Q-TWiST method provides a numerical C30. Qual Life Res 1996;5:212–222. way to compare the outcomes of therapies for (43) Sprangers MAG, te Velde A, Aaronson progressive disease in terms of both quality of NK. The construction and testing of the life (QoL) and length of survival (1). It is espe- EORTC colorectal cancer-specific quality cially useful in evaluating treatments that extend of life questionnaire module (QLQ-CR38). survival, but that do so at the cost of undesirable Eur J Cancer 1999;35:238–247. adverse effects (AEs) (2). The resulting numeri- (44) Stead ML, Brown JM, Velikova G, et al. cal summary of quality-adjusted life years is in- Development of an EORTC questionnaire tended to help clinicians and patients choose module to be used in health-related among treatment options. quality-of-life assessment for patients with multiple myeloma. Br J Haematol 1999;104:605–611. Conceptual Basis (45) Sigurdardóttir V, Brandberg Y, Sullivan M. The Q-TWiST method was developed to address Criterion-based validation of the EORTC the challenge of comparing cancer therapies that QLQ-C36 in advanced melanoma: the have differing profiles of efficacy and of AEs. CIPS questionnaire and proxy raters. Qual Cancer treatments that extend survival are often Life Res 1996;5:375–386. toxic, and when a treatment delays cancer remis- (46) Blazeby JM, Currie E, Zee B, et al. sion but at the cost of AEs, the Q-TWiST offers a Development of a questionnaire module to metric that balances the treatment efficacy against supplement the EORTC QLQ-C30 to AEs (3; 4). It allows patients and therapists to assess quality of life in patients with evaluate the trade-off using their own subjective hepatocellular carcinoma, the EORTC QLQ-HCC18. Eur J Cancer judgments: this is useful, for example, in compar- 2004;40:2439–2440. ing treatments with different AE profiles for a pa- (47) Arraras JI, Wright S, Greimel E, et al. tient who wishes to avoid AEs even at the expense Development of a questionnaire to evaluate of a shortened time to recurrence of the disease. the information needs of cancer patients: “The question is whether the insurance paid in the EORTC questionnaire. Patient Educ the form of early toxic effects is seen to have been Couns 2004;54:235–241. a good investment.” (5, p794). For example, (48) Taenzer PA, Speca M, Atkinson MJ, et al. choosing between short- or long-duration adju- Computerized quality-of-life screening in vant chemotherapy entails weighing the possible an oncology clinic. Cancer Pract delayed remission of the latter against the longer 1997;5:168–175. duration of experiencing treatment-related AEs. (49) Fayers PM, Weeden S, Curran D. EORTC This led Gelber and Goldhirsch in 1986 to pro- QLQ-C30 reference values. Brussels: pose the notion of gaining “time without symp- EORTC (ISBN 2-930064-11-0); 1998. (50) Perkins MRV, Devlin NJ, Hansen P. The toms of disease and toxicity of treatment”; the validity and reliability of the EQ-5D health adjustment for quality of survival (hence, Q- state valuations in a survey of Maori. Qual TWiST) was added in 1989. The Q-TWiST ap- Life Res 2004;13:271–274. proach explicitly recognizes that the pay-back (51) Kind P. Using standardised measures of from exposure to toxic treatments only emerges health-related quality of life: critical issues over time, and that different risk-benefit ratios ap- 560 Measuring Health ply at different stages following the start of treat- As with standard survival analysis, the Q- ment. The approach seeks to estimate when (if TWiST focuses on time, but it extends the con- ever) the advantage of treatment in terms of de- ventional analysis by partitioning survival into laying relapse shifts the balance of evidence in fa- predefined clinical states that correspond to dif- vor of the more toxic treatment (5, p794). fering levels of QoL (1). Three steps are involved in applying the method: defining the clinical Description states and deriving utilities for them, drawing By contrast with the other instruments reviewed survival curves on the basis of trial results, and in this book, the Q-TWiST is not a stand-alone thereby comparing treatments (1; 2). The first health measurement but instead provides an ana- step involves defining clinical health states that lytic method for summarizing estimates of QoL are relevant to the course of the disease and high- drawn from the results of clinical trials. It sum- light pertinent contrasts between treatments be- marizes the relative benefits of treatments for a ing compared. States typically include the chronic condition by partitioning a patient’s sur- experience of toxicity due to treatment with its vival time following treatment into a series of estimated utility weight, the period of time with- stages, each with an estimated mean duration out symptoms or toxicity (“TWiST”; utility as- (drawn from the results of clinical trials), and a sumed to be 1.0), and the phase of disease utility weight. In the case of cancer treatment, progression or relapse (Exhibit 10.7). Other survival is divided into phases: typically a period states can be defined as needed; an example of toxicity due to the treatment, a period of be- would be a period of impaired cognition follow- ing without symptoms, and a period of relapse. ing treatment for acute lymphoblastic leukemia Each phase is weighted by subjective utilities, (2). States can also be defined retrospectively, al- providing a composite QoL score. though this will be limited by the information Utility 1.0 0.5 TWiST Toxicity Relapse Death 0.0 Time Exhibit 10.7 The Three Stages of the Clinical Course of Cancer Described in the Q-TWiST  General Health Status and Quality of Life 561 1.0 Treatment A significantly better Utility for 0.5 Treatment A better toxicity (but not significantly so) Line of equivalence between treatments 0.0 0.0 0.5 1.0 Utility for relapse Exhibit 10.8 Utility Plot for a Q-TWiST Comparison of Two Treatments available in the data set. Gelber et al. note that Q-TWiST = uTOX × TOX + TWiST + uREL × REL, toxicity is generally the most difficult to define, and this should be based on symptoms directly where TOX refers to toxicity, REL is relapse, related to QoL (3, p293). Each state is assigned a and u refers to a utility weight. Exhibit 10.7 pro- utility weight between zero (“as bad as death”) vides a notional simplified illustration; the Q- and one (perfect health). Utilities can be ob- TWiST would be the area contained by the three tained from patients (e.g., during the clinical rectangles. trial), using a standard gamble or other utility In the third step, alternative treatments are scaling procedure, as described in Chapter 2 (6); evaluated by comparing Q-TWiST scores for alternatively, utilities can be taken from an exist- each; this can be presented graphically. The mean ing health index by selecting health profiles on expected duration of the toxicity and relapse that index that match the typical condition of phases can be taken from previous clinical trials, patients undergoing the therapy under investiga- but the utility judgments for any given patient tion. The utilities can be varied in a subsequent will not be known ahead of time. Therefore, sensitivity analysis. comparisons may be presented using a sensitivity In the second step, the results of clinical trials analysis that varies the utility weights for each are used to derive Kaplan-Meier survival curves state, showing combinations of values under for the different treatment options. The curves which each treatment would be considered supe- are partitioned to divide the area under the over- rior. Where utilities for two of the states (e.g., all survival plot into the toxicity, TWiST, and re- toxicity and relapse) are to be varied, the result lapse phases. The mean duration of each is can be presented as a two-dimensional plot with calculated and the Q-TWiST outcome is the sum axes that represent the two utilities, as illustrated of the durations of the clinical health states, each in Exhibit 10.8. The plot shows which treatment weighted by its utility coefficient (3, would be preferred for each pair of utility coeffi- pp283–284), cient values. A sloping threshold line designates 562 Measuring Health pairs of utility scores for which the two treat- toxicity may show an initial decline in net Q- ments will have equal Q-TWiST scores (the posi- TWiST, but show a significant advantage after a tion of this line depends on the relative durations time lapse, here roughly a year. of the toxicity and relapse states). A confidence Various authors have proposed enhancements region can then be calculated around this thresh- to the basic Q-TWiST. Cole et al. showed how old line, outside of which one or other treatment covariates may be included using Cox’s propor- is deemed significantly better (1, pp75–76). tional hazards model, which allows patients In a further refinement, the comparison be- with different characteristics to be profiled, tween treatments may be drawn over time. If the showing how prognostic factors affect treatment initially greater toxicity of one treatment is, in- benefits (7; 3, p289). Prognostic factors might deed, offset by longer eventual survival, a graph include tumor size, the number of lymph nodes such as Exhibit 10.9 can illustrate how the rela- involved, or the patient’s age; the equivalent of tive advantages of the treatments evolve over the Exhibit 10.9 is drawn for each prognostic group course of follow-up (3, p288). The horizontal (1, Figure 6). Sloan et al. have described a refine- axis shows time from commencement of treat- ment of the threshold plot that takes account of ment. The Q-TWiST scores for rival treatments biases in the underlying survival curves and im- are calculated at regular intervals and the differ- proves estimates of the significance of the differ- ence in scores plotted on the vertical axis, inter- ence between median Q-TWiST values for preted as quality-adjusted time gained by one treatments (8). Murray and Cole have provided treatment. The dotted lines show confidence formulas for sample size requirements and for bands plotted around the mean scores. As shown estimating variances in the Q-TWiST index (9). in the figure, an effective treatment that causes Gelber et al. have described a method to project 20 15 Quality- 10 adjusted survival advantage 5 for one treatment (in months) 0 -5 -10 0 5 10 15 20 25 30 35 40 45 50 60 Duration of follow-up (months) Exhibit 10.9 Q-TWiST Gain Function in Comparing Two Treatments  General Health Status and Quality of Life 563 survival estimates beyond the actual duration of his more general health policy model that also follow-up in the clinical trials used to derive the includes the QWB (19). Q-TWiST function (10). An approach for esti- The Q-TWiST moves beyond the traditional mating Q-TWiST values from censored observa- cross-sectional perspective of health measures. tions has been described (5), and Mounier et al. The investment metaphor afforded by the have discussed approaches to including data Q-TWiST highlights how patterns of advantage from studies with irregular follow-up delays, unfold over time, offering a more dynamic per- missing data, and informative (i.e., non random) spective than that provided by summative popu- drop-outs (11). lation health measures such as disability-adjusted life years. This type of thinking could be applied Reliability and Validity to many other aspects of life, such as decisions Because the Q-TWiST is not itself a health mea- about to how much time to spend in the educa- sure, the normal issues of reliability and validity tional system in the hope of subsequently greater do not apply. income. Alternative Forms References Although originally developed for adjuvant chemotherapy for cancer, the Q-TWiST method (1) Gelber RD, Goldhirsch A, Cole BF. has been applied in many other chronic condi- Evaluation of effectiveness: Q-TWIST. tions (3). These include multiple sclerosis (12) Cancer Treat Rev 1993;19(suppl A):73–84. (2) Gelber S, Gelber RD, Goldhirsch A. The and epilepsy, in which the dimensions included Q-TWiST method. Quality of Life drug toxicity, memory impairment, occurrence Newsletter 1996;16:3–4. of seizures, employment status, and other mor- (3) Gelber S, Gelber RD, Cole BF, et al. Using bidities and impairments (e.g., depression, activ- the Q-TWIST method for treatment ity limitations) (13). The Q-TWiST has been comparisons in clinical trials. In: Staquet used in studies of children (14) and in evaluating MJ, Hays RD, Fayers PM, eds. Quality of zidovudine treatment for HIV (15). life assessment in clincial trials: methods and practice. Oxford: Oxford University Commentary Press, 1998:281–296. The Q-TWiST offers an innovative approach to (4) Goldhirsch A, Gelber RD, Simes RJ, et al. summarizing health-related QoL life data, and Costs and benefits of adjuvant therapy in breast cancer: a quality-adjusted survival a practical way to integrate health measure- analysis. J Clin Oncol 1989;7:36–44. ment scales with survival analysis. It is widely (5) Gelber RD, Gelman RS, Goldhirsch A. A used in the oncology trials literature, for exam- quality-of-life-oriented endpoint for ple in deciding the optimal duration of therapy comparing therapies. Biometrics (16), in cost-benefit analyses (17), and in meta- 1989;45:781–795. analyses of treatment; an example is given by (6) Schwartz CE, Mathias SD, Pasta DJ, et al. an analysis of 47 breast cancer intervention tri- A comparison of two approaches for als (18). The Q-TWiST opens the door to assessing patient importance weights to recording multistate quality-adjusted life years; conduct an Extended Q-TWiST analysis. it extends the cross-sectional picture offered by Qual Life Res 1999;8:197–207. instruments such as the Health Utilities Index (7) Cole BF, Gelber RD, Goldhirsch A. Cox regression models for quality adjusted or the Quality of Well-Being scale (QWB) to survival analysis. Stat Med present patterns of quality of survival over 1993;12:975–987. time. The main goal of the QWB is to make (8) Sloan JA, Sargent DJ, Lindman J, et al. A comparisons across diseases; the Q-TWiST al- new graphic for quality adjusted life years lows comparisons across treatments for each (Q-TWiST) survival analysis: the Q- disease (13, p140). Kaplan, however, claims TWiST plot. Qual Life Res that the Q-TWiST can be viewed as a subset of 2002;11:37–45. 564 Measuring Health (9) Murray S, Cole B. Variance and sample size cost/utility studies in cancer. Cancer Treat calculations in quality-of-life—adjusted Rev 1993;19(suppl A):85–96. survival analysis (Q-TWiST). Biometrics 2000;56:173–182. (10) Cole BF, Gelber RD, Anderson KM. The Quality of Life Index Parametric approaches to quality-adjusted (W.O. Spitzer, 1980) survival analysis. International Breast Purpose Cancer Study Group. Biometrics 1994;50:621–631. The Quality of Life Index (QL Index) measures (11) Mounier N, Ferme C, Flechtner H, et al. the general well-being of patients who are termi- Model-based methodology for analyzing nally ill with cancer or other chronic diseases. It incomplete quality-of-life data and was intended as a brief instrument for evaluating integrating them into the Q-TWiST the effects of treatment and programs including framework. Med Decis Making palliative care, but it has subsequently been used 2003;23:54–66. more broadly. (12) Schwartz CE, Cole BF, Gelber RD. Mea- suring patient-centered outcomes in Conceptual Basis neurologic disease. Extending the Q- Spitzer’s index is one of the earliest scales to TWiST method. Arch Neurol be designated a quality of life measurement. It 1995;52:754–762. sought to extend outcome assessment beyond (13) Schwartz CE, Cole BF, Vickrey BG, et al. morbidity and mortality by providing a simple The Q-TWiST approach to assessing health-related quality of life in epilepsy. and quantified scale that might become the Qual Life Res 1995;4:135–141. equivalent of the neonatal Apgar score (1). The (14) Parsons SK, Gelber S, Cole BF, et al. theme of QoL was seen as pertinent for patients Quality-adjusted survival after treatment with terminal illness, for whom extending the for acute myeloid leukemia in childhood: a length of life is not feasible and for whom, there- Q-TWiST analysis of the Pediatric fore, the prime objective of treatment is to main- Oncology Group Study 8821. J Clin Oncol tain quality of life. 1999;17:2144–2152. Spitzer saw several dimensions to QoL, in- (15) Lenderking WR, Gelber RD, Cotton DJ, et cluding a positive mood state, supportive rela- al. Evaluation of the queality of life tionships, and the absence of physical or associated with zidovudine treatment in psychological distress (2). Hence, a measure- asymptomatic human immunodeficiency virus infection. N Engl J Med ment of QoL should consider physical, social, 1994;330:738–743. and emotional function; attitudes to illness; the (16) Nooij MA, de Haes JC, Beex LV, et al. adequacy of family interactions; and the burden Continuing chemotherapy or not after the of illness to the individual (3). induction treatment in advanced breast cancer patients. clinical outcomes and Description oncologists’ preferences. Eur J Cancer Potential themes for inclusion in the QL Index 2003;39:614–621. were selected empirically from an opinion sur- (17) Brown ML, Nayfield SG, Shibley LM. vey among patients with a chronic disease, their Adjuvant therapy for stage III colon cancer: relatives, health care professionals, and healthy economics returns to research and cost- people concerning the factors that enhance or effectiveness of treatment. J Natl Cancer detract from quality of life (4, p37). Fourteen Inst 1994;86:424–430. (18) Cole BF, Gelber RD, Gelber S, et al. themes were identified, including absence of Polychemotherapy for early breast cancer: symptoms, mental alertness and financial inde- an overview of the randomised clinical pendence. Questions measuring these were in- trials with quality-adjusted survival corporated in draft versions of the QL Index. analysis. Lancet 2001;358:277–286. Pilot testing led to the selection of the five (19) Kaplan RM. Quality of life assessment for themes in the eventual index: activity level (in-  General Health Status and Quality of Life 565 cluding occupation), activities of daily living, Stability was assessed by having patients rate feelings of healthiness, quality of social support, themselves daily for five days; their scores and psychological outlook (3). The resulting agreed completely 80% of the time and fell scale takes about two minutes to complete and is within one point 95% of the time (9). The QL administered by a physician or other health care Index was administered twice to patients with professional. This version is shown in Exhibit breast cancer who, in the judgment of a nurse, 10.10, and a self-administered version is shown had not changed physical or emotional func- in Exhibit 10.11. Scores of 0, 1, or 2 for each tioning over a two-week period. The QL Index category reflect increasing well-being and may (like a second scale, the Breast Cancer Ques- be summed to give a total score ranging from 0 tionnaire) showed a significant decline, suggest- to 10. In the interests of simplicity, differential ing either that the nurse’s rating missed a weights for the categories are not used. deterioration or that there is measurement error Along with the QL Index, Spitzer developed a in the scales (11). QL Uniscale. This is a visual analogue scale rang- ing from “lowest quality” to “highest quality”; it Validity may be included in the interviewer and self- The content validity of the QL Index was administered versions. A correlation of 0.70 be- checked by asking patients, healthy people, tween the QL Index and the uniscale was reported physicians, and researchers to judge its scope in a study of patients with breast cancer (5, p589). and design (2). Ten items from the initial item pool that were Reliability not used in the QL Index were formed into a Spearman rho correlations among the five items more extensive QoL scale. This instrument was ranged from 0.21 to 0.71; three coefficients administered with the QL Index to 476 patients were 0.5 or more (3, Table 1; 4, Table 4). Inter- and healthy people in Australia. The correlations nal consistency was assessed on 91 patients in an between the QL Index and the comprehensive Australian study, giving an alpha of 0.77 (3, scale were highest for samples of patients with p594). Alpha was also calculated for a sample of cancer (rho = 0.61–0.71), intermediate for other 261 patients in Canada, giving a value of 0.78, chronically sick patients, and low and of 0.85 for a subset of patients with cancer (rho = 0.29–0.49) for healthy people (3, Table 2). (2). Other estimates of internal consistency in- Replication of these analyses in a Canadian clude alpha = 0.66 (6), 0.77, and 0.80 (7). Item- study yielded correlations between 0.66 and total corrrelations ranged from 0.49 (activity) to 0.84 for cancer patients (2, Table 4). Correla- 0.86 (outlook) (8, Table 2). tions between the QL Index and the Uniscale Inter-rater reliability was studied by compar- were higher for cancer patients (0.49 to 0.61) ing pairs of ratings made by different physicians; than for healthy people (0.17 to 0.39). Spearman correlations for various samples Spitzer et al. reported mean scores for ranged from 0.74 to 0.84 (3; 4, Table 5). When healthy people and various patient groups, show- five raters judged each patient, they agreed com- ing that the index discriminates between healthy pletely on only 45% of ratings (coefficient of people and various categories of patients (3). concordance = 0.54) (9). Several studies have Scores on the QL Index have shown a decline compared the self-report and physician rating with age in several studies (1, p175). In a study versions of the QL Index. For 161 Australian pa- of congestive heart failure, the QL Index showed tients, the Spearman correlation was 0.61; for 51 greater ability to discriminate between experi- Canadian patients, it was 0.69 (3, p595). Other mental and control groups than the Sickness Im- studies obtained correlations of 0.45 (2) and pact Profile or the Quality of Well-Being Scale 0.72 (10), and a Kendall correlation of 0.72 (10, (12). Table 1). The Pearson correlation between self- The QL Index has been compared with the report and ratings of relatives was 0.63 for 261 Karnofsky Performance Status Scale in several Canadian patients (2). studies with variable results: correlations of 0.72  Exhibit 10.10 The Quality of Life Index: Clinician Rating Version Score each heading 2, 1 or 0 according to your most recent assessment of the patient. Activity During the last week, the patient • has been working or studying full-time or nearly so, in usual occupation; or managing own household; or participating in unpaid 2 or voluntary activities, whether retired or not • has been working or studying in usual occupation or managing own household or participating in unpaid or voluntary activities; 1 but requiring major assistance or a significant reduction in hours worked or a sheltered situation or was on sick leave • has not been working or studying in any capacity and not managing own household 0 Daily living During the last week, the patient • has been self-reliant in eating, washing, toileting and dressing; using public transport or driving own car 2 • has been requiring assistance (another person or special equipment) for daily activities and transport but performing light tasks 1 • has not been managing personal care nor light tasks and/or not leaving own home or institution at all 0 Health During the last week, the patient • has been appearing to feel well or reporting feeling “great” most of the time 2 • has been lacking energy or not feeling entirely “up to par” more than just occasionally 1 • has been feeling very ill or “lousy,” seeming weak and washed out most of the time or was unconscious 0 566 Support During the last week • the patient has been having good relationships with others and receiving strong support from at least one family member and/or 2 friend • support received or perceived has been limited from family and friends and/or by the patient’s condition 1 • support from family and friends occurred infrequently or only when absolutely necessary or patient was unconscious 0 Outlook During the past week the patient • has usually been appearing calm and positive in outlook, accepting and in control of personal 2 circumstances, including surroundings • has sometimes been troubled because not fully in control of personal circumstances or has been having 1 periods of obvious anxiety or depression • has been seriously confused or very frightened or consistently anxious and depressed or unconscious 0 QL-Index total How confident are you that your scoring of the preceding dimensions is accurate? Please ring [circle] the appropriate category. Absolutely confident Very confident Quite confident Not very confident Very doubtful Not at all confident 1 2 3 4 5 6 Reproduced from Spitzer WO, Dobson AJ, Hall J, Chesterman E, Levi J, Shepherd R, Battista RN, Catchlove BR. Measuring the quality of life of cancer patients: a concise QL-Index for use by physicians. J Chronic Dis 1981;34:591, Figure 2, Pergamon Press Ltd.  General Health Status and Quality of Life 567 Exhibit 10.11 The Quality of Life Index: Self-assessment Each of the next five headings has three choices. Under each heading put a circle around the number that best describes your quality of life during the last week. Activity—What is your “main” activity? 1. I work full-time (or nearly so) in my usual occupation or study full-time (or nearly so) or manage my own household or take part in as much unpaid or voluntary activity as I wish, whether retired or not. 2. I work or study in my usual occupation or manage my own household or participate in unpaid or voluntary activities; but I need a lot of help to do so or I work greatly reduced hours. 3. I do not work in any capacity, nor do I study, nor do I manage my own household. Daily living—Ability to look after yourself. 1. I am able to eat, wash, go to the toilet and dress without assistance. I drive a car or use public transport without assistance. 2. I can travel and perform daily activities only with assistance (another person or special equipment) but can perform light tasks. 3. I am confined to my home or an institution and cannot manage personal care or light tasks at all. Health—What is your state of health? 1. I feel well most of the time. 2. I lack energy or only feel “up to par” some of the time. 3. I feel very ill or “lousy” most of the time. Support—What support do you receive from others? 1. I have good relationships with others and receive strong support from at least one family member and/or friend. 2. The support I receive from family and friends is limited. 3. The support I receive from family and friends occurs infrequently or only when absolutely necessary. Outlook—How do you feel about your life? 1. I am basically a calm person. I generally look forward to things and am able to make my own decisions about my life and surroundings. 2. I am sometimes troubled and there are times when I do not feel fully in control of my personal life. I am anxious and depressed at times. 3. I feel frightened and completely confused about things in general. Adapted from an original provided by Dr. Spitzer. With permission. (13), 0.41, and 0.83 (6). A study of 45 elderly the typical correlation between self-administered patients reported moderate correlations with and interview versions of the QL Index of about three scales of activities of daily living (ADL): 0.70. The social support item correlated 0.72 0.52 with the Barthel Index, 0.45 with the Func- with Wortman’s social support scale (8, Table 5). tional Independence Measure, and 0.48 with The outlook item correlated −0.68 with the Katz’s Index of ADL (14, Table 2). Index scores Hamilton Rating Scale for Depression and −0.63 correlated 0.42 with a Breast Cancer Question- with the Brief Symptom Inventory anxiety scale naire (11). A Kendall rank correlation of 0.53 (8, Table 7). was reported between the Index scores and a vi- sual analogue scale (similar to the QL Uniscale). Alternative Forms Although this does not seem high enough to in- Morris et al. replaced the activity item by one on dicate that the QL Index could be replaced by mobility, arguing that the activity item was inap- the simpler scale, it is not very much lower than propriate for terminally ill patients; they also re- 568 Measuring Health worded items for verbal presentation. This in- be controlled for in research studies (2). It strument is termed the “HRCA-QL Index” (for should also be recognized that the QL Index may Hebrew Rehabilitation Center for the Aged) (15, be less sensitive to short-term changes than some pp57–58). These changes seem to offer little ad- other, disease-specific instruments. Its reliance vantage in terms of power or efficiency of the on physical functioning and on observer assess- QL Index, however (1, p173). ments may be at the expense of sensitivity to the The QL Index has been translated into impact of cancer on the patient’s social activities French (2), German (13), and Italian (16, p388). or emotional status (16, p389). The instrument is more suited to global assessments of seriously Reference Standards sick patients over the long term than to detailed Representative scores from healthy and sick re- evaluation of short-term outcomes (1, p182). spondents are given by Spitzer. Mean scores for The instrument is not suitable for healthy re- two samples of healthy people were 8.8 and 9.2; spondents, for whom survey measures such as means for those with chronic disease were 7.3, the SF-36 would be more appropriate (3). and for seriously ill patients, the mean was 3.3 (3, Table 3). References Commentary The emphasis in the QL Index is on practicality; (1) Wood-Dauphinee S, Williams JI. The Spitzer Quality-of-Life Index: its perfor- it is brief and easy to administer in a clinical set- mance as a measure. In: Osoba D, ed. ting and yet broad in scope: “The QL Index, like Effect of cancer on quality of life. Boca the Apgar score, should be thought of as a com- Raton, Florida: CRC Press, 1991:169–184. posite of dissimilar items” (1, p181). It has (2) Suissa S, Shenker SC, Spitzer WO. Measur- proved acceptable to clinicians and has been ing the quality of life of cancer and widely cited in the literature. The QL Index was chronically ill patients: cross-validation carefully developed, with extensive consultation studies of the Quality of Life Index. with patients and clinicians; the approach has [Manuscript] Montréal, Québec: served as a model for subsequent investigators Department of Clinical Epidemiology, (1). The QL Index has been used in Australia, McGill University; 1984. Great Britain, Canada, Germany, and the United (3) Spitzer WO, Dobson AJ, Hall J, et al. Mea- suring the quality of life of cancer patients: States. Reliability and validity results are good; a concise QL-Index for use by physicians. J the pattern of validity results suggests that the Chronic Dis 1981;34:585–597. instrument achieves its aims. (4) Spitzer WO. Quality of life. In: Burley D, A brief scale that is also broad in scope may Inman WHW, eds. Therapeutic risk: be expected to sacrifice some psychometric prop- perception, measurement, management. erties, but the internal consistency alphas of New York: John Wiley & Sons, 0.77 and above are reasonably high. The finding 1988:35–46. that agreement between patient and physician (5) Perez DJ, Williams SM, Christensen EA, et ratings is modest (in the range of 0.6 to 0.7) may al. A longitudinal study of health related not indicate a failure of reliability so much as re- quality of life and utility measures in mind us that the perspectives differ and both patients with advanced breast cancer. Qual Life Res 2001;10:587–593. types of rating are needed. The modest agree- (6) Mor V. Cancer patients’ quality of life over ment also indicates that the scale must be consis- the disease course: lessons from the real tently completed by one type of respondent in a world. J Chronic Dis 1987;40:535–544. before and after evaluative design. Suissa et al. (7) Mor V, Stalker MZ, Gralla R, et al. Day identified systematic variations in QL Index hospital as an alternative to inpatient care scores by age, sex, and disease categories, sug- for cancer patients: a random assignment gesting that these confounding variables should trial. J Clin Epidemiol 1988;41:771.  General Health Status and Quality of Life 569 (8) Williams JBW, Rabkin JG. The concurrent Conceptual Basis validity of items in the Quality-of-Life The Dartmouth/Northern New England Primary Index in a cohort of HIV-positive and HIV- Care Cooperative Information Project (the COOP negative gay men. Control Clin Trials project) at the Dartmouth Medical School in New 1991;12(suppl):129S–141S. Hampshire was established to create a practical (9) Slevin ML, Plant H, Lynch D, et al. Who system for measuring health status in physicians’ should measure quality of life, the doctor offices. Nelson et al. argued that, beyond safe- or the patient? Br J Cancer 1988;57:109–112. guarding biological function, care should be con- (10) Gough IR, Furnival CM, Schilder L, et al. cerned with improving patients’ physical, mental, Assessment of the quality of life of patients and role functions. The functioning of the person with advanced cancer. Eur J Cancer Clin as a whole is more important than that of sepa- Oncol 1983;19:1161–1165. rate organ systems, so indicators of organ system (11) Levine MN, Guyatt GH, Gent M, et al. function are not sufficient as indicators of health Quality of life in stage II breast cancer: an (3; 4). Several design requirements were made of instrument for clinical trials. J Clin Oncol the COOP Charts: they had to be reliable, valid, 1988;6:1798–1810. acceptable in routine practice, applicable to a (12) Tandon PK, Stander H, Schwarz RP, Jr. wide range of diagnoses, easily interpretable, and Analysis of quality of life data from a they had to provide clinically useful information randomized, placebo-controlled heart-failure trial. J Clin Epidemiol 1989;42:955–962. (1; 5, pp97–98). (13) Köster R, Gebbensleben B, Stützer H, et al. The COOP information model contains five Quality of life in gastric cancer. stages: screening, assessment, diagnosis, treat- Karnofsky’s scale and Spitzer’s index in ment, and monitoring, of which the charts form comparison at the time of surgery in a the screening stage (1, p61). cohort of 1081 patients. Scand J Gastroenterol 1987;22(suppl Used correctly, the Charts allow practitioners 133):102–106. to screen patients quickly and efficiently and (14) Rockwood K, Stolee P, Fox RA. Use of goal to highlight those individuals who might ben- attainment scaling in measuring clinically efit from a more comprehensive inquiry of important change in the frail elderly. J Clin Epidemiol 1993;46:1113–1118. functioning and health-related quality of life. (15) Morris JN, Suissa S, Sherwood S, et al. The screening instrument should be used pe- Last days: a study of the quality of life of riodically to monitor the progression of terminally ill cancer patients. J Chronic Dis chronic disease of a patient and to diagnose 1986;39:47–62. the onset of new disease. (6, p152) (16) Anderson RT, Aaronson NK, Wilkin D. Critical review of the international Wasson et al. illustrate how the screening infor- assessments of health-related quality of life. mation is used in routine clinical practice (7). Qual Life Res 1993;2:369–395. Time constraints in practice demand an instru- ment that can be completed and scored quickly, applied to a wide range of problems and diag- The COOP Charts for Primary Care noses, and provide clinically useful information. Practices With this in mind, the instrument was designed (Eugene C. Nelson, 1987) in the form of charts, similar in concept to the Purpose familiar Snellen charts used to measure visual The COOP Charts provide a rapid way to assess acuity. the health and functioning of patients in primary care practices (1–3). They are intended for use in Description routine clinical practice, rather than as research The original three charts covered physical, emo- instruments. tional, and role function (1). Successful pretest- 570 Measuring Health ing led to an expansion of the method to pro- class correlation (ICC) averaged 0.77 across the duce nine charts, each with a single question charts; the range was 0.50 to 0.98 (2, Table 1); about health during the past month. They are eight of the nine correlation coefficients were shown in Exhibit 10.12. Three charts cover 0.76 or above (four had kappa coefficients above function (physical, fitness, daily and social activ- 0.65) (5, Table 8.5). ities), three cover health perceptions (quality of Nelson et al. have reported reliability from life, overall health, and change in health condi- four samples (5, Table 8.5). One-hour retest re- tion), two cover symptoms and feelings (pain, liability averaged r = 0.93 for the nine charts, emotional status), and one (social support) is with a range of 0.78 to 0.98 (N = 53); in a dif- seen as a factor influencing health. Each chart ferent sample of 51 outpatients, reliability aver- includes a descriptive title, a question relating to aged 0.88 (range, 0.73–0.98) (2, Table 1). the past four weeks, and a five-point answer Retest reliability at two weeks was lower, aver- scale illustrated by simple pictures; a score of aging 0.67 (range, 0.42–0.88) (2, Table 1). Reli- five represents the most severe limitations. A ability at one week ranged from 0.67 (physical randomized trial that compared the charts with condition) to 0.82 (daily work) for the Dutch and without the pictures showed no difference in version of the charts (10, Table 9.1). Two-week mean scores between the two versions (8). retest coefficients were 0.74 for physical fitness, Nelson et al. originally recommended that the 0.75 for feelings, 0.78 for daily activities, 0.76 charts be administered by trained office staff (2, for social activities, and 0.67 for overall health p1121) but they can also be completed by the pa- (11, p531). tient in the waiting room in less than five minutes McHorney et al. compared the reliability of (2, p1119; 5). Administration by staff is typically the individual chart ratings and the global rat- done at the beginning of the office visit while rou- ings from the MOS instrument (12). The two in- tine data are being collected (1). The diagrams struments had the same number of scale levels can be displayed on the wall, or the patient may but only the COOP instrument had illustrations. be handed copies. The answer sheet is stored in The MOS ratings showed markedly higher relia- the patient’s medical record. The nine charts are bilities: for example, 0.57 compared with 0.37 considered as separate dimensions of function- for physical functioning and 0.65 compared ing, and an overall score is not calculated. with 0.54 for emotional functioning (12, Table Nelson et al. initially tested the charts on 117 2). The intercorrelations among scores for the patients (1), demonstrating their feasibility and nine charts averaged 0.37 (range 0.06 to 0.68) acceptability and collecting preliminary data on (13, p176). validity and reliability. The charts were revised and subsequently tested in a series of studies (2; Validity 5, Table 8.2). The largest included 2,349 pa- acceptability. The validity of the COOP tients participating in the RAND Medical Out- Charts has been tested in numerous studies, of- comes Study (MOS), in which the COOP Charts ten with large samples. Nelson evaluated the ac- were included. ceptability of the charts by interviewing office Concerning the interpretation of scores, Nel- staff and patients in ten practices that had been son suggested that “a chart score of 4 or 5 using them. Interviews with 225 patients indi- should always be considered abnormal.” (9, cated that 97% understood the charts, 89% en- p165). He gave a method for translating COOP joyed them, 93% liked the pictures, and 74% scores into equivalent scores on the SF-36 (9, rated the charts as useful (6, Table 10.3). A Table 3); this is also shown on the COOP Charts Swiss study compared charts with pictures to a Web site (see Address section). version without, and found few differences in re- sponse patterns or validity (14). Likewise, Jenk- Reliability inson et al. obtained virtually identical response Inter-rater agreement between tests administered rates and scores in versions of the charts with by a physician and a nurse measured by the intra- and without the illustrations in a large random- Exhibit 10.12 The Dartmouth COOP Charts (continued) 571 Exhibit 10.12 (continued) 572  General Health Status and Quality of Life 573 Exhibit 10.12 tients with minor and serious complaints, and psychiatric patients; they also compared this dis- crimination with that of MOS global indicators (12, Tables 3–8). The COOP Charts all proved capable of discriminating between categories of patient; in general, they performed less well than the equivalent MOS indicators for medical pa- tients but better for psychiatric patients. Nelson et al. also compared the discriminal ability of the charts and MOS scores. Interestingly, they found the two methods similar for physical illness, and that the charts were superior for emotional symptoms (5, pp111,116). The COOP Charts have been compared with various short forms of the MOS measures, in- cluding the SF-20 (2; 12; 16; 17). Nelson et al. compared the three original COOP physical, emotional, and role function charts to equivalent scales from the a 28-item short-form MOS in- strument (1). Convergent validity correlations (i.e., those between scales purporting to measure the same theme) were 0.71 (for physical func- tioning) and 0.74 (emotional) and 0.40 (daily activities) (1, Table 2; 1, Table 2); equivalent fig- ures from a study testing the revised version of the COOP Charts were 0.59, 0.69, and 0.60 (2, Adapted from an original provided by Dr. E. Nelson. With permission. Table 2; 9, Table 1). The average convergent cor- relations in two samples were 0.62 and 0.46 (2, Table 1). As expected, these exceeded correla- tions between scales measuring different attrib- ized study (15). Used in clinical settings, physi- utes; the average discriminant correlations were cians have reported that in one quarter of cases, 0.39 and 0.32 (2, p1117). Convergent validity the charts provided them with new information coefficients ranging from 0.48 to 0.78 were re- and that in 40% of these, the information led to ported from three samples (total sample size of a change in management (5, p117). For 13% of over 1,000) (5, Table 8.8). Two multitrait- patients, doctors reported that the charts had a multimethod analyses showed an average con- positive effect on communication but for 2% vergent coefficient of 0.60, and an average they had a negative effect (2, p1120). discriminant correlation of 0.16 (5, p111). In a comparison with the Nottingham Health Profile, concurrent validity. Nelson et al. compared the COOP Charts proved more sensitive to mi- the chart scores to a range of criterion variables nor deviations from complete well-being (18). in a study of 784 patients as part of the MOS. Four of the COOP Charts were compared with Convergent coefficients fell in the range 0.40 to the Nottingham scales in a Dutch study; conver- 0.50 for most charts (5, Table 8.6). The correla- gent correlation coefficients fell ranged from tions between chart scores and numbers of 0.35 to 0.58 (10, Table 9.2). Convergent corre- symptoms ranged from 0.25 (for “health lations between chart scores and the EuroQoL change”) to 0.51 (“overall health”) (2, p1117). EQ-5D scales include 0.60 for physical fitness, McHorney et al. evaluated the ability of each 0.45 for daily activities, and 0.72 for feelings COOP Chart to discriminate among medical pa- (11, Table 4). Equivalent correlations in the 574 Measuring Health Essink-Bot et al. study were 0.58 for daily activ- by performance based measures of gait and bal- ities (COOP) and Mobility (EuroQol), 0.83 for ance or the 50-foot walk time (16, Table 5). Siu Feelings (COOP) with Anxiety (EQ-5D), and et al. commented that although cross-sectional 0.58 for Daily activities (COOP) with Mobility studies have supported the convergent and dis- (EQ-5D) (19, p537). criminant validity of the SF-20 and COOP in- The correlations between the emotional con- struments, the results may not hold for changes dition chart and the General Health Question- in functional status (16, p1100). naire (GHQ) was 0.60 in a Dutch study. The Despite apparently covering the positive end correlation between the GHQ and the daily of the health continuum, the COOP charts have work section was 0.61, suggesting that the daily limitations in discriminating among healthy peo- work chart reflects emotional difficulties as ple (i.e., a ceiling effect), although this appears much as physical ones (10, Table 9.2). somewhat less severe than that for the Notting- A Swiss study reported a two-factor solution, ham Health Profile (21, p305). Seven of eight dividing the COOP into mental and physical chart replies differed significantly among those scales. The correlation between the mental scale with long-standing illness and those without (15, and the mental score of the SF-36 was 0.77, Table 6). Essink-Bot et al. compared the ability whereas the corresponding correlation for the of scores on four measures to discriminate be- physical scales was 0.74 (14, p409). tween people absent from work due to illness and others who remained at work. The SF-36 sensitivity to change. Regression analyses com- was the most discriminating (mean area under pared the sensitivity of the COOP Charts and the ROC curve 0.72), followed by the COOP SF-20 scales in reflecting the impact of selected Charts (mean area under the curve, 0.64), the chronic conditions on personal functioning. The EuroQol EQ-5D (mean area under the curve, variance explained by the physical function scale 0.61) and the Nottingham Health Profile (mean was slightly superior for the COOP Chart, area under the curve, 0.60) (19, Table 5). whereas for the emotional, role and overall scales, the SF-20 scales were better (markedly so predictive validity. A study of elderly patients for the emotional questions: R2 = 0.29 for the compared the validity of COOP and SF-20 COOP Charts versus 0.52 for the RAND ques- scores in predicting future health status; both tions) (2, Table 3). Correlations with number of provided significant and almost identical predic- symptoms ranged from 0.27 (social support) to tors of nursing home placement and hospitaliza- 0.51 (overall health) for 231 elderly people (5, tion (17). Wasson et al. noted that the predictive Table 8.7). In a study of hernia surgery, the value of a good emotional health score on the COOP charts provided comparable effect size COOP Charts lies somewhere between 96% and statistics to those for the SF-36; the physical, 99%: the emotional chart is adequate to rule out pain, and social function scales for both pro- psychiatric illness (7, pMS46). vided large effect size statistics in the range 1.0 to 1.85 (20, Table 4). Alternative Forms Siu et al. compared the convergent and dis- In 1988, the World Organization of National criminant validity of change scores on the Colleges, Academies, and Academic Associations COOP and SF-20. The average convergent va- of General Practitioners and Family Physicians lidity correlation was 0.37, and the average di- (WONCA) adopted the COOP Charts as the ba- vergent correlation was 0.18 (16, p1096). sis of their international system for measuring Moderate-to-strong convergent correlations functional status (22; 23). The WONCA group were observed only for pain, social, and mental modified the charts in several ways, changing the scores; convergent validity between change on titles and wording of the questions, omitting the the physical function measures was only 0.05 quality of life and social support charts, and (16, p1096). The COOP function score failed to changing the physical chart to cover only walk- detect changes in physical function as measured ing (23; 24; 25, p383). The COOP/WONCA  General Health Status and Quality of Life 575 charts are shown in an article by Sneeuw et al. standard deviation norms are available for a (26, p1215), and are available in over 20 lan- British community sample by gender, age, and guages from www.globalfamilydoctor.com. Four- social class (15, Tables 3–5). week test-retest reliability correlations for this version ranged from 0.42 to 0.62 (25, p384); in Commentary another study, one-day retest intraclass correla- The COOP Charts offer an imaginative and in- tion coefficients (ICCs) were 0.79 for the feelings novative approach to routine measurement in score and ranged from 0.84 to 0.89 for the other clinical practice. They are quick and easy to use, scales (26, Table 3). Mean ICCs between ratings they are acceptable to patients, and they are made by patients and family members were 0.54, judged clinically useful by physicians (1). In a whereas the ICC between patients and physician comparative review of five scales, McHorney ratings was 0.48 (26, p1210). In a simple exami- and Tarlov noted, “The COOP charts excelled nation of responsiveness, change scores over time in all standards pertaining to practical features: on each of the charts agreed with patient’s over- they were the shortest in length, the briefest to all ratings of whether their condition had im- complete and the easiest to score and interpret.” proved or deteriorated (26, Table 6). Other (21, p298). The visual metaphors are attractive information on validity, including correlations and may speed up administration; they may also with the Nottingham Health Profile and the SF- help to overcome language barriers in cross- 36, is available from the Global family doctor cultural applications, in areas where patients are Web site. marginally literate, or with patients unfamiliar The COOP Charts have been used in the with thinking abstractly (32). Methodologically, Netherlands (10), Japan (27), China (28) and the charts represent one side in the debate over Hong Kong with positive results (29). They have how brief a measure can be and still achieve the been adapted for use with North American Indi- precision required in clinical practice and re- ans (30). Along with the translation, the illustra- search (12). An impressive amount of data has tions are usually altered, at times offering been collected attesting to the validity and relia- intriguing cultural insights. The Dutch version, bility of the charts; the Dartmouth COOP proj- for example, depicts strenuous physical activity ect team has been a solid force in overseeing this. by a woman positively rushing up stairs two The comparison between the charts and the steps at a time (10, p134); equivalent activity for MOS instruments has in part been fueled by a the original COOP man is more sedate, at one desire to guide the balance between brevity and step at a time. The WONCA version is also accuracy of measurement. The results thus far available in Danish, Finnish, Norwegian, Span- seem to suggest that the single-item measures in ish, German, Hebrew, and Urdu (24). the charts can provide quite good indications of The COOP Chart approach has been adapted present state but are too coarse to detect minor for adolescents; six charts cover physical fitness, changes in function over time. The reliability feelings, school work, social support, family and validity studies have been undertaken in sev- communications, and health habits. The charts eral testing sites and offer a more extensive and appear acceptable to adolescents and they corre- varied testing protocol than applied in most reli- late well with questionnaire measures (31). The ability and validity studies. The low reliability drawings in the adolescent charts replace the results compared with the MOS global ratings stick figures used in the adult version with round give some cause for concern (12). heads from which the limbs protrude; they are The illustrations in the COOP Charts have shown in a review by Nelson (9, Figure 2). raised considerable interest, as have the smiling faces scales reviewed elsewhere in this chapter. Reference Standards The approach appears attractive to respondents, Mean scores and standard deviations are re- although the consistency of interpreting illustra- ported for a Swiss sample, by gender, and age- tions by different cultural groups is a point of group (14, Table 6). Mean, median, and concern. Furthermore, pictures are usually more 576 Measuring Health specific than a verbal description, and so may fo- of health status assessment measures in cus attention on a narrow interpretation of the clinical settings. Med Care question (32). McHorney et al. concluded that the 1992;30(suppl):MS23–MS41. illustrations may direct attention to psychological (5) Nelson EC, Landgraf JM, Hays RD, et al. distress rather than physical impairment: “Simi- The COOP Function Charts: a system to measure patient function in physicians’ larly, the use of a smiling face for no pain and a offices. In: WONCA Classification sad face with drooping shoulders for severe pain Committee, ed. Functional status measure- appears to have increased the sensitivity of the ment in primary care. New York: Springer- pain chart to emotional distress at the expense of Verlag, 1990:97–131. its sensitivity to physical health” (12, pMS263). (6) Landgraf JM, Nelson EC, Hays RD, et al. Although brief, the charts agree surprisingly Assessing function: does it really make a well with the lengthier MOS scales. The charts difference? A preliminary evaluation of the offer a brief screen that is accurate enough to acceptability and utility of the COOP give a global impression of a patient’s well-being Function Charts. In: WONCA but may miss changes in condition over time. Classification Committee, ed. Functional The ratings of physical function and of pain may status measurement in primary care. New York: Springer-Verlag, 1990:150–161. include a strong measure of psychological reac- (7) Wasson J, Keller A, Rubenstein L, et al. tion to any physical problems present. Benefits and obstacles of health status assessment in ambulatory settings: the Address clinician’s point of view. Med Care The COOP Charts are developed by the Dart- 1992;30(suppl):MS42–MS49. mouth/Northern New England COOP Project, (8) Larson CO, Hays RD, Nelson EC. Do the Dartmouth Medical School, Butler Building, HB pictures influence scores on the Dartmouth 7265, Hanover, NH 03755. Phone (603) 650- COOP Charts? Qual Life Res 1220; fax (603) 650-1331. An informative 1992;1:247–249. (9) Nelson EC, Wasson JH, Johnson DJ, et al. web site on the COOP Charts is at: http://www Dartmouth COOP functional health .dartmouth.edu/~coopproj/more_charts.html. assessment charts: brief measures for clinical practice. In: Spilker B, ed. Quality of life and pharmacoeconomics in clinical References trials. Philadelphia: Lippincott-Raven, (1) Nelson EC, Wasson J, Kirk J, et al. 1996:161–168. Assessment of function in routine clinical (10) Meyboom-de Jong B, Smith RJA. Studies practice: description of the COOP Chart with the Dartmouth COOP Charts in method and preliminary findings. J general practice: comparison with the Chronic Dis 1987;40(suppl 1):55S–63S. Nottingham Health Profile and the General (2) Nelson EC, Landgraf JM, Hays RD, et al. Health Questionnaire. In: WONCA The functional status of patients. How can Classification Committee, ed. Functional it be measured in physicians’ offices? Med status measurement in primary care. New Care 1990;28:1111–1126. York: Springer-Verlag, 1990:132–149. (3) Nelson EC, Landgraf JM, Hays RD, et al. (11) Stavem K, Jodalen H. Reliability and Dartmouth COOP proposal to develop and validity of the COOP/WONCA health demonstrate a system to assess functional status measure in patients with chronic health status in physicians’ offices. Final obstructive pulmonary disease. Qual Life report to Henry J. Kaiser Family Res 2002;11:527–533. Foundation. Hanover, NH: Department of (12) McHorney CA, Ware JE, Jr., Rogers W, et Community and Family Medicine, al. The validity and relative precision of Dartmouth Medical School, 1987. MOS short- and long-form health status (4) Greenfield S, Nelson EC. Recent scales and Dartmouth COOP charts. Med developments and future issues in the use Care 1992;30(suppl):MS253–MS265.  General Health Status and Quality of Life 577 (13) Westbury RC. Use of the Dartmouth assessment field trial. Aust Fam Physician COOP Charts in a Calgary practice. In: 1992;21:255–269. WONCA Classification Committee, ed. (23) Froom J. Preface: WONCA committee on Functional status measurement in primary international classification. Statement on care. New York: Springer-Verlag, functional status assessment Calgary, 1990:166–180. October 1988. In: WONCA Classification (14) Perneger TV, Chamot E, Etter J-F, et al. Committee, editor. Functional status mea- Assessment of the COOP charts with and surement in primary care. New York: without pictures in a Swiss population. Springer-Verlag, 1990:xiii–xvii. Qual Life Res 2000;9:405–414. (24) Scholten JHG, van Weel C. Manual for the (15) Jenkinson C, Mayou R, Day A, et al. use of the Dartmouth COOP functional Evaluation of the Dartmouth COOP charts health assessment charts/WONCA in mea- in a large-scale community survey in the suring functional status in family practice. United Kingdom. J Public Health Med In: Scholten JHG, ed. Functional status 2002;24:106–111. assessment in family practice. Lelystad: (16) Siu AL, Ouslander JG, Osterweil D, et al. Meditekst, 1992:17–50. Change in self-reported functioning in (25) Anderson RT, Aaronson NK, Wilkin D. older persons entering a residential care Critical review of the international facility. J Clin Epidemiol assessments of health-related quality of life. 1993;46:1093–1101. Qual Life Res 1993;2:369–395. (17) Siu AL, Reuben DB, Ouslander JG, et al. (26) Sneeuw KCA, Aaronson NK, Sprangers Using multidimensional health measures in MAG, et al. Value of caregiver ratings in older persons to identify risk of evaluating the quality of life of patients hospitalization and skilled nursing with cancer. J Clin Oncol placement. Qual Life Res 1997;15:1206–1217. 1993;2:253–261. (27) Shigemoto H. A trial of the Dartmouth (18) Coates AK, Wilkin D. Comparing the COOP Charts in Japan. In: WONCA Nottingham Health Profile with the Classification Committee, ed. Functional Dartmouth COOP Charts. In: Scholten status measurement in primary care. New JHG, ed. Functional status assessment in York: Springer-Verlag, 1990:181–187. family practice. Lelystad: Meditekst, (28) Murphy DD, Lam CL. Feasibility and 1992:81–86. acceptability of the COOP/WONCA charts (19) Essink-Bot M-L, Krabbe PFM, Bonsel GJ, for identification of functional limitations et al. An empirical comparison of four in rural patients of the People’s Republic of generic health status measures. The China. Int J Rehabil Res Nottingham Health Profile, the Medical 2001;24:207–219. Outcomes Study 36-item Short-Form (29) Lam CL, van Weel C, Lauder IJ. Can the Health Survey, the COOOP/WONCA Dartmouth COOP/WONCA charts be used charts, and the EuroQol instrument. Med to assess the functional status of Chinese Care 1997;35:522–537. patients? Fam Pract 1994;11:85–94. (20) Jenkinson C, Lawrence K, McWhinnie D, (30) Gilliland SS, Willmer AJ, McCalman R, et et al. Sensitivity to change of health status al. Adaptation of the Dartmouth COOP measures in a randomized controlled trial: Charts for use among American Indian comparison of the COOP charts and the people with diabetes. Diabetes Care SF-36. Qual Life Res 1995;4:47–52. 1998;21:770–776. (21) McHorney CA, Tarlov AR. Individual- (31) Wasson JH, Kairys SW, Nelson EC, et al. patient monitoring in clinical practice: are A short survey for assessing health and available health status surveys adequate? social problems of adolescents. J Fam Pract Qual Life Res 1995;4:293–307. 1994;38:489–494. (22) Landgraf JM, Nelson EC. Dartmouth (32) Palmer RH. Commentary: assessment of COOP primary care network. Summary of function in routine clinical practice. J the WONCA/COOP international health Chronic Dis 1987; 40(suppl 1):65S–69S. 578 Measuring Health tions; the final rating is a question concerning Single-Item Health Indicators overall health. Although no single author devel- (Various authors, circa 1965 oped these methods, much of the formal exami- onward) nation of the various scale formats was Purpose undertaken 30 years ago by Frank M. Andrews These methods provide subjective summary indi- and various associates (1; 7–9). The first four cators of a range of aspects of health: health or scales shown below were those found to be most quality of life in general, life satisfaction, or feel- valid in his analyses. They may be applied in rat- ings about specific aspects of health. Commonly ing QoL as a whole—“How happy are you these used in population surveys, they can also be used days?”—or to health in general, as well as to in clinical settings and, indeed, often form an other themes such as income or housing. In each opening question in a clinical interview: “How case, the scales assess the affective component of are you today?” They are also often included as QoL rather than the physical and social condi- a summary question in generic health measures. tions in which a person lives. According to the question phrasing, the scales may refer to the Conceptual Basis present or to the past or may be used to express Single-item summary ratings of health or quality the respondent’s hopes for the future. of life (QoL) require that the respondent inte- The four response scales tested by Andrews grate many aspects of his or her current experi- are: ence. Hence, they stress the subjective evaluative nature of QoL: “The quality of life is not just a 1. The Delighted-Terrible (D-T) Scale. This is matter of the conditions of one’s physical, inter- a seven-point scale ranging from delighted personal and social setting but also of how these to terrible; two alternative formats are are judged and evaluated by oneself and others” shown in Exhibit 10.13. Respondents are (1, p12). There is an extensive history of using given an instruction such as, “We want to single-item scales in the social sciences to rate find out how you feel about various parts themes such as life satisfaction, positive affect, of your life, and life in this country as you anxiety, and depression (2); this has subse- see it. Please indicate the feelings you have quently been extended to include health in gen- now—taking into account what has hap- eral. From this it has become clear that people pened in the last year and what you expect possess insights into their health that go beyond in the near future . . . How do you feel the reach of objective clinical ratings; simple about __________? (your health, job, self-ratings consistently explain variance in mor- etc.)” (9, p5). An alternative wording runs tality even after controlling for conventional risk “Below are some words and phrases that factors and other clinical information (3). A people use to identify various features of helpful review of the potential value of simple their lives. Each feature title has a scale self-ratings of health, and of the pitfalls in de- beside it that runs from ‘Terrible’ to ‘De- signing them, is given by Knäuper and Turner lighted’ in seven steps . . . Please check the (4). Meanwhile, Verbrugge has set out the crite- number on the scale beside each feature ria that a successful single-item indicator would that comes closest to describing how you need to fulfil (5). feel about __________” (10, p394). The D-T Scale is used to record responses in Description Lehman’s Quality of Life Interview Self-ratings of health, life satisfaction, and QoL (QOLI) (11) and also in the Quality of are frequently made on single-item response Life Scale for schizophrenia (12). scales that vary in design and format; early ex- 2. The Faces Scale. This is a seven-point scale amples included the studies by Campbell et al. consisting of stylized faces (Exhibit (6). The first five ratings reviewed are response 10.14). Each face consists of a circle with scales that may be used with a variety of ques- eyes that do not change and a mouth that  General Health Status and Quality of Life 579 Exhibit 10.13 The Delighted-Terrible Scale 7 6 5 4 3 2 1 MIXED MOSTLY (ABOUT EQUALLY MOSTLY DELIGHTED PLEASED UNHAPPY TERRIBLE SATISFIED SATISFIED AND DISSATISFIED DISSATISFIED) Adapted from Andrews FM, Withey SB. Social indicators of well-being: Americans' perceptions of life quality. New York: Plenum, 1976:371. With permission 1 2 3 4 5 6 7 Terrible Very Dissatisfying Mixed Satisfying Very Delightful Dissatisfying Satisfying Adapted from Michalos A. Satisfaction and happiness. Social Indicators Research 1980;8:394. varies from a smile of almost a half-circle expect to have.” Respondents are told, to a similar half-circle upside down, repre- “Here is a picture of a ladder. At the bot- senting gloom. Respondents were told, tom of this ladder is the worst situation “Here are some faces expressing various you might reasonably expect to have. At feelings . . . Which face comes closest to the top is the best you might expect to expressing how you feel about your have. The other rungs are in between . . . __________?” (9, p5). The faces scale has Where on the ladder is your (health, job, also been adapted to depict levels of pain marriage, home)? On which rung would (13, Figure 1). you put it?” The scale is often termed 3. Ladder Scale. This scale (shown in Exhibit “self-anchoring” because ratings are made 10.15) is drawn as a ladder with nine relative to each person’s conception of her rungs, derived from the ladder scale of own minimum and maximum life satisfac- Hadley Cantril (14). It has been used fre- tion. quently and the instructions show slight 4. The Circles Scale. The nine circles in this variations (15–17). The phrasing of An- scale are divided into eight slices, each drews and Withey is typical (1): the top containing a plus or minus sign. Circles rung is labeled “Best I could expect to are ordered so that they contain progres- have” and the bottom rung “Worst I could sively more pluses and fewer minuses. The Exhibit 10.14 The Faces Scale Reproduced from Andrews and Withey, page 376. With kind permission of Springer Science and Business Media. 580 Measuring Health Exhibit 10.15 The Ladder Scale Reproduced from Andrews and Withey, page 370. With kind permission of Springer Science and Business Media. instruction reads, “Here are some circles tinuum (e.g., health in general), running that we can imagine represent the lives of from the worst possible health to perfect different people. Circle 0 has all minuses health. The line may be marked at each in it, to represent a person who has all bad end with labels that indicate the range be- things in his or her life. Circle 8 has all ing considered; examples were given by pluses in it, to represent a person who has Cella and Perry (18). The respondent all good things in his or her life. Other cir- places a mark on the line to indicate his or cles are in between. Which circle comes her current health status. To save space on closest to matching how you feel about a page, a horizontal format is generally ___________?” (9). The Circles Scale is used, but an example that uses a vertical shown in Exhibit 10.16. format forms part of the EuroQol Quality of Life Scale (see Exhibit 10.38). To pro- In addition to the scales tested by Andrews, visual duce a smooth response distribution, the analogue response scales have been widely used VAS generally does not include numbers in health research. along the scale. This is because people of- ten favor numbers ending in zero or five, 5. Visual Analogue Scale (VAS). A 10- which produces a stepped distribution of centimeter line is used to represent a con- responses (19). Exhibit 10.16 The Circles Scale Reproduced from Andrews and Withey, page 370. With kind permission of Springer Science and Business Media.  General Health Status and Quality of Life 581 The final example of a single-item health mea- spondents provided an answer on retest that was sure includes a general question and a response identical or immediately adjacent to their previ- scale. ous answer (1). Two-year test-retest reliability in a community sample was 0.40 for the Ladder 6. Summary Self-Rating of Health. Wording Scale and 0.41 for an 11-point satisfaction rating varies, but a typical example would be “In scale. For respondents who reported no major general, would you say your health today changes in life circumstances, coefficients were is: Excellent? Very good? Good? Fair? 0.47 and 0.43, respectively (21). Test-retest reli- Poor?” (19). Common alternatives are ability for the VAS applied to rating various as- shown in Exhibit 10.17; more are re- pects of mood ranged from 0.5 (anxiety) to 0.83 ported by Idler and Benyamini (3). (distress) (18, p830). The retest intraclass corre- lation was 0.87 for the VAS applied to global Change is generally measured using differ- quality of life (22, p315). Kappa estimates of ence scores, but if the range of improvement is test-retest reliability for summary self-rating attenuated where the initial score is already ex- questions (with various alternative phrasings) treme, scores can be expressed as a proportion were derived from a small study of 52 elderly of the possible change (F. Andrews, personal people. Coefficients were kappa = 0.86 (overall communication, 1984). In place of a simple ordi- QoL), 0.69 (physical health status), and 0.84 nal scoring for the summary self-rating question, (level of happiness) (23, Table 5). Diehr et al. proposed scores of 95 for “excel- Agreement between self-ratings and judg- lent,” 90 for “very good,” 80 for “good,” 30 for ments by acquaintances averaged 0.33 for 16 ar- “fair,” and 15 for “poor.” They then scored eas rated on the D-T Scale. Correlations between dead as zero, thereby allowing the retention of alternative forms of the question stem (both us- people who died in a study (20). ing the D-T Scale) averaged 0.71 over 16 areas (24, Figure 1). Lehman et al. reported internal Reliability consistency reliabilities for the D-T Scales be- Retest reliability estimates from various surveys tween 0.74 and 0.87 on a sample of patients using scales 1 to 4 were combined by Andrews, with long-term mental illness (25, p1272). The who estimated an average test-retest reliability intraclass correlation between an interview and for each scale (applied twice in the same inter- a questionnaire version of the VAS was 0.87 (22, view) of about 0.70. He noted that 92% of re- p315). Exhibit 10.17 Examples of Single, Summary Item Health Scales, Showing Source References “In general, would you say your health is: Excellent, Very Good, Good, Fair, Poor?” (26; 27) “All in all, would you say your health is: Excellent, Good, Fair, Poor?” (28) “Do you think your health is [Good, Fair, Poor] for your age?” (29) “How is your health: Very poor, Poor, Neither poor nor good, Good, Very good?” (30) “Do you consider yourself to be: Very healthy, Fairly healthy, Sick, Very sick?” (31) “Would you say your health in general is: Excellent, Very Good, Good, Fair, Poor?” (32) “How would you rate your health at the present time: Excellent, Good, Fair, Poor, Bad?” (33; 34) “For your age, in general, would you say your health is: Excellent, Good, Fair, Poor, Bad?” (35) “How is your health, compared with others your age: Better, Same, Worse?” (36) “Taking everything into consideration, how would you rate your health in general these days? Very good, fairly good, not so good, poor, or very poor?” “How would you rate your health compared to others your age: Excellent (1) – Poor (5)” (37) 582 Measuring Health Validity with Spielberger’s state anxiety scale (2, Table Validity analyses for the first four formats just 1). described used a multimethod-multitrait study that assessed six aspects of well-being using the visual analogue scale. VAS summary ratings four methods (9). The results show median va- have been compared to multi-item measures, of- lidity coefficients of 0.82 for the D-T Scale, 0.82 ten showing close agreement. Cella and Perry, for the Faces Scale, 0.70 for the Ladder Scale, for example, reported a correlation of 0.58 be- and of 0.80 for the Circles Scale (9, Table 1). tween a VAS depression rating and the Beck De- The validity coefficients remained consistent pression Inventory, and a correlation of 0.52 when different aspects of QoL were assessed by between a VAS rating of anxiety and the Spiel- the same method (9). Several other single-item berger state anxiety score; an overall VAS mea- scales tested by Andrews showed much lower sure of distress correlated 0.63 with the Profile validity. of Mood States score (18, Table 1). A Canadian survey reported correlations of 0.63 between a the ladder scale. Although frequently used, VAS general health rating and both the Health this measure is rarely validated. Using this for- Utilities Index (HUI) and the EuroQol EQ-5D mat, Palmore and Kivett showed considerable (19, Table 8). The area under the ROC curve for stability in life satisfaction in a longitudinal the VAS question was virtually identical to those study of 378 community residents aged between obtained for the HUI and the EQ-5D (19, Table 46 and 70 (15). Self-rated health levels formed 16). Using the Beck Depression inventory, Kill- the strongest predictors of overall life satisfac- gore rated 142 students as depressed or not. A tion, accounting for two thirds of the explained VAS depression rating had a 77% agreement, variance (15; 16). Atkinson showed significant not significantly different from that achieved by associations between the Ladder Scale results the 15-item depression scale of the Profile of and life events (21). Sarason et al. found that the Mood States (79%) (40, Table 2). In a study of ladder scale explained 13% of the variance in patients with breast cancer, a VAS assessment scores representing satisfaction with social sup- of mood performed as well as an adjective port, and 19% of variance in self-rated availabil- check-list (41). ity of social support (38, Tables 2 and 3). The Using a community sample of 1,477 respon- ladder approach has been incorporated into a dents, Bélanger et al. compared a visual ana- Self-Anchoring Self-Esteem Scale (39). logue rating of “your health today” to a five-point scale rating (19). The correlation was the delighted-terrible scale. Lehman et al. 0.76, whereas the VAS correlated 0.69 with a showed this scale format was capable of showing summary question concerning “your health in consistent contrasts between patients with men- general.” A VAS rating of global QoL correlated tal disorders living in the community and popu- 0.67 with the health perceptions scale of the lation reference values derived from Andrews’s Short-Form-20 health survey, and 0.63 with the data (25). They also presented correlations be- mental health scale (22, Table 3). tween the subjective ratings and objective life In a study of responsiveness, the standardized circumstances, with coefficients falling in the response mean for a VAS recording global qual- range 0.07 to 0.57 (25, Table 3). Headey et al. ity of life was evaluated for patients with cancer reported a correlation of 0.67 between the D-T before and following surgery (22). The VAS, at Scale (rating “your life as a whole”) and a five- 0.47, was less responsive than the physical and item life satisfaction scale. The correlations with role scales of the SF-20, which ranged from 0.94 Bradburn’s scale were lower, at 0.35 for the pos- to 1.24. However, it was more responsive than itive affect scale, and −0.40 for the negative the SF-20 mental health and pain scales (22, scale. The D-T Scale correlated −0.50 with the Table 4). In a similar study with patients with 12-item General Health Questionnaire, −0.43 breast cancer, effect size statistics for the VAS (at with the Beck Depression Inventory, and −0.33 1.31 and 0.85) were comparable with those for  General Health Status and Quality of Life 583 the Functional Living Index-Cancer scale and for relations with measures of life satisfaction (46) the EuroQol EQ-5D (42, Table 2). (see Table 5.2 in the review of Neugarten’s scale). Correlations ranged from 0.40 to 0.47 the summary self-rating question. Numerous suggesting that less than one quarter of the vari- longitudinal studies have confirmed remarkably ance is shared by the two approaches. However, strong associations between self-rating scores single-item questions that are phrased to capture and subsequent mortality, even after controlling mood can show high correlations with more de- for a range of other risk factors (3; 32; 34; 36; tailed scales: a single-item depression question 43). Idler and Benyamini’s summary is impres- correlated 0.71 to 0.78 with the 21-item Beck sive: in 23 of 27 studies, the overall self-rating Depression Inventory in a study of 812 patients question showed an independent contribution to (47). With terminally ill patients, a single ques- explaining variance in mortality after control- tion “Are you depressed?” (sensitivity and speci- ling for age, socioeconomic status, and, in sev- ficity 100%) outperformed a VAS (sensitivity eral studies, chronic conditions and selected 72% and specificity 50%) and also the short- medical risk factors (3). Odds ratios for mortal- form Beck instrument (sensitivity 79% and ity typically lay between two and four for those specificity 71%) (48, Table 1). From an interna- who reported being in poor health (3, Table 1). tional review of the WHOQOL scale For example, the adjusted odds ratios for pre- (N = 11,830), Skevington et al. reported a corre- dicting four-year mortality in two samples (aged lation of 0.43 between the summary health ques- 65 and older) were 4.8 and 5.3 for men, and 3.0 tion and scores on the physical scale of the and 3.2 for women (33, Table 6). More extreme WHOQOL; the correlation was 0.17 for the values include a relative risk of 5.1 for nine-year psychological scale, 0.07 for the social health mortality among women who rated their health scale and 0.06 for the environmental scale (49, as poor (compared to those who rated it as ex- Table 4). In a study of treatment for intermittent cellent) (28). claudication, change scores before and after Gold et al., in an analysis of the National treatment were calculated for several QoL scales Health and Nutrition Examination Survey, and then compared. The intraclass correlation found that self-ratings were associated with age, between change scores measured on the rating race, and levels of education and income (44, scale and the Health Utilities Index (HUI3) was Table 1). Self-rated health predicted subsequent 0.41 and its correlation with the EQ-5D was mortality more strongly in people younger than comparable, at 0.39. The correlation between 65 years of age than older; it also predicted hos- the HUI3 and the EQ-5D was, however, much pital admissions over the subsequent four years. lower, at 0.30 (50, Table 6). Indeed, ratios of mean scores on the self-rating The single item question “In general, would question for various outcomes were somewhat you say your health is: Excellent, Very Good, stronger than those obtained for the Health Util- Good, Fair, Poor?” was used in the Short-Form- ities Index (44, Table 4). Idler and Angel also 36 and SF-12 measures and they have undergone found self-rated health to be less predictive of intensive testing. For example, in the Interna- mortality among elderly people (32). Jenkinson tional Quality of Life Assessment study, scaling et al. commented further on this, suggesting that analyses have shown nonlinearities in the re- elderly patients may have limited expectations sponse scale. The interval between “Excellent” for health improvement and so rate their health and “Very Good” is about half that between on a summary question relatively highly com- “Good” and “Fair” (26, p31). From these analy- pared with the impression gained from before ses, Ware suggested that “Excellent” be scored and after treatment changes on a scale such as 5.0; “Very Good” as 4.4; “Good” as 3.4; “Fair” the SF-36 (45). as 2.0 and “poor” as 1.0 (26, Table 6.1). Single-item overall summary questions may Overall, the impression is that the single-item correlate less strongly with measures of subjec- question offers a remarkably broad-ranging as- tive well-being. An example is given by the cor- sessment of health. Using an ordered logit re- 584 Measuring Health gression analysis, Schaafsma and Osoba showed A consistent finding is that these very simple that responses to the single question as used in summary health ratings hold surprising predic- the EORTC QLQ-C30 captured much wider tive validity, even for hard outcomes such as mor- sources of variance, physical, emotional and tality, and even when conventional risk factors cognitive, than the Karnofsky Performance Sta- are controlled (3; 33). Idler and Benyamini have tus scale did (51). discussed explanations, which include the possi- bility that insightful self-assessments may reflect Alternative Forms the impact of as yet undiagnosed disease; they The single-item self-rating of health has been in- may reflect health trajectories and not merely corporated into a measure called the Health and current health status; they may also reflect a per- Activities Limitations Index (HALex), in which son’s inner and social coping resources, or there self-rated health is combined with role limita- may be a self-fulfilling prophecy in which a per- tions presented on two axes, and utility scores son’s self-rating may influence their health behav- are awarded to the resulting combinations of rat- iors and determination to remain healthy (3). ings. This index can commonly be derived from Problems in using single-item scales include a data collected in existing surveys (52). lack of clarity over the definitions of the end- points of scales. Paterson, for example, found Commentary that people may vary in how they interpret “Best Single-item health measures have been tested in imaginable health state”—either the best the other areas, such as in screening for depression person could be in their present situation and (53) or physical activity levels (54), and the age, or else their health when they were young reader may wish to review the numerical and vi- and fit. It can also be difficult for respondents to sual analogue pain rating scales reviewed in translate multiple problems into a single, average Chapter 9. These scales are attractive for sur- rating (57, pp875–876). Care is required espe- veys: they are simple to apply and the nonverbal cially in using summary self-ratings with elderly format avoids translation issues in cross-cultural patients, for whom discrepancies with other research. Nonverbal scales can also work well health measures may occur due to their dimin- with children and others who would have diffi- ished expectations; this may occur especially culty completing a questionnaire. A Faces Scale when the question is phrased in terms of com- has, for example, been used in measuring pain in parisons with other people their age. children aged 3 to 15 years (55) and in measur- Single-item measures frequently perform al- ing anxiety in critically ill patients (56). The pic- most as well as multi-item scales, raising the ob- torial scales also have the virtue of being direct, vious question as to the cost-effectiveness of the of tapping into the feelings associated with QoL latter. In a discussion of single-item versus scales without passing through the intermediary of lan- measures of quality of life, Sloan et al. argued guage. Pictorial devices are commonly used in that the choice depends largely on the concep- qualitative research and should, perhaps, be used tual fit between the measure and the anticipated more frequently in the quantitative tradition. outcome of an intervention; scales offer detail at Measures of satisfaction, for example, can use the cost of respondent burden, while single-item pictorial formats: one scale measuring audience measures offer simplicity at the cost of detail satisfaction with a presentation used silhouette (58). For example, in studies of palliative care figures to represent a four-point scale running patients, or children, the simplicity of single- from “hats off ” (with a picture of a person rais- item scales may commend their use. ing their hat) through “thumbs up” and “so-so” (a person with hunched shoulders and splayed hands) to “thumbs down.” The VAS format and References perhaps the ladder scale arguably produce an (1) Andrews FM, Withey SB. Social indicators interval-scaled measurement, whereas the others of well-being: Americans’ perceptions of provide ordinal measures. life quality. New York: Plenum, 1976.  General Health Status and Quality of Life 585 (2) Headey B, Kelley J, Wearing A. Dimensions (17) Brown JS, Rawlinson ME, Hilles NC. Life of mental health: life satisfaction, positive satisfaction and chronic disease: affect, anxiety and depression. Soc Indicat exploration of a theoretical model. Med Res 1993;29:63–82. Care 1981;19:1136–1146. (3) Idler EL, Benyamini Y. Self-rated health (18) Cella DF, Perry SW. Reliability and and mortality: a review of twenty-seven concurrent validity of three visual-analogue community studies. J Health Soc Behav mood scales. Psychol Rep 1997;38:21–37. 1986;59:827–833. (4) Knäuper B, Turner PA. Measuring health: (19) Bélanger A, Berthelot J-M, Guimond E, et improving the validity of health al. A head-to-head comparison of two assessments. Qual Life Res 2003;12(suppl generic health status measures in the house- 1):81–89. hold population: McMaster Health Utilities (5) Verbrugge LM. A global disability Index (Mark3) and the EQ-5D. Statistics indicator. J Aging Stud 1997;11:337–362. Canada Report. Ottawa, (6) Campbell A, Converse PE, Rodgers WL. Ontario;2002:1–62. The quality of American life: perceptions, (20) Diehr P, Patrick D, Spertus J, et al. evaluations, and satisfactions. New York: Transforming self-rated health and the SF- Russell Sage, 1976. 36 scales to include death and improve (7) Andrews FM. Social indicators of perceived interpretability. Med Care life quality. Soc Indicat Res 2001;39:670–680. 1974;1:279–299. (21) Atkinson T. The stability and validity of (8) Andrews FM, McKennell AC. Measures of quality of life measures. Soc Indicat Res self-reported well-being: their affective, 1982;10:113–132. cognitive and other components. Soc (22) de Boer AEGM, van Lanschot JJB, Indicat Res 1980;8:127–155. Stalmeier PFM, et al. Is a single-item visual (9) Andrews FM, Crandall R. The validity of analogue scale as valid, reliable and measures of self-reported well-being. Soc responsive as multi-item scales in measur- Indicat Res 1976;3:1–19. ing quality of life? Qual Life Res (10) Michalos AC. Satisfaction and happiness. 2004;13:311–320. Soc Indicat Res 1980;8:385–422. (23) Raina P, Bonnett B, Waltner-Toews D, et (11) Lehman AF. The well-being of chronic al. How reliable are selected scales from mental patients: assessing their quality of population-based health surveys? An life. Arch Gen Psychiatry analysis among seniors. Can J Public 1983;40:369–373. Health 1999;90:60–64. (12) Heinrichs DW, Hanlon TE, Carpenter WT. (24) Crandall R. Validation of self-report mea- The Quality of Life Scale: an instrument sures using ratings by others. Sociol for rating the schizophrenic deficit Methods Res 1976;4:380–400. syndrome. Schizophr Bull (25) Lehman AF, Ward NC, Linn LS. Chronic 1984;10:388–398. mental patients: the quality of life issue. (13) Herr KA, Mobily PR, Kohout FJ, et al. Am J Psychiatry 1982;139:1271–1276. Evaluation of the faces pain scale for use (26) Ware JE, Jr., Kosinski M, Turner-Bowker with the elderly. Clin J Pain DM, et al. How to score Version 2 of the 1998;14:29–38. SF-12 Health Survey. Lincoln, Rhode (14) Cantril H. The pattern of human concerns. Island: QualityMetric Inc., 2002. New Brunswick, New Jersey: Rutgers (27) Ware JE, Jr., Snow KK, Kosinski M, et al. University Press, 1965. SF-36 Health Survey: manual and (15) Palmore E, Kivett V. Change in life interpretation guide. Boston: The Health satisfaction: a longitudinal study of persons Institute, New England Medical Center, aged 46–70. J Gerontol 1977;32:311– 1993. 316. (28) Kaplan GA, Camacho T. Perceived health (16) Palmore E, Luikart C. Health and social and mortality: a nine-year follow-up of the factors related to life satisfaction. J Health Human Population Laboratory cohort. Am Soc Behav 1972;13:68–80. J Epidemiol 1983;117:292–304. 586 Measuring Health (29) Jagger C, Clarke M. Mortality risk in the versus a multi-item scale: assessment of elderly: five-year follow-up of a total emotional well-being in an international population. Int J Epidemiol adjuvant breast cancer trial. Med Care 1988;17:111–114. 1996;34:234–248. (30) WHOQOL Group. The World Health Or- (42) Conner-Spady B, Cumming C, Nabholtz J- ganization Quality of Life Assessment M, et al. Responsiveness of the EuroQol in (WHOQOL): development and general breast cancer patients undergoing high dose psychometric properties. Soc Sci Med chemotherapy. Qual Life Res 1998;46:1569–1585. 2001;10:479–486. (31) Borawski EA, Kinney JM, Kahana E. The (43) Idler EL, Russell LB, Davis D. Survival, meaning of older adults’ health appraisals: functional limitations, and self-rated health congruence with health status and in the NHANES I Epidemiologic Follow- determinants of mortality. J Gerontol Soc up Study, 1992. First National Health and Sci 1996;51B:S157–S170. Nutrition Examination Survey. Am J (32) Idler EL, Angel RJ. Self-rated health and Epidemiol 2000;152:874–883. mortality in the NHANES-I Epidemiologic (44) Gold M, Franks P, Erickson P. Assessing Follow-up Study. Am J Public Health the health of the nation. The predictive 1990;80:446–452. validity of a preference-based measure and (33) Idler EL, Kasl SV, Lemke JH. Self- self-rated health. Med Care evaluated health and mortality among the 1996;34:163–177. elderly in New Haven, Connecticut, and (45) Jenkinson C, Jenkinson D, Shepperd S, et Iowa and Washington counties, Iowa, al. Evaluation of treatment for congestive 1982-1986. Am J Epidemiol heart failure in patients aged 60 years and 1990;131:91–103. older using generic measures of health (34) Idler EL, Kasl S. Health perceptions and status (SF-36 and COOP Charts). Age survival: do global evaluations of health Ageing 1997;26:7–13. status really predict mortality? J Gerontol (46) Lohmann N. Correlations of life 1991;46:S55–S65. satisfaction, morale and adjustment mea- (35) Mossey JM, Shapiro E. Self-rated health: a sures. J Gerontol 1977;32:73–75. predictor of mortality among the elderly. (47) McKenzie N, Marks I. Quick rating of Am J Public Health 1982;72:800–808. depressed mood in patients with anxiety (36) Ho SC. Health and social predictors of disorders. Br J Psychiatry mortality in an elderly Chinese cohort. Am 1999;174:266–269. J Epidemiol 1991;133:907–921. (48) Chochinov HM, Wilson KG, Enns M, et (37) Chipperfield JG. Incongruence between al. “Are you depressed?” Screening for health perceptions and health problems. J depression in the terminally ill. Am J Aging Health 1993;5:475–496. Psychiatry 1997;154:674–676. (38) Sarason IG, Sarason BR, Shearin EN. (49) Skevington SM, Lotfy M, O’Connell KA. Social support as an individual difference The World Health Organization’s variable: its stability, origins, and relational WHOQOL-BREF quality of life aspects. J Pers Soc Psychol assessment: psychometric properties and 1986;50:845–855. results of the international field trial. Qual (39) Carpenter JS. Applying the Cantril Life Res 2004;13:299–310. methodology to study self-esteem: (50) Bosch JL, Hunink MGM. Comparison of psychometrics of the Self-Anchoring Self- the Health Utilities Index Mark 3 (HUI3) Esteem Scale. J Nurs Meas and the EuroQol EQ-5D in patients treated 1996;4:171–189. for intermittent claudication. Qual Life Res (40) Killgore WDS. The visual analogue mood 2000;9:591–601. scale: can a single-item scale accurately (51) Schaafsma J, Osoba D. The Karnofsky Per- classify depressive mood state? Psychol Rep formance Status Scale re-examined: a cross- 1999;85:1238–1243. validation with the EORTC-C30. Qual Life (41) Hurny C, Bernhard J, Coates A, et al. Res 1994;3:413–424. Responsiveness of a single-item indicator (52) Bradley CJ, Kroll J, Holmes-Rovner M. The  General Health Status and Quality of Life 587 Health and Activities Limitation Index in assessing need for treatment, and monitoring patients with acute myocardial infraction. J progress. An assessment instrument needs to be Clin Epidemiol 2000;53:555–562. more detailed than a descriptive one, and a mon- (53) Avasarala JR, Cross AH, Trinkaus K. itoring instrument, in addition, must be sensitive Comparative assessment of Yale single to change (2, p1854). question and Beck Depression Inventory scale in screening for depression in multiple Description sclerosis. Mult Scler 2003;9:307–310. (54) Prochaska JJ, Sallis JF, Long B. A physical The FSQ was adapted from existing scales (in- activity screening measure for use with cluding the Sickness Impact Profile and the adolescents in primary care. Arch Pediatr RAND instruments) by researchers at the Beth Adolesc Med 2001;155:554–559. Israel Hospital in Boston and at the University of (55) McGrath PA, de Veber LL, Hearn MT. California Los Angeles. It covers physical func- Multidimensional pain assessment in tion (three activities of daily living [ADL] and children. In: Fields HL, Dubner R, Cervero six instrumental activities of daily living [IADL] F, eds. Advances in pain research and items), psychological function (five items), work therapy. Vol. 9. New York: Raven Press, performance (six items), social activity (three 1985:387–393. items), and quality of social interaction (five (56) McKinley S, Coote K, Stein-Parbury J. items). Four-, five- and six-point ratings are Development and testing of a faces scale for the assessment of anxiety in critically ill used, referring to health in the past month. Six patients. J Adv Nurs 2003;41:73–79. additional questions cover work status, bed dis- (57) Paterson C. Seeking the patient’s ability days, activity reductions, satisfaction perspective: a qualitative assessment of with sexual relationships, interpersonal relation- EuroQol, COOP-WONCA charts and ships, and feelings about health. The question- MYMOP. Qual Life Res naire takes about 15 minutes to complete (2, 2004;13:871–881. p1855). (58) Sloan JA, Aaronson N, Cappelleri JC, et al. The 34 items are scored by computer to pro- Assessing the clinical significance of single vide six summary scale scores and six single-item items relative to summated scores. Mayo scores (see Exhibit 10.18). The scale scores are Clin Proc 2002;77:479–487. standardized to a 0 to 100 range, with higher scores indicating better function. The computer The Functional Status prints a one-page summary report of responses, Questionnaire indicating areas that are of potential clinical (Alan M. Jette, 1986) concern based on cutting-points determined by a panel of experienced clinicians (1; 2). An illus- Purpose tration of routine clinical use of the FSQ is given The Functional Status Questionnaire (FSQ) is a by Rubenstein et al. (3). An illustration of scor- brief, self-administered instrument that assesses ing is available on the Web, at www.cebp.nl/ physical, psychological, and social role function- media/m410.pdf. The site also offers guides to ing in ambulatory care patients. It was designed interpreting scores in terms of so-called warning as a clinical tool to screen for disability and to zones that can be represented as cutting-points monitor changes in function among people seen (with lower scores indicating lower health). For at primary care practices (1; 2). basic ADL, the threshold is 87/88; intermediate ADL, 77/78; mental health, 70/71; work perfor- Conceptual Basis mance and social activities, 78/79; and quality of After the World Health Organization (WHO), interactions, 69/70. Jette defined functional disability in terms of de- parture from normal performance. Three com- Reliability mon purposes of monitoring functional Alpha internal consistency scores, reported from disability include describing a patient’s status, several studies on diverse groups of patients, Exhibit 10.18 The Functional Status Questionnaire Note: An asterisk indicates items for which scores are reversed. Category Item Physical function: During the past month have you had difficulty: Basic activities of daily living (ADL) Taking care of yourself, that is, eating, dressing or bathing? Moving in and out of a bed or chair? Walking indoors, such as around your home? Intermediate ADL Walking several blocks? Walking one block or climbing one flight of stairs? Doing work around the house such as cleaning, light yard work, home maintenance? Doing errands, such as grocery shopping? Driving a car or using public transportation? Doing vigorous activities such as running, lifting heavy objects or participating in strenuous sports? Responses: usually did with no difficulty (4), some difficulty (3), much difficulty (2), usually did not do because of health (1), usually did not do for other reasons (0). Psychological function: During the past month: Mental health Have you been a very nervous person? Have you felt calm and peaceful?* Have you felt downhearted and blue? Were you a happy person?* Did you feel so down in the dumps that nothing could cheer you up? Responses: all of the time (1), most of the time (2), a good bit of the time (3), some of the time (4), a little of the time (5), none of the time (6). Social/role function: During the past month have you: Work performance (for those employed, Done as much work as others in similar jobs?* during the preceding month) Worked for short periods of time or taken frequent rests because of your health? Worked your regular number of hours?* Done your job as carefully and accurately as others with similar jobs?* Worked at your usual job, but with some changes because of your health? Feared losing your job because of your health? Responses: all of the time (1), most of the time (2), some of the time (3), none of the time (4). Social activity Had difficulty visiting with relatives or friends? Had difficulty participating in community activities, such as religious services, social activities, or volunteer work? Had difficulty taking care of other people such as family members? Responses: Usually did with no difficulty (4), some difficulty (3), much difficulty (2), usually did not do because of health (1), usually did not do for other reasons (0). Quality of interaction Isolated yourself from people around you? Acted affectionate toward others?* Acted irritable toward those around you? Made unreasonable demands on your family and friends? Gotten along well with other people?* Responses: all of the time (1), most of the time (2), a good bit of the time (3), some of the time (4), a little of the time (5), none of the time (6). 588  General Health Status and Quality of Life 589 Exhibit 10.18 Single item questions: Which of the following statements best describes your work situation during the past month? Responses: working full-time; working part-time; unemployed; looking for work; unemployed because of my health; retired because of my health; retired for some other reason. During the past month, how many days did illness or injury keep you in bed all or most of the day? Response: 0–31 days. During the past month, how many days did you cut down on the things you usually do for one-half day or more because of your own illness or injury? Response: 0–31 days. During the past month, how satisfied were you with your sexual relationships? Responses: very satisfied; satisfied; not sure; dissatisfied; very dissatisfied; did not have any sexual relationships. How do you feel about your own health? Responses: very satisfied; satisfied; not sure; dissatisfied; very dissatisfied. During the past month, about how often did you get together with friends or relatives, such as going out together, visiting in each other’s homes, or talking on the telephone? Responses: every day; several times a week; about once a week; two or three times a month; about once a month; not at all. Reproduced from Jette AM, Davies AR, Cleary PD, et al. The Functional Status Questionnaire: reliability and validity when used in primary care. J Gen Intern Med 1986;1:144. With permission. have been summarized by Cleary and Jette (4). management by physicians who received, or did Studies of ambulatory patients have yielded al- not receive, FSQ profiles of the patients they pha coefficients for the basic ADL scale ranging were seeing. Of these physicians, 43% reported from 0.77 to 0.84, whereas values for the IADL that the FSQ information led them to alter their scale ranged from 0.82 to 0.89 (4, Table 1). Al- therapy; 97% reported that the information was pha values for the mental health scale ranged useful (3, Table 4). There were, however, no sig- from 0.77 to 0.88; those for the social activity nificant differences in health outcomes between scale ranged from 0.65 to 0.83, whereas results the experimental and control groups. for the quality of social interaction scale seem Jette reported the correlations among the six lower, at 0.42 to 0.79 (4, Table 1). Equivalent scales but did not present a factor analysis. The results from a study of 2,484 surgical patients correlations ranged from 0.14 to 0.75, with co- provided alpha coefficients for the ADL scale efficients above 0.62 among the three scales re- ranging from 0.69 to 0.82; those for the IADL quiring physical ability: basic ADL, IADL, and questions ranged from 0.71 to 0.93 (4, Table 2). social activity (2, Table 2). Similar correlations In a study of patients with Parkinson’s disease, were reported from a Swedish study (7, Table the overall alpha was 0.92; section values ranged VI). from 0.68 (social activity) to 0.87 (IADL) (5, Jette et al. present construct validity correla- p284). Finally, Reuben et al. reported an alpha tions for the FSQ scale scores and criterion vari- of 0.80 for the ADL scale and 0.81 for the IADL ables (1, Table 4). Illustrative results include: the (6, Table 2). The basic ADL scores correlated ADL scale of the FSQ correlated −0.33 with re- 0.73 with the IADL scores in a study of healthy stricted activity days and −0.44 with limitations seniors (6, p19). in work role. Equivalent correlations for the No information is available on test-retest sta- IADL scale were −0.36 and −0.61. Cleary and bility. Jette report correlations ranging from 0.54 to 0.60 between the IADL scale and a 0 to 100 Validity global health rating question; equivalent correla- As an indicator of the acceptability of the FSQ to tions for the mental health scale ranged from physicians, a randomized trial compared patient 0.40 to 0.51 (4, Tables 3 and 4). The mental 589 590 Measuring Health health scale correlated 0.45 with satisfaction placement patients and compared with that of the with health; quality of interaction correlated SIP, SF-36, and shortened Arthritis Impact Mea- 0.26 with number of social contacts. Equivalent surement Scales. The FSQ overall score ranked correlations were reported from the Swedish second in sensitivity to change (9, Table 4). study: the number of restricted activity days cor- related −0.42 with the ADL score and −0.44 Alternative Forms with the IADL score; the frequency of social The Quality of Life Instruments Database contacts correlated 0.29 with the FSQ quality of (QOLID) Web site lists translations into Dutch, interaction scale and 0.35 with the social activi- English for the United Kingdom, French for ties score (7, Table VI). Correlations of the FSQ France and Canada, German (11), Italian, and scores with the Current Health Perceptions scale Swedish (7). (http://www.qolid.org/public/FSQ of the SF-20 ranged from 0.30 to 0.46 (8, p667). .html). A correlation of 0.82 was obtained between the global dimension and the Sickness Impact Pro- Commentary file, and of 0.83 with the SF-36 in a study of 106 This measure was designed as a screen for dis- hip replacement patients (9, Table 3). The IADL ability for use in primary care settings and to be scale correlated 0.76 with the SF-36 physical capable of being scored and summarized using function scale (6, p19). The FSQ physical scores the doctor’s office computer. In addition to its correlated 0.76 with the modified Health Assess- use by physicians, Jette notes that the FSQ could ment Questionnaire. Reuben et al. reported a be adapted for clinical use by physical therapists correlation of 0.55 between the ADL score and a (2). The aim of providing a routine and system- seven-item physical performance measure; the atic review of functional status is similar to that equivalent correlation for the IADL section was of the COOP Charts, whereas the main differ- 0.45. The ADL scale correlated 0.47 with the ence between the two instruments lies in their Katz ADL scale and 0.70 with the Older Ameri- length. The Charts provide only summary rat- cans Resources and Services (OARS) instrument ings in each area compared with the FSQ’s more IADL scale; correlations for the IADL scale were conventional, psychometric approach of sum- lower, at 0.31 with the Katz and 0.59 with the ming responses to several questions to form re- OARS (6, Table 2). In a study of Parkinson’s dis- sponse scales. The FSQ overlaps with other ease, convergent correlations with the SF-36 scales such as the SF-36 in its measurement of scales were high: 0.75 and 0.77 for the ADL and physical function but contributes unique infor- IADL scales, 0.64 for social function, 0.54 for mation in its coverage of interactions with fam- health satisfaction, and 0.78 for mental health ily and friends (5). (5, Table 3). The carefully developed FSQ shares items The predictive validity of the FSQ has been with other scales. The mental health section, for tested in several studies. FSQ ADL and IADL example, includes the same questions as the SF- scores proved better able to predict outcomes of 36. The FSQ has been used in diverse settings, valvuloplasty and recurrences in patients with including rehabilitation, in studies of hospital aortic stenosis than the New York Heart Associ- practice variations, and as an outcome measure ation classification (10). In a four-year prospec- in randomized trials (4). Jette notes that, partly tive study of risk factors for mortality, the IADL because of the brevity of the scales, reliability is and quality of interaction scores were indepen- inadequate to make comparisons between indi- dently associated with mortality after control- vidual patients, although the FSQ may be ade- ling for age, race, marital status, and other quately reliable to study the same patient across variables (8). Correlations between the IADL time and to compare groups of patients. Further scale and risk of mortality was −0.22, whereas validity testing, including sensitivity to change, that for the social activities scale was −0.25 (4, is desirable, as is information on the practical as- Table 5). pects of using the FSQ routinely in medical prac- Sensitivity to change was tested for 106 hip re- tice (perhaps using computer instead of  General Health Status and Quality of Life 591 paper-and-pencil administration). Nonetheless, (10) Tedesco C, Manning S, Lindsay R, et al. this instrument shows good results and should Functional assessment of elderly patients be seriously considered for ambulatory care re- after percutaneous aortic balloon search. valvuloplasty: New York Heart Association classification versus Functional Status Questionnaire. Heart Lung Address 1990;19:118–125. Scoring software is available from Michael Mc- (11) Baker G, Gagnon D, McNulty P. The Coy, MD, Department of Medicine, UCLA, Los relationship between seizure frequency, Angeles, California, USA 90024. seizure type and quality of life: findings from three European countries. Epilepsy Res 1998;30:231–240. References (1) Jette AM, Davies AR, Cleary PD, et al. The The Duke Health Profile Functional Status Questionnaire: reliability (George R. Parkerson, 1990) and validity when used in primary care. J Gen Intern Med 1986;1:143–149. Purpose (2) Jette AM, Cleary PD. Functional disability The Duke Health Profile (DUKE) is intended as a assessment. Phys Ther brief and practical measure to evaluate patient- 1987;67:1854–1859. reported functional health status in primary care (3) Rubenstein LV, Calkins DR, Young RT, et settings. al. Improving patient function: a randomized trial of functional disability screening. Ann Intern Med Conceptual Basis 1989;111:836–842. The DUKE was derived from the Duke-UNC (4) Cleary PD, Jette AM. Reliability and Health Profile (DUHP), a 63-item measure de- validity of the Functional Status signed to measure functional status outcomes in Questionnaire. Qual Life Res primary care settings (1). The DUHP had four 2000;9:747–753. subscales: symptom status, and physical, emo- (5) Rubenstein LM, Voelker MD, Chrischilles tional, and social function that reflected the EA, et al. The usefulness of the Functional World Health Organization (WHO) definition Status Questionnaire and Medical of health (2, p8). Limitations in the DUHP stim- Outcomes Study Short Form in Parkinson’s ulated the development of the abbreviated ver- disease research. Qual Life Res sion. The limitations included grouping all 1998;7:279–290. (6) Reuben DB, Valle LA, Hays RD, et al. symptoms together rather than distributing them Measuring physical function in among physical and mental scales, reliance on community-dwelling older persons: a self-esteem as the sole indicator of emotional comparison of self-administered, health, and measuring social function only in interviewer-administered, and perfor- terms of social role performance (3, p1057). mance-based measures. J Am Geriatr Soc The abbreviated DUKE instrument did not 1995;43:17–23. change the wording of items but extended the (7) Einarsson G, Grimby G. Disability and WHO triad of physical, mental, and social clas- handicap in late poliomyelitis. Scand J sifications by adding self-esteem and self- Rehabil Med 1990;22:1–9. perceived health; positive and negative aspects of (8) Reuben DB, Rubenstein LV, Hirsch SH, et health are covered separately. al. Value of functional status as a predictor of mortality: results of a prospective study. Am J Med 1992;93:663–669. Description (9) Katz JN, Larson MG, Phillips CB, et al. The DUKE is a 17-item generic health status pro- Comparative measurement sensitivity of file from which six scales measure function: short and longer health status instruments. physical, mental, and social health, general Med Care 1992;30:917–925. health, perceived health, and self-esteem. Five 592 Measuring Health scales measure dysfunction: anxiety, depression, that the score is then reversed and the item also pain, and disability, plus an anxiety-depression included on one of the dysfunction measures. If scale that combines items from the anxiety and de- the response to an item is missing, scores cannot pression subscales. The questionnaire is self- be calculated for any scale that involves that completed; the time frame refers to the present item (2; 3, p42). or to the past week (3, p1062). The items are shown in Exhibit 10.19. The instrument is self- Reliability reported, but computer scannable and interviewer- Parkerson’s original article describing the DUKE administrated versions are illustrated in the reported correlations between each item and the manual of the DUKE (2, Figures 1 to 3). remainder of the multiitem scales; these were rel- The raw scores for each response are indi- atively low, ranging from 0.37 to 0.45 for the cated in the exhibit. The items may be used sep- physical health items, 0.38 to 0.45 for the men- arately, combined into a total quality of life tal health items, and 0.26 to 0.35 for the social score, or grouped into various measures, as fol- health items (3, p1062). Alpha coefficients for lows. Scores for the component measures are the eight multi-item measures ranged from 0.55 calculated by summing the raw scores for the to 0.78; test-retest coefficients for the 11 mea- relevant items, dividing by the maximum raw sures ranged from 0.30 to 0.78 and exceeded 0.5 score and then multiplying by 100 (this arith- for all except pain and disability (3, Table 2). Al- metic has been abbreviated in the summary of pha reliability for the scale scores in a study of each scale below). The resulting scales run from 314 ambulatory patients ranged from 0.49 to zero to 100, with high scores for health mea- 0.70, whereas test-retest coefficients ranged sures indicating good health and high scores for from 0.41 to 0.72 (4, Table 1). A small test- the dysfunction measures indicating poor health retest study on patients with musculoskeletal (2, Figure 5). The items to be added on each disorders reported reliability ranging from 0.35 scale, and the scaling factors for the six function to 0.69 (5). An intraclass coefficient of 0.59 for measures are as follows: the general health dimension was obtained for 49 patients with musculoskeletal disorders; this Physical health: items 8 to 12 (× 10) figure was lower than that found using the SF- Mental health: items 1, 4, 5, 13, 14 (× 10) 36, Sickness Impact Profile (SIP), or Nottingham Social health: items 2, 6, 7, 15, 16 (× 10) Health Profile (6). In a sample of healthy people, General health: (add above three scores and the alpha coefficient was 0.38 for the physical divide by 3) health scale (compared with 0.58 for the equiva- Perceived health: item 3 (× 50) lent scale in the Medical Outcomes Study Self-esteem: items 1, 2, 4, 6, 7 (× 10). (MOS) SF-20 instrument); the alpha coefficient for the mental health score was 0.47, compared In scoring the five dysfunction measures, the raw with 0.82 for the SF-20 scale (7, p681). scores for each item listed below are first sub- In all, results have accumulated from ten tracted from 2 to reverse the direction of each studies of reliability, giving results broadly com- raw score. parable with those already noted; these are sum- marized in the manual of the DUKE (2, Table 1). Anxiety: items 2, 5, 7, 10, 12, 14 (× 8.333) Several studies reporting internal consistency Depression: items 4, 5, 10, 12, 13 (× 10) were large; the total sample size from the nine Pain: item 11 (× 50) studies reporting alpha coefficients is over Disability: item 17 (× 50) 10,000 patients (2, Table 1). Anxiety-Depression: items 4, 5, 7, 10, 12, 13, 14 (× 7.143). Validity The correlations between selected DUKE scales Note that several items are counted on one (or and the equivalent scales from the DUHP instru- more than one) of the function measures and ment they were derived from were 0.72 (physical Exhibit 10.19 The Duke Health Profile, Showing the Raw Scores for Each Response Note: The scores shown are used in calculating the seven health measures. In calculating the four negative measures, the scoring for items 2, 4, 5, 7, 10–14 and 17 is reversed (see text). The scores are omitted from the version completed by the respondent. Instructions: Here are a number of questions about your health and feelings. Please read each question carefully and check (√) your best answer. You should answer the questions in your own way. There are no right or wrong answers. No, doesn’t Yes, describes Somewhat describe me exactly describes me me at all 1. I like who I am . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 2. I am not an easy person to get along with . . . . . . . . . . 0 1 2 3. I am basically a healthy person . . . . . . . . . . . . . . . . . . 2 1 0 4. I give up too easily . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 5. I have difficulty concentrating . . . . . . . . . . . . . . . . . . . 0 1 2 6. I am happy with my family relationships . . . . . . . . . . . 2 1 0 7. I am comfortable being around people . . . . . . . . . . . . . 2 1 0 Today would you have any physical trouble or difficulty None Some A Lot 8. Walking up a flight of stairs . . . . . . . . . . . . . . . . . . . . 2 1 0 9. Running the length of a football field . . . . . . . . . . . . . 2 1 0 During the past week: How much trouble have you had with: None Some A Lot 10. Sleeping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 11. Hurting or aching in any part of your body . . . . . . . . . 2 1 0 12. Getting tired easily . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 13. Feeling depressed or sad . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 14. Nervousness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 During the past week: How often did you: None Some A Lot 15. Socialize with other people (talk or visit with friends or relatives) . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 1 2 16. Take part in social, religious, or recreation activities (meetings, church, movies, sports, parties) . . . . . . . . . 0 1 2 During the past week: How often did you: None 1–4 Days 5–7 Days 17. Stay in your home, a nursing home, or hospital because of sickness, injury, or other health problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 0 Adapted from Parkerson GR, Jr. User’s guide for Duke Health Measures. Department of Community and Family Medicine, Duke University Medical Center, Durham, North Carolina, 2002. With permission. 593 594 Measuring Health health), 0.70 (mental health), and 0.61 (social); an 18-month follow-up study (8, Table 1). The the correlation of the overall scores was 0.86 (3, DUKE appears to suffer less from ceiling and p1063). floor effects than other measures do (i.e. it dis- A table summarizing convergent correlations criminates well at the extreme ends of the health between DUKE scales and other health measures continuum). Compared with the Nottingham is included in the Manual; coefficients typically Health Profile and the COOP Charts, the DUKE fall in the range 0.50 to 0.70 (2, Table 2). Park- appeared to have less of a ceiling effect; its per- erson et al. compared the DUKE and the Short- formance was comparable with that of the SF-36 form-20 survey in a sample of healthy students (9, pp305–6). Other predictive validity findings (7). A multitrait-multimethod analysis lent sup- are reported in the Manual (2, Table 4). port to the disability, pain, mental and perceived Parkerson summarized several studies of ef- health dimensions in which the convergent cor- fect size in the Manual (2, Tables 5 to 7). Com- relations clearly exceeded the divergent (7, Table pared with the SF-36, effect sizes were 3). Support for the physical and social health di- comparable for physical function scales on the mensions was less clear; physical health corre- two instruments. The DUKE mental health and lated most strongly with the SF-20 pain score, perceived health scales appeared more respon- and only 0.18 with the SF-20 physical function sive, whereas the SF-36 social function and pain score. Likewise, social functioning correlated scales appeared superior (2, Table 5). Standard- only 0.07 with the equivalent SF-20 score (7, ized response means for patients undergoing car- Table 3). Parkerson et al. discussed the contrast- diac rehabilitation patients ranged from 0.12 to ing content of the DUKE and SF-20 scales (7, 0.76 for the scales, with an overall value of 0.78 p682). A comparison of the DUKE and SIP (2, Table 7). However, a lower effect size of 0.34 showed an overall correlation of −0.70. Correla- was reported in a study of patients with muscu- tions between individual scales were: −0.63 for loskeletal disorders, smaller than that obtained the physical health scales, −0.48 for mental for other leading health measures (6). health and −0.41 for social, and +0.36 for dis- ability. It was notable that comparatively little contrast was found between convergent and dis- Alternative Forms criminant coefficients for the physical health The DUKE has been translated into seventeen measure, which correlated relatively highly with languages; a listing is included in the manual (2, most of the SIP scales (3, Table 3). Scores on the Appendix F) and on the Qolid Web site. The self-esteem measure correlated 0.80 with scores French translation has been validated (10). In on the Tennessee Self-Concept Scale; correla- linguistic terms, the European and Canadian tions with the other DUKE health measures (ex- French translations agree for just seven of the 17 cept physical) fell between 0.60 and 0.64 (3, items whereas the UK and US English versions Table 4). The Zung depression scale showed a agree for 15 of the 17 items. Castillan and Latin correlation of −0.70 with the DUKE mental American Spanish versions are close but differ health score, −0.58 with self-esteem and +0.63 greatly from the U.S. Spanish version. Such reve- with both the anxiety and depression scores (3, lations may offer a useful insight into interna- p1065). tional relations, or not. In terms of group comparisons, DUKE scores The seven-item subscale covering anxiety- showed some significant differences between pri- depression is sometimes used alone, recognizing mary care patients consulting for physical, men- the strong conceptual overlap between these two tal, or health maintenance reasons (3, Table 5). clinical entities (2, Figure 4). Information on Highly significant differences were found in predictive validity is available (11), whereas the DUKE scores between patients judged as having area under the ROC curve in identifying anxiety high and low disability (4, Table 2). DUKE was 0.72 and that for depression was 0.78 (2, scores also predicted use of ambulatory care in Table 3).  General Health Status and Quality of Life 595 Reference Standards Address Norms (means, standard deviations, floor and General information, and copies of the Duke ceiling percentages and percentiles) by age and scales, is available from http://healthmeasures sex are shown in the user’s guide (2, Appendices .mc.duke.edu. Information on translations is B and C), and in (12, Appendix B). These are available from the Quality of Life Instruments based on 3,521 health insurance policy holders Database (QOLID) web site: www.qolid.org/ and 1,916 primary care patients. Some reference public/DUKE.html. Copies of the Duke instru- data are available for patients with diabetes, pa- ment, including the user’s guide and administra- tients on hemodialysis, and others (2, p20). tion instructions, are also available at a charge Parkerson provided tables to convert raw scores from the Medical Outcomes Trust, http://www on each scale to reference-adjusted scores (2, .outcomes-trust.org/instruments.htm. Appendices D and E). Dr. G.R. Parkerson, MD, Department of Family and Community Medicine, Box 3886, Duke University Medical Center, Durham, Commentary North Carolina, USA 27710. The DUKE scale derives from the established Duke-UNC Health Profile and was created to fill the need for a brief and practical measure for pri- References mary care settings. The DUKE has been shown (1) Parkerson GR, Jr., Gehlbach SH, Wagner acceptable to patients as quicker and easier to EH, et al. The Duke-UNC Health Profile: complete than the SF-20 (13). It is longer and an adult health status instrument for more broad-ranging than the COOP Charts, but primary care. Med Care 1981;19:806–828. briefer than the SF-36 or Nottingham Health (2) Parkerson GR, Jr. Users guide for Duke Profile. Correlations with much longer scales are health measures. Durham, N.C.: moderate to high. The main distinctive feature of Department of Community and Family the DUKE is its inclusion of the self-esteem cate- Medicine, Duke University Medical Center, gory. The early evidence for reliability and valid- 2002. ity suggests that the emotional components are (3) Parkerson GR, Jr., Broadhead WE, Tse C- sound but that the physical health scale does not KJ. The Duke Health Profile: a 17-item perform in the manner expected. measure of health and dysfunction. Med Care 1990;28:1056–1072. The DUKE offers a broad-ranging instrument (4) Parkerson GR, Jr., Broadhead WE, Tse C- with a number of subscores. However, five of KJ. Quality of life and functional health of these share the same items so they cannot be primary care patients. J Clin Epidemiol seen as independent measures (an issue that also 1992;45:1303–1313. arises with the McMaster Health Index Ques- (5) Beaton DE, Hogg-Johnson S, Bombardier tionnaire). Nor are all the scales homogeneous: C. Evaluating changes in health status: the physical scale, for example, includes two reliability and responsiveness of five generic items on activities of daily living, one on sleep, health status measures in workers with one on tiredness, and one on pain. Because the musculoskeletal disorders. J Clin Epidemiol component scales lack specificity, it is not sur- 1997;50:79–93. prising that some convergent validity correla- (6) Beaton DE, Bombardier C, Hogg-Johnson S. Choose your tool: a comparison of the tions are of similar strength to the divergent psychometric properties of five generic health ones, and the low alpha coefficients are to be ex- status instruments in workers with soft pected. tissue injuries. Qual Life Res 1994;3:50–56. For use as a general outcome measure in pri- (7) Parkerson GR, Jr., Broadhead WE, Tse C- mary care settings, the DUKE is worthy of con- KJ. Comparison of the Duke Health Profile sideration, although precise interpretation of and the MOS Short-Form in healthy young some of the subscores may be uncertain. adults. Med Care 1991;29:679–683. 596 Measuring Health (8) Parkerson GR, Jr., Broadhead WE, Tse gan in 1972 as a study of alternatives to institu- CK. Health status and severity of illness as tional care for frail elderly people with a view to predictors of outcomes in primary care. maintaining their independence in the least re- Med Care 1995;33:53–66. strictive care environment (1). The program in- (9) McHorney CA, Tarlov AR. Individual- vested a major effort in developing an patient monitoring in clinical practice: are information system that included the OMFAQ as available health status surveys adequate? a broad-ranging patient assessment, along with Qual Life Res 1995;4:293–307. (10) Briancon S, Alla F, Mejat E, et al. [Mea- evaluation procedures to help ensure that services surement of functional inability and quality are tailored to needs (2). The OARS assessment of life in cardiac failure. Transcultural model considered three elements: personal func- adaptation and validation of the Goldman, tional status, a method for classifying services Minnesota and Duke questionnaires]. into their generic elements and counting the use [French]. Arch Mal Coeur Vaiss of each category of service, and a transition ma- 1998;90:1577–1585. trix to assess the impact of service packages for (11) Parkerson GR, Jr., Broadhead WE, Tse C- people according to functional level (1, p54). The KJ. Anxiety and depressive symptom transition matrix describes the proportions of identification using the Duke Health people in each functional class who progress to Profile. J Clin Epidemiol 1996;49:85–93. higher or lower functional classes over time given (12) Parkerson GR, Jr. User’s guide for the Duke Health Profile (DUKE). Durham, the provision of various types of support service. North Carolina: Duke University Medical Parts A and B of the OMFAQ cover the first two Center, 1994. of these elements; the transition matrices are be- (13) Chen AL-T, Broadhead WE, Doe EA, et al. ing developed empirically (3). Patient acceptance of two health status measures: the Medical Outcomes Study Description Short-Form General Health Survey and the The OMFAQ is a structured questionnaire that Duke Health Profile. Fam Med is divided into part A, the Multidimensional 1993;25:536–539. Functional Assessment Questionnaire, and part B, the Services Assessment Questionnaire. Part A The Older Americans Resources and includes five sections, covering social and eco- Services (OARS) Multidimensional nomic resources, mental and physical health, and activities of daily living. Sixty-six questions Functional Assessment are asked of the respondent and a further ten Questionnaire questions record judgments made by an infor- (Duke University, 1975, mant. Many of the questions have subparts, Revised 1988) making a total of 120 items. Based on informa- Purpose tion collected in the interview, the interviewer The OARS Multidimensional Functional Assess- makes five summary ratings, one for each of the ment Questionnaire (OMFAQ) was designed to sections. Part B, the Services Assessment Ques- assess the overall personal functional status and tionnaire, covers 24 categories of services re- service use of adults, and in particular of the el- ceived and needed; full definitions of each derly (1, p1). It can be used as a screening in- service are provided (1, pp39–44). The two strument, as an outcomes evaluation, and as a questionnaires can be used separately, but when measure in modeling the cost-effectiveness of al- the entire instrument is used, some of the service ternative approaches to providing care. use items are interspersed among the functional assessment sections to improve the flow of the Conceptual Basis interview (1, p3). The OARS Program forms the clinical facet of The OMFAQ evolved from revisions to an In- the Duke University Center for the Study of Ag- take Form, developed for an outpatient clinic ing and Human Development. The program be- population; it was later modified into a Commu-  General Health Status and Quality of Life 597 nity Service Questionnaire for assessing people baum (1, Tables 7–11). The weights were based at home (4). Although these instruments were on regression analyses of the questions that con- designed for the elderly, items were later added tributed most to each section score in the raters’ to the OMFAQ to make it suitable for all people scoring (1, Table 12). aged 18 years and over. Questions were drawn However they are derived, the section scores from existing instruments and the source of each can be presented in several ways. The five section is given in the manual. For example, it includes scores may be presented as a profile (2), or they the Short Psychiatric Evaluation Schedule, and may be added to form a Cumulative Impairment the Short Portable Mental Status Questionnaire Score (CIS) ranging from five (excellent function (SPMSQ) reviewed in Chapter 8. Over time, the in all areas) to 30 (totally impaired) (1, original questions were clarified and their an- pp46–49). The CIS gives equal weighting to swer categories updated for the questions on each section score. Scores below ten suggest ex- economic matters (such as the amount of rent cellent functioning; those over 18 indicate signif- paid); these revisions were published in 1988 (1, icant impairment in several areas (2). A third Chapter 8). approach is to dichotomize each of the five The OMFAQ must be administered by a scores into not impaired or impaired, giving 25, trained interviewer. Two-day interviewer train- or 32 permutations, and the respondent can be ing courses are available from the Duke Univer- classified into one of these profiles (1, p47). The sity Center for Aging, as described on the Web cutting-point for each section is left to the user site (see Address section). Further details on ad- and is chosen according to the purpose of the ministering the scale are given in the manual (2, classification. Fillenbaum suggested grouping p133), and in Fillenbaum’s book (1, Chapter 7). ratings one and two versus three through six if The OMFAQ was designed as a single interview the purpose is to compare those with unimpaired to be used in its entirety—its originators advise function against all others; if the focus is on against extracting particular sections (5). Part A more severe levels of disability, ratings one takes about 30 minutes to complete and the through four may be compared against five and whole interview takes about 45 minutes (2). The six (1, p47). The OARS manual, by contrast, entire scale is too long to present here; the earlier mentions grouping ratings one to three as unim- version is shown in the OARS manuals (2; 5) paired, and four to six as impaired (2). Finally, and the revised version in Fillenbaum’s book (1, instead of the profile, a simple count of the num- pp125–172). Exhibit 10.20 shows the contents ber of sections on which a patient shows signifi- of the OMFAQ; a rationale for the topics in- cant impairment (typically, scores of four or cluded is given by Pfeiffer. To illustrate the ques- more) may be used. tions, Exhibit 10.21 shows the activities of daily living (ADL) and instrumental activities of daily Reliability living (IADL) sections. Some reliability testing has been carried out on The OMFAQ may be scored either as a rating part A of the OMFAQ and on the Community scale or using computer-assigned scores. The Service Questionnaire from which it was derived first approach allows raters a measure of subjec- (2; 5). Because section scores on the OMFAQ tivity in interpreting responses; the computer are typically assigned by raters who review the method seeks to avoid possible inter-rater dis- questionnaire responses, the assessment of inter- crepancies (3). In the first mode, a rater reads the rater agreement is pertinent. Fillenbaum and answers that were given to each item and sum- Smyer reported inter-rater agreement for the marizes the level of function for that section on a OMFAQ for 11 raters who evaluated 30 patients six-point scale ranging from outstanding func- (4). Intraclass correlations were 0.66 for physi- tioning to complete impairment. The computer cal health, 0.78 for economic resources, 0.80 for scoring system seeks to replicate clinical judg- mental health, 0.82 for social resources, and ment by weighting each question in deriving the 0.87 for self-care. Raters were in complete same set of scores; it was summarized by Fillen- agreement for 74% of the ratings (1, p19). Exhibit 10.20 Contents of the OARS Multidimensional Functional Assessment Questionnaire Note: In addition to basic demographic and interview specific information, the OMFAQ includes two sections: Part A, Assessment of Individual Functioning, and Part B, Assessment of Services Utilization. Part A: Assessment of individual functioning Part A is divided into seven major sections. These sections, in order, with a listing of the number of primary questions (some questions include several items) and a description of their content, are: Section No. of questions Content Basic demographics 11 Address; date; interviewer; informant; place of interview; duration; sex; race; age; education; telephone number. Social resources 9 Marital status; resident companions; extent and type of contact with others; availability of confidante; perception of loneliness; availability, duration, and source of help. Economic resources 15 Employment status; major occupation of self (and of spouse, if married); source and amount of income; number of dependents; home ownership or rental, and cost; source and adequacy of financial resources; health insurance; subjectively assessed adequacy of income. Mental health 6 Short Portable Mental Status Questionnaire (SPMSQ), a ten-item test of organicity; extent of worry, satisfaction, and interest in life; assessment of present mental status and change in the past five years; fifteen-item Short Psychiatric Evaluation Schedule. Physical health 16 Physician visits, days sick, in hospital and/or nursing home in past six months; medications in past month; current illnesses and their extent of interference; physical, visual, and hearing disabilities; alcoholism; participation in vigorous exercise; self-assessment of health. Activities of daily living 15 Extent of capacity to: telephone, travel, shop, cook, do housework, take medicine, handle money, feed self, dress, groom, walk, transfer, bathe, and control bladder and bowels. Also, presence of another to help with ADL tasks. Informant assessments 10 Information on the focal person’s level of functioning on each of the five dimensions is sought from a knowledgeable informant. Specifically: Social: Capacity to get along with others; availability, duration, and source of help in time of need. Economic: Extent to which income meets basic self-maintenance requirements. Mental: Ability to make sound judgments, cope; interest in life; comparison with peers; change in past five years. Physical: Assessment of health; extent of interference of health problems. Interviewer section (a) interview specific 4 Sources of information; reliability of responses. (b) interview 15 Social: Availability and duration of help when needed; assessments adequacy of social relationships. Economic: Assessed adequacy of income; presence of reserves; extent to which basic needs are met. Mental: Ability to make sound judgments, cope; interest in life; behavior during interview. Physical: Whether obese or malnourished. Rating scales: Five six-point scales, one for each dimension. 598  General Health Status and Quality of Life 599 Exhibit 10.20 Part B: Services assessment For each of the twenty-four nonoverlapping services named below, enquiry is made into (a) utilization in the past six months, (b) intensity of present utilization (e.g., frequency), (c) service provider (e.g., self, family and friends, agency), and (d) perceived current need for service. 1. Transportation 13. Physical therapy 2. Social/recreational 14. Continuous supervision 3. Employment 15. Checking 4. Sheltered employment 16. Relocation and placement 5. Educational services, employment related 17. Homemaker-household 6. Remedial training 18. Meal preparation 7. Mental health 19. Administrative, legal, and protective 8. Psychotropic drugs 20. Systematic multidimensional evaluation 9. Personal care 21. Financial assistance 10. Nursing care 22. Food, groceries 11. Medical services 23. Living quarters (housing) 12. Supportive services and prostheses 24. Coordination, information, and referral Reproduced from Multidimensional functional assessment: the OARS methodology. A manual. 2nd ed. Durham, North Carolina: Duke University, Center for the Study of Aging and Human Development. 1978:13–15. Copyright Duke University Center for the Study of Aging and Human Development. With permission. Permission required to use. Reuben et al. reported an alpha of 0.68 for the each section in the questionnaire (4). Spearman IADL scale (6, Table 2). correlations between the OMFAQ and these rat- Other reliability data refer to the Community ings were 0.68 for the economic section, 0.67 Service Questionnaire, giving inter-rater Kendall for mental health, 0.82 for physical health, and coefficients of concordance between 0.70 and 0.89 for self-care capacity (1, Table 3; 4, Table 0.93, with 11 of 25 coefficients being 0.85 or 3). above (2, p32). Ratings of the same question- Several validity results are available for the naires made 12 to 18 months apart gave correla- Intake Form and the Community Survey Ques- tions between 0.47 and 1.00, with only six of 35 tionnaire, the earlier versions of the OMFAQ. coefficients lying below 0.80 (2, p32). Scores on physical and mental health from the Five week test-retest correlations for 30 elder- questionnaire were compared with ratings made ly subjects were 0.82 for the physical ADL ques- by psychiatrists and physicians’ assistants for 82 tions, 0.71 for the IADL questions, and 0.79 for patients. The Spearman correlation for mental those on economic resources. The test-retest cor- health was 0.62 and for physical health it was relation for the objective questions on social re- 0.70 (p< 0.001) (2, p27). OMFAQ scores have sources was 0.71 and that for the subjective been compared with health care expenditures, questions was 0.53. Coefficients for life satisfac- showing an increase in expenditures by a factor tion and mental health were lower: 0.42 and of 13 as the OMFAQ scores rose from the lowest 0.32, respectively (2, p30). Alpha internal con- to the highest level of impairment (2). Scores sistency for the IADL scale was 0.68, contrasting from three contrasting populations (985 com- with 0.81 for the Functional Status Question- munity residents, 78 patients, and 76 institution- naire IADL section (6, Table 2). alized elderly) showed clearly differing profiles on the instrument (7, Figure 1). In a prospective Validity study, scores on physical capacity significantly Fillenbaum and Smyer presented criterion valid- predicted mortality (relative risk of 7.9 for those ity results for the OMFAQ on 33 family medi- in poor health, compared to those rated as in ex- cine patients, using separate criterion ratings for cellent health) (8, Table 2). However, it was also Exhibit 10.21 The OARS Multidimensional Functional Assessment Questionnaire: ADL and IADL Sections Activities of daily living Now I’d like to ask you about some of the activities of daily living, things that we all need to do as a part of our daily lives. I would like to know if you can do these activities without any help at all, or if you need some help to do them, or if you can’t do them at all. [Be sure to read all answer choices if applicable in questions 56 through 69 to respondent.] Instrumental ADL 56. Can you use the telephone . . . 2 without help, including looking up numbers and dialing, 1 with some help (can answer phone or dial operator in an emergency, but need a special phone or help in getting the number or dialing), 0 or are you completely unable to use the telephone? – Not answered 57. Can you get to places out of walking distance . . . 2 without help (can travel alone on buses, taxis, or drive your own car), 1 with some help (need someone to help you or go with you when traveling) or 0 are you unable to travel unless emergency arrangements are made for a specialized vehicle like an ambulance? – Not answered 58. Can you go shopping for groceries or clothes [assuming subject has transportation] . . . 2 without help (taking care of all shoping needs yourself, assuming you had transportation), 1 with some help (need someone to go with you on all shopping trips), 0 or are you completely unable to do any shopping? – Not answered 59. Can you prepare your own meals . . . 2 without help (plan and cook full meals yourself ), 1 with some help (can prepare some things but unable to cook full meals yourself ), 0 or are you completely unable to prepare any meals? – Not answered 60. Can you do your housework . . . 2 without help (can you scrub floors, etc.), 1 with some help (can do light housework but need help with heavy work), 0 or are you completely unable to do any housework? – Not answered 61. Can you take your own medicine . . . 2 without help (in the right doses at the right time), 1 with some help (able to take medicine if someone prepares it for you and/or reminds you to take it), 0 or are you completely unable to take your medicines? – Not answered 62. Can you handle your own money . . . 2 without help (write checks, pay bills, etc.), 1 with some help (manage day-to-day buying but need help with managing your checkbook and paying your bills), 0 or are you completely unable to handle money? – Not answered Basic ADL 63. Can you eat . . . 2 without help (able to feed yourself completely), 1 with some help (need help with cutting, etc.), 0 or are you completely unable to feed yourself? – Not answered 600  General Health Status and Quality of Life 601 Exhibit 10.21 64. Can you dress and undress yourself . . . 2 without help (able to pick out clothes, dress and undress yourself ), 1 with some help, 0 or are you completely unable to dress and undress yourself? – Not answered 65. Can you take care of your own appearance, for example combing your hair and (for men) shaving . . . 2 without help, 1 with some help, 0 or are you completely unable to maintain your appearance yourself? – Not answered 66. Can you walk . . . 2 without help (except for a cane), 1 with some help from a person or with the use of a walker, or crutches, etc., 0 or are you completely unable to walk? – Not answered 67. Can you get in and out of bed . . . 2 without any help or aids, 1 with some help (either from a person or with the aid of some device), 0 or are you totally dependent on someone else to lift you? – Not answered 68. Can you take a bath or shower . . . 2 without help, 1 with some help (need help getting in and out of the tub, or need special attachments on the tub), 0 or are you completely unable to bathe yourself? – Not answered 69. Do you ever have trouble getting to the bathroom on time? 2 No 0 Yes 1 Have a catheter or colostomy – Not answered [If “Yes” ask a.] a. How often do you wet or soil yourself (either day or night)? 1 Once or twice a week 0 Three times a week or more – Not answered Reproduced from Multidimensional functional assessment: the OARS methodology. A manual. 2nd ed. Durham, North Carolina: Duke University, Center for the Study of Aging and Human Development, 1978:68. Copyright Duke University Center for the Study of Aging and Human Development. With permission. Permission required to use. noteworthy that the relative risk for a simple one each in the physical health, ADL, and IADL self-evaluation of overall health was even sections (1, Table 6). The reliability of the 11 stronger. factor scales ranged from alpha 0.52 to 0.87 (1, Although each section on the OMFAQ re- p25). Factor analyses of the physical function ceives a single, overall score, it was recognized questions shown in Exhibit 10.21 broadly con- that the sections cover different themes. Several firmed the appropriateness of their classification studies have examined the factor structure of the into ADL and IADL sections (9, Table 2). Analy- scales. Fillenbaum ran factor analyses on each sis of the OMFAQ mental health items suggested section separately and identified three factors in that they represent four dimensions: life satisfac- the social resources section, one in the economic tion, psychosomatic symptoms, alienation, and section, four in the mental health section, and cognitive deficit; items on affect were lacking 602 Measuring Health (10). Liang et al. argued that for screening pur- ever, a separate study found that the FAI re- poses the first three of these could be combined quired an average of 40 minutes to administer into a single score, but that the Short Portable (15, p244). Four-week test-retest reliability by Mental Status Questionnaire (SPMSQ) cognitive intraclass correlation was 0.81 for the cognitive rating should be treated separately (10, pP136). score, 0.71 for the ADL section, 0.55 for mental The ADL section showed a correlation of health, and 0.51 for physical health. Reliability rho = 0.65 with the Medical Outcomes Study for the social and economic resources sections (MOS) SF-20 physical functioning scale; the was lower, at 0.47 and 0.16. Inter-rater agree- equivalent correlation for the IADL section was ment ranged from 0.56 to 0.83 (14, Table 4). 0.67 (11, Table 3). The IADL scale correlated The correlation between FAI and OMFAQ sec- 0.33 with the Katz ADL scale and 0.59 with the tion scores was 0.77 for cognitive and ADL sec- Functional Status Questionnaire (FSQ) IADL tions, 0.59 for physical health, and 0.50 for scale; however, it correlated 0.70 with the FSQ mental health (14, Table 5). A small validity basic ADL scale; it correlated 0.56 with a seven- study of the FAI showed close agreement be- item measure of functional performance (6, tween FAI scores and independent clinical rat- Table 3). Five of the IADL items formed a ings of the same dimensions (16, p853). The FAI Guttman scale and scores on this were found to is able to distinguish between contrasting patient predict mental and physical health status one groups (15). year later (correlations between 0.48 and 0.51, A Brazilian version of the 15-item Short Psy- according to age). IADL scores also predicted chiatric Evaluation Schedule included in the mortality rates; for example, those unable to per- OMFAQ has been described; it had a sensitivity form any activities showed a 5.4 times higher of 61% at a specificity of 89% (17, p689). death rate than the sample as a whole (9, p704). A Rasch analysis found that the continence and Reference Standards bathing items did not fit a unidimensional scale, Norms from three U.S. random samples of peo- but that the remaining 12 ADL/IADL items did ple aged 60 and older, from the community, (12, p17). from adult day care, and from an institution, are given by Fillenbaum (1, Chapter 9). The mean Alternative Forms CIS summary score for community residents Pfeiffer et al. developed an abbreviated version aged 65 or older in Durham, North Carolina, of the OMFAQ called the Functional Assessment was 12.2, whereas it was 20.7 for institutional Inventory (FAI) (13). Note that this is completely residents (1, p47). distinct from the instrument with the same name Reference standards for the seven IADL ques- developed by Crewe and Athelstan and reviewed tions shown in Exhibit 10.21 were drawn from in this chapter. The two instruments were devel- three large studies and are available by sex and oped at the same time and were published in dif- age (9, Table 1). ferent locations, preventing the duplication of names from being detected. The FAI of Pfeiffer Commentary et al. omits most of the questions on medical ser- The development work for the OARS was thor- vices from part B of the OMFAQ, reduces the oughly carried out over a long period by a large number of answer categories for some items, multidisciplinary team. The OMFAQ has been uses a modified coding scheme, and includes cited as influencing the design of several subse- questions on life satisfaction and self-esteem not quent scales, including the CARE developed by found in the OMFAQ. Five scores are produced: Gurland et al. The instrument is being used in ADL impairment, physical health, mental various settings, reflecting the emphasis on de- health, economic resources, and social resources. veloping a multipurpose instrument (3). Fillen- The mean administration time for the FAI was baum proposed five of the IADL items as a brief 30.6 minutes in the home setting, compared screening instrument that showed good correla- with 44.6 for the OMFAQ (14, Table 6). How- tions with physical and mental health; these  General Health Status and Quality of Life 603 items could identify people for whom more ex- mending the OMFAQ as a valuable instrument tensive assessment is warranted (9). In its design, for providing a comprehensive profile of per- the OMFAQ combines elements of the clinical sonal functioning and service use. rating scale approach with the structured ques- tionnaire. It uses set question wording but al- Address lows scope for interpretation by the rater; it also The OARS web site is at http://www.geri.duke offers flexibility in scoring responses and in se- .edu/service/oars.htm. lecting cutting-points suited to the purpose at the time: “the same basic set of data can be ana- lyzed in several different ways to meet the needs References of its many different users” (2, p68). This flexi- (1) Fillenbaum GG. Multidimensional bility is also seen in the way that information functional assessment of older adults: the from family members or other informants can be Duke Older Americans Resources and Ser- used to supplement the picture obtained during vices procedures. Hillsdale, NJ: Lawrence the interview. Documentation is excellent, as is Erlbaum Associates, 1988. the provision of training opportunities. The (2) Multidimensional functional assessment: manuals provide clear details of the develop- the OARS methodology. A manual. 2nd ed. ment, administration, and quality of the instru- Durham, North Carolina: Duke University ment. The Data Archive at the Center at Duke Center for the Study of Aging and Human University holds a number of data sets that in- Development, 1978. (3) George LK, Fillenbaum GG. OARS clude the OARS instruments; these are available methodology: a decade of experience in to the public. The archive also keeps a register of geriatric assessment. J Am Geriatr Soc OARS users with descriptions of their projects 1985;33:607–615. and a bibliography of publications. (4) Fillenbaum GG, Smyer MA. The Much of the reliability testing refers to earlier development, validity, and reliability of the versions of the instrument and relies on small OARS Multidimensional Functional samples. The results suggest, rather than prove, Assessment Questionnaire. J Gerontol that the current version can be applied and 1981;36:428–434. scored in a consistent manner. It would be ad- (5) Fillenbaum G. Reliability and validity of vantageous to see reliability and validity studies the OARS Multidimensional Functional based on larger samples, and it would also be Assessment Questionnaire. In: Pfeiffer E, ed. Multidimensional Functional useful to derive reference standards from some Assessment: the OARS methodology. A of the larger studies in which the OMFAQ has manual. Durham, North Carolina: Duke been used. The ADL and IADL sections show University Center for the Study of Aging good validity and reliability, suggesting they are and Human Development, 1975. superior to many of the purpose-built instru- (6) Reuben DB, Valle LA, Hays RD, Siu AL. ments reviewed in Chapter 3. These questions Measuring physical function in were subsequently incorporated into Lawton’s community-dwelling older persons: a Multilevel Assessment Instrument (reviewed sep- comparison of self-administered, arately). Although the OARS team counsels interviewer-administered, and perfor- against applying these alone, they are increas- mance-based measures. J Am Geriatr Soc ingly used as separate scales. One limitation of 1995;43:17–23. (7) Maddox GL. Assessment of functional the ADL/IADL scales may lie in the response status in a programme evaluation and scales, which ask about independent perfor- resource allocation model. In: Holland mance of tasks. With cognitively impaired re- WW, Ipsen J, Kostrzewski J, eds. Measure- spondents, in particular, a task may still be ment of levels of health. Copenhagen: performed independently but more slowly and World Health Organization (WHO less efficiently (18, p406). Despite these minor Regional Publications European Series No. concerns, we have little hesitation in recom- 7), 1979:353–366. 604 Measuring Health (8) Lefrancois R, Lapointe M. Les indices de The Comprehensive Assessment and santé globale comme prédicteurs de la Referral Evaluation mortalité chez les personnes âgées: une étude prospective de six ans. Can J Public (Barry Gurland, 1977, Health 1999;90:66–71. Revised 1983) (9) Fillenbaum GG. Screening the elderly: a Purpose brief instrumental activities of daily living The Comprehensive Assessment and Referral measure. J Am Geriatr Soc 1985;33:698–706. Evaluation (CARE) is a semistructured interview (10) Liang J, Levin JS, Krause NM. Dimensions that evaluates the health and social problems of of the OARS mental health measures. J people aged 65 years and older (1; 2). It covers Gerontol 1989;44:P127–P138. psychiatric, medical, nutritional, economic, and (11) Carver DJ, Chapman CA, Thomas VS, et social problems and was intended to assess the al. Validity and reliability of the Medical individual’s need for care and preventive services Outcomes Study Short Form-20 and to indicate the likely prognosis. questionnaire as a measure of quality of life in elderly people living at home. Age Conceptual Basis Ageing 1999;28:169–174. The CARE was deliberately designed to be broad (12) Doble SE, Fisher AG. The dimensionality in scope, for several reasons. In assessing people and validity of the Older Americans living in the community, the interpretation of Resources and Services (OARS) Activities of Daily Living (ADL) Scale. J Outcome symptoms is more complex than among hospi- Meas 1998;2:4–24. talized patients: (13) Pfeiffer E, Johnson TM, Chiofolo RC. Functional assessment of elderly subjects in The situation is very different with regard to four service settings. J Am Geriatr Soc persons who have been randomly selected 1981;29:433–437. from the community based population. It (14) Cairl RE, Pfeiffer E, Keller DM. An cannot be assumed that their symptoms (if evaluation of the reliability and validity of any) have clinical significance, nor, if they do the Functional Assessment Inventory. J Am have significance, to which disciplinary do- Geriatr Soc 1983;31:607–612. main they might pertain. For example, (15) Pfeiffer BA, McClelland T, Lawson J. Use weight loss which may indicate depression in of the Functional Assessment Inventory to a hospitalized psychiatric patient may, in a distinguish among the rural elderly in five service settings. J Am Geriatr Soc community resident, just as well be normal 1989;37:243–248. (e.g., the person is on a reducing diet), due to (16) Robinson BE, Lund CA, Keller D, et al. a medical condition (e.g., a wasting disease), Validation of the Functional Assessment or due to a social condition (e.g., poverty, or Inventory against a multidisciplinary home lack of help in preparing food). (1, p18) care team. J Am Geriatr Soc 1986;34:851–854. Gurland et al. also noted that most elderly peo- (17) Blay SL, Ramos LR, Mari JJ. Validity of ple suffer from more than one health or social a Brazilian version of the Older problem, and each must be distinguished to de- Americans Resources and Services termine appropriate treatment (2). Treatment (OARS) mental health screening chosen depends on the combination of problems questionnaire. J Am Geriatr Soc encountered and may require continual monitor- 1988;36:687–692. (18) Doble SE, Fisk JD, Rockwood K. Assessing ing of all of them. the ADL functioning of persons with The conceptual model on which the CARE is Alzheimer’s disease: comparison of family based identifies a causal sequence of health informants’ ratings and performance-based problems in the elderly. The sequence begins assessment findings. Int Psychogeriatr with age, race, and social circumstances; these 1999;11:399–409. influence medical condition and cognitive states  General Health Status and Quality of Life 605 and thereby influence functional capacity. Func- memorize in part to enhance the flow of the in- tional capacity in turn may cause the person to terview (1). To clarify the meaning of questions seek care and may also lead to demoralization to a respondent who does not understand, the and family inconvenience (3). guide contains standard alternative phrasings of these questions. Description Some years after the initial development of Gurland described the CARE as follows: the CORE-CARE, a scoring system was pro- posed that provided summary scores for 22 “in- The CARE is a new assessment technique dicator scales” covering psychiatric and medical which is intended to reliably elicit, record, problems, service needs, and social conditions. grade and classify information on the health These scales were formed by selecting items and social problems of the older person. The from the original interview schedule on the basis CARE is basically a semi-structured inter- of expert judgment of face validity and impor- view guide and an inventory of defined rat- tance and using empirical data on internal con- ings. It is designated comprehensive because sistency collected from 445 randomly selected it covers psychiatric, medical, nutritional, elderly residents in New York and 396 in Lon- economic and social problems rather than the don (4). Exhibit 10.22 summarizes the 22 indi- interests of only one professional discipline. cator scales. Exhibit 10.23 gives an example of The style, scope and scoring of the CARE one of the scales. Copies of the questionnaire are makes it suitable for use with both patients available from Dr. Gurland; scoring instructions and non-patients, and a potentially useful aid may be obtained from the National Technical in determining whether an elderly person Information Service (see Address section). should be referred, and to whom, for a health The SHORT-CARE is intended as a simpler or social service. The CARE can also be em- instrument focusing on psychiatric impairment ployed in evaluating the effectiveness of that and physical disability. It contains 143 items service if given. (1, p10) drawn from six of the CORE-CARE scales: de- pression/demoralization, dementia, memory The items included in the CARE were drawn problems, sleep, somatic symptoms, and activity from existing instruments, including Wing’s Pre- limitation. The interview is in two parts. The sent State Examination, Gurland’s Structured first, which lasts about 30 minutes, contains 143 and Scaled Interview to Assess Maladjustment items; the second part, containing additional (SSIAM), the Older Americans Resources and items on depression and dementia, identifies the Services (OARS) Multidimensional Functional need for clinical intervention (5). Assessment Questionnaire (OMAFQ), the Men- tal Status Questionnaire, and various scales Reliability measuring activities of daily living (1). Develop- Alpha internal consistency scores for the 22 mental tests of the CARE were carried out in CORE-CARE indicator scales were high, rang- London and in New York. ing from 0.72 for retirement dissatisfaction to The original version of the CARE contained 0.95 for activity limitation (1, Table 2; 3, Tables 1,500 items and was administered in an inter- 1–5). Intercorrelations among the indicator view lasting about 90 minutes (2). Shortened scales were reported by Golden et al. (4). Teresi versions have subsequently been derived (2): the et al. have reported reliability of the cognitive CORE-CARE (329 items; see Exhibit 10.22), screening component using methods involving and the SHORT-CARE (143 items). item response theory (IRT). Results ranged from Interviewers receive detailed training (2). The 0.7 to 0.9 for various groups of respondents (6). interview is not completely standardized, thus A learning effect may set in during repeated ad- permitting the interviewer to alter the order of ministrations of the cognitive component (7). questions to suit each respondent. A manual The agreement between two raters for 30 in- gives questions that the interviewer is trained to terviews on the CORE-CARE gave kappas rang- 606 Measuring Health Exhibit 10.22 Indicator Scales of the CORE- Exhibit 10.23 An Example of an Indicator CARE Scale from the CORE-CARE Number Cognitive impairment of items 1. Doesn’t know his age 2. Doesn’t know the year of his birth Psychiatric problems 3. Doesn’t know the number of years in 1. Cognitive impairment 10 neighborhood 2. Depression/demoralization 29 4. Doesn’t know his address 3. Subjective memory problems 9 5. Doesn’t know the rater’s name, first try 6. Can’t recall the President’s name, current or 48 previous Physical problems 7. Doesn’t know the month 4. Somatic symptoms 34 8. Doesn’t know the year 5. Heart disorder 15 9. Doesn’t know the rater’s name, second try 6. Stroke effects 9 10. Failed knee-hand-ear test 7. Cancer 6 Adapted from Golden RR, Teresi JA, Gurland BJ. Develop- 8. Respiratory symptoms 6 ment of indicator scales for the Comprehensive Assessment 9. Arthritis 9 and Referral Evaluation (CARE) interview schedule. J Geron- 10. Leg problems 9 tol 1984;39:141–144, Table 1. 11. Sleep disorder 8 12. Hearing disorder 14 13. Vision disorder 11 14. Hypertension 4 15. Ambulation problems 27 was lower for the medical and physical dimen- 16. Activity limitation 39 sions and ranged widely, reflecting the specialty of the rater (1). Correlations ranged from 0.01 191 to 0.83. Agreement for the social dimensions Service needs was intermediate, with correlations ranging 17. Service utilization 15 from 0.48 to 0.92 (1, Table 3). 15 Inter-rater reliability for the SHORT-CARE was 0.76 for dementia, 0.94 for depression, and Environmental social problems 0.91 for disability (5, p167). Equivalent figures 18. Financial hardship 8 19. Dissatisfaction with neighborhood 8 in a second study ranged from 0.66 to 0.96 (8, 20. Fear of crime 18 p1016). Coefficient alphas for the same scales, 21. Social isolation 34 calculated on a population sample of 283 elderly 22. Retirement dissatisfaction 7 people, were 0.64, 0.75, and 0.84. 75 329 Validity Validity of the SHORT-CARE was reported by Adapted from Golden RR, Teresi JA, Gurland BJ. Develop- ment of indicator scales for the Comprehensive Assessment Gurland et al. (5). The convergent validity was and Referral Evaluation (CARE) interview schedule. J Geron- 0.33 for the depression scale, 0.51 for cognitive tol 1984;39:141–145, Tables 1, 2. impairment, and 0.70 for disability (3, Table 5). For 26 respondents, classified as psychiatric cases or normal by psychiatrists, the sensitivity of the combined SHORT-CARE psychiatric ing from 0.70 to 0.80 (4, p145). For the original scales was 100%; the specificity was 71%. Pre- CARE instrument, the intraclass correlation was dictive validity results were impressive in that used to measure agreement among four raters in the “diagnosed pervasive dementias had one applying the scale to videotaped interviews with year outcomes consistent with that expected of eight older women. Agreement was close for the dementia (e.g., death, institutional admission, psychiatric dimensions with correlations be- deterioration) in all cases” (5, p167). Those di- tween 0.82 and 0.97 (1, Table 3). Agreement agnosed as demented at initial interview had a  General Health Status and Quality of Life 607 mortality of 27% compared with 6% for an some form of psychiatric impairment. Kappa co- age-matched sample drawn from the remaining efficients were 0.76 and 0.78 for the two CARE interviewees. Those who were diagnosed as de- scales, giving sensitivity results of 93% and pressed during the initial interview had a higher 67%, respectively (3, p154). The CARE was use of psychotropic medication than others. compared with two criterion scales that assessed Gurland et al. assessed how well the depression family inconvenience and the extent to which the and dementia scales could discriminate between family had made plans to institutionalize the el- the two conditions. Using multivariate analyses, derly relative. Sensitivity and specificity analyses only three of 138 cases were misclassified; using yielded overall correct classification rates for the a simpler approach of setting cutting-points on CARE activity limitation scale ranging from 60 each scale, 22 cases (16%) were misclassified (9, to 89% (10, Tables 1, 2). Equivalent figures for p123). the cognitive impairment scale ranged from 0.44 The extensive validity results for the complete to 0.85, according to the cutting-point used (10, CARE instrument are presented under three Table 3). headings: construct, criterion, and predictive va- lidity. predictive validity. The predictive ability of 21 of the indicator scales to identify people who construct validity. Teresi et al. reported the subsequently died was examined; discrimination construct validity of the CARE, using multitrait- was significant for seven scales (p < 0.01), with multimethod matrices and path analyses. They odds ratios as high as 3.1 (10, Table 5). Logistic provided extensive details of the correlations be- regression was used to test the ability of several tween the CARE scales and data provided by CARE scales to predict death, a diagnosis of de- family informants and from global diagnostic mentia or depression, activity restriction, and ratings (3). The results provided strong evidence service use. The results, presented in terms of for the validity of the measures of functional ca- odds ratios attaching to each indicator scale, pacity (correlations ranging from 0.51 to 0.70). suggest that the likelihood of the outcomes, The validity coefficients for the medical scales given a high score on the CARE, are as much as ranged from 0.47 to 0.59. Validity of the service five to ten times greater than for those with low use scale was somewhat lower, although ade- scores. quate, with correlations falling between 0.40 and 0.75. The correlations for service needs were Alternative Forms lower, ranging from 0.29 to 0.54 (3, p150). Cor- Other forms of the CARE include the IN-CARE relations between CARE scales and the judg- for people in institutions (11), the MERGE- ments made by other informants were highest CARE, and GLOBAL-CARE (2). for the scales that assess behavior or are more An Italian version of the depression and de- objective (e.g., activity limitation, medical con- mentia indicator scales showed high reliability ditions, service use), with correlations falling in and validity (12). Inter-rater reliability (kappa the range 0.47 to 0.70 (3, Table 5). Agreement coefficient) was 0.83 and 0.96, respectively; sen- over the more subjective ratings (depression, ser- sitivity for the dementia scale was 77% at a vice needs) ranged from 0.30 to 0.60. Teresi et specificity of 96%; sensitivity and specificity for al. used IRT methods to identify differential item the depression scale were 95% and 92% (12, function between ethnic groups for the cognitive pp509–510). A Spanish version of the CARE section of the CARE (see Exhibit 10.23). The was produced in California (13). analysis identified some items that performed differently in the different groups (6). Commentary The validity testing of the CARE is extensive and criterion validity. The CARE depression and available results are impressive; the CARE scales cognitive impairment scales were compared with continue to be developed and refined. As a clinicians’ ratings of 26 cases, 16 of whom had broad-ranging assessment, it is valuable for as- 608 Measuring Health sessing elderly people living in the community, The SHORT-CARE: an efficient although it has been used in other settings with instrument for the assessment of reliable results. The original long form is nor- depression, dementia and disability. J mally replaced by one of the abbreviated ver- Gerontol 1984;39:166–169. sions that do not require such detailed training (6) Teresi JA, Kleinman M, Ocepek-Welikson K, et al. Applications of Item Response of interviewers. The validity of the SHORT- Theory to the Examination of the CARE is such that it can be used by psychiatrists Psychometric Properties and Differential or nonpsychiatrists as an accurate screening tool Item Functioning of the Comprehensive for depression and dementia. Assessment and Referral Evaluation The purpose of the CARE instruments is sim- Dementia Diagnostic Scale Among Samples ilar to that of the Multilevel Assessment Instru- of Latino, African American, and Non- ment (MAI) and the OARS Multidimensional Latino White Elderly. Res Aging Functional Assessment Questionnaire (OM- 2000;22:738–773. FAQ) also reviewed in this chapter. Of the three, (7) Di Bari M, Pahor M, Barnard M, et al. the MAI is perhaps the least well tested. The Evaluation and Correction for a ‘Training other two scales show similar levels of reliability Effect’ in the Cognitive Assessment of and validity; the OMFAQ has the advantage of Older Adults. Neuroepidemiology 2002;21:87–92. an extensive user’s manual and documentation, (8) Gurland BJ, Teresi J, Smith WM, et al. while the CARE has the advantage of being Effects of treatment for isolated systolic available in several shortened versions. hypertension on cognitive status and depression in the elderly. J Am Geriatr Soc Address 1988;36:1015–1022. Scoring instructions may be obtained from the (9) Gurland B, Golden R, Challop J. Unidimensional and multidimensional National Technical Information Service, U.S. approaches to the differentiation of Department of Commerce, Springfield, Virginia, depression and dementia in the elderly. In: USA 22151. Corkin S, Crowdon JH, Davis KL, et al., eds. Alzheimer’s disease: a report of progress. New York: Raven Press, References 1982:119–125. (1) Gurland B, Kuriansky J, Sharpe L, et al. (10) Teresi JA, Golden RR, Gurland BJ. The Comprehensive Assessment and Concurrent and predictive validity of Referral Evaluation (CARE) - rationale, indicator scales developed for the development and reliability. Int J Aging Comprehensive Assessment and Referral Hum Devel 1977;8:9–42. Evaluation interview schedule. J Gerontol (2) Gurland BJ, Wilder DE. The CARE 1984;39:158–165. interview revisited: development of an (11) Gurland B, Cross P, Defiguerido J, et al. A efficient, systematic clinical assessment. J cross-national comparison of the Gerontol 1984;39:129–137. institutionalized elderly in the cities of (3) Teresi JA, Golden RR, Gurland BJ, et al. New York and London. Psychol Med Construct validity of indicator-scales 1979;9:781–788. developed from the Comprehensive (12) Spagnoli A, Foresti G, MacDonald A, et al. Assessment and Referral Evaluation Italian version of the organic brain interview schedule. J Gerontol syndrome and the depression scales from 1984;39:147–157. the CARE: evaluation of their performance (4) Golden RR, Teresi JA, Gurland BJ. in geriatric institutions. Psychol Med Development of indicator scales for the 1987;17:507–513. Comprehensive Assessment and Referral (13) Lopez-Aqueres W, Kemp B, Plopper M, Evaluation (CARE) interview schedule. J et al. Health needs of the Hispanic elderly. Gerontol 1984;39:138–146. J Am Geriatr Soc 1984;32:191– (5) Gurland B, Golden RR, Teresi JA, et al. 198.  General Health Status and Quality of Life 609 The Multilevel Assessment many under the direction of M. Powell Lawton, who died in 2001. Other PGC scales reviewed in Instrument this book include the Physical Self-Maintenance (M. Powell Lawton, 1982) Scale and the Philadelphia Geriatric Center Purpose Morale Scale. The Philadelphia Geriatric Center The Multilevel Assessment Instrument (MAI) has been renamed the Madlyn and Leonard was designed to measure the overall well-being Abramson Center for Jewish Life and is part of of elderly persons living in the community. It the Polisher Research Institute. covers health problems, activities of daily living The MAI comprises seven dimensions with (ADL) skills, psychological well-being, environ- 147 items; a further 81 items cover medical and ment, and social interaction. demographic data, but these are not considered in this review. Items were taken from a wide variety Conceptual Basis of established indices (1), and the MAI incorpo- Lawton argued that to assess the quality of life of rates the ADL and Instrumental ADL (IADL) an elderly person, ratings must be made on four questions from the OMFAQ (see Exhibit 10.21). dimensions: behavioral competence, psychological The psychological domain includes questions on well-being, perceived quality of life, and objective morale and psychiatric symptoms, and the envi- quality of the environment (1). A fuller discussion ronmental dimension covers housing quality and of Lawton’s approach to defining quality of life personal security (1). Space does not permit listing among elderly people is given in his 1983 article all the MAI questions here, but a detailed instruc- (2). Reflecting this framework, the MAI built on tion manual that includes the complete instrument existing measurement instruments, notably the is available on the web, through the Polisher Re- OARS Multidimensional Functional Assessment search Institute at www.abramsoncenter.org/PRI Questionnaire (OMFAQ). Lawton argued that the /scales.htm. The seven dimensions are shown in MAI incorporated, but went beyond, existing in- Exhibit 10.24. Most of the dimensions are divided struments by considering environmental factors, into subscales—there are 14 in all (e.g., perceived by separating social interaction from personal environment includes housing quality, neighbor- pursuits such as hobbies, and by separating cogni- hood quality, personal security). Each subscale tive ability from psychological well-being (1). contains between three and 24 items. The functional abilities section of the MAI The instrument is administered in a home in- was based on the theme of “behavioral compe- terview that takes an average of 50 minutes to tence,” which Lawton viewed in terms of a hier- complete. Some information must be obtained archy of increasingly complex activities ranging from the elderly person, but much can be ob- from the basic biological functions required for tained from the spouse or other informant. As life maintenance, through perception and cogni- with the OARS MFAQ instrument, rating scores tion, followed by skills for physical self- are produced for the seven dimensions shown in maintenance, up to exploratory behavior and Exhibit 10.24. The ratings are based on re- complex social interactions (3). Within each of sponses to the questions, but also consider the these dimensions, a further hierarchy of compe- respondent’s behavior and other impressions tence is identified; thus, for example, the social gained by the interviewer (4, ppD1–D3). Scores behavior dimension ranges from sensory contact can also be derived in a more conventional man- through intimacy, to nurturance, and eventually ner from the question responses, added to give to creative leadership (2, Figure 1). The term scores for each dimension. Unweighted scores multilevel assessment is used to imply assessment were found to correlate well with more compli- on each of the levels of this hierarchy. cated scoring methods (1). Description Reliability This is one of several measurement scales devel- Lawton reported several studies of the agree- oped at the Philadelphia Geriatric Center (PGC), ment between two independent raters. Intraclass 610 Measuring Health Exhibit 10.24 The Scope of the Multilevel respondents ranged from 0.55 to 0.99 with the Assessment Instrument, Showing Numbers of exception of ADL, which had a correlation of Items in Each Subscale 0.35 (3, Table 2). The manual of the MAI pres- ents retest and alpha coefficients for each sec- Number tion; for the full length MAI these lie in the of items range 0.7 to 0.9 (4, Tables 9–15). 1. Physical health: Self-rated health 4 Use of health services 3 Validity Health conditions 26 The MAI showed a weak ability to distinguish Mobility 3 respondents living independently in the commu- Use of aids, prostheses, etc. 13 nity from those in institutional care. Correla- 2. Cognition: tions for the seven dimension scores fell between Mental status 6 0.05 and 0.54 (1, Table 2). Cognitive symptoms 4 Correlation with a psychologist’s indepen- 3. Activities of daily living: dent rating of 590 individuals was 0.23 for the IADL Scale 10 Personal Self-Maintenance Activities 6 cognition dimension of the MAI, although, curi- 4. Time use: ously, agreement between the psychologist’s rat- Social activities, sports, hobbies 18 ing of other dimensions and the MAI scores was 5. Social relations and interactions: higher, ranging from 0.56 to 0.69. Agreement Family 12 with ratings made by a housing administrator on Friends 5 180 respondents for the seven dimensions 6. Personal adjustment: ranged from 0.12 (for social interaction) to 0.59 Morale 9 (for ADL). The interviewers who applied the Psychiatric symptoms 3 MAI made their own summary ratings of each 7. Perceived environment: Housing quality 9 dimension, and these were correlated against Neighborhood quality 12 scores on the MAI. Coefficients ranged from Personal security 4 0.36 to 0.87 (1, Table 2). Lawton and Brody re- ported correlations between the IADL section 147 and the Physical Self-Maintenance Scale and a Adapted from Lawton MP, Moss M, Fulcomer M, Kleban variety of other measurements. The correlations MH. A research and service oriented Multi-level Assessment Instrument. J Gerontol 1982;37:96, Table 2. ranged from 0.36 to 0.62 (3, Table 6). Alternative Forms The MAI and its abbreviated forms are called “full length,” “middle length,” and “short correlations between ratings on the seven dimen- length.” They are described in the manual (4, sions ranged from 0.88 (for the IADL scale) to a Tables 1–8). low of 0.58 (for the social interaction scale) (1, p95). Agreement between two interviewers lay within a one-point discrepancy for 95% of the Reference Standards ratings on the summary scales for a sample of Mean scores and standard deviations are pre- 484 (1). sented for each dimension on the MAI in the Alpha internal consistency results for 590 re- manual (4, Tables 17–19). spondents ranged from 0.69 for the cognitive and psychiatric symptom scales to 0.93 for Commentary ADL. Two coefficients, those for health behav- The MAI was based on a clearly enunciated con- ior and personal security, were lower at 0.39 and ceptual framework and on existing measure- 0.57. Three-week test-retest reliabilities for 22 ments. The reliability findings are promising; yet  General Health Status and Quality of Life 611 the scale could clearly benefit from further test- The Self-Evaluation of Life Function ing and refinement. The validity results show Scale quite low correlations between the method and (Margaret W. Linn and independent assessments. Those scales showing Bernard S. Linn, 1984) low validity coefficients also showed low relia- bility scores, and Lawton noted that the psycho- Purpose metric properties of the social interaction and The Self-Evaluation of Life Function (SELF) Scale time use scales require improvement. The devel- was designed as a comprehensive measurement of opers of the MAI also experienced some prob- the physical, psychological, and social function- lems in selecting the most appropriate questions ing of people aged 60 years or older. The objec- for the environmental scales. The low validity tive was to develop a short multidimensional scale agreement between the psychologist’s rating of that elderly people could complete themselves. cognition and the corresponding MAI score is a cause for concern; it may be that the cognition Description measure discriminates only at very low levels of A pool of 130 items was derived from existing cognitive functioning. Lawton discussed the scales. On the basis of factor analysis, the items strengths and weaknesses of the MAI and ar- were reduced to 54 loading on six factors: activ- gued in favor of a simple scoring system and ities of daily living and physical disability (13 against abbreviating the scale. He noted that items), symptoms of aging (13 items), self- scale norms based on representative population esteem (seven items), social satisfaction (six samples are not yet available, and the social and items), depression (11 items), and personal con- environmental domains show particular need for trol (four items). Other questions cover diag- further research and development. He concluded noses, medications, sick days, and pain. that the physical health, cognition, and ADL di- Questions refer to the past month or current mensions of the MAI are the most robust. health; four-point response scales are used for This scale shows potential but lacks adequate all questions except diagnoses and medications. documentation and validity analysis. The MAI Factor scores are then used to weight each ques- needs to be refined but fills a potential niche for tion in forming section scores, and these are a scale to assess the well-being of the elderly liv- used in analysis. It takes respondents about 15 ing in the community. minutes to complete the scale. The scale, too long to reproduce here, is shown in Linn and Linn’s 1984 article (1). References The SELF was tested on 548 people 60 years (1) Lawton MP, Moss M, Fulcomer M, et al. of age or older and who were of at least moder- A research and service oriented Multilevel ate cognitive functioning (i.e., with scores above Assessment Instrument. J Gerontol 20 on the Mini-Mental State Examination). 1982;37:91–99. (2) Lawton MP. Environment and other Reliability determinants of well-being in older people. Test-retest intraclass correlations were calcu- Gerontologist 1983;23:349–357. lated for 101 elderly people after three to five (3) Lawton MP, Brody EM. Assessment of days. Results ranged from 0.99 to 0.36 for older people: self-maintaining and items, whereas section scores ranged from 0.96 instrumental activities of daily living. for physical disability to 0.59 for self-esteem (1, Gerontologist 1969;9:179–186. (4) Lawton MP, Moss M, Fulcomer M, et al. p609, Table 2). Multi-Level Assessment Instrument manual for full-length MAI. North Wales, Validity Pennsylvania: Polisher Research Institute, The six factors show modest intercorrelations: 2004. one third being 0.40 or higher (1, Table 2). The 612 Measuring Health scale was tested on four groups (i.e., institution- (2) Hawkins WE, Duncan DF, McDermott RJ. alized patients, mental health outpatients, those A health assessment of older Americans: receiving counseling, and community residents) some multidimensional measures. Prev and all factors discriminated significantly among Med 1988;17:344–356. groups, in the expected direction (1, Table 3). (3) Rohrer MJ, Cutler BS, Wheeler HB. Long- term survival and quality of life following Physical disability was the best discriminator. ruptured abdominal aortic aneurysm. Arch The SELF showed agreement with rating of im- Surg 1988;123:1213–1217. provement made by health care providers (p < 0.007). Predictive validity after 1 year was as- sessed for a variety of outcomes (1, Table 5). The The McMaster Health Index Questionnaire factor scores predicted the number of weeks in (Larry W. Chambers, 1976, institutions (R2 = 0.37) and visits to the physi- revised 1984) cian (R2 = 0.28). The most frequent predictors of outcome were physical disability and symptoms Purpose of aging (1, p611). Hawkins et al. obtained a The McMaster Health Index Questionnaire canonical correlation of 0.75 between the sub- (MHIQ) provides a profile of scores describing scales and a set of variables covering health prac- physical, emotional, and social function. The tices and demographic data (e.g., exercise, MHIQ is intended for use in health services eval- sleeping, education, age) (2). Unlike studies that uation and in clinical research, principally with use a single question, Hawkins et al. found an outpatients and those living in the community. association between self-reported health prac- tices and health status variables for the elderly. Conceptual Basis The World Health Organization definition of Alternative Forms health was used to guide the content, which cov- A 16-item version was adapted for telephone ad- ers physical, social, and mental well-being (1). ministration and used with patients after surgery for aneurysm (3). Description Items were drawn from a range of existing Commentary scales, including those of Bennett and Garrad, The SELF is a self-report multidimensional scale and the Katz Index of activities of daily living that appears to be acceptable to respondents. A (reviewed in Chapter 3). An early version of the scan of the literature suggests that it is used rea- MHIQ contained 150 items (2; 3); this was ab- sonably frequently, especially in studies of hear- breviated to the present 59-item version in the ing loss and in palliative care. Although the SELF early 1980s (4). The 59 items were selected on was based on established scales and shows prom- the basis of agreement with family physician rat- ise, it needs fuller formal testing. The scoring sys- ings of physical, mental, and social health, and tem is complicated and it would be interesting to of their sensitivity to change (1). Equal numbers see how well a simpler scoring method functions. of items cover physical, emotional, and social Linn and Linn concluded that the SELF scale is function. The physical items cover physical ac- useful for research and screening when a short, tivities, mobility, self-care, and communication. comprehensive, inexpensive, self-report measure The social function items cover general well- of the overall function of the elderly is needed. being, role performance, family participation, and relations with friends. The items on emo- tional function cover feelings of self-esteem, feel- References ings toward personal relationships, and thoughts (1) Linn MW, Linn BS. Self-Evaluation of Life about the future, and life events (5). The time Function (SELF) Scale: a short, comprehen- reference for most items is the present and most sive self-report of health for elderly adults. record performance rather than capacity. The J Gerontol 1984;39:603–612. questionnaire can be used in self-completed  General Health Status and Quality of Life 613 mode (20 minutes), or through personal or tele- tional, and 0.51 for the social scale for 40 pa- phone interviews. A comparison of these three tients with rheumatoid arthritis (5, p781). modes showed that physical function scores did not differ, nor did the size of change scores (6, Validity p474). However, for the emotional and social Chambers compared scores from contrasting scores self-completion gave slightly lower scores groups of patients. As hypothesized, physiother- than telephone administration (6, Table 1). apy patients scored significantly lower (i.e., Each item is scored by awarding one point to poorer functioning) than psychiatric patients on good function, and scores are added to provide the physical scale; the reverse held for the emo- the three scale scores. The responses that are tional and social scales (1, Figure 6.2). Family scored are shown in Exhibit 10.25, which also practice patients showed less physical disability indicates to which scale each item belongs. Note than did physiotherapy patients or those with that for items eight and nine, a point is awarded respiratory disease, but more emotional distress only if both of the “never” responses in either (1, Figures 6.3, 6.5). column are checked. Note, also, that the answer Physical function scores were compared to categories for question 35 are incorrect in some ratings made by occupational therapists for 40 published versions of the MHIQ (1; 4). The patients; the R2 was 0.26 (r= 0.51) (1, Figure items in question 42 are scored for both social 6.1). For 40 patients with rheumatoid arthritis, and emotional scales. Questions not answered MHIQ physical function scores were compared are scored to indicate poor function. The raw with clinical assessments and they showed an as- scores are standardized to a zero (extremely sociation with morning stiffness, with the poor function)-to-one (extremely good function) Ritchie Articular Index, and with age (no statis- scale; this involves dividing the raw score by 19 tical measure of association was provided) (5, for the physical scale and by 25 for the other Table 1). The social and emotional function two scales. A weighted scoring system has been scores showed no association with these vari- described for patients with chronic respiratory ables. Physical scores also correlated with a rat- disease (5); this correlated 0.98 with the un- ing of disease severity for 246 patients with weighted score (1, Figure 6.7). multiple sclerosis (rho = −0.70) (8, p307). The emotional function scores correlated 0.31 and Reliability 0.51 with Bradburn’s Affect Balance Scale on Intraclass one-week test-retest reliability coeffi- two occasions (6, p476). cients were 0.53 for the physical function scores, A randomized trial with physiotherapy pa- 0.70 for the emotional, and 0.48 for the social tients examined the sensitivity of the MHIQ to scores for 30 physiotherapy patients (1, p134). detecting change with treatment. The main ef- For 40 psychiatry patients, test-retest reliability fect of change over time was significant for the was 0.95 for the physical scores, 0.77 for the physical function scores, which also identified emotional, and 0.66 for social function scores. different amounts of improvement in different Fortin and Kérouac report inter-rater agreement patients (6, p474). The power of the MHIQ (its for the physical section of the original version of ability to detect differences of half a standard de- the MHIQ, to which they made “extensive viation) was greater than 0.9 for all three modes changes” (7, p129). Kendall’s coefficient of con- of administration (6, p474). Compared with the cordance between four raters was 0.71 (7, Table McMaster-Toronto Arthritis and Rheumatism 6). [Note that a figure of 0.80, representing (MACTAR) questionnaire, the MHIQ showed agreement between only two of these raters, is less sensitivity to change in patients with quoted in some reviews of the MHIQ (1, p134; rheumatoid arthritis (9). The MHIQ showed a 4, p162).] significant change in patients following hip re- Internal consistency was assessed using the placement surgery but a less marked change Kuder-Richardson formula, giving values of than that identified by the Arthritis Impact Mea- 0.76 for the physical scale, 0.67 for the emo- surement Scales (10, Table IV). Sensitivity to Exhibit 10.25 The McMaster Health Index Questionnaire Note: We have indicated the responses that are scored by letters following one of the response options for each question. These letters are omitted from the version given to the respondent. The responses marked “P” count one point to the physical score. In questions 8 and 9, both responses marked “p” must be checked to count one point. Responses marked “E” count one point to the emotional score; for questions 10–27, either of two responses in each question earns one point. Responses marked “S” count one point to the social function score. For question 34 there are several alternative answers; a maximum of one point can be awarded for this question. SECTION A: The questions in the first section ask about your health and whether you are able to do certain things. 1. Today, are you physically able to run a short distance, say 300 feet, if you are in a hurry? (This is about the length of a football field or soccer pitch.) 1 NO 2 YES P 2. Today, do you (or would you) have any physical difficulty at all with Difficulty No Difficulty a. walking as far as a mile? 1 2 P b. climbing up 2 flights of stairs? 1 2 P c. standing up from and/or sitting down in a chair? 1 2 P d. feeding yourself ? 1 2 P e. undressing? 1 2 P f. washing (face and hands), shaving (men), and/or combing hair? 1 2 P g. shopping? 1 2 P h. cooking? 1 2 P i. dusting and/or light housework? 1 2 P j. cleaning floors? 1 2 P 3. Today, are you physically able to take part in any sports (hockey, swimming, bowling, golf, and so forth) or exercise regularly? 1 NO 2 YES P 4. At present, are you physically able to walk out-of-doors by yourself when the weather is good? 1 NO 2 YES a. What is the farthest you can walk by yourself? b. Are you able to walk by yourself? 1 ONE MILE OR MORE P 4 BETWEEN ROOMS 2 LESS THAN 1 MILE BUT MORE THAN 5 ONLY WITHIN A ROOM 30 FEET (ABOUT THE SIDE OF A HOUSE) 6 CAN’T WALK AT ALL 3 LESS THAN 30 FEET 5. Today, do you (or would you) have any physical difficulty at all travelling by bus whenever necessary? (Circle your answer) 1 NO P 2 YES 6. Today, do you have any physical difficulty at all travelling by car whenever necessary? 1 NO P 2 YES 7. Do you have any physical difficulty at all driving a car by yourself? 1 NO (or do not have a licence) P Go to Q. 8 2 YES a. Is this because of a physical disability? 1 NO 2 YES 614 Exhibit 10.25 8. Do you wear glasses? 1 NO 2 YES a. Do you have any trouble seeing ordinary c. Do you have any trouble seeing ordinary newsprint when you wear your glasses? newsprint? 1 NEVER p 1 NEVER p 2 SOMETIMES 2 SOMETIMES 3 ALWAYS 3 ALWAYS b. Do you have a headache after watching d. Do you have a headache after watching television or television or reading when you wear your reading? glasses? } 1 NEVER p } 1 NEVER p 2 SOMETIMES 2 SOMETIMES Go to Q. 9 Go to Q. 9 3 ALWAYS 3 ALWAYS 9. Do you wear a hearing aid? 1 NO 2 YES a. Do you have trouble hearing in a normal c. Do you have trouble hearing in a normal conversation conversation with several other persons with several other persons? when you wear your hearing aid? 1 NEVER p 1 NEVER p 2 SOMETIMES 2 SOMETIMES 3 ALWAYS 3 ALWAYS b. Do you have trouble hearing the radio or tele- d. Do you have trouble hearing the radio vision when you wear your hearing aid? or television? 1 NEVER p 1 NEVER p 2 SOMETIMES 2 SOMETIMES 3 ALWAYS 3 ALWAYS SECTION B: Often people’s health affects the way they feel about life. For these next questions, please circle the choice that is closest to the way you feel about each statement. If you STRONGLY AGREE, circle 1 If you AGREE, circle 2 If you are NEUTRAL, circle 3 If you DISAGREE, circle 4 If you STRONGLY DISAGREE, circle 5 STRONGLY STRONGLY AGREE DISAGREE 10. I sometimes feel that my life is not very useful. 1 2 3 4 5 E 11. Everyone should have someone in his life 1 2 3 4 5 E whose happiness means as much to him as his own. 12. I am a useful person to have around. 1 2 3 4 5 E 13. I am inclined to feel that I’m a failure. 1 2 3 4 5 E 14. Many people are unhappy because they do not 1 2 3 4 5 E know what they want out of life. 15. In a society where almost everyone is out for 1 2 3 4 5 E himself, people soon come to distrust each other. 16. I am a quick thinker. 1 2 3 4 5 E (continued) 615 Exhibit 10.25 (continued) 17. Some people feel that they run their lives 1 2 3 4 5 E pretty much the way they want to and this is the case with me. 18. There are many people who don’t know what 1 2 3 4 5 E to do with their lives. 19. Most people don’t realize how much their 1 2 3 4 5 E lives are controlled by plots hatched in secret by others. 20. People feel affectionate toward me. 1 2 3 4 5 E 21. I would say I nearly always finish things once 1 2 3 4 5 E I start them. 22. When I make plans ahead, I usually get to 1 2 3 4 5 E carry out things the way I expected. 23. I think most married people lead trapped 1 2 3 4 5 E (frustrated or miserable) lives. 24. It’s hardly fair to bring children into the world 1 2 3 4 5 E the way things look for the future. 25. Some people feel as if other people push them 1 2 3 4 5 E around a good bit, and I feel this way too. 26. I am usually alert. 1 2 3 4 5 E 27. Nowadays a person has to live pretty much 1 2 3 4 5 E for today and let tomorrow take care of itself. SECTION C: This section contains some questions on general health and on your social activities. 28. How would you say your health is today? Would you say your health is (Circle your answer) 1 VERY GOOD 2 PRETTY GOOD S } 3 NOT TOO GOOD 29. Taking all things together, how would you say things are today? Would you say you are 1 VERY HAPPY 2 PRETTY HAPPY }S 3 NOT TOO HAPPY 30. In general, how satisfying do you find the way you’re spending your life today? Would you call it 1 COMPLETELY SATISFYING 2 PRETTY SATISFYING }S 3 NOT VERY SATISFYING 31. How would you say your physical functioning is today? (By this we mean the ability to move around, see, hear, and talk.) 1 GOOD 2 GOOD TO FAIR P} 3 FAIR 4 FAIR TO POOR 5 POOR 32. How would you say your social function is today? (By this we mean your ability to work, to have friends, and to get along with your family.) 1 GOOD 2 GOOD TO FAIR } S 3 FAIR 4 FAIR TO POOR 5 POOR 616 Exhibit 10.25 33. How would you say your emotional functioning is today? (By this we mean your ability to remain in good spirits most of the time and to be usually happy and satisfied with your life.) (Circle your answer) 1 GOOD 2 GOOD TO FAIR } E 3 FAIR 4 FAIR TO POOR 5 POOR 34. Are you presently working on a job for wages, either full- or part-time? 1 YES GO TO Q.35 S 2 NO . . a. Are you presently 1 ON VACATION S 2 ON SICK LEAVE 3 RETIRED 4 A STUDENT S 5 A HOUSEWIFE S 6 OTHER (please specify): 35. How much time, in a one-week period, do you usually spend watching television? } 1 NONE 2 LESS THAN THREE HOURS A WEEK S 3 LESS THAN TWO HOURS A DAY 4 TWO OR MORE HOURS A DAY 36. Which of the following describe your usual social and recreational activities? a. going to church? 1 NO 2 YES S b. going to a relative’s home? 1 NO 2 YES S c. any other activities? (please specify) S (yes) 37. Has anyone visited you in the last week? (Circle your answer) a. a relative? 1 NO 2 YES S b. a friend? 1 NO 2 YES S c. a religious group member? 1 NO 2 YES S d. a social agency representative? 1 NO S 2 YES 38. Do you have a telephone? 1 NO GO TO Q.41 2 YES S 39. Have you used your telephone in the last week to call a. a friend? 1 NO 2 YES S (continued) 617 618 Measuring Health Exhibit 10.25 (continued) b. a religious group member? 1 NO 2 YES S c. a social agency representative? 1 NO S 2 YES 40. Have you been called in the last week by a social agency representative? 1 NO S 2 YES 41. How long has it been since you last had a holiday? (Write in number “0” if presently on holidays.) _____ MONTHS or _____YEARS S (≤ 12 months) 42. During the last year, have any of the following things happened to you? a. separation from your spouse? 1 NO S,E 2 YES b. divorce? 1 NO S,E 2 YES c. going on welfare during the last year? 1 NO S,E 2 YES d. trouble getting along with friends/relatives during the last year? 1 NO S,E 2 YES e. retired from work during the last year? 1 NO S,E 2 YES f. some other problem or change in your life? (please specify) ______________________________ ______________________________ ______________________________ S, E (if no problems listed) Adapted from an original provided by Dr. L Chambers. With permission. change after myocardial infarction was lower however, variations in the components of the than that achieved by the SIP but slightly better questionnaire may not be reflected in the final than that for the Nottingham Health Profile (11, scores. The MHIQ has mainly been used in stud- Tables 2–4). ies at McMaster University in Ontario, Canada. The questionnaire has undergone a pro- Alternative Forms tracted development phase; Chambers noted that A Canadian French translation is available (7, he was reporting “major findings with the in- p129). strument some 22 years after it was initially con- ceived” (1, p132). This has compromised the Commentary consistency of the validity and reliability testing Two of the distinguishing features of the MHIQ in that several of the trials of the MHIQ used the are its wide range of coverage (including disabil- initial version, so results cannot be directly ap- ity and handicap and topics such as life events plied to the revised form. Some results are in un- that are not included in other scales) and the even published documents; other studies used only balance between physical, emotional and social parts of the questionnaire (8), or modified it (7), coverage. Because it only provides three scores, and some reports provide descriptive, rather  General Health Status and Quality of Life 619 than quantitative, evidence for validity (1, Fig- (7) Fortin F, Kérouac S. Validation of ures 6.2–6.5; 5). Fuller information on validity is questionnaires on physical function. Nurs needed, including comparisons with other, more Res 1977;26:128–135. recent scales as well as additional validation of (8) Harper AC, Harper DA, Chambers LW, et the emotional and social scales. al. An epidemiological description of physical, social and psychological problems The inclusion of the same items on both so- in multiple sclerosis. J Chronic Dis cial and emotional scales is unusual, although 1986;39:305–310. this appears in even more extreme form in the (9) Tugwell P, Bombardier C, Buchanan WW, DUKE instrument. Being counted twice, they et al. The MACTAR Patient Preference carry more weight than other items, although Disability Questionnaire: an individualized this approach may represent an honest recogni- functional priority approach for assessing tion of the difficulty of operationally separating improvement in physical disability in different dimensions of health. Even though this clinical trials in rheumatoid arthritis. J instrument has some attractive design features, Rheumatol 1987;14:446–451. the lack of reliability and validity data means (10) O’Boyle CA, McGee H, Hickey A, et al. that users will be likely to find an alternative Individual quality of life in patients undergoing hip replacement. Lancet scale that has undergone more extensive testing. 1992;339:1088–1091. (11) Taylor R, Kirby B, Burdon D, et al. The assessment of recovery in patients after References myocardial infarction using three generic (1) Chambers LW. The McMaster Health quality-of-life measures. J Cardiopulm Index Questionnaire: an update. In: Walker Rehabil 1998;18:139–144. SR, Rosser RM, eds. Quality of life assessment: key issues in the 1990s. The World Health Organization Dordrecht, Netherlands: Kluwer Academic Quality of Life Scale Publishers, 1993:131–149. (WHOQOL Group, 1994) (2) Chambers LW, Sackett DL, Goldsmith CH. Development and application of an index Purpose of social function. Health Serv Res The WHOQOL provides a subjective assessment 1976;11:430–441. of a broad definition of quality of life (QoL). (3) Sackett DL, Chambers LW, MacPherson Several applications were envisioned. In clinical AS, et al. The development and application applications, the WHOQOL would indicate the of indices of health: general methods and a impact of a health problem on a patient’s life, summary of results. Am J Public Health 1977;67:423–428. enhancing communication between doctor and (4) Chambers LW. The McMaster Health patient; it could then monitor progress after Index Questionnaire. In: Wenger NK, treatment and also be used as an outcome mea- Mattson ME, Furberg CD, et al., eds. sure in clinical trials. It could be used in health Assessment of quality of life in clinical surveys or in epidemiological studies to identify trials of cardiovascular therapies. New needs and guide policy formation, and subse- York: Le Jacq, 1984:160–164. quently to monitor the impact of health policies (5) Chambers LW, MacDonald LA, Tugwell P, on quality of life (1, p155; 2, pp53–54). et al. The McMaster Health Index Questionnaire as a measure of quality of Conceptual Basis life for patients with rheumatoid disease. J The motives for developing the WHOQOL in- Rheumatol 1982;9:780–784. (6) Chambers LW, Haight M, Norman G, et cluded concern over a perceived deterioration in al. Sensitivity to change and the effect of doctor-patient relationships, in which doctors mode of administration on health status lacked awareness of their patients’ feelings, and measurement. Med Care the recognition that clinical trial outcomes 1987;25:470–479. needed to incorporate quality of life (3). Atten- 620 Measuring Health tion to QoL was intended to counterbalance the established by international agreement among prevailing mechanistic model of medicine with a 15 participating centres in nine regions of the humanistic element; it was seen as “the missing world, rather than being developed by the WHO measurement in health” (2, p42). itself. In this “bottom-up” process, the WHO’s In the original protocol for developing the main role was to coordinate the development WHOQOL, the WHO’s Division of Mental work (1; 2; 5). Health set the goal of assessing subjective qual- Subsequently, the analysis of pilot data indi- ity of life, rather than symptoms or health condi- cated that a core set of themes was common to tions (1). At a time when most “QoL” all the countries involved, so the WHOQOL instruments were recording functional ability, team proposed a core set of 100 questions com- the WHOQOL chose to pay equal attention to mon to all centers. These could be supplemented function and to the person’s assessment of the by additional items as dictated by local circum- importance or the meaning of that level of func- stance or study topic (a conception comparable tion to their QoL (3, p2). Thus, for example, to that used for the European Organization for rather than record pain, questions would ask Research and Treatment of Cancer (EORTC) about the impact of pain on QoL, which was de- Quality of Life Questionnaire). The core module fined broadly, going beyond the person’s subjec- was seen as a “door” through which more de- tive evaluation of their health status and tailed questioning could occur for specific sub- functioning to include factors such as the safety groups. of the environment in which they live, their ac- cess to health services, and their spiritual well- Description being (2, p43; 4, p1). QoL was defined in terms The WHO prepared a detailed protocol that of “an individual’s perception of their position specified the general form and content of the in life in the context of the culture and value sys- WHOQOL; it would be a health profile measure tems in which they live and in relation to their covering broad domains of QoL, each divided goals, expectations, standards and concerns” (1, into facets. Domains and facets would be chosen p153; 5, p11). “It is a broad-ranging concept, as being applicable across cultures (1, incorporating in a complex way the person’s pp153–154; 3; 6). As a result of an iterative dis- physical health, psychological state, level of in- cussion process, six broad content domains were dependence, social relationships, personal beliefs defined, and definitions for 24 facets were re- and relationship to salient features of the envi- fined through focus group meetings in 15 field ronment . . . it includes both positive and nega- centers in nine regions of the world (2; 6, Table tive dimensions” (2, p43). This conception is 1; 7; 8; 9). The six original domains included intrinsically influenced by the respondent’s cul- physical and psychological health, level of inde- ture and value system; culture was viewed as an pendence, social relationships, environment, and integral element, rather than a variable to be ad- spirituality, religion, and personal beliefs (6, justed for to produce a culture free assessment Table 2). (3, p3). Each field center formed an item writing panel Taking a world-wide perspective, the WHO- that prepared objective and subjective questions QOL group’s original intention was not to pro- representing each facet of QoL. Questions were duce a single instrument, but to propose a to cover three complementary aspects of each methodology for producing equivalent QoL facet: perceived objective questions would de- measures for use in different cultures. The desire scribe the person’s behavior or capacity (e.g., was to encourage the development of measure- “How well do you sleep?”), self-reported subjec- ment tools that reflected local conceptions of tive questions that would cover satisfaction or QoL, and to avoid mere translation of an instru- dissatisfaction (“Are you satisfied with your ment developed in the United States or United sleep?”), and questions that would cover the per- Kingdom for use in other countries. The content ceived importance of the topic (“How important of the instrument and the study methods were is sleep to your overall quality of life?”) (1, Table  General Health Status and Quality of Life 621 1; 2, p52). Questions were phrased to be applica- The WHOQOL-100 is scored as a profile ble to healthy people as well as those with severe with scores for the 24 facets, six domain scores, impairments; positive terms were used wherever and a general QoL and health perceptions score possible to escape from the problem-centered fo- based on the four questions pertaining to global cus of many instruments (3). QoL and general health. All scores are oriented The questions written in each field center such that higher values denote better quality of were pooled, producing about 1800 undupli- life. Facet scores sum the relevant question cated items, from which 276 covering 30 facets scores, after reversing the scoring for negative were selected for a first test version (2, pp46–47; items. The raw scores (range, 4–20) can option- 9). Field trials involved over 4,500 respondents ally be transformed to a zero to 100 scale to rep- in the 15 centers and led to production of the fi- resent the percentage of the achievable score. nal, 100-item version (9, Table 4). Item selection Domain scores sum the facet scores; because the procedures are described in detail (3; 9), as are domains contain different numbers of facets, the some of the issues related to translation (10). raw domain score is divided by the number of The 100 core items (which are cross-culturally facets included to facilitate comparison between applicable) can be supplemented by additional domains. items of relevance to local settings (e.g., skin color in India, or security in Israel) (3, p3). A Reliability statistical procedure for examining whether lo- Alpha coefficients for the WHOQOL-100 phys- cal items contribute beyond the 100 core items is ical health domain in three large samples ranged described by Skevington et al. (11). from 0.86 to 0.88; values for the psychological For the objective questions, five-point re- domain ranged from 0.79 to 0.82; those for the sponse scales cover intensity, capacity, or fre- social domain ran from 0.72 to 0.73, whereas quency (see Exhibit 10.26). Note that the the environmental coefficients were all 0.85 (13, Exhibit shows the “international English” word- Table 3). Other samples produced alpha values ing; minor variations are used in the Australian, ranging from 0.82 to 0.95 (4, Table 3). A British British and U.S. versions. These local versions study obtained figures ranging from 0.87 to should be obtained from the national representa- 0.95 for the domain scores, with an overall al- tives, from whom permission for use is also re- pha of 0.97 (14, p457). A study of patients with quired. The time frame refers to the past two depression gave alpha values for the domains weeks (3, p4). For the subjective questions a ranging from 0.82 to 0.91, with an overall alpha satisfied–dissatisfied scale was used. Intermediate of 0.96 (15, Table 4). scale points were chosen for each language after In a more detailed analysis of internal consis- a psychometric scaling process designed to en- tency, Bonomi et al. used multitrait analysis to sure equivalence across cultures (2, p48; 12). investigate the appropriateness of the original The WHOQOL-100 takes ten to 20 minutes to placement of items on facets and facets on do- complete, with more time needed for elderly or mains. They found that questions on pain and sick respondents (4, p3). It is intended for self- on energy correlated more highly with the inde- administration but can be administered by an in- pendence dimension than with the physical; this terviewer where necessary. appears to result from question wording (e.g., Early in the field studies, it was found that the pain is evaluated in terms of how much it pre- six domains shown in the Exhibit could be re- vents the person from undertaking normal activ- duced to four by merging domains one and ities) (4, p6). three, and two and six (7, p4; 9, Table 8; 13, Test-retest correlations after a two- to eight- Table 1). Nonetheless, some publications group week delay were 0.66 for physical health, 0.72 the 100 items into six domains, whereas others for psychological, 0.76 for social, and 0.87 for now use four. The issue continues to be investi- environmental scores (13, p555). Intraclass cor- gated, and the structure shown in the Exhibit relations measuring two-week retest reliability was current as of May, 2004. were 0.83 for the physical domain, 0.84 for psy- Exhibit 10.26 Domains, Facets and Items Included in the WHOQOL-100 and WHOQOL-BREF Notes: (1) The questions are shown under their domains and facets to illustrate the content of the WHOQOL. However, the sequence of questions is changed in field versions (obtainable from the national study centers). (2) The question wording shown here is the generic English-language version and is intended for purposes of illustration only. (3) “(BREF)” after an item identifies the 26 items included in the WHOQOL-BREF Eight different answer scale are used. The appropriate scale for each question is indicated by a code letter after each question. A. 1 Not at all; 2 A little; 3 A moderate amount; 4 Very much; 5 An extreme amount B. 1 Not at all; 2 Slightly; 3 Moderately; 4 Very; 5 Extremely C. 1 Not at all; 2 Slightly; 3 Moderately; 4 Very well; 5 Extremely D. 1 Not at all; 2 A little; 3 Moderately; 4 Mostly; 5 Completely E. 1 Very dissatisfied; 2 Dissatisfied; 3 Neither satisfied nor dissatisfied; 4 Satisfied; 5 very satisfied F. 1 Very unhappy; 2 Unhappy; 3 Neither happy nor unhappy; 4 Happy; 5 Very happy G. 1 Very poor; 2 Poor; 3 Neither poor nor good; 4 Good; 5 Very good H. 1 Never; 2 Seldom; 3 Quite often; 4 Very often; 5 Always Overall Quality of Life and General Health How would you rate your overall quality of life? (response scale G) (BREF) How satisfied are you with your quality of life? (E) In general, how satisfied are you with your life? (E) How satisfied are you with your health? (E) (BREF) Domain 1: Physical Health 1. Pain and discomfort How often do you suffer physical pain? (H) Do you worry about your pain or discomfort? (A) How difficult is it for you to handle any pain or discomfort? (B) To what extent do you feel that physical pain prevents you from doing what you need to do? (A) (BREF) 2. Energy and fatigue Do you have enough energy for everyday life? (D) (BREF) How easily do you get tired? (B) How satisfied are you with the energy that you have? (E) How bothered are you by fatigue? (B) 3. Sleep and rest How well do you sleep? (G) Do you have any difficulties with sleeping? (A) How satisfied are you with your sleep? (E) (BREF) How much do any sleep problems worry you? (A) Domain 2: Psychological 4. Positive feelings How much do you enjoy life? (A) (BREF) Do you generally feel content? (H) How positive do you feel about the future? (B) How much do you experience positive feelings in your life? (A) 622 Exhibit 10.26 5. Thinking, learning, memory and concentration How would you rate your memory? (G) How satisfied are you with your ability to learn new information? (E) How well are you able to concentrate? (C) (BREF) How satisfied are you with your ability to make decisions? (E) 6. Self-esteem How much do you value yourself? (A) How much confidence do you have in yourself? (A) How satisfied are you with yourself? (E) (BREF) How satisfied are you with your abilities? (E) 7. Bodily image and appearance Are you able to accept your bodily appearance? (D) (BREF) Do you feel inhibited by your looks? (C) Is there any part of your appearance that makes you feel uncomfortable? (A) How satisfied are you with the way your body looks? 8. Negative feelings How often do you have negative feelings, such as blue mood, despair, anxiety, depression? (H) (BREF) How worried do you feel? (B) How much do any feelings of sadness or depression interfere with your everyday functioning? (A) How much do any feelings of depression bother you? (A) Domain 3: Level of Independence 9. Mobility How well are you able to get around? (G) (BREF) How satisfied are you with your ability to move around? (E) How much do any difficulties in mobility bother you? (A) To what extent do any difficulties in movement affect your way of life? (A) 10. Activities of daily living To what extent are you able to carry out your daily activities? (D) To what extent do you have difficulty in performing your routine activities? (A) How satisfied are you with your ability to perform your daily living activities? (E) (BREF) How much are you bothered by any limitations in performing your everyday living activities? (A) 11. Dependence on medication or treatment How dependent are you on medications? (D) How much do you need any medication to function in your daily life? (A) How much do you need any medical treatment to function in your daily life? (A) (BREF) To what extent does your quality of life depend on the use of medical substances or medical aids? (A) 12. Working capacity Are you able to work? (D) Do you feel able to carry out your duties? (D) How would you rate your ability to work? (G) How satisfied are you with your capacity for work? (E) (BREF) (continued) 623 Exhibit 10.26 (continued) Domain 4: Social Relations 13. Personal relationships How alone do you feel in your life? (B) Do you feel happy about your relationship with your family members? (F) How satisfied are you with your personal relationships? (E) (BREF) How satisfied are you with your ability to provide for or support others? (E) 14. Social support Do you get the kind of support from others that you need? (D) To what extent can you count on your friends when you need them? (D) How satisfied are you with the support you get from your family? (E) How satisfied are you with the support you get from your friends? (E) (BREF) 15. Sexual activity How would you rate your sex life? (G) How well are your sexual needs fulfilled? (B) How satisfied are you with your sex life? (E) (BREF) Are you bothered by any difficulties in your sex life? (B) Domain 5: Environment 16. Physical safety and security How safe do you feel in your daily life? (B) (BREF) Do you feel you are living in a safe and secure environment? (B) How much do you worry about your safety and security? (B) How satisfied are you with your physical safety and security? (E) 17. Home environment How comfortable is the place where you live? (B) To what extent does the quality of your home meet your needs? (D) How satisfied are you with the conditions of your living place? (E) (BREF) How much do you like it where you live? (A) 18. Financial resources Have you enough money to meet your needs? (D) (BREF) Do you have financial difficulties? (A) How satisfied are you with your financial situation? (E) How much do you worry about money? (A) 19. Health and social care: availability and quality How easily are you able to get good medical care? (B) How would you rate the quality of social services available to you? (G) How satisfied are you with your access to health services? (E) (BREF) How satisfied are you with the social care services? (E) 20. Opportunities for acquiring new information and skills How available to you is the information you need in your day-to-day life? (D) (BREF) To what extent do you have opportunities for acquiring the information that you feel you need? (D) How satisfied are you with your opportunities for acquiring skills? (E) How satisfied are you with your opportunities to learn new information? (E) 624  General Health Status and Quality of Life 625 Exhibit 10.26 21. Participation in and new opportunities for recreation/leisure To what extent do you have the opportunity for leisure activities? (D) (BREF) How much are you able to relax and enjoy yourself? (D) How much do you enjoy your free time? (A) How satisfied are you with the way you spend your spare time? (E) 22. Physical environment (pollution/noise/traffic/climate) How healthy is your physical environment? (B) (BREF) How concerned are you with the noise in the area you live in? (A) How satisfied are you with your physical environment (e.g., pollution, climate, noise, attractiveness)? (E) How satisfied are you with the climate of the place where you live? (E) 23. Transport To what extent do you have adequate means of transport? (D) To what extent do you have problems with transport? (A) How satisfied are you with your transport? (E) (BREF) How much do difficulties with transport restrict your life? (A) Domain 6: Spirituality/Religion/Personal beliefs 24. Spiritual Do your personal beliefs give meaning to your life? (A) To what extent do you feel your life to be meaningful? (A) (BREF) To what extent do your personal beliefs give you the strength to face difficulties? (A) To what extent do your personal beliefs help you to understand difficulties in life? (A) From an original provided by Dr. S. Chatterji, WHO, Geneva. With permission. chological, 0.96 for independence, 0.88 for so- mains each made an independent contribution to cial, 0.92 for environment, and 0.86 for spiritual the overall QoL score (13, Table 5). This finding domains (4, p7). In a study of patients who had was later confirmed using structural equation undergone liver transplantation, a control group modeling, which also confirmed the fit of the who received no intervention showed no signifi- various facets onto their respective domains (17, cant change in scores over a three-month period Figures 2 and 3). With minor variations, these (16, Table 2). findings held for all 15 study centers included in the test development process. Validity Concurrent validity has been assessed for the Initial factor analyses based on the pilot data U.S. version. Correlations with equivalent sub- from the initial trial of the WHOQOL-100 iden- scales on the SF-36 were generally high, com- tified four factors, in which the independence monly ranging from 0.6 to 0.7 (4, Table 4). and physical domains were merged, and the spir- However, the social domain scores correlated itual domain was merged with that of the psy- more highly (0.55) with the SF-36 mental health chological (9, Table 8). Confirmatory factor scale than with the SF-36 social functioning analysis subsequently showed that a structure scale (0.30), and the WHOQOL physical do- based on four domains was somewhat superior main tended to correlate with all of the SF-36 to the six—domain structure; the four domains scales, suggesting that it may lack specificity (4, loaded on a higher order factor representing Table 4). In the same study, correlations between global QoL (9, Figure 1; 13, p555). The four do- the WHOQOL and the Subjective Quality of 626 Measuring Health Life Profile (SQLP) were higher: 0.70 for the Alternative Forms two physical scales, 0.66 for the two social An abbreviated version, the WHOQOL-BREF, scales, whereas the WHOQOL environmental contains 26 of the 100 items, one for each facet domain correlated 0.73 with the SQLP material and two general items (13); the items are shown life dimension. The WHOQOL spiritual domain in Exhibit 10.26. The WHOQOL-BREF is avail- scores correlated 0.42 with the SQLP spirituality able from www.who.int/mental_health/media/en scale, and 0.46 with the social life scale (4, Table /76.pdf. The general items are not included in 5). In a British study, correlations with the SF-36 the scoring (20). Scoring instructions are given included 0.58 between SF-36 bodily pain and on the Web site. For healthy people, the BREF WHOQOL independence, 0.70 for SF-36 physi- takes less than five minutes to complete (20, cal functioning and WHOQOL independence, p303). It has been tested in several large samples. and 0.64 for SF-36 mental health scores and the Correlations between domain scores on the WHOQOL psychological domain (18, Table 7). BREF and on the full WHOQOL ranged from The WHOQOL pain questions correlated −0.68 0.89 to 0.95 in one study (13, p554) and from with the McGill Pain Questionnaire present pain 0.82 to 0.92 in another (16, p122). About 95% intensity scale; regression analyses showed that of the total facet score variance in the 100-item it was principally the evaluative (or emotional) measure is captured by the BREF (3, p6). component of pain that contributed to predict- Several studies have reported internal consis- ing QoL. The WHOQOL pain scale correlated tency, including an analysis of 11,830 question- 0.60 with the overall WHOQOL score, and the naires from 23 countries (20). Results are shown duration of a person’s pain was also strongly in Table 10.4. and inversely associated with QoL scores (19, One study reported that alpha values for the p401). BREF fell between 0.04 and 0.06 lower than the In a study of responses before and following corresponding figures for the WHOQOL-100 childbirth, Bonomi et al. reported effect size sta- (13, Table 3). Four-week retest reliability values tistics of 0.86 for physical domain and 0.91 for ranged from 0.64 (social) to 0.79 (psychologi- the independence domain; none of the other do- cal) (21, p756). mains showed a change in scores (4, p8). The The question of whether the WHOQOL- physical, psychological, and environmental do- BREF should be presented as a single score has main scores can discriminate significantly be- been addressed. A Rasch analysis indicated that tween healthy and chronically sick groups, the complete WHOQOL-BREF should not be whereas the independence, social and spiritual considered a single scale, although the four do- scores may not (4, Table 6). Similarly, a study of main scores did fit a two-parameter item re- patients before and after a liver transplantation sponse theory model (22). A four-factor solution operation showed standardized response means was derived from a study in Taiwan, reflecting of 1.0 for physical, 0.96 for psychological, 0.84 psychological, social, environmental, and physi- for environmental, and 0.70 for social (16, Table cal dimensions; not all of the items, however, 3). Each domain of the WHOQOL proved sensi- loaded on the expected factors (21, Table 3). Us- tive to changes following antidepressant phar- ing data from a different sample, Skevington et macotherapy; correlations of before and after al. identified items that appear not to be specific changes between the WHOQOL and the Beck to their intended domains (20, p305). However, Depression Inventory (BDI) included 0.496 confirmatory factor analyses using data from 23 (physical domain), 0.55 (psychological), and countries showed that the BREF items did, in- 0.40 (independence and social) (15, Table 1). A deed, load on their intended four factors, and second, similar, study obtained correlations with that these all contribute to an overall QoL do- the BDI of 0.49 for the physical domain, 0.69 main (20, p306). for the psychological, and 0.54 for social rela- Discriminant validity analyses show that the tionships and 0.53 for general quality of life (18, BREF physical domain scores best discriminate Table 4). between healthy and sick respondents, and sex  General Health Status and Quality of Life 627 Table 10.4 Alpha Internal Consistency Coefficients for the WHOQOL-BREF Physical Psychological Social Environment (7 items) (6 items) (3 items) (8 items) N Source 0.82 0.81 0.68 0.80 11,830 (20, Table 2) 0.82 0.75 0.66 0.80 4,802 (13, Table 3) 0.84 0.77 0.69 0.80 3,882 (13, Table 3) 0.80 0.76 0.66 0.80 2,369 (13, Table 3) 0.74 0.81 0.76 0.85 136 (21, Table 2) and age explained less than 3% of the overall Reference Standards variance (20, p303). All four domain scores dis- Mean facet and domain scores, with standard criminated significantly between HIV-infected deviations, by sex, age-group, and for healthy patients and healthy respondents (21, Table 4). versus sick respondents, were drawn from the Standardized response means were calculated in initial testing of the WHOQOL-100 (N = 4787) a study of patients who had undergone liver (9, Table 7). For the WHOQOL-BREF, mean transplantation before and after the operation. scores and standard deviations for the domains The results were slightly lower than the figures are available for 23 countries, and scores by sex obtained for the complete WHOQOL-100, es- and age group for the countries combined (20, pecially for the social domain: values were 0.92 Tables 6 and 7). Danish population norms and for physical, 0.91 for psychological, 0.74 for percentile scores are available, by age and sex environmental, and 0.43 for social (16, Table (22, Table 3). 3). A separate, 32-item WHOQOL instrument Commentary covers spirituality, religion and personal beliefs More than with any other measure reviewed in (SRPB) and has been tested in 18 centers. Infor- this book, the development of the WHOQOL mation is available from www.who.int/mental has been transparent and collaborative. As Skev- _health/media/en/620.pdf. The WHO regional ington et al. observed, “The WHOQOL is an in- office for Europe (www.euro.who.int/ageing) is strument that has been designed by the uers for developing a version suitable for adults more the users (3, p3). Before any questions were than 60 years old, the WHOQOL-OLD. A 31- written, an international group of experts con- item WHOQOL-HIV instrument was developed vened by the WHO mapped out the objectives using the WHOQOL group’s international col- and full details of the procedures to be observed. laborative approach (23; 24, pp345–346). It is For the most part, these guides were followed available from www.who.int/mental_health and input was received from people in many /media/en/556.pdf. A child health questionnaire countries. A similar international collaborative has been developed following the format of the approach was used in developing the Organisa- WHOQOL (25), and a pain module is also tion for Economic Co-operation and Develop- available (26). ment (OECD) scale and the European Organiza- The WHOQOL is available in over 40 lan- tion for Research and Treatment of Cancer guages, with more adaptations under way. Many Quality of Life Questionnaire C30 (EORTC of the studies reviewed in this section provide QLQ-C30), but the WHO collaboration in- comparative results from different countries, but volved a broader range of countries, and used a reports are appearing on the validity and relia- more “bottom up” approach. bility of different linguistic versions (mainly of The strengths of the WHOQOL include its the BREF): Arabic (27), Korean (28), and Tai- universality. A standard method was used in wanese (21). each field setting to develop and test the ques- tions, and the core instrument is globally appli- 628 Measuring Health cable yet retains the option of adding culture- evant to policy, and in routine clinical practice. specific items (2, p51). Although it was origi- Because these varied purposes imply contrasting nally anticipated that separate versions of the design features, few (if any) instruments could instrument would be required for different coun- successfully serve so many purposes. To date, ev- tries, an important empirical finding was that a idence for the validity of the WHOQOL focuses core set of items proved acceptable in 15 widely mainly on its factorial structure and, to a lesser different centers (9; 17). Although each trial cen- extent, its correlations with other scales; more ter included additional items considered to be of targeted studies are required to indicate whether local relevance, few of these items performed the instrument can function in the other in- better than the core items in subsequent item tended applications. In a related vein, the criti- analyses (9, p1582). Hence, the WHOQOL has cism has been made that the content of the permitted empirical investigation of the struc- WHOQOL was not justified theoretically, and ture of QoL in different cultures. that topics relevant to some of the intended ap- A second virtue of the WHOQOL is its plications may have been omitted (10, p141). breadth of coverage; the inclusion of spirituality For example, it takes a deliberately subjective and of environmental factors ensured that the approach and few items directly assess func- concept of QoL extends beyond the focus on tional limitations or disability. Used as an out- disability that is common in other instruments. come measure, therefore, it will primarily The inclusion of importance ratings is reminis- indicate whether a patient feels better rather than cent of the approach used in scales such as Sara- whether she has recovered; this is compounded son’s Social Support Questionnaire. The original because the items on the physical scale appear to intention had been to develop QoL scores that be nonspecific (4). A study of the BREF in pa- used the importance questions to weight the var- tients with HIV disease seemed to confirm that ious facets in forming domain scores. This relevant themes are omitted and that additional would have allowed people in different cultures disease-specific items would make it more sensi- to place appropriate weights on the various tive to change (21). Williams (31) questioned the facets in expressing their overall QoL; this has value of including several similar items on each generated some interesting analyses of the rela- topic and, indeed, several of the questions do tive perceived importance of the facets in differ- seem repetitive and differ only in shades of ent cultures (29). However, combining the scores meaning. The development of the WHOQOL- was abandoned because it would be impossible BREF has addressed this issue and time will tell to distinguish between people with fair function whether the 100-item version survives given that in facets rated as important and others with poor its internal consistency is only slightly higher function in areas rated as unimportant (30). than that of the BREF, and its sensitivity to Skevington and O’Connell have analyzed the change does not appear greatly superior. correspondence between the objective and the The WHOQOL, one of the newer generic in- importance items and found that for only five struments, has benefited from exceedingly care- facets—mobility, social support, financial re- ful initial development work. It shows promise sources, freedom from anxiety or depression, in many ways, but hard evidence for its validity and work—did the importance items appear to is still limited. Furthermore, it is now in the contribute to explaining overall QoL ratings more challenging second phase of development, (30). This may suggest the need for future modi- in which final versions are established and multi- fications of the content of the WHOQOL. ple tests of validity for different applications are Among the potential concerns about the undertaken. It often occurs that during this WHOQOL, one could include the possibly ex- phase, when the novelty has worn off, the sup- aggerated optimism about its range of applica- port required to complete the project is lost. tion. The WHO suggests that it can be used as a There are warning signs: as of early 2004, the survey tool, an outcome measure for clinical tri- WHO Web sites describing the WHOQOL are als, an instrument for collecting information rel- outdated and the contact links to the WHO per-  General Health Status and Quality of Life 629 sonnel fail to work. The user gains the impres- (7) World Health Organization. WHOQOL: sion that the WHOQOL program is no longer a measuring quality of life. priority. Momentum needs to be maintained, WHO/MSA/MNH/PSF/97.4. Geneva: and it is hoped that the WHO or another agency World Health Organization, Division of will be able to marshal the support to bring this Mental Health; 1997:1–10. (8) WHOQOL Group. The World Health Or- scale to full maturity. ganization quality of life assessment (WHOQOL): position paper from the Address World Health Organization. Soc Sci Med The WHO Web site for the WHOQOL is at www 1995;41:1403–1409. .who.int/evidence/assessment-instruments/qol/. (9) WHOQOL Group. The World Health Or- To obtain copies of the instrument, readers are ganization Quality of Life Assessment advised to contact WHOQOL regional represen- (WHOQOL): development and general tatives who are listed on the Web site. Transla- psychometric properties. Soc Sci Med tions of the instrument are listed on its Web site, 1998;46:1569–1585. but copies may be obtained locally. For example, (10) Skevington SM. Advancing cross-cultural the Australian version of the WHOQOL-100 can research on quality of life: observations be downloaded from www.acpmh.unimelb.edu drawn from the WHOQOL development. .au/whoqol/instruments.html. Qual Life Res 2002;11:135–144. (11) Skevington SM, Bradshaw J, Saxena S. Selecting national items for the References WHOQOL: conceptual and psychometric (1) WHOQOL Group. Study protocol for the considerations. Soc Sci Med World Health Organization project to 1999;48:473–487. develop a quality of life assessment (12) Skevington SM, Tucker C. Designing instrument (WHOQOL). Qual Life Res response scales for the cross-cultural use in 1993;2:153–159. health care: data from the development of (2) WHOQOL Group. The development of the the UK WHOQOL. Br J Med Psychol World Health Organization quality of life 1999;72:51–61. assessment instrument (the WHOQOL). (13) WHOQOL Group. Development of the In: Orley J, Kuyken W, eds. Quality of life World Health Organization WHOQOL- assessment: international perspectives. BREF quality of life assessment. Psychol Berlin: Springer-Verlag, 1994:41–57. Med 1998;28:551–558. (3) Skevington SM, Sartorius N, Amir M, et (14) Skevington SM. Measuring quality of life al. Developing methods for assessing in Britain: introducing the WHOQOL-100. quality of life in different cultural settings. J Psychosom Res 1999;47:449–459. Soc Psychiatry Psychiatr Epidemiol (15) Skevington SM, Wright A. Changes in the 2004;39:1–8. quality of life of patients receiving (4) Bonomi AE, Patrick DL, Bushnell DM, et antidepressant medication in primary care: al. Validation of the United States’ version validation of the WHOQOL-100. Br J of the World Health Organization Quality Psychiatry 2001;178:261–267. of Life (WHOQOL) instrument. J Clin (16) O’Carroll RE, Smith K, Couston M, et al. Epidemiol 2000;53:1–12. A comparison of the WHOQOL-100 and (5) World Health Organization. Report of the the WHOQOL-BREF in detecting change meeting of investigators on quality of life. in quality of life following liver MNH/PSF/92.6. Geneva: WHO Division transplantation. Qual Life Res of Mental Health; 1992:1–52. 2000;9:121–124. (6) Szabo S. The World Health Organization (17) Power M, Harper A, Bullinger M, et al. quality of life (WHOQOL) assessment The World Health Organization instrument. In: Spilker B, ed. Quality of WHOQOL-100: tests of the universality of life and pharmacoeconomics in clinical quality of life in 15 different cultural trials. Philadelphia: Lippincott-Raven, groups worldwide. Health Psychol 1996:355–362. 1999;18:495–505. 630 Measuring Health (18) Skevington SM, Carse MS, Williams AC. The WHO quality of life assessment Validation of the WHOQOL-100: pain instrument (WHOQOL-Bref ): the management improves quality of life for importance of its items for cross-cultural cancer pain patients. Clin J Pain research. Qual Life Res 2001;10:711–721. 2001;17:264–275. (30) Skevington SM, O’Connell KA. Can we (19) Skevington SM. Investigating the identify the poorest quality of life? relationship between pain and discomfort Assessing the importance of quality of life and quality of life, using the WHOQOL. using the WHOQOL-100. Qual Life Res Pain 1998;76:395–406. 2004;13:23–34. (20) Skevington SM, Lotfy M, O’Connell KA. (31) Williams JI. Ready, set, stop. Reflections The World Health Organization’s on assessing quality of life and the WHOQOL-BREF quality of life WHOQOL-100 (U.S. version). J Clin assessment: psychometric properties and Epidemiol 2000;53:13–17. results of the international field trial. Qual Life Res 2004;13:299–310. The Sickness Impact Profile (21) Fang C-T, Hsiung P-C, Yu C-F, et al. Validation of the World Health Organiz- (Marilyn Bergner, 1976, ation quality of life instrument in patients Revised 1981) with HIV infection. Qual Life Res Purpose 2002;11:753–762. The Sickness Impact Profile (SIP) indicates the (22) Noerholm V, Groenvold M, Watt T, et al. Quality of life in the Danish general changes in a person’s behavior due to sickness population—normative data and validity of (1). It is broadly applicable and was intended for WHOQOL-BREF using Rasch analysis and use in measuring the outcomes of care, in health item response theory models. Qual Life surveys, in program planning and policy forma- Res 2004;13:531–540. tion, and in monitoring patient progress (2). (23) WHOQOL Group. Initial steps to developing the World Health Organiz- Conceptual Basis ation’s quality of life instrument Conceptual development originated from the (WHOQOL) module for international observation that the ultimate aim of most assessment of HIV/AIDS. AIDS Care health care is to reduce sickness or modify its 2003;15:347–357. effect on everyday activities (3). “Sickness” de- (24) Skevington SM, O’Connell KA. Measuring notes the individual’s own experience of illness, quality of life in HIV and AIDS: a review of the recent literature. Psychol Health perceived through its effect on daily activities, 2003;18:331–350. feelings, and attitudes. In this sense, sickness (25) Jirojanakul P, Skevington SM. Developing differs from disease, which denotes a profes- a quality of life measure for children aged sional definition of illness based on clinical ob- 5–8 years. Br J Health Psychol servations (4). 2000;5:299–321. The SIP measures health status by assessing (26) Mason VL, Skevington SM, Osborn M. the way sickness changes daily activities and be- Development of a pain and discomfort havior. Its concentration on behavior holds sev- module for use with the WHOQOL-100. eral advantages over recording feelings or Qual Life Res 2004;13:1139–1152. clinical reports. Behaviors are observable and so (27) Younis MS, AlKaisr H, Young A. Quality accessible to external validation; feelings are of life among college students in Iraq. variable and can be hard to measure. Even if Quality of Life Newsletter 2003;30:3–4. (28) Min SK, Kim KI, Lee CI, et al. clinical parameters may not improve in a chronic Development of the Korean version of disease, improvements in behavioral limitations WHO Quality of Life scale and may be achieved. Behavioral reports can be ob- WHOQOL-BREF. Qual Life Res tained whether or not a patient is receiving care; 2002;11:593–600. they may also be less subject to cultural bias (29) Saxena S, Carlson J, Billington R, et al. than reports of feelings (1).  General Health Status and Quality of Life 631 The items in the SIP all concentrate on pp334–341). Exhibit 10.27 indicates the scope changes in performance rather than capacity (3). of the SIP and illustrates the items used. The behaviors included in the profile are consid- Respondents check only the items that de- ered significant from individual, social, and scribe them on a given day and are related to health care points of view. They are held to rep- their health, although the actual medical condi- resent universal patterns of limitations that may tion is not an issue (1). Item weights indicate the be affected by sickness or disease, regardless of relative severity of limitation implied by each the specific conditions, treatment, individual statement. The weights were derived from equal- characteristics, or prognosis concerned (1). appearing interval scaling procedures involving more than 100 judges (6; 7). Description Work on the SIP began in 1972; statements de- The scaling permits the calculation of an SIP scribing changes in behavior attributable to sick- percent score which is the sum of the scale ness were compiled from professionals, from values of items checked divided by the sum of interviews with healthy and sick laypeople and the scale values for all items multiplied by from a literature review. Following a succession 100. This SIP percent score may be calculated of field trials (total sample = 1,108), the proto- for the entire SIP (the SIP overall score) as type, containing 312 statements grouped into 14 well as for each of the categories. (8, p58) categories of activities, was refined to a final ver- sion with 136 statements in 12 categories. The The 12 categories may be scored separately, coverage of the scale is as follows: or (as shown in the exhibit) two dimension scores may be formed: ambulation, mobility, Dimensions Categories and body care and movement can be summed to (number of items) (number of items) form a physical score, and social interaction, Physical (45) Ambulation (12) alertness, emotional behavior, and communica- Mobility (10) tion can form a psychosocial dimension score Body Care and (6). The remaining five categories are scored sep- Movement (23) arately. Psychosocial (48) Communication (9) The profile can be administered by an inter- Alertness viewer in 20 to 30 minutes (9), or it can be self- Behavior (10) administered. A small study suggested that it Emotional may be feasible to administer the physical di- Behavior (9) mension to frail patients by telephone (10). It Social Iteraction (20) takes five to ten minutes to score the SIP (11, Independent Sleep and Rest (7) p1004). User’s and trainer’s manuals are avail- categories (43) Eating (9) able (see Address section). Self-report has been Work (9) compared with proxy report (typically by the Home caregivers of elderly respondents), often showing Management (10) considerable differences. Rothman et al. found Recreation and significant differences between self- and proxy Pastimes (8) ratings for both physical and psychosocial scores, with proxies rating patients as more im- The SIP is composed of statements such as “I paired (12, p119). DeBruin et al. concluded that have difficulty reasoning and solving problems” the two sources of information cannot be re- and “I do not walk at all,” each of which de- garded as interchangeable (11, p1009). scribes a change in behavior and specifies the ex- tent of limitation. The full instrument is too long Reliability to reproduce here; a copy is included in the ap- In a comprehensive discussion of reliability, Pol- pendix to the book by Wenger et al. (5, lard et al. reported the results of various tests of Exhibit 10.27 The Sickness Impact Profile: Categories and Selected Items Items describing Dimension Category behavior related to Selected items Independent SR Sleep and Rest I sit during much of the day categories I sleep or nap during the day E Eating I am eating no food at all, nutrition is taken through tubes or intra- venous fluids I am eating special or different food W Work I am not working at all I often act irritable toward my work associates HM Home management I am not doing any of the mainte- nance or repair work around the house that I usually do I am not doing heavy work around the house RP Recreation and pastimes I am going out for entertainment less I am not doing any of my usual physical recreation or activities Physical A Ambulation I walk shorter distances or stop to rest often I do not walk at all M Mobility I stay within one room I stay away from home only for brief periods of time BCM Body care and movement I do not bathe myself at all, but am bathed by someone else I am very clumsy in body movements Psychosocial SI Social interaction I am doing fewer social activities with groups of people I isolate myself as much as I can from the rest of the family AB Alertness behavior I have difficulty reasoning and solving problems, for example, making decisions, learning new things I sometimes behave as if I were confused or disoriented in place or time, for example, where I am, who is around, directions, what day it is EB Emotional behavior I laugh or cry suddenly I act irritable and impatient with myself, for example, talk badly about myself, swear at myself, blame myself for things that happen C Communication I am having trouble writing or typing I do not speak clearly when I am under stress Reproduced from Bergner M, Bobbitt RA, Carter WB, Gilson BS. The Sickness Impact Profile: development and final revision of a health status measure. Med Care 1981;19:789, Table 1. With permission. 632  General Health Status and Quality of Life 633 the 235-item and 146-item versions of the SIP version had a lower alpha of 0.81 (6, p793). In a (13; 14), whereas reliability results for the final study involving patients in long-term care facili- 136-item version were reported by Bergner et al. ties, the alpha for the overall score was 0.95, (6). Test-retest reliability in these various trials whereas section score alphas ranged from 0.59 was consistently high (range, 0.88–0.92) for the to 0.93 (19, Table 7). An intraclass correlation overall scores. Reliability was higher for the of 0.92 was reported on a sample of patients interviewer-administered version (0.97) than with musculoskeletal disorders (20). Alpha val- the self-administered (0.87). Reproducibility of ues for the category scores ranged from 0.60 to the individual items averaged 0.50, and for the 0.84 in another sample of residents in long-term 12 category scores it was 0.82 (6, pp793, 796). care facilities (21, Table 1). The summary by De- The reliability of overall scores did not vary by Bruin et al. cites alpha coefficients ranging from type or level of sickness; the short form was as 0.91 to 0.95 for the overall score, from 0.84 to reliable as the long form (6; 13; 14). Deyo et al. 0.93 for dimension scores, and from 0.60 to applied the SIP in a study of 79 patients with 0.90 for category scores (11, Table 2). arthritis (mean age, 57 years) (15). Test-retest Inter-rater reliabilities of r = 0.92 and reliability (Spearman correlation) was 0.91 for kappa = 0.87 have been reported for the overall 23 patients (kappa 0.87) (15, Table 6). Lower score (11, Table 2). intra-class correlation values, of 0.61 for the overall score and ranging from 0.30 to 0.93 for Validity the dimension scores, were reported for Spanish- Early validation studies compared the SIP with and English-speaking samples (16, Table 1). other subjective ratings and with clinical assess- Nanda et al. obtained an intraclass retest coeffi- ments. The SIP scores correlated 0.69 with a cient of 0.77 (range 0.61 to 0.88 for dimension self-assessment of functional limitation; 0.63 scores) (17, Table 4). DeBruin et al. summarized with a self-assessment of sickness; 0.50 with a results from several studies: test-retest reliability clinician’s assessment of limitation; and 0.40 for the overall SIP score ranges from 0.75 to with a clinician’s assessment of sickness (6, 0.92; reliabilities for dimension scores range Table 4). Responses to the SIP were compared from 0.79 to 0.91; and category score relia- with clinical indicators for patients in diagnostic bilities range from 0.50 to 0.95 (11, Table 2). groups in which clinical tests could be expected Beckerman et al. reported intraclass retest corre- to reflect patient functioning to varying extents. lations ranging from 0.92 (overall score) to For 15 patients who had hip replacement, for ex- 0.899 (physical dimension) and 0.85 (psychoso- ample, the overall SIP score correlated 0.81 with cial dimension) down to 0.63 (recreation and an index of physical functioning. A similar com- pastimes) (18, Table 2). They also provided an parison for 15 patients with rheumatoid arthritis innovative approach to illustrating retest reliabil- yielded a correlation of 0.66. For patients with ity in terms of the “smallest real difference” in hyperthyroid disease, the overall SIP score corre- scores (whether over time, or between patients) lated 0.41 with thyroid function measurements that should be viewed as indicating a real con- (6, Table 8). Read et al. compared SIP scores trast, beyond unreliability. The results suggest with indicators of overall health; correlations that a change of six or more items on the physi- were 0.53 with the number of self-reported cal dimension, or of nine or more on the psy- symptoms, 0.57 with a mental health scale, 0.63 chosocial, would form a reliable indication of with time achieved on treadmill, and 0.34 with change (18, Table 2). forced expiratory capacity (22, pS15). Deyo et Internal consistency has been widely exam- al. reported correlations between SIP physical ined for the U.S. version, and also for many of and psychosocial scores and several indicators of the translated versions of the SIP. Alpha coeffi- disease severity for 79 arthritic patients (15). cients for the overall score were 0.97 for the The results are shown in Table 10.5. Correla- 235-item version and 0.94 for the final 136-item tions with the American Rheumatism Associa- version, although a mailed, self-administered tion’s four-point functional rating were 0.36 for 634 Measuring Health Table 10.5 Correlations between SIP Scores and Criterion Variables SIP Overall Score Physical Dimension Psychosocial Dimension Duration of rheumatoid arthritis 0.26* 0.36† 0.03 Morning stiffness 0.23* 0.15 0.32† Hematocrit −0.29* −0.26* −0.12 Sedimentation rate (ESR) 0.36* 0.44† 0.11 Anatomic stage 0.17 0.31† −0.01 Evidence of mental health problems 0.11 −0.03 0.27† * p<0.05 † p<0.01 Adapted from Deyo RA, Inui TS, Leininger JD, et al. Measuring functional outcomes in chronic disease: a comparison of traditional scales and self-administered health status questionnaire in patients with rheumatoid arthritis, Med Car. 1983;21:187, Table 2 the SIP physical dimension and 0.02 for the psy- low correlations presumably reflect the broader chosocial dimension (15, Table 3). Correlations scope of the SIP overall score; other studies show with neurological tests in a sample of patients clearly that the association increases as succes- with head injury were modest, falling in the sively narrower samples of SIP items are com- range of 0.25 to 0.40; this was explained in pared with the ADL scale. Thus, the ADL scale terms of the narrow focus of the neurological correlated 0.64 with a combined score for the tests, compared with the broad scope of the SIP five SIP categories that included ADL behaviors (23, p54). DeBruin et al. found correlations with (8, p65). A higher correlation of 0.74 was ob- self-assessed health status mainly ranging from tained between the SIP physical dimension alone 0.55 to 0.65; correlations with clinicians’ ratings and the ADL in a study in a long-term care facil- were lower, typically in the range 0.45 to 0.55 ity; the equivalent correlation with the Barthel (11, Table 3). The convergent and discriminant Index was also 0.74 (21, pS162). An even higher validity of the SIP subscores were illustrated by a correlation of 0.88 was found between the ADL comparison of patients in pain and patients with score and a score based on the 14 SIP items most depression (24, Table 3). There was no differ- closely matching the content of the ADL scale; ence between the two groups on sleep and rest, the equivalent correlation with the Barthel Index mobility, or recreation (both groups scoring as was 0.90 (21, pS163). This consistent pattern of moderately disabled); the patients with depres- correlations illustrates the contrast between a sion scored significantly higher on emotional be- specific scale (here, the function-specific ADL havior, home management, and alertness, items) and a generic scale like the SIP. whereas patients in pain scored significantly SIP scores correlated 0.55 with the National higher on body care, social interaction, and am- Health Interview Survey questions on activity bulation. limitation (6, p795). The overall scores of the The SIP has been compared with most lead- SIP and the Arthritis Impact Measurement ing health indexes; DeBruin et al. show correla- Scales (AIMS) correlated 0.83; their physical tions of the SIP overall score with 13 other scales correlated 0.86, and their psychosocial measures, almost all exceeding 0.50 (11, Table scales correlated 0.65 (16, p964). A correlation 3). The rank correlation between SIP overall of 0.73 with the abbreviated AIMS was reported scores and Katz’s Index of Activities of Daily (25, Table 3). In a sample of 27 elderly men, the Living (ADL) was 0.46 for 73 rehabilitation pa- physical function scales of the SIP and the SF-36 tients (8, p65) and 0.42 in a sample of patients correlated 0.78 and social function scores corre- from long-term care facilities (19, Table 6). The lated 0.67 (9, p693). In a sample of elderly am-  General Health Status and Quality of Life 635 bulatory patients, correlations between the SIP proved less sensitive to change than that of the and the SF-36 were likewise high: −0.75 be- SF-36, the AIMS, and the Functional Status tween the SIP total score and the SF-36 physical Questionnaire (25, Table 4). Sensitivity to summary scale, and −0.38 between the SIP and change following myocardial infarction was the SF-36 mental scale. The SIP psychosocial modest, but slightly better than that for the NHP score correlated −0.57 with the SF-36 mental or the McMaster Health Index Questionnaire scale (26, Table 4). Convergent correlations be- (33, Tables 2 to 4). The SIP proved no better tween SIP subscales and equivalent scales on the than a seven-point patient self-rating scale in re- Health Utilities Index ranged from 0.43 to 0.72 flecting clinically judged changes, and both and were markedly higher than the correlations showed similar correlations with physical signs between scales measuring different themes (27, such as hematocrit, erythrocyte sedimentation Table 9). rate (ESR), and grip strength (32, Table 5). A The SIP psychosocial subscale correlated 0.72 similar caution was raised by MacKenzie et al., with the Carroll Rating Scale for Depression; the who found that the SIP was comparatively insen- correlation between the physical scale and the sitive to change (as judged by asking patients di- Carroll was 0.44 (28, p793). A correlation of rectly about their change in health status) (34). It 0.54 between the SIP overall score and the Geri- identified deterioration more reliably than im- atric Depression Scale has been reported (19, provement, and variance in patients who appar- Table 6). Correlations with the Katz Adjustment ently remained stable was large (34, p436). For Scale were 0.45 for the overall score and 0.57 patients with low back pain, briefer disease- for the withdrawal subscale (29, Table 5). An- specific scales were found to be more responsive derson et al. comment that “Perhaps reflecting to change than the complete SIP (35). DeBruin et its behavioral content, the SIP psychosocial al. concluded “a definite conclusion about the component has demonstrated only moderate responsiveness of the SIP cannot yet be correlations (r = 0.40 to 0.60) with traditional drawn. . . . Findings suggest that the instrument depression and anxiety measures” (30, p375). is not sensitive to small, daily changes in a pa- One factor analysis of the SIP identified two tient’s situation, but changes occurring over a main factors, distinguishing the physical and longer period of time seem to be mirrored in SIP psychosocial subscales (28). Other analyses have scores” (11, p1011). A more positive conclusion also only partially replicated the original factor can be drawn from comparison by Liang et al. of structure (17, p587). Several other studies have five health measures: the effect size for the SIP reported factor analyses of the SIP combined was second only to that of the AIMS for mobil- with other measures; very few report on the SIP ity and overall scores, and the SIP was also sec- alone (11, p1007). ond on the social score, this time to the Health The ability of the SIP to reflect change has Assessment Questionnaire (36, Tables 2, 3). been reported in various studies; a review of seven studies has been undertaken (31). An ef- Alternative Forms fect size of 0.52 was reported for musculoskele- The SIP was adapted for use in England and re- tal patients, somewhat lower than that of the named the Functional Limitations Profile (FLP). SF-36 or the Nottingham Health Profile (NHP) Linguistic changes were made and scale weights (20). Deyo and Inui compared changes in SIP were recalculated, although these agreed closely scores over time with clinicians’ ratings of with the original weights for the United States change in the status of arthritic patients. The SIP (37). The FLP is the version normally used in showed only a 50% sensitivity to clinically esti- British studies. In a study of 105 patients with mated improvement and a 43% sensitivity in de- arthritis, the FLP overall score correlated −0.66 tecting clinical deterioration (32, Table 4). with grip strength, 0.62 with the Ritchie index, Sensitivity to change for patients who had un- 0.29 with ESR, and 0.49 with a rating of morn- dergone hip replacement was compared with ing stiffness (38, Table 2). Sensitivity to im- that of other measures; the SIP overall score provement in disease state over 15 months was a 636 Measuring Health modest 55% (specificity, 56%), whereas sensi- version for use in long-term care facilities; this tivity to deterioration was 73% (specificity correlated 0.98 with the complete SIP (19). 69%). These results for the FLP compared Roland and Morris took 24 items relevant to low closely with those obtained for the Health As- back pain from the SIP to form the Roland Dis- sessment Questionnaire. Information on the FLP ability Questionnaire (39); this version may be may be obtained from http://depts.washington more responsive to change in patients with low .edu/yqol/docs/FLPOrderForm.pdf. back pain than the complete SIP (35) and it has The French MAPI Research Institute has co- subsequently been modified by Patrick et al. ordinated translation of the SIP into 11 Europe- (40). Temkin et al. proposed a version of the SIP an languages, including Dutch (41; 42), Spanish for patients with head injuries. This added new (16; 43–46), French (47), Swedish (48), Italian items relevant to head injury, deleted those items (49), and other languages (11; 30). Weights for not relevant to individual respondents (e.g., the Italian (49) and Spanish versions (50) have items about work for those not working), and al- been compared with the original values, indicat- tered the relative weighting between sections ing certain items (e.g., on social interaction) for (23; 29). Although these modifications some- which the translation does not seem to produce what improved discriminant validity, Temkin et culturally equivalent items. al. concluded that the advantages were not “suf- The length of the SIP has stimulated several ficiently large or consistent to provide a practical abbreviations, which seem to have converged advantage over the SIP . . . Until other factors, on a 68-item version, the SIP68 (31; 51–53). such as emotional status and response style, are Items that loaded most highly in a principal better controlled, little benefit is likely to be ob- components analysis were retained; an overall tained from creating disease-specific psychoso- score runs from zero (best health) to 68. The cial measures” (29, pS44). items are shown in reference (53, pp444–445). Modifications to the SIP instructions have The items form six subscales: somatic auton- been proposed for use with patients in long-term omy, mobility control, psychological autonomy care facilities (21). These included abbreviation, and communication, emotional stability, social adding examples, and interpreting the theme of behavior, and mobility range. The first two impairments being “due to your health” for peo- form a physical dimension with 29 items; the ple with chronic conditions. The SIP items were third and fourth comprise a psychological di- changed from statements into questions by mension with 17 items, and the last two sub- Johnson et al. to facilitate administration in an scales form a social dimension with 22 items. interview (55). Unweighted dimension scores count the num- ber of items checked (52). Nanda et al. re- Commentary ported a correlation of 0.94 between the SIP The SIP was developed with exemplary care and and the SIP68; the retest intraclass correlation thoroughness, and with the death of Marilyn was 0.88 for the SIP68 overall score, compared Bergner at the end of 1992, the health measure- to 0.77 for the complete SIP (17, Table 4). Co- ment field lost one of its most respected leaders. efficient alpha was 0.92 (53, p442) and princi- The quality of the SIP is tacitly acknowledged in pal components analysis confirmed the that it frequently serves as the criterion against presumed structure of the instrument (53, which other scales are evaluated. An illustration Table 3). Correlations with equivalent SF-36 of the many types of study in which the SIP has scores fell in the range −0.49 to −0.58 (17, been used is given by DeBruin et al. (11, Table 1). Table 5). The SIP68 appears to be as responsive The SIP illustrates the philosophy of the to change as is the SIP136 (31). generic instrument: it seeks to be applicable in Other abbreviations and condition-specific any country, to all age groups, and to any med- versions have been produced for particular dis- ical condition. It appears to achieve this objec- eases, including a 64-item version for patients tive. For example, the study that modified the with rheumatoid arthritis (54), and a 66-item SIP for patients with head injuries found that the  General Health Status and Quality of Life 637 revised version performed little better than the generic SIP; the authors concluded that the stan- References dard version should be used (23). Nonetheless, (1) Bergner M, Bobbitt RA, Kressel S. The this broad applicability may come at a cost. The Sickness Impact Profile: conceptual full SIP is long, and although it can be used suc- formulation and methodology for the cessfully in general practice settings (41), the development of a health status measure. Int SIP68 version should be considered, as it appears J Health Serv 1976;6:393–415. to perform almost equally well. (2) The Sickness Impact Profile: a brief The reliability results for the SIP are good, summary of its purpose, uses and administration. Seattle: University of and it appears valid as a discriminative method, Washington, Department of Health Ser- particularly in group analyses. Correlations with vices, 1978. various clinical assessments are moderate to (3) Gilson BS, Bergner M, Bobbitt RA, et al. good, falling in the range 0.40 to 0.65; correla- The Sickness Impact Profile: final tions with self-ratings are higher and follow a development and testing, 1975–1978. logical pattern in which correlations between Seattle: Department of Health Services, narrow (e.g., function-specific) scales and the University of Washington, 1979. SIP overall are modest, but successively nar- (4) Gilson BS, Gilson JS, Bergner M, et al. The rower selections of SIP items correlate increas- Sickness Impact Profile: development of an ingly highly with the specific scale. The evidence outcome measure of health care. Am J gives strong support for the concurrent validity Public Health 1975;65:1304–1310. (5) Bergner M. The Sickness Impact Profile of the SIP. Its use as an evaluative index for use (SIP). In: Wenger NK, Mattson ME, in detecting changes, however, appears less cer- Furberg CD, et al., eds. Assessment of tain. The emphasis on behaviors and functional quality of life in clinical trials of status may limit sensitivity to change because cardiovascular therapies. New York: these are not sensitive to minor changes that LeJacq, 1984:152–159. might be picked up by questions on mood or (6) Bergner M, Bobbitt RA, Carter WB, et al. feelings. Length may actually be a disadvantage The Sickness Impact Profile: development here, for diluted among 136 items, a change in and final revision of a health status mea- one area is unlikely to be reflected in changes in sure. Med Care 1981;19:787–805. the overall score. The SIP focuses on relatively (7) Carter WB, Bobbitt RA, Bergner M, et al. severe levels of disability and so, compared with Validation of an interval scaling: the Sickness Impact Profile. Health Serv Res scales such as the SF-36, it will suffer ceiling ef- 1976;11:516–528. fects and will be less able to show improvements (8) Bergner M, Bobbitt RA, Pollard WE, et al. in function (9). The Sickness Impact Profile: validation of a The SIP is well established, has been exten- health status measure. Med Care sively used, and is appropriate where a compre- 1976;14:57–67. hensive assessment is required; it can be applied (9) Weinberger M, Samsa GP, Hanlon JT, et to populations with a wide range of levels of al. An evaluation of a brief health status sickness. measure in elderly veterans. J Am Geriatr Soc 1991;39:691–694. Address (10) Morishita L, Boult C, Ebbitt B, et al. Copies of the SIP, including an 89-page manual Concurrent validity of administering the Geriatric Depression Scale and the physical and administration instructions, are available at functioning dimension of the SIP by cost from the Medical Outcomes Trust, www. telephone. J Am Geriatr Soc outcomes-trust.org/instruments.htm. 1995;43:680–683. Information on translations is available from (11) DeBruin AF, De Witte LP, Stevens F, et al. the MAPI Research Institute, 27 rue de la Sickness Impact Profile: the state of the art Villette, 69003 Lyon, France (www.mapi of a generic functional status measure. Soc -research-inst.com/). Sci Med 1992;35:1003–1014. 638 Measuring Health (12) Rothman ML, Hedrick SC, Bulcroft KA, et modifications to the Sickness Impact Profile al. The validity of proxy-generated scores for head injury. J Clin Epidemiol as measures of patient health status. Med 1988;41:47–57. Care 1991;29:115–124. (24) Pilowsky I, Bassett DL. Pain and depression. (13) Pollard WE, Bobbitt RA, Bergner M, et al. Br J Psychiatry 1982;141:30–36. The Sickness Impact Profile: reliability of a (25) Katz JN, Larson MG, Phillips CB, health status measure. Med Care Comparative measurement sensitivity of 1976;14:146–155. short and longer health status instruments. (14) Pollard WE, Bobbitt RA, Bergner M. Med Care 1992;30:917–925. Examination of variable errors of measure- (26) Andresen EM, Rothenberg BM, Kaplan ment in a survey-based social indicator. Soc RM. Performance of a self-administered Indicat Res 1978;5:279–301. mailed version of the Quality of Well-Being (15) Deyo RA, Inui TS, Leininger JD, et al. (QWB-SA) questionnaire among older Measuring functional outcomes in chronic adults. Med Care 1998;36:1349–1360. disease: a comparison of traditional scales (27) Costet N, Le Gales C, Buron C, et al. and a self-administered health status French cross-cultural adaptation of the questionnaire in patients with rheumatoid health utilities indexes Mark 2 (HUI2) and arthritis. Med Care 1983;21:180–192. 3 (HUI3) classification systems. Qual Life (16) Hendricson WD, Russell IJ, Prihoda TJ, et Res 1998;7:245–256. al. An approach to developing a valid (28) Brooks WB, Jordan JS, Divine GW, et al. Spanish language translation of a health- The impact of psychologic factors on mea- status questionnaire. Med Care surement of functional status: assessment 1989;27:959–966. of the Sickness Impact Profile. Med Care (17) Nanda U, McLendon PM, Andresen EM, 1990;28:793–804. et al. The SIP68: an abbreviated Sickness (29) Temkin NR, Dikmen S, Machamer J, et al. Impact Profile for disability outcomes General versus disease-specific measures: research. Qual Life Res 2003;12:583–595. further work on the Sickness Impact Profile (18) Beckerman H, Roebroeck ME, Lankhorst for head injury. Med Care GJ, et al. Smallest real difference, a link 1989;27(suppl):S44–S53. between reproducibility and (30) Anderson RT, Aaronson NK, Wilkin D. responsiveness. Qual Life Res Critical review of the international 2001;10:571–578. assessments of health-related quality of life. (19) Gerety MB, Cornell JE, Mulrow CD, et al. Qual Life Res 1993;2:369–395. The Sickness Impact Profile for Nursing (31) de Bruin AF, Diederiks JPM, De Witte LP, Homes (SIP-NH). J Gerontol et al. Assessing the responsiveness of a 1994;49:M2–M8. functional status measure: the Sickness (20) Beaton DE, Bombardier C, Hogg-Johnson Impact Profile versus the SIP68. J Clin S. Choose your tool: a comparison of the Epidemiol 1997;50:529–540. psychometric properties of five generic (32) Deyo RA, Inui TS. Toward clinical health status instruments in workers with applications of health status measures: soft tissue injuries. Qual Life Res sensitivity of scales to clinically important 1994;3:50–56. changes. Health Serv Res (21) Rothman ML, Hedrick S, Inui T. The 1984;19:275–289. Sickness Impact Profile as a measure of the (33) Taylor R, Kirby B, Burdon D, et al. The health status of noncognitively impaired assessment of recovery in patients after nursing home residents. Med Care myocardial infarction using three generic 1989;27(suppl):S157–S167. quality-of-life measures. J Cardiopulm (22) Read JL, Quinn RJ, Hoefer MA. Measur- Rehabil 1998;18:139–144. ing overall health: an evaluation of three (34) MacKenzie CR, Charlson ME, DiGioia D, important approaches. J Chronic Dis et al. Can the Sickness Impact Profile mea- 1987;40(suppl 1):S7–S21. sure change? An example of scale (23) Temkin N, McLean A, Jr., Dikmen S, et al. assessment. J Chronic Dis Development and evaluation of 1986;39:429–438.  General Health Status and Quality of Life 639 (35) Deyo RA, Centor RM. Assessing the (47) Chwalow AJ, Lurie A, Bean K. A French responsiveness of functional scales to version of the Sickness Impact Profile (SIP): clinical change: an analogy to diagnostic stages in the cross cultural validation of a test performance. J Chronic Dis generic quality of life scale. Fundam Clin 1986;39:897–906. Pharmacol 1992;6:319–326. (36) Liang MH, Fossel AH, Larson MG. (48) Sullivan M, Ahlmen M, Archenholtz B. Comparisons of five health status Measuring health in rheumatic disorders by instruments for orthopedic evaluation. Med means of a Swedish version of the Sickness Care 1990;28:632–642. Impact Profile: results from a population (37) Patrick DL, Sittampalam Y, Somerville SM, study. Scand J Rheumatol et al. A cross-cultural comparison of health 1986;15:193–200. status values. Am J Public Health (49) Marchionni N, Ferrucci L, Baldasseroni S, 1985;75:1402–1407. et al. Item re-scaling of an Italian version (38) Fitzpatrick R, Newman S, Lamb R, et al. A of the sickness impact profile: effect of age comparison of measures of health status in and profession of the observers. J Clin rheumatoid arthritis. Br J Rheumatol Epidemiol 1997;50:195–201. 1989;28:201–206. (50) Badia X, Alonso J. Re-scaling the Spanish (39) Roland M, Morris R. A study of the version of the Sickness Impact Profile: an natural history of back pain. Part I: opportunity for the assessment of cross- Development of a reliable and sensitive cultural equivalence. J Clin Epidemiol measure of disability in low-back pain. 1995;48:949–957. Spine 1983;8:141–144. (51) de Bruin AF, Buys M, De Witte LP, et al. (40) Patrick DL, Deyo RA, Atlas SJ, et al. The sickness impact profile: SIP68, a short Assessing health-related quality of life in generic version. First evaluation of the patients with sciatica. Spine reliability and reproducibility. J Clin 1995;20:1899–1909. Epidemiol 1994;47:863–871. (41) Jacobs HM, Luttik A, Touw-Otten (52) de Bruin AF, Diederiks JPM, De Witte LP, FWMM, et al. Measuring impact of et al. The development of a short generic sickness in patients with nonspecific version of the Sickness Impact Profile. J abdominal complaints in a Dutch family Clin Epidemiol 1994;47:407–418. practice setting. Med Care (53) Post MWM, de Bruin A, de Witte L, et al. 1992;30:244–251. The SIP68: a measure of health-related (42) Jacobs HM, Luttik A, Touw-Otten FW, et functional status in rehabilitation medicine. al. The Sickness Impact Profile (SIP): Arch Phys Med Rehabil 1996;77:440–445. results of an evaluation study of the Dutch (54) Sullivan M, Ahlmen M, Bjelle A, et al. version. Ned Tijdschr Geneeskd Health status assessment in rheumatoid 1990;134:1950–1954. arthritis. II. Evaluation of a modified (43) Esteva M, Gonzalez N, Ruiz M. Reliability Shorter Sickness Impact Profile. J and validity of a Spanish version of the Rheumatol 1993;20:1500–1507. Sickness Impact Profile. Arthritis Rheum (55) Johnson J, King K, Murray R. Measuring 1992;35(suppl):S219. the impact of sickness on functions of (44) Gilson BS, Erickson D, Chavez CT, et al. A radiation therapy patients. Oncol Nurs Chicano version of the Sickness Impact Forum 1983;10:36–39. Profile. Cult Med Psychiatry 1980;4:137–150. (45) Deyo RA. Pitfalls in measuring the health The Nottingham Health Profile status of Mexican Americans: comparative (Sonja Hunt, 1981) validity of the English and Spanish Sickness Impact Profile. Am J Public Health Purpose 1984;74:569–573. The Nottingham Health Profile (NHP) was de- (46) Badia X, Alonso J. Reliability of the signed to give a brief indication of perceived Spanish version of the Sickness Impact physical, social, and emotional health problems Profile. Qual Life Res 1994;3:65. (1). Originally intended for use in primary med- 640 Measuring Health ical care settings, the NHP has also been used to hobbies, and holidays (i.e., vacations) (7). Part II assess need for care in health surveys and has is optional because some items (e.g., work, sex been used as an outcome measure in clinical tri- life) may not be applicable (8, p5), and it is now als (2). rarely used. Yes/no responses are used through- out. The NHP is self-administered and takes Conceptual Basis about ten minutes to complete. Exhibit 10.28 The design and content of the NHP were influ- shows the 1995 version of the scale items, the enced by the Sickness Impact Profile (SIP). One section to which each pertains, and the scale difference, however, is that the NHP asks about weights. feelings and emotional states directly, rather Hunt and McEwen recommend that section than through changes in behavior. The emphasis scores in Part I be presented as a profile. The is on the respondent’s subjective assessment of items were scaled for severity using a paired her health status: this is seen as the major factor comparisons technique involving 1,200 outpa- predicting use of medical services and satisfac- tient interviews (7; 9). The weights are trans- tion with outcomes (3). The questions reflect the formed to yield scores from 0 (no problems) to World Health Organization (WHO) definition 100 (all items checked) for each scale. An often of disability and the profile may be viewed as an used simpler scoring system counts the number indicator of perceived distress. of affirmative responses in each section. In three samples of respondents, the weighted and un- Description weighted scores correlated 0.98 or higher (10, The pool of items used in constructing the NHP p1415), raising the question of the usefulness of was formed through surveys of 768 patients the weights. Although not recommended by the with acute and chronic ailments, supplemented developers of the NHP, some users present re- by items drawn from other health indexes such sults as an overall score, and O’Brien et al. sug- as the SIP. Each item refers to gested three options for this which all provide a zero-to-one (poor-to-good health) overall score departures from “normal” functioning be- (11). First, the proportion of the 38 items to cause, in the field of health especially, it is which an affirmative response is given is sub- easier to obtain, record, and provide some tracted from 1. Second, the item weights for measure of departures from the norm than it items answered affirmatively are added, and this is to specify the norm itself. Respondents total is divided by 600 (the sum of all item were asked to answer “Yes” or “No” accord- weights); the result is subtracted from 1. Third, ing to whether or not they feel the item ap- because there are different numbers of state- plies to them “in general.” (4, p282) ments in the dimensions, a weight can be used to give each section the same contribution to the The original version of the NHP (called the Not- overall score. In this method, weights for items tingham Health Index) contained 33 items; it answered affirmatively are summed within each was tested on rehabilitation patients (5) and on section, and the section scores are multiplied by patients undergoing hip surgery (2). The revised one sixth times the number of items in that sec- version, called the NHP, includes two parts. Part tion; the resulting section scores are added and I contains 38 items grouped into six sections: this sum subtracted from 1 (11, ppS152–S153). physical abilities (in Exhibit 10.28 designated These three scores provided virtually identical PA, with 8 items), pain (P = 8 items), sleep results (11, pS148). Replications of the original (S = 5), social isolation (SI = 5), emotional reac- scaling exercise have been undertaken in other tions (ER = 9), and energy level (EL = 3 items) settings (12). (6). Part II provides a brief indicator of handicap Part II of the NHP is scored by summing the and contains seven items that record the effect of number of positive responses; no item weights health problems on occupation, jobs around the are used. house, personal relationships, social life, sex life, Exhibit 10.28 The Nottingham Health Profile, Showing the Sections on Which Each Item Is Scored and the Weights for Positive Responses Note: EL = energy level, P = pain, ER = emotional reactions, S = sleep, SI = social isolation, PA = physical abilities. The columns headed Section and Weight are not included in the version used by the repondent. Part I Listed below are some problems people might have in their daily lives. Read the list carefully and put a tick in the box under Yes for any problem that applies to you at the moment. Tick the box under No for any problem that does not apply to you. Please answer every question. If you are not sure whether to answer Yes or No, tick whichever answer you think is more true at the moment. Yes No Section Weight I’m tired all the time EL 39.20 I have pain at night P 12.91 Things are getting me down ER 10.47 Yes No I have unbearable pain P 19.74 I take tablets to help me sleep S 22.37 I’ve forgotten what it’s like to enjoy myself ER 9.31 Yes No I’m feeling on edge ER 7.22 I find it painful to change position P 9.99 I feel lonely SI 22.01 Yes No I can only walk about indoors PA 11.54 I find it hard to bend PA 10.57 Everything is an effort EL 36.80 Yes No I’m waking up in the early hours of the morning S 12.57 I’m unable to walk at all PA 21.30 I’m finding it hard to make contact with people SI 19.36 Remember if you are not sure whether to answer “Yes” or “No” to a problem, tick whichever answer you think more true at the moment. Yes No The days seem to drag ER 7.08 I have trouble getting up and down stairs and steps PA 10.79 I find it hard to reach for things PA 9.30 Yes No I’m in pain when I walk P 11.22 I lose my temper easily these days ER 9.76 I feel there is nobody I am close to SI 20.13 Yes No I like awake for most of the night S 27.26 I feel as if I’m losing control ER 13.99 I’m in pain when I’m standing P 8.96 (continued) 641 Exhibit 10.28 (continued) Yes No Section Weight I find it hard to dress myself PA 12.61 I soon run out of energy EL 24.00 I find it hard to stand for long (e.g., at the kitchen sink, waiting for a bus) PA 11.20 Yes No I’m in constant pain P 20.86 It takes me a long time to get to sleep S 16.10 I feel I am a burden to people SI 22.53 Yes No Worry is keeping me awake at night ER 13.95 I feel that life is not worth living ER 16.21 I sleep badly at night S 21.70 Yes No I’m finding it hard to get on with people SI 15.97 I need help to walk about outside (e.g., a walking aid or someone to support me) PA 12.69 Yes No I’m in pain when going up and down stairs or steps P 5.83 I wake up feeling depressed ER 12.01 I’m in pain when I’m sitting P 10.49 Part II Now we would like you to think about the activities in your life which may be affected by health problems. In the list below, tick Yes for each activity in your life which is being affected by your state of health. Tick No for each activity which is not being affected, or which does not apply to you. Is your present state of health causing problems with your. . . Yes No Job or work? (That is, paid employment) Looking after the home? (Examples: cleaning and cooking, repairs, odd jobs around the home, etc.) Social life? (Examples: going out, seeing friends, going to the pub, etc.) Home life? (That is: relationships with other people in your home) Sex life? Interests and hobbies? (Examples: sports, arts and crafts, do-it-yourself, etc.) Holidays? (Examples: summer or winter holidays, weekends away, etc.) Now please go back to page 1 and make sure that you have answered “Yes” or “No” to every question, in all pages of this questionnaire. Thank you for your help. Adapted from an original provided by Dr. SP McKenna. Copyright SM Hunt, J McEwen, SP McKenna. With permission. 642  General Health Status and Quality of Life 643 Reliability scales from the SF-36, and the EuroQol EQ-5D For the original version, correlations of items items identified two dimensions, physical and with section scores were examined in two stud- mental (19, Table 4). ies (2; 5). The four-week test-retest reliability of Part I concurrent validity. The correlation between of the revised NHP was reported by Hunt et al. scores on the first version and the McGill Pain for 58 patients with arthritis and 93 others with Questionnaire was 0.74, with a range from 0.50 peripheral vascular disease; coefficients for the for the social activities section to 0.78 for the six sections ranged from 0.75 to 0.88 (1, Table pain section. The overall score correlated 0.65 2); (13). Test-retest reliability for the seven items with a physiotherapist’s disability rating; corre- in Part II ranged from 0.44 to 0.86 in one sam- lations for each item ranged from 0.16 to 0.47 ple and from 0.55 to 0.89 in the other (1, Table (2, Table 3). Equivalent data were collected in 3). In a study of patients with amputated limbs, the rehabilitation study, correlations being retest agreement ranged from intraclass correla- drawn between the questionnaire and a set of tion (ICC) 0.83 (for pain and emotional scores) ratings made by a physician (5). Kendall correla- to 0.81 (mobility), 0.75 (sleep, energy), and 0.64 tions for the walking questions ranged from (social isolation) (14). 0.47 to 0.56, whereas those for dressing and Spearman correlations among domain scores self-care ranged between 0.40 and 0.48; coeffi- ranged from 0.32 (sleep and social isolation) to cients for body movements ranged from 0.44 to 0.70 (pain and physical mobility) in a sample of 0.55 (5, Table 4). The emotional questions cor- patients with cardiac disease, most ranged from related 0.71 (Spearman rho) with scores on the 0.41 to 0.58 (15, Table 4). Equivalent analyses General Health Questionnaire (GHQ) (20, using a sample of patients with chronic venous p167). Using the GHQ as a criterion, the sensi- leg ulceration were somewhat lower, ranging tivity of the NHP for detecting depression was from 0.23 to 0.60 (16, p696). Internal consis- examined. The best result was obtained by cal- tency estimates for the section scores ranged culating a summary score for the entire NHP less from 0.62 to 0.82, with an overall alpha coeffi- the physical mobility section; sensitivity was cient of 0.72 (17). Similar results have been re- 91% and specificity 82% (20, p167). Although ported (range, 0.63–0.81) (16, Table 2). the emotional questions alone achieved high sen- Reliability measured by intraclass correlation sitivity, specificity was greatly increased by in- was 0.95, with an effect size of 0.52 in people cluding responses to the other four sections. A with musculoskeletal disorders (18). The high separate study found a correlation of 0.76 be- reliability balanced by modest effect size was at- tween the 12-item GHQ scores and the emo- tributed in part to the fact that many respon- tional reactions scale of the NHP (8, p49). dents achieved the maximum score; the ceiling Several studies have compared the NHP and effect on the NHP is discussed later in this sec- the SF-36. A head-to-head comparison was made tion. for a sample of patients with chronic obstructive pulmonary disease (21). The average convergent Validity validity correlation for equivalent sub scales was In the initial development, considerable attention 0.63; criterion correlations were slightly higher was paid to content validity. Items were based for the SF-36 scales, although the areas under on patients’ descriptions of their experience; for- the ROC curves were comparable. There was a mal tests of linguistic clarity were applied in an greater ceiling effect for the NHP and alpha in- effort to simplify the item wording. ternal consistency results for each section were A factor analysis of the 33 items in the origi- consistently higher for the SF-36. Rasch analyses nal version identified eight factors, reflecting showed similar severity ranges for the two pain, mobility, body movement, sleep, anxiety, scales, although the NHP had more even item getting out of the house, loneliness, and depres- spacing. In a study by Essink-Bot et al., conver- sion (2). A factor analysis of the NHP scales, the gent validity correlations with the scales of the 644 Measuring Health SF-36 were lower: 0.47 for Energy (NHP) and sky scale were almost identical (15, Table 4). Vitality (SF-36), 0.43 for the two pain scales, Hunt et al. summarized studies that applied the 0.46 for the emotional scales, 0.32 for the social NHP to nine different patient groups; the results scales, and 0.67 for the mobility (NHP) and showed clinically plausible contrasts in scores physical functioning (SF-36) scales (19, p537). (1). Essink-Bot et al. compared migraine patients Correlations were higher in a study of patients with matched controls and reported areas under with disease involving chronic airway limitation; the ROC curve ranging from only 0.52 to 0.59, convergent correlations ranged from 0.54 to with a mean of 0.55 (17). All six sections on the 0.80 for male patients but were lower for fe- NHP showed significant differences in patients males (22, Tables 2 and 3). In a study of revascu- before and after treatment for leg ulcers (16, larization surgery for patients with lower limb Table 5). Likewise, all six sections showed ischemia, the NHP discriminated better than did strong contrasts before and after heart surgery the SF-36 in terms of pain and mobility and was involving organ transplant procedures (11, more responsive to change (23). Both scales Table 4). By contrast, Hunt et al. demonstrated were reliable, and the SF-36 showed less skewed no contrast in the status of a group of patients a scores. However, in a study of revascularization week before surgery and 6 to 8 weeks after (7, for intermittent claudication, the SF-36 was Table VII). Essink-Bot et al. (19) compared the found somewhat more responsive, although the ability of scores on four measures to discrimi- NHP pain scale was again superior and there nate between people absent from work due to ill- was good general agreement between the instru- ness and others. The SF-36 was the most ments (24). discriminating (mean area under the ROC curve [AUC], 0.72), followed by the COOP Charts discriminant validity. The NHP has repeatedly (mean AUC, 0.64), the EuroQol EQ-5D (mean been shown to discriminate between different AUC, 0.61), and the NHP (mean AUC, 0.60) types of patient and between patients before and (19, Table 5). The NHP has been evaluated in after treatment. The original Nottingham Health patients with dementia, 94% of whom could Index showed marked differences between complete the forms with some guidance. Agree- arthritic patients before and after hip replace- ment with forms completed by the patients’ rela- ment surgery (2). The revised NHP distinguished tives was, however, low (27). For patients with between hip replacement patients and their rheumatoid arthritis, NHP scores have been spouses who served as a comparison group (25). shown to correlate highly with disease activity as Correlations with the Harris Hip Score are as measured by the Disease Activity Score, but less high as −0.82 (26, p603). The NHP showed con- highly with the number of swollen joints or trasts between rehabilitation patients with phys- morning stiffness (28). ical and mental handicaps (5, Table 5) and between stroke patients and healthy controls sensitivity to change. Formal calculations of (20, Table 1). All six sections of the NHP effect size have begun to appear in the literature. showed significant differences (p< 0.001) be- In the study of patients with chronic leg ulcers, tween four groups of elderly people, ranging effect sizes were reported for each NHP section from those with chronic illness and physical, so- compared with size of the ulcer (effect sizes cial, and emotional disabilities, to physically fit ranged from 0.19 for sleep to 0.35 for mobility), people (4). In a study of patients with cardiac and mobility (range, 0.2 for emotion to 1.3 for disease, NHP domain scores discriminated sig- physical mobility) (16, Table 4). Standardized nificantly between patients in different New response means (SRM) after 12 weeks of treat- York Heart Association (NYHA) classes, and ac- ment ranged from 0.68 (pain) to 0.14 (social iso- cording to Karnofsky scores (15, Tables 2, 3). lation) (Table 7). In a study of migraine, the Spearman correlations with the NYHA classifi- NHP showed small-to-medium effect sizes, but cation ranged from 0.30 (social isolation) to somewhat larger than those for the EuroQol 0.52 (energy level); correlations with the Karnof- EQ-5D (29, Tables 3 to 5). Likewise, in a study  General Health Status and Quality of Life 645 of asthma, SRMs for the section scores on the across sections, 0.58–0.86), and test-retest relia- NHP were slightly lower than those for the SF- bilities ranging from 0.69 to 0.85 for the six sec- 36, but higher than those for the EQ-5D Index tions (45; 46). An abbreviated version of the (30, Table 5). Sensitivity to change following Spanish translation has been derived using Rasch myocardial infarction was lower than that analysis; it contains 11 physical and 11 psycho- achieved by the SIP and by the McMaster Health logical items that can be scored as separate Index Questionnaire (31, Tables 2 to 4). Congle- scales, or used together as an overall score (47). ton et al. reported limited sensitivity to change Translations that have also developed their in a sample of cystic fibrosis patients (32). own weighting schemes include Swedish (48) There is some indication, acknowledged by and French (49). Correlations between the origi- Hunt, that the NHP has a ceiling effect, whereby nal and the French weights exceeded 0.9 for it is not sensitive to minor levels of disability and most sections, although the correlation for pain cannot distinguish between levels of good was only 0.76 and that for emotional reactions health. A comparison with the Dartmouth was 0.58 (49). The Swedish version of the NHP COOP Charts, for example, showed that the has been extensively used, and considerable in- COOP Charts were more sensitive to minor de- formation is available on its reliability and valid- viations from complete well-being (33; 34, ity (48; 50–53). Rank correlations of item p305). Similarly, in a sample of healthy people, weights between the Swedish version and the scores on the SF-36 were less skewed than those original weights are in the high 0.80s: valuations on the NHP, such that the SF-36 identified mi- of health states seem similar in both cultures nor levels of discomfort missed by the Notting- (54, p373). Test-retest reliability at 4 weeks was ham scale (35). Conversely, in another review, 0.92; internal consistency for the sections ranged the SF-36 showed more floor effects than the from 0.34 (social isolation) to 0.81 (emotional NHP did (i.e., the NHP was more sensitive to reactions) (54, p373). variations at the negative end of the health con- Based on their experience in producing so tinuum) (34, pp305–306). Ceiling effects on the many versions, Hunt and her group have pro- NHP have also occurred in studies of patient posed guidelines for the development of cultur- groups, such as patients with leg ulcers for ally equivalent versions of a measurement (39; whom between 25 to 68% achieved scores on 55). the various sections that indicated the highest levels of health (16, Table 2). Reference Standards Reference scores for the NHP are available for Alternative Forms healthy people (by age-group, sex, and social Translations are listed on the Quality of Life In- class) and for various categories of patients (1, struments Database (QOLID) Web site Table 1; 56; 57). (http://www.qolid.org/public/NHP.html). Trans- lations include most European languages Commentary (36–38), Arabic (39), and Urdu (40). A Danish Although now dated, the NHP remains one of version showed an overall test-retest reliability the more frequently used measures, especially in of 0.93, with reliabilities for the sections ranging Europe. Originally intended for use in primary from 0.76 to 0.86 (41). Alpha internal consis- care and in and epidemiological studies, it has tency coefficients for a Dutch version ranged been applied to many groups of people in med- from 0.55 (physical mobility) to 0.81 (pain); ical and nonmedical settings: to miners and fire- four of six coefficients exceeded 0.70 (42, Table men in good physical health (58), to elderly 9.2). The Greek version had retest reliability co- people and patients frequently going to see their efficients ranging from 0.77 (pain) to 0.86 general practitioners (59), and in a study of so- (sleep, social isolation, mobility) (43, Table 5). cial class differences in perceived health (1). The Spanish versions have been tested (44); one strengths of the method include its simplicity showed an internal consistency of 0.91 (range and its broad coverage. The NHP, for example, 646 Measuring Health elicits fewer missing values than rival generic signment of items to separate sections has been scales (19). It was not intended for use with hos- criticized as arbitrary, and the correlations pitalized patients, nor as an outcome for clinical among items in some categories are lower than trials (36). those between categories (60, Tables 5, 6). Sec- The NHP has, however, encountered spirited ond, there have been criticisms of the weighting criticism; McEwen and McKenna summarize system used (54, p371). For example, Jenkinson some of the comments (36). The first issue de- noted that a person who cannot walk will re- rives from confusion over the applications for ceive a less severe weighted score than one who which it is best suited. Some early articles sug- has difficulty with stairs and can only walk gested that the NHP would be suitable for use in around indoors (10). Another example concerns health surveys, but subsequent use indicated a scores for those who do not work at all, com- clear ceiling effect whereby healthy respondents pared with others who do work, but with limita- or those with minor ailments tend to show per- tions. The criticism has also been made that fect scores and hence have no scope for improve- items on several of the NHP sections do not ment; comparison with other instruments shows show a wide range of scores (10). that such scores do not necessarily indicate an ab- Certain of the norms quoted by Hunt et al. sence of problems (1; 33). Community surveys do not form smooth trends across the groups typically find that around two thirds of the popu- sorted by age, sex, or social class; either the pop- lation record no problems on the NHP (19); a ulations used were not homogeneous, or the lower figure of 46% was found in one British sur- samples were simply too small to derive stable vey (60). Sleep and energy level form the mildest estimates. The norms may need to be smoothed degree of disorder and therefore the most com- before using them as reference standards. mon problems (61); the range of disability cov- The NHP continues to be used and tested and ered by each dimension is uneven. Hunt and still holds considerable utility as a clinical instru- McKenna subsequently suggested that the NHP ment. Until the advent of the SF-36, it was one should not be used as a survey instrument (62). of the most popular instruments in Europe. Sev- The ceiling effect results from its original design eral studies that have compared the instruments as an instrument for people seeking care, deliber- show them to be similar, with perhaps a slight ately focusing on health problems rather than edge for the SF-36. Time will tell how the turf is positive well-being; only later was the NHP used divided between them. in healthy populations. A balanced view would hold that the NHP is suitable as a survey tool only in populations, such as the elderly, in which References there are likely to be people with significant dis- (1) Hunt SM, McEwen J, McKenna SP. Mea- ability (8, p3; 63). Kind and Carr-Hill suggest suring health status: a new tool for that if the NHP is used as a screen for disability, clinicians and epidemiologists. J R Coll it is unnecessarily long and contains considerable Gen Pract 1985;35:185–188. redundancy (60, p909). Readers should consider (2) McDowell IW, Martini CJM, Waugh W. A alternative scales carefully before using the NHP method for self-assessment of disability as a survey instrument in general populations. before and after hip replacement Debate also surrounds the scoring. First, operations. Br Med J 1978;2:857–859. there is discussion over the choice between a to- (3) Hunt SM, McKenna SP, McEwen J, et al. The Nottingham Health Profile: subjective tal score and scores for each section. Hunt et al. health status and medical consultations. argued that an overall score was inappropriate, Soc Sci Med 1981;15:221–229. because the large number of possible routes to (4) Hunt SM, McKenna SP, McEwen J, et al. an intermediate score would not provide an in- A quantitative approach to perceived terpretable picture of a person’s disability (1). health status: a validation study. J However, if section scores are used, then the sec- Epidemiol Community Health tions should measure different things. The as- 1980;34:281–286.  General Health Status and Quality of Life 647 (5) Martini CJ, McDowell I. Health status: (17) Juva K, Verkkoniemi A, Viramo P, et al. patient and physician judgments. Health APOE epsilon4 does not predict mortality, Serv Res 1976;11:508–515. cognitive decline, or dementia in the oldest (6) Hunt SM, McEwen J. The development of old. Neurology 2000;54:412–415. a subjective health indicator. Sociol Health (18) Beaton DE, Bombardier C, Hogg-Johnson Illn 1980;2:231–246. S. Choose your tool: a comparison of the (7) Hunt SM, McEwen J, McKenna SP, et al. psychometric properties of five generic Subjective health assessments and the health status instruments in workers with perceived outcome of minor surgery. J soft tissue injuries. Qual Life Res Psychosom Res 1984;28:105–114. 1994;3:50–56. (8) Hunt SM, McKenna SP, McEwen J. The (19) Essink-Bot M-L, Krabbe PFM, Bonsel GJ, Nottingham Health Profile: user’s manual. et al. An empirical comparison of four Rev ed. [Manuscript] Manchester, England: generic health status measures. The 1989. Nottingham Health Profile, the Medical (9) McKenna SP, Hunt SM, McEwen J. Outcomes Study 36-item Short-Form Weighting the seriousness of perceived Health Survey, the COOOP/WONCA health problems using Thurstone’s method charts, and the EuroQol instrument. Med of paired comparisons. Int J Epidemiol Care 1997;35:522–537. 1981;10:93–97. (20) Ebrahim S, Barer D, Nouri F. Use of the (10) Jenkinson C. Why are we weighting? A Nottingham Health Profile with patients critical examination of the use of item after a stroke. J Epidemiol Community weights in a health status measure. Soc Sci Health 1986;40:166–169. Med 1991;32:1413–1416. (21) Prieto L, Alonso J, Ferrer M, et al. Are (11) O’Brien BJ, Buxton MJ, Ferguson BA. results of the SF-36 health survey and the Measuring the effectiveness of heart Nottingham Health Profile similar? A transplant programmes: quality of life data comparison in COPD patients. J Clin and their relationship to survival analysis. J Epidemiol 1997;50:463–473. Chronic Dis 1987;40:S137–S153. (22) Crockett AJ, Cranston JM, Moss JR, et al. (12) Prieto L, Alonso J. Exploring health The MOS SF-36 health survey preferences in sociodemographic and questionnaire in severe chronic airflow health related groups through the paired limitation: comparison with the comparison of the items of the Nottingham Nottingham Health Profile. Qual Life Res Health Profile. J Epidemiol Community 1996;5:330–338. Health 2000;54:537–543. (23) Klevsgard R, Froberg BL, Risberg B, et al. (13) Hunt SM, McKenna SP, Williams J. Nottingham Health Profile and Short-Form Reliability of a population survey tool for 36 Health Survey questionnaires in patients measuring perceived health problems: a study with chronic lower limb ischemia: before of patients with osteoarthrosis. J Epidemiol and after revascularization. J Vasc Surg Community Health 1981;35:297–300. 2002;36:310–317. (14) Demet K, Guillemin F, Martinet M, et al. (24) Wann-Hansson C, Rahm Hallberg I, Nottingham Health Profile: reliability in a Risberg B, et al. A comparison of the sample of 542 subjects with major Nottingham Health Profile and Short Form amputation of one or several limbs. 36 Health Survey in patients with chronic Prosthet Orthot Int 2002;26:120–123. lower limb ischaemia in a longitudinal (15) O’Brien BJ, Buxton MJ, Patterson DL. perspective. Health Qual Life Outcomes Relationship between functional status and 2005;2:9. health-related quality-of-life after myocar- (25) McKenna SP, McEwen J, Hunt SM, et al. dial infarction. Med Care Changes in the perceived health of patients 1993;31:950–955. recovering from fractures. Public Health (16) Franks PJ, Moffatt CJ. Health related 1984;98:97–102. quality of life in patients with venous (26) Garellick G, Malchau H, Herberts P. ulceration: use of the Nottingham Health Specific or general health outcome mea- Profile. Qual Life Res 2001;10:693–700. sures in the evaluation of total hip 648 Measuring Health replacement. A comparison between the developing health-related quality of life Harris Hip Score and the Nottingham instruments for international clinical Health Profile. J Bone Joint Surg 1998;80- studies and outcomes research. B:600–606. Pharmacoeconomics 1996;10:336–355. (27) Bureau-Chalot F, Novella J-L, Jolly D, et (38) European guide to the Nottingham Health al. Feasibility, acceptability, and internal Profile. Montpellier: The European Group consistency reliability of the Nottingham for the Quality of Life and Health Mea- Health Profile in dementia patients. surement, 1992. Gerontology 2002;48:220–225. (39) Hunt SM. Cross-cultural issues in the use (28) Houssien DA, McKenna SP, Scott DL. The of socio-medical indicators. Health Policy Nottingham Health Profile as a measure of 1986;6:149–158. disease activity and outcome in rheumatoid (40) Ahmad WIU, Kernohan EEM, Baker MR. arthritis. Br J Rheumatol 1997;36:69–73. Cross-cultural use of socio-medical (29) Essink-Bot M-L, van Royen L, Krabbe P, et indicators with British Asians. Health al. The impact of migraine on health Policy 1989;13:95–102. status. Headache 1995;35:200–206. (41) Thorsen H, McKenna SP, Gottschalck L. (30) Oga T, Nishimura K, Tsukino M, et al. A The Danish version of the Nottingham comparison of the responsiveness of Health Profile: its adaptation and different generic health status measures in reliability. Scand J Prim Health Care patients with asthma. Qual Life Res 1993;11:124–129. 2003;12:555–563. (42) Meyboom-de Jong B, Smith RJA. Studies (31) Taylor R, Kirby B, Burdon D, et al. The with the Dartmouth COOP Charts in assessment of recovery in patients after general practice: comparison with the myocardial infarction using three generic Nottingham Health Profile and the General quality-of-life measures. J Cardiopulm Health Questionnaire. In: WONCA Rehabil 1998;18:139–144. Classification Committee, ed. Functional (32) Congleton J, Hodson ME, Duncan-Skingle status measurement in primary care. New F. Do Nottingham Health Profile scores York: Springer-Verlag, 1990:132–149. change over time in cystic fibrosis? Respir (43) Vidalis A, Syngelakis M, Papathanasiou Med 1998;92:268–272. M, et al. The Greek version of the (33) Coates AK, Wilkin D. Comparing the Nottingham Health Profile: features of its Nottingham Health Profile with the adaptation. Hippokratia 2002;6(suppl Dartmouth COOP Charts. In: Scholten 1):79–82. JHG, ed. Functional status assessment in (44) Alonso J, Prieto L, Anto JM. The Spanish family practice. Lelystad, The Netherlands: version of the Nottingham Health Profile: a Meditekst, 1992:81–86. review of adaptation and instrument (34) McHorney CA, Tarlov AR. Individual- characteristics. Qual Life Res patient monitoring in clinical practice: are 1994;3:385–393. available health status surveys adequate? (45) Alonso J, Anto JM, Moreno C. Spanish Qual Life Res 1995;4:293–307. version of the Nottingham Health Profile: (35) Brazier JE, Harper R, Jones NMB, et al. translation and preliminary validity. Am J Validating the SF-36 Health Survey Public Health 1990;80:704–708. questionnaire: new outcome measure for (46) Badia X, Alonso J, Brosa M, et al. primary care. Br Med J Reliability of the Spanish version of the 1992;305:160–164. Nottingham Health Profile in patients with (36) McEwen J, McKenna SP. Nottingham stable end-stage renal disease. Soc Sci Med Health Profile. In: Spilker B, ed. Quality of 1994;38:153–158. life and pharmacoeconomics in clinical (47) Prieto L, Alonso J, Lamarca R, et al. Rasch trials. Philadelphia: Lippincott-Raven, measurement for reducing the items of the 1996:281–286. Nottingham Health Profile. J Outcome (37) Anderson RT, Aaronson NK, Bullinger M. Meas 1998;2:285–301. A review of the progress towards (48) Hunt SM, Wiklund I. Cross-cultural  General Health Status and Quality of Life 649 variation in the weighting of health (61) Kind P, Gudex CM. Measuring health statements: a comparison of English and status in the community: a comparison of Swedish valuations. Health Policy methods. J Epidemiol Community Health 1987;8:227–235. 1994;48:86–91. (49) Bucquet D, Condon S, Ritchie K. The (62) Hunt SM, McKenna SP. Validating the SF- French version of the Nottingham Health 36. Br Med J 1992;305:645. Profile. A comparison of items weights (63) Hunt SM. Nottingham Health Profile. In: with those of the source version. Soc Sci Wenger NK, Mattson ME, Furberg CD, Med 1990;30:829–835. eds. Assessment of quality of life in clinical (50) Wiklund I, Romanus B, Hunt SM. Self- trials of cardiovascular therapies. New assessed disability in patients with arthrosis York: Le Jacq, 1984:165–169. of the hip joint. Reliability of the Swedish version of the Nottingham Health Profile. The Short-Form-36 Health Survey Int Disabil Stud 1988;10:159–163. (51) Wiklund I. The Nottingham Health (RAND Corporation and John Profile—a measure of health-related quality E. Ware Jr., 1990, revised 1996) of life. Scand J Prim Health Care Purpose 1990;8(suppl 1):15–18. The 36-item short form of the Medical Out- (52) Wiklund I, Welin C. A comparison of different psychosocial questionnaires in comes Study questionnaire (SF-36) was designed patients with myocardial infarction. Scand as a generic indicator of health status for use in J Rehabil Med 1992;24:195–202. population surveys and evaluative studies of (53) Wiklund I, Karlberg J. Evaluation of health policy. It can also be used in conjunction quality of life in clinical trials. Control with disease-specific measures as an outcome Clin Trials 1991;12:204S–216S. measure in clinical practice and research (1). (54) Anderson RT, Aaronson NK, Wilkin D. Critical review of the international Conceptual Basis assessments of health-related quality of life. The SF-36 derived from the work of the RAND Qual Life Res 1993;2:369–395. Corporation of Santa Monica during the 1970s. (55) Hunt SM, Alonso J, Bucquet D, et al. RAND’s Health Insurance Experiment com- Cross-cultural adaptation of health mea- pared the impact of alternative health insurance sures. Health Policy 1991;19:33–44. systems on health status and utilization (2; 3, (56) Hunt SM, McEwen J, McKenna SP. Perceived health: age and sex comparisons p2:3). The outcome measures developed for the in the community. J Epidemiol Community study have been widely used and several are de- Health 1984;38:156. scribed in this book. They were subsequently re- (57) Hunt SM, McKenna SP. The Nottingham fined and used in RAND’s Medical Outcomes Health Profile user’s manual, revised. Study (MOS), which focused more narrowly on Manchester: Galen Research and care for chronic medical and psychiatric condi- Consultancy, 1991. tions (4; 5). The MOS surveys were comprehen- (58) McKenna SP, Hunt SM, McEwen J. sive, covering 40 physical and mental health Absence from work and perceived health concepts, and several abbreviated forms were among mine rescue workers. J Soc Occup produced. An 18-item scale was produced in Med 1981;31:151–157. 1984, followed by the 20-item short-form (SF- (59) McKenna SP, Hunt SM, McEwen J, et al. 20) in 1986. The SF-36 was constructed to an- Looking at health from the consumers’ point of view. Occup Health (Lond) swer criticisms of limitations in the SF-20; it 1980;32:350–355. covered the eight most important of the original (60) Kind P, Carr-Hill R. The Nottingham 40 concepts (3, p2:3). The SF-20 was reviewed Health Profile: a useful tool for in the 1996 edition of Measuring Health, but it epidemiologists? Soc Sci Med has largely been replaced by a more recent 1987;25:905–910. abridgement, the SF-12, which is reviewed sepa- 650 Measuring Health rately in this edition. Note that the physical bourne give an extended description of the origins functioning items in the SF-36 are also con- of the SF-36 and of its links with the other MOS tained in the MOS Physical Functioning Mea- instruments (1). The SF-36 includes multi-item sure, reviewed in Chapter 3; the mental health scales to measure the following eight dimensions questions are derived from the Mental Health (the question numbers refer to those in Exhibit Inventory that is described in Chapter 5. Fuller 10.29): details of the origins of the items in the SF-36 are given in the SF-36 manual, which shows the derivation of each item (3, Table 3.4). The devel- PF: Physical functioning (ten items in ques- opment of the SF-36 is now coordinated by tion 3) QualityMetric, Incorporated, in Rhode Island RP: Role limitations due to physical health (see Address section), although RAND contin- problems (four items in question 4) ues to provide information on its version of the BP: Bodily pain (questions 7 and 8) SF-36, designated the “RAND 36-item Health SF: Social functioning (questions 6 and 10) Survey 1.0” (www.rand.org/health/surveys/sf36 MH: General mental health, covering psy- item/). chological distress and well-being (five As a generic instrument, the SF-36 was de- items: questions 9b, 9c, 9d, 9f, and 9h) signed to be applicable to a wide range of types RE: Role limitations due to emotional prob- and severities of health conditions. Generic in- lems (questions 5a, 5b, and 5c) struments are useful for monitoring patients VT: Vitality, energy or fatigue (four items: with multiple conditions, for comparing the questions 9a, 9e, 9g, and 9i) health status of patients with different condi- GH: General health perceptions (five items: tions, and for comparing patients with the gen- questions 1 and 11a to 11d). eral population (4, p912). Ware et al. argued that a generic measure should cover both physi- cal and mental concepts and should measure In addition, the second question covers change each concept in several contrasting ways. These in health status over the past year; this is not include behavioral functioning, perceived well- counted in scoring the eight dimensions but is being, social and role disability, and personal used to estimate change in health from a cross- evaluations of health in general (6, p3:2). Mea- sectional administration of the SF-36 (6, p9:15). sures of behavioral functioning and role limita- Version 1 of the SF-36 was described in the tions include questions on work, self-care, 1996 edition of this book. To address criticism mobility, among other topics. Perceived well- of the layout and wording of some items, Ver- being is subjective and cannot be completely in- sion 2 changed the dicohotomous answers for ferred from behavior; hence, the SF-36 includes the role questions to five-point scales, and questions on feeling states. The questions on slightly altered the wording of several items (7, overall evaluation of health provide a summary Figure 1). Modifications to the SF-36 wording indicator and capture the impact of health prob- over time are indicated in the manual (6, Table lems not directly covered by the other questions 3.4). Exhibit 10.29 shows Version 2. (6, p3:3). The standard instrument uses a four-week recall period, but an acute version uses a one- Description week recall and is suitable for use when the The items in the SF-36 were drawn from the measure is administered repeatedly on a weekly original 245-item MOS questionnaire; nine basis. The SF-36 may be self-administered or items from the SF-20 were retained in the SF-36 used in personal or telephone interviews; whereas a further five were reworded, as were machine-readable forms and instruction sheets several of the answer categories. Several of the for each version are available from Quality- mental health items originated from the General Metric. McHorney et al. compared mail and Well-Being Schedule of Dupuy. Ware and Sher- telephone survey administration; mail was sig- Exhibit 10.29 The Short-Form-36 Health Survey, Version 2 Your Health and Well-Being This survey asks for your views about your health. This information will help keep track of how you feel and how well you are able to do your usual activities. Thank you for completing this survey! For each of the following questions, please mark an X in the one box that best describes your answer. 1. In general, would you say your health is: Excellent Very Good Good Fair Poor ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 2. Compared to one year ago, how would you rate your health in general now? Much better Somewhat better About the Somewhat worse Much worse now than one now than one same as one now than one now than one year ago year ago year ago year ago year ago ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 3. The following items are about activities you might do during a typical day. Does your health now limit you in these activities? If so, how much? Yes, Yes, No, Not Limited Limited A Limited A Lot Little At All ▼ ▼ ▼ a. Vigorous activities, such as running, lifting 1 2 3 heavy objects, participating in strenuous sports b. Moderate activities, such as moving a table, 1 2 3 pushing a vacuum cleaner, bowling, or playing golf c. Lifting or carrying groceries 1 2 3 d. Climbing several flights of stairs 1 2 3 e. Climbing one flight of stairs 1 2 3 f. Bending, kneeling, or stooping 1 2 3 g. Walking more than a mile 1 2 3 h. Walking several blocks 1 2 3 i. Walking one block 1 2 3 j. Bathing or dressing yourself 1 2 3 4. During the past 4 weeks, have you had any of the following problems with your work or other regular daily activities as a result of your physical health? Yes No ▼ ▼ a. Cut down on the amount of time you spent on work or other 1 2 activities b. Accomplished less than you would like 1 2 c. Were limited in the kind of work or other activities 1 2 d. Had difficulty performing the work or other activities 1 2 (for example, it took extra effort) (continued) 651 Exhibit 10.29 (continued) 5. During the past 4 weeks, have you had any of the following problems with your work or other regular activities as a result of any emotional problems (such as feeling depressed or anxious)? Yes No ▼ ▼ a. Cut down on the amount of time you spent on work or other 1 2 activities b. Accomplished less than you would like 1 2 c. Did work or other activities less carefully than usual 1 2 6. During the past 4 weeks, to what extent has your physical health or emotional problems interfered with your normal social activities with family, friends, neighbors, or groups? Not at all A little bit Moderately Quite a bit Extremely ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 7. How much bodily pain have you had during the past 4 weeks? None Very mild Mild Moderate Severe Very Severe ▼ ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 6 8. During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)? Not at all A little bit Moderately Quite a bit Extremely ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 9. These questions are about how you feel and how things have been with you during the past 4 weeks. For each question, please give the one answer that comes closest to the way you have been feeling. How much of the time during the past 4 weeks . . . All A Good A Little of the Most of Bit of the Some of of the None of Time the Time Time the Time Time the Time ▼ ▼ ▼ ▼ ▼ ▼ a. Did you feel full of pep? 1 2 3 4 5 6 b. Have you been a very 1 2 3 4 5 6 nervous person? c. Have you felt so down 1 2 3 4 5 6 in the dumps that nothing could cheer you up? d. Have you felt calm and 1 2 3 4 5 6 peaceful? e. Did you have a lot of energy? 1 2 3 4 5 6 f. Have you felt downhearted 1 2 3 4 5 6 and blue? g. Did you feel worn out? 1 2 3 4 5 6 h. Have you been a happy 1 2 3 4 5 6 person? i. Did you feel tired? 1 2 3 4 5 6 652  General Health Status and Quality of Life 653 Exhibit 10.29 10. During the past 4 weeks, how much of the time has your physical health or emotional problems interfered with your social activities (like visiting friends, relatives, etc.)? All of the Most of the Some of the A little of the None of the time time time time time ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 11. How TRUE or FALSE is each of the following statements for you? Definitely Mostly Don’t Mostly Definitely True True Know False False ▼ ▼ ▼ ▼ ▼ a. I seem to get sick a little easier 1 2 3 4 5 than other people b. I am as healthy as anybody I 1 2 3 4 5 know c. I expect my health to get worse 1 2 3 4 5 d. My health is excellent 1 2 3 4 5 Thank you for completing these questions! From Ware JE Jr, Kosinski M, Gandek B. SF-36 Health Survey: manual and interpretation guide. Lincoln, RI: QualityMetric Inc., 2000: B6-B11. With permission. nificantly cheaper, and provided a higher re- scales (3, Chapter 6), and a norm-based ap- sponse rate; mail also identified a higher level of proach that adjusts these raw scores to have a disability (8). The questions generally take five mean of 50 and a standard deviation of 10 (14, to ten minutes to complete; elderly respondents Chapter 5). may require up to 15 minutes (9; 10). Self- The first step is to check for out-of-range val- administration appears acceptable and feasible ues, and then to orient all item scores so that for most patients, although the optical scan high scores correspond to better health. The forms may be difficult for people with vision codes shown in the exhibit are replaced for sev- problems (11). A version for computerized ad- eral items. In the first approach, values for items ministration appears acceptable to respondents 1, 7, and 8 are recoded, using weights derived (12). Nonresponse rates averaged 3.9% across from Likert analyses. For item 1, excellent is the 36 items in a large study of people with scored 5.0, very good = 4.4, good = 3.4, chronic conditions (13, p46). Detailed adminis- fair = 2.0, and poor = 1.0. For item 7, none = 6.0, tration instructions are contained in Chapter 4 very mild = 5.4, mild = 4.2, moderate = 3.1, se- of the manual. vere = 2.2, and very severe = 1.0. Scores for item 8 take account of the answers to item 7: if no scoring. Two sets of scores are derived from pain is recorded on either item, then item 8 is the SF-36: a profile of eight section scores, and scored 6. If item 8 is answered not at all, but two summary scores, one for the physical com- item 7 > none, then item 8 is scored 5. For the re- ponent (PCS) and one for the mental component maining categories of item 8, a little bit = 4, (MCS) summary scores. For each set of scores, moderately = 3, quite a bit = 2, and extremely = 1. two alternative approaches may be used in cal- However, if item 7 was not answered, the values culating scores: a normal, additive approach for item 8 are as follows: not at all = 6.0, a little that produces zero-to-100 scores for the eight bit = 4.75, moderately = 3.3, quite a bit = 2.25, 654 Measuring Health and extremely = 1 (3, Table 6.3). Next, scores for ing the PCS and MCS scores by 10 and adding items on each of the eight scales (see above) are 50 to the product. added to give scale scores. Finally, these are lin- The influence of health economics and the de- early transformed to a 0-to-100 scale (6, Table mand for estimates of quality-adjusted survival 6.11; 13; 15). The formula is have led to the development of ways to trans- form SF-36 scores into a utility score. One ap- (actual score − lowest proach is to use regression analysis on a data set Transformed possible score) × 100 = containing both SF-36 scores and a utility mea- scale possible raw sure; this has been undertaken using the Health score range Utilities Index (HUI) (16), the Quality of Well- A missing value is given for a scale if over half its Being scale (17), and the EuroQol EQ-5D (18). items are missing; where fewer items are miss- Alternatively, selected permutations of response ing, these are replaced by that respondent’s categories from the SF-36 can be valued using a mean scores on the remaining items in the scale utility estimation procedure (e.g., standard gam- (13, p44). However, the scoring software from ble). Utility scores for other permutations not di- www.sf-36.com provides a more sophisticated rectly evaluated can be interpolated using adjustment for missing values (14, p41). regression methods. For example, a regression The zero-to-100 scales can also be trans- method for transforming SF-36 scores into formed into norm-based scores that translate equivalent scores on the Quality of Well-Being raw scores into a position on the population dis- scale has been described (17). tribution of scores and so requires population norms. Ideally, norms should be taken from the Reliability country in which the study is being undertaken. McHorney et al. based comprehensive analyses This simplifies interpretation, because the user of item response, reliability and validity on a no longer has to remember the (different) norms sample of 3,445 patients with chronic medical for each scale. It also has the advantage of ensur- or psychiatric conditions drawn from the MOS ing comparability across different versions of the study (13). Item analyses confirmed the assign- SF-36 (14, p26). First, the zero to 100 scores for ment of the items to the eight scales; this was the eight subscales are standardized using a z replicated in different patient groups (13, p51). score transformation, which involves subtract- In the International Quality of Life Assessment ing the population mean score for that scale project (IQOLA), studies in 11 countries from each respondent’s score, and dividing the showed that items generally correlated more difference by the population standard deviation. highly with their own scales than with others For example, the raw score for Physical Function (19, Table 5). is transformed by subtracting 83.29094 and di- Alpha internal consistency coefficients for the viding by 23.75883; the formulas are given in eight scales have been reported from many stud- the scoring manual (14, Table 6.12). Next, to ies; Ware et al. listed results from 14 (6, Table give a mean of 50 and standard deviation of 10, 7.2), and then from 11 countries under the the z score is multiplied by 10 and 50 is added to IQOLA project (19, Table 6). Combining results the product. from these studies, the median alpha reliability To produce the PCS physical summary score, for all scales exceeds 0.80, except for the two- the z scores for each of the eight scales is multi- item social functioning scale (0.76); in the study plied by a factor score coefficient and the result- of Essink- Bot et al., the mean alpha for the SF- ing scores summed over the eight subscales. The 36 scales was 0.84, compared with 0.72 for the same is done for the MCS score and formulas are Nottingham Health Profile (20, pp 528–529). given in the manual (14, p51). Finally, the PCS Typical results are illustrated in Table 10.6, in and MCS scores are translated into T-scores which the first line shows values from the study (with a mean of 50 and a SD of 10) by multiply- of over 163,000 elderly people. Slightly lower  Table 10.6 Cronbach Alpha Coefficients for SF-36 Scales from Various Studies General Physical Role, Social Mental Role, Health Function Physical Bodily Pain Functioning Health Emotional Vitality Perceptions (PF) (RP) (BP) (SF) (MH) (RE) (VT) (GH) Reference 0.93 0.91 0.88 0.85 0.83 0.88 0.87 0.83 (25, Table 5) 655 0.88 0.90 0.80 0.77 0.82 0.80 0.88 0.83 (26, Table 1) 0.93 0.84 0.82 0.85 0.90 0.83 0.87 0.78 (13, Table 7) 0.93 0.96 0.85 0.73 0.95 0.96 0.96 0.95 (22, Table 2) 0.90 0.88 0.82 0.76 0.83 0.80 0.85 — (27, Table 3) 0.93 0.89 0.90 0.68 0.84 0.82 0.86 0.81 (6) 0.92 0.89 0.86 0.80 0.86 0.86 0.86 0.83 (28, Table 1) 0.92 0.95 0.85 0.85 0.84 0.92 0.84 0.80 (7, Table 2) 0.91 0.86 0.81 0.56 0.80 0.83 0.84 0.66 (24, Table 1) 0.93 0.88 0.85 0.82 0.86 0.83 0.83 0.77 (29, Table 1) 656 Measuring Health values were reported by Kurtin et al. (11, Table a medical disorder and a psychological one (6, 5). All scales appear sufficiently reliable for com- pp9:21–9:23; 30, pp255–259). From these paring groups, and the physical functioning analyses, McHorney et al. provided guidelines scale appears reliable enough for comparing in- for interpreting the eight scales. The physical dividuals. The intraclass correlation was 0.85 functioning and mental health scales are rela- for patients with musculoskeletal problems (21). tively pure, being specific to medical or to psy- Item-total correlations typically lie in the middle chiatric disorders. The two role scales chiefly 0.70s (6, Table 5.2). reflect physical or mental conditions, but not ex- Two-week test-retest correlations exceeded clusively. By design, the social functioning and 0.8 for physical function, vitality, and general vitality scales reflect both physical and mental health perceptions; the lowest coefficient was conditions. The general health perceptions scale 0.6 for social function (22, Table II). Assessing appears to be most sensitive to physical health agreement, the mean of the differences in scores problems. Nerenz et al. compared patient SF-36 did not exceed one point on the 100-point scale ratings with physicians’ ratings of the eight di- (22, p162). Test-retest correlations for the scales mensions; these fell between 0.39 and 0.64 (23, after a delay of 6 months ranged between 0.60 pMS117). In general, physicians rated patients and 0.90, except for the pain dimension, with a as healthier than the patients did themselves. correlation of 0.43 (23, Table 5). Slightly lower Predictive validity for physician visits was results were reportd in a British study, in which greatest for the pain and health perceptions 6-month retest Spearman correlations ranged scales; the mental health and role-emotional from 0.28 (SF) to 0.70 (VT) (24, Table 2). scales best predicted hospital admissions, Ware et al. provided a series of tables indicat- whereas vitality, general health perceptions, and ing the estimated sample size requirements, physical function scales best predicted mortality based on the reliability of the scales, for showing over four years (10, pp576–578). a given difference in scores as being statistically Principal components analyses of the MOS significant (3, Tables 7.4 to 7.8). scale scores indicated that a general health di- mension was common to all eight SF-36 scales, Validity explaining 55% of the variance (30, p254). The SF-36 manual presents criterion validity in- When a two-component solution is extracted, formation on the scales, comparing scale scores physical and mental dimensions are clearly dis- to ability to work, symptoms, use of health care, tinguished; the GH and VT scales tend to load and with a range of criteria for the mental health on both, with GH loading mainly on the physi- scale (6, Chapter 9). Each comparison suggested cal factor and VT on the mental factor (6, Table significant and consistent associations with the 9.12; 30, Table 1; 31, Tables 2 to 4). A slightly validation criteria. Item 2, on self-reported different two-factor solution was derived from change, was evaluated in a study that applied a analysis of the eight scales by Kazis et al. The general health rating twice, at an interval of one PF, RP, BP, GH, and VT loaded on one factor, year (6, Table 9.11). There was substantial and the SF, RE, and MH loaded on the second, agreement, although there were some errors at again supporting the interpretation of physical the ends of the scale: 6.9% of those who said and emotional dimensions (32, Table 3). This they were much better had worsened, and 3.4% was further supported in the analysis of over of those who reported being much worse had 163,000 respondents in the Health Outcomes improved. McHorney et al. compared SF-36 Study (25, p19). An analysis of the individual scale scores for patients with varying levels of items provided an eight-factor solution that cor- medical and psychiatric conditions and with responded to the original structure of the SF-36. combinations of both. The scales discriminated However, a nine-factor solution fit the data sig- between types and levels of disease and were nificantly better (33). Correlations among the also able to distinguish people with only a eight scales were reported by Nerenz et al. (23, chronic medical condition from those who had Table 2). These broadly correspond to the  General Health Status and Quality of Life 657 groupings found by McHorney, except that the SF-36 dimension scores with those of the Not- SF scale was more closely associated with physi- tingham Health Profile. Although the correla- cal than mental functioning in the Nerenz study. tions between comparable dimensions in the two A structural equation modeling analysis using a instruments exceeded those between noncompa- large data set assembled from ten countries gen- rable dimensions, the coefficients were not high erally supported the conceptual structure of the (0.52 for the physical scale, 0.55 for pain, 0.67 SF-36 as comprising eight first-order domains, for mental health, 0.68 for vitality) (22, Table three second-order domains (i.e., physical and IV). mental health, and well-being, which comprised The SF-36 appears sensitive to change: an ef- the GH and VT scales), and a single underlying fect size of 0.67 in a study of patients with mus- construct of health (34, p1185). Jette et al. pro- culoskeletal disorders was higher than that for vided an interesting analysis in which they com- the Nottingham Health Profile, SIP, or the Duke bined items from the PF scale, the Functional DUHP (21). Effect sizes in a study of hip re- Independence Measure (FIM), the Minimum placement (6–12 months after the operation) Data Set (MDS), and the Outcome and Assess- were 0.82 (overall), 1.45 for the pain scale, 1.33 ment Information Set for Home Health Care for PF, 1.22 for RP, and 0.8 for social. These (OASIS). Using Rasch analysis, they presented were markedly higher than corresponding statis- the scope of coverage of the four scales on a zero tics for the London Handicap Scale (39, Table (greatest disability) to 100 (no disability) scale. 4). In another study of effect sizes for hip re- The PF scale items covered higher levels of func- placements, the effect size for the PCS was 1.26; tion, ranging from 50 to 100 (35, Figure 2). The that for the pain scale was 1.73 and PF was 1.37. FIM ranged from 25 to 86, whereas the MDS These figures were higher than figures for the and OASIS scales covered almost the entire dis- HUI-2 or HUI-3 (40, Table 2). Effect sizes for ability range. detecting long standing illness in elderly patients Correlations with the Sickness Impact Profile ranged from 0.31 (MH) to 0.96 (PF); these were (SIP) were 0.78 for 106 hip surgery patients (36, comparable with the effect size for the EuroQol Table 3); in a separate study the correlations EQ-5D (24, Table 5). In a comparison of sensi- were 0.73 for overall scores, 0.78 for PF, and tivity to change in a study of hip replacement 0.67 for SF (N = 25 elderly men) (9, Table 2). A patients, the ranking of five instruments de- comparison of the SF-36 and the Quality of pended on whether the overall scores or the Well-Being Scale (QWB) for 916 respondents in- physical or psychological scores were used (36, dicated that the SF-36 accounted for 55.7% of Table 4). Using overall scores, SF-36 was more the variance in the QWB scores; conversely, the sensitive to change than the SIP but less than the QWB accounted for 63.7% of the variance in Arthritis Impact Measurement Scales or the SF-36 physical functioning scores, for 33.8% of Functional Status Questionnaire. The physical variance in general health perceptions, and for score of the SF-36 was more sensitive than all 49.2% of physical role functioning (37). Corre- but one of the other measures, but the psycho- lations of the eight scales of the British version logical score was the least sensitive to change. In of the SF-36 with the EuroQol Quality of Life a study of migraine, the SF-36 produced larger Index ranged from 0.48 to 0.60 (p<0.01) (38, effect sizes than did the COOP charts (41, Ta- p173). Ware et al. present a table listing correla- bles 2, 6). Likewise, the SF-36 was slightly supe- tions between the SF-36 and 15 other health rior to the NHP in a study of myocardial measures. Correlations for the mental health infarction (42). Jenkinson et al. commented that scale range from 0.51 to 0.82 with the corre- the SF-36 may not be adequately sensitive to sponding scales in other leading measures; change when used in survey research (43), al- equivalent correlations for the physical function though it was superior to the EQ-5D (44). scale range from 0.52 to 0.85 (6, Table 9.16). Essink-Bot compared the ability of scores on Brazier et al. present a full multitrait- four measures to discriminate between people multimethod correlation matrix comparing the absent from work due to illness and others. The 658 Measuring Health SF-36 was the most discriminating (mean area correlation of 0.69 with the HUI, and 0.66 with under the ROC curve 0.72), followed by the the EQ-5D were reported in a study of rheuma- COOP Charts (mean area under the curve tology patients (48). [AUC], 0.64), the EuroQol EQ-5D (mean AUC, 0.61) and the Nottingham Health Profile (mean Alternative Forms AUC, 0.60) (20, Table 5). In a study of patients There are several abbreviations of the SF-36, of undergoing hernia surgery, the SF-36 and COOP which the SF-12 has become sufficiently well- charts provided comparable effect size statistics; tested to merit a separate entry in this chapter; values for the physical, pain, and SF scales for the second edition of Measuring Health also both instruments ranged from 1.0 to 1.85 (45, contained a separate entry for the SF-20, but Table 4). The SF-36 manual presents confidence that instrument has now largely been supplanted intervals and estimates of the sample sizes re- by the SF-12. Among the other abbreviations, quired to detect differences of various sizes un- the SF-8 uses one question to represent each of der a variety of statistical designs for each SF-36 the SF-36 domains and requires only one to two scale (6, Tables 7.4–7.9). Ferguson et al. have minutes to complete. Physical and mental com- calculated reliable change index thresholds for ponent scores are produced, along with eight the eight subscales by age and sex (46, Table 2). single-item domain scores (49, p16). Unlike the Values (in terms of T-scores) ranged from 7.35 SF-12, however, items in the SF-8 are not a sub- for the PF scale to 15.8 for the SF scale. set of those in the SF-36. Ware et al. reported a McHorney and others have reviewed ceiling test-retest reliability of 0.73 for the PCS-8 and and floor effects (10, Table 1; 13, p53; 19, Table 0.74 for the MCS-8 (50). The scale may be 7; 25, Table 6). Both effects are most common in scored using an online scoring service at the RP and RE scales, but the ceiling effect is www.qualitymetric.com. In a study of migraine worse, and PF, BP, and SF produce important patients, correlations between the SF-8 and cor- ceiling effects (see, for example, reference 19, responding SF-36 scores ranged from 0.67 (role- Table 6). However, in a sample of relatively emotional scale) to 0.84 (pain) (51, Table 2). healthy people, the SF-36 identified minor levels Minor variants of the SF-36 include the acute of discomfort that were missed by the Notting- version in which the time referent is the past ham Health Profile (22). In a general population week. The questions are otherwise identical. The sample, the SF-36 demonstrated less of a ceiling Veterans SF-36 Health Survey has made changes effect than the EuroQol, on which 64 to 95% of to the RP and RE scales and altered the response respondents achieved maximum scores on the di- options for various questions (32; 52). The mensions of the instrument, compared with 37 RAND Corporation has presented an alternative to 72% for the various dimensions of the SF-36 version of the SF-36 that differs in detail of the (38, p173). wording of two questions and in the scoring O’Brien et al. compared scores on the SF-6D method (53). A condition-specific variant of the (described later in this section) with the HUI for SF-36 has been described for use with patients a sample of 310 patients with cardiac disease. who had undergone knee replacement surgery, The results showed poor agreement between the in which patients were asked to report only two scales: the HUI showed a much wider range those limitations in function due to their knee of utility scores (−0.21 to 1.00) than the SF-6D conditions (26). A comparison of the condition- (range, 0.3–0.95). O’Brien et al. obtained an specific and generic versions indicated that some intraclass correlation between the SF-6D and the of the condition-specific scales (e.g., pain, role HUI of 0.42 (R2 = 0.34) (47, p978). The correla- limitations) were more sensitive to the effects of tions among the six dimensions of the SF-6D treatment than were the equivalent generic were higher than for the HUI, but the low corre- scales, which reflected the combined effect of the lations are an intentional design feature of the knee condition and comorbidities (26, HUI. Similar findings in terms of a narrower dis- pMS248). A 30-item version for use with HIV tribution of scores for the SF-6D and a Pearson patients, the MOS-HIV, takes five to ten minutes  General Health Status and Quality of Life 659 to administer and has been widely used in clini- 3 and 5; 22, Table III). Several other sets of cal trials (54–56, p339). British norms are available (24, Table 3; 28, Brazier et al. have developed an abbreviated Table 3; 72; 73). Canadian norms are available version of the SF-36 that can be scored using a (74), as are figures from a survey of 42,000 Aus- utility-based approach. This is called the SF-6D, tralian women (75), and norms from Romania and includes 8 items from the SF-12 and 3 more (76). Table 10.7 illustrates some of these results. from the SF-36 covering six dimensions: physi- The RAND figures in the table were based on cal functioning, role limitations, social function- 2,471 responses from the MOS and use the ing, pain, mental health, and vitality. Between RAND scoring approach described earlier. two and six levels are specified in each dimen- An interpretation of effect sizes can be de- sion (57, Figure 1; 58, Table 1). Utility weights rived from the comparison of patients with dif- have been developed using a standard gamble ferent types of chronic disease. McHorney et al. (57; 58). noted that a difference of 23 points on the phys- Through the IQOLA project, the SF-36 has ical functioning scale reflects the impact of a been translated and adapted for use in many lan- complicated chronic medical condition, while a guages, including Swedish (59–61), German, difference of 27 points on the mental health Spanish, French, Italian, Danish (62), Dutch, and scale is equivalent to the impact of serious de- Japanese (6, p12:7; 63; 64, p381). A Hebrew pressive symptoms (30, p261). version has been described (65). A Chinese ver- sion has been designed for use in the United Commentary States (66–68). A British-English version alters The SF-36 achieved prominence in a meteoric the phrasing of six items; for example, “block” rise; the New England Health Institute estimated was standardized to “100 yards” and “feeling that by 1992 a million forms were being admin- blue” was changed to “feeling low” (7; 22; 27). istered each year, even before evidence on relia- An alternative form of the five-item mental bility and validity had accumulated. The health scale has been validated; this is useful in publications followed; from a handful in 1991 to studies requiring repeated administrations of the 250 in 1997, 300 in 1998 and 400 in 1999. The scale over short time periods (69). instrument clearly met a need and was also care- fully promoted; there are several advantages in Reference Standards this. Attention is being paid to ensuring standard Norms for the general U.S. population were pre- administration. This was facilitated through the sented for seven age groups and by gender in the Medical Outcomes Trust, a nonprofit organiz- original SF-36 manual, as were norms for 13 dif- ation created to support the development and ferent medical conditions (6, Chapter 10; 8). distribution of standardized outcome measures. The manuals for version 2 provide extensive sets Permission to use the SF-36 should be obtained of norms for the U.S. population (3, pp10:14 to from QualityMetric, which provides updates on 10:38; 14, pp63–102). U.S. norms for each sub- its administration, scoring, and interpretation. scale are also shown in (70, Table 1). Reference The SF-36 is in the public domain, and no royal- standards for people aged 65 and older are pre- ties are required for using it. The IQOLA project sented for the eight scales by disability status, was established to translate and adapt the SF-36 based on a sample of 177,714 participants in the for use outside the United States (61); the project Health Outcomes Study (25, Figure 1). The evaluates the psychometric properties of the in- same source also gives means and standard devi- strument, gathers general population norms and ations for each item (Table 1). Norms for both documents findings in accessible documents. The the SF-36 and SF-12 from a smaller U.S. sample SF-36 Manual is comprehensive and exemplary of 2329 are available (71, Tables 4 to 12). and a necessity for anyone wishing to administer Norms from large British samples show mean the instrument (3). Attention to detail is out- scores by age, gender, socioeconomic class, and standing. For example, 15 internal consistency by the presence of chronic conditions (7, Tables checks are programmed into the SF-36 scoring  Table 10.7 Population Norms for SF-36 Scales. Figures indicate mean scores on the 0–100 scale (standard deviations in parentheses) General Physical Role, Social Mental Role, Health Function Physical Bodily Pain Functioning Health Emotional Vitality Perceptions Sample (PF) (RP) (BP) (SF) (MH) (RE) (VT) (GH) Reference RAND general U.S. 70.6 53.0 70.8 78.8 70.4 65.8 52.2 57.0 (53) population (27.4) (40.8) (25.5) (25.4) (22.0) (40.7) (22.4) (21.1) Elderly U.S. males 52.0 — — 66.2 — — — — (9, Table 2) (30.7) (35.6) 660 U.S. 84.2 81.0 75.2 83.3 74.7 81.3 60.9 72.0 (6) general (23.3) (34.0) (23.7) (22.7) (18.1) (33.0) (21.0) (20.3) population Canada general 85.8 82.1 75.6 86.2 77.5 84.0 65.8 77.0 (74, Table 1) population (20.0) (33.2) (23.0) (19.8) (15.3) (31.7) (18.0) (17.7) Australian 85.1 79.6 70.7 81.4 72.1 77.0 58.1 71.9 (75, Table 1) women (18.7) (35.2) (23.8) (23.7) (18.0) (36.3) (20.9) (20.6) 45–49 years British males 89.8 89.0 81.3 84.7 74.3 88.1 60.8 70.9 (7, Table 3) (18.8) (21.1) (22.2) (22.6) (17.2) (19.9) (18.9) (20.3) British females 86.7 85.8 77.0 81.3 70.1 84.1 55.9 71.3 (7, Table 3) (20.2) (22.5) (23.4) (23.6) (18.7) (21.8) (19.9) (20.5)  General Health Status and Quality of Life 661 to identify respondents who may be making in- ated? Is “in general” intended to exclude specific ternally inconsistent responses (6, p7:16). The health problems? Or, does “in general” refer to manuals provide clear guidelines for administra- a time preceding the current, acute symptoms tion, details on scoring and an extensive sum- they are experiencing? Some items seem inap- mary of validity results and normative data (3; propriate for elderly people (“. . . vigorous activ- 14). ities, such as running, lifting heavy objects”; The history of the SF-36 reflects the chal- “How much did pain interfere with your normal lenges inherent in designing any general health work?”) (79). Question 10 refers to the impact measurement. The tradeoff between covering of physical health or emotional problems on so- many topics superficially and providing detailed cial activities. “Physical health” suggests a con- coverage of a few translates into comprehensive- tinuum, whereas the locution “emotional ness versus precision. In this vein, the scope of problems” is clearly negative and may even im- the first abbreviation of the original MOS instru- ply that the respondent did experience emotional ments, the SF-20, was soon criticized as too nar- problems. It is also curious that two items (num- row. There ensued an explosion of rival bers 6 and 10 in the Exhibit) should so closely abbreviations beginning with a 30-item short replicate each other. form (54); subsequently, 36-, 38-, and 56-item A debate has arisen over the appropriateness versions appeared (77). The SF-36 now replaces of the scoring system for the PCS and MCS, these; the SF-20 has largely been supplanted by which was based on orthogonal factor-loadings. the more recent SF-12 and SF-8 surveys. A com- The goal of this was to provide independent parison of the shortened versions is included in physical and mental scores. Taft et al. argued the QualityMetric Web site. that the negative weights given to physical items The scope of the SF-36 is broad and the on the mental scale (and vice-versa) could con- “quality of life” label appears justified as factors taminate each scale. That is, a perfect score on outside the health domain correlate with scores. the MCS can be obtained only when the respon- The SF-36 is one of the few generic scales to in- dent scores 100 on each scale in the MCS, and clude items on positive health (20). As with any there are no subtractions due to scores on the brief generic instrument, the scope of the SF-36 physical scale (i.e., the physical score must be can be criticized, and some reviewers have com- zero, and vice-versa for the PCS calculation) mented on the absence measures of cognitive (80). Ware and Kosinski responded, arguing function and distress (77). The physical activity mainly on empirical grounds, that this does not items focus on gross motor activities such as occur frequently enough to be an issue (81), but walking, bending, and kneeling, whereas coordi- other authors have noted the same problem with nated actions that might be captured by items the subscale scoring (82; 83). Nortvedt et al., for such as shopping or cooking are not covered (64, example, showed that the SF-36 scoring pro- p380). Floor effects were reported for the role vided a correlation of 0.10 between PCS and scales in version 1 (11; 64, p379), suggesting MCS, whereas RAND scoring of the same items that these scales might not detect further deterio- produced a correlation of 0.70 between the ration in the condition of relatively sick patients. scales (83, Table 2). The revisions in Version 2 appear to have re- With the active debate over the choice be- duced the problem (7, p49). The use of the SF- tween health indexes and health profiles, the 36 with cognitively impaired elderly people has comparisons between the SF-6D and the HUI are received some attention, and it appears that it instructive. Although both used variants of a can be used reliably by such patients (10, p579). standard gamble approach to derive utility Various comments have been made on item weights, the agreement between the two meth- wording and Lessler gave a summary (78). The ods is low (47; 48). This may in part reflect dif- first question concerning health in general may fering item content: the HUI focuses on a be troublesome: is the respondent to undertake a ‘within a the skin’ definition of health, whereas form of averaging where their health has fluctu- the SF questions cover handicap and function in 662 Measuring Health the person’s environment. The SF instruments (6) Ware JE, Jr., Snow KK, Kosinski M, et al. SF- also cover positive aspects of health not included 36 Health Survey: manual and interpretation in the HUI. There were also significant method- guide. Boston: The Health Institute, New En- ological differences in the way weights were de- gland Medical Center, 1993. rived (47, p980). The result is that quality- (7) Jenkinson C, Stewart-Brown SL, Petersen S, et al. Assessment of the SF-36 version 2 adjusted life years estimated from the two in- in the United Kingdom. J Epidemiol struments cannot be directly compared and it is Community Health 1999;53:46–50. not clear which provides the more accurate pic- (8) McHorney CA, Kosinski M, Ware JE, Jr. ture. Comparisons of the costs and quality of It is certain, however, that the SF-36 will con- norms for the SF-36 health survey collected tinue to be the leading general health measure by mail versus telephone interview: results for many years to come. from a national survey. Med Care 1994;32:551–567. Addresses (9) Weinberger M, Samsa GP, Hanlon JT, et The SF-36 web site is at www.sf-36.org, but ex- al. An evaluation of a brief health status tensive information on ordering copies of the measure in elderly veterans. J Am Geriatr scale, manuals, etc., is available from www Soc 1991;39:691–694. .qualitymetric.com/products/SFSurveys.shtml. (10) McHorney CA. Measuring and monitoring general health status in elderly persons: Information on translations is available from practical and methodological issues using the International Quality of Life Assessment the SF-36 Health Survey. Gerontologist (IQOLA) http://www.iqola.org/ 1996;36:571–583. SF-36 Health Survey, The Health Institute, (11) Kurtin PS, Davies AR, Meyer KB, et al. New England Medical Center Hospitals, Box Patient-based health status measures in 345, 750 Washington Street, Boston, Massachu- outpatient dialysis: early experiences in setts, USA 02111 developing an outcomes assessment program. Med Care 1992;30(suppl):MS136–MS149. References (12) Ryan JM, Corry JR, Attewell R, et al. A (1) Ware JE, Sherbourne CD. The MOS 36- comparison of an electronic version of the Item Short-Form Health Survey (SF-36). I. SF-36 general health questionnaie to the Conceptual framework and item selection. standard paper version. Qual Life Res Med Care 1992;30:473–483. 2002;11:19–26. (2) Lohr KN, Brook RH, Kamberg CJ, et al. (13) McHorney CA, Ware JE, Jr., Lu JFR, et al. Use of medical care in the RAND Health The MOS 36-item Short-Form Health Insurance Experiment. Diagnosis- and ser- Survey (SF-36): III. Tests of data quality, vice-specific analyses in a randomized scaling assumptions, and reliability across controlled trial. Med Care diverse patient groups. Med Care 1986;24(suppl):S1–S87. 1994;32:40–66. (3) Ware JE, Jr., Kosinski M, Gandek B. SF-36 (14) Ware JE, Jr., Kosinski M, Dewey JE. How Health Survey: manual and interpretation to score version 2 of the SF-36 Health guide. Lincoln, Rhode Island: Survey (standard and acute forms). QualityMetric Inc., 2002. Lincoln, Rhode Island: QualityMetric, Inc., (4) Stewart AL, Greenfield S, Hays RD, et al. 2005. Functional status and well-being of patients (15) Medical Outcomes Trust. How to score the with chronic conditions. JAMA SF-36 Short-Form Health Survey. Boston: 1989;262:907–913. The Medical Outcomes Trust, 1992. (5) Tarlov AR, Ware JE, Jr., Greenfield S, et al. (16) Nichol MB, Sengupta N, Globe DR. The Medical Outcomes Study: an Evaluating quality-adjusted life years: application of methods for monitoring the estimation of the Health Utility Index results of medical care. JAMA (HUI2) from the SF-36. Med Decis Making 1989;262:925–930. 2001;21:105–112.  General Health Status and Quality of Life 663 (17) Fryback DG, Lawrence WF, Martin PA, et (27) Jenkinson C, Wright L, Coulter A. al. Predicting Quality of Well-being scores Criterion validity and reliability of the SF- from the SF-36: results from the Beaver 36 in a population sample. Qual Life Res Dam Health Outcomes Study. Med Decis 1994;3:7–12. Making 1997;17:1–9. (28) Garratt AM, Ruta DA, Abdalla MI, et al. (18) Franks P, Lubetkin EI, Gold MR, et al. The SF36 health survey questionnaire: an Mapping the SF-12 to preference-based outcome measure suitable for routine use instruments - Convergent validity in a low- within the NHS? Br Med J income, minority population. Med Care 1993;306:1440–1444. 2003;41(11):1277–1283. (29) van der Heijden PGM, van Buuren S, (19) Gandek B, Ware JE, Jr., Aaronson NK, et Fekkes M, et al. Unidimensionality and al. Tests of data quality, scaling reliability under Mokken scaling of the assumptions, and reliability of the SF-36 in Dutch language version of the SF-36. Qual eleven countries: results from the IQLA Life Res 2003;12:189–198. project. J Clin Epidemiol (30) McHorney CA, Ware JE, Jr., Raczek AE. 1998;51:1149–1158. The MOS 36-Item Short-Form Health (20) Essink-Bot M-L, Krabbe PFM, Bonsel GJ, Survey (SF-36): II. Psychometric and et al. An empirical comparison of four clinical tests of validity in measuring generic health status measures. The physical and mental health constructs. Nottingham Health Profile, the Medical Med Care 1993;31:247–263. Outcomes Study 36-item Short-Form (31) Ware JE, Jr., Kosinski M, Gandek B, et al. Health Survey, the COOOP/WONCA The factor structure of the SF-36 Health charts, and the EuroQol instrument. Med Survey in 10 countries: results from the Care 1997;35:522–537. IQOLA project. J Clin Epidemiol (21) Beaton DE, Bombardier C, Hogg-Johnson 1998;51:1159–1165. S. Choose your tool: a comparison of the (32) Kazis LE, Lee A, Spiro A, et al. Measure- psychometric properties of five generic ment comparisons of the Medical health status instruments in workers with Outcomes Study and Veterans SF-36 soft tissue injuries. Qual Life Res Health Survey. Health Care Financ Rev 1994;3:50–56. 2004;25:43–58. (22) Brazier JE, Harper R, Jones NMB, et al. (33) Wolinsky FD, Stump TE. A measurement Validating the SF-36 Health Survey model of the Medical Outcomes Study 36- questionnaire: new outcome measure for item Short-Form Health Survey in a clinical primary care. Br Med J sample of disadvantaged, older, black and 1992;305:160–164. white men and women. Med Care (23) Nerenz DR, Repasky DP, Whitehouse FW, 1996;34:537–548. et al. Ongoing assessment of health status (34) Keller SD, Ware JE, Jr., Bentler PM. Use of in patients with diabetes mellitus. Med structural equation modeling to test the Care 1992;30(suppl):MS112–MS123. construct validity of the SF-36 Health (24) Brazier JE, Walters SJ, Nicholl JP, et al. Survey in ten countries: results from the Using the SF-36 and Euroqol on an elderly IQOLA project. J Clin Epidemiol population. Qual Life Res 1998;51:1179–1188. 1996;5:195–204. (35) Jette AM, Haley SM, Ni P. Comparison of (25) Gandek B, Sinclair SJ, Kosinski M, et al. functional status tools used in post-acute Psychometric evaluation of the SF-36 care. Health Care Financ Rev Health Survey in Medicare managed care. 2003;24:13–24. Health Care Financ Rev 2004;25(4): (36) Katz JN, Larson MG, Phillips CB, et al. 5–25. Comparative measurement sensitivity of (26) Kantz ME, Harris WJ, Levitsky K, et al. short and longer health status instruments. Methods for assessing condition-specific Med Care 1992;30:917–925. and generic functional status outcomes (37) Fryback DG, Dasbach ED, Klein R, et al. after total knee replacement. Med Care Health assessment by SF-36, Quality of 1992;30(suppl):MS240–MS252. Well-Being Index, and time tradeoffs: 664 Measuring Health predicting one measure from another. Med (48) Conner-Spady B, Suarez-Almazor ME. Decis Making 1992;12:348. Variation in the estimation of quality- (38) Brazier J, Jones N, Kind P. Testing the adjusted life-years by different preference- validity of the Euroqol and comparing it based instruments. Med Care with the SF-36 health survey questionnaire. 2003;41:791–801. Qual Life Res 1993;2:169–180. (49) Ware JE, Jr., Kosinski M, Turner-Bowker (39) Harwood RH, Ebrahim S. A comparison DM, et al. How to score Version 2 of the of the responsiveness of the Nottingham SF-12 Health Survey. Lincoln, Rhode extended activities of daily living scale, Island: QualityMetric Inc., 2002. London handicap scale and SF-36. Disabil (50) Ware JE, Kosinski M, Dewey JE, et al. Rehabil 2000;22:786–793. How to score and interpret single-item (40) Blanchard C, Feeny D, Mahon JL, et al. Is health status measures: a manual for users the Health Utilities Index responsive in of the SF-8 Health Survey. Lincoln, Rhode total hip arthroplasty patients? J Clin Island: Quality Metric Incorporated, Epidemiol 2003;56:1046–1054. 2001. (41) Essink-Bot M-L, van Royen L, Krabbe P, et (51) Turner-Bowker DM, Bayliss MS, Ware JE, al. The impact of migraine on health Jr., et al. Usefulness of the SF-8 Health status. Headache 1995;35:200–206. Survey for comparing the impact of (42) Brown N, Melville M, Gray D, et al. migraine and other conditions. Qual Life Comparison of the SF-36 health survey Res 2003;12:1003–1012. questionnaire with the Nottingham Health (52) Kazis LE, Ren XS, Lee A. Health status in Profile in long-term survivors of a myocar- VA patients: results from the Veterans dial infarction. J Public Health Med Health Study. Am J Medical Qual 2000;22:167–175. 1999;14:28–38. (43) Jenkinson C, Layte R, Coulter A, et al. (53) RAND Health Sciences Program. RAND Evidence for the sensitivity of the SF-36 36-Item Health Survey 1.0. Santa Monica: health status measure to inequalities in RAND Corporation, 1992. health: results from the Oxford healthy (54) Wu AW, Rubin HR, Matthews WC. A lifestyles survey. J Epidemiol Community health status questionnaire using 30 items Health 1996;50:377–380. from the Medical Outcomes Study: (44) Jenkinson C, Gray A, Doll H, et al. preliminary validation in persons with Evaluation of index and profile measures of early HIV infection. Med Care health status in a randomized controlled 1991;29:786. trial. Comparison of the Medical (55) Wu AW, Hays RD, Kelly S, et al. Outcomes Study 36-item Short-Form Applications of the Medical Outcomes Health Survey, EuroQol, and disease Study health-related quality of life mea- specific measures. Med Care sures in HIV/AIDS. Qual Life Res 1997;35:1109–1118. 1997;6:531–554. (45) Jenkinson C, Lawrence K, McWhinnie D, (56) Skevington SM, O’Connell KA. Measuring et al. Sensitivity to change of health status quality of life in HIV and AIDS: a review measures in a randomized controlled trial: of the recent literature. Psychol Health comparison of the COOP charts and the 2003;18:331–350. SF-36. Qual Life Res 1995;4:47–52. (57) Brazier J, Usherwood T, Harper R, et al. (46) Ferguson RJ, Robinson AB, Splaine M. Use Deriving a preference-based single index of the reliable change index to evaluate from the UK SF-36 health survey. J Clin clinical significance in SF-36 outcomes. Epidemiol 1998;51:1115–1128. Qual Life Res 2002;11:509–516. (58) Brazier J, Roberts J, Deverill M. The (47) O’Brien BJ, Spath M, Blackhouse G, et al. estimation of a preference-based measure A view from the bridge: agreement of health from the SF-36. J Health Econ between the SF-6D utility algorithm and 2002;21:271–292. the Health Utilities Index. Health Econ (59) Sullivan M, Karlsson J, Bengtsson C, et al. 2003;12:975–981. Health-related quality of life in Swedish  General Health Status and Quality of Life 665 populations: validation of the SF-36 Health health summary scales. Boston: The Health Survey. Qual Life Res 1994;3:95. Institute, New England Medical Center, (60) Taft C, Karlsson J, Sullivan M. Perfor- 1995. mance of the Swedish SF-36 version 2.0. (72) Jenkinson C, Coulter A, Wright L. Short Qual Life Res 2004;13:251–256. Form 36 (SF-36) Health Survey (61) Ware JE, Jr., Gandek B. Overview of the questionnaire: normative data from a large SF-36 Health Survey and the International random sample of working age adults. Br Quality of Life Assessment (IQLA) project. Med J 1993;306:1437–1440. J Clin Epidemiol 1998;51:903–912. (73) Jenkinson C, Wright L, Coulter A. Quality (62) Bjorner JB, Kreiner S, Ware JE, Jr., et al. of life measurement in health care: a Differential item functioning in the Danish review of measures and population norms translation of the SF-36. J Clin Epidemiol for the UK SF-36. Oxford: Health Services 1998;51:1189–1202. Research Unit, University of Oxford, (63) Aaronson NK, Acquadro C, Alonso J, et 1993. al. International quality of life assessment (74) Hopman WM, Towheed T, Anastassiades (IQOLA) Project. Qual Life Res T, et al. Canadian normative data for the 1992;1:349–351. SF-36 health survey. Can Med Assoc J (64) Anderson RT, Aaronson NK, Wilkin D. 2000;163:265–271. Critical review of the international (75) Mishra G, Schofield MJ. Norms for the assessments of health-related quality of life. physical and mental health component Qual Life Res 1993;2:369–395. summary scores of the SF-36 for young, (65) Lewin-Epstein N, Sagiv-Schifter T, Shabtai middle-aged and older Australian women. EL, et al. Validation of the 36-item Short- Qual Life Res 1998;7:215–220. form Health Survey (Hebrew version) in (76) Mihaila V, Enachescu D, Davila C, et al. the adult population of Israel. Med Care General population norms for Romania 1998;36:1361–1370. using the Short Form 36 health survey (SF- (66) Ren XS, Amick B, Zhou L, et al. 36). Quality of Life Newsletter Translation and psychometric evaluation of 2001;26:17–18. a Chinese version of the SF-36 Health (77) Hays RD, Shapiro MF. An overview of Survey in the United States. J Clin generic health-related quality of life mea- Epidemiol 1998;51:1129–1138. sures for HIV research. Qual Life Res (67) Chang DF, Chun CA, Takeuchi DT, et al. 1992;1:91–97. SF-36 Health Survey: tests of data quality, (78) Lessler JT. Choosing questions that people scaling assumptions, and reliability in a can understand and answer. Med Care community sample of Chinese Americans. 1995;33(suppl):AS203–AS208. Med Care 2000;38:542–548. (79) Lloyd A. Assessment of the SF-36 version 2 (68) Yu J, Coons SJ, Draugalis JR, et al. in the United Kingdom [letter]. J Epidemiol Equivalence of Chinese and US-English Community Health 1999;53:651. versions of the SF-36 Health Survey. Qual (80) Taft C, Karlsson J, Sullivan M. Reply to Life Res 2003;12:449–457. Drs. Ware and Kosinski. Qual Life Res (69) McHorney CA, Ware JE, Jr. Construction 2001;10:415–420. and validation of an alternate form (81) Ware JE, Kosinski M. Interpreting SF-36 general mental health scale for the summary health measures: a response. Medical Outcomes Study Short-Form 36- Qual Life Res 2001;10:405–413. item Health Survey. Med Care (82) Wilson D, Parsons J, Tucker G. The SF-36 1995;33:15–28. summary scales: problems and solutions. (70) Taft C, Karlsson J, Sullivan M. Do SF-36 Soz Praventivmed 2000;45:239–246. summary component scores accurately (83) Nortvedt MW, Riise T, Myhr K-M, et al. summarize subscale scores? Qual Life Res Performance of the SF-36, SF-12, and 2001;10:395–404. RAND-36 summary scales in a multiple (71) Ware JE, Jr., Kosinski M, Keller SD. How sclerosis population. Med Care to score the SF-12 physical and mental 2000;38:1022–1028. 666 Measuring Health The Short-Form-12 Health Survey for use must be obtained from QualityMetric. The scale can be administered in two to three minutes (John E. Ware, Jr., 1994, (2, p7). revised 1998) The principal scores from the SF-12 are a Purpose physical health composite score (PCS-12) and a This abbreviation of the SF-36 Health Survey mental health score (MCS-12). In addition, in was designed to be broad ranging but brief the second version, an eight-domain profile can enough for practical use in large-scale surveys be produced, providing scores for Physical Func- and yet still reproduce the physical and mental tioning, Role Physical, Bodily Pain, General scores of the complete Survey. Its main applica- Health, Vitality, Social Functioning, Role Emo- tion is in surveys and in outcome studies where tional, and Mental Health (3, pp12, 36). The space constraints prevent use of the SF-36. scoring approach was changed between the first and second versions of the SF-12. Version 2 may Conceptual Basis be scored in a conventional manner in several Although the SF-36 includes eight dimensions, steps: first, out-of-range values are treated as the physical and mental summary scores account missing; next, scores for some items are reversed for 80 to 85% of reliable variance in the eight and some are recalibrated; total scores for each scores, so reducing the number of health dimen- domain are then calculated and, finally, these are sions does not seriously compromise validity. transformed into a zero to 100 scale. Full in- Hence, Ware et al. argued that an abbreviated structions are given in the scoring manual, version that covers just physical and mental di- Chapter 6 (3, pp29–38). Alternatively, norm- mensions of health should prove valid and prac- based scoring uses regression weights to stan- tical for general use (1, p221). The SF-12 was dardize each of the eight scale scores to a mean developed with the goals of accounting for at of 50 and standard deviation (SD) of 10, using least 90% of the variance in the SF-36 physical weights based on the general U.S. population (4, and mental summary scores, of providing sum- pp437–438), although reference values are avail- mary scores that would coincide with the aver- able for other countries. PCS and MCS scores age scores on the complete SF-36, and of being are likewise based on factor score weights for brief enough to be printed on a single page and combining the eight scale scores (3, pp45–59). administered in less than two minutes. The norm-based scoring method provided a sig- nificantly stronger correlation with SF-36 scores Description than a simpler equal-interval scoring approach The first major abbreviation of the SF-36 was the (1, p222). For simplicity, scoring can also be SF-20, which was reviewed in the second edition done online, at http://www.qualitymetric.com/ of Measuring Health. Subsequently, Ware et al. products/products.shtml. The online scoring found that ten items from six of the eight SF-36 system can also handle incomplete data, which scales reproduced 90% of the variance in the phys- normally prevents the calculation of PCS or ical and mental scores; they added one item each MCS scores. from the remaining two scales, thus forming the To facilitate use in cost-effectiveness studies, a 12-item version that covers the same dimensions as pilot study has derived utility scores for the SF-12 the original SF-36 (2, p6). The first version of the by mapping the PCS and MCS scores onto the SF-12 was produced in 1994, and a second version EuroQol EQ-5D and the Health Utilities Index was presented in 1998. The items of version 2 are (5). Regression analyses explained an R2 of 0.51 shown in Exhibit 10.30, which shows a generic for the Health Utilities Index (HUI)-3 and 0.59 U.S. version; this is also available on the Quality- for the EQ-5D (5, Table 2). Scores for the EQ-5D Metric web site (http://www.qualitymetric.com/ and the HUI that were estimated from the SF-12 demos/SF12v2.html), which includes other forms. correlated 0.77 and 0.71 with the actual EQ-5D These differ in terms of the recall period: acute and HUI, respectively (5, Table 3). In a similar (one week) and standard (four week). Permission manner, utilities have been calculated for remis- Exhibit 10.30 The Short-Form-12 Health Survey, Version 2 Your Health and Well-Being This survey asks for your views about your health. This information will help keep track of how you feel and how well you are able to do your usual activities. Thank you for completing this survey! For each of the following questions, please mark an X in the one box that best describes your answer. 1. In general, would you say your health is: Excellent Very Good Good Fair Poor ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 2. The following questions are about activities you might do during a typical day. Does your health now limit you in these activities? If so, how much? Yes, Yes, No, Not Limited Limited A Limited A Lot Little At All ▼ ▼ ▼ a. Moderate activities, such as moving a table, pushing a vacuum cleaner, bowling, or playing golf . . . . . . . 1 1 3 b. Climbing several flights of stairs . . . . . . . . . . . . . . . 1 2 3 3. During the past 4 weeks, how much of the time have you had any of the following problems with your work or other regular daily activities as a result of your physical health? All of Most Some A Little None the of the of the of the of the Time Time Time Time Time ▼ ▼ ▼ ▼ ▼ a. Accomplished less than you would like . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 b. Were limited in the kind of work or other activities . . . . . . . . . . . . . . . . . 1 2 3 4 5 4. During the past 4 weeks, how much of the time have you had any of the following problems with your work or other regular daily activities as a result of any emotional problems (such as feeling depressed or anxious)? All of Most Some A Little None the of the of the of the of the Time Time Time Time Time ▼ ▼ ▼ ▼ ▼ a. Accomplished less than you would like . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5 b. Did work or other activities less carefully than usual . . . . . . . . . . . 1 2 3 4 5 5. During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)? Not at all A little bit Moderately Quite a bit Extremely ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 (continued) 667 668 Measuring Health Exhibit 10.30 (continued) 6. These questions are about how you feel and how things have been with you during the past 4 weeks. For each question, please give the one answer that comes closest to the way you have been feeling. How much of the time during the past 4 weeks . . . All of Most Some A Little None the of the of the of the of the Time Time Time Time Time ▼ ▼ ▼ ▼ ▼ a. Have you felt calm and peaceful? . . . . 1 2 3 4 5 b. Did you have a lot of energy? . . . . . . . 1 2 3 4 5 c. Have you felt downhearted and depressed? . . . . . . . . . . . . . . . . . . 1 2 3 4 5 7. During the past 4 weeks, how much of the time has your physical health or emotional problems interfered with your social activities (like visiting friends, relatives, etc)? All of Most Some A little None the of the of the of the of the time time time time time ▼ ▼ ▼ ▼ ▼ 1 2 3 4 5 Adapted from Ware JE Jr., Kosinski M, Turner-Bowker DM, Gandek B. How to score version 2 of the SF-12 Health Survey. Lincoln, R.I., QualityMetric Inc., October 2002 sion of depression as measured by the SF-12 (6). Cronbach’s alpha was 0.84 for the MCS Information is available from www.quality items and 0.81 for the PCS (8, p3). metric.com/products/Utility_Index.aspx A factor analysis confirmed the placement of physical and mental health items on two sepa- rate factors; physical and mental scores corre- Reliability lated 0.18 (9, Tables 4 and 5). In Ware’s original description, test-retest relia- Ware et al. provided power tables, indicating bility for the PCS-12 was 0.89 in the United the sample sizes required to demonstrate a given States and 0.86 in the United Kingdom. Coeffi- difference in SF-12 scores before and after an cients for the MCS were 0.76 and 0.77 (1, intervention as statistically significant (3, p225). From the manual describing version 2 of pp66–70). the SF-12, theta reliability estimates range from 0.73 to 0.87 across the eight scales, whereas the Validity value for the PCS-12 was 0.89 and that of the The 12 items from version 1 predicted PCS MCS-12, 0.86 (3, p64). scores on the complete SF-36 (R2 = 0.91), whereas The SF-12 was compared with the SF-36 in a the R2 for the MCS was 0.92 (1, p224). Ware et sample of patients with arthritis. Intraclass relia- al. compared SF-12 scores with SF-36 scores de- bility correlations were 0.75 for the SF-12 ver- rived from the same data set; the correlation be- sion, compared with 0.81 for the full SF-36. The tween the SF-12 and SF-36 PCS was 0.95 and correlation between the two scales was 0.94 (7). was 0.97 for the MCS (3, p22). They reported The correlation between the PCS scores on the that, in a number of international studies, the two instruments in another study was 0.95; the PCS correlations ranged from 0.94 to 0.96, and correlation for the MCS was 0.97 (1, p231). those for the MCS ranged from 0.94 to 0.97 (3,  General Health Status and Quality of Life 669 p27). Agreement between scores was also high; item functioning (DIF) in the 12 items using a scores on average differed by less than one large population sample (N = 11,626) (12). DIF point, and the percentile values agreed within was evaluated by gender, age, race, and educa- less than one point. The standard deviations for tion; most items exhibited DIF although the the SF-12 were often lower than those for the physical and emotional role items exhibited the SF-36 (2, Tables 4–12). This finding of equiva- least (12, Table 5). They found, for example, lence was largely replicated in a British study of that adjusting for DIF “reduced the effects of three different patient groups (10). In an ex- black race on mental health”, suggesting that tended comparison of the ability of the SF-36 previous findings of a racial contrast in mental and SF-12 to discriminate between diagnostic health scores may partly reflect differing inter- groups, Ware et al. evaluated the validity of the pretation of items by different racial groups (12, SF-12 in comparison to that of the complete SF- pIII–82). In the same study, a confirmatory fac- 36, using a relative validity statistic. Across tor analysis suggested that a simple two-factor several comparisons of different medical condi- model did not fit the data. Instead, allowing gen- tions, they reported a median relative validity of eral health, energy, and social activities items to 0.67 for the PCS-12, compared with the most load on both physical and mental factors discriminating subscale of the SF-36. The me- brought the fit within acceptable boundaries (12, dian for the MCS-12 was 0.97, i.e., it performed Table 3). The correlation between the physical virtually as well as did the best subscale in the and mental factors was 0.57 after controlling for complete instrument (1, p229). These analyses demographic and clinical factors and for DIF were subsequently extended in evaluating ver- (12, pIII–81) sion 2 (3, Table 12.1 to 12.12). The overall im- Effect sizes were estimated for a comparison pression is that the SF-12 performs remarkably between patients with serious physical condi- well compared with the SF-36. tions, compared with others with minor com- Scores on the RAND version differed signifi- plaints. The PCS-12 gave an effect size of 0.87, cantly for patients with diabetes of differing du- whereas the MCS-12 was 0.13. The equivalent ration and severity (4, Table 3). Physical health results for the complete SF-36 were 0.91 and scores discriminated significantly between pa- 0.18: higher, but only marginally so. In a com- tients ranked in the different levels of the New parison between patients with serious physical York Heart Association (NYHA) classification; or mental conditions and those with more minor the mental scale showed significant differences mental problems, the effect size was 1.30 for the only between NYHA levels 1 + 2 and 3 + 4 (9, PCS-12 compared with 1.45 for the SF-36 PCS. Table 6). For the MCS-12 the result was 0.61 for both In a study of patients with heart failure, the measures (1, Table 2). An Australian study of SF-12 showed neither ceiling nor floor effects, heart and stroke patients demonstrated that the whereas two disease-specific instruments, the SF-12 is able to show differences between diag- congestive heart failure questionnaire and the nostic groups and length of hospital stay (8, Minnesota Living with Heart Failure question- Table 4). naire, did (9, Table 3). The SF-12 appears to suf- fer less of a ceiling effect than the EuroQol Alternative Forms EQ-5D: it discriminated among people who Through the International Quality of Life As- scored in perfect health on the EQ-5D (11, Table sessment project, the SF-12 has been translated 9). The PCS-12 correlated 0.55 with the EQ-5D and adapted for use in 36 countries (http://www VAS score whereas the MCS-12 correlated 0.41 .qualitymetric.com/products/surveys/SF12v2.sht (11, p163). ml). Languages include Spanish, French, Ger- Physical health scores correlated 0.60 with man, Italian, Japanese, Chinese, Korean, and the HUI-2, whereas the MCS-12 correlated 0.59 Vietnamese (1, p232) and validation studies (4, Table 7). have been undertaken in most of these. Fleishman and Lawrence studied differential 670 Measuring Health Reference Standards questions on memory or cognition may prove a For version 1 of the SF-12, mean PCS-12 scores limitation (14). The short forms have been criti- were 50.12; mean MCS-12 scores were 50.04 in cized as being unable to discriminate among a U.S. population sample (N = 2329) (2, Table 4). more severe levels of disability (15), although Equivalent data from the larger 2000 Medical this has been improved in version 2 of the SF-12 Expenditure Panel Survey (N = 11,626) gave (3, p25). There has been concern over comple- mean PCS scores of 50.06 (SD, 10.39) and 51.43 tion rates for the instrument in community stud- for the MCS (SD, 9.1) (12, pIII–79). This survey ies. In an Australian study, only 78% completed also provided mean scores by age group, sex, ed- all 12 items (8, p2); in a British study, 15.9% of ucation, race, and for various chronic health con- 55,000 respondents gave incomplete responses ditions (12, Table 1). Ware et al. reported mean (13, Table 1). scores and standard deviations for four categories Overall, however, the SF-12 appears to be re- of medical conditions (1, Table 1). markably effective as a brief but broad-ranging For version 2, the scoring manual provides a instrument, certainly suitable for survey use and huge array of norms by age, sex, and for a range probably also sensitive to change as an evalua- of medical conditions (3, pp73–123). This has tive instrument. to represent the most extensive set of norms of any scale reviewed in this book. Address U.K. population norms are available from a Information on the SF-12 can be obtained study of over 60,000 respondents, by sex, age, from http://www.sf-36.org/tools/sf12.shtml and and ethnic group (13, Tables 2–5). Reference from the QualityMetric site, at http://www values for an Australian sample show percentiles .qualitymetric.com/products/SFSurveys.shtml. for PCS and MCS scores by sex and age group (8, Figure 2). References (1) Ware JE, Jr., Kosinski M, Keller SD. A 12- Commentary Item Short-Form Health Survey: construc- The SF-12 was based on the extensively tested tion of scales and preliminary tests of and internationally recognized Medical Out- reliability and validity. Med Care comes Study measures. The items are well writ- 1996;34:220–233. ten; the instrument was carefully developed and (2) Ware JE, Jr., Kosinski M, Keller SD. How to score the SF-12 physical and mental has been tested in large population studies. Doc- health summary scales. Boston: The Health umentation and the Web site are excellent. The Institute, New England Medical Center, improvements in scale precision from version 1 1995. to 2 of the SF-12 were illustrated by Ware et al. (3) Ware JE, Jr., Kosinski M, Turner-Bowker (3, Figure 5.2). In survey applications, the SF-12 DM, et al. How to score Version 2 of the substantially reproduces the PCS and MCS scale SF-12 et al. Health Survey. Lincoln, Rhode values obtained by the SF-36, yet imposing only Island: QualityMetric Inc., 2002. one third of the respondent burden. Used with (4) Maddigan SL, Feeny DH, Johnson JA. large samples, for example, in health surveys, the Construct validity of the RAND-12 and loss of precision is unlikely to be significant. Health Utilities Index Mark 2 and 3 in type Ware et al. noted “the SF-12 versions [of the 2 diabetes. Qual Life Res 2004;13:435–448. PCS and MCS] define fewer levels and pool less (5) Franks P, Lubetkin EI, Gold MR, et al. reliable variance and should, therefore, be ex- Mapping the SF-12 to preference-based pected to yield less reliable assignments of indi- instruments—Convergent validity in a low- viduals to those levels.” (1, p231). income, minority population. Med Care Several general comments have been made 2003;41:1277–1283. about the abbreviated versions of the SF-36. In (6) Lenert LA, Sherbourne CD, Sugar C, et al. assessing elderly populations, the omission of Estimation of utilities for the effects of  General Health Status and Quality of Life 671 depression from the SF-12. Med Care resources; by permitting comparisons across 2000;38:763–770. treatments and medical conditions, it is intended (7) Hurst NP, Ruta DA, Kind P. Comparison to measure the performance of a hospital or of the MOS short form-12 (SF12) health health service as a whole (1; 2). status questionnaire with the SF36 in patients with rheumatoid arthritis. Br J Conceptual Basis Rheumatol 1998;37:862–869. The Disability and Distress Scale originated in a (8) Lim LLY, Fisher JD. Use of the 12-item Short-Form (SF-12) Health Survey in an program to measure hospital output. A hospital Australian heart and stroke population. was viewed as a system that organizes resources Qual Life Res 1999;8:1–8. (e.g., staff, equipment) to change the state of pa- (9) Bennett SJ, Oldridge NB, Eckert GJ, et al. tients between admission and discharge. The Discriminant properties of commonly used output of the system is recorded in terms of the quality of life measures in heart failure. difference between utilities assigned to the states Qual Life Res 2002;11:349–359. of health on admission and discharge; the effect (10) Jenkinson C, Layte R, Jenkinson D, et al. A of death is incorporated by adding the difference shorter form health survey: can the SF-12 in utilities for death and residual morbidity replicate results from the SF-36 in among survivors (1, pp134–135; 3). Utilities are longitudinal studies? J Public Health Med not seen as absolute but as expected to vary ac- 1997;19:179–186. (11) Johnson JA, Coons SJ. Comparison of the cording to the patient’s morbidity state, attrib- EQ-5D and SF-12 in an adult US sample. utes (e.g., age, occupation, marital status) and Qual Life Res 1998;7:155–166. also by the characteristics of the raters. (12) Fleishman JA, Lawrence WF. Demographic Based on interviews with doctors, econo- variation in SF-12 scores: true differences mists, and health administrators, Rosser identi- or differential item functioning? Med Care fied disability and distress as the two 2003;41:III-75–III-86. characteristics used to define the severity of a (13) Jenkinson C, Chandola T, Coulter A, et al. health condition. An assessment of the construct validity of the SF-12 summary scores across ethnic Description groups. J Public Health Med The Disability and Distress Scale can be adminis- 2001;23:187–194. (14) Carver DJ, Chapman CA, Thomas VS, et tered by a clinician as a rating scale, or through al. Validity and reliability of the Medical a self-completed questionnaire. Two dimensions Outcomes Study Short Form-20 are rated: disability, which includes eight cate- questionnaire as a measure of quality of gories ranging from no disability to unconscious; life in elderly people living at home. Age and subjective distress, with four categories (see Ageing 1999;28:169–174. Exhibit 10.31). The disability dimension covers (15) Bindman AB, Keane D, Lurie N. Measur- mobility and social function and includes ele- ing health changes among severely ill ments that a sample of physicians considered patients: the floor phenomenon. Med Care when judging severity (1, p136). Patients in the 1990;28:1142–1152. first seven disability categories are also rated on the four-category distress scale, giving 7 × 4, or 28 possible combinations, to which ‘uncon- The Disability and Distress Scale scious’ is added, making 29 categories. Because (Rachel M. Rosser, 1978) unconscious people are assumed not to be ac- Purpose tively suffering, they are classified as being free Rosser’s Disability and Distress Scale provides a from distress (4, p56). numerical summary index of health status that For the clinician-rating version, exact ques- can be used in estimating quality-adjusted life tions are not specified, but the patient is classi- years (QALYs). The scale is designed for evalua- fied on the basis of observation, interview, chart tion, planning, and allocation of health service review, or other health status measurements. 672 Measuring Health Exhibit 10.31 The Disability and Distress Scale: Descriptions of States of Illness Note: The Disability rating describes the extent to which a patient is judged to be unable to pursue the activities of a normal person at the time at which the classification is made. Patients in Class 2 are slightly disabled, but performance in their normal work is not limited. This degree of disablement affects social activities and personal relations. It includes such conditions as mild cosmetic defects, slight injuries and diseases which may interfere with hobbies but not with essential activities, and some of the less severe psychiatric states which cause some social disablement. The Distress rating describes the patient’s pain, mental suffering in relation to disablement, anxiety, and depression. Disability 8 Unconscious 7 Not in 8 but confined to bed 6 Not in 7 but confined to chair or wheelchair or able to move around in the home only with support from an assistant 5 Not in 6 but unable to undertake any paid employment. Unable to continue any education. Old people confined to home except for escorted outings and short walks and unable to do shopping. Housewives only able to perform a few simple tasks 4 Not in 5 but choice of work or performance at work very severely limited. Housewives and old people able to do light housework only, but able to go out shopping 3 Not in 4 but severe social disability and/or slight impairment of performance at work. Able to do all housework except very heavy tasks 2 Not in 3 but slight social disability 1 No disability Distress Pain and mental suffering D Severe C Moderate B Mild A None Adapted from Rosser RM. A health index and output measure. In: Walker SR, Rosser RM, eds. Quality of life: assessment and ap- plication. Lancaster: MTP Press, 1987:Table 7.1. Clinicians familiar with the patient can make the Exhibit 10.32 The Disability and Distress rating in about ten seconds; patient acceptability Scale: Valuation Matrix for Health States is not an issue as no questions need be asked. Rosser warns against using ratings by nonclini- Distress rating cians, who may be unduly influenced by factors A B C D such as patient denial (2, p141). 1 1.000 0.995 0.990 0.967 Several approaches to calculating values for 2 0.990 0.986 0.973 0.932 the 29 states were tested, including an analysis 3 0.980 0.972 0.956 0.912 of the monetary value of court awards for injury 4 0.964 0.956 0.942 0.870 compensation (3). The weights normally used 5 0.946 0.935 0.900 0.700 were derived from a magnitude estimation scal- 6 0.875 0.845 0.680 0.000 7 0.677 0.564 0.000 −1.486 ing task, setting optimal health at 1.0 and death 8 −1.028 at 0.0, providing an interval scale (5). This ap- proach does not assume that death is the worst Adapted from Rosser RM. A health index and output measure. In: Walker SR, Rosser RM, eds. Quality of life: assessment possible state, and states judged worse than and application. Lancaster: MTP Press, 1987:Table 7.2. death can receive negative valuations. Rosser ar-  General Health Status and Quality of Life 673 gued that phenomena such as suicide, with- pensation awarded to patients with varying types drawal of life support, and living wills indicate of disability as a result of injury (8, p63). The that death should not be the worst state on the correlation was 0.81; the main discrepancy lay in scale (2). The original weights were based on the more severe states, where the psychometric judgments made by a mixed group of 70 doc- valuations were much more severe (1, Figure 7.3; tors, nurses, patients, and health volunteers (6). 5). They are shown in Exhibit 10.32. Because of the The cost utilities for various orthopedic inter- small size of this sample, however, the weights ventions as calculated by the Rosser index and have been revisited; Williams and Kind, for ex- the EuroQol were compared by James et al. The ample, altered some of the valuations (7). two instruments provided different rank-orders for the operations; both instruments recorded Reliability significant differences between preoperative and Test-retest reliability of the scaling procedure postoperative scores, and the Rosser scale was estimated for 50 volunteers and the percent- showed more agreement between patient and age agreement was 97.2%; internal consistency physician ratings (perhaps because it does not of the ratings was estimated for ten raters and contain subjective aspects of health that are in- nine of the ten were internally consistent (6, cluded in the EuroQol) (9, pp186–187). p351). Alternative Forms Validity The distress rating described here was used in Rosser has paid considerable attention to the va- most of Rosser’s work but was subsequently ex- lidity of the scale weights derived from the mag- tended to distinguish five levels of pain and five nitude estimation procedure. The extent to levels of emotional distress (1, p137). This gives which this provided a true ratio scale was tested 175 possible states in the Index of Health- by plotting the presumed ratio values against Related Quality of Life (IHQL) (10; 11, values calculated under a simpler ordinal as- pp147–148). The hierarchical structure of the sumption; the results showed that the expected index divides global health-related quality of life log-linear relationship was obtained (1, p140). into dimensions comprising disability, discom- Rosser also showed that when the data were fort, and distress. The disability dimension is processed as a paired comparisons exercise, the further divided into dependence and dysfunc- resulting interval scale could be transformed tion; the discomfort dimension includes pain and into a ratio scale with a close fit (r = 0.93) (8, symptoms, whereas the distress dimension is di- p63). These results offered some evidence that vided into dysphoria, disharmony, and fulfil- the ratio scale had, in fact, been achieved. Scale ment (11, p150). Three questionnaires have weights were calculated for subsets of the origi- been designed for the IHQL, including a self- nal 70 judges, and there was no difference ac- completed version, a version used by a trained cording to the age, social class, or other observer, and a version completed by a relative characteristics of the judges (6). Comparing (11, p82). scale weights derived from different types of pa- A self-completed questionnaire for the Dis- tients suggested that patients with psychiatric ability and Distress Scale was proposed by the disorders made more negative valuations of the Health Economics group at York University in health states than medical patients (1, Figure 2; England (12; 13). This called the Health Mea- 5). However, of 14 such comparisons made, only surement Questionnaire, covers mobility, self- two reached statistical significance (6, Table 4). care, usual activities, social relationships, and The assignment of weights therefore appears rel- feelings (13, p87). It is reproduced in Appendix atively robust. 1 of the book by Patrick and Erickson (14, An external validation of the scale values was pp377–379). The questions permit classification obtained from comparisons with levels of com- into the 29 states, and Rosser’s utility weights 674 Measuring Health are then applied. Scores correlated −0.43 with the General Health Questionnaire; correlations References with section scores on the Nottingham Health (1) Rosser RM. A health index and output Profile ranged from −0.23 to −0.62 (13, measure. In: Walker SR, Rosser RM, eds. pp88–89). Quality of life: assessment and application. Lancaster, U.K.: MTP Press, Commentary 1987:133–160. Compared with the Quality of Well-Being Scale (2) Rosser RM. Recent studies using a global (QWB), Rosser’s scale is briefer and simpler to approach to measuring illness. Med Care 1976;14(suppl):138–147. administer; it deliberately combines ratings of (3) Rosser RM, Watts VC. The measurement function with subjective distress. It has, however, of hospital output. Int J Epidemiol not been as fully examined for validity; we know 1972;1:361–368. little about how it compares with the QWB or to (4) Rosser R. A history of the development of other scales. The QWB is likely to be more sensi- health indicators. In: Teeling-Smith G, ed. tive to minor deviations from well-being owing Measuring the social benefits of medicine. to its inclusion of mild symptoms. Because the London: Office of Health Economics, Rosser classification contains no positive health 1983:50–62. items, it is unlikely to discriminate among minor (5) Kind P, Rosser R, Williams A. Valuation of levels of disability, and so is unlikely to be suit- quality of life: some psychometric evidence. able for community health surveys. In one repre- In: Jones-Lee MW, ed. The value of life and safety. Amsterdam: Elsevier North- sentative population sample, 74.6% of people Holland, 1982:159–170. reported no disability and 54.6% reported no (6) Rosser R, Kind P. A scale of valuations of distress on the scale (13, Table 3). states of illness: is there a social consensus? Although primarily intended as a research Int J Epidemiol 1978;7:347–358. tool for policy formation, the scale can also be (7) Williams A, Kind P. The present state of used in individual case studies. Administered at play about QALYs. In: Hopkins A, ed. admission and discharge, it can be used, for ex- Measures of the quality of life. London: ample, to describe the output of hospital services; Royal College of Physicians, 1992:21–34. Rosser gave a telling illustration of the use of the (8) Rosser R. Issues of measurement in the scale in peer review, drawing comparisons of the design of health indicators: a review. In: output of medical and surgical specialties in a Culyer AJ, ed. Health indicators. Amsterdam: North Holland Biomedical teaching hospital and comparing the progress Press, 1983:36–81. made by patients seen by different clinicians (9) James M, St Leger S, Roswell KV. within each specialty (1). She noted that the scale Prioritising elective care: a cost utility “seems to be yielding information and insights analysis of orthopaedics in the north west which cannot be obtained from routinely- of England. J Epidemiol Community available data and which are relevant to clinical Health 1996;50:182–189. practice and Health Service management and (10) Rosser R. The history of health related policy” (1, p151). The scale offers a direct com- quality of life in 101⁄2 paragraphs. J R Soc parison of the progress made by patients under Med 1993;86:315–318. different teams and highlights differences in cri- (11) Rosser R, Cottee M, Rabin R, et al. Index teria for admission, length of stay, and discharge. of Health-related Quality of Life. In: Hopkins A, ed. Measures of the quality of As originally formulated, however, the method is life and the uses to which such measures not expected to be sufficiently sensitive to serve may be put. London: Royal College of as an outcome indicator for clinical trials; in- Physicians of London, 1992: 81–89. stead, its intent is more to compare results across (12) Williams A. Applications in management. trials (1, p158). Time will tell whether the modi- In: Teeling-Smith G, ed. Measuring health: fied Index of Health-Related Quality of Life of- a practical approach. Chichester, U.K.: fers a more sensitive scale. John Wiley, 1988:225–243.  General Health Status and Quality of Life 675 (13) Kind P, Gudex CM. Measuring health gram by the number of Well-Years gives its status in the community: a comparison of relative efficiency or “cost-effectiveness.” (9, methods. J Epidemiol Community Health p64) 1994;48:86–91. (14) Patrick DL, Erickson P. Health status and In the years since the QWB was developed, the health policy: quality of life in health care concept of QALYs, (see Chapter 2) has come to evaluation and resource allocation. New be used as an alternative to cost-benefit analysis York: Oxford University Press, 1993. for allocating health care resources; well-years form the units through which QALYs can be The Quality of Well-Being Scale measured. “Quality-adjusted life years integrate (Formerly the Index of Well-Being) mortality and morbidity to express health status (J.W. Bush and R.M. Kaplan, 1973, in terms of equivalents of well-years of life” (10, Revised 1976, 1994) p66). The General Health Policy Model and the Purpose QWB Scale are based on a three-component The Quality of Well-Being Scale (QWB) is a model of health (4). The assessment of health health index that summarizes a person’s current begins with an objective appraisal of current symptoms and disability in a single number that functional status, based on performance. Sec- represents a judgment of the social undesirabil- ond, a value reflecting the relative desirability or ity of the problem and expresses it in terms of utility is associated with each functional level (1; quality-adjusted life years (QALYs). This value 11). These values are anchored at zero, which can be adjusted to reflect the likely prognosis of represents death. Otherwise, if mortality were any existing medical condition. The QWB is in- ignored, the death of a disabled patient would tended for use as an outcome indicator and in es- appear to improve the net population estimate of timating present and future need for care (1; 2). health status (12). Third, health implies a con- It can be applied to individuals and to popula- sideration not only of present state but also of tions and can be used with any type of disease. the future prognosis for any illness present. This permits a distinction to be drawn, for example, Conceptual Basis between two people with similar current func- The QWB is part of a General Health Policy tional levels, one of whom has a malignancy: in- Model that defines an approach to quantifying formation crucial in assessing future need for the output of a health care system (3–8). To health care. The prognostic component can also compare treatments for different types of dis- reflect the notion of positive health (3). ease, the model requires an index to quantify health status that combines mortality with esti- Description mates of the quality of life among survivors. Originally called the Health Status Index (1), the This is the purpose of the QWB, which scale was renamed the Index of Well-Being (12) and later the Quality of Well-Being Scale to quantifies the health output of any treatment stress its focus on quality of life (9). Various in terms of the years of life, adjusted for their modifications have been made to the QWB over diminished quality, that it produces or saves. the years; the 1994 version is described here. Thus, a “Well-Year” can be defined concep- The procedure for classifying an individual tually as the equivalent of a year of com- may be described in three stages, corresponding pletely well life, . . . A disease that reduces to the three components of the model just de- the health-related quality of life by one-half, scribed. for example, will take away .500 Well-Years over the course of one year. If it affects two assessing functional status. A structured inter- people, it will take away 1.0 Well-Year view is used to record symptoms and medical (=2 × .500). . . . Dividing the cost of a pro- problems experienced on each of the previous 676 Measuring Health eight days, and to classify the respondent’s level 10.33, there are three categories on the mobility of functioning. The interview schedule is repro- and physical activity scales and five on the social duced in a book by Patrick and Erickson (13, activity scale. Note that this is an abbreviation of pp392–404). The questions were derived from the earlier version of the QWB described in the the Health Interview Survey and from the U.S. first edition of this book. The social activity Social Security Administration Survey of the scale retains five categories as these are needed to Disabled. The interview is structured so that derive QWB scores from other survey instru- screening questions lead to more detailed investi- ments. The respondent is placed into one level of gation of problems that are identified. According each scale, giving 3 × 3 × 5, or 45 possible combi- to the respondent’s level of health, the current nations plus death, making 46 function levels. abbreviated interview takes seven minutes or The interview also records the presence of longer; the earlier version took 10 to 30 minutes symptoms or problem complexes, or “CPX.” (14; 15, p963). The questions on function cover Note that these refer to problems on the previ- performance rather than capacity; self-reports ous day, so the past tense is used in the ques- are used rather than observation. Questions tions. By recording symptoms, even where these cover three dimensions of functioning: mobility are not sufficient to restrict activity levels, the and confinement (e.g., in the hospital or institu- QWB is sensitive to minor deviations from com- tion); physical activity, especially ambulation; plete well-being. There are currently 27 symp- and social activity, which includes work, house- tom or problem complexes, as shown in Exhibit keeping, and self-care. As shown in Exhibit 10.34 (Ex 10.34), increased from 23 in previous Exhibit 10.33 Dimensions, Function Levels, and Weights of the Quality of Well-Being Scale Step Step definition Weight Mobility Scale (MOB) 5 No limitations for health reasons −.000 4 Did not drive a car, health related; did not ride in a car as usual for age (younger −.062 than 15 yr), health related, and/or did not use public transportation, health related; or had or would have used more help than usual for age to use public transportation, health related 2 In hospital, health related −.090 Physical Activity Scale (PAC) 4 No limitations for health reasons −.000 3 In wheelchair, moved or controlled movement of wheelchair without help from −.060 someone else; or had trouble or did not try to lift, stoop, bend over, or use stairs or inclines, health related; and/or had any other physical limitation in walking, or did not try to walk as far as or as fast as others the same age are able, health related 1 In wheelchair, did not move or control the movement of wheelchair without help −.077 from someone else, or in bed, chair, or couch for most or all of the day, health related Social Activity Scale (SAC) 5 No limitations for health reasons −.000 4 Limited in other (e.g. recreational) role activity, health related −.061 3 Limited in major (primary) role activity, health related −.061 2 Performed no major role activity, health related, but did perform self-care activities −.061 1 Performed no major role activity, health related, and did not perform or had more −.106 help than usual in performance of one or more self-care activities, health related Reproduced from an original supplied by Dr. Kaplan. With permission.  General Health Status and Quality of Life 677 versions of the QWB by the addition of four social activity weight − CPX weight. Weights are mental health symptoms. The interview takes negative, indicating the reduction in well-being about 18 minutes, slightly less than the Sickness implied by each functional category or CPX. Impact Profile (SIP) (16, Table 2). The CPX weights range from −0.0 (no prob- Copies of the interview schedule, too long to lems) to −0.727 (death), as shown in Exhibit reproduce here, and a 30-page interviewer man- 10.34. Readers should be aware that there are in- ual can be ordered at cost from Dr. Kaplan. An consistencies in the weights quoted in different interview format has generally been used be- articles describing the QWB; the weights listed cause early tests showed that a self-administered in Exhibit 10.34 are correct. Where multiple version identified only 45% of disabilities com- problems exist, the one considered most undesir- pared with 89% for the interviewer version (17, able is scored (21); where a problem complex is p466). Kappa coefficients of agreement between present but is unknown, a value of −0.257 is the two versions ranged from 0.48 to 0.58 for used. A person who has died receives a CPX the three sections (18, p134). Anderson et al. score of −.727, and the lowest score on the three concluded that the additional cost of the inter- functional scales, giving an overall score of zero. viewer version is warranted (18, p134). More re- W represents a person’s well-being at a point cently, however, further attention has been paid in time. To indicate QALYs, W is multiplied by to developing a self-administered version, the the time spent in that state of health; this calcu- QWB-SA, which takes about ten to 14 minutes lation is repeated and the results summed as the to complete (8, p511; 19; 20). person’s symptoms or functional level change over the time period being considered. Kaplan scaling the responses. Preference weights for gives the example of a person who is well for each function level were derived originally from 65.2 years, then experiences disability for 4.5 an equal-appearing interval scaling task involv- years (W = 0.59), followed by bed-ridden disabil- ing 867 raters (12). The weights reflect social ity for 1.9 years (W = 0.34). Although the person preferences or judgments of the relative impor- has actually lived a total of 71.6 years, the QWB tance that members of society associate with indicates that this is equivalent to each function level (12; 21); they do not consider (65.2 × 1.0 + 4.5 × 0.59 + 1.9 × 0.34), or 68.5 the patient’s diagnosis. Bush et al. emphasized years of healthy life (3, p41). This calculation that preference ratings should be collected em- can be extended to groups of patients and used pirically rather than be based on the assump- as a program output indicator (12, p493; 10, tions or clinical experience of the researchers Table 1). (22). Category weighting was considered more appropriate than a magnitude estimation ap- indicating prognosis. W can be adjusted to re- proach that did not yield a true ratio scale and flect prognoses. The prognostic adjustment re- gave counterintuitive values (21); further valida- flects an estimate of the time spent at each level tion of the scaling procedure has been reported of disability for patients with a given medical by Kaplan and Ernst (23). Similarly, weights condition, based on empirical studies of pa- were derived for the symptom and problem com- tients’ progress. These durations are expressed plexes, as described in various articles (21; in terms of the probabilities of future transitions 24–26). to worse or better levels of function and well- The overall QWB score ranges from one being within a fixed time (1; 21). The “Well-Life (complete well-being) to zero (death), although Expectancy” (E) weights W by the expected du- the scoring does permit values below zero, to ration of stay in each functional level, summed represent states worse than death, such as a pro- across levels of disability (12, p484). This gives longed vegetative existence (9). Using the an estimate only, adequate for use with groups weights shown in Exhibits 10.33 and 10.34, the of patients rather than with individual patients. QWB score, known as W, is calculated as: W= At present, transition weights are only available 1− mobility weight − physical activity weight − for certain diseases or conditions, such as the 678 Measuring Health Exhibit 10.34 Symptom and Problem Complexes (CPX) for the Quality of Well-Being Scale CPX No. CPX description Weight 1 Death [not on respondent’s card] −.727 2 Loss of consciousness such as seizure (fits), fainting, or coma (out cold or knocked out) −.407 3 Burn over large areas of face, body, arms, or legs −.387 4 Pain, bleeding, itching, or discharge (drainage) from sexual organs—does not include −.349 normal menstrual bleeding 5 Trouble learning, remembering, or thinking clearly −.340 6 Any combination of one or more hands, feet, arms, or legs either missing, deformed −.333 (crooked), paralyzed (unable to move), or broken—includes wearing artificial limbs or braces 7 Pain, stiffness, weakness, numbness, or other discomfort in chest, stomach (including −.299 hernia or rupture), side, neck, back, hips, or any joints or hands, feet, arms, or legs 8 Pain, burning, bleeding, itching, or other difficulty with rectum, bowel movements, or −.292 urination (passing water) 9 Sick or upset stomach, vomiting or loose bowel movement, with or without chills, or −.290 aching all over 10 General tiredness, weakness, or weight loss −.259 11 Cough, wheezing or shortness of breath, with or without fever, chills, or aching all over −.257 12 Spells of feeling upset, being depressed, or of crying −.257 13 Headache, or dizziness, or ringing in ears, or spells of feeling hot, nervous or shaky −.244 14 Burning or itching rash on large areas of face, body, arms, or legs −.240 15 Trouble talking, such as lisp, stuttering, hoarseness, or being unable to speak −.237 16 Pain or discomfort in one or both eyes (such as burning or itching) or any trouble seeing −.230 after correction 17 Overweight for age and height or skin defect of face, body, arms, or legs, such as scars, −.188 pimples, warts, bruises or changes in color 18 Pain in ear, tooth, jaw, throat, lips, tongue; several missing or crooked permanent −.170 teeth—includes wearing bridges or false teeth 19 Took medication or stayed on a prescribed diet for health reasons −.144 20 Wore eyeglasses or contact lenses −.101 21 Breathing smog or unpleasant air −.101 22 No symptoms or problems [not on respondent’s card] −.000 23 Standard symptom/problem −.257 24 Trouble sleeping −.257 25 Intoxication −.257 26 Problems with sexual interest or performance −.257 27 Excessive worry or anxiety −.257 Reproduced from an original supplied by Dr. Kaplan. With permission. likelihood of developing mental retardation fol- liability coefficient obtained when judges re- lowing various forms of phenylketonuria (27). assessed scale values for the function levels was 0.90 (9; 25). The preference weights obtained Reliability from different judges at different times showed Considerable attention has been paid to the sta- small, although systematic, variations (14). A bility of estimates of preference weights. The re- replication of the original preference rating exer-  General Health Status and Quality of Life 679 cise, using 288 patients with rheumatoid arthri- demonstrated that the QWB is capable of show- tis, again indicated that raters achieved high lev- ing significant treatment effects in a study of pa- els of consistency in their ratings, and agreed tients with chronic obstructive pulmonary closely with the original weights. QWB scores disease (32), in a trial of zidovudine (AZT) for for 132 patient scenarios based on the original AIDS (33), and in a trial of dietary and exercise weights correlated 0.94 with scores using the interventions for patients with diabetes (34). In a weights provided by the patients with arthritis trial of auranofin, the QWB proved a slightly (28, p978). more sensitive outcome indicator than the physi- Kaplan and Bush noted that in applying the cal and pain scores of the Arthritis Impact Mea- scale in assessing over 50,000 person-days, the surement Scales (AIMS) (33, Table 6; 35, Table classification accuracy exceeded 96% (9, p70). 7). Correlations between the QWB and the Stability of the W statistic was studied by cor- AIMS were −0.57 for the AIMS physical score, relating the rating of the first day with the −0.40 for the pain score, and −0.17 for the mean of ratings made on eight subsequent AIMS psychological score (33, Table 4). Liang et days. Correlations in excess of 0.93 were ob- al. compared the sensitivity of five health mea- tained (14). A summary of stability coefficients sures to change and found that the QWB was for QWB scores on consecutive days cited 0.96 ranked in the middle of the five for most com- for a general population sample, 0.93 for parisons (36, Table 2). healthy children, 0.90 for a mixed sample of The QWB has been evaluated in several stud- patients, and 0.96 for diabetic patients (8, ies of respiratory conditions. For 44 patients Table 1). with cystic fibrosis, QWB scores correlated sig- nificantly with pulmonary function tests: 0.55 with forced expiratory volume in one second Validity (FEV1) and 0.58 with peak oxygen consumption Arguments for the content validity of the QWB (Vo2max) (33, pS38; 37, Table 3). In a study of note its broad scope and that it is the only in- patients with chronic obstructive pulmonary dis- strument to consider mortality, symptoms, and ease, QWB scores correlated 0.51 with FEV1 and problems as well as functional levels—all of 0.41 with a treadmill test of exercise tolerance which are central components of the concept of (32, Tables 4, 5). In the study by Read et al., health (12; 29). In a qualitative study, the QWB QWB scores correlated 0.32 with time achieved was judged to have omitted several aspects of on treadmill and 0.32 with FEV1; the variance in quality of life noted as being important to pa- QWB scores explained (R2) was 0.35 (16, p15S). tients with Alzheimer’s disease, but had a more These correlations were lower than equivalent comprehensive coverage than the EuroQol EQ- results for the Sickness Impact Profile, for which 5D or the Health Utilities Index (30). the R2 was 0.62. Kaplan et al. reported correlations of −0.75 The QWB correlated 0.46 with a modified between QWB scores and the number of re- version of Jette’s Functional Status Index; omit- ported symptoms, and of −0.96 between the ting the QWB symptom component raised this QWB and the number of chronic health prob- association to 0.57. This was explained because lems (12, pp497–498). The correlation with the the symptom component appeared to capture number of physician contacts in the preceding variation towards the healthy end of the scale eight days was −0.55. Read et al. obtained a that was missed by Jette’s scale (15, lower correlation of −0.48 with number of self- pp962–963). The QWB correlated −0.55 with reported symptoms (16, p15S). QWB scores the SIP and 0.57 with a ten-point rating of vary significantly across New York Heart Asso- health over the previous six days (16, p14S). In a ciation functional classes (31). study of patients with prostate disease, correla- Because the functional scales of the QWB of- tions with the subscales of the Short-Form 36 in- fer relatively coarse indicators, sensitivity to strument ranged from 0.67 (SF-36 physical change has been widely reviewed. It has been function) to 0.58 (SF-36 role performance and 680 Measuring Health vitality scales), to 0.55 (mental health) and 0.50 relationship between social class and functional (bodily pain). The general health perceptions status (39). correlation was the lowest, at 0.28 (38, Figure A version of the QWB for children has shown 1). Kaplan et al. summarized four more studies good validity results (4; 40). A self-administered that presented correlations between the QWB version of the QWB provides both an overall in- and the SF-36 subscales. The physical function dex score and section scores in the manner of the scale of the SF-36 generally correlates most health profile measures and also includes more strongly (range, 0.51–0.69), followed by the categories of symptoms and problemsm (see Ad- role-physical scale (range, 0.28–0.64) and the dress section). general health perception scale (0.37 to 0.53) (8, Table 3). Finally, in a sample of elderly ambula- Reference Standards tory patients, the QWB correlated −0.45 with Some data on QWB norms for the general U.S. the SIP, whereas correlations with the SF-36 population are presented by Erickson et al. (41); summary scales were 0.47 with the physical, and unfortunately they only cover three broad age 0.22 with the mental scale (19, Table 4). categories. Mean scores for three categories of Given that the QWB does not contain items patient derived from the U.S. National Health on psychological functioning, its sensitivity to Interview Survey were reported by Kaplan et al. psychological distress is of interest. Correlations (8, Figure 1). with the Center for Epidemiologic Studies De- pression Scale were −0.31 (35, Table 5) and a Commentary correlation of 0.33 with a mental health scale The QWB was one of the earliest of the broad- has been reported (16, p15S). Interestingly, spectrum numerical indexes of health and the changes in QWB scores before and after treat- first to confront the conceptual and methodolog- ment for arthritis correlated more highly with ical issues of combining length and quality of life changes in the AIMS psychological score than in in a single score. The metric of the QWB, evalu- the AIMS physical score (35, Table 6), suggest- ating an intervention in terms of the well-year ing that as an outcome measure, the QWB does equivalents gained, is powerful and seems valu- identify changes in emotional well-being. Per- able in guiding decision-making by patients (42). haps because of its emphasis on role function- The QWB has also exerted a major influence on ing, the QWB correlated 0.40 with employment the design of other measures. It has been used in status (16, p15S). numerous cost-effectiveness studies (1; 4; 11; 27; 43) and the Kaplan and Bush summary ta- Alternative Forms bles comparing the cost-effectiveness of screen- Reynolds et al. have suggested a simplified scor- ing programs per well-year gained set an ing procedure for their version of the index example that has since been followed in other called the Function Status Index (FSI) (29). They fields (9). The prognostic dimension remains in- also recorded behavior only on the previous day novative, allowing the QWB to be applied rather than over the previous eight days and then equally to acute and chronic disease states, separated the social activity scale and self-care whereas other scales cannot adequately compare scales. From a sample of 8,036 respondents in the two because of their different implications Alabama, a correlation of −0.61 was obtained for the future need for care. The task of deriving between the FSI and the number of chronic the required transition probabilities for all dis- health problems reported (29, Table 2). A gamma eases is daunting (44), although it may be feasi- coefficient of −0.53 was obtained with the num- ble to estimate these for particular conditions; ber of physician contacts (29, Table 3). Reynolds for example, the progression of physical medi- et al. also reported a correlation of −0.48 with a cine patients is being described by groups work- health worry scale (29, pp279, 281). Harkey et ing with the Functional Independence Measure al. applied the FSI in a study (N = 16,569) of the (45).  General Health Status and Quality of Life 681 The QWB is also distinctive in its inclusion of estimation (48). They argued that anchoring symptoms; this seems to improve its sensitivity the scale at zero, for death, caused a distortion to minor deviations from complete well-being. It and precluded measurement of states worse may therefore be ideal for evaluating policies in than death. They proposed an alternative scal- healthy populations with lower levels of morbid- ing procedure, although this correlated highly ity. Used as a survey instrument, the QWB classi- (−0.81) with the values produced by Kaplan et fies a higher proportion of the population as sick al. (48, Table 8). than some other measures do. For example, Er- Other commentaries on the QWB have noted ickson et al. compared QWB scores with esti- the relative complexity of interviewer training mates of population health status based on an that requires one to two weeks, although after indicator of activity limitation, showing that training is complete, the scale does not appear only about 5% of the population aged between unduly difficult to administer (16). The new self- 45 and 64 years had no functional limitation on administered version will overcome this criti- the QWB, compared with 75% when activity cism. limitation was used as the criterion (41, Figure As economic constraints on medical care in- 1). They argued that activity limitation measures crease, indexes such as the QWB will take an in- overestimate health status; the reverse argument creasingly important place in evaluation of is that the QWB may be identifying trivial devia- patient care. Of the scales that provide estimates tions from well-being. of QALYs, the QWB is the most widely used. Its Criticisms of any health measurement focus advantages lie in its clear conceptual approach, on two concerns: the items included and the attention to scaling, and widespread use. scoring method. The coverage of the QWB is ori- ented strongly toward physical problems: the Address mobility dimension concentrates on ability to Interview Schedule and Manual: R.M. Kaplan, get around or use transport; the social activity PhD, Professor and Chief, Division of Health dimension taps role functioning and self-care; Care Sciences 0622, School of Medicine, Univer- and few of the symptom and problem complexes sity of California, San Diego, La Jolla, Califor- are explicitly oriented toward emotional dis- nia, USA 92093-0622. tress. The criticism has therefore been made that Copies of the QWB, including a 300-page the QWB underrepresents mental health (46, manual and administration instructions, are avail- p94). Kaplan, however, responds that the divi- able at cost from the Medical Outcomes Trust, sion into mental and physical health is artificial, www.outcomes-trust.org/instruments.htm. since the two affect each other (3, p45; 4). The significant correlations with measures of mental well-being seem to bear this out. References Criticisms of the QWB scoring focus on the (1) Fanshel S, Bush JW. A Health-Status Index scale weights. Concern was raised over appar- and its application to health-services ent anomalies, whereby some more disabled outcomes. Oper Res 1970;18:1021–1065. states appear preferable to some less disabled (2) Chen MM, Bush JW, Patrick DL. Social ones (47). Bush et al. responded to this criti- indicators for health planning and policy cism, pointing out that desirability may differ analysis. Policy Sci 1975;6:71–89. (3) Kaplan RM. New health promotion between acute and chronic conditions. In acute indicators: the General Health Policy conditions, for example, “being in a wheel- Model. Health Promot 1988;3:35–49. chair is sometimes more comfortable (and (4) Kaplan RM, Anderson JP. A General therefore more desirable) than struggling to Health Policy Model: update and walk with limitations (e.g., with crutches)” applications. Health Serv Res (22). Haig et al. criticized the use of category 1988;23:203–235. rating to derive weights, rather than magnitude (5) Kaplan RM, Anderson JP, Ganiats TG. The 682 Measuring Health Quality of Well-Being Scale: rationale for a (17) Anderson JP, Bush JW, Berry CC. single quality of life index. In: Walker SR, Classifying function for health outcome Rosser RM, eds. Quality of life assessment: and quality-of-life evaluation: self-versus- key issues in the 1990s. Dordrecht, interviewer modes. Med Care Netherlands: Kluwer Academic Publishers, 1986;24:454–470. 1993:65–94. (18) Anderson JP, Bush JW, Berry CC. Internal (6) Kaplan RM. Application of a General consistency analysis: a method for studying Health Policy Model in the American the accuracy of function assessment for health care crisis. J R Soc Med health outcome and quality of life 1993;86:277–281. evaluation. J Clin Epidemiol (7) Kaplan RM, Anderson JP. The general 1988;41:127–137. health policy model: an integrated (19) Andresen EM, Rothenberg BM, Kaplan approach. In: Spilker B, ed. Quality of life RM. Performance of a self-administered and pharmacoeconomics in clinical trials. mailed version of the Quality of Well-Being Philadelphia: Lippincott-Raven, (QWB-SA) questionnaire among older 1996:309–322. adults. Med Care 1998;36:1349–1360. (8) Kaplan RM, Ganiats TG, Sieber WJ, et al. (20) Kaplan RM, Sieber WJ, Ganiats TG. The The Quality of Well-Being Scale: critical Quality of Well-Being Scale: comparison of similarities and differences with SF-36. Int the interviewer-administered version with a J Qual Health Care 1998;10:509–520. self-administered questionnaire. Psychol (9) Kaplan RM, Bush JW. Health-related Health 1997;12:783–791. quality of life measurement for evaluation (21) Kaplan RM, Bush JW, Berry CC. Health research and policy analysis. Health Status Index: category rating versus Psychol 1982;1:61–80. magnitude estimation for measuring levels (10) Kaplan RM. Health-related quality of life of well-being. Med Care with applications in nutrition research and 1979;17:501–523. practice. Clin Nutr 1988;7:64–70. (22) Bush JW, Anderson JP, Kaplan RM, et al. (11) Bush JW, Fanshel S, Chen MM. Analysis “Counterintuitive” preferences in health- of a tuberculin testing program using a related quality-of-life measurement. Med Health Status Index. Socioecon Plann Sci Care 1982;20:516–525. 1972;6:49–68. (23) Kaplan RM, Ernst JA. Do category rating (12) Kaplan RM, Bush JW, Berry CC. Health scales produce biased preference weights status: types of validity and the Index of for a health index? Med Care Well-Being. Health Serv Res 1983;21:193–207. 1976;11:478–507. (24) Patrick DL, Bush JW, Chen MM. Toward (13) Patrick DL, Erickson P. Health status and an operational definition of health. J health policy: quality of life in health care Health Soc Behav 1973;14:6–23. evaluation and resource allocation. New (25) Patrick DL, Bush JW, Chen MM. Methods York: Oxford University Press, 1993. for measuring levels of well-being for a (14) Kaplan RM, Bush JW, Berry CC. The Health Status Index. Health Serv Res reliability, stability, and generalizability of 1973;8:228–245. a Health Status Index. Proceedings of the (26) Blischke WR, Bush JW, Kaplan RM. Social Statistics Section, American Successive intervals analysis of preference Statistical Association, 1978. measures in a Health Status Index. Health (15) Ganiats TG, Palinkas LA, Kaplan RM. Serv Res 1975;10:181–198. Comparison of Quality of Well-Being Scale (27) Bush JW, Chen MM, Patrick DL. Health and Functional Status Index in patients Status Index in cost effectiveness: analysis with atrial fibrillation. Med Care of PKU program. In: Berg RL, ed. Health 1992;30:958–964. status indexes. Chicago: Hospital Research (16) Read JL, Quinn RJ, Hoefer MA. Measur- and Educational Trust, 1973:172–209. ing overall health: an evaluation of three (28) Balaban DJ, Sagi PC, Goldfarb NI, et al. important approaches. J Chronic Dis Weights for scoring the Quality of Well- 1987;40(suppl 1):S7–S21. Being instrument among rheumatoid  General Health Status and Quality of Life 683 arthritics: a comparison to general (40) Bradlyn AS, Harris CV, Warner JE, et al. population weights. Med Care An investigation of the validity of the 1986;24:973–980. quality of Well-Being Scale with pediatric (29) Reynolds WJ, Rushing WA, Miles DL. The oncology patients. Health Psychol validation of a Function Status Index. J 1993;12:246–250. Health Soc Behav 1974;15:271–288. (41) Erickson P, Kendall EA, Anderson JP, et al. (30) Silberfeld M, Rueda S, Krahn M, et al. Using composite health status measures to Content validity for dementia of three assess the nation’s health. Med Care generic preference based health related 1989;27(suppl):S66–S76. quality of life instruments. Qual Life Res (42) Kaplan RM, Atkins CJ. The well-year of 2002;11:71–79. life as a basis for patient decision-making. (31) Ganiats TG, Browner DK, Dittrich HC. Patient Educ Couns 1989;13:281–295. Comparison of Quality of Well-being scale (43) Weinstein MC, Stason WB. Foundations of and NYHA functional status classification cost-effectiveness analysis for health and in patients with atrial fibrillation. New medical practices. N Engl J Med York Heart Association. Am Heart J 1977;296:716–721. 1998;135:819–824. (44) Chen MM, Bush JW, Zaremba J. (32) Kaplan RM, Atkins CJ, Timms R. Validity Effectiveness measures. In: Shuman L, of a Quality of Well-Being Scale as an Speas R, Young J, eds. Operations research outcome measure in chronic obstructive in health care—a critical analysis. pulmonary disease. J Chronic Dis Baltimore: Johns Hopkins University Press, 1984;37:85–95. 1975:276–301. (33) Kaplan RM, Anderson JP, Wu AW, et al. (45) Long WB, Sacco WJ, Coombes SS, et al. The Quality of Well-Being Scale: Determining normative standards for applications in AIDS, cystic fibrosis, and Functional Independence Measure arthritis. Med Care transitions in rehabilitation. Arch Phys 1989;27(suppl):S27–S43. Med Rehabil 1994;75:144–148. (34) Kaplan RM, Hartwell SL, Wilson DK, et (46) Hays RD, Shapiro MF. An overview of al. Effects of diet and exercise interventions generic health-related quality of life mea- on control and quality of life in non- sures for HIV research. Qual Life Res insulin-dependent diabetes mellitus. J Gen 1992;1:91–97. Intern Med 1987;2:220–228. (47) Anderson GM. A comment on the Index of (35) Kaplan RM, Kozin F, Anderson JP. Mea- Well-Being. Med Care 1982;20:513–515. suring quality of life in arthritis patients (48) Haig THB, Scott DA, Wickett LI. The (including discussion of a general health- rational zero point for an illness index with decision model). Qual Life Cardiovasc ratio properties. Med Care Care 1988;4:131–139. 1986;24:113–124. (36) Liang MH, Fossel AH, Larson MG. Comparisons of five health status The Health Utilities Index instruments for orthopedic evaluation. Med (George W. Torrance, David Feeny; Care 1990;28:632–642. (37) Orenstein DM, Nixon PA, Ross EA, et al. 1986, 1990) The quality of well-being in cystic fibrosis. Purpose Chest 1989;95:344–347. The Health Utilities Index (HUI) system is a (38) Frosch D, Porzsolt F, Heicappell R, et al. generic preference-based measure of health sta- Comparison of German language versions tus intended for use in clinical outcomes mea- of the QWB-SA and SF-36 evaluating outcomes for patients with prostate disease. surement, in health surveys, in program Qual Life Res 2001;10:165–173. planning, and in resource allocation. The utility (39) Harkey J, Miles DL, Rushing WA. The scores provide a summary index of health- relation between social class and functional related quality of life on a zero to 1.0 scale; they status: a new look at the drift hypothesis. J may be used in cost-utility analyses to calculate Health Soc Behav 1976;17:194–204. quality-adjusted life years and in population 684 Measuring Health studies as quality weights for calculating quality- not logically constrain his level on another, so adjusted life expectancy (1, p250). that any combination of levels across the attrib- utes should be biologically possible (3). This also Conceptual Basis suggests little overlap in concepts across the at- The HUI separates the classification of a person’s tributes. The desire to achieve structural inde- health state from the application of preference pendence influenced the choice of attributes to weights to that state. In classifying health states, be measured; working ability, for example, was the HUI records functional capacity rather than not covered because it would depend on a cer- performance. “The intent is to document the ex- tain level of physical function. The HUI3 system tent to which deficits in health status . . . inhibit achieved full structural independence, whereas or prohibit normal functioning, rather than to HUI2 has partial independence. For instance, report the level at which an individual chooses to level 5 of mobility (no use of arms or legs) pre- function” (1, p241). Feeny et al. argued that per- cludes being in level 1 of self-care. formance measures confound three concepts: the person’s health or physical capacity, her oppor- Description tunities and choices, and her preferences. “Thus, The HUI has evolved over time; a summary is people with the same underlying capacity who given by Patrick and Erickson (5, pp381–389). face different opportunity sets or have different The original version (HUI Mark 1) was devel- preferences may be assessed as having different oped to assess the health of children who had health status in a system that relies on perfor- been in neonatal intensive care units; it was later mance . . .” (1, p243; 2, p496). extended for more general pediatric applications The HUI covers several aspects of health, (6). The Mark 1 version is now seldom used. which are termed ‘attributes’ rather than ‘dimen- The Mark 2 was developed to measure treatment sions’ so as to emphasize the link with multiat- outcomes for childhood cancer (7). It assessed tribute utility theory (3). However, in keeping seven attributes: sensation (e.g., vision, hearing, with the focus on capacity, the coverage is inten- speech), mobility, emotion, cognition, self-care, tionally restricted to “within the skin” abilities pain, and fertility. Three to five levels of function (physical and emotional) and thus excludes were defined for each attribute. The most recent questions about role performance or social inter- version, Mark 3, extends the range of the Mark action. This corresponds to a focus on health- 2 and alters the attributes included. In part this related quality of life, rather than quality of life was due to an overlap between the self-care and in general (1, pp240–241; 2, p491; 3, p376). other dimensions, which implied redundancy Feeny et al. argue that although it is important to and complicated the scoring system. The Mark 3 measure social interaction as it reflects health covers eight attributes: vision, hearing, speech, status, the two should be measured separately so ambulation, dexterity, emotion, cognition, and that the relationships can be analysed in a non- pain—each with five or six levels, described in tautological fashion (2, pp495–496). Similarly, Exhibit 10.35. The Mark 2 version is still used the HUI omits positive aspects of emotional and may be the more appropriate instrument well-being, such as mastery or the sense of co- when it is necessary to assess self-care or when herence (4). fertility is important. The Mark 2 also covers the The scaling system used to estimate utilities type of analgesic required for pain relief, also influenced the content of the HUI. Within whereas the Mark 3 covers activity disruption each attribute, several levels of function are spec- due to pain (3, p378). The HUI2 emotion scale ified, making possible a large number of permu- focuses on worry and anxiety; the HUI3 focuses tations of levels across the attributes. The task of on happiness versus depression. The standard estimating utilities for large numbers of states is questionnaires distributed by HUI include ques- simplified if the health status classifications are tions to classify people in both the HUI2 and 3 chosen to be structurally independent. This systems (see Address section). means that a person’s level on one attribute does The HUI3 classification system is shown in Exhibit 10.35 The HUI3 Classification System Note: The questions from which the classification is derived are shown on the HUI Web site. Attribute Level Description Vision 1 Able to see well enough to read ordinary newsprint and recognize a friend on the other side of the street, without glasses or contact lenses. 2 Able to see well enough to read ordinary newsprint and recognize a friend on the other side of the street, but with glasses. 3 Able to read ordinary newsprint with or without glasses but unable to recognize a friend on the other side of the street, even with glasses. 4 Able to recognize a friend on the other side of the street with or without glasses but unable to read ordinary newsprint, even with glasses. 5 Unable to read ordinary newsprint and unable to recognize a friend on the other side of the street, even with glasses. 6 Unable to see at all. Hearing 1 Able to hear what is said in a group conversation with at least three other people, without a hearing aid. 2 Able to hear what is said in a conversation with one other person in a quiet room without a hearing aid, but requires a hearing aid to hear what is said in a group conversation with at least three other people. 3 Able to hear what is said in a conversation with one other person in a quiet room with a hearing aid, and able to hear what is said in a group conversation with at least three other people, with a hearing aid. 4 Able to hear what is said in a conversation with one other person in a quiet room, without a hearing aid, but unable to hear what is said in a group conversation with at least three other people even with a hearing aid. 5 Able to hear what is said in a conversation with one other person in a quiet room with a hearing aid, but unable to hear what is said in a group conversation with at least three other people even with a hearing aid. 6 Unable to hear at all. Speech 1 Able to be understood completely when speaking with strangers or friends. 2 Able to be understood partially when speaking with strangers but able to be understood completely when speaking with people who know me well. 3 Able to be understood partially when speaking with strangers or people who know me well. 4 Unable to be understood when speaking with strangers but able to be understood partially by people who know me well. 5 Unable to be understood when speaking to other people (or unable to speak at all). Ambulation 1 Able to walk around the neighbourhood without difficulty, and without walking equipment. 2 Able to walk around the neighbourhood with difficulty; but does not require walking equipment or the help of another person. 3 Able to walk around the neighbourhood with walking equipment, but without the help of another person. 4 Able to walk only short distances with walking equipment, and requires a wheelchair to get around the neighbourhood. 5 Unable to walk alone, even with walking equipment. Able to walk short distances with the help of another person, and requires a wheelchair to get around the neighbourhood. 6 Cannot walk at all. (continued) 685 686 Measuring Health Exhibit 10.35 (continued) Dexterity 1 Full use of two hands and ten fingers. 2 Limitations in the use of hands or fingers, but does not require special tools or help of another person. 3 Limitations in the use of hands or fingers, is independent with use of special tools (does not require the help of another person). 4 Limitations in the use of hands or fingers, requires the help of another person for some tasks (not independent even with use of special tools). 5 Limitations in use of hands or fingers, requires the help of another person for most tasks (not independent even with use of special tools). 6 Limitations in use of hands or fingers, requires the help of another person for all tasks (not independent even with use of special tools). Emotion 1 Happy and interested in life. 2 Somewhat happy. 3 Somewhat unhappy. 4 Very unhappy. 5 So unhappy that life is not worthwhile. Cognition 1 Able to remember most things, think clearly and solve day to day problems. 2 Able to remember most things, but have a little difficulty when trying to think and solve day to day problems. 3 Somewhat forgetful, but able to think clearly and solve day to day problems. 4 Somewhat forgetful, and have a little difficulty when trying to think or solve day to day problems. 5 Very forgetful, and have great difficulty when trying to think or solve day to day problems. 6 Unable to remember anything at all, and unable to think or solve day to day problems. Pain 1 Free of pain and discomfort. 2 Mild to moderate pain that prevents no activities. 3 Moderate pain that prevents a few activities. 4 Moderate to severe pain that prevents some activities. 5 Severe pain that prevents most activities. Exhibit 10.35. Unlike other health measures, from the person’s usual state; for use as an out- there is no single questionnaire, but (as with a come measurement, current status is typically as- clinician rating scales), various approaches may sessed. Different recall periods may also be used be used to classify respondents into the func- (e.g., one, two, four weeks) according to the pur- tional levels shown in the Table. Alternative pose. Most versions include 31 questions, but questionnaires are available from the HUI Web skip patterns reduce this number in an actual in- site (www.fhs.mcmaster.ca/hug/index.htm). There terview and administration takes an average of is flexibility in the format of the questions and two minutes for a community population, rang- different versions cover self-administration or by ing from 1.5 for people with no health problems an interviewer; a proxy version is available for to five minutes or more for people with multiple patients unable to respond themselves; different problems (8; 9). versions are available for adults and for children. Each different combination of levels from For health surveys, questions may cover changes each attribute forms a unique health state. Util-  General Health Status and Quality of Life 687 Exhibit 10.36 HUI3 Single Attribute Utility Functions Level Vision Hearing Speech Ambulation Dexterity Emotion Cognition Pain 1 1 1 1 1 1 1 1 1 2 0.95 0.86 0.82 0.83 0.88 0.91 0.86 0.92 3 0.73 0.71 0.67 0.67 0.73 0.73 0.92 0.77 4 0.59 0.48 0.41 0.36 0.45 0.33 0.70 0.48 5 0.38 0.32 0 0.16 0.20 0 0.32 0 6 0 0 0 0 0 ity weights for these were derived in two main health states (11, p714). This produced the eight stages. First, utility weights were estimated for single-attribute utility functions shown in Ex- each level of each attribute (shown in Exhibit hibit 10.36. These are scaled on a 0.0 to 1.0 10.36) and second, a scoring function translates scale, which does not reflect the differing relative the resulting profile of weights into an overall in- weights of each attribute in forming a multiat- dex score, as outlined in Exhibit 10.37. tribute health profile. These weights are only The attribute weights were generated in a used when comparing single attributes (e.g., preference scaling task employing a general pop- pain levels before and after treatment). ulation sample of 504 adults from Hamilton, Because of the large number of possible per- Ontario (10). For simplicity, most of the weight- mutations in the HUI, it was not feasible to score ing used visual analog are scales (VAS) (termed a each multiattribute health state directly. Instead, “feeling thermometer”), and the resulting value a subset of the possible states was scored using a scores were converted to utilities using a power standard gamble scaling procedure to generate function (1; 10). The power function was de- weights, and then weights for the remaining per- rived from a comparison between VAS and util- mutations were estimated using a multiattribute ity scores derived from a separate standard scaling analysis. Multiattribute utility theory ex- gamble scaling task undertaken for a selection of tends the von Neumann-Morgenstern utility the- Exhibit 10.37 HUI3 Multi-Attribute Utility Function on Dead-to-Healthy Scale Vision Hearing Speech Ambulation Dexterity Emotion Cognition Pain Level b1 b2 b3 b4 b5 b6 b7 b8 1 1 1 1 1 1 1 1 1 2 0.98 0.95 0.94 0.93 0.95 0.95 0.92 0.96 3 0.89 0.89 0.89 0.86 0.88 0.85 0.95 0.90 4 0.84 0.80 0.81 0.73 0.76 0.64 0.83 0.77 5 0.75 0.74 0.68 0.65 0.65 0.46 0.60 0.55 6 0.61 0.61 0.58 0.56 0.42 The multi-attribute utilities are assumed to refer to a chronic condition that will last a lifetime, judged on a scale running from dead (value 0.0) to healthy (value 1.0). To calculate the multi-attribute utility, take the value in Exhibit 10.37 for each attribute (b1 through b8) that corresponds to the person’s level on that attribute. The formula is: Utility = 1.371 (b1 * b2 * b3 * b4 * b5 * b6 * b7 * b8) − 0.371 As an example, for someone with a level 2 hearing rating, and a level 3 ambulation problem and level 3 pain, but otherwise healthy: Utility = 1.371 (0.95 * 0.86 * 0.90) − 0.371 = 0.637 688 Measuring Health ory to cover the utility weighting of combina- ranged from 0.7 to 1.0 for the attribute scores of tions of states. Based on empirical analyses, the Mark 2 instrument (15, Table 3). In a three- Feeny et al. used the multiplicative form, which day retest study on hospital patients, kappa fig- allows for interactions between the individual ures ranged from 0.61 to 0.94, except for Pain, attributes in forming the overall score. Thus, for which was 0.48 (16, Table 10). example, the effect of being both blind and deaf Grootendorst et al. analyzed the impact of might be rated as more severe than either condi- proxy versus self-report, and of interviewer- tion alone, but less severe than the sum of the versus self-administration on responses using two (10, pp116–117). The multiplicative multi- 24,000 survey responses to the HUI3 pain and attribute scoring function was used to estimate emotion scales (17). A question on employment scores for all permutations of the health states; status was used as a comparator. Weighted the calculation is shown in Exhibit 10.37. Using kappa indexes for inter-rater agreement were the calculation instructions shown below the Ex- 0.29 and 0.31 for emotion and pain, respec- hibit, the mildest departure from perfect health tively, compared with 0.82 for employment (17, (wearing glasses) has a utility of 0.973, whereas Table 5). The proxy respondents tended to re- the utility for the lowest level on all eight attrib- port less dysfunction than the self-reports. utes is −0.36, thus judged to be markedly worse Agreement for mode of administration was 0.53 than being dead. for emotion and 0.47 for pain, compared with It will be noted that the weights in Exhibit 0.91 for employment status (17, Table 9). The 10.36 for Cognition levels 2 and 3 appear to be observed disagreement for the emotion scale was reversed; this is correct, and it arose because the subdivided into 58% due to random error, 35% categories were first defined on a conceptual ba- due to rater, and 7% due to mode of administra- sis and only later were the empirical weights de- tion. For the pain scale, 31% was due to random rived. Indeed, in a French study, the magnitude error, 10% to rater type, and 59% to the mode of the weights follows the rank ordering shown of administration (17, Table 11). Mathias et al. in Table 1 (12, Table 5). The utility weights were obtained higher inter-rater agreement in a com- cross-validated by comparing the estimated parison of ratings by patients who had suffered a weights for a small selection of states against stroke and by their caregivers. Kappa values for weights measured directly in a separate scaling selected individual questions ranged from 0.37 task. For the Mark 2, the highest discrepancy to 0.80 (18, Table 1), whereas ICC values for at- was 0.08 and the mean difference for three tribute utility scores ranged from 0.39 to 0.81 health states was 0.002 (11, Table 10; 13, Table (Table 2). 2). For the Mark 3, the mean difference for three Agreement between different sets of raters for marker states was 0.01, and the intraclass corre- the utility scoring task has been reported. A lation (ICC) between them was 0.91 (10, p124). study of 144 ratings made by parents of children The ICC agreement for 73 states across two sep- with cancer, compared with other parents, arate surveys was 0.88 (10, p125). showed an ICC of 0.99 between their judg- ments; the mean difference was only 0.018 (19). Reliability In another study, correlations among three ob- One-month retest reliability for HUI Mark 3 servers were 0.8 or higher (20). A cautionary scores in a population survey application note has been raised, however, by Feeny et al., (N = 506) gave an ICC of 0.77 (1, p240). This who report that using HUI scores to reflect utili- overall figure masked variations in estimates for ties as measured directly by the standard gamble individual questions, from kappa = 0.18 to 0.77, is appropriate at a group level, but that consider- and estimates for the eight attribute scores from able variability occurs at the individual level so 0.14 to 0.73 (14). In a study of mothers report- that HUI scores “are not a good substitute for ing on their child’s health after a two- to four- directly measured utility scores at the individual week delay, weighted kappa test-retest reliability level” (21).  General Health Status and Quality of Life 689 Validity own change in health over a 12-month period, face validity. Empirically, all levels of each at- the HUI showed a substantially smaller effect tribute have been represented by responses in size than the EQ-5D, and slightly smaller than population surveys (3; 13, p515). The indepen- that of the SF-6D (30, Table 4). In a study of ef- dence of the attributes in the HUI3 was sup- fect sizes for patients with hip replacements, the ported in analyses of a community health survey. effect size for the HUI2 was 1.00; that for the The highest Kendall correlation between attrib- HUI3 was 1.19. Effect sizes for the pain scales utes was 0.35; only four of 28 correlations were were 1.0 and 1.3 for the HUI2 and HUI3. These 0.20 or greater, compared with ten of ten equiv- findings were lower than the corresponding fig- alent correlations for the EuroQol EQ-5D (9). In ures for the SF-36 (31, Table 2). a small qualitative study of content validity, the HUI3 omitted many aspects of quality of life concurrent validity. Gold et al. derived scores noted as being important to patients with that approximated the Mark 1 HUI from data Alzheimer’s disease and had a less comprehen- collected in the 1982 National Health Examina- sive coverage than the Quality of Well-Being tion Survey (N = 10,163) (32). The HUI scores scale (22). correlated 0.52 with a simple, three-point self- rating of health. As expected, both measures discriminant evidence. Three of the six HUI2 were associated with age, educational level, so- subscales and the overall utility score discrimi- cioeconomic status, and the presence of chronic nated significantly between children receiving health conditions. However, despite the much active cancer therapy and those on follow-up finer gradation of the HUI, the statistical signifi- care (15, Table 5). Using self-report data from a cance of the associations between the self-rated population survey, and after adjustment for de- health question and most chronic conditions mographic factors and the number of chronic was stronger than that for the HUI (32, Table 3). conditions, mean HUI3 scores were 0.538 for This close correspondence between the HUI and people who had had a stroke, 0.765 for people an overall health rating has been replicated (33, with arthritis, and 0.925 for people with neither Table 1). Using data from a Canadian national (23, p294). HUI2 scores distinguished between survey, Kopec et al. reported associations be- children aged eight years who had been born tween HUI scores and self-rated health, activity with a subnormal birth-weight and those with restrictions, drug use, the presence of chronic normal birth-weight (24). conditions, age, and income level (34). HUI2 and HUI3 scores discriminate between Because of their similarity, several studies severity levels of patients with Alzheimer’s dis- have compared the HUI3 with the EuroQol EQ- ease. Mark 2 scores ranged from 0.73 for bor- 5D. In a community sample of 1,477 respon- derline cases to 0.14 for those with advanced dents, Bélanger et al. compared the two scales disease (25); Mark 3 scores declined from 0.47 to a VAS rating “your health today” (8). The for borderline to −0.23 for end-stage Alzheimer’s HUI overall scores correlated 0.69 with the EQ- disease (26). Similarly, HUI scores varied across 5D scores, and 0.63 with the VAS (8, Table 8). severity levels of Parkinson’s disease (27, Table Correlations between scores for comparable di- 2). The HUI2 and HUI3 were compared in a mension on the HUI and EQ-5D included 0.56 study of patients with diabetes; the HUI3 pro- for pain, 0.42 for emotion, and 0.47 for mobil- vided better discrimination of more severe states ity (8, Table 10). Kendall correlations confirmed than the HUI2 did (28; 29, Tables 4–6). the independence of the HUI attributes; only Mark 3 scores showed significant differences two of 28 correlations exceeded 0.25 (8, Table between children before, during, and after re- 12). The area under the ROC curve for the HUI ceiving chemotherapy; the effect size was 2.8 (3, in distinguishing respondents who had been pp380–381). However, in a study of patients hospitalized in the past year was 0.66, virtually with rheumatologic disorders who rated their identical to the 0.64 for the EQ-5D and 0.66 for 690 Measuring Health the VAS (8, Table 16). Areas under the ROC predictive validity. In the Gold et al. study, curve for detecting the presence of chronic con- HUI1 scores were compared with health out- ditions were 0.70 for the HUI, 0.71 for the EQ- comes recorded five years later. Baseline HUI1 5D, and 0.69 for the VAS; for activity scores were associated with subsequent declines restrictions, values were 0.83 for the HUI, 0.84 in health, with hospital admissions, and with for the EQ-5D, and 0.81 for the VAS (8, Table mortality. However, the gradient in scores across 16). A statistical comparison of HUI and EQ- these outcome categories was steeper for the self- 5D scores for a range of subpopulations sug- rating of health than for the HUI scores (32, gested that the two scales provided virtually Table 4). Gold et al. concluded that the two identical ratings for all but the healthiest groups measures were complementary in predicting the (8, pp43–46). A Dutch study obtained a rank outcomes, but that they did “not fully cover the correlation of 0.90 between the scores on the waterfront of domains of HRQOL” (32, p174). EQ-5D and the HUI, although the mean scores The responsiveness of the HUI3 to changes on the HUI were 0.16 points lower than those in health status over two years was tested in a for the EQ-5D for six case descriptions (35, population survey (N = 81,804) (33). Baseline Table 2). Also remarkable in that study was the HUI3 scores discriminated between those who contrast in preference scores between the HUI2 remained healthy and those who were subse- and the HUI3 (Table 2). By contrast, a U.S. quently hospitalized, or those who became re- study found that mean scores for the HUI were stricted in activities; it also identified those who 0.11 points higher than the EQ-5D scores. In developed a severe chronic illness, but not those this study, the intraclass correlation between who developed a moderate condition (33, scores was 0.49 before treatment, 0.66 one p566). For each of these outcomes, the stan- month after treatment, and 0.78 at three dardized response means for HUI3 were com- months; the ICC between change scores for the parable with, but slightly lower than, those of a two scales over time was only 0.30 (36, p597). single-item overall health rating (33, Tables In a study of patients with Parkinson’s disease, 5–7). A before- and after-study of patients who the correlation between the HUI and the EQ5-D had hip replacement surgery showed the HUI3 was 0.74; the correlations between the EQ5-D to be more responsive to improvement in health and measures of disease severity and quality of status than the HUI2 or the SF-36, but less so life were in general higher than those for the than disease-specific instruments such as the HUI (27, p105). For patients with rheumato- Harris Hip Scale, the Western Ontario and Mc- logic disorders, the correlation between HUI Master Universities Osteoarthritis Index and EQ-5D was 0.68 (30, Table 3). (WOMAC), or the McMaster-Toronto Arthritis Spearman coefficients comparing HUI2 Patient Preference Disability Questionnaire scores to the Pediatric Oncology Quality of Life (MACTAR) (31). (POQOL) scale for children with cancer ranged from −0.41 (self-care dimension), to −0.51 (pain) Alternative Forms and −0.53 (emotion). Equivalent correlations A questionnaire that incorporates the Mark 3 with the Child Behavior Checklist were −0.48 version is shown in the Appendix of the book by (emotion) and −0.29 (pain) (15, Table 4). In the Patrick and Erickson (5, pp385–389). same study, the VAS score for the HUI showed a The HUI Web site provides information on stronger correlation with the POQOL overall translations of questionnaires, which are avail- score, at −0.70, than did the HUI utility score, at able in English, French, Spanish, Italian, Ger- −0.44 (15, p427). Correlations between HUI2 man, Dutch and Japanese; Polish, Swedish, overall scores and the SF-12 were 0.60 for the Norwegian, Finnish and Danish translations are SF-12 physical score and 0.59 for the mental being planned. Validity information is available score; equivalent figures for the HUI3 were 0.64 on French translations (16; 37–39), and Cana- and 0.60, respectively (29, Tables 7 and 8). dian French (15). A Spanish version has a re-  General Health Status and Quality of Life 691 ported alpha of 0.76 and a correlation of 0.79 nity surveys. This derives from a limitation of with the EQ-5D (40). the EQ-5D scoring, in which it is not possible to achieve a score between 0.88 and 1.00; of re- Reference Standards spondents rated 1.0 by the EQ-5D, the HUI3 The HUI3 has been used in several Canadian na- rated many as falling between 0.9 and 0.99 (8, tional surveys, and reference scores are available p24). On a technical note, Austin et al. described for community and institutionalized populations a modified regression approach, Tobit analysis, by age and sex (3, p381). which can improve estimates in the presence of Furlong et al. quote mean scores for various ceiling effects (43; 44). patient groups (3, Table 2) and reproduce the As with other health measures, it seems prob- frequency distributions for each attribute from a able that the various modes of administering the French study (3, Table 3). HUI (e.g., personal interviews, telephone, mail) may not provide equivalent results (45). Relia- Commentary bility results suggest that variations may be The HUI has received widespread attention and greatest for the pain and emotion scales (17). frequent use in survey research, especially in Studies making more than one administration of Canada. The basic articles describing the instru- the HUI should probably use the same mode of ment give a clear indication of the steps taken in administration on each occasion. developing the scale, especially the utility scaling A limitation of the EuroQol EQ-5D may be component. Indeed, the impression is that the that it has relatively few response steps for each authors are more interested in the utility scaling item, making it insensitive to minor variations in procedures than in question wording and con- health. By contrast, descriptions of the HUI3 of- tent validity. The Web site (www.fhs.mcmaster ten cite as an advantage the fact that it can rate .ca/hug/index.htm) contains basic information 972,000 unique health states. However, this may on the HUI, an extensive bibliography, and order have limited practical relevance, because very forms for unpublished technical reports, which many fewer actually occur. In the 1990 Ontario are available at cost. The HUI scoring function Health Survey (N = 68,394), 1,755 health states has been applied to various other data collection were used, but 129 health states accounted for methods, such as the Minimum Data Set (41) almost 95% of responses. In the 1991 Canadian and the SF-12 (42). General Social Survey (N = 11,567), only 12 The wide range of states of poor health make health states occurred among more than 1% of it unlikely that the HUI will suffer floor prob- respondents; taken together, these 12 accounted lems. However, the HUI deliberately focuses on for 75% of all responses. Nonetheless, Feeny et capacity; the arguments in support of this are al. note that several states were identified that cogent, but users should recognize that this gives would not have been picked up by the EuroQol it a narrower scope than other instruments (e.g., or by Rosser’s Disability Index (1, p244). It also SF-36). It may also lead to ceiling problems in remains somewhat disconcerting that several that the HUI may not be sensitive to variations studies have indicated that a single summary at the positive end of the health continuum. This question or a VAS can provide discrimination re- may restrict its application in prognostic studies sults as good as those of the HUI3 and the EQ- and in measuring outcomes where performance 5D. above mere competence is relevant, although The utility scoring system has applications many other instruments also suffer ceiling prob- beyond assessing individual health status. In a lems (1, p243). For example, comparisons with modeling study, for example, typical health out- the EQ-5D suggest similar performance on va- comes for a therapy can be taken from published lidity tests, although the HUI3 is more sensitive reports, mapped onto the HUI classification sys- to variation at the healthy end of the spectrum tem and then scored, thus permitting cost-utility and so may be more suitable for use in commu- analyses. The preference scores can also be used 692 Measuring Health in calculating quality-adjusted life expectancy (13, p517). References Different health indexes take different ap- (1) Feeny D, Torrance GW, Furlong WJ. proaches to estimating preferences, and the de- Health Utilities Index. In: Spilker B, ed. bate remains active. Feeny et al. recommend Quality of life and pharmacoeconomics in utilities rather than values, arguing that because clinical trials. Philadelphia: Lippincott- future health is uncertain, utilities are the appro- Raven, 1996:239–252. priate measure of preference. By contrast, the (2) Feeny D, Furlong WJ, Boyle M, et al. Quality of Well-Being scale used values derived Multi-attribute health status classification systems: Health Utilities Index. from VAS tasks, whereas the EuroQol used VAS Pharmacoeconomics 1995;7:490–502. and time trade-off methods. Feeny et al. show (3) Furlong WJ, Feeny DH, Torrance GW, et empirically how the approaches differ in their al. The Health Utilities Index (HUI) system relative weighting of major and minor disabili- for assessing health-related quality of life in ties: values pay relatively greater attention to mi- clinical studies. Ann Med nor disabilities compared with utilities (1, Table 2001;33:375–384. 7). In terms of the multiattribute scoring func- (4) Richardson CG, Zumbo BD. A statistical tion, the HUI uses a multiplicative model examination of the Health Utility Index- whereas the Quality of Well-Being scale uses the Mark III as a summary measure of health simpler additive model. Torrance et al. argue status for a general population health that the additive model tends to underestimate survey. Soc Indicat Res 2000;51:171–192. (5) Patrick DL, Erickson P. Health status and the impact of that disability in situations where a health policy: quality of life in health care person has a single, severe disability (13, p513). evaluation and resource allocation. New There remains disagreement over methodologi- York: Oxford University Press, 1993. cal details of how best to estimate utilities, but (6) Cadman D, Goldsmith C. Construction of the team developing the HUI have contributed social value or utility-based health indices: significantly to pushing this field forward. the usefulness of factorial experimental In the context of the debate over health in- design plans. J Chronic Dis dexes versus health profile measures, some infor- 1986;39:643–651. mation has been collected in a comparison (7) Feeny D, Furlong W, Barr RD, et al. A between the HUI and the SF-6D. The latter is a comprehensive multiattribute system for rescored version of the SF-36 that provides utility classifying the health status of survivors of childhood cancer. J Clin Oncol scores. Although both used variants of a standard 1992;10:923–928. gamble approach to derive utility weights, the (8) Bélanger A, Berthelot J-M, Guimond E, et agreement between the two methods is low. This al. A head-to-head comparison of two may in part reflect the contrast in content: the generic health status measures in the house- HUI focuses on ‘within a the skin’ definition of hold population: McMaster Health Utilities health, whereas the SF questions cover handicap Index (Mark3) and the EQ-5D. Statistics and function. It may also reflect methodological Canada Report. Ottawa, Ontario: differences in the way weights were derived (46, 2002;1–62. p980). The implication is that we should not yet (9) Houle C, Berthelot J-M. A head-to-head consider utility scores to be equivalent across in- comparison of the Health Utilities Index struments; further work is required to standard- Mark 3 and the EQ-5D for the population living in private households in Canada. ize the utility metric and also to indicate which Quality of Life Newsletter 2000;24:5–6. health index provides superior results. (10) Feeny D, Furlong W, Torrance GW, et al. Multiattribute and single-attribute utility Address functions for the Health Utilities Mark 3 William Furlong, Health Utilities Inc., 88 Syden- system. Med Care 2002;40:113–128. ham Street, Dundas, Ontario L9H 2V3, Canada. (11) Torrance GW, Feeny DH, Furlong WJ, et E-mail: [email protected]. al. Multiattribute utility function for a  General Health Status and Quality of Life 693 comprehensive health status classification (22) Silberfeld M, Rueda S, Krahn M, et al. system. Health Utilities Index Mark 2. Content validity for dementia of three Med Care 1996;34:702–722. generic preference based health related (12) Le Galès C, Buron C, Costet N, et al. quality of life instruments. Qual Life Res Development of a preference-weighted 2002;11:71–79. health status classification system in (23) Grootendorst P, Feeny D, Furlong W. France: the Health Utilities Index 3. Health Health Utilities Mark 3. Evidence of Care Manag Sci 2002;5:41–51. construct validity for stroke and arthritis in (13) Torrance GW, Furlong W, Feeny D, et al. a population health survey. Med Care Multi-attribute preference functions: 2000;38:290–299. Health Utilities Index. Pharmacoeconomics (24) Saigal S, Feeny D, Furlong WJ, et al. 1995;7:503–520. Comparison of the health-related quality of (14) Boyle MH, Furlong W, Feeny D, et al. life of extremely low birth weight children Reliability of the Health Utilities Index— and a reference group of children at age Mark III used in the 1991 cycle 6 eight years. J Pediatr 1994;125:418–425. Canadian General Social Survey Health (25) Neumann PJ, Kuntz KM, Leon J, et al. Questionnaire. Qual Life Res Health utilities in Alzheimer’s disease: a 1995;4:249–257. cross-sectional study of patients and (15) Trudel JG, Rivard M, Dobkin PL, et al. caregivers. Med Care 1999;37:27–32. Psychometric properties of the Health (26) Neumann PJ, Sandberg EA, Araki SS, et al. Utilities Index Mark 2 system in paediatric A comparison of HUI2 and HUI3 utility oncology patients. Qual Life Res scores in Alzheimer’s disease. Med Decis 1998;7:421–432. Making 2000;20:413–422. (16) Costet N, Le Gales C, Buron C, et al. (27) Siderowf A, Ravina B, Glick HA. French cross-cultural adaptation of the Preference-based quality-of-life in patients health utilities indexes Mark 2 (HUI2) and with Parkinson’s disease. Neurology 3 (HUI3) classification systems. Qual Life 2002;59:103–108. Res 1998;7:245–256. (28) Maddigan SL, Feeny DH, Johnson JA. A (17) Grootendorst PV, Feeny DH, Furlong W. comparison of the health utilities indices Does it matter whom and how you ask? mark 2 and mark 3 in type 2 diabetes. Med Inter- and intra-rater agreement in the Decis Making 2003;23:489–501. Ontario Health Survey. J Clin Epidemiol (29) Maddigan SL, Feeny DH, Johnson JA. 1997;50:127–135. Construct validity of the RAND-12 and (18) Mathias SD, Bates MM, Pasta DJ, et al. Health Utilities Index Mark 2 and 3 in type Use of the Health Utilities Index with 2 diabetes. Qual Life Res stroke patients and their caregivers. Stroke 2004;13:435–448. 1997;28:1888–1894. (30) Conner-Spady B, Suarez-Almazor ME. (19) Wang Q, Furlong W, Feeny D, et al. How Variation in the estimation of quality- robust is the Health Utilities Index Mark 2 adjusted life-years by different preference- utility function? Med Decis Making based instruments. Med Care 2002;22:350–358. 2003;41:791–801. (20) Gemke RJBJ, Bonsel GJ. Reliability and (31) Blanchard C, Feeny D, Mahon JL, et al. Is validity of a comprehensive health status the Health Utilities Index responsive in measure in a heterogeneous population of total hip arthroplasty patients? J Clin children admitted to intensive care. J Clin Epidemiol 2003;56:1046–1054. Epidemiol 1996;49:327–333. (32) Gold M, Franks P, Erickson P. Assessing (21) Feeny D, Blanchard C, Mahon JL, Bourne the health of the nation. The predictive R, Rorabeck C, Stitt L et al. Comparing validity of a preference-based measure and community-preference-based and direct self-rated health. Med Care standard gamble utility scores: Evidence 1996;34:163–177. from elective total hip arthroplasty. Int J (33) Kopec JA, Schultz SE, Goel V, et al. Can Technol Assess Health Care the Health Utilities Index measure change? 2003;19:362–372. Med Care 2001;39:562–574. 694 Measuring Health (34) Kopec JA, Williams JI, To T, et al. Measur- when one can do better than ‘1’? Qual Life ing population health: correlates of the Res 2000;9:911–914. Health Utilities Index among English and (45) Verrips GH, Stuifbergen MC, den Ouden French Canadians. Can J Public Health AL, et al. Measuring health status using 2000;91:465–470. the Health Utilities Index: agreement (35) Oostenbrink R, Moll HA, Essink-Bot M-L. between raters and between modalities of The EQ-5D and the Health Utilities Index administration. J Clin Epidemiol for permanent sequelae after meningitis: a 2001;54:475–481. head-to-head comparison. J Clin Epidemiol (46) O’Brien BJ, Spath M, Blackhouse G, et al. 2002;55:791–799. A view from the bridge: agreement (36) Bosch JL, Hunink MGM. Comparison of between the SF-6D utility algorithm and the Health Utilities Index Mark 3 (HUI3) the Health Utilities Index. Health Econ and the EuroQol EQ-5D in patients treated 2003;12:975–981. for intermittent claudication. Qual Life Res 2000;9:591–601. The EuroQol EQ-5D Quality of Life (37) Costet N, Le Galès C, Buron C, et al. French cross-cultural adaptation of the Scale Health Utilities Indexes Mark 2 (HUI2) (The EuroQol Group, 1990, revised and 3 (HUI3) classification systems. 1993) Clinical and Economic Working Groups. Purpose Qual Life Res 1998;7:245–256. (38) Le GC, Buron C, Costet N, et al. The European Quality of Life (EuroQol) mea- Development of a preference-weighted sure expresses health status in a single index health status classification system in score; it is intended for use in evaluative studies France: the Health Utilities Index 3. Health such as drug trials and in policy research (1). It Care Manag Sci 2002;5:41–51. is a generic scale intended to form one compo- (39) Le Galès C, Costet N, Gentet JC, et al. nent of a measurement battery supplemented, Cross-cultural adaptation of a health status for example, by disease-specific questions. classification system in children with cancer. First results of the French adaptation of the Conceptual Basis Health Utilities Index Marks 2 and 3. Int J From the outset in 1987, the EuroQol has been Cancer Suppl 1999;12:112–118. developed with cross-cultural applications in (40) Ruiz M, Rejas J, Soto J, et al. Adaptation mind. The instrument was developed by the in- and validation of the Health Utilities Index Mark 3 into Spanish and correction norms ternational European Quality of Life Group, for Spanish population. Med Clin (Barc) which has since grown to include members from 2003;120:89–96. the United States, Canada, and Japan. Observing (41) Wodchis WP, Hirdes JP, Feeny DH. the great number of rival instruments that had Health-related quality of life measure not been systematically compared, they sought based on the minimum data set. Int J to develop a standard instrument that could be Technol Assess Health Care used in international studies (2). Recognizing 2003;19:490–506. that no measurement method can suit every ap- (42) Franks P, Lubetkin EI, Gold MR, et al. plication, they designed the EuroQol to have a Mapping the SF-12 to preference-based core set of generic quality of life items that could instruments - Convergent validity in a low- be supplemented for particular disease applica- income, minority population. Med Care tions. The core set would permit cross-national 2003;41:1277–1283. (43) Austin PC, Escobar M, Kopec JA. The use comparisons and would encourage methodologi- of the Tobit model for analyzing measures cal standardization. of health status. Qual Life Res The design of the EuroQol balanced the de- 2000;9:901–910. sire to cover all topics that might be relevant (44) Grootendorst P. Censoring in statistical against producing an instrument brief and sim- models of health status: what happens ple enough to be practical (3). The measure was  General Health Status and Quality of Life 695 designed as a self-completed questionnaire that can be converted into a summary index using a would be acceptable for use in postal surveys utility-weighted scoring system. The weights and would represent health status as a single ag- may use the respondent’s own expressed prefer- gregated index value (2; 4, p202). Because the ences using a zero to 100 scale that indicates her instrument would be used chiefly for overall overall valuation of her current state of health decision-making, brevity and ease of administra- derived from the VAS rating of her “own health tion were of paramount importance, so the num- state today.” In this case, part 1 of the question- ber of health states in each dimension was naire merely provides descriptive information. intentionally kept small (4). Alternatively, a utility score can be based on es- tablished scale values; the EuroQol group has Description devoted considerable attention to the best way to The current version of the EuroQol, the EQ-5D, derive these weights (3). Weights have been de- covers five dimensions of health: mobility, self- rived from studies in several countries (9, p338), care, usual activities, pain/discomfort, and anxi- including the United States (10) and Japan (11; ety/depression (see Exhibit 10.38). For each 12); the EuroQol web site lists 21 sets of scoring dimension, three items describe no problems, weights. Kind has compared weights from the moderate problems, or severe problems. The United Kingdom, Sweden, the Netherlands, and closest match is chosen to describe the respon- Norway (4, Figure 3). Weights derived from a dent’s status on each dimension, providing 35, or VAS approach have been compared with values 243 possible health state combinations. In addi- derived from the time trade-off (13; 14, Table 2) tion, a self-rating of overall health is obtained and a test-retest study of the ratings gave an from a 20-cm visual analogue scale (VAS), an- intra-class correlation of 0.78 (14, p528). chored at zero (worst imaginable health state) Exhibit 10.39 illustrates scale weights and a and 100 (best imaginable health state). scoring system derived from 3,000 interviews in In addition to the self-report version, the EQ- a representative community sample in the United 5D exists in observer, proxy (5), and telephone Kingdom. As with the Quality of Well-Being response versions. Agreement between patient- Scale, the values are expressed as negative completed and proxy versions is moderately weights that are subtracted from the maximum good (kappa 0.53), but less good among more score of 1.0, which represents well-being. In ad- severely ill patients unable to complete the scale dition to subtracting the values shown in the themselves (6, Table 1). Elderly people are likely table, a constant of 0.081, shown at the foot of to require interview administration: an estimated the table, is subtracted if one or more dimen- 11% at age 65 and 37% at age 75 (7, p6). The sions is scored at two or three. If one or more di- EQ-5D can be administered in about 90 seconds mensions is scored three, a further constant of in a community-dwelling sample (8). Note that 0.269 is subtracted. Thus, for example, a re- earlier versions of the EuroQol were somewhat sponse pattern of someone with extreme anxiety different (1, pp178–179); many of the available (profile 11113) scores 1.0 − 0.081 − 0.236 validity and reliability results refer to the earlier − 0.269 = 0.414, whereas a pattern of 11222 versions. Exhibit 10.38 shows the EQ-5D self- would score 1.0 − 0.081 − 0.036 − 0.123 − 0.071 classification system and VAS (9, Figure 2). = 0.689. Dolan described the derivation of these In scoring the EQ-5D, a person’s health state weights (15). is first summarized by a five-digit profile, indi- Those who use the EQ-5D are asked to regis- cating the score on each dimension. Thus, a pro- ter their research online at www.euroqol.org. file denoted as 12311 would represent no difficulties with mobility, pain or anxiety/de- Reliability pression, but some problems with self-care and Test-retest reliability for the dimensions in the an inability to perform usual activities. The re- original version of the EuroQol were reported to sulting profile can be graphed, as illustrated by range from 0.69 to 0.94 (16, Table 1). Retest re- Rabin and deCharro (9, Figures 3 and 4), or it liability for the current version has been reported Exhibit 10.38 The EuroQol Quality of Life Best imaginable Scale health state By placing a tick (thus ✓ ) in at least one box in each 100 group below, please indicate which statements best describe your own healthstate today. Mobility 1 I have no problems in walking about 9•0 2 I have some problems in walking about 3 I am confined to bed Self-Care 8•0 1 I have no problems with self-care 2 I have some problems washing or dressing myself 3 I am unable to wash or dress myself 7•0 Usual Activities (e.g., work, study, housework, family or leisure activities) 1 I have no problems with performing my 6•0 usual activities (e.g. work, study, housework, family or leisure activities. 2 I have some problems with performing my usual activities Your own health 5•0 today 3 I am unable to perform my usual activities Pain / Discomfort 1 I have no pain or discomfort 4•0 2 I have moderate pain or discomfort 3 I have extreme pain or discomfort Anxiety / Depression 3•0 1 I am not anxious or depressed 2 I am moderately anxious or depressed 3 I am extremely anxious or depressed 2•0 Visual analogue scale To help people say how good or bad a health state is, we have drawn a scale (rather like a thermometer) on which the best state you can imagine is marked by 1•0 100 and the worst state is marked by 0. We would like you to indicate on this scale how good or bad is your own health today, in your opinion. Please do this by drawing a line from the box “Your own health today” to whichever point on the scale 0 indicates how good or bad your health state is. Worst imaginable health state Adapted from an original provided by Dr. A. Williams. With permission. 696  General Health Status and Quality of Life 697 Exhibit 10.39 Dimensions and Weights for the occasions for patients with rheumatoid arthritis EuroQol as proposed by Dolan (15) (21, Table III). The EuroQol correlated 0.51 Note: the weights indicated are subtracted from 1.00 with depression scores and 0.44 with anxiety scores from the Hospital Anxiety and Depres- Item weights sion scale. Correlations between the EuroQol In- Dimensions 1 2 3 dex score and the VAS were 0.37 and 0.54 on two testing occasions (21, Table III). Mobility −0.0 −.069 −.314 The EQ-5D was compared with the Barthel Self-care −0.0 −.104 −.214 Index in a Dutch study of patients who had suf- Usual activity −0.0 −.036 −.094 Pain/discomfort −0.0 −.123 −.386 fered a stroke; Barthel scores explained 54 to Anxiety/depression −0.0 −.071 −.236 59% of the variance in EQ scores, and observed constants −0.0 −.081 −.269 EQ scores correlated (intraclass correla- Figures supplied by Prof. A Williams. Reproduced with per- tion = 0.70) with those predicted from the mission. Barthel (22, Tables 3 and 4). Those authors noted that this implies that Barthel scores could substitute for EQ-5D when the latter cannot be as 0.86 for group level coefficients averaged over obtained. Convergent correlations with the health states and 0.90 for a coefficient derived Medical Outcomes Study HIV scale lay in the from individual correlations considering all range 0.50 to 0.65 (23, Table 3). Convergent health states simultaneously (17, p1543). Kappa correlations with the COOP Charts include 0.39 values for seven-day retest stability were 0.29 for between EQ-5D mobility and COOP physical anxiety/depression, 0.40 for mobility, 0.58 for scales, 0.59 between the two daily activities pain, and 0.61 for usual activities (18, p90). A scales, 0.70 between EQ anxiety/depression and Spearman retest correlation over 6 months was COOP feelings, and 0.74 between the two pain 0.67 (19, Table 2). scales (7, Table 6). In a study by Bélanger, the Kendall correlations among the dimensions Health Utilities Index (HUI-3) and EQ-5D on the EQ-5D ranged from 0.24 to 0.64 (19, scores correlated 0.69; the EQ-5D correlated Table 11). 0.63 with a VAS rating of “your health today” (24, Table 8). Correlations between equivalent Validity dimension scores on the HUI and EQ-5D in- In a small qualitative study of content validity, cluded 0.56 for pain, 0.42 for emotion, and 0.47 the EQ-5D was judged to omit aspects of quality for mobility (24, Table 10). A Dutch study ob- of life noted as being important to patients with tained a strong rank correlation of 0.90 between Alzheimer’s disease and to have a less compre- the scores on the EQ-5D and the HUI, but pref- hensive coverage than the Quality of Well-Being erence scores on the HUI-3 were significantly scale (20). Evidence for the construct validity of lower than those for the EQ-5D, by an average the preliminary version of the EuroQol was of 0.16 points for six case descriptions (25, drawn from the pattern of responses across age, Table 2). A statistical comparison of EQ-5D and gender, and socioeconomic groups, and between HUI scores suggested that the two scales pro- those who had recently sought health care ser- vided virtually identical ratings for all but the vices and those who had not (1, Table 2). The in- healthiest groups (24, pp43–46). In a study of strument identified significant contrasts in the 100 patients with Parkinson’s disease, the corre- anticipated direction for most of these compar- lation between the EQ5-D and the HUI was isons, across most of the six dimensions. 0.74; the correlations between the EQ5-D and measures of disease severity and quality of life concurrent validity. Correlations between the were in general higher than those for the HUI preliminary version of the EuroQol and three (26, p105). Other correlations include 0.69 be- subscores on the Health Assessment Question- tween the HUI3 and the EQ-5D Index and 0.56 naire ranged from 0.46 to 0.76 on two testing with the VAS (27, Table 1). In the same study, 698 Measuring Health correlations for the EQ Index were 0.64 with the scores. Scores on the Functional Living Index- physical score and 0.52 for the mental score on Cancer (FLIC) showed variability for patients the SF-36. Other comparisons of the EQ Index who scored at the ceiling on the EQ-5D (33, and the SF-36 include 0.80 with the SF-36 phys- p481). ical function scale, 0.83 with the SF-36 pain scale, 0.70 with vitality, and 0.50 with the men- discriminant validity. The area under the tal health score (28, Table 3). In a study of pa- ROC curve for the EQ-5D in distinguishing re- tients with arthritis, the EQ-5D correlated −0.78 spondents who had been hospitalized in the past with the Health Assessment Questionnaire year was 0.64, almost identical to 0.66 for the (HAQ); the correlation for the VAS scale was HUI and 0.66 for the VAS (24, Table 16). Areas 0.61 (29, Table 5). The overall score on the EQ- under the ROC curve for detecting the presence 5D correlated only −0.11 with the General of chronic conditions were 0.71 for the EQ-5D, Health Questionnaire; the correlation for the 0.70 for the HUI, and 0.69 for the VAS; for ac- mood section was also low, at 0.30 (30, p316). tivity restrictions, values were 0.84 for the EQ- 5D, 0.83 for the HUI, and 0.81 for the VAS (24, ceiling effects. Applied in general population Table 16). samples, the EQ-5D produces a marked ceiling effect, often with 50 to 65% of the sample re- sensitivity to change. Findings on the respon- porting a 11111 profile (1, Table 5; 30, Table 1; siveness of the EQ-5D are somewhat mixed (33). 31, Table 3). Bélanger et al. compared the EQ- In a study of patients with rheumatologic disor- 5D with the HUI3 in a community sample ders who rated their own change in health over a (N = 1,477) (24). The EQ-5D classified 48% as 12-month period, the correlation between EQ- being in full health compared with 26% on the 5D and HUI3 was 0.68; the EQ-5D showed a HUI (24, p18). The contrast reflects the scoring substantially larger effect size than the HUI or gap in the EQ-5D between 0.88 and 1.00; many the SF-6D (34, Tables 3 and 4). In a study of of those people rated 1.0 on the EQ-5D obtained chemotherapy for breast cancer patients, effect HUI scores between 0.9 and 0.99. The EuroQol size results for the EQ-5D (mostly > 0.8) were and the SF-36 were applied in a general popula- comparable with those obtained using the FLIC tion and those achieving perfect scores on each (33, p482). However, a single VAS score showed EQ dimension were divided into two groups on an effect size as high as, or higher than either the basis of their SF-36 scores. In each instance, FLIC and EQ-5D (33, Table 2). In a study of there were significant differences between the HIV patients, the EQ-5D Index proved less sensi- groups in terms of medical care consultation tive to change than the VAS did (23, Tables 4 and rates and selected demographic variables sug- 5). In a study of patients being treated for arthri- gesting that the EQ-5D may miss relevant varia- tis, the EQ-5D Index and the VAS proved equally tion among relatively healthy people (1, Tables sensitive to change over time, and moreso than 6A and 6B). In a similar analysis comparing the the HAQ: standardized response means of 1.0 EQ-5D with the SF-12, 50% of a community for the EQ scales compared with 0.41 for the sample reported maximal EQ scores. These re- HAQ (29, Table 7). A study of patients with hip spondents were divided at the median SF-12 fractures recorded a standardized effect size of component scores; the resulting groups differed 1.37 for the EQ-5D, compared with 0.89 for the in terms of number of chronic health problems, SF-36; the corresponding standardized response mean VAS score, age, and employment status means were 0.90 and 0.82, respectively (35, (31, Table 9). Similarly, Johnson et al. noted that Table 5). Conversely, in a study of patients re- the SF-12 identified as restricted some people ceiving therapy for prostate disorders, the EQ-5D rated as healthy on the EQ-5D (32). Although showed no significant change over time, in con- the SF-12 and SF-36 are sensitive to relevant dif- trast to the SF-36 and to disease-specific mea- ferences missed by EuroQol Index scores of 1.0, sures (36). Similarly, the SF-36 proved more the effect seems less pronounced for the EQ-VAS responsive (i.e., standardized response means,  General Health Status and Quality of Life 699 0.41–0.57 for the subscale scores) than the EQ- center at Erasmus University, Rotterdam, coordi- 5D (response mean, 0.32) in a study of patients nates development work. In 1998, a three-year with asthma (37, Table 5). The effect size for project, EQ-net, was funded by the European identifying elderly patients with long-standing ill- Union to support instrument development and ness was 0.85 for the EQ-5D, compared with harmonization of the various valuation projects. 0.96 for the physical functioning scale on the SF- The business management center is developing 36, and 0.77 on the SF-36 physical role function guidelines for standard reporting of the EQ-5D, scale (19, Table 5). In a study of treatment for and software packages for analysis are in prepa- sleep apnea, the effect size for the EQ-5D index ration. was only 0.23 (0.27 for the VAS), versus 0.63 for Compared with other health index measures the physical component score on the SF-36 and such as the Quality of Well-Being Scale or the 1.33 on the Patient-Generated Index (38, Table HUI, the EuroQol is a simpler instrument with 2). Jenkinson et al. commented that the content fewer scale levels, a brief time referent, and no of the EQ-5D probably is not sensitive to the coverage of symptoms. Unlike the others, it can types of health concerns raised by patients expe- be presented as a profile or as an index. riencing sleep apnea (38). Various concerns have been expressed over the EQ-5D. Its categories are relatively coarse: Alternative Forms although in principle it can generate 243 permu- The EuroQol was developed simultaneously in tations of the 15 items (plus the headings uncon- Dutch, English, Finnish, Norwegian, and scious and dead), many of these permutations Swedish versions. As of 2004, it is available in are implausible (e.g., confined to bed, yet able to 33 approved translations, with 14 more awaiting perform usual role activities), so many fewer approval (see www.euroqol.org; Chinese 9; 39, states occur in empirical studies. Compounding Table 2). A study of the Singapore version re- this, the scoring system produces awkward gaps ported reliability and the results of testing 13 va- in the final scale: below 1.00, the next highest lidity hypotheses (18). score is 0.883. In a Canadian community survey, The possible benefits of including items on this meant that the EQ-5D (in contrast to the cognition have been explored (40). HUI) could offer no discrimination among al- most half of the respondents (8). Furthermore, Reference Standards the substantial value of 0.269 is subtracted if a Brazier et al. showed the proportions of people level 3 occurs within at least one dimension, reporting a problem in each dimension of the which can lead to a bimodal distribution if sub- EuroQol by age group, sex, and socioeconomic stantial numbers of respondents report a level 3 class from a survey in Sheffield, England (1, (33, p485). Other health indexes have addressed Table 2). Equivalent results have been reported the challenge of coarse gradations in different from a national survey of 3,395 British adults ways. Rosser’s index includes distress and Ka- (41, Tables 2–4). Mean index scores by age- plan’s includes symptoms to give finer discrimi- group for people aged 75 and older were re- nation, but the EuroQol includes only functional ported by Brazier et al. (19, Table 4). status questions. Hence, the EQ-5D is well suited to patients with major morbidity (42, Commentary p382) but is relatively insensitive to variations in The EuroQol was carefully developed by a well- well-being at the upper ends of the health con- organized international team that has paid care- tinuum. ful attention to supporting the continued The relative lack of responsiveness of the EQ- development and testing of the instrument. In 5D may be due to the ceiling effect and the 1995, the EuroQol Foundation was created to coarseness of the response scales for each item. disseminate information on the instrument and Several studies have found that even quite sick to keep track of studies in which it is used (9, patients can score at the ceiling of the scale (23, Figure 6). A EuroQol Business Management p279); this is a problem shared with the Notting- 700 Measuring Health ham Health Profile (43). Essink-Bot compared Further information on the EuroQol can be the ability of scores on four measures to discrim- obtained from Frank de Charro, PhD, EuroQol inate between people who were absent from Group Business Management, PO Box 4443, work due to illness and others who were not. The 3006 AK Rotterdam, Netherlands, E-mail: SF-36 was the most discriminating (mean area [email protected]. under the ROC curve [AUC], 0.72), followed by the COOP Charts (mean AUC, 0.64), the EQ-5D References (mean AUC, 0.61) and the Nottingham Health Profile (mean AUC, 0.60) (43, Table 5). (1) Brazier J, Jones N, Kind P. Testing the validity of the Euroqol and comparing it The EQ-5D is simpler to complete than rival with the SF-36 health survey questionnaire. methods, and yet in three postal surveys only 59, Qual Life Res 1993;2:169–180. 66, and 60% of the questionnaires that were re- (2) EuroQol Group. EuroQol: a new facility turned were fully usable (2, pp205–206). There for the measurement of health-related has been some criticism of the relatively direct quality of life. Health Policy language used in the emotional questions: Pater- 1990;16:199–208. son cited some elderly respondents who objected (3) Brooks R. EuroQol: the current state of to the term “anxious or depressed” and did not play. Health Policy 1996;37:53–72. check that item even though they recorded equiv- (4) Kind P. The EuroQol instrument: an index alent symptoms on the COOP Charts (44, of health-related quality of life. In: Spilker p876). Other comments have noted the lack of B, ed. Quality of life and pharmacoeconomics in clinical trials. detail in the health state descriptions used in the Philadelphia: Lippincott-Raven, scaling task (45). 1996:191–201. A psychometric concern with the EQ-5D lies (5) Badia X, Diaz-Prieto A, Rué M, et al. Mea- in the level of correlation between the health di- suring health and health state preferences mension scores, which do not seem independent. among critically ill patients. Intensive Care In a sample of 1,477 community-dwelling peo- Med 1996;22:1379–1384. ple, Kendall correlations between dimension (6) Dorman PJ, Waddell F, Slattery J, et al. Are scores ranged from 0.25 (mobility and anxi- proxy assessments of health status after ety/depression) and 0.64 (mobility and usual ac- stroke with the EuroQol questionnaire tivities). Five of ten correlations exceeded 0.40 feasible, accurate, and unbiased? Stroke (8). By comparison, no correlation between the 1997;28:1883–1887. (7) Coast J, Peters TJ, Richards SH, et al. Use dimensions of the HUI exceeded 0.40. This sug- of the EuroQoL among elderly acute care gests that the dimensions of the EQ-5D are rela- patients. Qual Life Res 1998;7:1–10. tively generic and thus overlap; there is consider- (8) Houle C, Berthelot J-M. A head-to-head able redundancy between them and they cannot comparison of the Health Utilities Index be precisely interpreted. Mark 3 and the EQ-5D for the population Of the leading generic health index measures, living in private households in Canada. the EQ-5D is the briefest and simplest to admin- Quality of Life Newsletter 2000;24:5–6. ister and pays for this in terms of a lack of sensi- (9) Rabin R, de Charro FT. EQ-5D: a measure tivity to change and to minor departures from of health status from the EuroQol group. well-being. It appears best suited to studies of Ann Med 2001;33:337–343. patients who are sicker. Where a single, numeri- (10) Johnson JA, Coons SJ, Ergo A, et al. Valuation of the EuroQol (EQ-5D) health cal index summary of health is required, users states in an adult US sample. will have to choose carefully among the Euro- Pharmacoeconomics 1998;13:421–433. Qol, the HUI, the Rosser and Kaplan scales. (11) Tsuchiya A, Ikeda S, Ikegami N, et al. Estimating an EQ-5D population value set: Address the case of Japan. Health Econ The EuroQol web site is at http://www.euroqol 2002;11:341–353. .org/. (12) Hisashige A, Mikasa H, Katayama T.  General Health Status and Quality of Life 701 Description and valuation of health-related life in people enrolled in an AIDS clinical quality of life among the general public in trial. Qual Life Res 2002;11:273–282. Japan by the EuroQol. J Med Invest (24) Bélanger A, Berthelot J-M, Guimond E, et al. 1998;45:123–129. A head-to-head comparison of two generic (13) Dolan P, Sutton M. Mapping visual health status measures in the household analogue scale health state valuations onto population: McMaster Health Utilities Index standard gamble and time trade-off values. (Mark3) and the EQ-5D. Statistics Canada Soc Sci Med 1997;44:1519–1530. Report. Ottawa, Ontario: 2002;1–62. (14) Gudex C, Dolan P, Kind P, et al. Health (25) Oostenbrink R, Moll HA, Essink-Bot M-L. state valuations from the general public The EQ-5D and the Health Utilities Index using the visual analogue scale. Qual Life for permanent sequelae after meningitis: a Res 1996;5:521–531. head-to-head comparison. J Clin Epidemiol (15) Dolan P. Modeling valuations for EuroQol 2002;55:791–799. health states. Med Care (26) Siderowf A, Ravina B, Glick HA. 1997;35:1095–1108. Preference-based quality-of-life in patients (16) Uyl-de Groot CA, Rutten FFH, Bonsel GJ. with Parkinson’s disease. Neurology Measurement and valuation of quality of 2002;59:103–108. life in economic appraisal of cancer (27) Lubetkin EI, Gold MR. Areas of decrement treatment. Eur J Cancer in health-related quality of life (HRQOL): 1994;30A:111–117. comparing the SF-12, EQ-5D, and HUI3. (17) van Agt HME, Essink-Bot M-L, Krabbe Qual Life Res 2003;12:1059–1067. PFM, et al. Test-retest reliability of health (28) Myers C, Wilks D. Comparison of Euroqol state valuations collected with the EuroQol EQ-5D and SF-36 in patients with chronic questionnaire. Soc Sci Med fatigue syndrome. Qual Life Res 1994;39:1537–1544. 1999;8:9–16. (18) Luo N, Chew LH, Fong KY, et al. Validity (29) Hurst NP, Kind P, Ruta D, et al. Measur- and reliability of the EQ-5D self-report ing health-related quality of life in questionnaire in English-speaking Asian rheumatoid arthritis: validity, patients with rheumatic diseases in responsiveness, and reliability of EuroQol Singapore. Qual Life Res 2003;12:87–92. (EQ-5D). Br J Rheumatol (19) Brazier JE, Walters SJ, Nicholl JP, et al. 1997;36:551–559. Using the SF-36 and Euroqol on an elderly (30) Badia X, Schiaffino A, Alonso J, et al. population. Qual Life Res Using the EuroQol 5-D in the Catalan 1996;5:195–204. general population: feasibility and (20) Silberfeld M, Rueda S, Krahn M, et al. construct validity. Qual Life Res Content validity for dementia of three 1998;7:311–322. generic preference based health related (31) Johnson JA, Coons SJ. Comparison of the quality of life instruments. Qual Life Res EQ-5D and SF-12 in an adult US sample. 2002;11:71–79. Qual Life Res 1998;7:155–166. (21) Hurst NP, Jobanputra P, Hunter M, et al. (32) Johnson JA, Pickard AS. Comparison of Validity of Euroqol—a generic health the EQ-5D and SF-12 health surveys in a status instrument—in patients with general population survey in Alberta, rheumatoid arthritis. Br J Rheumatol Canada. Med Care 2000;38:115–121. 1994;33:655–662. (33) Conner-Spady B, Cumming C, Nabholtz J- (22) van Exel NJA, Scholte op Reimer WJM, M, et al. Responsiveness of the EuroQol in Koopmanschap MA. Assessment of post- breast cancer patients undergoing high dose stroke quality of life in cost-effectiveness chemotherapy. Qual Life Res studies: the usefulness of the Barthel Index 2001;10:479–486. and the EuroQol-5D. Qual Life Res (34) Conner-Spady B, Suarez-Almazor ME. 2004;13:427–433. Variation in the estimation of quality- (23) Wu AW, Jacobson DL, Frick KD, et al. adjusted life-years by different preference- Validity and responsiveness of the EuroQol based instruments. Med Care as a measure of health-related quality of 2003;41:791–801. 702 Measuring Health (35) Tidermark J, Bergström G, Svensson O, et (45) Insinga RP, Fryback DG. Understanding al. Responsiveness of the EuroQol (EQ 5- differences between self-ratings and D) and the SF-36 in elderly patients with population ratings for health in the displaced femoral neck fractures. Qual Life EuroQOL. Qual Life Res 2003;12:611–619. Res 2003;12:1069–1079. (36) Jenkinson C, Gray A, Doll H, et al. Evaluation of index and profile measures of Conclusion health status in a randomized controlled trial. Comparison of the Medical This chapter reviews some of the most recently Outcomes Study 36-item Short-Form developed measurement scales; they are in many Health Survey, EuroQol, and disease ways the showpiece of current health measure- specific measures. Med Care 1997;35:1109–1118. ment technology. Most have avoided the pitfalls (37) Oga T, Nishimura K, Tsukino M, et al. A illustrated by other instruments reviewed in the comparison of the responsiveness of book: lack of conceptual formulation, poor reli- different generic health status measures in ability and validity, and a tendency toward the patients with asthma. Qual Life Res haphazard development of alternative versions 2003;12:555–563. that are not strictly comparable. They also repre- (38) Jenkinson C, Stradling J, Petersen S. How sent the major research instruments that are be- should we evaluate health status? A ing used in evaluative studies to measure comparison of three methods in patients physical well-being: they are frequently selected presenting with obstructive sleep apnoea. in preference to those reviewed in Chapter 3. Qual Life Res 1998;7:95–100. Some reservations should be raised, however. (39) de Charro FT, Rabin R. EQ-5D from the EuroQol Group: an update. Quality of Life Despite the high standard of many quality of life Newsletter 2000;25:6–7. scales, health-related quality of life remains a (40) Krabbe PFM, Stouthard MEA, Essink-Bot somewhat flexibly defined concept. There is a M-L, et al. The effect of adding a cognitive loose consensus over the content for such in- dimension to the EuroQol multiattribute dexes, but we still lack a solid conceptual formu- health-status classification system. J Clin lation for their content. The work of the World Epidemiol 1999;52:293–301. Health Organization in developing the WHO- (41) Kind P, Dolan P, Gudex C, et al. Variations QOL has made a strong contribution in this in population health status: results from a context, and it may gain a broader influence. We United Kingdom national questionnaire have progressed since Spitzer’s complaint over survey. Br Med J 1998;316:736–741. the proliferation of quality of life measurements (42) Anderson RT, Aaronson NK, Wilkin D. during the 1980s: Critical review of the international assessments of health-related quality of life. Qual Life Res 1993;2:369–395. We are now hopefully at the peak of an epi- (43) Essink-Bot M-L, Krabbe PFM, Bonsel GJ, demic of quality of life measurement et al. An empirical comparison of four schemes. This embarrassment of riches bewil- generic health status measures. The ders clinicians and even investigators. It does Nottingham Health Profile, the Medical not favor in-depth work dedicated to valida- Outcomes Study 36-item Short-Form tion, and it militates against understanding Health Survey, the COOOP/WONCA and acceptance of these types of measures for charts, and the EuroQol instrument. Med Care 1997;35:522–537. clinical research and clinical practice . . . I (44) Paterson C. Seeking the patient’s believe that most of what needs to be done in perspective: a qualitative assessment of health services research can be accomplished EuroQol, COOP-WONCA charts and with three or at most four or five thoughtfully MYMOP. Qual Life Res developed and carefully validated measures of 2004;13:871–881. health status.” (1, p469)  General Health Status and Quality of Life 703 Perhaps the term “general health measurements” mental illnesses, especially those living in the should be applied to scales that cover physical, community (2). It describes the person’s general mental, and social functioning, whereas “quality living situation, relationships, finances, and of life” should be reserved for scales that are even work, rather than forming a health status mea- broader in scope, providing, in addition, overall sure, and so appeared unsuitable for detailed re- ratings of environmental quality, subjective well- view here. being, and life satisfaction. The emphasis in the coming years should continue to be less on devel- oping new measurements of these themes than on References agreeing on definitions and then on analyzing which of the existing indexes best reflect those (1) Spitzer WO. State of science 1986: quality of life and functional status as target concepts. Only when a definite gap in the coverage variables for research. J Chronic Dis of existing indexes is identified should we attempt 1987;40:465–471. to develop new quality of life measurements. (2) Lehman AF. Quality of Life Interview Among those scales not included in this chap- (QOLI). In: Sederer LI, Dickey B, eds. ter, we may mention Lehman’s Quality of Life Outcomes assessment in clinical practice. Interview (QOLI). This is used to describe the Baltimore: Williams and Wilkins, current life circumstances of people with severe 1996:117–119.  11 Recommendations and Conclusions The Current Status of Health to hope that the comments contained in reviews Measurement such as this book will encourage the fuller devel- opment of the less adequate methods. Two devel- In the ten years since the second edition of this opments merit especial mention: the increase in book was written, marked improvements have international standardization of measures, and been made in the quality, sophistication, and the development of banks of items in place of set standardization of health measurement scales. measurement instruments. Health measurement has increasingly taken ad- One of the clearest developments in recent vantage of the technical advances in test con- years has been the collaborative efforts to develop struction established in the social sciences, internationally comparable instruments. There is although the application of this knowledge to a growing trend away from merely translating health measurements remains somewhat uneven. English-language instruments into other lan- In the 1987 edition of Measuring Health, we ar- guages (which was carried to extremes by instru- gued that pain scales were the most successful in ments such as the General Health Questionnaire exploiting more sophisticated scaling tech- or the Functional Assessment of Cancer Ther- niques, whereas physical disability measures apy). Instead, the World Health Organization were for the most part unsophisticated. Over the (WHO) developed the WHOQOL measure si- intervening years, measurements of physical dis- multaneously in a wide variety of countries and ability have begun to catch up, and scales such cultures, in a genuinely democratic process of as the Functional Autonomy Measurement Sys- discussion and negotiation over the content of tem (SMAF) reviewed in Chapter 3 contain the instrument. The EuroQol offers a similar, al- some very innovative approaches. It is also good though narrower, example. In part stimulated by to be able to report a marked increase in com- multinational pharmaceutical trials that require parisons between measurement instruments: in- equivalent outcome measures in several lan- creasingly we now know how to map one scale guages, organizations such as the Mapi Research onto another, illustrating the strengths and limi- Institute in Lyon, France, have been developed. tations of each. Major steps have been made in The Mapi Institute includes an information re- scoring health measures; item response theory source center that collects and distributes infor- and econometric scaling methods are becoming mation on health measures. It possesses a common and norm-referenced scoring is begin- collection of questionnaires and can assist users ning to be applied. Although econometric and in finding an instrument for their needs. The psychometric approaches come from different Mapi also coordinates the translation of existing academic traditions, their meeting in the health instruments into other languages; their Web site measurement field appears to be encouraging offers a valuable source of information on the some melding of the two. Kaplan et al., for ex- rapidly evolving field of translation. Mapi pro- ample, illustrated how common ground may be duces a Quality of Life Newsletter that encour- found between them (1). All in all, things are ages rapid dissemination and exchange of moving, and in the right direction. We continue information on health outcome measurement 704  Recommendations and Conclusions 705 (see www.mapi-research-inst.com). Also in Eu- semble information on item characteristics for rope, the Harmonization Project for Instruments every item in our current leading tests; they then in Dementia (EURO-HARPID) has undertaken need to be cross-calibrated to understand the empirical comparisons across language versions conditional probabilities of responses to items of dementia scales and has established standard from different existing tests. An example was administration procedures (2). In the United given by Ware et al. in separate studies for mea- States, the Medical Outcomes Trust (www.out sures of headache severity that compares items comes-trust.org/) was incorporated in 1992 to drawn from five different measures (4; 5), while promote the science and application of outcome Cella has described a vision of the future of item measures; as part of this mission, it has developed banking in health measurement (http://out an instrument library and has proposed quality comes.cancer.gov/conference/irt/cella_et_al.pdf). guidelines for instrument developers (3). Dynamic testing will no doubt evolve rapidly, The other innovation has been to take advan- but an example of its potential is given on the tage of computers to administer tests, not merely website www.amIhealthy.com (accessed in late to administer existing instruments, but to cus- 2004). tomize sets of items drawn from a range of mea- Although we may complain about the weak- sures. The goal is to administer items that ness and lack of coordinated development work maximize the information gained from each: in certain areas of health measurement, it is also successive items are selected as being optimal in true that a universal perfect index can never ex- providing the most information, after consider- ist. It is quite wrong to imagine one set of ques- ing the responses to previous items. This cus- tions suited to all diseases, all individuals, and tomization by computerized adaptive testing all applications. Such an instrument would have (CAT) can save time and achieve greater effi- to make so many compromises it would proba- ciency of test information for a given test length bly not be suitable for any particular application. and can provide automatic scoring, greater pri- Fundamentally different scales will be required vacy for the respondent, and the possibility of for policy analysis and for individual patient providing immediate feedback. The future vision evaluation; we will continue to have generic and for health measurement is that item banks will disease-specific scales or item pools, health in- contain information on hundreds of items, indi- dexes and health profiles, subjective and objec- cating their psychometric and scaling properties tive measures. Each has its place, although and relationship to other items. The application certain quality control procedures can be fol- of item response theory has been central to the lowed in developing health indexes of any type. analysis of item characteristics across existing Given some successes and some areas of weak- measurement instruments. A CAT algorithm ness, what should be done to strengthen this then selects items from the bank so that redun- field? dant questions are not asked (if a person cannot walk a block, there is little reason to ask him if he can run 100 meters). Using the logic of a Guidelines for Developing Health screening test, sensitive (but perhaps not spe- Measurements cific) filter items can identify areas in which the respondent reports problems and these can be Ensuring the quality of health measurements is probed in more detail, whereas further questions supremely important, because decisions affect- on areas in which they report no problems can ing the welfare of patients and the expenditure automatically be skipped. The great gain is in ef- of massive public funds are based on the results ficiency: by asking only questions that are perti- of such measures; and pressure to monitor the nent to the respondent’s current level of health, outcomes of treatment is virtually universal. much greater precision and discriminal ability is Over the nearly 20 years since the first edition of achieved for any set length of interview. To Measuring Health, we have witnessed the steady achieve this goal, we need to systematically as- growth in a science of health measurement, 706 Measuring Health equivalent to psychometrics or econometrics. in the standards of the American Psychological Several medical disciplines have now established Association (14), which guided the following measurement standards (6–10), and some are suggestions: undertaking comparative reviews of measure- ments against set review standards (11). Many 1. Published articles or a manual should in- measures now include administration manuals; clude a full description of the purpose of several are excellent (12; 13). More sophisticated the method, specifying the population for analytic procedures, such as item response the- which it is designed, the populations on ory and differential tem functioning, are becom- which it has been tested, and the intended ing the norm. New approaches, such as item use for the data collected. banking described above, reflect a sensible con- 2. Make the definitive version of the instru- centration on further improving current meth- ment readily available to users, ideally ods in place of developing new scales. through the Web, or by distributing it at For those who do develop new instruments, cost. However, a lesson from several mea- however, guidelines can address four common sures reviewed in this book is that variant shortcomings in health indexes: unclear concep- versions proliferate if central control is tual formulation of the method, inadequate de- not applied, causing confusion where velopment and testing of the instrument, these variants are still described in publi- inadequately detailed published descriptions of cations as being the original scale. Many it, and a lack of leadership in ensuring continued measures therefore allow users free access development and promotion of the method. but to require them to register and obtain This book has repeatedly mentioned the de- permission to use the instrument, and sirability of a conceptual definition of the topic agree to apply it in a standard manner. It being measured. This is intended to stress the is hoped that, the health measurement role of measurement in scientific discovery: as field will avoid the tendency among psy- science ultimately tests theories, we must know chological test publishing houses of copy- what theoretical orientation each health index righting scales (some of which were represents. This goal has been quite well developed using public funds and have achieved in the various fields of mental health been used freely for years) and erecting measurement, and to a limited extent in quality needless barriers to their use. The argu- of life measures. Many measurements of func- ment that scales can only be interpreted tional disability, however, pay little more than lip by a qualified expert does not apply in the service to the idea of a conceptual approach to field of general health measurements or the topic: the WHO definition may be given screening scales: these are not diagnostic passing mention, or the distinction between dis- instruments. A current copy of the scale ability and handicap. Although a useful start, should be included in the manual or Web this could be refined to indicate more closely site; some sites make it needlessly difficult what questions should be included and why. to figure out which is the definitive version Providing adequate information on how to of a measure. administer and interpret a measure is crucial in 3. Measures should be given a name that ac- achieving standardization. It is always hard to curately conveys their content; some scales describe a measurement in sufficient detail of activities of daily living were never even within the constraints of a journal article, so named, making it difficult for subsequent most measures produce a Web site or manual de- users to indicate which scale they used. If scribing the method and its administration. If a a scale has been revised or abbreviated, measurement has been adequately tested, it is this should be indicated by a version num- certainly worth the additional effort of prepar- ber or a change in the name. At the same ing a manual to describe it. Several pertinent time, authors of instruments should delay suggestions on preparing manuals are included presenting revised versions until there is a  Recommendations and Conclusions 707 clear advantage in doing so. The descrip- Survey should be heeded; many validation tion of the method should also outline studies were wasted because they argued pitfalls in its administration and interpre- over what turned out not to be compara- tation and should show how high and low ble versions of the instrument; the Life scores are to be interpreted. Authors Satisfaction Index has also suffered in this should remain cautious over making exag- way. Users commonly want an abbrevi- gerated claims of the applicability of their ated version of a measurement and will scales; a lot of confusion was caused, for abbreviate it themselves if published short- example, when it was claimed that the forms are not available. Goldberg success- Nottingham Health Profile could serve as fully forestalled this problem by a survey instrument—a claim that later recommending standard abbreviations for had to be withdrawn. the General Health Questionnaire, as 4. A rationale should be given for the design Ware has done for the reports on SF in- of the instrument; what conceptual defini- struments. Both of these scales have also tion of the topic of measurement does it estimated the loss of precision that may reflect? For example, descriptions of an arise from using the abbreviated versions. anxiety measurement should outline the 7. Instructions must be clear enough to en- theoretical approach it takes to the con- sure standard administration and scoring cept. Specification in this detail has been of the method. Responses can be very sen- achieved for relatively few of the indices sitive to the way in which a measure is pre- that we review; the scales of Bradburn, of sented, and even for a self-administered Bush and Kaplan, and of Melzack provide questionnaire, the precise instructions for examples of clear conceptual formula- the respondent should be shown. Likewise, tions. the setting in which the method was ad- 5. The way in which questions were selected ministered during the validation studies should be indicated: where did they come must be described. A good example of at- from and how were they sampled? Re- tention to detail of this type was given in viewers often have a difficult time in the Structured and Scaled Interview to As- trying to trace the ancestry of subcompo- sess Maladjustment. Published descrip- nents in a scale; there should be no shame tions should provide details of how in simply stating where the items came missing data should be handled, of how from. The manual should describe the change scores should be calculated (e.g., as procedures used to develop the instru- absolute changes or as a percentage of the ment; good illustrations of this are con- initial score?) When there are alternative tained in descriptions of the General approaches to scoring, the advantage of Health Questionnaire, of the SF-36, and each needs to be given and their agreement of the Sickness Impact Profile. The con- shown; Melzack did this for the McGill tent of most successful questionnaires has Pain Questionnaire. generally been established by a team of cli- 8. All measures should assemble reference nicians, measurement experts, and pa- scores from several populations. Ideally, tients. these should include a population sample 6. Revisions to the method should be clearly of healthy people and samples from vari- explained and data on the reliability and ous patient groups, to provide yardsticks validity of the latest version should be pre- against which to interpret the results of sented. Academic pressures encourage the subsequent studies. This has been done for publication of preliminary versions; the the SF-36 (12), whereas reference stan- danger is that users may continue to use dards showing the typical improvement outdated versions of a measurement. The following treatment have been described cautionary tale of the Health Opinion for the Functional Independence Measure 708 Measuring Health (15; 16). Various other measures have a correlation coefficient is not sufficient been applied in large studies, but these to indicate the agreement between them. have not been exploited for the produc- Much more useful information can be tion of reference standards. gained from using item response theory 9. The validity and reliability testing should methods to illustrate the range of cover- examine both the internal structure of the age of two (or more) instruments. An ex- method (its internal consistency and fac- ample was given by Jette et al., who used tor structure) and also its relation to alter- Rasch analysis to place rival scales on a native measurements of the same concept. common metric, illustrating the range of Measures of psychological well-being coverage of each scale (19). (Chapter 5) are often validated almost ex- 12. Each measurement method should be clusively in terms of their factor structure, tested by users other than the original while we know little of how they relate to authors, indicating that it holds a wide diagnostic criteria. In criterion validation, appeal and that it can be used success- the rationale for the selection of the crite- fully by others to provide consistent re- rion scores must be given, and attention sults. It is almost a law of health must be paid to their validity. The ex- measurement that the original authors pected level of correlation should be speci- achieve higher validity figures than sub- fied before the study is undertaken; good sequent users do. examples include the work of McHorney 13. The most successful instruments are et al. on the SF-36 (17). All too often au- those for which the authors take long- thors present a list of miscellaneous corre- term responsibility for the further refine- lation coefficients from which they ment of the method. Few, if any, conclude, magically, that the results measures are perfect when first pub- demonstrate the construct validity of the lished: most of the leading scales have instrument. This is logically inadequate undergone revisions. Some scales (e.g., and it precludes refutation of validity. Hamilton Rating Scale for Depression) Construct validity should include tests of clearly filled a need and were of good discriminant and of convergent validity, quality; no one, however, took responsi- always with an advance indication of how bility for continuing the development ef- much discrimination is expected. More fort. When there is no coordination of formal testing requires the construction of after-publication development work, a multitrait-multimethod matrix and the users make piecemeal changes in an un- specification of patterns of correlations coordinated fashion and rival versions within the various sectors of the matrix. proliferate, making comparisons be- 10. As is increasingly the case, formal statis- tween studies uncertain. This is unfortu- tics should be reported on the standard nate, because the desire to improve a error of the measurement, on sensitivity method attests to its basic soundness (for to change, effect sizes and reliable change otherwise it would be abandoned) and to (18). the demand for it. Like children, health 11. While there has been a recent increase in measures require care through adoles- the numbers of studies that compare rival cence and those who develop measures measurement methods, there is still a ten- should recognize the need to remain in dency in some fields to avoid head-to- charge of them over the long term; al- head comparisons of scales. Ultimately, though this poses funding difficulties, this is what users need to guide their several teams have managed and exam- choice among alternative methods. Fur- ples may be found in every chapter in this thermore, when measures are compared, book.  Recommendations and Conclusions 709 Final Remarks attributes and review criteria. Qual Life Res 2002;11:193–205. These guidelines are exacting and may repre- (4) Ware JE, Jr., Kosinski M, Bjorner JB. Item sent an ideal that only a few scales have banking and the improvement of health status measures. Quality of Life Newsletter achieved. And yet the difficulty of the task 2004; Fall (special issue):2–5. should not condone inferior work: why should (5) Bjorner JB, Kosinski M, Ware JE, Jr. Using health be measured any less accurately than item response theory to calibrate the personality or achievement? The impression as Headache Impact Test (HIT) to the metric of early in 2005 is that exciting advances have of traditional headache scales. Qual Life been made and that there are strong prospects Res 2003;12:981–1002. for the well-being of this field. Increasingly, (6) Task Force on Standards for Measurement comparative information on the various scales in Physical Therapy. Standards for tests and in each domain of health is being assembled, measurements in physical therapy. Phys and innovative approaches such as item bank- Ther 1991;71:589–622. ing combined with computer-assisted test ad- (7) Johnston MV, Keith RA, Hinderer SR. ministration may lead to a shift away from the Measurement standards for interdisciplinary medical rehabilitation. historical insularity and rivalry of the different Arch Phys Med Rehabil measurement tools. The journal Quality of Life 1992;73(suppl):S3–S23. Research has assumed a leadership position in (8) Kraemer HC. Evaluating medical tests: publishing articles on measurement. Organiza- qualitative and objective guidelines. tions such as the Mapi Institute are coordinat- Newbury Park, California: Sage, 1992. ing the exchange of ideas and supporting (9) American Physical Therapy Association. training. The international perspective of many Standards for tests and measurements in scales is exciting. The Web can be used as a physical therapy practice. Phys Ther valuable repository of information on the mea- 1991;71:589–622. sures, and many scales have excellent Web (10) Staquet M, Berzon R, Osoba D, et al. sites; others have not. It can be further used for Guidelines for reporting results of quality dynamic testing and scoring of measures, and of life assessments in clinical trials. Qual Life Res 1996;5:496–502. it is obvious that information sources such as (11) Peña-Casanova J. Alzheimer’s Disease this book should be made accessible on the Assessment Scale-Cognitive in clinical Web so that they may be kept continuously up practice. Int Psychogeriatr 1997;9(suppl to date. 1):105–114. (12) Ware JE, Jr., Kosinski M, Gandek B. SF-36 Health Survey: manual and interpretation References guide. Lincoln, Rhode Island: (1) Kaplan RM, Feeny D, Revicki DA. QualityMetric Inc., 2002. Methods for assessing relative importance (13) Parkerson GR, Jr. Users guide for Duke in preference based outcome measures. health measures. Durham, North Carolina: Qual Life Res 1993;2:467–475. Department of Community and Family (2) Verhey FR, Houx P, van Lang N, et al. Medicine, Duke University Medical Center, Cross-national comparison and validation 2002. of the Alzheimer’s Disease Assessment (14) American Psychological Association. Scale: results from the European Standards for educational and Harmonization Project for Instruments in psychological testing. Washington, DC: Dementia (EURO-HARPID). Int J Geriatr American Psychological Association, 1985. Psychiatry 2004;19:41–50. (15) Granger CV, Hamilton BB. The Uniform (3) Scientific Advisory Committee and the Data System for Medical Rehabilitation Medical Outcomes Trust. Assessing health report of first admissions for 1991. Am J status and quality-of-life instruments: Phys Med Rehabil 1993;72:33–38. 710 Measuring Health (16) Long WB, Sacco WJ, Coombes SS, et al. physical and mental health constructs. Determining normative standards for Med Care 1993;31:247–263. Functional Independence Measure (18) Erickson P. Assessment of the evaluative transitions in rehabilitation. Arch Phys properties of health status instruments. Med Rehabil 1994;75:144–148. Med Care 2000;38(suppl II):II-95–II-99. (17) McHorney CA, Ware JE, Jr., Raczek AE. (19) Jette AM, Haley SM, Ni P. Comparison of The MOS 36-Item Short-Form Health functional status tools used in post-acute Survey (SF-36): II. Psychometric and care. Health Care Financ Rev clinical tests of validity in measuring 2003;24:13–24.  Glossary of Technical Terms Alpha Cronbach’s alpha is a generalized formula cates the degree to which two or more sets of used to express the internal consistency relia- observations fit a linear relationship. There bility of a test. are various formulas for estimating the Area under the ROC Curve (AUC). See Receiver strength of the correlation; in each case the Operating Characteristic Curve. range lies between −1 and +1. A correlation Attenuation of Correlations Validity correlations close to zero indicates no association between between two scales are always limited by the the observations; as correlations rise, it be- unreliability (i.e., random error) in each scale. comes more possible to predict the value of A correction for this attenuation may be ap- the second observation from a knowledge of plied to show what the underlying association the first. The formula most commonly en- between the scales would be. A formula is countered is Pearson’s r, suited to data mea- shown on page 35. sured at an interval or ratio scale level. Category Scaling See Scaling. Kendall’s tau and Spearman’s rho correlation Coefficient of Concordance While a rank order formulas may be used to indicate the asso- correlation shows the agreement between ciation between variables measured at the two sets of rankings, Kendall’s coefficient of ordinal level, and are termed “rank order cor- concordance (W) provides a measure of the relations.” relationship among several rankings of ob- Criterion Validity Validity indicated by compar- jects or individuals. It is the nonparametric ing the results obtained using a measurement equivalent of the intraclass correlation. scale with a “criterion standard” or indicator Concurrent Validity Validity indicated by com- of the true situation. paring a scores on a measurement with those Discriminant Analysis A multivariate statistical obtained by applying alternative, equivalent procedure that indicates how adequately a set measurements at the same time. of variables (here, typically, the replies to Construct Validity Used when there is no crite- questions in a health measurement) differen- rion against which to evaluate the validity of tiate between two or more groups of people a measurement. Validity assessed by compar- who are known to differ on some characteris- ing the results of several contrasting tests of tic (here, typically being sick or well). The validity (including concurrent, convergent, analysis selects the set of questions that show and divergent validation studies) with predic- the most marked contrast in the pattern of tions from a theoretical model. replies between the groups (i.e., the most dis- Content Validity The extent to which a measure- criminative questions). ment covers all aspects of the topic it pur- Discriminant Validity The extent to which ports to measure. scores on a measurement distinguish between Convergent Validity The extent to which two or individuals or populations that would be ex- more instruments that purport to be measur- pected to differ (e.g., people with or without ing the same topic agree with each other. a disease). Correlation A measure of association that indi- Effect Size An standardized indicator of the abil- 711 712 Measuring Health ity of scores on a measure to distinguish be- eralizes this procedure and expresses the tween two groups (typically, cases and con- agreement among more than two raters. Un- trols). A common formula for effect size is like the Pearson correlation, however, the (M1 − M2)/s, where M1 and M2 are the mean intraclass coefficient is a measure of agree- scores for the two groups and s is the pooled ment that records the average similarity of standard deviation. Being standardized, effect raters’ actual scores on the ratings being com- sizes may be directly compared between dif- pared. ferent measurement instruments. Effect size Item The term “item” is used to refer to individ- statistics can be translated into sensitivity and ual questions or response phrases in any specificity values. Effect sizes of 0.2 to 0.49 health measurement. It replaces the more ob- are generally considered small; 0.5 to 0.79 vious term “question” simply because not all are moderate, and 0.8 or above are large (see response categories are actually phrased as Kazis LE et al. Effect sizes for interpreting questions: some use rating scales and others changes in health status. Med Care 1989; use agree/disagree statements. 27(suppl):S178–S189). Item-Characteristic Curve A graphical plot of the Equal-Appearing Interval Scales See Scaling. probability that respondents will respond Factor Analysis Factor analysis, like principal positively to an item across levels of the char- component analysis, is a mathematical tech- acteristic (e.g., depression) being measured. nique that reduces a large number of interre- The resulting curves slope more or less lated observations to a smaller number of steeply upward in an ogive form from the common dimensions or factors. A factor is a lower left to the upper right corner of a cluster of variables (here, items on a health graph. The placement of the curve along the measurement instrument) that are highly re- horizontal axis indicates the severity of the lated to each other. As an example, a factor characteristic (here, depression) that is identi- analysis of questions asked to assess intelli- fied by the item. The steepness of the curve gence might identify discrete groups of ques- where it crosses the center point of the verti- tions that assess verbal ability, numerical cal axis indicates the sharpness with which ability, and visual-spatial judgments. The fac- the item discriminates people higher and tors are composed of measurements or vari- lower on the characteristic being measured. ables that intercorrelate but that are distinct This is related to the sensitivity and specificity from variables on other factors. of the item. The curves are produced by item Goal Attainment Scaling An evaluation method response theory analyses (q.v.). that assesses the efficacy of a program in at- Item Response Theory (IRT) A theory that taining predetermined goals. guides empirical analyses of the connection Guttman Scaling See Scalogram Analysis. between items and the underlying character- Internal Consistency See Reliability. istic (or “latent trait”, q.v.) being measured. Inter-rater Reliability The extent to which re- Generalized linear analytic methods based on sults obtained by different raters or inter- IRT are used to produce scores for respon- viewers using the same measurement method dents, interval-scaled values for the response will agree. The agreement is appropriately categories for each item, and also indicators calculated using the intraclass correlation of item discrimination (see Item- when the measurement provides interval-level Characteristic Curve). Based on a measure- data. ment model originally developed by Georg Interval Scale See Scales of Measurement. Rasch, IRT is used to analyze items during Intraclass Correlation In testing the reliability test development to produce a health measure of a measurement, correlation coefficients that taps a single dimension of health and to such as Pearson’s r may be used to com- select an optimum set of items evenly spaced pare the ratings of a number of patients made across the continuum being measured. by two raters. The intraclass correlation gen- Item-Total Correlation The correlation between  Glossary of Technical Terms 713 each item or question in a health measure- ues; the curve that results is known as a re- ment and the total score; used as an indica- ceiver operating characteristic (ROC) curve. tion of the internal consistency or The likelihood ratio expresses how much the homogeneity of the scale, suggesting how far odds of the patient having the disease in- each question contributes to the overall crease following a positive test score, or de- theme being measured. crease following a negative test score. This Kappa As a coefficient of agreement between way of presenting validity data may aid in se- two raters, kappa expresses the level of agree- lecting the optimal cutting point, as described ment that is observed beyond the level that by McNeil BJ et al. (Primer on certain ele- would be expected by chance alone. A typical ments of medical decision making. N Engl J formula is Med 1975;293:211–215.) See also Sensitiv- ity, Specificity. K = (po − pc)/(1 − pc), Logit A unit of measurement used in item re- where po is the observed proportion of agree- sponse theory analyses (q.v.). A logit is a ment and pc is the proportion of agreement transformation of a probability value into a expected by chance alone. Chance agreement linear continuum. may be thought of as the agreement that Magnitude Estimation An approach to scaling would occur if one rater merely guessed or (q.v.) in which the respondent expresses as a flipped a coin to make his ratings. The pc is ratio how much better or worse each state be- assessed as follows: ing scaled is compared to a reference stan- dard. pc = p1p2 + (1 − p1)(1 − p2), Minimally Important Difference (MID), or Min- where p1 is the probability of rater 1 diagnos- imally Clinically Important Difference ing a case, and p2 is the equivalent probability (MCID) The smallest change on a health out- for the second rater. Although in theory, the come measure that would be considered im- range of kappa is from 0 to 1, in practice its portant by patient or clinician. A difference upper value is limited by the sensitivity and that would, for example, be considered as in- specificity of the test. (See Grove WM et al. dicating progress in treatment, or (if it was in Reliability studies of psychiatric diagnosis. a negative direction) as justifying a change in Theory and practice. Arch Gen Psychiatry therapy. This concept was introduced to 1981;38:408–413.) counterbalance an exclusive emphasis on sta- Kendall’s Tau See Correlation. tistical inference: in very large studies, rela- Latent Trait The unobservable continuum (e.g., tively unimportant changes may be pain, or health) that items in a test are in- statistically significant. The term “clinically” tended to measure. Latent trait analysis evalu- was dropped from MCID to avoid the im- ates how consistently the items measure a pression that the importance of a difference single trait. See Item Response Theory. in scores was determined only by the clini- Likelihood Ratio An approach to summarizing cian. the results of sensitivity and specificity analy- Multitrait-Multimethod Matrix A format for ses for various cutting-points on diagnostic presenting validity and reliability correlations or screening tests that also takes account of in which the agreement among several mea- the prevalence of the condition in the popula- surement methods (hence, multimethod) as tion under study. Each cutting-point pro- applied to several traits (multitrait) is shown duces a value for the true positive ratio (i.e., in a manner that simplifies the interpretation sensitivity) and the false positive ratio (i.e., 1- of construct validity. It is assumed, for exam- specificity). The ratio of true to false positives ple, that the correlations between different is the likelihood ratio for each cutting-point. measurement methods will be higher when These values may be plotted on a graph applied to the same topic of measurement whose axes show true and false positive val- than when applied to different topics. (A 714 Measuring Health clear example of the approach and the under- internal consistency of the method: how far lying assumptions is given by Campbell DT, the questions it contains all measure the Fiske DW. Convergent and discriminant vali- same theme. dation by the multitrait-multimethod matrix. Reliable Change, or Reliable Individual Change Psychol Bull 1959;56:81–105). Because of random error in any measure- Ordinal Scale See Scales of Measurement. ment, a small change in scores (e.g., before Path Analysis A procedure for testing causal hy- and after a treatment) might be due merely to potheses that indicates the extent to which a chance error in measurement. Using the stan- hypothesized causal pattern fits empirical dard error of measurement (q.v.), it is possi- data. Path analysis could, for example, be ble to calculate the width of a band of scores used in analyzing a set of data to calculate the either side of zero change, beyond which a relative strength of the causal influence ex- change in scores is unlikely to be due to erted by smoking, obesity, sedentary living, chance alone. The reliable change index and cholesterol levels in predicting cardiovas- (RCI) is the smallest number of scale points’ cular disease. The strength of each causal in- change that represents a real change, as op- fluence is indicated by path coefficients, posed to chance variation. Reliable change which are derived from standardized regres- scores are usually reported in the form of a sion coefficients. threshold value (e.g., ±4). Pearson Correlation See Correlation. Response Shift The tendency of patients who are Positive Predictive Value The proportion of all gravely sick to alter their expectations about the people who were identified by a measure- health and quality of life. People make sub- ment or screening test as apparently having jective ratings of health compared with an in- the disease who actually do have it. ternal frame of reference that can shift as a Predictive Validity The accuracy with which a person experiences illness. Thus, a person’s measurement predicts some future event, judgment of the severity of a pain may alter such as mortality (see Chapter 2, page 31). after they have experienced very severe pain. Rank Order Correlation See Correlation. A patient may deal with a terminal illness by Rasch Analysis See Item Response Theory recalibrating their expectations so may thus Receiver Operating Characteristic (ROC) Curve provide paradoxically positive responses to a An analysis of the validity of a screening test quality of life questionnaire. that combines indicators of sensitivity and Responsiveness For outcome measurements, the specificity. The true-positive rate (sensitivity) ability to detect change is a crucial character- is plotted graphically against the false- istic, and many indicators of responsiveness positive rate (1-specificity) for a range of have been proposed. These are generally cutting-scores on the screening test. A statisti- identical to effect size measures; the numera- cal summary of the overall performance of a tor is the raw score change (typically before test is given by calculating the area under the and after treatment). The denominator may ROC curve (AUC); this statistic runs from be the standard deviation of the measure at 0.5, indicating prediction no better than baseline or the standard error of scores; care- chance, to 1.0 (perfect accuracy). See also ful reading of the article is often required to Likelihood Ratio. see which approach was used. See Chapter 2, Reliability The proportion of variance in a mea- pages 37–39. surement that is not error variance (see page Rho Spearman’s rho correlation formula is used 40). Reliability can be assessed in many to indicate the association between variables ways, each of which differs in the definition measured at the ordinal level, and is termed a it implies of error variance. Commonly, reli- “rank order correlation.” See Scales of Mea- ability refers to the stability of a measure- surement; Correlation. ment: how far it will give the same results on Ridit Sometimes used in presenting the results of separate occasions. This is influenced by the health indices, a ridit is a way of expressing  Glossary of Technical Terms 715 the observed score relative to an identified ful to speak of one value being so many population (hence “ridit”). The average ridit times greater or less than another value. calculated for the group of interest shows the All arithmetical operations, including probability that a member of that group is multiplication and division, may be ap- “worse off ” than someone in the identified, plied, and all types of statistical analysis reference population distribution. As an ex- may be used. ample, if the average ridit for a subgroup is Scaling A set of procedures used to assign nu- 0.625, then 62.5% of the people in the refer- merical weights to indicators of health status ence population have a better score (e.g., are to reflect the quality of life or severity of dis- less sick) than the average individual in the ability implied. It is conventional to use subgroup. weights in scoring the answers to questions ROC See Receiver Operating Characteristic. (for example, mild pain might be scored 1, Scalability Coefficient See Scalogram Analysis. and severe pain 6.) Less commonly, weights Scales of Measurement The mathematical quali- may be used in rating the relative severity of ties of numerical measurement scales vary different questions, typically giving more and are of four main types. weight to more serious disabilities in forming 1. Nominal scales. Numbers are assigned ar- a summary score. By no means do all health bitrarily with no implication of an inher- indexes use item weights of this type: many ent order to their categories, as in simply count up symptoms or areas of dis- telephone numbers. Such scales may only ability and so are viewed as an ordinal scale. be used as classifications; no statistical Where item scaling is used, methods for analyses may be carried out that use the deriving weights are of two broad types– numerical characteristics of the scale. category scaling (such as Thurstone’s “equal- 2. Ordinal scales. Classification into a scale appearing intervals” procedure), which pro- that implies a distinct order among the duces weights at an interval scale level, and categories (e.g., building numbers on a magnitude estimation, which provides a ratio street), but where there is no assumption scale. An index that uses category scaling is concerning the relative distance between the Sickness Impact Profile (Chapter 10); the adjacent values. Statistical methods such Pain Perception Profile is a scale that uses as rank order correlations may be used, magnitude estimation (Chapter 9). These il- but addition and subtraction, or calcula- lustrate the psychometric approach to scal- tion of averages, may not be appropriate. ing; the econometric approach estimates 3. Interval scales. Interval scales are so weights (termed “utilities,” q. v.) for patterns named because the distance between adja- of scores rather than individual items. More cent numbers in one region of the scale is detail is given in Chapter 2. assumed to be equal to the distance be- Scalogram Analysis Also known as Guttman tween adjacent numbers at another region scaling, this method of analysis is used to se- of the scale (as in Fahrenheit or Celsius lect questions that lie in a hierarchical order scales). Addition and subtraction are per- such that agreeing to an item at the severe end missible, but not multiplication or division of the scale implies a positive answer to all of such scales; statistical analyses such as other items on the scale. For example, a scale the Pearson correlation, factor analysis, or of functional ability may ask about walking discriminant analysis may be used with in- ability using items like “I can walk a block or terval scales. more,” “I can walk 100 yards,” “I can walk 4. Ratio scales. A ratio scale is an interval to my front door,” and “I can move around scale with a true zero point, so ratios be- in my room.” Statistical techniques developed tween values are meaningfully defined. by Guttman and others analyze the pattern of Examples include weight, height, and in- responses to such items and show how come, because in each case it is meaning- closely they lie in a consistent hierarchy of 716 Measuring Health severity: are there respondents who said, “I a measurement means that an individual can walk a block or more” but said they score may not be precisely correct; some error could not walk 100 yards? If so, the intention is to be expected. The amount of error that of the scale to measure varying levels of a sin- may be expected, due to chance alone, is rep- gle dimension may not have been met, per- resented by the standard error of measure- haps because some respondents do not ment (SEM), the standard deviation of errors. understand 100 yards, or because they feel A perfectly reliable instrument would have a that a “block or more” may be less than 100 SEM of zero and all variation would reflect yards. Whatever the reason, the items may be true differrences. Hence, the SEM can be ap- reworded during the test development pro- proached in terms of variances: it is the over- cess. The hierarchical consistency is evaluated all variance less the variance between by coefficients of scalability and of repro- subjects. Alternatively, it may be estimated as ducibility that range from 0.0 to 1.0. The co- the standard deviation × √ (1 − reliability) , us- efficient of scalability indicates how far the ing estimates of standard deviation and of re- questions form a cumulative scale, and liability from a large population sample. should exceed 0.6 if the scale is truly unidi- Standard Gamble An approach to scaling utili- mensional. The coefficient of reproducibility ties (q.v.) in which the judges are asked to shows how accurately the scale score indi- make the imaginary choice between remain- cates a person’s entire pattern of responses. In ing in an impaired condition for the rest of a valid scale the coefficient of reproducibility their lives, versus undergoing an operation will fall above 0.9. that is likely to cure them, but which also Sensitivity The ability of a measurement or carries a specified risk of operative mortality. screening test to identify those who have a The operative mortality risk is adjusted until condition, calculated as the percentage of all the judge has an equal preference for the op- cases with the condition who are judged by eration or for remaining with the chronic the test to have the condition. A test with condition. The level of risk accepted to es- high sensitivity will miss few real cases, and cape the condition is used as a marker of the so produce few “false-negative” judgments. person’s utility for the condition. This leads to a mnemonic, as the complement Standard Scoring One cannot make a direct of seNsitivity is the false Negative rate. comparison between scores on different mea- Spearman Correlation See Correlation. surement instruments unless they use an Specificity The ability of a measurement to cor- equivalent range of scores. A way to achieve rectly identify those who do not have the con- this equivalence is to transform scores into a dition in question. The word “specificity” standard metric that permits direct compari- refers to how narrowly a test is targeted: does son between different scales. This requires it only identify people with a particular type knowing the mean score and standard devia- of disease (is it specific to that condition)? As tion from a suitable reference group (these a mnemonic, the complement of sPecificity is are shown in the reviews of many of the in- the false Positive rate: the numbers of people struments). To calculate a standard score, the the test falsely classifies as having the disease. mean score from the reference group is sub- The bane of all screening tests is that as tracted from the individual’s score, and the sensitivity rises, specificity falls. A metaphor result divided by the standard deviation. This might be using a flashlight to find your keys. produces a positive score for those above the A flashlight with a wide beam (sensitive) will mean; the overall mean of the standard scores light up a large area and hence increase the will be zero and the measurement unit is stan- chance of finding the keys, but it will also dard deviations from the population mean of shine on sticks and stones, etc., which will whatever you are measuring (e.g., pain, de- distract you. pression, elevated blood pressure). The result Standard Error of Measurement Unreliability in is a z score, which can be compared to statis-  Glossary of Technical Terms 717 tical tables to estimate what fraction of peo- different health states. Approaches that de- ple in the population would achieve a score as rive from psychometrics estimate values, high or higher than the person’s score (this re- which refer to a person’s enduring prefer- quires making an assumption that the distri- ences, under conditions of certainty. The bution is normal). If required, the z scores econometric approach focuses on choices and can be further transformed to provide a scale records utilities, which refer to the strength with any chosen mean and standard devia- of a person’s preference for a particular out- tion; this is called a T-score. For example, IQ come when making a choice under conditions scores have traditionally been presented with of uncertainty. This is relevant in clinical de- a mean of 100 and a standard deviation of cision analyses and studies of patient choices 15. The conversion is purely cosmetic: it is between alternative therapies for which the somehow more comforting to learn that one’s outcome lies in the future and is uncertain. IQ is 95 than −0.02. Validity Narrowly, the extent to which an assess- Tau See Correlation. ment method measures what it is intended to. Test-Retest Reliability The stability (i.e., re- More generally, the range of interpretations peatability) of a measurement is evaluated in that can be appropriately placed on a mea- terms of the agreement between a measure- sure. For example, socio economic status may ment applied to a sample of people and the serve as a valid indicator of risk of premature same measurement repeated later (typically mortality, even though it was not collected one or two weeks afterward). Assuming that for that purpose. A fuller discussion of valid- the state being measured has not changed, ity is given on page 30. any change in scores can be regarded as error Visual Analogue Scale A broadly applicable for- variance, and hence the level of agreement is mat for a measurement scale in which the re- used as an indicator of reliability. spondent places a mark at a point on a 10-cm Time Trade-off An approach to scaling utilities line that indicates the intensity of his re- (q.v.) in which judges are asked how many sponse. Phrases are printed at the ends of the years of life they would be willing to give up line (e.g., “no pain” and “pain as bad as you to recover from a specified chronic health can imagine”) to indicate the scope of the condition. The disutility of the condition is scale (see page 478). indicated by the number of years sacrificed. Yule’s Q A correlation formula used for estimat- Utilities Two conceptual approaches have been ing the association between two binary vari- used in scaling (q.v.) the relative severity of ables. This page intentionally left blank  Index Abbreviated Mental Test validity comparison to State-Trait Anxiety validity comparison to Informant Inventory (STAI), 323 Questionnaire on Cognitive Decline in the validity comparison to Zung Self-rating Elderly, 452, 453 Depression Scale, 347 activities of daily living (ADL), 28 affective responses, 151, 207–8 Older Americans Resources and Services age, 39 Multidimensional Functional Assessment agnosia, 394 Questionnaire, 600–602 agreement versus association, 36, 40–41 scales, 55–58 AIDS-Health Assessment Questionnaire, 116 See also Instrumental Activities of Daily AIMS. See Arthritis Impact Measurement Living (IADL) scales Scales Adapted General Well-Being Schedule Index alarm bell pain model, 470 (AGWBI), 245 alpha coefficient, 43–44, 45–46, 296t, 308t, ADAS-Cog. See Alzheimer’s Disease Assessment 549t, 655t, 711 Scale Alzheimer’s disease, 395, 465–67 ADAS-L. See Alzheimer’s Disease Assessment Alzheimer’s Disease Assessment Scale (ADAS), Scale-Late 396, 398t, 411–15 ADL. See activities of daily living alternative forms, 414 Affect Balance Scale (ABS), 151, 208, 209, conceptual basis, 411–12 211t, 221–24, 272 description, 412 alternative forms, 223 exhibit, 412 conceptual basis, 221–22 purpose, 411 description, 222 reference standards, 414 exhibit, 222 reliability, 412–13 purpose, 221 review, 411–17 reference standards, 223 scale format, 412 reliability, 222–23 validity, 413–14 review, 221–25 Alzheimer’s Disease Assessment Scale-Late scale format, 222 (ADAS-L), 414 validity, 223 American College of validity comparison to Beck Depression Neuropsychopharmacology (ACNP), 327, Inventory (BDI), 338 391 validity comparison to Center for anxiety checklist, 327 Epidemiologic Studies Depression Scale depression checklist, 327 (CES-D), 353 American Psychological Association (APA), validity comparison to Depression Adjective 706 Check Lists (DACL), 366 Anhedonia, 276 validity comparison to Life Satisfaction Index anxiety, 273–328 (LSI), 234 definitions of, 273–76 validity comparison to McMaster Health and depression, 276–77, 331 Index Questionnaire (MHIQ), 613 measurements, 277 719 720 Index anxiety (continued) ASQ. See Anxiety Scale Questionnaire psychological versus medical models, 274 Asthma Quality of Life Questionnaire, 522 theoretical approaches to, 274–76 Athelstan, Gary T., 538 types of, 274–75 attenuation of correlations, 711 See also specific measurement scales attitudes, 25 Anxiety Adjective Check List, 364 AUC. See Area under the Receiver Operating anxiety disorders, 273 Characteristic Curve Anxiety Scale Questionnaire (ASQ), 275 Anxiety Sensitivity Index (ASI), 275 back pain, 28, 518 Anxiety Status Inventory (ASI), 278, 302–6 Back Pain Classification Scale (BPCS), 28, 474, conceptual basis, 302 475t exhibit, 303 alternative forms, 505 purpose, 302 conceptual basis, 501–2 reference standards, 304–5 description, 502 reliability, 304 exhibit, 503–4 review, 302–6 purpose, 501 scale format, 303–4 reliability, 502 validity, 304 review, 501–6 aphasia, 394 scale format, 503 apraxia, 394 validity, 502, 504 Area under the Receiver Operating BAI. See Beck Anxiety Inventory Characteristic Curve (AUC), 33, 714 Barthel, D. W., 66 Arthritis Impact Measurement Scales (AIMS), Barthel Index, 56, 59t, 66–72 523, 524t, 526–32 alternative forms, 72 alternative forms, 531–32 conceptual basis, 66 conceptual basis, 526 description, 66–67 description, 526–27 exhibit, 67 exhibit, 528–30 guidelines, 69–70 purpose, 526 reliability, 67, 70 reference standards, 532 review, 66–74 reliability, 527 scale format, 67, 68–70 review, 526–534 scoring, 68, 71 scale format, 528–530 validity, 36, 70–71 validity, 527, 530–31 validity comparison to the European Quality validity comparison to Functional Status of Life Measure, 697 Index, 102 validity comparison to Functional validity comparison to Health Assessment Independence Measure, 146, 147 Questionnaire, 115 validity comparison to the Kenny Self-Care validity comparison to McMaster Health Evaluation, 82 Index Questionnaire, 613 validity comparison to Physical Self- validity comparison to Quality of Well-Being Maintenance Scale, 87 Scale, 679–680 validity comparison to PULSES Profile, 63 validity comparison to SF-36, 657 validity comparison to Sickness Impact validity comparison to Sickness Impact Profile, 634 Profile, 634 validity comparison to Spitzer Quality of Life Åsberg, M. Index, 567 Comprehensive Psychopathological Rating validity comparison to Functional Autonomy Scale, 391 Measurement System, 139 Cronholm-Ottosson Depression Rating Scale, BAS-D. See Brief Assessment Schedule— 391 Depression Montgomery-Åsberg Depression Rating Scale BCDRS. See Brief Carroll Depression Rating (MADRS), 382 Scale ASI. See Anxiety Sensitivity Index BCRS. See Brief Cognitive Rating Scale  Index 721 BDI-PC. See Beck Depression Inventory for validity comparison to the General Health Primary Care Questionnaire, 265, 268 BDI-II. See Beck Depression Inventory validity comparison to General Well-Being Bech-Rafaelsen Melancholia Scale (BRMS), Schedule, 244 374 validity comparison to Geriatric Depression validity comparison to Hamilton Rating Scale, 361 Scale for Depression, 372 validity comparison to Hamilton Rating validity comparison to Montgomery-Åsberg Scale for Depression, 372–73 Depression Rating Scale, 385 validity comparison to Hospital Anxiety and Beck, A. T. Depression Scale, 297–98 Beck Anxiety Inventory (BAI), 306 validity comparison to Manifest Anxiety Beck Depression Inventory (BDI), 335 Scale, 283 Beck Anxiety Inventory (BAI), 277, 278t, validity comparison to Montgomery-Åsberg 306–12, 318t Depression Rating Scale, 385 alpha internal consistency coefficients, 308t validity comparison to Positive and Negative alternative forms, 310–11 Affect Scale, 228 conceptual basis, 306–7 validity comparison to single-item rating description, 307 scale, 583 purpose, 306 validity comparison to State-Trait Anxiety reference standards, 311 Inventory, 323, 325 reliability, 307 validity comparison to visual analogue review, 306–13 depression scale, 391 validity, 307–10, 309t validity comparison to World Health validity comparison to Beck Depression Organization Quality of Life Scale, 626 Inventory, 337, 339 validity comparison to Zung Self-Rating validity comparison to Depression Anxiety Depression Scale, 347 Stress scales, 318 and visual analogue scales, 391 validity comparison to State-Trait Anxiety Beck Depression Inventory for Primary Care Inventory, 324 (BDI-PC), 339 Beck Depression Inventory (BDI), 318t, 333, Behavior Rating Scale (BRS), 457, 459 334t, 335–41, 392 Bennett, A. E., 88 alternative forms, 339–40 Disability Interview Schedule, 88 conceptual basis, 335–36 Benton Controlled Oral Word Association Test, description, 336 465 purpose, 335 Benton Temporal Orientation Test, 465 reference standards, 340 Benton Visual Retention Test, 465 reliability, 336–37 Bergner, Marilyn, 520, 630 review, 335–44 Sickness Impact Profile (SIP), 630 validity, 337–39 Berle Index, 153 validity comparison to Affect Balance Scale, Bias in measurement, 25–27, 34 223 defined, 25 validity comparison to Beck Anxiety reflecting personality, 25 Inventory, 309–10 social desirability, 25 validity comparison to Carroll Rating Scale spectrum, 33 for Depression, 388 verification, 34 validity comparison to Delighted-Terrible work-up, 34 scale, 582 Biobehavioral Pain Profile, 517 validity comparison to Depression Adjective biomedical pain models, 470 Check Lists, 366 bipolar disorder, 330 validity comparison to Depression Anxiety Blessed, G. Stress scales, 318 Blessed Dementia Scale (BLS-D), 420 validity comparison to Functional Living Information-Memory-Concentration Test Index - Cancer, 544 (IMC), 417 722 Index Blessed Dementia Scale (BLS-D), 396, 398t, conceptual basis, 491 420–22 description, 491, 494 review, 420–423 exhibit, 492–94 scale format, 422 purpose, 491 validity comparison to Cambridge Mental reliability, 494 Disorders of the Elderly Examination, 462, review, 491–96 463 scale format, 492–94 validity comparison to Clock Drawing task, validity, 495 409 Brief Psychiatric Rating Scale (BPRS) validity comparison to Modified Mini- validity comparison to Social Adjustment Mental State test, 446 Scale, 191–92 validity comparison to the Physical Self- Brief Symptom Inventory (BSI) Maintenance Scale, 87 validity comparison to Beck Anxiety Blessed Information-Memory-Concentration Inventory, 309 Test (IMC), 396, 398t, 417–19 validity comparison to Patient Evaluation validity comparison to Alzheimer’s Disease Conference System, 104 Assessment Scale, 413 validity comparison to Spitzer Quality of Life Blessed-Roth Dementia Scale, 420 Index, 567 BLS-D. See Blessed Dementia Scale Brink, T. L., 359 BNT. See Boston Naming Test Geriatric Depression Scale (GDS), 359 Boston Naming Test (BNT) BRMS. See Bech-Rafaelsen Melancholia Scale validity comparison to Dementia Rating Broadhead, W. E., 168 Scale, 401 Functional Social Support Questionnaire, validity comparison to Modified Mini- 168 Mental State test, 446 broad-spectrum generic health measures, 13 Box Pain Scale, 477 Brody, Elaine M., 84 BPCS. See Back Pain Classification Scale Physical Self-Maintenance Scale (PSMS), 84 BPI. See Brief Pain Inventory BRS. See Behavior Rating Scale BPRS. See Brief Psychiatric Rating Scale BSI. See Brief Symptom Inventory Bradburn, N. M. Bush, J.W. Affect Balance Scale (ABS), 221 Quality of Well-Being Scale (QWB), 675 model of well-being, 207–8 Branch, L. G. Cambridge Mental Disorders of the Elderly Social Disability Questionnaire, 203 Examination (CAMDEX), 396, 397, 398t, Brandt, P. A. 460–63 Personal Resource Questionnaire, 157, 204 alternative forms, 463 Brief Assessment Schedule—Depression conceptual basis, 460 (BAS-D), 333, 334t, 378–81 description, 460–61 alternative forms, 381 purpose, 460 conceptual basis, 378 reference standards, 463 description, 378–79 reliability, 461–62 exhibit, 379, 380 review, 460–64 reliability, 379 scale format, 462 review, 378–82 validity, 462–63 scale format, 379 validity comparison to Alzheimer’s Disease validity, 379–80 Assessment Scale, 414 Brief Carroll Depression Rating Scale (BCDRS), CAMCOG 388 review, 460–464 Brief Cognitive Rating Scale (BCRS), 466 scale format, 462 validity comparison to Alzheimer’s Disease validity comparison to Blessed Dementia Assessment Scale, 413 Scale, 421 Brief Pain Inventory (BPI), 474, 475t, 491–95 validity comparison to Clifton Assessment alternative forms, 495 Procedures for the Elderly, 458  Index 723 validity comparison to Modified Mini- Center for Epidemiologic Studies Depression Mental State test, 446 Scale (CES-D), 333, 334t, 350–56 CAMDEX. See Cambridge Mental Disorders of alternative forms, 354–55 the Elderly Examination conceptual basis, 350–51 Canberra Interview for the Elderly (CIE), 396, description, 351 467 exhibit, 352 cancer purpose, 350 European Organization for Research and reference standards, 355 Treatment of Cancer Quality of Life reliability, 351 Questionnaire, 551–56 review, 350–358 Functional Assessment of Cancer Therapy, scale format, 352 546–50 validity, 351–54 Functional Living Index, 541–45 validity comparison to Beck Depression Quality-adjusted Time Without Symptoms Inventory, 337–38 and Toxicity Method, 559–63 validity comparison to Depression Adjective Cancer Rehabilitation Evaluation System Check Lists, 366 (CARES) validity comparison to General Well-Being validity comparison to European Schedule, 244, 245 Organization for Research and Treatment validity comparison to Geriatric Depression of Cancer Quality of Life Questionnaire Scale, 361 QLQ-C30, 555 validity comparison to Illness Behavior Cantril, H. Questionnaire, 512 Ladder scale, 579, 580, 582 validity comparison to Quality of Well-Being capacity wording, 15–16 Scale, 680 CAPE. See Clifton Assessment Procedures for validity comparison to Social Adjustment the Elderly Scale, 191 CARE. See Comprehensive Assessment and validity comparison to State-Trait Anxiety Referral Evaluation Inventory, 323 CARES. See Cancer Rehabilitation Evaluation validity comparison to Zung Self-rating System Depression Scale, 347 Carroll, B. J. CERAD. See Consortium to Establish a Carroll Rating Scale for Depression (CRS), Registry for Alzheimer’s Disease 387 CES-D. See Center for Epidemiologic Studies Carroll Rating Scale for Depression (CRS), Depression Scale 333, 334t, 387–90 Chambers, Larry W., 612 alternative forms, 388 checklists, 206–7 conceptual basis, 387 CIE. See Canberra Interview for the Elderly description, 387–88 Circles Scale, 579–80 exhibit, 389 circumplex models of affect, 208 purpose, 387 Clare, Anthony W., 194 reliability, 388 Clark, L.A., 225, 276 validity, 388 classical test theory, 40 validity comparison to Sickness Impact Cleeland, Charles S., 491 Profile, 634 Clifton Assessment Procedures for the Elderly CASI. See Cognitive Abilities Screening (CAPE), 397, 398t, 456–59 Instrument alternative forms, 459 categorical scales, 18 conceptual basis, 456–57 CCCE. See Cross-Cultural Cognitive description, 457–58 Examination exhibit, 457 CCSE. See Cognitive Capacity Screening purpose, 456 Examination reference standards, 459 CDR. See Clinical Dementia Rating reliability, 458 Cella, David, 546, 705 review, 456–460 724 Index Clifton Assessment Procedures for the Elderly description, 605 (CAPE) (continued) exhibit, 606 scale format, 457 and the Older Americans Resources and survey version, 459 Services Multidimensional Functional validity, 458–59 Assessment Questionnaire, 602 validity comparison to Informant purpose, 604 Questionnaire on Cognitive Decline in the reliability, 605–6 Elderly, 452 review, 604–8 Clifton Assessment Schedule, 457 scale format, 606 Clinical Dementia Rating (CDR) validity, 606–7 validity comparison to Short Portable Mental validity comparison to Zung Self-rating Status Questionnaire, 428 Depression Scale, 347 clinical neuropsychological tests, 395 Comprehensive Psychopathological Rating Clock Drawing Test, 396, 398t, 407–11 Scale, 382, 391 conceptual basis, 407 computerized adaptive testing, 705 description, 407–9 Concordance, 711 purpose, 407 concurrent validity, 31, 32, 711 reference standards, 409–10 Consortium to Establish a Registry for reliability, 409 Alzheimer’s Disease (CERAD), 396 validity, 409 confirmatory factor analysis, 37 coefficient of concordance, 711 consistency. See reliability coefficient omega, 43 construct validity, 34–37, 39, 708, 711 coefficient theta, 43 content saturation, 35 cognition content validity, 30–31, 711 definition of, 394 convergent validity, 34, 36, 44, 711 mental status testing, 394–469 COOP Charts for Primary Care Practice, 523, and pain, 470 524t, 569–76 screening tests, 465–66 alternative forms, 574–75 See also specific measurement scales conceptual basis, 569 Cognitive Abilities Screening Instrument description, 569–70 (CASI), 466 exhibit, 571–73 Cognitive Assessment Scale (CAS), 457, 459 purpose, 569 Cognitive Behavior Rating Scale, 396 reference standards, 575 Cognitive Capacity Screening Examination reliability, 570 (CCSE), 396, 398t, 404–6 review, 569–77 review, 404–7 scale format, 571–73 scale format, 405 validity, 570, 573–74 validity comparison to the Functional validity comparison to European Quality of Activities Questionnaire, 108 Life Measure, 697, 700 cognitive impairment, 395–96 validity comparison to Nottingham Health Community Adaptation Schedule, 203 Profile, 644–45 Community Adjustment Profile System, 204 validity comparison to SF-36, 657–58 community care for elderly, 56 WONCA version, 574–75 comparative techniques, 20 CORE-CARE, 607 Comprehensive Assessment and Referral Cornell Scale for Depression in Dementia, 380 Evaluation (CARE), 464, 523, 525t, Corrected General Health Questionnaire, 264 604–8 correlation, 35–36, 711 alternative forms, 607 intraclass, 712 as basis for Brief Assessment Schedule— item-total, 712–13 Depression, 378–79, 381 interpretation of, 35–36 and the Brief Assessment Schedule- versus agreement, 36, 40 Depression, 378, 381 correlational approaches, 22, 34–36 conceptual basis, 604–5 correlation coefficients, 35, 41  Index 725 Covi Anxiety Scale, 327 validity comparison to Cognitive Capacity validity comparison to Hamilton Anxiety Screening Examination, 406 Rating Scale, 288 validity comparison to Information-Memory- Crewe, Nancy M., 538 Concentration Test, 419 Functional Assessment Inventory, 538 validity comparison to Mini-Mental State CRIES pain scale, 518 Examination, 432 criterion valildity, 31–34, 708, 711 Dementia Scale. See Blessed Dementia Scale Cronbach, L. J. depression, 329–93 coefficient alpha, 42–46, 549t, 655t, 711 and anxiety, 276–77, 331 Cronholm-Ottosson Depression Rating Scale, 391 classifications of, 330–31 Cross-Cultural Cognitive Examination (CCCE), definition of, 329 465 diagnostic criteria for, 329–31 cross-cultural testing, 395 measurement of, 332–33 cross-modality matching, 17–18 scales, 39, 332, 391 CRS. See Carroll Rating Scale for Depression See also specific measurement scales cutting-point, 32–33 Depression Adjective Check Lists (DACL), 43, 333, 334t, 364–68 DACL. See Depression Adjective Check Lists alternative forms, 367 Dartmouth COOP Charts. See COOP Charts conceptual basis, 364 for Primary Care Practice description, 364–65 Dartmouth Pain Questionnaire, 488 exhibit, 365 Dastoor, D. P. purpose, 364 Hierarchic Dementia Scale, 467 reference standards, 367 DASS. See Depression Anxiety Stress Scales reliability, 365–66 decision analysis, 19–20, 23–25, 26 review, 364–69 Delayed Memory Test scale format, 365 validity comparison to Mental Status validity, 366–67 Questionnaire, 425 validity comparison to Beck Depression Delayed Word Recall Test (DWR), 466 Inventory, 338 Delighted-Terrible Scale, 578, 579, 582 validity comparison to Center for delirium, 394 Epidemiologic Studies Depression Scale, dementia 353 definition of, 394 validity comparison to General Well-Being diagnosis of, 467 Schedule, 243 mental status testing, 394–469 validity comparison to Zung Self-Rating scales, 705 Depression Scale, 347 screening tests, 465–67 Depression Anxiety Stress Scales (DASS), 277, See also specific measurement scales 278t, 313–19 Dementia Rating Scale (DRS), 396, 398t, alternative forms, 317 399–403 conceptual basis, 313–14 alternative forms, 402 correlation with other measures, 318t conceptual basis, 399 description, 314, 316 description, 399 exhibit, 315–16 exhibit, 400 purpose, 313 purpose, 399 reference standards, 317 reference standards, 402 reliability, 316–17 reliability, 399–401 review, 313–19 review, 399–404 scale format, 315–16 scale format, 400 validity, 317 validity, 401–2 validity comparison to Beck Anxiety validity comparison to Cambridge Mental Inventory, 309 Disorders of the Elderly Examination, 462, validity comparison to Positive and Negative 463 Affect Scale, 228 726 Index Depression Anxiety Stress Scales (DASS) DUHP. See Duke-University of North Carolina (continued) (UNC) Health Profile validity comparison to State-Trait Anxiety DUKE-Anxiety-Depression Scale, 327 Inventory, 323 Duke Health Profile (DUKE), 523, 524t, 591–95 Depression Extender Check List, 367 alternative forms, 594 Depression Status Inventory, 348 conceptual basis, 591 Descaretes, René, 470 description, 591–92 descriptive classification, 12, 13 exhibit, 593 diagnostic indices, 13 purpose, 591 Diagnostic Interview Schedule (DIS), 396 reference standards, 595 Diagnostic and Statistical Manual of the reliability, 592 American Psychological Association review, 591–96 (DSM), 330, 332 scale format, 593 diagnostic tests, 55 validity, 592, 594 diagnostic work-up bias, 34 validity comparison to Duke Social Support DIF. See differential item functioning and Stress Scale, 172 difference, 42 Duke Social Support and Stress Scale differential item functioning (DIF), 39 (DUSOCS), 153, 154t, 170–73 Differential Reliability Index, 35 conceptual basis, 170 DIS. See Diagnostic Interview Schedule description, 170 disability exhibit, 171–72 definition of, 29, 706 purpose, 170 evolution of measurements, 55–58 reliability, 170, 172 index of, 28 review, 170–73 Rasch model for, 21 scale format, 171–72 Disability and Distress Scale, 523, 525t, validity, 172 671–74 Duke Social Support Index (DSSI), 204 alternative forms, 673–74 Duke University Center for the Study of Aging conceptual basis, 671 and Human Development, 596 description, 671–73 Duke-University of North Carolina (UNC) exhibit, 672 Functional Social Support Questionnaire purpose, 671 (DUFSS), 153, 154t, 168–70 review, 671–75 review, 168–170 scale format, 672 scale format, 169 validity, 673 Duke-University of North Carolina (UNC) validity comparison to Health Utilities Index, Health Profile (DUHP) 691 validity comparison to Duke Social Support Disability Interview Schedule, 59t, 88–89 and Stress Scale, 172–73 review, 88–89 validity comparison to Duke-UNC scale format, 90–91 Functional Social Support Questionnaire, disablement, 29 168 discriminant analysis, 711 Dupuy, H. J. discriminant validity, 32, 711 General Well-Being Schedule (GWB), 240 discriminative indexes, 13 Dupuy scale. See General Well-Being Schedule distress, 207, 240, 247 DUSOCS. See Duke Social Support and Stress divergent validity, 35, 36 Scale DRS. See Dementia Rating Scale DWR. See Delayed Word Recall Test DSM. See Diagnostic and Statistical Manual of Dyadic Effectiveness Scale, 160 the American Psychological Association Dysthymia, 330 DSSI. See Duke Social Support Index DUFSS. See Duke-University of North Carolina East Boston Memory Test, 466 (UNC) Functional Social Support validity comparison to Short Portable Mental Questionnaire Status Questionnaire, 427  Index 727 Eastern Cooperative Oncology Group (ECOG) exhibit, 552–53 performance Status Scale, 547, 554 purpose, 551 validity comparison to Functional reference standards, 556 Assessment of Cancer Therapy, 549 reliability, 553–54 EBAS. See Even Briefer Scale for Depression validity, 554–55 ECAQ. See Elderly Cognitive Assessment European Quality of Life Measure (EuroQol Questionnaire EQ-5D), 520, 523, 525t, 694–700, 704 ECOG. See Eastern Cooperative Oncology alternative forms, 699 Group conceptual basis, 694–95 econometrics, 19–20, 23–25, 704, 717 description, 695 effect size, 37–39, 711–12 exhibit, 696–97 elderly purpose, 694 community care for, 56 reference standards, 699 Comprehensive Assessment and Referral reliability, 695, 697 Evaluation, 604–8 review, 694–702 Geriatric Depression Scale, 359–62 scale format, 696 Informant Questionnaire on Cognitive validity, 697–99 Decline in the Elderly, 449–55 validity comparison to Barthel Index, 71 mental status testing, 394–469 validity comparison to COOP Charts, 573–74 Multilevel Assessment Instrument, 609–11 validity comparison to Functional Living Older Americans Resources and Services Index - Cancer, 544 Multidimensional Functional Assessment validity comparison to Health Utilities Index, Questionnaire, 596–603 689–91 Philadelphia Geriatric Center Morale Scale, validity comparison to Nottingham Health 236–39 Profile, 644–45 Social Dysfunction Rating Scale, 196–99 validity comparison to SF-12, 666, 669 See also dementia validity comparison to SF-36, 657–58 Elderly Cognitive Assessment Questionnaire validity comparison to single-item rating (ECAQ), 465 scale, 583 Elemente der Psychophysik (Fechner), 16 validity comparison to visual analogue scale, EMAS. See Endler Multidimensional Anxiety 582–83 Scales evaluative indexes, 13 emic and etic translations, 14 Even Briefer Scale for Depression (EBAS) empirical method, 27 review, 379–82 EMSQ. See Extended Mental Status scale format, 380 Questionnaire Executive Interview (EXIT), 409, 466 Endler Multidimensional Anxiety Scales EXIT. See Executive Interview (EMAS), 275, 327–28 Extended Barthel Index, 72 EORTC. See European Organization for Extended Mental Status Questionnaire Research and Treatment of Cancer (EMSQ), 425 (EORTC) Quality of Life Questionnaire Extended Rating Scale for Dementia, 396 QLQ-C30 Extended Scale for Dementia, 402 Equal-Appearing Intervals scaling method, 20 exploratory factor analysis, 37 error, 39, 40 Eslinger battery, 465 Faces Scale, 578–79 European Organization for Research and FACT. See Functional Assessment of Cancer Treatment of Cancer (EORTC) Quality of Therapy Life Questionnaire QLQ-C30, 523, 524t, factor analysis, 14, 36–37, 43, 44, 712 551–56 factorial validity, 36–37 alpha internal consistency figures, 554t FAI. See Functional Assessment Inventory alternative forms, 555 Fairbank, Jeremy, 498 conceptual basis, 551 Oswestry Low Back Pain Disability description, 551, 553 Questionnaire, 498 728 Index FAM. See Functional Assessment Measure validity, 547, 549 FAQ. See Functional Activities Questionnaire validity comparison to Functional Living fear, 273, 274 Index-Cancer, 544 Fechner, Gustav, 16, 17 Functional Assessment of Multiple Sclerosis Feeny, David, 683 (FAMS), 549 Health Utilities Index (HUI), 683 Functional Autonomy Measurement System FIM. See Functional Independence Measure (SMAF), 56, 60t, 122–40, 704 FLIC. See Functional Living Index—Cancer alternative forms, 139 FLP. See Functional Limitations Profile conceptual basis, 122 Folstein, Marshal, 429 description, 122–23 Mini-Mental State Examination, 429 exhibit, 124–38 FONE-FIM, 146 purpose, 122 Forer, Stephen K., 95 reference standards, 139 Functional Status Rating System, 95 reliability, 123 Freud, Sigmund, 274, 275 review, 141–49 Fries, James F., 111 scale format, 124–38 Health Assessment Questionnaire (HAQ), validity, 123, 139 111 functional classification, 13 FSQ. See Functional Status Questionnaire functional disability, 29, 30, 56, 587, 706 function, 12 Functional Disability Index, 112 Functional Activities Questionnaire (FAQ), 60t, Functional Independence Measure (FIM), 57, 108–10, 464, 465 60t, 97, 141–47, 707 review, 108–11 alternative forms, 146 scale format, 109–10 conceptual basis, 141–42 validity comparison to Cognitive Capacity description, 142 Screening Examination, 406 exhibit, 143–44 validity comparison to Mini-Mental State and the Minimum Data Set, 141 Examination, 433 purpose, 141 Functional Assessment Inventory (FAI) (Crewe), reference standards, 146 523, 524t, 538–41 reliability, 142, 144 alternative forms, 540 review, 141–149 conceptual basis, 538–39 scale format, 143–145 description, 539–40 summary chart, 145 exhibit, 539 validity, 144, 146 purpose, 538 validity comparison to Modified Mini- reliability, 540 Mental State Examination, review, 538–41 validity comparison to Barthel Index, 72 scale format, 539 validity comparison to Patient Evaluation validity, 540 Conference System, 104 Functional Assessment Inventory (Pfeiffer), 602 validity comparison to PULSES Profile, 63 Functional Assessment Measure (FAM), 146 validity comparison to Functional Autonomy Functional Assessment of Cancer Therapy Measurement System, 139 (FACT), 523, 524t, 546–50 validity comparison to Spitzer Quality of Life alternative forms, 549 Index, 567 conceptual basis, 546 WeeFIM, 146 Cronbach alpha coefficients, 549t functional level, 28–29 description, 546–47 functional limitations, 29 exhibit, 548 Functional Limitations Profile (FLP), 92, purpose, 546 635–36 reference standards, 549 Functional Living Index—Cancer (FLIC), 523, reliability, 547 524t, 541–45 review, 546–51 alternative forms, 545 scale format, 548 conceptual basis, 541  Index 729 description, 541, 544 Garrad, Jessie, 88 exhibit, 542–43 Garron, David C., 501 purpose, 541 gate control theory of pain, 471 reliability, 544 GDS. See Geriatric Depression Scale review, 541–46 Gelber, R. D. scale format, 542–43 Quality-Adjusted Time Without Symptoms validity, 544 and Toxicity Method Index, 559 validity comparison to European Quality of General Health Questionnaire (GHQ), 209, Life Measure, 698 211t, 259–69, 271, 272, 707 validity comparison to Functional alternative forms, 268 Assessment of Cancer Therapy, 547, 549 conceptual basis, 259 validity comparison to visual analogue scale, description, 259, 264 582–583 exhibit, 260–63 Functional Social Support Questionnaire, 168 purpose, 259 review, 168–70 reliability, 264–65 scale format, 169 review, 259–71 Functional Status Index (FSI), 60t, 100–103, 680 scale format, 260–63 alternative forms, 102 validity, 265–68, 266t, 267t conceptual basis, 100 validity comparison to Affect Balance Scale, description, 100, 102 223 exhibit, 101 validity comparison to Beck Depression purpose, 100 Inventory, 338 reliability, 102 validity comparison to Center for review, 100–103 Epidemiologic Studies Depression Scale, scale format, 101 353, 354 validity, 102 validity comparison to COOP Charts, 574 Functional Status Questionnaire (FSQ), 523, validity comparison to Delighted-Terrible 524t, 587–91 scale, 582 alternative forms, 590 validity comparison to European conceptual basis, 587 Organization for Research and Treatment description, 587 of Cancer Quality of Life Questionnaire, exhibit, 588–89 555 purpose, 587 validity comparison to Functional Living reliability, 587, 589 Index—Cancer, 544 validity, 589–90 validity comparison to General Well-Being validity comparison to Health Assessment Schedule, 243 Questionnaire, 115 validity comparison to Hospital Anxiety and validity comparison to Quality of Well-Being Depression Scale, 297–99 Scale, 679 validity comparison to Interview Schedule for Functional Status Questionnaire (FSQ) Social Interaction, 182 review, 587–91 validity comparison to Mental Health scale format, 588–89 Inventory, 252 validity comparison to Arthritis Impact validity comparison to Nottingham Health Measurement Scales, 530 Profile, 643 validity comparison to Index of ADL, 77 validity comparison to Social Adjustment validity comparison to Older Americans Scale, 191–92 Resources and Services ADL scale, 602 general health status, 520–703 validity comparison to Sickness Impact General Health Rating Index (GHRI), 254 Profile, 635 Generalizability theory, 42 Functional Status Rating System (FSRS), 56, General Well-Being Schedule (GWB), 209, 59t, 95–97, 464 211t, 240–46, 272, 650 review, 95–97 alternative forms, 244–45 scale format, 96–97 conceptual basis, 240 730 Index General Well-Being Schedule (GWB) (continued) goal attainment scaling, 712 description, 240 Goldberg, D. P. exhibit, 241–43 Clinical Interview Schedule, 265 reference standards, 245 General Health Questionnaire, 259 reliability, 240 Granger, Carl V. review, 240 Functional Independence Measure, 141 scale format, 241–43 Long-Range Evaluation System, 72 validity, 240, 243–44 PULSES Profile, 62 generic health measures, 13 Graphic Rating Scale, 477 Geriatric Depression Scale (GDS), 333, 334t, Gurel, Lee 359–62, 391 Physical and Mental Impairment-of-Function alternative forms, 362 Evaluation, 535 conceptual basis, 359 Gurland, Barry J., 199 description, 359 Comprehensive Assessment and Referral exhibit, 360 Evaluation (CARE), 604 purpose, 359 Structured and Scaled Interview to Assess reference standards, 362 Maladjustment, 199 reliability, 359–60 Guttman scaling, 22, 43, 715–16 review, 359–364 GWB. See General Well-Being Schedule scale format, 360 validity, 360–62 Hachinski Ischemia Score, 467 validity comparison to Beck Anxiety HADS. See Hospital Anxiety and Depression Inventory, 310 Scale validity comparison to Beck Depression HALex. See Health and Activities Limitations Inventory, 338 Index validity comparison to Brief Assessment Hamilton, Byron B., 141 Scale, 380 Hamilton Anxiety Rating Scale (HARS), 278t, validity comparison to Center for 286–93 Epidemiologic Studies Depression Scale, alternative forms, 288, 291–92 354 conceptual basis, 286 validity comparison to Sickness Impact description, 286 Profile, 635 exhibit, 287, 289–91 validity comparison to Zung Self-Rating purpose, 286 Depression Scale, 347 reliability, 286 Geriatric Evaluation by Relative Rating review, 286–94 Instrument (GERRI), 396 scale format, 287 validity comparison to Alzheimer’s Disease structured interview guide, 289–91 Assessment Scale, 413 validity, 288 Geriatric Mental State Examination (GMS) validity comparison to Beck Anxiety validity comparison to CAMDEX, 462 Inventory, 309 GERRI. See Geriatric Evaluation by Relative validity comparison to Zung Self-Rating Rating Instrument Anxiety Scale, 304 GHQ. See General Health Questionnaire Hamilton Anxiety Scale for Generalized GHRI. See General Health Rating Index Anxiety (HAS-G), 292 Gilleard, C. J., 456 Hamilton Anxiety Scale for Panic Attacks GLOBAL-CARE, 607 (HAS-P), 292 Global Deterioration Scale, 466 Hamilton Depression Inventory, 373 validity comparison to Clock Drawing task, Hamilton Rating Scale for Depression (HRSD), 409 14, 333, 334t, 369–75, 392, 708 validity comparison to Mental Status alternative forms, 373–74 Questionnaire, 424 conceptual basis, 369 validity comparison to Mini-Mental State description, 369, 371 Examination, 432 exhibit, 370  Index 731 purpose, 369 social, 150–205 reliability, 371–72 See also health measurement review, 369–78 Health and Activities Limitations Index scale format, 370 (HALex), 584 validity, 372–73 Health and Nutrition Examination Survey validity comparison to Beck Depression (HANES), 240 Inventory, 337–38 Health Assessment Questionnaire (HAQ), 56, validity comparison to Carroll Rating Scale 60t, 111–17 for Depression, 388, 390 AIDS-HAQ, 116 validity comparison to Center for alternative forms, 116 Epidemiologic Studies Depression Scale, Childhood HAQ, 116 354 conceptual basis, 111–12 validity comparison to Geriatric Depression description, 112 Scale, 361 exhibit, 113–14 validity comparison to Hamilton Anxiety Fibromyalgia HAQ, 116 Rating Scale, 288 purpose, 111 validity comparison to Hospital Anxiety and reliability, 112, 115 Depression Scale, 298 review, 111–119 validity comparison to Montgomery-Åsberg scale format, 113–114 Depression Rating Scale, 386 validity, 115–16 validity comparison to Spitzer Quality of Life validity comparison to Arthritis Impact Index, 567 Measurement Scales, 530–531 validity comparison to visual analogue validity comparison to European Quality of depression scale, 391 Life Measure EQ-5D, 697–698 validity comparison to Zung Self-Rating validity comparison to Functional Status Depression Scale, 347 Questionnaire, 590 and visual analogue scales, 391 health index(es), 522–23 handicap, 55–149 conceptual basis for, 29–30 definition of, 29, 706 construction of, 27–28 See also disability; specific measurement definition of, 15 scales and numerical scaling techniques, 16 HANES. See Health and Nutrition subjective and objective data, 26 Examination Survey theories behind, 706 happiness, 151, 206, 521 health indicators, 11–12 HAQ. See Health Assessment Questionnaire health measurements Harmonization Project for Instruments in adequacy of, 11 Dementia, 414, 705 bias in, 25–27 HARS. See Hamilton Anxiety Rating Scale conceptual bases for, 27–30 Harvey, Richard F. current status of, 704–5 Patient Evaluation Conference System evaluating, 7–8 (PECS), 103 generic versus specific, 13 Hasegawa, K. guidelines for developing, 705–8 Dementia Screening Scale, 466 index versus profile, 13–15, 21, 44 HAS-G. See Hamilton Anxiety Scale for reliability of, 40 Generalized Anxiety subjective versus objective, 13–15, 25, 26 HAS-P. See Hamilton Anxiety Scale for Panic theoretical and technical foundations of, 10–46 Attacks types of, 12–16 Havighurst, Robert J., 231 See also specific measurement scales health Health Measurement Questionnaire, 673 definition of, 10–11 Health Opinion Survey (HOS), 26, 28, 206, general status, 520–703 208, 209, 210, 211t, 212–15, 271, 707 numerical estimates of, 18–25 alternative forms, 214 single-item indicators, 578–84 conceptual basis, 212 732 Index Health Opinion Survey (HOS) (continued) Hiearchic Dementia Scale, 467 description, 212–13 HIS-GWD, 244–45 main variants, 213 Hopkins Symptom Checklist (HSCL), 327, 391 original version, 212 and the Katz Adjustment Scales, 174 purpose, 210, 212 validity comparison to Beck Depression reliability, 213–14 Inventory, 337 review, 210–16 validity comparison to the General Health scale format, 212 Questionnaire, 266 validity, 214 validity comparison to the Positive and validity comparison to General Well-Being Negative Affect Scale, 228 Schedule, 245 HOS. See Health Opinion Survey Health Perceptions Questionnaire (HPQ), 209, Hospital Anxiety and Depression Scale 211t, 253–58 (HADS), 277, 278t, 294–300, 318t, 333 alternative forms, 258 alpha internal consistency coefficients, conceptual basis, 254 296t description, 254 alternative forms, 299 exhibit, 255–56 comparison to Zung Self-rating Anxiety purpose, 253–54 Scale, 304 reference standards, 258 conceptual basis, 294 reliability, 254, 257, 257t description, 294 review, 253–58 exhibit, 295 scale format, 255–56 purpose, 294 validity, 257–58 reliability, 296 health profile, 15, 522–23 review, 294–302 Health Status Index. See Quality of Well-Being scale format, 295 Scale validity, 296–99 Health Utilities Index (HUI), 523, 525t, validity comparison to Depression Anxiety 683–92 Stress Scales, 318 alternative forms, 690–91 validity comparison to European conceptual basis, 684 Organization for Research and Treatment description, 684, 686–88 of Cancer Quality of Life Questionnaire, exhibit, 685–87 555 purpose, 683–84 validity comparison to European Quality of reference standards, 691 Life Measure EQ-5D, 697 reliability, 688 validity comparison to the General Health review, 683–94 Questionnaire, 265, 267 scale format, 685–86 validity comparison to Montgomery-Åsberg validity, 689–90 Depression Rating Scale, 385–86 validity comparison to European Quality of validity comparison to Positive and Negative Life Measure EQ-5D, 697–98 Affect Scale, 228 validity comparison to SF-12, 666, 669 HPQ. See Health Perceptions Questionnaire validity comparison to SF-36, 657 HRCA-QL Index, 567–68 validity comparison to single-item rating HRSD. See Hamilton Rating Scale for scale, 583 Depression validity comparison to visual analogue scale, HSCL. See Hopkins Symptom Checklist 582 HUI. See Health Utilities Index Hébert, Réjean Hunt, Sonja Functional Autonomy Measurement System Nottingham Health Profile (NHP), 639 (FAMS), 122 hello-goodbye effect, 26 IADL. See Instrumental Activities of Daily Henderson, Scott Living (IADL) scales Interview Schedule for Social Interaction, IBQ. See Illness Behavior Questionnaire 180 ICC. See intraclass correlation  Index 733 ICD. See International Classification of Disease Informant Questionnaire on Cognitive Decline ICF. See International Classification of in the Elderly (IQCODE), 397, 398t, Functioning, Disability and Health 449–55 ICIDH. See International Classification of alternative forms, 453–54 Impairment, Disability and Handicap conceptual basis, 449 idiographic approach to measurement, 14, description, 449, 451 27 exhibit, 450–51 Illness Behavior Assessment Schedule, 512 purpose, 449 Illness Behavior Questionnaire (IBQ), 471, 474, reference standards, 454 475t, 508–13 reliability, 451 alternative forms, 512 review, 449–456 conceptual basis, 508 scale format, 450–451 description, 508, 510–11 validity, 451–53 exhibit, 509–10 validity comparison to Clifton Assessment purpose, 508 Procedures for the Elderly, 458 reference standards, 512–13 validity comparison to Mini-Mental State reliability, 511 Examination, 432 review, 508–14 Information-Memory-Concentration Test scale format, 509–10 (IMC). See Blessed Information-Memory- validity, 511–12 Concentration Test impairment, 29 Instrumental Activities of Daily Living (IADL) impairment scales, 55 scales, 56–149, 600–602 Incapacity Status Scale, 63 See also specific scales IN-CARE, 607 internal consistency, 42–45, 714 Index of ADL. See Index of Independence in International Classification of Disease (ICD), 33 Activities of Daily Living International Classification of Functioning, Index of Health-Related Quality of Life Disability and Health (ICF), 29 (IHQL), 673 International Classification of Impairment, Index of Independence in Activities of Daily Disability and Handicap (ICIDH), 29 Living, 56, 59t, 74–77 inter-rater reliability, 712 conceptual basis, 74 interval scales, 19, 20, 715 description, 74, 76 Interview Schedule for Social Interaction (ISSI), evaluation form, 75 153, 154t, 180–83 review, 74–78 alternative forms, 182 scale format, 75–76 conceptual basis, 180 scoring and definitions, 76 description, 180–82 validity, 77 exhibit, 181 validity comparison to Barthel Index, 72 purpose, 180 validity comparison to Functional reference standards, 182 Independence Measure, 146 reliability, 182 validity comparison to Functional Living review, 180–84 Index - Cancer, 544 scale format, 181 validity comparison to Older Americans validity, 182 Resources and Services ADL scale, 602 intraclass correlation (ICC), 41, 45, 712, 714 validity comparison to Sickness Impact Inventory of Socially Supportive Behaviors Profile, 634 (ISSB), 204 validity comparison to Spitzer Quality of Life IPAT Anxiety Scale, 323 Index, 567 IQCODE. See Informant Questionnaire on Index of Well-Being. See Quality of Well-Being Cognitive Decline in the Elderly Scale IRT. See item response theory index scores, 15 Ischemic Score, 467 indicators, 11–12 ISSB. See Inventory of Socially Supportive Infant mortality rate (IMR), 11 Behaviors 734 Index ISSI. See Interview Schedule for Social Kendall’s tau, 711 Interaction Kenny Self-Care Evaluation, 56, 59t, 78–84 item, 12, 712 conceptual basis, 78 item analysis, 28, 31 description, 78, 82 differential item functioning (DIF), 39 exhibit, 79–83 item-characteristic curve, 22, 712 purpose, 78 item response analyses, 37 reliability, 82 item response theory (IRT), 21–23, 44, 704, review, 78–84 705, 708, 712 scale format, 79–83 item-total correlation, 712–13 validity, 82, 84 Kuder-Richardson formula, 43 Jacobs, John W. Kutner Morale Scale, 233 Cognitive Capacity Screening Examination (CCSE), 404 Ladder Scale, 579, 580, 582 Jellinek, Hollis M., 103 Lambeth Disability Screening Questionnaire, Jette, Alan M. 57, 59t, 89–92 Functional Status Index (FSI), 100 conceptual basis, 89 Functional Status Questionnaire (FSQ), 587 description, 91 Jorm, Anthony exhibit, 90–91 Informant Questionnaire on Cognitive purpose, 89 Decline in the Elderly (IQCODE), 449 reliability, 91 just noticeable differences, 16–17, 26 review, 89–91–93 scale format, 92 Kahn, Robert L. validity, 91–92 Mental Status Questionnaire (MSQ), 423 Langner 22–item scale. See Twenty-Two Item Kaplan, R. M. Screening Score of Psychiatric Symptoms Quality of Well-Being Scale (QWB), 675 Langner, T. S. Kappa coefficient, 42, 45, 713 Twenty-two Item Screening Score of Karnofsky Performance Status Scale Psychiatric Symptoms, 216 validity comparison to Functional Living language, 14 index-Cancer, 544 latent trait analysis, 21–23, 713 validity comparison to Nottingham Health Lawton, M. Powell Profile, 644 Multilevel Assessment Instrument (MAI), validity comparison to Quality of Life Index, 609 565 Philadelphia Geriatric Center Morale Scale, Katz, M. M., 173 236 Katz, S. Physical Self-Maintenance Scale (PSMS), 84 Index of Independence in Activities of Daily Leavitt, F. Living, 74 Back Pain Classification Scale (BPCS), 501 Katz Adjustment Scales, 153, 154t, 173–76 Lehman Quality of Life Interview (QOFI), 703 alternative forms, 176 Levine-Pilowsky Depression Questionnaire, 512 conceptual basis, 174 life satisfaction, 521 description, 174 Life Satisfaction Index (LSI), 206, 209, 211t, exhibit, 175 231–35, 707 purpose, 173 alternative forms, 234 reference standards, 176 conceptual basis, 231 reliability, 174–75 description, 231–32 review, 173–77 exhibit, 232 scale format, 175 purpose, 231 validity, 175 reference standards, 234 validity comparison to Sickness Impact reliability, 233 Profile, 635 review, 231–36 Kendall’s coefficient of concordance, 711 scale format, 232  Index 735 validity, 233–34 reliability, 280 validity comparison to Philadelphia Geriatric review, 279–285 Center Morale Scale, 237 scale format, 281–82 validity comparison to Zung Self-Rating validity, 280, 283 Depression Scale, 347 validity comparison to Beck Depression Life Satisfaction Rating Scale, 233 Inventory, 338 likelihood ratio, 713 validity comparison to Functional Linn, Margaret W., 97, 196, 611 Assessment of Cancer Therapy, 547 Rapid Disability Rating System, 97 validity comparison to State-Trait Anxiety Self-Evaluation of Life Function Scale Inventory, 323 (SELF), 611 validity comparison to Twenty-Two Item Social Dysfunction Rating Scale (SDRS), Score, 219 196 validity comparison to Zung Self-Rating logit, 713 Anxiety Scale, 304 London Handicap Scale, 657 Mapi Research Institute, 3, 704, 709 London Psychogeriatric Rating Scale, 396 Maryland Disability Index, 66 Long-Range Evaluation System, 72 Mattis, S. Lovibond, P. F. Dementia Rating Scale (DRS), 399 Depression Anxiety Stress Scales (DASS), McCann, Cairbre B., 62 313 McFarlane, Allan H. Low Back Pain Rating Scale, 518 Social Relationship Scale (SRS), 155 Low Back Pain Symptom Checklist McGill Comprehensive Pain Questionnaire, review, 501–6 488 scale format, 503 McGill Pain Questionnaire (MPQ), 471, 473, LSI. See Life Satisfaction Index 474, 475t, 483–89, 707 Lubin, Bernard alternative forms, 487–88 Depression Adjective Check Lists (DACL), conceptual basis, 483–84 364 description, 484 Depression Extender Check List, 367 exhibit, 485–86 State-Trait Depression Adjective Check Lists, Pearson correlations, 487t 367 purpose, 483 reference standards, 488 MAACL. See Multiple Affect Adjective Check reliability, 485 List review, 483–491 MacDonald, A. J. D. scale format, 485 Brief Assessment Scale-Depression (BAS-D), scoring methods, 484–85 378 short form MPQ, 487–488 Macmillan, Allister M. validity, 486–87 Health Opinion Survey (HOS), 210 validity comparison to European MADRS. See Montgomery-Åsberg Depression Organization for Research and Treatment Rating Scale of Cancer Quality of Life Questionnaire, magnitude estimation, 18, 20, 713 555 Mahoney, Florence I. validity comparison to Functional Living Barthel Index, 66 Index—Cancer, 544 MAI. See Multilevel Assessment Instrument validity comparison to Nottingham Health Major Depression Rating Scale, 374 Profile, 643 Manifest Anxiety Scale (TMAS), 274–75, 277, validity comparison to visual analogue pain 278t, 279–84 rating scale, 479 alternative forms, 283 validity comparison to WHO Quality of Life conceptual basis, 279–80 Scale, 626 description, 280 McHorney, C. A., 650, 654, 708 exhibit, 281–82 MCID. See Minimal Clinically Important purpose, 279 Difference 736 Index McMaster Health Index Questionnaire Medical Outcomes Study SF-36, 520, 523, (MHIQ), 523, 525t, 612–19 525t, 649–62, 708 conceptual basis, 612 alternative forms, 658–59 description, 612–13 conceptual basis, 649–50 exhibit, 614–18 Cronbach alpha coefficients, 655t measurement(s) description, 650, 653–54 purpose, 612 exhibit, 651–53 reliability, 613 population norms for scales, 660t review, 612–619 purpose, 649 scale format, 614–618 reference standards, 659 validity, 613, 618 reliability, 654, 656 validity comparison to Arthritis Impact scoring, 653–54 Measurement Scales, 531 validity, 656–58 validity comparison to Nottingham Health Medical Outcomes Study Social Support Profile, 645 Survey, 165 validity comparison to Sickness Impact Medical Outcomes Trust, 705 Profile, 635 Meenan, Robert F. McMaster-Toronto Arthritis and Rheumatism Arthritis Impact Measurement Scales (MACTAR), 613, 690 (AIMS), 526 MDS. See Minimum Data Set Melzack, R. measurement gate control theory of pain, 471, 484 anxiety, 277 See also McGill Pain Questionnaire defined, 10 (MPQ) error in, 39–42 memory, 394, 395 objective, 15 Mental Function Index (MFI), 465–66 quality of, 30–46 Mental Health Inventory (MHI), 209 scales of, 715 Mental Health Inventory, 252 of social adjustment, 151 review, 247–253 subjective, 12, 15, 16–18, 25–27 scale format, 249–251 theoretical bases for, 16–18 validity comparison to the General Health See also health measurement; reliability; Questionnaire, 266 validity; specific instruments and scales validity comparison to Positive and Negative Medical Outcomes Study Pain Measures, 474, Affect Scale, 228 475t, 496–98 mental illness, 216 Medical Outcomes Study Physical Functioning Mental Impairment Measure (MIM) Measure, 56, 60t, 119–22 validity comparison to Mental Status alternative forms, 120 Questionnaire, 424 conceptual basis, 119–20 mental status description, 120 measurements of, 395 exhibit, 121 Mental Status Questionnaire (MSQ), 423–25 purpose, 119 review, 423–426 reliability, 120 scale format, 423 validity, 120 validity comparison to Blessed Information- Medical Outcomes Study SF-12, 523, 525t, Memory-Concentration test, 419 666–70 validity comparison to Physical Self- alternative forms, 669 Maintenance Scale, 87 conceptual basis, 666 validity comparison to Short Portable Mental description, 666, 668 Status Questionnaire, 428 exhibit, 667–68 mental status testing, 394–469 purpose, 666 See also specific scales reference standards, 670 MERGE-CARE, 607 reliability, 668 methodological classifications, 12, 13–15 validity, 668–69 MFI. See Mental Function Index  Index 737 MHI5. See Mental Health Inventory validity comparison to Zung Self-Rating Middlesex Hospital Questionnaire, 391 Depression Scale, 347 Mini Examen Cognitivo (MEC), 435 MMMS. See Modified Mini-Mental State Minimal Clinically Important Difference Examination (MCID), 46, 713 MMPI. See Minnesota Multiphasic Personality Minimally Important Difference (MID), 46, Inventory 713 MMSE. See Mini-Mental State Examination Minimum Data Set (MDS), 142 Modified Barthel Index, 67 Mini-Mental State Examination (MMSE), 397, Modified Health Assessment Questionnaire, 116 398t, 429–37, 465–66 Modified Mini-Mental State Examination alternative forms, 434–35 (3MS), 398t, 441–47, 466 conceptual basis, 429–30 alternative forms, 447 description, 430 conceptual basis, 441 exhibit, 431 description, 441 purpose, 429 exhibit, 442–45 reference standards, 435 purpose, 441 reliability, 430–31, 431t reference standards, 447 review, 429–441 reliability, 446 scale format, 431 review, 441–49 sensitivity and specificity, 433t scale format, 442–45 validity, 431–34 validity, 446–47 validity comparison to Modified Mini- validity comparison to Blessed Dementia Mental State Examination, 446 Scale, 421 validity comparison to Alzheimer’s Disease validity compariston to Mini-Mental State Assessment Scale, 413 Examination, 446 validity comparison to Blessed Information- Mohs, Richard C., 411 Memory-Concentration test, 419 Montgomery-Åsberg Depression Rating Scale validity comparison to Cambridge Mental (MADRS), 333, 334t, 382–86 Disorders of the Elderly Examination, 462, alternative forms, 386 463 conceptual basis, 382 validity comparison to Clifton Assessment description, 382 Procedures for the Elderly, 458 exhibit, 383–85 validity comparison to Clock Drawing task, purpose, 382 409 reliability, 382, 385 validity comparison to Cognitive Capacity review, 382–87 Screening Examination, 406 scale format, 383–85 validity comparison to Dementia Rating validity, 385–86 Scale, 401 validity comparison to Geriatric Depression validity comparison to Functional Activities Scale, 361 Questionnaire, 108 validity comparison to Hamilton Rating validity comparison to Mental Status Scale for Depression, 372 Questionnaire, 424 validity comparison to Hospital Anxiety and validity comparison to Informant Depression Scale, 298 Questionnaire on Cognitive Decline in the mood disorders, 329, 330 Elderly, 451–454 moral anxiety, 274 validity comparison to Short Portable Mental morale, 231, 521 Status Questionnaire, 428 mortality indicators, 10–12, 15 Minnesota Multiphasic Personality Inventory Moskowitz, Eugene (MMPI) PULSES Profile, 62 validity comparison to Beck Depression MOS Pain Measures, 496 Inventory, 338 MOS Physical Function Measure, 650 validity comparison to Hamilton Rating review, 119–122 Scale for Depression, 373 scale format, 121 738 Index MOS Social Support Survey, 153, 154t, review, 639–49 165–67 scale format, 641–42 MOS-HIV, 658–59 validity, 643–45 validity comparison to European Quality of validity comparison to Dartmouth COOP Life Measure EQ-5D, 697 Charts, 573–74 MPQ. See McGill Pain Questionnaire validity comparison to Duke Health Profile, MSQ. See Mental Status Questionnaire 592–94 multiattribute utility theory, 24–25 validity comparison to European Quality of multidimensional scaling, 37 Life Measure EQ-5D, 699–700 Multilevel Assessment Instrument (MAI), 523, validity comparison to Medical Outcomes 525t, 609–11 Study SF-36, 654, 657–58 review, 609–11 validity comparison to Sickness Impact scale format, 610 Profile, 635 Multiple Affect Adjective Check List numerical scaling techniques, 16 (MAACL), 364 validity comparison to Beck Depression OARS. See Older Americans Resources and Inventory, 338 Services multitrait-multimethod matrix, 713–14 OASIS. See Outcome and Assessment multitrait scaling analysis, 44 Information Set Older Americans Resources and Services NCSE. See Neurobehavioral Cognitive Status (OARS) Instrumental Activities of Daily Examination Living scale, 596, 600–603 negative affect, 207–9, 221, 225 validity comparison to Functional Autonomy Nelson, Eugene C. Measurement System, 139 Dartmouth COOP Charts, 569 validity comparison to Functional Status Neugarten, Bernice L. Questionnaire, 590 Life Satisfaction Index, 231 OARS Multidimensional Functional Neumann/Morgenstern model of decision- Assessment Questionnaire (OMFAQ), 82, making, 23 202, 523, 524t, 596–603 Neurobehavioral Cognitive Status Examination alternative forms, 602 (NCSE), 466 conceptual basis, 596 neuromatrix model of pain, 471 description, 596–97 neurotic anxiety, 274 exhibit, 598–99, 600–601 Newcastle Dementia Scale, 417 purpose, 596 Newcastle Memory, Information, and reference standards, 602 Concentration Test, 417 reliability, 597, 599 NHP. See Nottingham Health Profile review, 596–604 nociception, 470, 471 scale format, 598–601 nominal scales, 18, 715 validity, 599, 601–2 nomothetic approach to measurement, 14, 27, objective anxiety, 274, 275 28 objective measurement, 15 nonspecific psychological distress, 207 observed score variance, 45 Norbeck Social Support Questionnaire, 204 OECD. See Organization for Economic norms, 151, 152 Cooperation and Development (OECD) Nottingham Health Profile (NHP), 523, 525t, Long-Term Disability Questionnaire 639–46 Older Americans Resources and Services alternative forms, 645 (OARS), conceptual basis, 640 OMC. See Orientation-Memory-Concentration description, 640 Test exhibit, 641–42 OMFAQ. See OARS Multidimensional purpose, 639–40 Functional Assessment Questionnaire reference standards, 645 onion principle, 11 reliability, 643 ordinal scales, 18–19, 715  Index 739 Organization for Economic Cooperation and purpose, 514 Development (OECD) Long-Term review, 514–17 Disability Questionnaire, 57, 59, 93–95 scale format, 516 review, 93–95 reliability, 515, 517 scale format, 93 validity, 517 Orientation-Memory-Concentration Test PAIRS. See Pain and Impairment Relationship (OMC) Scale validity comparison to Clock Drawing task, PAMIE. See Physical and Mental Impairment- 409 of-Function Evaluation validity comparison to Mental Status PANAS-C. See Positive and Negative Affect Questionnaire, 424 Scale validity comparison to Mini-Mental Status PANAS-X. See Positive and Negative Affect Examination, 432 Scale Oswestry Low Back Pain Disability panic attacks, 275 Questionnaire, 474, 475t, 498–501 Parkerson, George R. review, 498–501 Duke Health Profile (DUKE), 591 scale format, 499 Duke Social Support and Stress Scale validity comparison to visual analogue pain (DUSOCS), 170 rating scale, 479 PAS. See Psychogeriatric Assessment Scales Oucher Index, 518 path analysis, 714 Outcome and Assessment Information Set Patient Evaluation Conference System (PECS), (OASIS), 144 60t, 103–7 conceptual basis, 103–4 PAD. See Pain and Distress Scale description, 104 PADL. See Performance Activities of Daily exhibit, 105–7 Living purpose, 103 pain, 470–519 reliability, 104 analogue methods of measurement, 474 review, 103–7 approaches to measurement, 472–73 scale format, 105–7 back, 28, 518 validity, 104 behavioral measurement, 472–74 Patient-Generated Index (PGI), 13, 522 definition of, 472, 517 validity comparison to European Quality of questionnaire techniques, 473 Life Measure EQ-5D, 699 research, 27 Patrick, Donald L. theoretical approaches, 470–72 Lambeth Disability Screening Questionnaire, treatments for, 472 89 visual analogue rating scales, 477–81 Pattie, A. H. See also specific measurement scales Clifton Assessment Procedures for the Pain and Distress Scale (PAD), 474, 475t, Elderly (CAPE), 456 506–7 Pearson correlations, 36, 41, 45, 487t, 711 review, 507–7 PECS. See Patient Evaluation Conference scale format, 507 System Pain and Impairment Relationship Scale Pediatric Oncology Quality of Life (PAIRS), 517, 518 Questionnaire (POQOL) Pain Disability Index, 518 validity comparison to Health Utilities Index, Pain Drawing Test 690 validity comparison to Back Pain performance, 152 Classification Scale, 504 Performance Activities of Daily Living (PADL), Pain Perception Profile (PPP), 473, 474, 475t, 56. See also Instrumental Activities of 514–17 Daily Living (IADL) scales conceptual basis, 514–15 performance measures of function, 57 description, 515 performance versus capacity question wording, exhibit, 516 15–16, 57, 66–67 740 Index Personal Capabilities Questionnaire, 540 Pilowsky, I. Personality and Social Network Adjustment Illness Behavior Questionnaire (IBQ), 508 Scale, 204 policy analysis, 24 personality traits, 25 POMS. See Profile of Mood States Personal Resource Questionnaire, 157, 204 Popoff Index of Depression, 391 Pfeffer, Robert I. POQOL. See Pediatric Oncology Quality of Functional Activities Questionnaire (FAQ), Life Questionnaire 108 positive affect, 207–9, 221, 225 Mental Function Index (MFI), 465 Positive and Negative Affect Scale (PANAS), Pfeiffer, Eric 209, 211t, 225–30, 272, 318t Functional Assessment Inventory, 602 alternative forms, 228–29 Short Portable Mental Status Questionnaire conceptual basis, 225–26 (SPMSQ), 426 description, 226 PFQ. See Present Function Questionnaire exhibit, 227 PGI. See Patient-Generated Index purpose, 225 PGWB. See Psychological General Well-Being reference standards, 229 Index reliability, 226–27 Philadelphia Geriatric Center review, 225–31 Delayed Memory Test, 425 scale format, 227 Philadelphia Geriatric Center Morale Scale, validity, 227–28 207, 209, 211t, 236–39 validity comparison to Affect Balance Scale, conceptual basis, 236 222 description, 236–37 validity comparison to Social Adjustment exhibit, 238 Scale, 191–92 purpose, 236 validity comparison to Mental Health reference standards, 239 Inventory, 252 reliability, 237 positive predictive value, 714 review, 236–40 power law, 17–18 scale format, 238 PPP. See Pain Perception Profile validity, 237–39 predictive indexes, 13 validity comparison to Life Satisfaction predictive validity, 31–32, 714 Index, 233 preference-based health measures, 13 Physical and Mental Impairment-of-Function Present Function Questionnaire, (PFQ), 396 Evaluation (PAMIE), 523, 524t, 535–38 Present State Exam (PSE), 396 review, 535–38 validity comparison to Katz Adjustment scale format, 536–38 Scales, 175 physical disability. See disability validity comparison to Social Adjustment Physical Function Measure Scale, 191–92 review, 119–22 validity comparison to Social Functioning scale format, 121 Schedule, 179 Physical Self-Maintenance Scale (PSMS), 59t, primary biological functions, 56 84–88 prognostic measures, 13 alternative forms, 87 Profile of Mood States (POMS) conceptual basis, 84 validity comparison to European description, 84–87 Organization for Research and Treatment exhibit, 85–86 of Cancer (EORTC) Quality of Life purpose, 84 Questionnaire QLQ-C30, 554, 555 reliability, 87 validity comparison to Functional review, 84–88 Assessment of Cancer Therapy, 547, 549 scale format, 85–86 validity comparison to visual analogue scale, validity, 87 582 validity comparison to Multilevel Assessment PSE. See Present State Exam Instrument, 610 PSMS. See Physical Self-Maintenance Scale  Index 741 psychiatry, 151 exhibit, 566–67 psychodynamic pain theories, 471 purpose, 564 Psychogeriatric Assessment Scales (PAS), 396 reference standards, 568 Psychological General Well-Being Index reliability, 565 (PGWB), 244 review, 564–569 Psychological Mental Health Index, 245 scale format, 566–67 psychological well-being, 206–72, 708 validity, 565, 567 psychometrics, 16–23, 704, 717 validity comparison to Functional psychophysical experiments, 25 Independence Measure, 146 psychophysics, 16–18 Quality of Life Index Self-Assessment, 567 psychotherapy, 199 Quality of Life Interview (QOLI), 703 PULHEMS Profile, 62 Quality of Life Questionnaire (QLQ-C30) PULSES Profile, 55, 59t, 62–66 review, 551–59 alternative forms, 63 scale format, 552–53 conceptual basis, 62 Quality of Life Research, 709 description, 62–63 Quality of Well-Being Scale (QWB), 523, 525t, exhibit, 64 675–81 reliability, 63 alternative forms, 680 review, 62–66 conceptual basis, 675 revised version, 65 description, 675–78 scale format, 64–65 exhibit, 676, 678 validity, 36, 63 purpose, 675 validity comparison to Barthel Index, 71–72 reference standards, 680 validity comparison to Functional reliability, 678–79 Independence Measure, 146 review, 675–83 scale format, 676, 678 QALY. See quality-adjusted life year validity, 679–80 QL Index. See Quality of Life Index validity comparison to European QLQ-C30. See Quality of Life Questionnaire Organization for Research and Treatment QOLI. See Quality of Life Interview of Cancer Quality of Life Questionnaire, Q-TWiST. See Quality-Adjusted Time Without 697, 699 Symptoms and Toxicity Method validity comparison to Functional Status quality-adjusted life year (QALY), 23–25 Index, 102 Quality-Adjusted Time Without Symptoms and validity comparison to Health Assessment Toxicity (Q-TWiST), 523, 559–63 Questionnaire, 115 alternative forms, 563 validity comparison to Health Utilities Index, conceptual basis, 559–60 692 description, 560–63 validity comparison to Health Perceptions exhibit, 560–62 Questionnaire, 257 purpose, 559 validity comparison to Quality-Adjusted review, 559–64 Time Without Symptoms and Toxicity, 563 quality of life, 520–703 validity comparison to Rosser Disability and definition of, 521, 522 Distress Scale, 673–74 measurements, 151, 520–22 validity comparison to Medical Outcomes quality-adjusted life year, 23–24 Study SF-36, 657 single-item indicators, 578–84 validity comparison to Spitzer Quality of Life as subjective, 15, 26 Index, 565 See also specific measurement scales Quality receiver operating characteristic curve Quality of Life Index (QL Index), 523, 524t, (QROC), 33 564–68 question wording, 13, 14 alternative forms, 567–68 performance versus capacity, 15 conceptual basis, 564 Quick-Functional Living Index—Cancer, 545 description, 564–65 QWB. See Quality of Well-Being Scale 742 Index Radloff, L. S. RCI. See reliable change index Center for Epidemiologic Studies Depression RDS. See Raskin depression scale Scale (CES-D), 350 rebound effect, 26 RAND Corporation, 649 receiver operating characteristic curve (ROC), RAND Corporation Mental Health Inventory 26–27, 33, 713, 714 (MHI), 211t, 247–53, 272 reliability, 30, 39–46, 714 alternative forms, 252 alpha coefficient, 43–44 conceptual basis, 247–48 coefficient theta, 43 description, 248 assessing, 39–46 exhibit, 249–51 definition of, 40, 714 purpose, 247 Guttman analysis, 21, 715 reference standards, 252 internal consistency, 42–45, 708 reliability, 248 interpreting coefficients, 44–46 review, 247–53 inter-rater, 40, 712 scale format, 249–51 kappa coefficient, 45 validity, 248, 252 relation to alternative measurements, 708 validity comparison to the General Health of single-item health indicators, 581 Questionnaire, 266 smallest real difference (SRD), 42 validity comparison to the RAND Spearman-Brown, 44 Corporation Social Health Battery, 162 split half, 43 RAND Corporation Short-Form-36 Health Standard error of measurement, 41–42 Survey, 649 test-retest, 40, 717 RAND Corporation Social Health Battery, 153, See also specific instruments and scales 154t, 161–65 reliable change, 714 conceptual basis, 161 Reliable Change Index (RCI), 42, 52, 714 description, 162 reliable change/reliable individual change, 714 exhibit, 163–64 Remington, Marina purpose, 161 Social Functioning Schedule (SFS), 177 reference standards, 162 repeatability, 42 reliability, 162 repertory grid technique, 14–15 review, 161–65 Research Mattis Dementia Rating Scale scale format, 163–64 (RMDRS), 42 scoring method, 164 response shift, 26, 522, 714 validity, 162 responsiveness, 37–39, 42, 714 random errors, 40 responsiveness statistics, 37–39 Rapid Disability Rating System, 60t, 97–100 Revised PGC Morale Scale, 238 description, 97, 98 rho, 711, 714 exhibit, 98 ridit, 714–15 reference standards, 99 risk, 20, 24 reliability, 99 Ritchie Articular Index review, 97–100 validity comparison to Health Assessment scale format, 98 Questionnaire, 115 validity, 99 Rivermead ADL scale, 57 Rasch, Georg RMDRS. See Research Mattis Dementia Rating Rasch latent trait analysis 19, 21–23 Scale Rasch measurement model, 15, 21–22, 42, ROC. See receiver operating characteristic 708 curve Raskin depression scale (RDS), 391 Roland Disability Questionnaire, 479, 518, validity comparison to Montgomery-Åsberg 636 Depression Rating Scale, 385 Roland Morris Pain Questionnaire Rating of Social Disability, 203 validity comparison to Oswestry Low Back rating scales, 13–14 Pain Disability Questionnaire, 500 ratio scales, 19, 715 roles, 151–52  Index 743 Rosen, Wilma G. scores, 15, 716–17 Alzheimer’s Disease Assessment Scale screening tests (ADAS), 411 cognitive, 44, 395, 465–66 Rosser, Rachel M. for emotional disorders, 37 Disability and Distress Scale, 671 validity of, 32–34 Index of Health-Related Quality of Life SDRS. See Social Dysfunction Rating Scale (IHQL), 673 SDS. See Self-Rating Depression Scale Roth, Sir Martin SDT. See Signal Detection Theory Cambridge Mental Disorders of the Elderly SEIQOL. See Schedule for the Evaluation of Examination (CAMDEX), 460 Quality of Life Rotterdam Symptom Checklist (RSCL) Self-Anchoring Scale, 579 validity comparison to Hospital Anxiety and Self-Evaluation of Life Function (SELF), 523, Depression Scale, 297 525t, 611–12 RSCL. See Rotterdam Symptom Checklist Self-Rating Anxiety Scale (SAS), 277, 278t, Ruesch, J. 302–6 Rating of Social Disability, 203 alternative forms, 304 conceptual basis, 302 SAD Index for the Clinical Assessment of Pain, description, 302–4 518 exhibit, 305 St. Thomas’s Pain Questionnaire, 500 purpose, 302 Sandoz Clinical Assessment-Geriatric (SCAG), reference standards, 304–5 467 reliability, 304 validity comparison to Hospital Anxiety and review, 302–6 Depression Scale, 297 scale format, 305 Sarason, Irwin G. validity, 304 Dyadic Effectiveness Scale, 160 Self-Rating Depression Scale (SDS), 333, 334t, Social Support Questionnaire (SSQ), 158 344–49 SAS. See Social Adjustment Scale alternative forms, 347–48 SCAG. See Sandoz Clinical Assessment- conceptual basis, 345 Geriatric description, 345 Scaled General Health Questionnaire, 259 exhibit, 346 scales of measurement, 18–20, 715 purpose, 344 scaling methods, 18–25, 715 reference standards, 348 comparative techniques, 20 reliability, 345 econometric methods, 23–25 review, 344–50 equal-appearing intervals, 20 scale format, 346 magnitude estimation, 20 validity, 345–47 psychometric methods, 20–23 validity comparison to Beck Depression standard gamble, 24 Inventory, 338 time trade-off, 24 validity comparison to Depression Adjective utilities, 23, 23 Check Lists, 366 visual analogue scales, 391, 477–83, validity comparison to Duke Health Profile, 580–83 594 scalogram analyses, 21, 715. See also Guttman validity comparison to General Well-Being scaling Schedule, 243–245 Schedule for Affective Disorders and validity comparison to Geriatric Depression Schizophrenia (SADS), 267, 391 Scale, 361 Schedule for the Evaluation of Quality of Life validity comparison to Hamilton Rating (SEIQOL), 522 Scale for Depression, 373 Schipper, H. validity comparison to Hospital Anxiety and Functional Living Index-Cancer, 541 Depression Scale, 298 Schoening, Herbert A. validity comparison to Illness Behavior Kenny Self-Care Evaluation, 78 Questionnaire, 512 744 Index Self-Rating Depression Scale (SDS) (continued) Short-Form-36 Health Survey (SF-36) validity comparison to Manifest Anxiety review, 649–665 Scale, 283 scale format, 651–653 validity comparison to Self-rating Anxiety validity comparison to Arthritis Impact Scale, 304 Measurement Scales, 531 validity comparison to visual analogue validity comparison to Dartmouth COOP depression scale, 391 Charts, 574 and visual analogue scales, 391 validity comparison to European Quality of self-rating methods, 332, 391–92 Life Measure EQ-5D, 698–99 SEM. See Standard Error of Mmeasurement validity comparison to Nottingham Health sensibility, 31 Profile, 643–45 sensitivity, 27, 32–34, 38, 713, 716 validity comparison to Quality of Well-Being sensitivity to change, 37–39, 714 Scale, 679–80 sensory decision theory. See decision analysis validity comparison to Sickness Impact SES. See Standardized Effect Size Profile, 635 SFS. See Social Functioning Schedule validity comparison to Social Maladjustment SF-12. See Short-Form-12 Health Survey Schedule, 196 SF-36. See Short-Form-36 Health Survey validity comparison to WHO Quality of Life Shader, R. I. Instrument, 625–26 Sandoz Clinical Assessment-Geriatric Short-Form-36 Health Survey. See Medical (SCAG), 467 Outcomes Study SF-36 Sherbourne, Cathy Short Portable Mental Status Questionnaire Medical Outcomes Study Pain Measures, (SPMSQ), 397, 398t, 426–28 496 review, 426–29 Social Support Survey, 165 scale format, 426 Short Blessed Test (SBT), 419 validity comparison to Clock Drawing task, Short-Cambridge Mental Disorders of the 409 Elderly Examination (CAMDEX), 463 validity comparison to Cognitive Capacity Short-Comprehensive Assessment and Referral Screen Examination, 406 Evaluation (CARE), 605 validitiy comparison to Older Americans validity comparison to Zung Self-rating Resources and Services Multidimensional Depression Scale, 347 Functional assessment Questionnaire, 602 Short-Form-6D (SF-6D), 658–59 validity comparison to Informant Short-Form-8 (SF-8), 658 Questionnaire on Cognitive Decline in the Short-Form-12 Health Survey (SF-12) Elderly, 452 review, 666–71 Short Psychiatric Evaluation Schedule, 602 scale format, 667–68 Short Test of Mental Status (STMS), 465 validity comparison to EuroQol EQ-5D, Sickness Impact Profile (SIP), 518, 520, 525t, 698 630–37 validity comparison to SF-36, 669 alternative forms, 635–36 Short-Form-12 Health Survey. See Medical conceptual basis, 630–31 Outcomes Study SF-12 correlations between scores and criterion Short-Form-20 Health Survey (SF-20), 649 variables, 634t validity comparison to COOP Charts, description, 631 573–74 exhibit, 632 validity comparison to Duke Health Profile, purpose, 630 592–94 reliability, 631, 633 validity comparison to Functional Status review, 630–39 Questionnaire, 590 scale format, 632 validity comparison to Older Americans SIP68, 636 Resources and Services ADL scale, 602 validity, 633–35 validity comparison to visual analogue scale, validity comparison to Arthritis Impact 582 Measurement Scales, 531,  Index 745 validity comparison to Duke Health Profile, reliability, 184, 191 592, 594 review, 184–94 validity comparison to European scale format, 185–91 Organization for Research and Treatment validity, 191–92 of Cancer Quality of Life Questionnaire Social Disability Questionnaire, 203 QLQ-C30, 555 Social Dysfunction Rating Scale (SDRS), 153, validity comparison to Functional Status 154t, 196–99 Questionnaire, 590 review, 196–99 validity comparison to Health Assessment scale format, 198 Questionnaire, 115 Social Evaluation and Social Support Schedule, validity comparison to Health Perceptions 183 Questionnaire, 257 Social Functioning Scale, 204 validity comparison to Katz Adjustment Social Functioning Schedule (SFS), 153, 154t, Scales, 175 177–80 validity comparison to McMaster Health description, 177 Index Questionnaire, 618 exhibit, 178–79 validity comparison to Philadelphia Geriatric purpose, 177 Center Morale Scale, 237 reliability, 179 validity comparison to Quality of Well-Being review, 177–80 Scale, 679–80 scale format, 178–79 validity comparison to SF-36, 657 validity, 179 validity comparison to Spitzer Quality of Life social health, 150–205 Index, 565 Social Maladjustment Schedule, 153, 154t, SIGH-A, 288 194–96 SIGH-D. See Structured Interview Guide for review, 194–96 Depression scale format, 195 SIDAM. See Structured Interview for the social networks, 152–53 Diagnosis of Dementia of the Alzheimer social reforms, 11 Type Social Relationship Scale (SRS), 153, 154t, SI-HDRS, 374 155–57 Signal Detection Theory (SDT), 26–27 review, 155–58 single indicator, 15 scale format, 156 single-item health measures social roles, 151–52 review, 578–87 social support, 56, 152–53 scale formats, 579–581 Social Support Questionnaire (SSQ), 153, 154t, Single-Item Indicators of Well-Being, 524t 158–61 single score, 15 alternative forms, 160 SIP. See Sickness Impact Profile conceptual basis, 158 Sister Kenny Institute, 78 description, 158 SMAF. See Functional Autonomy Measurement exhibit, 159–60 System reliability, 158 smallest real difference (SRD), 42 review, 158–61 Snaith, R. P., 294 scale format, 159–60 social adjustment, 150, 151–52 validity, 158, 160 Social Adjustment Scale (SAS), 153, 154t, Social Support Survey 184–93, 202 review, 165–67 alternative forms, 192 scale format, 159–60 conceptual basis, 184 Somatic Symptom Inventory (SSI) description, 184 validity comparison to mental Health exhibit, 185–91 Inventory, 252 as most carefully developed scale, 204 Sorant Battery purpose, 184 validity comparison to Mental Status reference standards, 192 Questionnaire, 424 746 Index Spearman’s rho, 711, 714 validity comparison to Functional Living specific health measures, 13 Index - Cancer, 544 specificity, 32–34, 35, 713, 716 validity comparison to General Health spectrum bias, 33 Questionnaire, 268 Spence, N. D., 508 validity comparison to General Well-Being Spielberger, C. D. Schedule, 243–44 State-Trait Anxiety Inventory (STAI), 319 validity comparison to Hospital Anxiety and Spitzer, W. O., 564, 702 Depression Scale, 298 Quality of Life Index (QL Index), 564 validity comparison to Illness Behavior SQLP. See Subjective Quality of Life Profile Questionnaire, 512 SQMSQ. See Short Portable Mental Status validity comparison to Positive and Negative Questionnaire Affect Scale, 228 SRD. See smallest real difference validity comparison to Zung Self-Rating SRM. See Standardized Response Mean Depression Scale, 347 SRS. See Social Relationship Scale validity comparision to Depression Anxiety SSI. See Somatic Symptom Inventory Stress Scales, 317–18 SSIAM. See Structured and Scaled Interview to State-Trait Depression Adjective Check Lists, Assess Maladjustment 367 SSQ. See Social Support Questionnaire Stern, Isaac, 520 STAI. See State-Trait Anxiety Inventory Stevens, S. S., 17 Standard Error of Measurement (SEM), 41–42, Stewart, Anita 716 Physical Functioning Measure, 119 standard gamble, 24, 716 Social Support Survey, 165 Standardized Effect Size (SES), 38 STMS. See Short Test of Mental Status Standardized Response Mean (SRM), 38–39 Stockton Geriatric Rating Scale, 457 standard scoring, 716–17 Storandt Battery, 465 Stanford Health Assessment Questionnaire. Stroke Unit Mental Status Examination See Health Assessment Questionnaire (SUMSE), 466 (HAQ) Structured and Scaled Interview to Assess State-Trait Anxiety Inventory (STAI), 275, 277, Maladjustment (SSIAM), 153, 154t, 278t, 318t, 319–26 199–203, 707 alternative forms, 324–25 conceptual basis, 199 conceptual basis, 319–20 description, 199–200 description, 320–21 exhibit, 201–2 exhibit, 321 purpose, 199 internal consistency coefficients, 322t reliability, 200 purpose, 319 review, 199–203 reference standards, 325 scale format, 201–2 reliability, 321 validity, 200 review, 319–27 Structured Interview for the Diagnosis of STAI-C, 324 Dementia of the Alzheimer Type (SIDAM), validity, 322–24 467 validity comparison to Affect Balance Scale, Structured Interview Guide for Depression 223 (SIGH-D), 374 validity comparison to Beck Anxiety Structured Interview Guide for the Hamilton Inventory, 308–9 Anxiety Rating Scale, 289–91 validity comparison to Beck Depression subjective measurements, 12, 15, 16–18, 25–27 Inventory, 338 Subjective Quality of Life Profile (SQLP) validity comparison to Center for validity comparison to WHO Quality of Life Epidemiologic Studies Depression Scale, Instrument, 625–26 353 summary self-ratings of health validity comparison to Delighted-Terrible review, 578–87 scale, 582 scale formats, 579–81  Index 747 SUMSE. See Stroke Unit Mental Status review, 216–21 Examination scale format, 217–18 Swift, Jonathan, 520 validity, 219 System Check List-90 (SCL-90) validity comparison to Center for validity comparison to Beck Anxiety Epidemiologic Studies Depression Scale, Inventory, 309 353 validity comparison to Center for validity comparison to the General Well- Epidemiologic Studies Depression Scale, Being Schedule, 245 354 Tyrer, P.J. symptom checklist approach, 206–7 Social Functioning Schedule (SFS), 277 Symptom Questionnaire, 328 systematic errors, 40 UDS. See Uniform Data System for Medical Rehabilitation Taylor, J. A. Uniform Data System for Medical Taylor Manifest Anxiety Scale (TMAS), 279 Rehabilitation (UDS), 72, 141, 147 Taylor Manifest Anxiety Scale (TMAS). See unimpaired consciousness, 394 Manifest Anxiety Scale unreliability, 40, 716 Telephone Interview for Cognitive Status utilities (TICS), 435 defined, 19–20, 717 Tellegen, A., 225 in econometric scaling, 23–25 Teng, Evelyn utility scaling methods, 24 Cognitive Abilities Screening Instrument, 434, 447, 466 validity Modified Mini-Mental State Test, 441 assessing, 30–39 Tennessee Self-Concept Scale central theme of, 11 validity comparison to Duke Health Profile, concurrent, 31–34, 711 594 construct, 34–37, 39, 708, 711 Test Anxiety Inventory content, 30–31, 711 validity comparison to State-Trait Anxiety convergent, 34–36, 44, 711 Inventory, 324 correlational evidence of, 34–36, 711 test-retest reliability, 717 criterion, 31–34, 708, 711 3MS. See Modified Mini-Mental State definition of, 30, 717 Examination discriminant, 32, 35, 37–39, 711 time trade-off, 717 divergent, 35, 36 Timed Up and Go test (TUG), 57 effect size, 37–39, 714 TMAS. See Manifest Anxiety Scale face, 31 Torrance, George W. factorial, 36–37 Health Utilities Index (HUI), 683 group differences and sensitivity to change, trait anxiety, 275, 320 37–39 traits, 275 likelihood ratios, 33 translation, 14 multitrait-multimethod matrix, 713–14 true score, 40, 44, 45 predictive, 31–32, 714 TUG. See Timed Up and Go test of psychological measures, 208, 708 Tursky, B. receiver operating characteristic curve Pain Perception Profile (PPP), 514 analyses, 26–27, 33, 714 Tursky Pain Perception Profile. See Pain of screening test, 34 Perception Profile (PPP) sensitivity and specificity, 27, 32, 716 Twenty-Two Item Screening Score of Psychiatric of single-item health indicators, 582–84 Symptoms, 209, 211t, 216–20, 271 See also specific instruments and scales description, 216, 219 values, 19–20, 24 exhibit, 217–18 VAS. See Visual Analogue Scales purpose, 216 Veit, C. T., 247, 248 reliability, 219 verbal report, 12 748 Index verification, 34 WHO Disability Questionnaire, 94 Veterans SF-36 Health Survey, 658 WHOQOL. See WHO Quality of Life Scale Visual Analogue Pain Rating Scales (VAS), WHO Quality of Life Scale-BREF, 626–27 695–98, 717 WHO Quality of Life Scale-HIV, 627 review, 477–83 WHO Quality of Life Scale-OLD, 627 scale format, 478, 480 WHO Quality of Life Scale (WHOQOL), 520, validity comparison to Health Assessment 525t, 619–29, 702, 704 Questionnaire, 115 alternative forms, 626–27 validity comparison to Oswestry Low Back conceptual basis, 619–20 Pain Disability Questionnaire, 500 description, 620–21 Visual Analogue Scales (VAS), 18, 391, 474, exhibit, 622–25 475t, 477–83, 580–83 purpose, 619 alternative forms, 479–80 reference standards, 627 conceptual basis, 477 reliability, 621, 625 definition of, 717 review, 619–29 description, 477–78 scale format, 622, 25 exhibit, 478, 480 validity, 625–26 purpose, 477 validity comparison to the McGill Pain reference standards, 480 Questionnaire, 487 reliability, 478–79 validity comparison to single-item rating and single-item health indicators, 580–81, scale, 583 582–83 WHYMPI. See West Haven-Yale validity, 479 Multidimensional Pain Inventory Wisconsin Brief Pain Questionnaire. See Brief Ware, J.E., Jr., 247–48, 253, 254, 649, 666, Pain Inventory 705, 707 WONCA, 574–75 Health Perceptions Questionnaire (HPQ), World Health Organization Quality of Life 253 Scale. See WHO Quality of Life Scale Mental Health Inventory (MHI), 247 (WHOQOL) Short-Form-12 Health Survey (SF-12), 666 World War II, 12 Short-Form-36 Health Survey, 649 (SF-36) Watson, D., 225, 276 Yesavage, Jerome A., 359 Positive and Negative Affect Scale (PANAS), Yule’s Q, 717 225 Wechsler Memory Scale, 465 Zigmond, A. WeeFIM, 146 Hospital Anxiety and Depression Scale Weissman, Myrna M. (HADS), 294 Social Adjustment Scale (SAS), 184 Zung, W.W. K. well-being Anxiety Status Inventory (ASI), 302 measurement of, 151, 206–8 Depression Status Inventory, 348 psychological, 206–72, 708 Pain and Distress Scale (PAD), 506 See also health; quality of life Self-Rating Anxiety Scale (SAS), 302 West Haven-Yale Multidimensional Pain Self-Rating Depression Scale (SDS), 344 Inventory (WHYMPI), 518 Zung Pain and Distress Scale. See Pain and Whiteley Index of Hypochondriasis Distress Scale (PAD) validity comparison to Illness Behavior Zung Self-Rating Depression Scale. See Self- Questionnaire, 508 Rating Depression Scale (SDS)