Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Umme S Warda

In 1954, the California Department of Corrections and Rehabilitation established the Parole Outpatient Clinic (POC) program to assist parolees with mental health problems and, as a consequence, reduce recidivism rates among this population. Since its inception until October 1, 2000, parole agents were primarily responsible for referring parolees to the POCs for services. Referrals would be made if the parolee had a history of mental illness (usually indicated by the receipt of mental health services while in prison), or if the parole agent perceived that the parolee showed signs of mental instability. However, under this approach a substantial proportion of otherwise eligible parolees were either not identified or not provided appropriate services.

FINAL REPORT ON THE MENTAL HEALTH SERVICES CONTINUUM PROGRAM OF THE CALIFORNIA DEPARTMENT OF CORRECTIONS AND REHABILITATION—PAROLE DIVISION Submitted to The California Department of Corrections and Rehabilitation Division of Parole Prepared by The UCLA Integrated Substance Abuse Program Neuropsychiatric Institute David Farabee, Principal Investigator Dave Bennett, Analyst David Garcia, Analyst Umme Warda, Statistician Joy Yang, Study Coordinator June 30, 2006 Final Report on the MHSCP Evaluation STATE OF CALIFORNIA DEPARTMENT OF CORRECTIONS AND REHABILITATION Kingston Prunty Undersecretary D.L. Runnels Chief Deputy Secretary Adult Operations Thomas Hoffman Director Division of Adult Parole Operations Marilyn Kalvelage Assistant Deputy Director Division of Adult Parole Operations Robert Ambroselli Associate Director Division of Adult Parole Operations Distributed by the: Division of Adult Parole Operations 1515 S Street, Rm 212 North Sacramento, CA 95814 (916) 327-4612 i Final Report on the MHSCP Evaluation TABLE OF CONTENTS Preface iii Executive Summary 1 I. MHSCP: Historical Context and Overview A. Background 6 B. Program Design and Description 7 II. Impact Evaluation A. Identification and Assessment of Eligible Inmates 10 B. Clinic Attendance 12 C. Characteristics of MHSCP and Non-MHSCP Parolees 15 D. Trends in Data Quality 16 E. Reduction in EOP Parolee Caseloads 16 III. Outcome Evaluation (12-Month Return to Custody) A. Comparisons by MHSCP Participation Status 18 B. Time in Program 21 IV. Cost Analysis 22 V. Clinician Interviews 23 VI. Conclusions 31 References 33 Appendix A: Figures 7 & 8 (Survival Curves) 34 Appendix B: Trends in PATS Data Quality 37 ii Final Report on the MHSCP Evaluation PREFACE In May of 2002, the California Department of Corrections and Rehabilitation—Parole and Community Services Division (in consultation with the Research Branch) selected the Integrated Substance Abuse Programs (ISAP) at the University of California, Los Angeles to conduct a process and outcome evaluation of the Mental Health Services Continuum Program (MHSCP). The MHSCP, according to its design, was to be applied to all eligible inmates released on or after October 1, 2000. This report summarizes findings from the fourth and final year of this evaluation, and is submitted pursuant to the approved scope of services which calls for an annual report on or before June 30th of each project year. iii Final Report on the MHSCP Evaluation Executive Summary Background In 1954, the California Department of Corrections and Rehabilitation established the Parole Outpatient Clinic (POC) program to assist parolees with mental health problems and, as a consequence, reduce recidivism rates among this population. Since its inception until October 1, 2000, parole agents were primarily responsible for referring parolees to the POCs for services. Referrals would be made if the parolee had a history of mental illness (usually indicated by the receipt of mental health services while in prison), or if the parole agent perceived that the parolee showed signs of mental instability. However, under this approach a substantial proportion of otherwise eligible parolees were either not identified or not provided appropriate services. To enhance the Department’s ability to identify and treat mentally ill parolees, the Mental Health Services Continuum Program (MHSCP) was developed by the Parole and Community Services Division (P&CSD) in July of 2000. According to its design, the MHSCP was to be applied to all eligible inmates released on or after October 1, 2000. The purpose of this report is to summarize the results of UCLA’s ongoing process and outcome evaluation of the MHSCP. Program Design and Description The MHSCP was designed to reduce the symptoms of mental illness among parolees by providing timely, cost-effective mental health services that optimizes their level of individual functioning in the community thereby reducing recidivism and improving public safety. The MHSCP is designed to include: • Pre-release needs assessment of paroling mentally ill inmates. • Pre-release benefits eligibility and application assistance. • Expanded and enhanced post-release mental health treatment for mentally ill parolees. • Improved continuity of care from the institution's Mental Health Service Delivery System to the community-based parolee outpatient clinics. • Increased assistance for successful reintegration into the community upon discharge from parole. • A standardized program in all four-parole regions. According to the MHSCP design, regional Transitional Case Management Program—Mental Illness (TCMP-MI) social workers are to conduct face-to-face assessments with eligible inmates within 90 days of the inmates’ estimated parole release date, and update this assessment information within 30 days of the inmates’ release. The TCMP-MI social worker then merges the assessment information into the Parole Automated Tracking System (PATS) database. This information is verified by the TCMP-MI liaison and forwards this information to the appropriate POC headquarters. Once received, a POC-MHSCP liaison consults with the inmates’ parole 1 Final Report on the MHSCP Evaluation agent of record (AOR) and schedules an initial appointment. For Enhanced Outpatient Program (EOP) parolees, this appointment is scheduled to occur within 3 working days of release; for Correctional Clinical Case Management System (CCCMS) parolees, the initial appointment is scheduled to occur within 7 working days of release. In general, the jurisdictions of the TCMPI-MI social workers are divided into northern and southern regions, with Kern County Department of Public Health serving as the headquarters for the northern region, and the University of California at San Diego serving as the headquarters for the southern region. Some exceptions to this regional approach (e.g., including San Quentin State Prison in the southern region) were made to achieve balance between the regional caseloads and to reduce costs. Upon leaving the institution, parolees return to one of four parole regions (typically based on the county of commitment). The headquarters for these regions are located in Sacramento (Region I), Oakland (Region II), Los Angeles (Region III), and Diamond Bar (Region IV). Impact Evaluation Except where otherwise indicated, the current evaluation focused on inmates who were released from prison between July 1, 2001 and June 30, 2005 (N= 60,912). Highlights of this portion of the evaluation are summarized below: Pre-Release Assessments • Overall, 54.5% of the eligible pool of releases in the study sample had received a face-to- face assessment prior to release, and the percentage of inmates who are assessed has increased over time. Another important consideration is the time between the date that an eligible MHSCP inmate appears on the Offender Information Services (OIS) List and the date that he or she is released. For the current study sample, 19.2% of the MHSCP- eligible inmates appeared on the OIS List within 45 days of release. In such circumstances, the TCMP social workers are able to conduct face-to-face assessments with 23.4%. When inmates appear on the OIS List with at least 45 days before their actual release date, they are approximately 2.4 times more likely to be assessed (58.7%). • Overall, EOP inmates were significantly more likely to receive a pre-release assessment than CCCMS inmates (59.3% versus 53.7%). • For parolees who had appeared on the OIS List at least 45 days prior to release, assessment rates increased from 55.2% from July 1, 2001 to December 31, 2001 to 65.9% for those released between January 1, 2004 to June 30, 2004. However, the percentage of eligible inmates who were assessed prior to release has declined since that time. Parole Outpatient Clinic (POC) Attendance • Inmates who were assessed prior to release were significantly more likely to attend a POC at least once than those who did not receive a pre-release assessment (63.8% versus 42.8%, respectively). • The likelihood of being admitted to a POC did not differ between EOP (55.4%) and CCCMS (54.9%) parolees. • The average number of POC sessions attended was 6.1 (SD=8.1), and ranged from 1 to 2 Final Report on the MHSCP Evaluation 254 sessions. Overall, TCMP-assessed parolees attended a significantly greater number of POC sessions (mean=6.3) than non-assessed parolees (mean=5.9 sessions). • Controlling for the effects of other background variables, receiving a pre-release assessment by a TCMP-MI social worker was associated with more than a two-fold increase in the odds of attending a POC at least once following release from prison. Characteristics of MHSCP and Non-MHSCP Parolees • As found in the first three annual reports, the likelihood of an inmate’s being assessed prior to release did not appear to be systematically related to his or her background characteristics. In other words, assessments are not conducted on inmates who are considered “best bets,” nor are they biased toward those who are most impaired. Reductions in EOP Caseloads • One of the policy shifts that occurred in conjunction with the MHSCP initiative was the reduction in EOP parolee caseloads to 40:1. Effective July 1, 2001, EOP parolees are to be contacted by their parole agent on the first workday following release and interviewed by the third workday. In addition, parole agents are to make a home call within six workdays following release, (and four per quarter for the remainder of parole), have two face-to-face contacts per month, two collateral interviews per month, one random drug test per month, and a case review at 30 days following release and every 90 days after • that. Based on our analysis of parole supervision records, we found that 61.4% of EOP parolees had an initial contact with their parole agent within five days of release; 17.8% made initial contact six or more days after release; and 20.9% of EOP parolees were never seen by their parole agent. In addition, we found that, on a monthly basis, parole agents who supervise EOP parolees conduct an average of nearly two face-to-face parole office visits, nearly three collateral contacts, and over one drug test. These frequencies indicate that the reduction in EOP caseloads is having its intended effect on parolee monitoring. Outcome Evaluation The analyses in this section focus on offenders who were released between July 1, 2001 and December 31, 2004. This cohort was selected to allow us to examine the effects of the MHSCP transitional process after one year of implementation, while still allowing a minimum of 12 months at risk in the community. • The likelihood of being returned to custody (for any reason) during the first 12 months following release from custody was associated with being younger, male, African- American, having been initially committed for a property crime, and having more serious mental health problems. Sixty-one percent of EOP parolees were returned to prison within 12 months, relative to 53.4% of CCCMS parolees. • After controlling for these five background variables (and parole region), receiving a pre- release assessment by a TCMP social worker was associated with a 13% reduction in the odds of being returned to custody within 12 months; having one or more POC contacts following release was associated with a 34% reduction in the odds of being returned to 3 Final Report on the MHSCP Evaluation prison within 12 months. Time in Program • Based on data on OIS-listed CCCMS/EOP releases from July 1, 2001-December 31, 2004 (N=48,917), 43.4% had no POC contact, 14.6% had one POC visit, 19.2% had 2-4 POC visits, 10.4% had 5-8 POC visits, and 12.4% had nine or more. • Consistent with previous research, our analysis revealed a strong relationship between the number of POC sessions attended and recidivism risk. Specifically, the greater number of POC contacts a CCCMS/EOP parolee has, the less likely he or she is to be returned to prison. For example, 36.2% of parolees with 9 or more POC contacts were returned to custody within 12 months, compared to 65.8% of parolees with no POC contacts. Cost Analysis • Based on the reduced number of incarceration days, we estimate that pre-release assessments conducted by TCMP-MI social workers resulted in savings of $875 for each EOP parolee and $344 for each CCCMS parolee, relative to non-assessed parolees. • Regarding the cost savings associated with POC attendance, we calculated that having one or more POC contacts following release was associated with savings of $4,890 per EOP parolee and $2,876 for each CCCMS parolee, relative to those with no POC contact. Clinician Interviews Because the majority of data for this evaluation came from official records, we sought to supplement these findings by soliciting perceptions of clinicians (i.e., psychiatrists, psychologists, and social workers) based at the POCs. Although the response rate was low, we opted to include these findings in the report to reveal possible areas for further improvement. Three major themes ran through the majority of the 23 MHSCP POC clinician interviews: 1) The need for improved functionality of the patient computer database, 2) The importance of transferring mental health and psychiatric medication records from the prison to the POC, and 3) The need to improve coordination and communication between clinicians within the institution and those at the POCs to ensure that clients do not fall through the cracks. Conclusions The current evaluation has focused on the impact of the MHSCP on those inmates who were OIS-listed, but these offenders only account for 53.4% of eligible releases overall. This percentage has increased, however, since we began tracking these data in July of 2001. As we have indicated in previous annual reports, the overall effectiveness of the MHSCP program depends on the ability of the California Department of Corrections and Rehabilitation to further increase the percentage of eligible inmates who appear on the OIS list prior to being released on parole. According to our analysis of EOP parolees’ supervision records, the decision to reduce the size of EOP caseloads to 40:1 (effective July 1, 2001) appears to have had a positive impact on the immediacy and frequency of parole agent contacts. Among eligible offenders who were identified on the OIS list prior to release, the TCMP process 4 Final Report on the MHSCP Evaluation continues to show promising results, both with regard to increasing the likelihood of attending a POC upon release as well as reducing the likelihood of being returned to custody. Specifically, after controlling for background characteristics, we found that receiving a pre-release assessment by a TCMP-MI social worker doubled the odds that a parolee would attend a POC at least once following release from prison. In turn, parolees who attended a POC following release from prison showed a 20% reduction in the odds of being returned to prison within 12 months, relative to parolees who did not attend a POC. Similar trends were found when predicting how long parolees remained out of prison. Because receiving a TCMP-MI assessment and attending a POC were both related to significantly fewer days in prison during the 12 months following release, we conducted a basic analysis comparing the cost savings associated with avoided re-incarceration days (within 12 months) by TCMP-MI pre-release assessment and POC participation. These analyses revealed that pre-release assessments conducted by TCMP-MI social workers produced annual savings of $875 for each EOP parolee and $344 for each CCCMS parolee, relative to non-assessed parolees. Regarding the cost savings associated with POC attendance, we calculated that having one or more POC contacts following release was associated with an annual savings of $4,890 per EOP parolee and $2,876 for each CCCMS parolee, relative to those with no POC contact. This result is consistent with findings in the broader offender rehabilitation literature showing that the greatest savings can be achieved by targeting the most seriously impaired offenders (Andrews, Bonta, & Hoge, 1990; Knight et al., 1999; Wexler, Melnick, & Cao, 2004). Lastly, semi-structured interviews conducted with a sample of POC clinicians indicated that over half (57%) of the respondents believed that the MHSCP has resulted in improved services for parolees. On the other hand, a large number of clinicians expressed the need to further improve the transfer of information from psychiatrists in the institutions to those based in the POCs. 5 Final Report on the MHSCP Evaluation FINAL REPORT ON THE EVALUATION OF THE MENTAL HEALTH SERVICES CONTINUUM PROGRAM OF THE CALIFORNIA DEPARTMENT OF CORRECTIONS—PAROLE DIVISION The purpose of this report is to summarize the results of UCLA’s process and outcome evaluation of the MHSCP during the first four years of implementation. Specifically, this report describes the historical context of the MHSCP project, effectiveness of the program with regard to ensuring continuity of mental health services for eligible parolees, and the association between MHSCP program participation and 12-month recidivism outcomes. I. MHSCP: Historical Context and Overview A. Background The research findings regarding the relationship between mental illness and crime vary considerably—often as a function of the definitions used and the populations studied. Not surprisingly, the associations between mental illness and crime tend to be more pronounced when studied among the general population. For example, according to one analysis of the Epidemiologic Catchment Area (ECA) data from the early 1980s (N>10,000), people with serious mental illness (i.e., Axis I diagnoses) were more than five times as likely to report engaging in violent behaviors as those without serious mental illness (see Monahan, 1996). In contrast, an attempt to predict general and violent recidivism among parolees from a maximum- security inpatient psychiatric unit showed that psychotic parolees were less likely than non- psychotic (but mentally ill) parolees to be rearrested for any offense, and equally likely to be rearrested for a violent offense (approximately 70% at three years post-release; Villeneuve & Quinsey, 1995). Similarly, Hodgins and Cote (1993) found that the criminal careers of mentally disordered and non-mentally disordered offenders differed little. However, the combination of antisocial personality disorder (ASP) and serious mental illness was associated with a significant increase in the frequency of non-violent arrests. Other researchers have suggested that the presumed association between mental illness and criminality is an artifact of the use of arrest records as a proxy for actual offenses. Mentally ill offenders may be more vulnerable to detection and arrest than non-mentally ill offenders. Therefore, they are more likely to be cycled through the criminal justice system for minor offenses (Teplin, 1984). A possible moderating variable in these studies is the effect of post- release mental health services. If, in fact, serious mental illness is associated with risk of recidivism, then the ongoing provision of needed psychiatric services to mentally ill parolees should result in improved functioning and fewer arrests. Indeed, several studies have supported this relationship. In one study of post-release mentally ill offenders, recidivism was directly related to the receipt of fewer services that the clients reported they needed (Solomon, Draine, & Meyerson, 1994). More recently, Berecochea and Liu (1999) found that, among mentally ill parolees in California, each additional parole outpatient clinic service was associated with an increase of 21 days on parole (i.e., reduced risk of recidivism). While the research findings regarding mental illness and criminality appear somewhat inconsistent, the association between substance use comorbidity and crime, particularly violent crime, is not. In a study of hospitalized psychiatric patients (N=101), alcohol and cocaine abusers were significantly more likely to have homicidal ideation and homicidal plans (Salloum 6 Final Report on the MHSCP Evaluation et al. 1996). Moreover, in a recent study of involuntarily admitted psychiatric patients with severe mental illness (N=331), Swartz et al. (1998) found that, whereas a diagnosis of schizophrenia or another psychotic disorder was not predictive of serious violence, the interaction of medication non-adherence and substance dependence was associated with a more than two-fold increase in the likelihood of committing violent acts, relative to those with either of these problems alone. The Mental Health Services Continuum Program In 1954, the California Department of Corrections established the Parole Outpatient Clinic (POC) program to assist parolees with mental health problems and, as a consequence, reduce recidivism rates among this population. Since its inception until October 1, 2000, parole agents were primarily responsible for referring parolees to the POCs for services. Referrals would be made if the parolee had a history of mental illness (usually indicated by the receipt of mental health services while in prison), or if the parole agent perceived that the parolee showed signs of mental instability. However, under this approach a substantial proportion of otherwise eligible parolees were either not identified or not provided appropriate services. To enhance the Department’s ability to identify and treat mentally ill parolees, the Mental Health Services Continuum Program (MHSCP) was developed by the Parole and Community Services Division (P&CSD) in July of 2000. According to its design, the MHSCP was to be applied to all eligible inmates released on or after October 1, 2000. However, based on a preliminary evaluation of the MHSCP, Bureau of State Audits (BSA; 2001) reported that (1) the program had failed to serve almost 40% of its target population, (2) clinicians were able to meet with their parolees during the scheduled time frame about 54% of the time, (3) the MHSCP database failed to identify almost 39% of the parolees who were eligible for MHSCP services, and (4) the MHSCP database did not allow for tracking of the time clinicians spent on each patient. The current report builds upon this earlier research and that of the first three years of the UCLA evaluation. It is the view of the Department that the P&CSD has since had an opportunity to address these and other implementation issues that mitigated its effectiveness in its early stages. B. Program Design and Description The MHSCP was designed to reduce the symptoms of mental illness among parolees by providing timely, cost-effective mental health services that optimizes their level of individual functioning in the community thereby reducing recidivism and improving public safety. The MHSCP is designed to include: • Pre-release needs assessment of paroling mentally ill inmates. • Pre-release benefits eligibility and application assistance. • Expanded and enhanced post-release mental health treatment for mentally ill parolees. • Improved continuity of care from the institution's Mental Health Service Delivery System to the community-based parolee outpatient clinics. • Increased assistance for successful reintegration into the community upon discharge from parole. 7 Final Report on the MHSCP Evaluation • A standardized program in all four-parole regions. Population Served The MHSCP target population consists of parolees who were receiving mental health treatment in the institutions under the Mental Health Services Delivery System prior to release to parole. The MHSCP target population also consists of those parolees who have been in a Mental Health Crisis Bed and those releasing from any Department of Mental Health facility. The criteria for admission to both the institution's and parole's mental health treatment programs is a diagnosis of one or more of the following Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) psychiatric disorders: • Schizophrenia (all subtypes) • Delusional Disorder • Schizophreniform Disorder • Schizoaffective Disorder • Substance-Induced Psychotic Disorder (exclude intoxication and withdrawal) • Psychotic Disorder Due To A General Medical Condition • Psychotic Disorder Not Otherwise Specified • Major Depressive Disorders • Bipolar Disorders I and II • Medical Necessity (any other major mental illness diagnosis which requires treatment due to the acuity or severity of the illness) The following mental health designations are used to determine the level of treatment need for inmates/parolees who require mental health services delivered by POC: 1. Correctional Clinical Case Management System (CCCMS) designation requires one or more of the above-referenced DSM IV diagnoses, and: • Stable functioning in the community; • Global Assessment of Functioning Score (GAF) above 50. 2. Enhanced Outpatient Program (EOP) designation requires one or more of the above referenced DSM IV diagnoses, and: • Acute onset or significant deterioration of a serious mental disorder characterized by increased delusional thinking; • Hallucinatory experiences, marked changes in affect and vegetative signs with definitive impairment of reality testing and/or judgment; • Dysfunctional or disruptive social interaction including withdrawal, bizarre or disruptive behavior, extreme defensiveness, inability to respond to instruction, provocative behavior toward others as a consequence of a serious mental disorder; 8 Final Report on the MHSCP Evaluation • Impairment of Activities of Daily Living (ADL) including eating, and personal hygiene, maintenance of dwelling, and ambulation as a consequence of a serious mental disorder. • Global Assessment of Functioning Score (GAF) of 50 or less. 3. Mental Health Crisis Bed (MHCB) designation: • An inmate experiencing a mental health crisis may be placed in a MHCB for inpatient, treatment and stabilization; • Length of stay may be up to 10 days, unless the inmate is approved for a longer stay by the chief psychiatrist, or designee; • Criteria for removal from the MHCB include stabilization and the ability to function in a less restrictive environment; i.e. EOP or CCCMS. 4. Priority of Services: • EOP inmates • MHCB inmates • Department of Mental Health facility releases on medication and receiving clinical treatment • CCCMS inmates receiving only clinical treatment • CCCMS inmates not on medication AND who did not receive clinical treatment within six (6) months prior to release will NOT receive services from the MHSCP According to the MHSCP design, regional Transitional Case Management Program—Mental Illness (TCMP-MI) social workers are to conduct face-to-face assessments with eligible inmates within 90 days of the inmates’ EPRD, and update this assessment information within 30 days of the inmates’ EPRD. The TCMP-MI social worker then merges the assessment information into the Parole Automated Tracking System (PATS) database. This information is verified by the TCMP-MI liaison and forwards this information to the appropriate POC headquarters. Once received, a POC-MHSCP liaison consults with the inmates’ parole agent of record (AOR) and schedules an initial appointment. For EOP parolees, this appointment is scheduled to occur within 3 working days of release; for CCCMS parolees, the initial appointment is scheduled to occur within 7 working days of release. In general, the jurisdictions of the TCMPI-MI social workers are divided into northern and southern regions, with Kern County Department of Public Health serving as the headquarters for the northern region, and the University of California, San Diego serving as the headquarters for the southern region. Some exceptions to this regional approach (e.g., including San Quentin State Prison in the southern region) were made to achieve balance between the regional caseloads and to reduce costs. Upon leaving the institution, parolees return to one of four parole regions (typically based on the county of commitment). The headquarters for these regions are located in Sacramento (Region I), Oakland (Region II), Los Angeles (Region III), and Diamond Bar (Region IV). 9 Final Report on the MHSCP Evaluation II. Impact Evaluation While the primary purpose of this evaluation is to examine the impact of MHSCP participation on recidivism, it is also important to assess patient-level data, including background characteristics, program participation, services received, and program discharge status. It is also important to examine the characteristics of the otherwise eligible parolees who were not served by the MHSCP program to determine whether there are any systematic biases in the referral and screening process by which inmates take part in the MHSCP; this cohort will also serve as a comparison group for the outcome evaluation. It should be noted that the impact evaluation is not directly concerned with determining the effectiveness of the program on recidivism; rather, the focus of this component is to describe the “pipeline” of patient flow and to characterize the continuity of services of MHSCP program participants relative to eligible parolees who do not participate in the program. It should be noted that, except where otherwise indicated, the analyses below are based upon releases from July 1, 2001 through June 30, 2005.1 Thus, the primary analysis sample consisted of 60,912 cases. When appropriate, this aggregate sample is divided into nine release cohorts: July 1, 2001-December 31, 2001 (Cohort 1; n=6,633), January 1, 2002-June 30, 2002 (Cohort 2; n=7,107), July 1, 2002-December 31, 2002 (Cohort 3; n=7,516), January 1, 2003-June 30, 2003 (Cohort 4; n=7,746), July 1, 2003-December 31, 2003 (Cohort 5; n=8,064), January 1, 2004-June 30, 2004 (Cohort 6; n=7,649), July 1, 2004-December 31, 2004 (Cohort 7; n=8,024), and January 1, 2005-June 30, 2005 (Cohort 8; n=8,173). A. Identification and Assessment of Eligible Inmates According to its design, the initial identification of MHSCP-eligible inmates is based upon monthly listings generated by the Offender Information Services Branch (OISB). The OIS List provides basic information on CCCMS and EOP inmates who are within 120 days of their estimated release date. However, because these estimated release dates are often inaccurate, not all eligible inmates appear on the OIS List. As a result, the sample frame for the present evaluation is limited to those inmates who appeared on the OIS List prior to release. It should be noted that not all inmates appear on the OIS list prior to release. Over the time period covered by the present evaluation, 55.2% of releases were identified on the OIS list before being discharged from prison. However, the percentage of inmates who appear on the OIS list has increased across the five release cohorts evaluated thus far, from 51.3% of inmates released in the July 1, 2001-December 31, 2001 cohort to 62.0% of inmates in the January 1, 2005-June 30, 2005 cohort—a 21% increase over that period. An analysis comparing offenders who appeared on the OIS list prior to release with those who did not revealed only modest differences between the two groups, with the exception that OIS-listed inmates were less likely to be male (82.7% versus 87.3%; Chi-square [1, N=110,414]=685.1, p<.0001); more likely to be committed for a property offense (36.0% versus 33.4%; Chi-square [1, N=110,414]=76.1, p<.0001); and more likely to be classified as an EOP (12.7% versus 6.5%; Chi-square [1, N=110,414]=1193.5, p<.0001). Because gender, commitment offense, and severity of mental health problems have all been demonstrated to predict recidivism, the findings of this report 1 Two groups of inmates were excluded from these analyses: (1) non-felons, and (2) inmates paroling from receptions centers. 10 Final Report on the MHSCP Evaluation (which are based only on OIS-listed offenders) should not be generalized to all released offenders with mental disorders. One of the primary goals of the MHSCP process is to increase the percentage of eligible inmates who are assessed prior to leaving the institution. To determine program effectiveness in this regard, we conducted two sets of analyses using the Parole Automated Tracking System database. In the first analysis, “assessed” was operationalized by whether the TCMP-MI social worker had reported having at least one face-to-face meeting with the OIS-listed inmate prior to release. In the second analysis, we examined the percentages of inmates assessed depending on whether they appeared on the OIS List at least 45 days prior to release. Overall, 54.5% of the eligible pool of releases in the study sample had received a face-to-face assessment prior to release, with the percentages showing a general decline over the course of 2005. Another important consideration is the time between the date that an eligible MHSCP inmate appears on the Offender Information Services (OIS) List and the date that he or she is released. For the current study sample, 19.2% of the MHSCP-eligible inmates appeared on the OIS List within 45 days of release. In such circumstances, the TCMP social workers are able to conduct face-to-face assessments with only 23.4%. When inmates appear on the OIS List with at least 45 days before their actual release date, they are approximately 2.4 times more likely to be assessed (58.7%). Figure 1 shows the trends in assessment percentages over the nine release cohorts of interest.2 Trend lines are provided for overall assessment percentages as well as the assessment percentages for inmates who appeared on the OIS List at least 45 days prior to their release date. Overall, EOP inmates were more likely to receive a pre-release assessment than CCCMS inmates (59.3% versus 53.7%; Chi-square [1, N=56,475]=82.2, p<.0001). Figure 1: Percent of MHSCP-Eligible Cases Assessed Prior to Release (N=56,475) 100 90 % Attending POC 80 70 65.3 68.4 67.4 65.9 64.2 60.7 60 55.2 60.3 60.5 60 54.6 54.2 53.5 50 46.7 47.5 44.2 41.1 40 34.7 30 31.1 20 19.7 21.1 13.6 12.8 10 10.4 0 1 2 3 4 2 03 04 05 /0 /0 /0 /0 /0 0/ 0/ 0/ 31 30 31 31 31 /3 /3 /3 2/ 2/ 2/ 2/ 6/ -6 -6 -6 -1 -1 -1 -1 2- 03 04 05 01 02 03 04 /0 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 7/ 7/ 7/ 7/ OIS Listed 45+ Days <45 Days 2 Unlike prior reports, this analysis was limited to offenders who were designated as being CCCMS or EOP at the time of release. 11 Final Report on the MHSCP Evaluation B. Clinic Attendance To assess clinic attendance, we analyzed POC attendance as a dichotomous outcome, categorizing parolees by whether they had at least one POC visit following release versus none at all.3 In addition, to allow for comparisons across all years prior to the initiation of the MHSCP program, POC attendance records were extracted from the Clipper database as well as the PATS database, since the latter was only recently adopted and, in many cases, parolee attendance records continued to be entered into the Clipper database even after the PATS database was in place. Thus, for parolees released from 1997 through 2002, POC clinic attendance was determined by the presence of a POC visit in either the Clipper or PATS databases. Figure 2 shows POC attendance rates for the cohorts specific to this evaluation (July 1, 2001-December 31, 2005). The average number of POC contacts (for those who attended at least once) is only shown for the first seven cohorts, since the follow-up periods were truncated for the latter cohorts. Overall, these data indicate that the percentage of CCCMS/EOP parolees who attend a POC at least once following release from prison decreased between the July-December 2001 cohort to the January-Jun 2005 cohort, from 54.5% to 45.2%. The likelihood of being admitted to a POC did not differ between EOP (55.4%) and CCCMS (54.9%) parolees (Chi-square [1, N=56,475]=0.76, p=.38). The average number of POC sessions was relatively stable over time at slightly under four sessions, though this, too, showed a decline in 2005. Among parolees who were admitted to a POC, the average number of sessions attended was 5.7 (SD=7.9). Figure 2: Percentage of CCCMS/EOP Parolees with One or More POC Contacts (N=56,475) 100 4.5 Avg. No. of Sessions 90 4 % Attending POC 3.9 80 3.7 3.6 3.7 3.5 3.5 70 56.7 55.6 58.7 58.7 58.93.2 3 54.5 52.4 60 2.5 2.4 45.2 50 40 2 30 1.5 20 1 10 0.5 0 0 1 2 3 4 2 03 04 05 /0 /0 /0 /0 /0 0/ 0/ 0/ 31 30 31 31 31 /3 /3 /3 2/ 2/ 2/ 2/ 6/ -6 -6 -6 -1 -1 -1 -1 2- 03 04 05 01 02 03 04 /0 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 1/ 7/ 7/ 7/ 7/ % Attending POC Sessions 3 It should be noted that these data are based on clinic attendance records as reported in the PATS database. Some POC psychiatrists have indicated having difficulty entering these cases on occasion. Therefore, our estimates of POC attendance may be lower than the actual rates. In addition, determining whether a parolee was admitted to a POC within the prescribed time period proved to be a difficult task, given that these appointments are based on the inmates’ earliest possible release dates (EPRD). EPRDs are not precise release dates, but rather serve as best estimates of an inmate’s anticipated release. As a result, these dates are often inaccurate, occurring well before or after the tentatively scheduled POC intake appointment. 12 Final Report on the MHSCP Evaluation As indicated in the first annual report, POC attendance (for the combined cohorts) reveals slight but statistically significant variation in POC attendance by parole region. Figure 3 shows a regional distribution of the percentage of releases who had at least one POC contact following release from prison. Figure 3: Percent of Releases with One or More POC Visits by Parole Region (N=59,885) 100 90 80 70 62.1 60 55.6 50.6 50.3 % 50 40 30 20 10 0 Region I Region II Region III Region IV Although the above variations in POC attendance may reflect actual regional differences in attendance rates, some of the variation may also be an artifact of clinic- and regional-based differences in implementing the new reporting system. Although we have attempted to control for this potential source of bias by combining data from the Clipper and PATS databases, it is possible that the transition from one database to another may have resulted in a temporary suppression of reporting. Figure 4: Pre-Release Assessment by POC Attendance Distributed by Region of Parole (N=60,912) 100 90 80 71.8 70 60.9 63.8 58 60 51.1 43 44.4 % 50 38.3 40 30 20 10 0 Region I Region II Region III Region IV Non-Assessed Assessed 13 Final Report on the MHSCP Evaluation The impact of pre-release assessments on POC attendance. The bivariate relationship between pre-release assessments and POC attendance (defined as one or more visits) reveals that at least once than those who do not receive a pre-release assessment (42.8; χ2 [1, inmates who are assessed prior to release are significantly more likely (63.8%) to attend a POC N=60,912]=2,692.6, p<.0001). Not controlling for other differences between the assessed and non-assessed groups, this indicates that receiving a pre-release assessment is associated with a 49% increase in the likelihood of attending a POC at least once after release.4 This effect appeared to be consistent across parole regions (see Figure 4). Predicting POC attendance in a multivariate model. Because the assessed and non-assessed groups are not randomly assigned, it is important to control for the potentially confounding influences of other variables in order to isolate the effects of the intervention. Table 1 shows the results of a multivariate logistic regression model conducted to test the relationship between the pre-release assessments and POC attendance (as a dichotomous outcome), after controlling for the effects of age, gender, race, commitment offense and severity of mental health problems (CCCMS versus EOP). Table 1: Odds Ratios Predicting Any POC Attendance Among Releases Between July 1, 2001- December 31, 2005 (N=60,912) __________________________________________________________________ Variables OR 95% CI* Age 1.01 1.0, 1.01 ** Male (ref: female) 0.74 0.7, 0.78 ** African-American (ref: other races) 0.96 0.93, 0.99 * Property Offender (ref: other offenders) 0.91 0.89, 0.95 * EOP (vs. CCCMS) 1.06 1.03, 1.13 ** Region I (ref: Reg. IV) 1.40 1.33, 1.44 ** Region II (ref: Reg. IV) 0.88 0.85, 0.93 ** Region III (ref: Reg. IV) 0.85 0.83, 0.91 ** TCMP Assessed 2.25 2.37, 2.52 ** * p < .05; ** p <.001 level. The overall model was statistically significant (-2 Log L=93,710.8, Chi-Square [df=9]=2,968.8, p<.0001). With the exception of race/ethnicity, all of the control variables were significantly predictive of POC attendance. Taken as a single profile, those who attended a POC at least once tended to be older, female, non-property offenders, classified as EOP (rather than CCCMS), and have been released to Parole Region I. However, after controlling for these background 4 It should be noted that the 36% figure refers to the percent increase, rather than the percentage point difference (17.7%) between assessed and non-assessed groups. 14 Final Report on the MHSCP Evaluation variables, we find that receiving a pre-release assessment by a TCMP-MI social worker was associated with more than a two-fold increase in the odds of attending a POC at least once following release from prison. C. Characteristics of MHSCP and Non-MHSCP Parolees One of the approaches we used to assess the effectiveness of the MHSCP with regard to recidivism involved a comparison between TCMP-assessed and non-TCMP-assessed inmates. Given that the MHSCP is still a relatively new program, we sought to capitalize on the fact that not all eligible inmates are assessed by a TCMP-MI social worker prior to release. Obviously, such an approach assumes that whether an eligible inmate was assessed is essentially a random process (i.e., there are no systematic differences between those who are and those who are not assessed). To verify this assumption, we compared the assessed and non-assessed groups (collapsing across the nine cohorts representing releases between July 1, 2001-December 31, 2005) with regard to selected background variables, parole region, and mental health status. Table 2: Comparability of Assessed and Non-Assessed Inmates, July 1, 2001-December 31, 2005 (N=60,912) __________________________________________________________________ Variable Non-Assessed Assessed Total N 28,590 32,322 60,912 Demographics Gender (% male)* 82.4 82.9 82.7 Age (mean, SD) * 38.1 (9.1) 38.3 (9.2) 38.2 (9.2) Race/ethnicity** African American 33.6 34.7 34.1 Hispanic 22.7 19.8 21.3 White 40.1 42.4 41.2 Other 3.7 3.1 3.4 Parole Region * Region I 27.1 24.9 26.0 Region II 23.8 21.7 22.7 Region III 23.2 23.4 23.3 Region IV 26.0 30.1 28.0 Offense Category* Violent 24.1 25.8 24.9 Property 36.5 35.9 36.2 Drug 29.6 28.2 28.9 Other 9.8 10.1 10.0 Mental Health Status* CCCMS 79.7 81.1 80.4 EOP 10.8 14.1 12.4 None given 9.6 4.8 7.2 *p<.0001 As shown in Table 2, the assessed and non-assessed groups showed statistically significant differences on all five background domains. However, it is important to keep in mind that 15 Final Report on the MHSCP Evaluation statistical significance is determined not only by the magnitude of differences, but by sample size as well. With large sample sizes such as this, even subtle group differences that are not clinically relevant can be statistically significant. In such cases, it is more appropriate to examine these group profiles for patterns of differences that appear to reflect systematic biases in selection. Using this approach, we can see that whether an inmate is assessed prior to release appears to be unrelated to his or her background characteristics. In other words, assessments are not conducted on inmates who are considered most likely to succeed, nor are they biased toward those who are most impaired. The profiles in Table 2 indicate that non-assessments tend to occur at random, or at least for reasons external to the inmate. Therefore, non-assessed inmates appear to be a justifiable comparison group for the outcome portion of the present evaluation. D. Trends in Data Quality Maintaining data quality in the PATS database is critical for both the implementation and evaluation of the MHSCP program. To monitor data quality, a set of items from the PATS database was identified by the UCLA analysts to be monitored over time. These items (as shown in Appendix B) consist of the following: 1. OIS listed but no TCMP/POC entry 2. OIS listed but already released 3. OIS listed without EPRD in PATS database 4. OIS listed less than 45 days from estimated release date 5. OIS listed at least 45 days prior to release, but not assessed by a TCMP-MI 6. Assessment date listed occurs after inmate was released on parole 7. POC appointment date occurs while offender is still in the institution, and 8. OIS listed with missing level of psychiatric severity (MHLevel variable indicating CCCMS or EOP status) As seen in Appendix B, the rates of inconsistencies (and/or logical errors) in the PATS database have dropped substantially since 2001. During calendar year 2005, none of the items listed above had an error rate exceeding 3%. E. Reductions in EOP Parolee Caseloads One of the policy shifts that occurred in conjunction with the MHSCP initiative was the reduction in EOP parolee caseloads to 40:1. Effective July 1, 2001, EOP parolees are to be contacted by their parole agent on the first workday following release and interviewed by the third workday. In addition, parole agents are to make a home call within six workdays following release, (and four per quarter for the remainder of parole), have two face-to-face contacts per month, two collateral interviews per month, one random drug test per month, and a case review at 30 days following release and every 90 days after that. Packets requesting data for 1,239 EOP parolees were mailed to 157 unit supervisors on May 31 (76% of all units with EOP caseloads) and June 1, 2005 (24% of all units). Each packet included a memo from Deputy Director L’Etoile notifying Unit Supervisors about the Evaluation of the MHSCP, a letter from the Principal Investigator requesting copies of four sets of records contained in each EOP file (Face sheet [CDC Form 1503-A], Contact Sheet [CDC Form 1244], Records of Supervision [CDC Form 1650D], and any Activity Reports [CDC Form 1502] 16 Final Report on the MHSCP Evaluation currently in the file), and a list of EOP parolees assigned to the unit. From June 1, 2005 to August 22, 2005, 133 parole units returned 1,014 out of 1,239 EOP files (an 82% return rate). Seventeen EOP cases were transferred to other parole units within the state. In addition, 22 EOPs were discharged from parole and 4 cases were transferred to U.S. Immigration and Naturalization Services units. Percentages of EOP files returned broken down by region are as follows: Region I (Sacramento)—92%, Region II (Oakland)—85%, Region III (Los Angeles)—71%, and Region IV (Diamond Bar)—87%. To account for parolees whose first three days following release overlapped with a weekend, we expanded the window of time for initial contact to five days. Using this time frame, we found that 61.4% of EOP parolees had an initial contact with their parole agent within five days of release; 17.8% made initial contact six or more days after release; and 20.9% of EOP parolees were never seen by their parole agent. Among those seen (N=807), the most common type of initial contact was at the parole office (66%), followed by collateral contact (52%), home visit (19%), jail visit (5%), or phone contact (5%). Table 3, shows the average number of office visits, collateral contacts, and drug tests conducted per month. Overall, these frequencies indicate that the reduced caseloads and closer supervision required for EOP parolees are being implemented with fidelity. On a monthly basis, parole agents who supervise EOP parolees conduct an average of nearly two face-to-face parole office visits, nearly three collateral contacts, and over one drug test. Table 3: Frequency of Parole Contacts/Actions Per Month for EOP Parolees (N=807) No. Actions Per Month Mean SD Range Parole Office Visits 1.9 1.7 0-15 Collateral Contacts 2.8 2.1 0-15 Drug Tests 1.2 0.8 0-5 III. Outcome Evaluation (12-Month Return to Custody) The three primary outcomes for this evaluation are (1) recidivism, (2) time to recidivism, and (3) correctional costs. It should be noted that recidivism is a general term referring to subsequent offending after release. It can be measured in many ways, including self-reported offenses, official arrest records, convictions, or returns to custody. Even among those who are returned to custody, there can be substantial variation in the reasons why (e.g., violation of parole conditions, pending revocation, or for a new offense). For the present evaluation, analyses of recidivism outcomes are based on simple return to custody, regardless of the reason, unless subsequent analyses reveal that alternative recidivism definitions produce substantively different results. In addition, recidivism is based on release cohorts, rather than at the individual level. Consequently, one parolee can account for multiple returns. Regarding correctional costs, this evaluation includes an estimate of the incremental cost savings associated with avoided incarceration costs with the receipt of the MHSCP transitional services. 17 Final Report on the MHSCP Evaluation This evaluation does not include a formal cost-benefit analysis, but rather provides an estimate of correctional costs saved based on the number of days in custody that are avoided as a consequence of participating in the MHSCP. Only correctional cost savings (rather than savings to other agencies or to society) are estimated. A. Comparisons by MHSCP Participation Status The analyses in this section focus on offenders who were released between July 1, 2001 and December 31, 2004. This cohort was selected to allow us to examine the effects of the MHSCP transitional process after one year of implementation, while still allowing a minimum of 12 months at risk in the community. In the first analysis, we conducted a multivariate logistic regression to predict the likelihood that a CCCMS/EOP parolee would be returned to custody within 12 months of release. To accomplish this, we included six control variables (age, gender, race/ethnicity, commitment offense, parole region, and severity of mental health problems) and the two primary variables of interest with regard to the MHSCP transitional process: (1) whether the CCCMS/EOP inmate was assessed by a TCMP social worker prior to release and (2) whether he or she had at least one POC contact following release. The results of this analysis are presented in Table 4. The overall model was statistically significant (-2 Log L=65,797, Chi-square [df=10]=1,690.5, p<.0001). Odds ratio estimates for the 10 predictors are shown in Table 4. Table 4: Odds Ratios Predicting 12-Month Return to Custody for July 1, 2001-December 31, 2004 Release Cohort (N=48,243) __________________________________________________________________ Variables OR 95% CI Age 0.99 0.98, 0.99 ** Male (ref: female) 1.28 1.22, 1.34 ** African American (ref: other races) 1.26 1.21, 1.31 ** Property Offender (ref: other offenders) 1.16 1.12, 1.21 ** EOP (vs. CCCMS) 1.35 1.28, 1.43 ** Parole Region I (vs. IV) 1.18 1.12, 1.25 ** Parole Region II (vs. IV) 1.07 1.02, 1.13 * Parole Region III (vs. IV) 0.51 0.48, 0.54 ** TCMP Assessed 0.99 0.96, 1.03 Attended POC 0.80 0.77, 0.83 ** * p<.01; ** p<.001 We can see that the likelihood of being returned to custody (for any reason) was associated with being younger, male, African-American, having been initially committed for a property offense, and having more serious mental health problems. In fact, according to this analysis, EOP parolees have 35% greater odds than CCCMS of being returned to custody within the first year 18 Final Report on the MHSCP Evaluation following release. (Overall, 60.9% of EOP parolees were returned to prison within 12 months, relative to 53.4% of CCCMS parolees; Chi-square [1, N=56,475]=150.9, p<.0001). In addition, using Parole Region IV as a reference point, we see that parolees returned to Region I are more likely to be returned to custody within 12 months, while those released to Region III are only about half as likely to be returned to custody during this time frame. However, for purposes of this outcome evaluation, our greatest interest lies in understanding how the MHSCP-related variables are associated with recidivism. After controlling for these five background variables, receiving a pre-release assessment by a TCMP social worker was not associated with a significant reduction in the odds of being returned to custody within 12 months, but having one or more POC contacts following release was associated with a 20% reduction in the odds of recidivating during this time period.(It should be noted that by computing this same model without including the POC attendance variable, the effect of TCMP assessment is statistically significant.) Because recidivism risk has been shown to vary by parole region, additional bivariate analyses were conducted to ascertain the extent to which the effects of pre-release assessments and POC attendance occurred across all four regions. As can be seen in Figures 5 and 6, the patterns are consistent across parole regions, and thus do not appear to be artifacts of regional policy differences. However, it should be noted that, while the difference between assessed and non- assessed groups is statistically significant overall and for Regions II, III, and IV, the difference was not statistically significant for Region I. Figure 5: Return to Custody (12 Months) by Assessment Status and Parole Region, July 1, 2001- December 31, 2004 (N=48,917) 70 64.2 62.8 61.7 61.7 59.9 61.3 60 47.1 50 42.9 40 % 30 20 10 0 Region I Region II Region III Region IV Non-Assessed Assessed 19 Final Report on the MHSCP Evaluation Figure 6: Return to Custody (12 Months) by POC Attendance and Parole Region, July 1, 2001- December 31, 2004 (N=48,917) 100 90 80 70 64.2 62.2 62.9 61.7 64.1 58.1 60 52.1 % 50 38.4 40 30 20 10 0 Region I Region II Region III Region IV Did Not Attend POC Did Attend POC Another variable of interest is the length of time a parolee remains in the community prior to being returned to custody (if ever). Our comparison of group means reveals a statistically significant difference in parole days between assessed and non-assessed parolees, as well as between those who did and did not attend a POC upon release. Specifically, we found that parolees who had received a pre-release assessment had an average of 8.4 additional days on parole, and parolees who had one or more POC contacts had an additional 74.6 days on parole (see Table 5). Table 5: Mean Number of Days on Parole by Assessment and POC Attendance* No Yes Difference MHSCP Variable Mean (SD) Mean (SD) Assessed by TCMP-MI 208.9 (131.0) 217.3 (130.4) 8.4 *** Attended POC 171.0 (131.0) 243.1 (122.1) 72.1 *** *Maximum number of days is truncated at 366; Means are based on actual days not in custody for all releases, not estimates derived from the proportional hazard model described below; ***p<.0001 Using a proportional hazard model, the next analysis sought to predict the length of time CCCMS/EOP parolees successfully remained out of prison after their index release. This analysis allowed us to test for the effects of the MHSCP transitional process on parolees’ survival, that is, time to first re-incarceration. Consistent with the logistic regression models summarized earlier in this report, this model included age, gender, race/ethnicity, region, and severity of mental health problems as control variables. In addition, the model included whether the inmate had received a pre-release assessment and whether he or she had attended a POC 20 Final Report on the MHSCP Evaluation upon release from prison.5 The results of the full Cox regression appear in Table 6. In addition, Figures 7 and 8 (Appendix A) depict the survival curves for OIS-listed CCCMS/EOP parolees by pre-release assessment status and POC attendance. Table 6: Results of Proportional Hazard Model Predicting Risk of Recidivism Over Time Among CCCMS/EOP Parolees Released July 1, 2001-December 31, 2004 (N=46,521) __________________________________________________________________ Variables Parameter Est. Hazard Ratio Age -0.01 0.99 *** Male (ref: female) 0.33 1.39 *** African American (ref: white) -0.06 0.94 *** Hispanic (ref: white) -0.17 0.84 *** Other Race (ref: white) -0.19 0.83 *** EOP (ref: CCCMS) 0.23 1.26 *** TCMP Assessed 0.01 1.01 Attended POC -0.59 0.55 *** *p<.05; ** p<.001; *** p<.0001 The parameter estimates in Table 6 represent the magnitude and direction of the effects of the independent variables on the outcome (i.e., days on parole). Parameters with a negative sign indicate a longer time on parole. As in the previous multivariate models, most of the background variables are statistically significant (with the exception of being African American). However, even after controlling for these potentially confounding effects we see that attending a POC remains strongly predictive of days on parole. Specifically, parolees who attended a POC had a proportional hazard of only .55 that of those who did not attend a POC. Interestingly—and in contrast to previous reports—receiving a pre-release assessment was not independently predictive of time spent on parole. B. Time in Program As noted earlier, Berecochea and Liu (1999) found a positive relationship between the number of parole outpatient clinic services and the length of time a CCCMS/EOP parolee remained out of prison. This time-in-program effect has been observed in other studies as well—particularly evaluations of substance abuse treatment. 5 The number of days on parole was fitted by a proportional hazard (Cox) model. The data were right-censored at 366 days. 21 Final Report on the MHSCP Evaluation Figure 9: Return to Custody (12 Months) by Number of POC Visits, July 1, 2001--December 31, 2004 (N=48,917) 100 90 80 65.2 62.4 70 58.8 60 51.2 % RTC 50 36.2 40 30 20 10 0 0 1 2 to 4 5 to 8 9+ POC Visits In the current study, we examined the relationship between the number of times a parolee was present for a scheduled POC visit and the likelihood of being returned to custody within 12 months of release. Based on data on OIS-listed CCCMS/EOP releases from July 1, 2001- December 31, 2004 (N=48,917), 43.4% had no POC contact, 14.6% had one POC visit, 19.2% had 2-4 POC visits, 10.4% had 5-8 POC visits, and 12.4% had nine or more. Figure 9 shows 12- month recidivism results across these five groups. Consistent with previous research, our analysis revealed a strong relationship between the number of POC sessions attended and recidivism risk. Specifically, the greater number of POC contacts a CCCMS/EOP parolee has, the less likely he or she is to be returned to prison. Interestingly, the greatest decline in recidivism risk occurs for parolees who have attended a POC more than four times. IV. Cost Analysis As we have seen above, inmates who have received MHSCP transitional services were less likely than inmates who did not receive these services to have been returned to prison during the 12 months following release. Overall, based on the July 1, 2001-December 31, 2004 release cohort, TCMP-assessed parolees spent an average of 8.4 more days on parole during the year following release than did non-TCMP-assessed parolees. Likewise, parolees who attended a POC one or more times following release spent 72.1 more days on parole than did parolees with no POC contact. However, because CCCMS and EOP differ substantially in their incarceration costs, we have calculated number of days on parole separately for these groups (see Table 7). 22 Final Report on the MHSCP Evaluation Table 7: Additional Days on Parole and Estimated Correctional Costs by Mental Health Status* TCMP-Assessed Any POC Contact Est. Daily Est. Daily Classification (Days) (Days) Incarceration Costs Parole Cost EOP 10.5 58.7 $93.3 $9.99 CCCMS 8.9 74.3 $48.7 $9.99 *Maximum number of days is truncated at 366 To estimate annual correctional cost savings associated with the MHSCP, we assumed daily incarceration costs of $93.30 per day for EOP inmates and $48.70 per day for CCCMS inmates (CDCR, 2003). From these figures, we subtracted the estimated daily cost of parole supervision ($9.99), which resulted in estimated daily cost savings of $83.31 for each additional day an EOP parolee remains under community supervision (relative to being incarcerated), and $38.71 for each additional day a CCCMS parolee remains under community supervision. Based on these assumptions, we calculate that pre-release assessments conducted by TCMP-MI social workers resulted in an annual savings of $874.65 for each EOP parolee and $344.43 for each CCCMS parolee, relative to non-assessed parolees. Regarding the cost savings associated with POC attendance, we calculated that having one or more POC contacts following release was associated with an annual savings of $4,889.70 per EOP parolee and $2,876.15 for each CCCMS parolee, relative to those with no POC contact. V. Clinician Interviews Because the majority of data for this evaluation came from official records, we sought to supplement these findings by soliciting perceptions of clinicians (i.e., psychiatrists, psychologists, and social workers) based at the POCs. To accomplish this, we drew a 50% sample from each of these three staff categories in each of the four parole regions. This resulted in a sample frame of 74 subjects. The UCLA staff research associate as able to contact 62 of these potential respondents. Of these, 23 agreed to participate in the interview. Although this is an admittedly low response rate, we have included these largely qualitative findings to provide an anonymous forum in which POC clinicians could voice their opinions regarding the MHSCP program and how it might be improved. Three major themes ran through the majority of the 23 MHSCP POC clinician interviews: 1) The need for improved functionality of the MHCAS computer database, 2) The importance of mental health and psychiatric medication records from the prison, and 3) The need to improve coordination and communication between clinicians within the institution and those at POC to ensure that clients do not fall through the cracks. The following summary provides more detail about these major themes and clinicians’ suggestions to improve the MHSCP program. Q1. How has your job changed as a result of the new MHSCP program? Clinicians’ responses to this question dealt mainly with use of and dependence on the MHCAS database, problems with client no-shows/communication with TCMP, and lack of mental health and medication records from the prison. Six clinicians felt that they could not say how their job had changed, as they had not been employed by POC before the implementation of the MHSCP 23 Final Report on the MHSCP Evaluation program. Seven respondents alluded to their use of and increased reliance on the MHCAS computer database. Although some clinicians felt that the MHCAS is an improvement, which is saving clinicians time looking for files and trying to decipher handwritten notes, other clinicians were more ambivalent. One said, “The wonderfully confused computer system has complicated my job because it’s so unwieldy…but it does put all the clients in one database. You don’t have to look for a file, if you can get to the information in the database.” Three clinicians said that the computer had interfered with their work because of frequent breakdowns or other problems. Two others stated that they spend more time doing data entry. Five clinicians stated that they have had problems with no-shows who were scheduled using auto-generated appointments. According to one clinician, “99.9% of those preset appointments are no-shows…We don’t know what the parolees are being told. There should be better communication between TCMP and the POC line staff so that we’re on the same page.” Five clinicians also referred to the lack of mental health records and information about medications from prison. One clinician who had worked with POC when the MHSCP program was first implemented said, “When they first started the program, maybe half had medications written under the medication tab. Now, none of them have that information.” Another clinician suggested that the institution mail a packet that includes the medication slip. Two clinicians said that because they lacked the medication records from the institution, they had to rely on the client for that information. Q2. Do you believe that the MHSCP program has resulted in improved services for the CDCR parolees? Over half (13) of the respondents said “yes” to this question. Four responded that they did not know. Three did not respond, and 3 indicated that they believed the MHSCP had not improved services for parolees. Q2a. Explain. Five clinicians said that the MHSCP program has enhanced their ability to deliver services to clients. Access to statewide POC records through the MHCAS database helped clinicians make informed evaluations and decisions about their clients’ treatment (4), particularly with medication management (2). The MHCAS database also fostered greater communication among clinicians, however only one clinician specifically stated that there was better communication among clinicians both inside and outside of the institution. Although two clinicians felt that the MHSCP program has made their work more organized in that they do not need to spend time looking for client files, two clinicians pointed out that some necessary client information is either inaccurate or missing. Nine clinicians felt that the MHSCP program was not being used to its capacity (Don’t Know = 3, No Response = 3, Yes = 3). Four noted that they do not receive much information about the clients from the TCMP social worker in the institution. Suggestions for improvement were taking care of appointment scheduling glitches (2), having a computer database back up system (1), and allowing the mental health staff in the prison to have access to the MHCAS database. (1) Two (DK = 1, NR = 1) said that they couldn’t answer the questions as they had nothing to compare the MHSCP program against. 24 Final Report on the MHSCP Evaluation Three (No = 3) explained that they didn’t feel that MHSCP had improved services because they either saw no change (1), did not feel the MHSCP program was being put into practice (1), or felt that the research generated from the MHSCP program could not be validated or used by clinics (1). Five (DK = 2, NR = 1, Y = 2) commented on how the computer database had considerable problems that interfered with their work. Q3. What is your typical response if a parolee does not keep his or her scheduled appointment? Most clinicians reported that they notify the parole agent (16). Two of these respondents also notify the parole agent supervisor. Agents also typically reschedule the appointment and send the appointment letter to the parolee and his/her agent of record (14). One clinician does not reschedule the second appointment until the client’s parole status and address are verified. The absence is recorded in the computer database (5), and interventions are carried out based on the number of absences (5). Often, clients’ cases are closed after three no-shows (2). Q3a. What are the most important barriers that keep patients from attending their scheduled appointments? The most commonly reported barrier that keeps patients from attending their scheduled appointments is lack of access to or money for transportation (15). Patient irresponsibility, resistance, or lack of motivation was another common reason reported by clinicians (12). Clients’ mental illness (7) or problems with medication compliance (2) were also cited as contributing factors. Use of illegal drugs poses another important barrier (6), as does lack of family support (4), being homeless or having unstable housing (3), and work responsibilities (3). Clinicians also attribute no-shows to the lack of coordination and resources within the parole department (3). Clients are often unaware of their auto-generated appointments (5). Out-of-date addresses (1) and clients being returned to custody or placed in another controlled environment, such as a hospital (1), are other barriers. One clinician reported that a number of parolees are scheduled for POC appointments inappropriately. They may have been prescribed a psychiatric medication in prison, putting them into the prison mental health system, but they do not have a primary mental health problem. Many of the parolees who are in the mental health system inappropriately have drug abuse problems and use their status for secondary gain, for example to collect SSI (1). Q4. For what percentage of new admissions to your POC are you able to access parolee data collected by a TCMP social worker as part of the pre-release interview? The average estimate was 59%. Q5. How does having this information impact the way you conduct the Initial Mental Health Examination? The majority of the clinicians reported that the data that they receive from the TCMP social worker is not very useful, questionable, or sometimes not in the database (14). - “Often, a lot of the data is wrong…Generally, the person who is in custody does not give valid data. I ignore it. It is totally irrelevant.” - “Because of cycling in and out violating, the TCMP social workers don’t have time to collect good information. I don’t trust the data not because of the social workers, but because of the data that they’re looking at. It’s not uncommon for a person to have a 25 Final Report on the MHSCP Evaluation diagnosis with a psychotic disorder. The parolee fools the social worker into not having to work, getting social benefits, etc. Ten to 15% of my caseload is committed to getting better, 25-30% benefit from regular contact—they may not get better, but they don’t get worse. For the rest of my caseload, it’s not an effective use of resources because they’re not mentally ill, or they’re so badly brain-damaged from the use of drugs—it’s more of a social problem.” - “Sometimes there’s no information. Other times it may give me information on their history of substance abuse, meds, and diagnostic information. It’s rare when there’s nothing of any use and rare when there’s really helpful information. Most of the time it’s not very helpful.” - “Most of my initial evaluations feel as if they’re cold interviews. It’s like a stranger walking in without any history.” Some of these clinicians said that it would be helpful to get records describing the clients’ psychiatric history in order to better help the client (3). Others felt it would be helpful to get a record of the clients’ risk of violence (2). Some clinicians reported that some information collected by the TCMP social worker does help with the overall evaluation or provides important background on the mental health of the client (10). Again, clinicians felt that it would be helpful to have more detailed diagnostic information and information relating to the clients’ psychiatric history and medications. Three clinicians reported that the information from the TCMP social worker helps them keep the client in check/make sure the client is not malingering (3). Two clinicians felt that the information from the TCMP social workers could potentially make the initial POC evaluations more efficient by their not needing to spend so much time obtaining clients’ mental health history (2). Q6. Is your clinic currently serving EOPs? All respondents reported that their POC served EOP patients. Q7. For what percentage of EOP parolees are you able to hold an Interdisciplinary Treatment Team meeting within the 30 days following completion of an Initial Mental Health Examination? On average, the respondents reported that they were able to do this 64% of the time. Q8. Who normally participates in these meetings? (2 missing) Most respondents reported that the parolee, parole agent, social worker, and either a psychologist or psychiatrist normally attend the IDTT (13). Three others said that the parolee, parole agent and a clinician or case manager were involved in the IDTT. Five others said that the parolee and either a psychologist, psychiatrist, social worker or case manager participated in the IDTT, without mention of the parole agent. Two clinicians could not respond because they had never participated in an IDTT. Two clinicians noted that the IDTT does not always occur with all of the participants at one time: - “The parole agent, psychiatrist, me, and the parolee, but not always at the same time or in the same room.” - “The parole agent, clinician, and psychiatrist. We’re seldom in the office at the same time. We don’t sit down in the same room all together.” 26 Final Report on the MHSCP Evaluation One clinician reported that formal team meetings do not take place: - “Team meetings don’t exist here. We don’t have the time. We consult between patients. We have no choice. IDTTs are a myth.” While another reported that IDTTs consist of more than just one meeting: - “These meetings are ongoing and happen all the time.” Q9. For what percentage of the patients on your caseload do you participate in a discharge planning process within 120 days of discharge from parole? The average estimate was 71%. Q10. Who normally participates in these meetings? (4 missing) Responses to this question were as follows: Parolee and social worker = 6 Parolee and psychologist = 3 Parolee and psychiatrist = 1 Parolee, social worker/psychologist, and psychiatrist = 3 Parolee, clinician(s) (consisting of one or more of the following: social worker, psychologist, psychiatrist), parole agent = 6 Q11. What information in the MHCAS do you find most helpful? Nearly all of the respondents reported that the progress notes/case notes, previous evaluations, or records of psychiatric history, particularly from the institution, were the most helpful information in the MHCAS (20). Two of these clinicians reported that they would like to see more of the data entered by the TCMP social workers under specific MHCAS tabs, instead of just under the case notes area of the database. Clinicians also found information about prescriptions and medication history very helpful (15). Diagnoses (5), response to treatment and treatment compliance (3), appointment schedule (3), current address (2), and the benefit file (2) were also reported to be helpful. Only one clinician in each of the following areas reported that these areas of the MHCAS were helpful: criminal history, demographic information, the community release file, and vocational plans or housing. Q12. What, if any, information in the MHCAS do you not find helpful? Five clinicians said that they did not find any information not helpful, or that most of the information in the MHCAS was helpful. Two others did not know what they did not find helpful. Eight clinicians commented that many of the MHCAS tabs are not used, incomplete, or unreliable, and thus not helpful. - “I find that the MHCAS is incomplete, unreliable, and not congruent with my own experience with these people. I can’t use it to validate my experience. My experience is 27 Final Report on the MHSCP Evaluation more reliable…Most is not helpful. I don’t even use it. I can’t count on its reliability and when it is there it’s not accurate. As far as the software itself, it works pretty well. It is a useful scheduling and note-keeping tool.” - “There’s not enough information in there. People aren’t putting it in. We’re not getting enough [information] from the prisons. I never see their medications in there. I don’t have a clue what meds they’re on.” - “The medication tab is pretty useless for what they were taking in prison. That’s rarely filled out. They could get rid of that and put it in the social worker discharge note or psychiatric discharge note, under the case note, instead of having to look up meds. It’s filled in less than half of the time.” - “There are a lot of tabs we are not able to get to because we are so busy. Ninety-nine percent of the time we live in three tabs: case notes, appointments, and address—because their address is always changing.” The following areas of the MHCAS were reported to be not helpful by clinicians: Demographic section (5) Community release planning (5) Goals (4) Release needs assessment (4) Benefit file (4) Revocation codes (4) Body marks (3) Physical health (3) Location tab (2) Mental health classification (2) Again, most felt that these areas are not helpful because they are either not used or inaccurate. Other problems with the MHCAS included problems with un/scheduling people from groups, incorrect auto-generated appointments, repetitive DSM-IV diagnosis fields, and inaccurately filled out checkboxes (e.g., the PC 290 box is not checked when the client is a PC 290). Q13. Is there any additional information you would like to see included in the MHCAS database? Ten clinicians reported that they would like the sections of the MHCAS to be filled in more completely and provide more detail about their clients’ mental health history. - “…the tabs are fine. The problem is that the data is either missing or inadequate…We don’t know the client’s medical issues. If things were entered in and entered accurately, it would be nice, but nothing needs to be added.” - “More treatment-detailed information. I don’t get the medications in the database. I seldom have an evaluation from the social worker, psychiatrist, or psychologist before they are released.” 28 Final Report on the MHSCP Evaluation - “I’d like to see information from their mental health treatment team in prison. I’d like to see their meds…” - “Consistency. Sometimes there’s information and sometimes there isn’t.” - “I would like to have a whole history of these people. What meds, what kinds of treatment they received in the prison and they can’t usually give you that information.” One clinician felt that the database had too many sections and that a data entry clerk was needed to fill in these areas of the MHCAS. - “It’s got a lot of information—too much…I don’t have time to fill it all out…Every time one [client] goes in or out there’s paperwork involved. I’m not a data entry clerk…we don’t have the data entry support here. That’s what you need—a receptionist and data entry clerk.” Eight clinicians reported that the capacity to link the MHCAS with information from other criminal justice databases and other records would make the MHCAS more useful. One of these clinicians felt that the database would be more helpful if it had the capacity to store scanned reports, including previous prison treatment reports and the client’s cumulative summary (Cum Sum), which is accessible to the parole agent. Another said, - “It would be nice to somehow have our records linked with the prison records and also with the parole agents database, for instance, a link into IPTS to get the current information of the client so that any changes (address, violations, revocations from parole) would automatically alert our system. The OBIS program will tell you where the client is at any given time. If we were alerted, we would be able to prevent parolees from falling through the cracks.” Clinicians also reported that they would like the MHCAS to be more user-friendly (4). For example, one clinician felt that too much time was spent moving through the alphabetically organized series of tabs and suggested listing the most frequently used tabs in the front to save time. Another commented that the progress notes and evaluations are hard to read. And another suggested updating the MHCAS to be able to reflect more than one clinician working with a patient, so that all clinicians who are working with a particular client receive credit for working with the client. Another suggestion was to add attendance categories including “absent-excused,” “absent-unexcused,” “cancelled by therapist,” and “cancelled by parolee.” Two clinicians said that they would like the MHCAS to provide more comprehensive background information on clients and one suggested that it provide psychological testing results. Four clinicians had nothing to add. Q14. Do you have a preference for data provided as case notes (i.e. narratives), or as pre- established response options (like checklists or check-off boxes)? Most of the respondents (13) indicated a preference for case notes; four preferred checklists; and 6 respondents said that they prefer a combination of both. Q14a. In what circumstances is one better than the other? Seven clinicians felt that narratives were more helpful than checklists in nearly all circumstances. Eleven said that both narratives and checklists were useful. Two overwhelmingly 29 Final Report on the MHSCP Evaluation preferred the checklists. Narratives were reported to be better for providing a more in-depth psychological history of the client. Checklists were seen as good for yes/no questions, served as reminders for the clinician on areas to cover in an evaluation, saved time, and were good for standardizing evaluations. One problem with the case notes in the MHCAS is that clinicians can only enter up to a certain number of bytes and the computer does not alert the clinician when s/he has exceeded the allowable space limit. Q15. In your opinion, how could the MHSCP program be further improved? Most clinicians felt that the MHSCP program could be improved by having an integrated system with better communication among clinicians inside and outside of the institution and parole agents (13). - “By having a system that links the three divisions of the CDCR—prisons, POC, and parole agents. Until we do that we’re not going to reach our potential. We need an integrated system.” - “[By] sharing of prison records and parole agent records.” - “I wish someone would look at the whole system. Clients are getting some care, but we’re mostly putting out fires.” - “More information from the TCMP social worker and just that real useful information such as how the inmate is doing in prison, what medication they have been taking or not taking…I thought that the TCMP promised before that they would help inmates have a clear plan for housing—that’s not happening. EOPs have no idea where they’re going. Many parolees are housed in drug-infested hotels downtown. We lose people, absconding in that poor housing environment. Many times I see notes from the TCMP social worker such as “Needs to be done.” I would like them to initiate more concrete help so the agent and we will know how to help the parolee. What I wonder, well, if psychiatrists or the social worker in prison can give some kind of input, I think that would be helpful. They’ve been seeing the person for one to two years, maybe. I’m afraid that the TCMP social worker just doesn’t know. It’s a little difficult for them to pass on useful information to us. I’m wondering if the clinician can meet with the TCMP social worker and the client so that they can get a better handle on the condition of the client.” - “It’s often difficult to contact the individual prisons and the prison pharmacies about what the parolee was given at the gate. Sometimes it’s difficult to get that information at the parole level.” - “There should be some type of regular dialogue with the TCMP.” - “Well, again, most importantly to have the knowledge of what was going on with them in prison.” Ten clinicians thought that the MHSCP program could be improved by improving the functionality of the MHCAS database. Suggestions included scanning information into the database (rather than photocopying records), improving the network connection speed (reducing down time), eliminating database glitches and shutdowns, and rethinking auto- generated appointments since the discharge date is often incorrect, or clients are not aware of their appointments. Other suggestions for improvement included: 1. Upgrading office technology, including fax machines dedicated to clinicians to fax pharmacies and newer computers (3) 30 Final Report on the MHSCP Evaluation 2. Focusing on training clinicians inside and outside of the institution (3) 3. Improving data reliability, in particular data related to psychiatric history, diagnoses, and medications (3) 4. Hiring clerical support (3) 5. Improving housing resources for clients (3) 6. Reducing caseloads (2) 7. Improving training of correctional officers in the area of mental health (1) 8. Requiring clients with drug abuse problems to go to drug abuse treatment (1) VI. Conclusions The current evaluation has focused on the impact of the MHSCP on those inmates who were OIS-listed, but these offenders only account for 56.8% of eligible releases overall. This percentage has increased, however, since we began tracking these data in July of 2001. As we have indicated in previous annual reports, the overall effectiveness of the MHSCP program depends on the ability of the California Department of Corrections and Rehabilitation to further increase the percentage of eligible inmates who appear on the OIS list prior to being released on parole. According to our analysis of EOP parolees’ supervision records, the decision to reduce the size of EOP caseloads to 40:1 (effective July 1, 2001) appears to have had a positive impact on the immediacy and frequency of parole agent contacts. Among eligible offenders who were identified on the OIS list prior to release, the TCMP process continues to show promising results, both with regard to increasing the likelihood of attending a POC upon release as well as reducing the likelihood of being returned to custody. Specifically, after controlling for background characteristics, we found that receiving a pre-release assessment by a TCMP-MI social worker doubled the odds that a parolee would attend a POC at least once following release from prison. In turn, parolees who attended a POC following release from prison showed a 34% reduction in the odds of being returned to prison within 12 months, relative to parolees who did not attend a POC. Similar trends were found when predicting how long parolees remained out of prison. Because receiving a TCMP-MI assessment and attending a POC were both related to significantly fewer days in prison during the 12 months following release, we conducted a basic analysis comparing the cost savings associated with avoided re-incarceration days (within 12 months) by TCMP-MI pre-release assessment and POC participation. These analyses revealed that pre-release assessments conducted by TCMP-MI social workers produced annual savings of $1,749.51 for each EOP parolee and $592.26 for each CCCMS parolee, relative to non-assessed parolees. Regarding the cost savings associated with POC attendance, we calculated that having one or more POC contacts following release was associated with an annual savings of $5,440.14 per EOP parolee and $3,147.12 for each CCCMS parolee, relative to those with no POC contact. This result is consistent with findings in the broader offender rehabilitation literature showing that the greatest savings can be achieved by targeting the most seriously impaired offenders (Andrews, Bonta, & Hoge, 1990; Knight et al., 1999; Wexler, Melnick, & Cao, 2004). Lastly, semi-structured interviews conducted with a sample of POC clinicians indicated that over half (57%) of the respondents believed that the MHSCP has resulted in improved services for 31 Final Report on the MHSCP Evaluation parolees. On the other hand, a large number of clinicians expressed the need to further improve the transfer of information from psychiatrists in the institutions to those based in the POCs. 32 Final Report on the MHSCP Evaluation References Andrews, D.A., Bonta, J., & Hoge, R.D. (1990). Classification for effective rehabilitation: Rediscovering psychology. Criminal Justice and Behavior, 17, 19-52. Berecochea, J., & Liu, R. (1999). Seriously Mentally Disordered Offenders and Recidivism. Sacramento, CA: California Department of Corrections. Bureau of State Audits (2001). Department of Corrections: Though Improving, the Department Still Does Not Identify and Serve All Parolees Needing Outpatient Clinic Program Services, but Increased Caseloads Might Strain Services. Sacramento, CA: California State Auditor. (2001-104). California Department of Corrections (2003). CDC Facts. Sacramento, CA: Author. Knight, K., Simpson, D.D., & Hiller, M.L. (1999). Three-year reincarceration outcomes for in- prison therapeutic community treatment in Texas. The Prison Journal, 79(3), 337-351. Monahan, J. (1996). Mental illness and violent crime. Research Preview. Washington, DC: Department of Justice. Salloum, I.M., Daley, D.C., Cornelius, J.R., & Kirisci, L. (1996). Patterns of suicidality and alcohol use in alcoholics with major depression. Alcoholism: Clinical & Experimental Research, 20(8), 1451-1455. Swartz, M.S., Swanson, J.W., Hiday, V.A., Borum, R., Wagner, R., & Burns, B.J. (1998). Violence and severe mental illness: The effects of substance abuse and nonadherence to medication. American Journal of Psychiatry, 155(2), 226-231. Teplin, L. (1984). Managing disorder: Police handling of the mentally ill. In L.A. Teplin (Ed.), Mental health and criminal justice (pp. 157-175). Beverly Hills, CA: Sage. Villenueve, D.B., & Quinsey, V.L. (1995). Predictors of general and violent recidivism among mentally disordered inmates. Criminal Justice and Behavior, 22(4), 397-410. Wexler, H.K., Melnick, G.D., & Cao, Y. (2004). Risk and prison substance abuse treatment outcomes: A replication and challenge. The Prison Journal, 84 (1), 106-120. 33 Final Report on the MHSCP Evaluation Appendix A: (Figures 7 & 8) Survival Curves Predicting 12-Month Recidivism as a Function of TCMP Assessment and POC Attendance 34 Final Report on the MHSCP Evaluation Figure 7 35 Final Report on the MHSCP Evaluation Figure 8: 36 Final Report on the MHSCP Evaluation Appendix B: Trends in PATS Data Quality 37 Final Report on the MHSCP Evaluation Inconsistency Report Inconsistency: 2001 % 2002 % 2003 % 2004 % 2005 % OIS listed but no TCMP/POC entry 2322 17.9 317 1.9 293 1.6 3 0.0 728 3.0 OIS listed but already released 126 0.5 29 0.1 358 1.1 426 0.9 830 0.7 OIS listed without EPRD in TCMP/POC 1808 7.8 1529 5.4 1271 4.0 1430 3.0 530 0.4 OIS listed with less than 45 days from estimated release 2128 9.2 2774 9.7 2160 6.8 2559 5.4 2597 2.0 OIS listed with more than 45 days and no assessment 4632 20.0 4164 14.6 4738 14.9 5729 12.1 2577 2.0 Assessment entry during a parole term 45 0.2 86 0.3 122 0.4 284 0.6 567 0.4 POC appt during an Institution term 146 0.6 837 2.9 920 2.9 1171 2.5 355 0.3 OIS listed with missing MHLevel in TCMP/POC 864 3.7 464 1.6 668 2.1 1339 2.8 1052 0.8 OIS Population 12976 16299 18738 20399 24064 TCMP/POC Population 23162 28478 31746 47348 126841 38

RELATED PAPERS

RELATED TOPICS

Log In

Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Final Report on the Mental Health Services Continuum Program of the California Department of Corrections and Rehabilitation—Parole Division

Related Papers

RELATED PAPERS

RELATED TOPICS