FINAL
REPORT
FEBRUARY

CRIMINAL JUSTICE
INTERVENTIONS FOR
OFFENDERS WITH MENTAL
ILLNESS: EVALUATION OF
MENTAL HEALTH COURTS
IN BRONX AND BROOKLYN,
NEW YORK

2012

FINAL REPORT
Shelli B. Rossman
Janeen Buck Willison
Kamala Mallik-Kane
KiDeuk Kim
Sara Debus-Sherrill
P. Mitchell Downey
This report was prepared under ASP BPA 2004BF022,
Task Requirement T-014, Task Order 2006 TO096 for
the National Institute of Justice

This report was prepared under National Institute of
Justice (NIJ) Task Order #2004-DD-BX-1123.
Ridge House Final Report

Acknowledgments 
The authors thank the many people whose support and contributions made this research
possible. Foremost, we wish to acknowledge the contributions and assistance of the
dedicated staff of the Bronx and Brooklyn Mental Health Court programs, the New York
City Department of Health and Mental Hygiene (DOHMH), and the New York State
Division of Criminal Justice Services (DCJS). While it is impossible to name the many
staff who assisted us with this study, we would like to highlight the special contributions
of the following individuals:
Bronx TASC Mental Health Court Program
Judge John P. Collins; Dr. Charles Amrhein; Cynthea Kimmelman-DeVries; Ryna Dery
Brooklyn Mental Health Court
Judge Matthew D’Emic; Lucille Jackson
Center for Court Innovation
Michael Rempel; Carol Fisler; Sarah Fritsche
New York City Department of Health and Mental Hygiene
Dr. Daniel Selling; Dominic Domingo; Patricia Brown; Louise Cohen; Suzette Gordon;
Dr. Jason Hershberger; Dr. Charles Luther; Olivette Burton; Jacqueline Rivera
New York State Division of Criminal Justice Services
David J. van Alstyne; Renee Konicki; Diane Cavin
We also extend our appreciation to Marilyn Moses, Social Science Analyst, at the
National Institute of Justice, for her support and guidance throughout the study, and to
Ruby Qazilbash, Senior Policy Advisor for Substance Abuse and Mental Health, the
Bureau of Justice Assistance, for sharing her insights and expertise.
Additionally, we thank Nancy LaVigne, Director of the UI Justice Policy Center, Julie
Samuels, Rayanne Hawkins, Lynn Johnson, Katie Johnson, and Aaron Horvath for their
support and assistance in completion of the final report.
Although we appreciate the contribution of those noted above, and any others
inadvertently omitted, the authors acknowledge their responsibility for any errors within
this report.
This report is dedicated to the late Laura Winterfield, Ph.D. the project’s original
Principal Investigator.

TABLE OF CONTENTS 
 
CHAPTER 1.   EVALUATION OF CRIMINAL JUSTICE INTERVENTIONS FOR OFFENDERS WITH MENTAL 
ILLNESS: STUDY CONTEXT AND LITERATURE REVIEW.............................................................................. 1 
THE MENTAL HEALTH TASK ORDER: GOALS AND OBJECTIVES .................................................................................. 1 
OVERVIEW OF THIS REPORT .............................................................................................................................. 2 
REVIEW OF THE LITERATURE .............................................................................................................................. 3 
What are Mental Health and Mental Illness? ....................................................................................... 3 
The Relationship Between Mental Illness and Crime ............................................................................ 4 
The Justice System’s Response to Mental Illness .................................................................................. 6 
The Emergence of Problem‐Solving Courts ........................................................................................... 9 
Mental Health Courts: An Overview .................................................................................................... 12 
Current Status of Mental Health Courts .............................................................................................. 13 
Mental Health Court Research ............................................................................................................ 15 
Process Evaluations ......................................................................................................................................... 16 
Outcome Evaluations ...................................................................................................................................... 17 

Limitations of Existing Research ......................................................................................................... 19 
CHAPTER 2.  EVALUATION DESIGN AND METHODS ............................................................................ 20 
THE PROPOSED STUDY ................................................................................................................................... 20 
FINAL EVALUATION DESIGN ............................................................................................................................ 22 
Process Evaluation .............................................................................................................................. 22 
Field Visits ....................................................................................................................................................... 23 
Stakeholder Interviews ................................................................................................................................... 23 
Courtroom Observation .................................................................................................................................. 24 

Impact Evaluation ............................................................................................................................... 26 
Defining the Treatment and Comparison Groups ........................................................................................... 26 
Data Acquisition and File Construction ........................................................................................................... 27 
Mental Health Court Program Data ................................................................................................................ 28 
Criminal Justice Records Data ......................................................................................................................... 30 
Mental Health Records Data ........................................................................................................................... 30 

Cost Analysis ....................................................................................................................................... 30 
DATA ANALYSIS: OVERVIEW ........................................................................................................................... 31 
CHAPTER 3. BRONX AND BROOKLYN MENTAL HEALTH COURT PROGRAMS AND BUSINESS‐AS‐USUAL . 32 
BRONX MENTAL HEALTH COURT ..................................................................................................................... 32 
Administrative Structure and Staffing ................................................................................................. 32 
The Role of the Judge ...................................................................................................................................... 33 
The Role of the Prosecution ............................................................................................................................ 33 
The Role of the Defense .................................................................................................................................. 34 
The Role of the Clinical Team .......................................................................................................................... 34 
The Role of Treatment Providers .................................................................................................................... 36 

Program Structure and Target Population .......................................................................................... 37 
Target Population ............................................................................................................................................ 37 
Demographic Profile ....................................................................................................................................... 40 

Eligibility Criteria ................................................................................................................................. 42 
Referral ............................................................................................................................................... 43 
Screening ............................................................................................................................................. 44 
Entrance into the Mental Health Court ............................................................................................... 45 
Court Proceedings ............................................................................................................................... 47 
Clinical Treatment and Case Management ......................................................................................... 50 
Program Exit ....................................................................................................................................... 53 

i

Stakeholder Views of the Bronx Mental Health Court ........................................................................ 56 
BROOKLYN MENTAL HEALTH COURT ................................................................................................................ 57 
Administrative Structure and Staffing ................................................................................................. 58 
Role of the Judge ............................................................................................................................................. 58 
The Role of the Prosecution ............................................................................................................................ 58 
The Role of the Defense .................................................................................................................................. 59 
The Role of the Clinical Team .......................................................................................................................... 59 
The Role of Treatment Providers .................................................................................................................... 60 

Program Structure and Operations ..................................................................................................... 61 
Target Population and Eligibility Criteria ............................................................................................ 63 
Demographic Profile ....................................................................................................................................... 64 

Referral ............................................................................................................................................... 67 
Eligibility Screening and Program Acceptance .................................................................................... 67 
Brooklyn Mental Health Court Proceedings ........................................................................................ 70 
Compliance ..................................................................................................................................................... 72 
Noncompliance ............................................................................................................................................... 73 

Clinical Treatment and Case Management ......................................................................................... 74 
Mental Health Court Participation and Outcomes .............................................................................. 74 
Program Exit ....................................................................................................................................... 75 
Comparison of Successful and Failed Participation in the Mental Health Court ................................. 78 
Stakeholder Views of the Brooklyn Mental Health Court .................................................................... 80 
“BUSINESS‐AS‐USUAL” MENTAL HEALTH SERVICES RECEIVED BY CRIMINAL DEFENDANTS IN BROOKLYN AND THE BRONX, 
NEW YORK .................................................................................................................................................. 80 
Program Operations ............................................................................................................................ 82 
Arrest and Admission to City Jails ................................................................................................................... 82 
Referral to Mental Health Services ................................................................................................................. 84 

Mental Health Services ....................................................................................................................... 84 
Assessment and Evaluation ............................................................................................................................. 84 
Mental Health Treatment in Jail ...................................................................................................................... 87 
Discharge Planning Services In Jail .................................................................................................................. 89 

RELEASE AND AFTERCARE ............................................................................................................................... 90 
ALTERNATE RELEASE AND DISCHARGE PLANNING SCENARIOS ................................................................................ 91 
BARRIERS TO CONTINUOUS CARE IN THE COMMUNITY ......................................................................................... 92 
COLLABORATION AND COORDINATION WITH MENTAL HEALTH COURTS................................................................... 93 
HOW DO MENTAL HEALTH COURT DEFENDANTS COMPARE TO THE BROADER UNIVERSE OF DEFENDANTS WITH MENTAL 
ILLNESS? ..................................................................................................................................................... 95 
Bronx Mental Health Court ................................................................................................................. 95 
Brooklyn Mental Health Court ............................................................................................................ 96 
Conclusions About Mental Health Court Participants and Other Jail Inmates With Mental Illness ................ 97 

CHAPTER 4. IMPACT ANALYSES AND FINDINGS .................................................................................... 98 
ANALYTIC OVERVIEW ..................................................................................................................................... 98 
MEASURES .................................................................................................................................................. 98 
ANALYTIC STRATEGIES ................................................................................................................................... 99 
RESULTS FROM THE BRONX MENTAL HEALTH COURT ........................................................................................ 101 
Development of Comparison Groups ................................................................................................ 101 
Does Mental Health Court Reduce Recidivism? ................................................................................ 105 
What Explains Recidivism in Bronx Mental Health Court? ................................................................ 107 
How Soon to Re‐Offend? ................................................................................................................... 110 
RESULTS FROM THE BROOKLYN MENTAL HEALTH COURT ................................................................................... 112 
Development of Comparison Groups ................................................................................................ 112 
Does Mental Health Court Reduce Recidivism? ................................................................................ 117 
What Explains Recidivism in Brooklyn Mental Health Court? ........................................................... 117 
How Soon to Re‐Offend? ................................................................................................................... 120 

ii

LIMITATIONS .............................................................................................................................................. 123 
SUMMARY ................................................................................................................................................. 124 
CHAPTER 5. COST‐BENEFIT ANALYSIS ................................................................................................. 126 
COST‐BENEFIT ANALYSIS AS PART OF THE MENTAL HEALTH COURT EVALUATION .................................................... 126 
Challenges in Conducting a Cost‐Benefit Analysis ............................................................................ 126 
Implications for Practice ................................................................................................................... 128 
SUGGESTED STRATEGY FOR CONDUCTING A MENTAL HEALTH COURT COST‐BENEFIT ANALYSIS ................................. 128 
Staff Costs ......................................................................................................................................... 129 
Data Collection .............................................................................................................................................. 129 
Analysis ......................................................................................................................................................... 129 

Treatment Costs ................................................................................................................................ 131 
Data Collection .............................................................................................................................................. 131 
Analysis ......................................................................................................................................................... 132 

Criminal Justice Costs ........................................................................................................................ 133 
Data Collection .............................................................................................................................................. 133 
Analysis ......................................................................................................................................................... 134 

CONCLUSIONS ............................................................................................................................................ 135 
CHAPTER 6.  WHAT HAVE WE LEARNED FROM THE EVALUATION OF CRIMINAL JUSTICE INTERVENTIONS 
FOR OFFENDERS WITH MENTAL ILLNESS? IMPLICATIONS FOR POLICY, PRACTICE, AND FUTURE 
RESEARCH ......................................................................................................................................... 136 
SUMMARY FINDINGS ................................................................................................................................... 136 
LIMITATIONS .............................................................................................................................................. 142 
Outcomes Limited to Recidivism ....................................................................................................... 142 
Analysis Limited to Program Impacts ................................................................................................ 143 
Potential Sample Bias ....................................................................................................................... 143 
Motivation and Other Potential Source of Bias ................................................................................. 144 
IMPLICATIONS FOR POLICYMAKERS, PRACTITIONERS, AND RESEARCHERS ............................................................... 145 
CONCLUSIONS ............................................................................................................................................ 147 
REFERENCES ...................................................................................................................................... 148 

 
TABLE OF FIGURES 
 
FIGURE 2.1. DATA COLLECTION PLAN: EVALUATION OF CRIMINAL JUSTICE INTERVENTIONS FOR OFFENDERS WITH MENTAL 
ILLNESS .................................................................................................................................................. 29 
FIGURE 3.1. CASE FLOW DIAGRAM OF BRONX MENTAL HEALTH COURT ...................................................................... 38 
FIGURE 3.2. BROOKLYN MENTAL HEALTH COURT CASE FLOW ................................................................................... 62 
FIGURE 3.3.  MENTAL HEALTH SERVICES PROVIDED DURING PRETRIAL DETENTION ....................................................... 83 
FIGURE 4.1. PROPENSITY SCORES, BY BRONX MENTAL HEALTH COURT AND COMPARISON GROUPS ............................... 105 
FIGURE 4.2. TIME TO RE‐ARREST BY TREATMENT CONDITION .................................................................................. 112 
FIGURE 4.3. PROPENSITY SCORES, BY BROOKLYN MENTAL HEALTH COURT AND COMPARISON GROUPS .......................... 116 
FIGURE 4.4. DIFFERENCE BETWEEN TREATMENT AND COMPARISON GROUPS OVER TIME .............................................. 122 

 
 

 

iii

TABLE OF TABLES
TABLE 3.1. CURRENT OFFENSES AND CRIMINAL HISTORY OF BRONX MHC PARTICIPANTS .............................................. 39
TABLE 3.2. MENTAL HEALTH INDICATORS AMONG A SUBSET OF BRONX MHC PARTICIPANTS WHO RECEIVED JAIL-BASED
MENTAL HEALTH SERVICES FROM DOHMH, 2005-2006 .............................................................................. 40
TABLE 3.3. SELF-REPORTED SUBSTANCE ABUSE BY BRONX MHC PARTICIPANTS........................................................... 41
TABLE 3.4. DEMOGRAPHIC, SOCIOECONOMIC, AND HEALTH CHARACTERISTICS OF BRONX MHC PARTICIPANTS ................. 42
TABLE 3.5. REFERRAL SOURCES FOR BRONX MHC PARTICIPANTS .............................................................................. 43
TABLE 3.6. TIME FROM ARREST TO BRONX MHC PARTICIPATION .............................................................................. 44
TABLE 3.7. PLEA CHARGES AND SENTENCES AT ENTRY TO THE BRONX MHC ................................................................ 46
TABLE 3.8. TYPES OF BRONX MHC HEARINGS OBSERVED ........................................................................................ 48
TABLE 3.9. COURTROOM PARTICIPATION DURING BRONX MHC HEARINGS ................................................................ 49
TABLE 3.10. INTERMEDIATE SANCTIONS ISSUED IN THE BRONX MHC ......................................................................... 51
TABLE 3.11. MENTAL HEALTH TREATMENT RECEIVED BY BRONX MHC PARTICIPANTS .................................................. 52
TABLE 3.12. CASE OUTCOMES AMONG BRONX MHC PARTICIPANTS ......................................................................... 54
TABLE 3.13. DURATION OF BRONX MHC PARTICIPATION ........................................................................................ 55
TABLE 3.14. SENTENCING OUTCOMES AMONG SUCCESSFUL BRONX MHC COMPLETERS ............................................... 55
TABLE 3.15. DEMOGRAPHIC CHARACTERISTICS OF BROOKLYN MHC PARTICIPANTS, 2002-2006 .................................... 65
TABLE 3.16. MENTAL HEALTH CHARACTERISTICS OF BROOKLYN MHC PARTICIPANTS.................................................... 66
TABLE 3.17. CASE PROCESSING STATISTICS AMONG BROOKLYN MHC PARTICIPANTS .................................................... 70
TABLE 3.18. TYPES OF BROOKLYN MHC HEARINGS OBSERVED ................................................................................. 71
TABLE 3.19. COURTROOM INTERACTIONS OBSERVED DURING BROOKLYN MHC HEARINGS ............................................ 72
TABLE 3.20. SPECIALIZED TREATMENT AND SERVICES PROVIDED THROUGH THE BROOKLYN MHC.................................... 76
TABLE 3.21. CASE OUTCOMES FOR BROOKLYN MHC PARTICIPANTS .......................................................................... 77
TABLE 3.22. SENTENCING OUTCOMES AMONG BROOKLYN MHC GRADUATES............................................................. 77
TABLE 3.23. CHARACTERISTICS OF BROOKLYN MHC GRADUATES AND FAILURES, 2002-2006 ....................................... 79
TABLE 3.24. CHARACTERISTICS OF BROOKLYN AND BRONX ARRESTEES WHO RECEIVED MENTAL HEALTH SERVICES FROM
DOHMH IN JAIL DURING 2005-2006........................................................................................................ 85
TABLE 3.25. TIME TO SELECTED MENTAL HEALTH SERVICE MILESTONES ..................................................................... 86
TABLE 3.26. TIME SERVED IN JAIL ........................................................................................................................ 87
TABLE 3.27. INDICATORS OF MENTAL HEALTH STATUS ............................................................................................ 88
TABLE 3.28. MENTAL HEALTH DIAGNOSES............................................................................................................ 88
TABLE 3.29. SUBSTANCE ABUSE DIAGNOSES ......................................................................................................... 89
TABLE 4.1. DESCRIPTIVE STATISTICS OF KEY MEASURES .......................................................................................... 102
TABLE 4.2. AVERAGE TREATMENT EFFECT OF MENTAL HEALTH COURT ON RECIDIVISM ................................................ 106
TABLE 4.3. OFFENSE TYPE FOR RE-ARREST BY TREATMENT AND COMPARISON GROUPS ............................................... 107
TABLE 4.4. LOGISTIC REGRESSIONS PREDICTING RE-ARREST AND RE-CONVICTION (N=1,128) ...................................... 108
TABLE 4.5. SURVIVAL ANALYSIS PREDICTING THE HAZARD OF RE-ARREST (N=1,128) .................................................. 111
TABLE 4.6 DESCRIPTIVE STATISTICS OF KEY MEASURES ........................................................................................... 114
TABLE 4.7. AVERAGE TREATMENT EFFECT OF MENTAL HEALTH COURT ON RECIDIVISM ................................................ 117
TABLE 4.8. OFFENSE TYPE FOR RE-ARREST BY TREATMENT AND COMPARISON GROUPS ............................................... 118
TABLE 4.9. LOGISTIC REGRESSIONS PREDICTING RE-ARREST AND RE-CONVICTION (N=606) ......................................... 119
TABLE 4.10. SURVIVAL ANALYSIS PREDICTING THE HAZARD OF RE-ARREST (N=606) ................................................... 121
TABLE 5.1. SAMPLE OF HOW STAFF COSTS WOULD BE CALCULATED ........................................................................ 130
TABLE 5.2. SAMPLE OF TREATMENT RECORDS FOR CLIENT A ................................................................................... 132
TABLE 5.3. SAMPLE TREATMENT COSTS FOR CLIENT A ........................................................................................... 133
TABLE 5.4. COMMONLY USED COSTS OF ARREST (DOLLARS IN 2005) ....................................................................... 134

iv

Appendices 
 
Appendix A – Study Design Modifications, 2005‐2008 (New York only) 
Appendix B – Bronx and Brooklyn MHC Court Characteristics 
Appendix C – Courtroom Observation Instruments 
Appendix D – Administrative Data Variable Inventory 
Appendix E – MHC and DOHMH Screening Instruments 
Appendix F – Bivariate Analysis of MHC Participants and Nonparticipants 
Appendix G –List of Covariates 
Appendix H –Propensity Score Matching and Sensitivity Analysis 

 

 

v

CHAPTER 1.  EVALUATION OF CRIMINAL JUSTICE INTERVENTIONS FOR 
OFFENDERS WITH MENTAL ILLNESS: STUDY CONTEXT AND LITERATURE 
REVIEW  
For the past five decades, U.S. law enforcement, courts, and correctional agencies have
experienced steady growth in the number of offenders with mental illness who fall under
their purviews. Various reasons have been suggested for this trend, such as: 1) the
deinstitutionalization of persons with mental illness that occurred during the 1960s and
early 1970s (i.e., the systematic shift in resources available for treating affected
populations in community-based settings, rather than in residential, state-run psychiatric
hospitals), 2) changes in civil commitment statutes that make it more difficult to
involuntarily place affected individuals in psychiatric hospital settings, 3) the evolution of
psychotropic medications that ideally make it possible to treat even the most severe
disorders within outpatient programs, and 4) law enforcement strategies that have focused
police resources on drug and low-level quality-of-life crimes (Denckla and Berman 2001;
Fisher, Silver, and Wolff 2006; Pogrebin and Poole 1987; Teplin 1984).
The Mental Health Task Order: Goals and Objectives  
By the early 2000s, if not before, it became increasingly clear that the criminal justice
system had become the primary public response to inappropriate behaviors by the
mentally ill, and that persons with mental illness were over-represented within criminal
justice populations. In response, federal agencies offered support for programming and
services targeting offenders with mental illness. One such initiative was undertaken by
the National Institute of Justice (NIJ), which commissioned an evaluation of two distinct
approaches to handling offenders with mental illness in the criminal justice system: 1)
the Brooklyn Mental Health Court (BMHC), a specialized problem-solving court
operating in the Kings County (New York) Supreme Court, and 2) the Pinellas County
Mentally Ill Diversion Program (MIDP), operating in the 6th Judicial Circuit’s Public
Defender’s Office in Clearwater, Florida.
In October 2005, researchers in the Justice Policy Center at the Urban Institute (UI)
began a three-year evaluation of those two programs. Although both interventions
primarily targeted offenders with mental illness booked into local jails and awaiting
disposition of their charges, the two interventions differed significantly both in their
approach toward such offenders and their operational structures. As such, the initial
project was conceived as two separate evaluations, unified by common research questions
about the process, potential offender-level impact, and costs of these two models of
criminal justice intervention for offenders with mental illness.1

1

UI’s proposed research designs for both the New York and Florida projects consisted of process, impact,
and cost-effectiveness components, using a quasi-experimental approach. The New York impact evaluation
was expected to rely solely on administrative data to construct comparable treatment and comparison

1

Evaluation activities in each study site began in early 2006 following a two-month design
phase and approval of the final research approach by NIJ. Several issues hampered
sample recruitment and data collection, leading to a number of modifications to both the
project and research design. As noted in Chapter 2 (Evaluation Design and
Methodology), the most pressing issues occurred in the Florida site, where a prospective
sampling approach was necessary in order to obtain informed consent and a signed
Release of Information for treatment and comparison group cases. Instability with the
program model, insufficient case flow, and funding uncertainties impeded treatment
group recruitment, while administrative barriers hampered progress in the comparison
site. As a result of the difficulties encountered in the field, the Florida site was eliminated
from the study in January 2007. Subsequently, NIJ and UI researchers reached an
agreement, in August 2008, to replace the Florida program with the Bronx (NY) Mental
Health Court program. Prolonged contract negotiations gave rise to further modifications
in the scope of the evaluation, including elimination of both 1) client focus groups and 2)
courtroom observations that would have enabled comparison of standard (i.e., “businessas-usual”) court processing of mentally ill offenders to that of the Bronx and Brooklyn
mental health (treatment) courts.
Thus, the current project examines the impact of two New York City mental health courts
on participant outcomes. The goal of the current study is to determine if participation in
mental health court reduces subsequent criminal justice involvement⎯namely,
recidivism as measured by new arrests and new convictions. It is important to note that
the study treats the two evaluations as separate and distinct (i.e., an impact evaluation of
the Brooklyn Mental Health Court and an impact evaluation of the Bronx Mental Health
Court, not as a single cross-site study); site data are not pooled. Separate analyses have
been conducted using distinct site-specific treatment and comparison groups. The same
administrative data sources (i.e., New York State Division of Criminal Justice Services,
New York City Department of Health and Mental Hygiene, program databases
maintained by each court) and analytic techniques (propensity score matching) were used
to construct equivalent comparison groups. Additionally, data needed to perform the
intended cost analyses were not accessible within the resource constraints of the project
as it evolved; as a result, the study provides guidance for conducting future cost analyses
of mental health court programs.
Overview of this Report 
This report presents findings from the process and impact evaluations of the Brooklyn
and Bronx Mental Health Court (MHC) programs. In the remainder of this chapter, we
present a brief review of relevant literature. First, in this chapter, we briefly review the
considerable literature on mental health and mental illness, the relationship between
mental illness and crime, justice system responses to offenders with mental health
problems or mental illness, and problem-solving courts and their conceptual foundation;
then, we examine the emerging literature specific to mental health courts, including the
research on mental health court effectiveness.
groups, while the Florida evaluation anticipated a prospective sample of treatment cases drawn from MIDP
and comparison cases sampled from a neighboring jurisdiction.

2

Chapter 2 discusses the design, data sources, and analytic strategy employed by the
current study. The current design is contrasted against the original research approach to
highlight key modifications, and a brief discussion of the factors necessitating these
modifications is also provided.
Chapter 3 explores the Brooklyn and Bronx MHCs, as well as the Business-As-Usual
(BAU) alternative for processing offenders with mental illness in the New York City
criminal justice system. All three descriptions are supplemented by findings from the
analysis of program data.
Chapter 4 describes how the quasi-experimental samples were constructed using
propensity score matching (PSM). It also details the impact evaluation findings with
respect to recidivism, looking separately at the effects of each mental health court.
Chapter 5 identifies the type of data needed to conduct a defensible cost-effectiveness
evaluation. In the absence of these data, we provide guidance for practitioners (and other
researchers) regarding our recommendations for establishing information systems that
will support future analyses of this type.
Chapter 6 summarizes the study’s objectives and approach, key findings, limitations, and
recommendations for future research.
Review of the Literature  
What are Mental Health and Mental Illness? 
The Department of Health and Human Services (DHHS) regards mental health and
mental illness as points on a continuum (DHHS 1999):
•

Mental health constitutes a state of successful performance of mental function,
resulting in productive activities, fulfilling relationships with other people, and the
ability to adapt to change and to cope with adversity. While some of the key
ingredients of mental health may be readily identifiable, mental health is
nonetheless difficult to define, in part because the concept of “health,” itself, is
tied to cultural and subcultural values. Thus, the meaning of being mentally
healthy is subject to diverse interpretations that are rooted in value judgments that
vary across time and space.

•

Mental illness refers collectively to all diagnosable mental disorders, which are
health conditions characterized by alterations in thinking, mood, or behavior
(either independently or in combination) associated with distress or impaired
functioning. For example, antisocial personality disorder (ASPD) is an Axis II
disorder characterized by a pervasive pattern of disregard for, and violation of, the
rights of others that begins in childhood or early adolescence and continues into
adulthood (APA 2000: 645–650). Similarly, depression constitutes a mental
disorder largely marked by alterations in mood, while attention-

3

deficit/hyperactivity disorders are mental disorders largely identified with
alterations in behavior (overactivity) or thinking (inability to concentrate).
Alterations in thinking, mood, or behavior contribute to a host of problems—
patient distress, impaired functioning, or heightened risk of death, pain, disability,
or loss of freedom.
Determinations of eligibility to receive publicly-funded mental health treatment services
also may entail classifications of individuals as having serious mental illness (SMI) or
severe mental illness or severe and persistent mental illness (SPMI). SMI generally refers
to mental disorders that interfere with some area of social functioning, such as school or
work. By contrast, SPMI encompasses more seriously affected persons, typically
including such diagnoses as schizophrenia, bipolar disorders, and severe forms of
depression, panic disorder, or obsessive-compulsive disorder, particularly where
individuals exhibit high levels of functional impairment.
In addition to mental health and mental illness, it is not uncommon to see the term
“mental health problems” used in the literature (and, at times, within this report). The
phrase is commonly used to denote circumstances in which the individual’s signs and
symptoms are of insufficient intensity or duration to meet the criteria for any mental
disorder. Virtually everyone has experienced mental health problems on some occasions
when under duress (e.g., during a bereavement period). Though potentially not as serious
as mental illness, mental health problems, nonetheless, may warrant active health
promotion, prevention, and treatment efforts.
Lastly, in this report, and in the broader literature on offenders with mental health
impairments, the term “co-occurring disorders” is often referenced (see, for example,
Hills 2000, Prins and Draper 2009). Here, it is defined as circumstances in which
individuals simultaneously experience mental illness and substance abuse disorders.
The Relationship Between Mental Illness and Crime 
The deinstitutionalization of persons with mental illness that occurred during the 1960s
and early 1970s rested on two assumptions: that needed services for the mentally ill
would be available, and that those services would be accessible. Unfortunately, neither
assumption proved true (Council of State Governments 2002). One of the unintended
consequences of the shift in public policy is that it is reportedly far more difficult for
many people with mental illness to access the mental health treatment system. States
closed or shrank their psychiatric hospitals without adequately funding community-based
treatment. As a result, many people with mental illness live in the community, where they
are unable to access adequate support services or medication.
The problems associated with insufficient or inaccessible treatment resources may be
further exacerbated by the stigmatization of persons with mental illness and by their own
improper behavior. “Stigma leads others to avoid living, socializing or working with,
renting to, or employing people with mental disorders, especially severe disorders such as
schizophrenia. It reduces…access to resources and opportunities (e.g., housing, jobs) and

4

leads to low self-esteem, isolation, and hopelessness…. Nearly two-thirds of all people
with diagnosable mental disorders do not seek treatment. Stigma surrounding the receipt
of mental health treatment is among the many barriers that discourage people from
seeking treatment” (DHHS 1999: 6). Then, as Prins and Draper (2009) point out, if they
publicly behave in ways that are symptomatic of untreated mental disorders (e.g.,
evidencing public intoxication, creating public disturbances, or engaging in “nuisance”
offenses), a bad situation may escalate as other citizens or the police intervene.
Additionally, individuals with mental illness are at increased risk of 1) developing
substance abuse disorders during their lifetimes and 2) experiencing homelessness⎯each
of which increases their visibility to law enforcement, along with increasing the
likelihood that they will become mired in the criminal justice system. Hence, as
previously noted, the criminal justice system is inundated with individuals who exhibit
diverse mental health anomalies. For example, estimates reveal that individuals with
serious mental illness are more than three times more likely to be housed in jails and
prisons than in hospitals (Torrey, Kennard et al. 2010). Los Angeles County Jail, Cook
County Jail, and Rikers Island all house more people with mental illness than any U.S.
psychiatric facility (CSG Undated). Further, studies in local jurisdictions have found that
jail inmates with severe mental illness are likely to spend significantly more time in jail
than other inmates who have the same charges, but who do not evidence severe mental
illness.
Although estimated prevalence rates for mental illness vary from study to study, research
has repeatedly demonstrated that correctional populations have higher rates of mental
illness than the general population (James and Glaze 2006; Lamb and Weinberger 1998;
Steadman, Osher et al. 2009; Teplin 1994). For example, James and Glaze (2006)
obtained nationally representative estimates by compiling data from surveys of the
nation’s inmates in jails and prisons; their estimates indicate that a disproportionate
number of incarcerated individuals (64 percent of jail inmates, 56 percent of state
prisoners, and 45 percent of federal prisoners) experience mental health problems,2 as
compared to 11 percent of the general population. Rates of severe mental illness
reportedly are also higher for incarcerated populations. For instance, 24 percent of jail
inmates and 15 percent of state prisoners have psychotic disorders, as compared with 3
percent of the general population. Research has found female inmates to have even higher
rates of mental illness (Abram, Teplin, and McClelland 2003; James and Glaze 2006;
Steadman et al. 2009). Furthermore, mental health problems are often compounded by
co-morbid substance problems. For example, approximately 74 percent of state prisoners
and 76 percent of jail populations with mental health problems also reported substance
dependence or abuse (James and Glaze 2006).
Many researchers and advocates assert that individuals with mental illness are trapped in
a “revolving door” of the criminal justice system, cycling in and out of correctional
facilities due to their mental illness and lack of treatment. Conversely, others have
claimed that mental health has little relation to criminal behavior and vice versa, citing
2

“Mental health problems” were defined as the occurrence of a mental health diagnosis, treatment, or
symptoms fitting DSM-IV criteria for a mental health disorder within the past 12 months.

5

the fact that the majority of individuals with mental illness do not commit crimes.
Regardless, incarcerated individuals with mental health problems have more extensive
criminal histories (James and Glaze 2006) and higher levels of criminal activity postrelease (Baillargeon Binswanger et al. 2009; CSG Undated; Mallik-Kane and Visher
2008). Findings have been mixed on whether individuals with mental illness (or certain
diagnoses) have higher rates of violence, although it appears that substance abuse may
explain higher rates of violence when they are found (Elbogen and Johnson 2009; Fazel
Gulati et al. 2009; Steadman, Mulvey et al. 1998).
Skeem and colleagues (2009) assert: 1) there is mixed evidence of mental health
diversion successfully reducing criminal behavior even when evidence-based mental
health treatment is used; 2) studies have found no significant relation between symptom
reduction and reduced recidivism when symptoms did improve; 3) systems-level data
show no link between the likelihood of incarceration for individuals with mental illness
and the availability of mental health services, psychiatric inpatient beds, or funding of
public mental health services; and 4) the rise in persons with mental illness who were
incarcerated (4 percent) from 1950 to 2000 did not match the reduction of individuals
with mental illness living in psychiatric institutions (23 percent) during this same period.
As a result, they suggest a moderated mediation model that indicates a small number of
individuals encounter legal trouble directly due to their mental health symptoms, while
the majority of offenders with mental illness come in contact with the legal system for the
same reason as other offenders who are not mentally ill: criminogenic needs.3 In essence,
Skeem and colleagues contend that individuals with mental illness are at higher risk for
these criminogenic needs, which would explain the disproportionately high rates of
mental illness among the incarcerated population. In support of this model, two studies
(Girard and Wormith 2004, Skeem et al. 2009) have found that offenders with mental
illness score higher on the Levels of Service Inventory/Case Management Instrument
(LS/CMI), an assessment tool used to assess risk and criminogenic needs, compared to
offenders without mental illness.
The Justice System’s Response to Mental Illness 
Until the mid-1990s, most suspects with mental illness could expect to be processed by
the criminal justice system in the same manner as suspects who were not experiencing
mental health issues. However, justice system actors increasingly have sought solutions
for balancing traditional objectives (e.g., public safety, punishment, incapacitation) with
innovative responses designed to meet the special needs of this population. For example,
numerous police agencies now operate crisis intervention teams (CIT)4 designed to
productively interact with suspects exhibiting mental illness. Other criminal justice
responses to mental illness include the expansion of jail screening procedures, mental
3

Criminogenic needs are dynamic risk factors predictive of criminal activity that can be targeted in
rehabilitative treatment. For example, the “Central Eight” criminogenic needs include: substance abuse,
history of antisocial behavior, personality, cognition, peers, and circumstances regarding family/marriage,
school/work, and recreation (Andrews and Bonta 2010).
4
CIT are specialized law enforcement programs developed to provide more effective, compassionate, and
safer approaches that police officers can use in interacting with people who suffer from mental illnesses or
developmental disabilities.

6

health
h pods in loccal jails, and
d mental heallth housing uunits at correectional faciilities. Also,
withiin the past deecade, two federal
fe
fundin
ng initiative s—the Subsstance Abusee and Mental
Healtth Administrration’s (SAM
MHSA’s) Targeted Cappacity Expannsion jail divversion
progrram, and the Mental Heaalth Courts grant
g
program
m administerred by the B
Bureau of
Justicce Assistance (BJA) and
d SAMHSA—
—have proviided resourcces to variouus locales forr
innov
vative approaaches to han
ndling offend
ders with meental illness.
Along these liness, Munetz an
nd Griffin
(2006
6: 544 ) concceptualized The
T
Sequential Intercept Model as
a a
frameework that id
dentifies a seeries of
pointts of intercep
ption at whicch
interv
ventions can
n be made to prevent
indiv
viduals with mental
m
healtth problems
eitherr from enteriing the criminal justice
system
m or penetraating more deeply
d
over
time. The interception points are: law
enforrcement and emergency services;
initiaal detention and
a initial heearings;
jails, courts, foren
nsic evaluatiions, and
foren
nsic commitm
ments; reentrry from
jails, state prisons, and forenssic
a commun
nity
hospiitalization; and
corrections and community support.5
The idea
i
is to inteercept most people at
the eaarly points, with
w decreassing
numb
bers at each subsequent point.
p
Below
w, we briefly
y describe th
he initial
three intercepts, as
a these prov
vide the
backd
drop for con
nsideration off mental heaalth courts.
Ideally, people with
w mental illlnesses wou
uld seek or bbe referred foor appropriatte treatment
beforre they engag
ged in seriou
usly anti-soccial or criminnal activitiess. However, even if locall
mental health sysstems had ab
bundant and comprehenssive resourcees, some peoople with
seriou
us mental diisorders undo
oubtedly wo
ould come too the attentioon of the poliice before
reach
hing service providers.
p
Hence,
H
pre-arrrest diversioon programss are likely too be the firstt
pointt of intercepttion (at least within the criminal
c
justtice arena).
Sincee deinstitutio
onalization, law
l enforcem
ment agenciees have playyed an increaasingly
important role in the managem
ment of perssons with meental illness:: experts estiimate that
betweeen seven an
nd ten percen
nt of patrol officer
o
encouunters involvve individuaals
experriencing men
ntal health crrises (some, but not all, of whom aree psychotic oor
dangeerous). Geneerally speakiing, law enfo
orcement offficers have cconsiderablee discretion in
resolv
ving interacttions with peeople who haave mental ddisorders; thhis includes ttheir
5

Also see the GAIN
NS Center Sequ
uential Intercep
pt for Developinng Criminal Juustice-Mental H
Health
Partneerships, availab
ble online at: htttp://gains.prainc.com/pdfs/inntegrating/GAIINS_Sequentiaal_Intercept.pddf

7

authority to issue warnings, convey unruly persons to hospitals or service centers (where
feasible), and make arrests. Mental health advocates suggest that arrest probably should
be the option of last resort; however, when police officers lack knowledge of alternatives
or are unable to access supportive services, they may regard arrest as the best disposition
for people who clearly cannot be left on the street.
While mental health systems and law enforcement agencies historically have not worked
closely together (e.g., they have engaged in little joint planning, cross training, or planned
collaboration in the field), several strategies have been implemented by police
departments⎯with or without the participation of local mental health systems⎯to more
effectively deal with persons with mental illness in the community who are in crisis and
who can be managed using police-level or pre-arrest diversion (level 1 intercept). These
include 1) mobile crisis teams of mental health professionals, 2) mental health workers
employed by the police to provide on-site and telephone consultation to officers in the
field, 3) teaming of specially trained police officers with mental health workers from the
public mental health system to address crises in the field, and 4) creation of a team of
police officers who have received specialized mental health training and who then
respond to calls thought to involve people with mental disorders.
In communities that lack pre-arrest or police-level diversion (intercept 1), individuals
with mental health issues who commit less serious crimes may be considered for postarrest diversion (i.e., diverted to treatment at the initial hearing stage⎯level 2 intercept)
as an alternative to prosecution, particularly if they are nonviolent, low-level
misdemeanants with symptomatic mental illness. In communities that have implemented
strong intercept 1 programs, candidates for post-arrest diversion may be offenders who
have committed more serious crimes.
There are several post-arrest diversion strategies in practice. In some, the courts employ
mental health workers who screen individuals after arrest, either in local jails or at
courthouses. These mental health practitioners then advise the courts about the possible
presence of mental illness and suggest options for assessment and treatment, which could
include diversion alternatives or treatment as a condition of probation. Alternatively,
some courts have developed collaborative relationships with the public mental health
system, which provides staff to conduct assessments and facilitate links to community
service providers.
Optimally, a majority of offenders with mental illness are filtered out of the criminal
justice system in intercepts 1 and 2, receive needed treatment and avoid future crime and
incarceration. In reality, however, it is clear that both local jails and state prisons house
substantial populations of persons with mental illness.
To stem the flow of the persons with impaired mental health into jails and prisons, one
approach that has received considerable attention at the level of intercept 3 is the
establishment of a separate docket or court program specifically to address the needs of
individuals with mental illness who come before the criminal court. These specialjurisdiction courts⎯mental health courts⎯ limit punishment and instead focus on linking
defendants to community-based treatment and other problem-solving strategies to help

8

such individuals desist from future crimes and avoid further involvement in the criminal
justice system.
The Emergence of Problem‐Solving Courts 
Although there are now various forms of problem-solving courts, including mental health
courts, reentry courts, and veterans’ courts, they share a common origin rooted in the
wave of crime and violence associated with illegal drug use, particularly the epidemic of
crack cocaine, in the 1980s. As penalties for drug possession and sales were toughened,
and drug offenders were arrested and prosecuted in unprecedented numbers, the justice
system was overwhelmed by burgeoning caseloads. Then in June 1989, drug treatment
courts (or, simply, drug courts) essentially emerged as a grassroots movement from a
model implemented as a partnership among the Court, the State Attorney’s Office, and
the Public Defender’s Office in Miami-Dade County, FL. The original drug court was
developed to deal with drug-related crimes and drug-using offenders by offering courtmonitored drug treatment to reduce both defendants’ drug use and the constant recycling
of such offenders through the court system. As the program demonstrated success, drug
courts mushroomed throughout the U.S. (see, for example, Rossman 2011 for detailed
discussion of the evolution of the model), and later spawned adaptations such as the
mental health courts that are the focus of this study.
Virtually, all problem-solving courts are community-justice partnerships that include
public agencies and community organizations such as drug treatment and social services
providers. Participants attend regularly held judicial status hearings or court sessions,
receive access to comprehensive treatment services (including mental health or substance
abuse treatment, as well as other services such as employment assistance, family
services), participate in frequent monitoring (such as home visits, drug testing), and
receive sanctions for behavioral infractions, or conversely, incentives for achievements.
The programs’ integration of behavioral modification principles from psychology,
together with recurring courtroom experiences that include the interaction between the
judge and the participant, the public aspect of being sanctioned or complimented, and the
collaborative approach among the “key stakeholders” (including prosecution and
defense) are considered essential elements of problem-solving courts.

9

Problem-solving courts differ from
conventional court case processing in a
number of fundamental ways (Berman and
Feinblatt 2005; Casey and Rottman 2003;
Farole, Puffett, et al. 2005; Office of Justice
Programs and National Association of Drug
Court Professionals 1997):
1) Voluntary participation.
2) A non-adversarial, problem-solving
focus.
3) Integration of treatment services that
ideally represent a continuum of
outpatient and residential treatment, as
well as support groups, with treatment
assignment and frequency of
attendance depending on participants’
particular needs.
4) Intensive supervision of the treatment
process by judges and case managers.
5) Direct conversational interaction
between defendants and the judge
during frequent status hearings.
6) Graduated sanctions, such as more
frequent court appearances or
increased drug testing, are used to
monitor compliance and respond to
problems.
7) Routine drug testing.
8) A team approach to decision-making.

The Drug Court Model
In 1997, the Office of Justice Programs and the
National Association of Drug Court Professionals
described ten basic elements of drug courts,
including:
•

Integration of alcohol and drug treatment
with justice system case processing.

•

Use of a non-adversarial approach through
which prosecution and defense promote
public safety, while protecting defendants’
rights to due process.

•

Early identification and prompt placement of
eligible participants in the program.

•

Access to a continuum of treatment,
rehabilitation, and related services.

•

Frequent alcohol and drug testing.

•

A coordinated strategy among the judge,
prosecution, defense, and treatment
providers to oversee participants’
compliance.

•

Ongoing judicial interaction with each
participant.

•

Monitoring and evaluation to measure
achievement of program goals and gauge
effectiveness.

•

Continuing interdisciplinary education to
promote effective planning, implementation,
and program operations.

•

Partnerships with public agencies and
community-based organizations to support
effectiveness.

9) Incentives to motivate and acknowledge accomplishments.
As noted by Roman, Rossman, and Rempel (2011), the drug court model and other
problem-solving courts have adapted approaches that are consistent with several
prominent crime reduction theories, including:

10

•

Therapeutic jurisprudence posits that legal rules and procedures can be used to
improve psychosocial outcomes, an idea supported by a growing research
consensus that coerced treatment is as effective as voluntary treatment (Anglin,
Brecht, and Maddahian 1990; Belenko 1999; Collins and Allison 1983; DeLeon
1988a, 1988b; Hubbard, Marsden, et al. 1989; Lawental, McClellan, et al. 1996;
Siddall and Conway 1988; Trone and Young 1996).

•

Procedural justice theory predicts that individuals are as concerned about fair
procedures and respectful treatment by legal authorities as they are about the
outcomes of their interactions with the criminal justice system. The literature
suggests that individuals’ judgments of procedural fairness shapes their
perceptions of the legitimacy of and satisfaction with legal authorities, which in
turn influences their compliance with the law and decisions made by those in
authority (Casper, Tyler, and Fisher 1988; Folger 1977; Gottfredson, Kearley, et
al. 2007; Lind 1982; Lind, Kanfer, and Earley 1990; Lind and Tyler 1988;
Thibaut and Walker 1975; Tyler’s 1984, 1990, 2003). Procedural effects
apparently are independent of outcomes produced. Thus, individuals who
perceive they have been treated fairly by the system can demonstrate better
procedural outcomes (e.g., compliance with court mandates), regardless of the
outcome of their case (e.g., the length of the sentence).

•

Deterrence theory holds that the threat or experience of a punishment for an
infraction will reduce the likelihood that the infraction will be repeated; i.e.,
actual or threatened sanctions should deter crime (see Roman, Rossman and
Rempel 2011 for more detailed review of this literature). General deterrence
holds that by increasing the probability that a particular infraction will be
punished, misbehavior will be reduced. Specific deterrence posits that an
individual’s own punishment experience will affect his/her future behavior. For
all facets of deterrence, the goal is to increase expectations that infractions will be
punished; expectations can be changed either by directly punishing an individual,
making highly visible punishments of others, or simply by increasing individuals’
beliefs that punishment will follow an infraction. Three aspects of
punishment─perceived certainty, severity, and celerity of the possible
sanctions─are hypothesized to affect would-be offenders’ decision-making and to
be correlated with offending (Andenaes 1974, Gibbs 1975). The theory is
typically regarded as involving two linkages: 1) a perceptual link, where potential
offenders form perceptions about the risks of committing the crime based on
information regarding sanction policy and other experiences, and 2) a behavioral
link, where the potential offenders’ perceptions of sanction risk influence their
behavior (Paternoster 1987, Scheider 2001). Drug courts employ graduated
sanctions—incrementally more stringent responses to continuing infractions—as
mechanisms to deter future offending. There have been few direct studies of the
effectiveness of deterrence on client outcomes. Marlowe, Festinger, et al. (2005b)
found correlational evidence that drug court clients with higher “elevated”
perceptions of deterrence had better outcomes than those with lower levels of
perceived deterrence while Gottfredson, Kearley, et al. (2007) report that both

11

procedural justice and deterrence contribute to better drug court participant
outcomes.
•

Social learning theory posits that humans learn behaviors (both positive and
negative) from their environment; hence, publicly rewarding pro-social behaviors
can reinforce those behaviors in group settings (Akers and Sellers 2008). Drug
courts often combine deterrence-based approaches with positive rewards for good
conduct based on social learning theory.
Mental Health Courts: An Overview  

While the initiators of the first mental health courts
used the drug court model for inspiration, there are
important differences between the two models. In
fact, some support for mental health courts
developed due to observations that some offenders
with mental illness were not achieving success in
drug courts. Drug courts often have more
formalized sets of goals and phase progression
than mental health courts, and may be more
willing to employ sanctions in response to
noncompliance.

What is a Mental Health Court?
Mental health courts are defined as
specialized court dockets⎯ for certain
defendants with mental illness⎯that
substitute a problem-solving model in place of
traditional court processing. Participants are
identified through mental health screening and
assessments and voluntarily participate
in a judicially supervised treatment plan
developed jointly by a team of court staff and
mental health professionals. Incentives reward
adherence to the treatment plan or other court
conditions, non-adherence may be sanctioned,
and success or graduation is defined according
to predetermined criteria.

The first mental health court was established in
Broward County, FL, in 1997 (Goldkamp and
Source: Council of State Governments 2008
Irons-Guynn 2000) due to a growing concern for
offenders with mental illness that was precipitated
by an increase in suicides in the local jail. A judge convened a task force of mental
health and criminal justice stakeholders to examine methods for better integrating the
mental health and criminal justice systems. During this time, a grand jury also published
a report on the treatment of offenders with mental illness, which further focused attention
on this issue. The Broward County Mental Health Court began on an as-needed-basis for
misdemeanor cases. Within the first two years, the court grew to handle nearly 900
cases. Other mental health courts soon followed in Seattle, WA; Anchorage, AK; and San
Bernardino, CA (Goldkamp and Irons-Guynn 2000).
Federal legislation followed later to reinforce the importance of this innovative court
model. In 2000, America’s Law Enforcement and Mental Health Project funded the
Bureau of Justice Assistance (BJA)/Substance Abuse and Mental Health Services
Administration (SAMHSA) Mental Health Courts Program, which provides financial
support to local and state governments that are interested in implementing mental health
courts (Public Law 106-515). The program has funded 123 mental health courts in 39
states, as of 2011 (BJA undated). The Mentally Ill Offender Treatment and Crime
Reduction Act of 2003 authorized an additional $50 million for collaborative initiatives
between criminal justice and mental health agencies, including programs such as mental
health courts (Public Law 108-414). In addition, the BJA has selected five mental health
courts to serve as learning sites: Akron, OH; Bonneville County, ID; Bronx County, NY;

12

Dougherty Superior Court, GA; and Washoe
County, NV. Other jurisdictions in the country
can contact or visit these sites to learn about how
to develop and implement a mental health court.
The federally-funded Criminal Justice/Mental
Health Consensus Project (administered by the
Council of State Governments) also provides
educational resources and technical assistance on
mental health courts.
Beyond the federal level, mental health courts
cannot succeed without local or state support.
Individual jurisdictions choose to pursue these
initiatives and may also provide funding,
dependent on available resources. With this mix
of local, state, and federal support, the mental
health court model has spread widely and can be
found in every region of the country today.
Current Status of Mental Health Courts 
As of 2011, the Criminal Justice/Mental Health
Consensus Project listed more than 240 courtbased mental health interventions within their
Local Programs Database. Mental health courts
are located primarily in the Western (37 percent)
and Southern (37 percent) regions, with fewer
courts in the Midwest (15 percent) or Northeast
(11 percent) (CSG 2005a). As mentioned
previously, court characteristics vary across
jurisdictions with regard to a number of key court
elements (CSG 2005b):

Key Elements of a Mental Health Court
The following are some defining features of
mental health courts (Almquist and Dogg 2009;
Thompson, Osher, and Tomasini-Joshi 2008):
Serve defendants with mental illness
• Provide diversion for justice-involved
individuals with mental health problems
Involve stakeholders from multiple fields
• Involve stakeholders from the criminal
justice, mental health, substance abuse, and
related fields during the planning stages
and administration
Voluntary and informed participation
• Require informed consent to participate
from defendants
Link participants to community-based services
• Identify appropriate mental health
treatment options in the community and
coordinate entry into these services
(typically achieved through case
management services)
Monitor treatment compliance
• Use court setting to monitor treatment
compliance; track participation in
treatment through regular meetings with
defendants and communication with
treatment providers
Use of sanctions and incentives
• Use sanctions and incentives to encourage
treatment engagement and court
compliance
Therapeutic jurisprudence
• Rely on the principles of therapeutic
jurisprudence to view the court as a
potentially therapeutic experience

•

Size. There is a wide range of caseload
sizes across mental health courts; e.g., a
survey of 90 mental health courts found
annual caseloads ranged from 3 to 852 individuals, with a median size of 36
active clients (Redlich, Steadman, et al. 2006).

•

Entry point. The point of entry into the mental health court also varies across
jurisdictions. Courts generally follow one of two models: the pre-adjudication
model and the post-adjudication model. In the pre-adjudication model,
prosecution is deferred until the client completes the program. In the postadjudication model, the defendant’s participation in the court is contingent upon a
guilty plea. The post-adjudication model is more common among mental health
courts, particularly among those that accept felony offenders (Almquist and Dodd
2009). On the one hand, pre-adjudication models can cause challenges with

13

future trials if a client fails to successfully complete the mental health court
program (e.g., the prosecutor may be unable to find witnesses or the witnesses
may not recall salient details of the incident/case). On the other hand, clients are
left with little protection if they fail the program in courts using a postadjudication model.
•

Eligibility criteria. Mental health courts generally have two types of eligibility
criteria: clinical eligibility and legal eligibility. In terms of clinical eligibility,
courts vary with respect to which mental health problems are acceptable for
inclusion in their programs. For instance, some courts use a legally-defined term
of serious/severe and persistent mental illness (SPMI), while others use a clinical
definition such as Axis I disorders. Courts also vary regarding how severe
symptoms may be and still qualify for inclusion. The acceptance of traumatic
brain injuries, developmental disorders, and Axis II or personality disorders also
are controversial clinical eligibility issues. Some courts purposely exclude these
diagnoses, whereas other courts are willing to accept participants with these
conditions on a case-by-case basis.
In terms of legal eligibility, most courts have restrictions on certain types of crime
(e.g., rape, arson). Half of surveyed mental health courts in 2005 admitted both
misdemeanor and felony offenders (Redlich et al. 2006). Two-thirds of BJAfunded mental health courts automatically deny violent offenses (CSG 2005b).
However, more courts are accepting serious and violent crimes than at the outset
of the mental health court movement. Past criminal history is another point of
diversity across courts. While some courts disqualify defendants for extensive
criminal histories, other courts view this as a symptom of the “revolving door”
and do not exclude solely on this basis. There are also different philosophies on
the role mental illness must play in the current charge. Some courts require that
the mental illness be directly tied to the offense, while others merely require that
the defendant have mental illness, regardless of its relation to the crime.

•

Screening. In addition to the eligibility criteria, themselves, courts vary in terms
of how they determine individuals’ mental health needs and how comprehensive
their screening measures are in terms of detecting and identifying mental health
problems. Some jurisdictions complete relatively informal assessments, while
other jurisdictions conduct full, systematic psychiatric evaluations with collected
information and records from other external sources (e.g., previous medical
records, family interviews).

•

Program staffing and services. Mental health courts differ in terms of how the
court team is constructed (e.g., rotating vs. dedicated attorneys, internal vs.
external treatment and case management services) and forms of treatment,
depending on that particular jurisdiction’s court and community resources.

•

Supervision and monitoring. Once accepted into a mental health court program,
clients may experience vastly different forms of supervision and court monitoring.

14

Supervision in the community may be the responsibility of treatment providers,
probation officers, mental health court personnel, or other criminal justice
agencies (Redlich et al. 2006). Additionally, judicial status hearings may range
from multiple times per week to quarterly appointments, although most courts (91
percent) have participants begin with weekly to monthly hearings (Redlich et al.
2006).
•

Sanctioning. If participants are noncompliant with court requirements, there may
be a variety of sanctions used. Often, the most severe sanction is return to jail.
Courts also vary on their use of this controversial punishment, with 8 percent of
courts reporting that they never use jail as a sanction, and the largest portion of
courts (39 percent) using jail sanctions in 5 to 20 percent of cases (Redlich et al.
2006). Mental health courts with more felony offenders, and with more frequent
status hearings, tend to use jail sanctions more often.

•

Program completion. A study of 400 participants from four mental health courts
found 48 percent had graduated and another 23 percent remained in the program
one year after entry into the court (Redlich et al. 2010). As is the case with drug
court and other problem-solving court programs, if participants fail to comply
with program requirements to the extent that their sanctioning options are
exhausted, they may be terminated from the program. Redlich and colleagues
(2010) found nearly one-third (29 percent) of participants in their study were
terminated from the programs; differences were seen among courts with
termination rates ranging from 17 to 41 percent.

Although mental health courts have become a popular approach for responding to
offenders with mental illness, some concerns have surfaced. For instance, Sarteschi and
colleagues (2011) discuss thorny issues surrounding the extent of voluntariness and
participant comprehension, expungement practices for reduced or eliminated charges in
post-adjudication models, possible gender and race selection biases, and “creaming”
practices where mental health courts only accept clients with few risk factors who would
be expected to succeed regardless of program participation. The issue of “creaming” is
particularly important when considering evaluation methodology. Since admission into
these programs can sometimes be dependent on the subjective assessments of multiple
stakeholders (e.g., judge, prosecutor, clinical team, etc.), there is a risk of selection bias
where more “promising” candidates expected to succeed and present less danger to the
community are chosen for participation (Wolff et al. 2011). This presents a challenge to
evaluation studies comparing outcomes to other offenders with mental illness who may
already have a higher risk for reoffending than those participating in the mental health
court from the outset.
Mental Health Court Research 
Mental health court research is still relatively sparse; however, existing process and
outcome studies and meta-analyses provide preliminary support for the viability of this
approach. The literature documents a mix of single and multisite studies in which the
majority of multisite studies are descriptive or process evaluations. Most of the reported

15

outcome studies use quasi-experimental designs, with the exception of one study
(Cosden, Ellens, et al. 2003) that implemented random assignment. Studies also employ a
mix of self-reported information and administrative records for outcome measures, and
most studies have follow-up periods of up to two years from program entry.
Process Evaluations 
Descriptive or process evaluations of mental health courts describe the way the courts
function or examine intermediary goals or mechanisms, such as engagement and
delivery/receipt of treatment and other services. One area of interest for these studies is
describing the populations served by mental health courts. For instance, a study of a
mental health court in North Carolina documented demographic and criminal history
differences from North Carolina arrestees, with more misdemeanant, nonviolent, female,
white, and older offenders found in the mental health court (Hiday, Moore, et al. 2005).
Similar findings regarding the higher likelihood of female, white, and older individuals
within mental health courts were replicated in another study on seven courts (Steadman,
Redlich, et al. 2005). Petrila and colleagues (2001) also identified some special needs of
the mental health court population in Broward County, with approximately one-quarter of
participants evidencing homelessness, and 29 percent having co-morbid substance abuse.
Another question explored by process evaluations is the role of activities or outputs
within mental health courts. Lacking an understanding regarding whether intermediary
mechanisms (i.e., mental healthcare, court monitoring by a specialized judge, etc.) are
being successfully implemented, it is difficult to interpret subsequent outcome findings.
Two areas in particular have been studied: courtroom procedures and provision of mental
health services. Court observations have revealed an atmosphere marked by less
formality and the absence of an adversarial orientation, as compared to more traditional
courts (Petrila Poythress, et al. 2001). Respondents in this same study believed the judge
was the primary source of success in their court.
A few researchers indicated they found a reluctance to sanction program participants
within the courts they studied, a characteristic these authors contrasted with the drug
court model (Griffin, Steadman, and Petrila 2002; Petrila et al. 2001). In our view, this
may be overtly or subtly due to a key distinction in the way the court teams perceive the
participants, the nature of their anti-social behavior, and their abilities to conform with
required activities. Despite the potential for overlap in their presenting problems, MHC
participants must have mental health problems or mental illness, neither of which is
illegal. By contrast, drug court participants must exhibit alcohol or substance abuse, both
of which constitute illegal behaviors.
In terms of mental healthcare, past studies have shown that MHC clients are more likely
to engage in treatment than they were prior to participating in the mental health court
program (Ridgely, Engberg, et al. 2007) and compared to a similar group of offenders
with known mental health problems (Boothroyd, Poythress, et al. 2003; Trupin and
Richards 2003). Respondents in Petrila and colleagues’ study (2001) of the Broward
County MHC acknowledged potential challenges regarding the program’s capacity and
available services in the community. Similar issues with community resources also were

16

discussed in a descriptive study of four courts in different states (Watson, Hanrahan, et al.
2001).
A few studies also have examined procedural justice issues within mental health courts.
While participants often report positive views on issues such as procedural justice and
coercion based on their experience with mental health court (O’Keefe 2006; Poythress,
Petrila, et al. 2002; Redlich, Hoover, et al. 2010; Wales, Hiday, and Ray 2010), evidence
has shown that participants may not fully understand the voluntary nature of these
diversion programs or other important information about how the courts function
(Redlich, Hoover, et al. 2010).
One study of particular interest is an evaluation of the Brooklyn Mental Health Court
describing the functioning of the court within its first 28 months of existence (O’Keefe
2006). The report provides a comprehensive portrait of the court’s functioning at that
point in time. A selection of these observations follows. The court began with a
restriction against violent offenses, but slowly incorporated more of these types of felony
cases, an evolution towards accepting more severe crimes that other courts have also
experienced (Redlich, Steadman, et al. 2005). The report assesses the referral/screening
process and finds it to be inconsistent, lengthy, and at times unclear. At the time, no
universal screening was in place, and referrals had to be reviewed by the Assistant
District Attorney, which could take weeks. Of those admitted to the court, the primary
diagnoses were bipolar disorder (28 percent), schizophrenia (26 percent), and major
depression (25 percent); nearly half of participants had co-morbid substance abuse. Like
other studies, respondents reported challenges providing treatment within the resources
available in the community and indicated this contributed to delays in receiving
treatment. Courtroom observations were characterized by high levels of direct
communication between the judge and MHC participants, including hand-shaking and
verbal, positive feedback.
Outcome Evaluations 
While the previously described research is important to understanding how courts
function, additional research is necessary to know whether the courts achieve their
ultimate goals, particularly in relation to mental health and criminal justice outcomes.
Studies have shown positive impacts on clinical outcomes. As described previously,
mental health court participants are more likely to engage in treatment; however,
evidence of clinical improvement (e.g., reduction in mental health symptoms) is more
ambiguous. Mental health court participants may experience improvements in substance
abuse (Cosden et al. 2003) and level of “functioning”6 compared with other offenders
with mental health diagnoses (Cosden et al. 2003; O’Keefe 2006; Trupin and Richards

6

All three studies used the Global Assessment Functioning (GAF) to assess level of functioning among
clients. The GAF score is a numerical measurement of the individual’s psychological, social, and
occupational functioning based on clinical judgment (American Psychiatric Association 2000).

17

2003). However, a review7 of eight studies with clinical outcomes found only two with
clinical improvement (Sarteschi, Vaugh, and Kim 2011).
Findings have also been mixed regarding criminal justice outcomes, although a consensus
seems to be building in favor of mental health courts. O’Keefe (2006), for example,
conducted a small outcome study of 37 Brooklyn Mental Health Court participants who
had completed the program and found that arrest rates declined after participation (from
27 to 16 percent) compared to the12 months prior to court enrollment, although this
change was non-significant (possibly due to low statistical power given the small sample
size). Sarteschi and colleagues (2011) conducted a meta-analysis of 18 studies published
before July 2009. Their analysis showed that mental health court participants had better
criminal justice outcomes than similar comparison groups. The lack of rigorous study
design prevents strong conclusions, though; studies with better methodological design
found smaller impacts, revealing a possible influence of selection bias.
Examples of positive findings from individual studies include that mental health court
participants spend fewer days in jail post-release than they had before going through the
specialized court (Cosden, Ellens, et al. 2005; Ridgely et al. 2007). In addition, they
spend less time in jail than other offenders with mental illness (Christy, Poythress, et al.
2005; Trupin and Richards 2003). In terms of recidivism, mental health court
participants have more days before re-arrest compared with other arrestees with mental
disorders (McNiel and Binder 2007) and experience fewer arrests than they did before
participating (Christy et al. 2005; Herinckx, Swart, et al. 2005) and compared to other
offenders with mental illness (Cosden et al. 2003; McNiel and Binder 2007; Moore and
Hiday 2006).
While these studies have shown positive impacts of mental health courts, other studies
have been more equivocal. Some research has shown similar outcomes for both mental
health court clients and other individuals with mental illness processed more traditionally
in terms of symptom reduction (Boothroyd et al. 2005), time spent in jail (Cosden et al.
2003), and time to re-arrest (Christy et al. 2005). Most outcome studies examine
individual courts, which may account for conflicting findings across studies. Many also
do not include process components, making it difficult for researchers to isolate possible
causes of differing outcomes.
A study published after Sarteschi et al.’s meta-analysis (2011) examined the criminal
justice outcomes of mental health court participants in four jurisdictions compared with
propensity score-matched controls (Steadman et al. 2011). This study has a stronger
methodological design than many of the earlier studies, using modified propensity score
matching to address potential selection bias into mental health courts. Compared with
matched offenders with mental illness undergoing traditional processing, mental health
court participants across the four jurisdictions were less likely to be arrested, had a larger
reduction in arrest rate, and spent fewer days incarcerated during the 18 months after
program entry (intervention group) or jail admission (comparison group). For those
7

The authors attempted to conduct a meta-analysis of these studies, but the effect sizes were too
heterogeneous to produce a valid estimate.

18

involved in the mental health court, substance use, schizophrenia or depression
diagnoses, lack of pre-MHC treatment, and more significant criminal history were related
to worse criminal justice outcomes.
Limitations of Existing Research  
Methodological weaknesses of extant studies (e.g., sole reliance on self-reported
outcomes, lack of random assignment, and short-term follow-up) make it difficult to
reach confident conclusions. Most of the existing outcome studies examine single courts,
which may account for conflicting findings across studies; however, meta-analyses help
to produce overall estimates of mental health courts’ effectiveness.
In addition to these shortcomings, little research has been conducted on questions of
efficiency and cost. One study (Ridgeley et al. 2007) investigated costs for a mental
health court in Allegheny County. This study found that the jurisdiction’s mental health
court costs were similar to those of the traditional court system. The authors speculated
that it was likely that the mental health court might become less costly over time.
Given the state of the field, additional research is warranted to fill gaps in knowledge and
help stimulate informed decision-making by practitioners who are either implementing
new mental health courts or strengthening their existing efforts. The current study is
intended to support this objective by: 1) contributing a two-site process and outcome
evaluation of mental health courts in New York City, using sophisticated analytic
techniques to control for selection bias (the largest methodological threat to mental health
court evaluation research) and 2) presenting guidance to support future economic
analyses.

19

CHAPTER 2. EVALUATION DESIGN AND METHODS 
This study consists of process and impact evaluation components that draw on multiple
data sources to describe the two New York court programs (their operations, structures,
and client caseload) and assess the impact of each program on participant criminal justice
outcomes. The impact evaluation features a quasi-experimental design and employs
propensity score matching (PSM) to construct equivalent comparison groups for each
court’s treatment sample. In this section, we describe the study’s methods and sample
construction. Limitations and considerations associated with the final design are
discussed in Chapter 6.
The Proposed Study 
In October 2005, NIJ selected UI to evaluate two distinct criminal justice interventions
for offenders with mental illness: the Brooklyn Mental Health Court, a specialized
problem-solving court operating in the Supreme Court in Brooklyn, NY, and the Pinellas
County Mentally ill Diversion Program (MIDP), operating in the Public Defender’s
Office in Clearwater, FL. The evaluation was intended to answer the following questions:
•

Does either model reduce participants’ criminal justice involvement?

•

Does either model increase access to, participation in, or retention in mental
health treatment?

•

Is either model cost-effective compared to business-as-usual, and is one more cost
effective than the other?

Because the two interventions differed significantly in their approaches toward offenders
with mental illness and their operational structures, UI developed separate and slightly
different evaluation design strategies for each.8 Evaluation activities began in each site in
early 2006, following NIJ’s approval of UI’s proposed evaluation strategies.
Although UI encountered early challenges in both locations, the most pressing issues
occurred in the Florida treatment site where evaluation activities were hampered by
funding uncertainties and on-going substantive and structural changes to the MIDP
intervention (e.g., changes in target population, screening and assessment procedures,
staff roles and responsibilities, and services), as well as administrative barriers and legal
constraints that impeded the research. Following several months of negotiations with NIJ,
the decision was made in January 2007 to cease data collection and evaluation activities
in Florida and to explore adding a second New York City mental health court program to
the evaluation in order to optimize resources.
8

Although both strategies employed quasi-experimental designs with impact and process evaluation
components, as well as cost-benefit and transferability analyses, the New York evaluation relied on
administrative records data and a retrospective sample, while the Florida study required a prospective
sampling approach in two jurisdictions (treatment and comparison). Evaluation design reports for both sites
were presented to NIJ for review and approval in early 2006.

20

Ultimately, the initial research plan was modified, as follows:
•

A second New York City treatment court site—Bronx Mental Health Court—was
approved by NIJ in September 2008, and formally agreed to participate in the
evaluation in January 2009.

•

The proposed network analysis (initially designed to support the transferability
analysis) was eliminated due to concerns that this type of assessment is costly and
likely would yield little value in a study of this nature.

•

Focus groups with both MHC participants and, separately, comparison group
offenders were eliminated to conserve evaluation resources.9

•

Comparison court sites were eliminated from the courtroom observation task to
conserve evaluation resources; courtroom observation of both the Brooklyn and
Bronx MHC sites was retained. The rationale for this modification was that
examining differences between the two court models, rather than between the
mental health courts and traditional courts, would yield more relevant
information.

•

The proposed economic analyses task was re-purposed. Since researchers were
unable to access information about post-incarceration services and utilization for
either the treatment or comparison groups, the cost analysis shifted from an
assessment of whether mental health courts are cost-effective relative to
conventional case processing to development of a practitioner tool to guide
jurisdictions interested in conducting costs analyses (including the type of
information to collect and how to estimate critical program and criminal justice
costs). This is discussed more fully in Chapter 5.

While most of the original expectations about the Brooklyn evaluation held regarding
data acquisition and sample construction, it become clear during the design phase that
several proposed evaluation features were not viable, notably the ability to measure
retention and utilization of post-incarceration treatment services (Appendix A provides a
table listing both the original and final designs for the Brooklyn Mental Health Court,
facilitating quick comparison). Although early discussions with court partners suggested
that it would be possible to access post-release service use data from the New York State
Department of Health (DOH) Office of Medicaid Management, which contains service
data (provider, type of service, recipient, dates of service) for all services paid by
9

Additionally, it was thought that participant focus groups, even with the addition of comparison ‘businessas-usual’ subjects, would not add much to our understanding of the courts’ operations. Other
studies⎯namely, Policy Research Associates’ (PRS’) qualitative evaluation of the Brooklyn Mental Health
Court, on-going in 2006—were using a similar approach, and findings were expected before the end of this
evaluation. With funding from the National Science Foundation, PRS was examining participant
perceptions, understandings, and appreciation of BMHC procedures and requirements; as part that study,
PRA was interviewing BMHC participants at court intake, and then examining some defendant-level
outcomes (such as bench warrants) at 12 months. Analysis focused on participant understandings in
relationship to specific case outcomes.

21

Medicaid, agency policies prohibited the release of individual-level data for research
purposes (e.g., data cannot even be released to the agency’s own divisions), unless the
research is directly related to improvements in the delivery of the Medicaid program,
itself. Without post-detention service information, the impact evaluation for the Brooklyn
court site (and, ultimately, the Bronx court evaluation, as well) was limited to criminal
justice outcomes only. Sufficient data could not be collected to support economic
analyses, either, for reasons discussed in Chapter 6.
Final Evaluation Design  
The final evaluation design focused primarily on measuring the impact of the two court
programs on participant recidivism. The study features a quasi-experimental design that
includes process and impact analyses. Both evaluation components relied on multiple
data sources. Semi-structured interviews with program staff and key partners, repeated
courtroom observation of the Brooklyn and Bronx MHCs, and document review
informed the process evaluation. The impact evaluation relied on administrative data
drawn from the two New York City MHC programs, the New York City Department of
Health and Mental Hygiene (DOHMH), and the New York State Division of Criminal
Justice Services (DCJS); court program data also supported the descriptive analyses
presented in Chapter 3. The study’s key components and their objectives are described
below.
Process Evaluation 
The objective of the process evaluation was to:
•

Document the operational structure of the mental health courts and how the two
MHCs differed from business-as usual (BAU) in their respective jurisdictions.

•

Identify any significant changes made to the program model during the study
period, and explore the rationale for those changes.

•

Examine factors that impeded or facilitated either program’s ability to achieve
intended objectives.

As noted above, the process evaluation drew on multiple data sources, but relied
primarily on 1) semi-structured interviews with program staff, key criminal justice
partners, and a limited set of mental health treatment providers⎯both those contracted to
serve mental health court participants and staff at the NYC Department of Health and
Mental Hygiene (DOHMH) to document program operations, policies and procedures, as
well as business-as-usual; and 2) courtroom observations to explore the in-court
dynamics of the two mental health court programs and how program philosophy
manifested. Program materials and documents also were important sources of secondary
information for this evaluation component. For example, the Mental Health Court
program data described in the Impact Evaluation section, below, were used to generate
process evaluation findings presented in Chapter 3.

22

Field Visits 
Project researchers made a series of site visits early in the study (October and November
2005) to the two original programs to document court operations and gather information
for the impact analysis. These visits focused on understanding the logic and underlying
philosophy of the Brooklyn MHC program and the FL site that was subsequently dropped
from the research. Following the resumption of evaluation activities late in 2008, the
research team visited both the Bronx and Brooklyn MHCs twice between January and
October 2009 to: 1) document extant program operations; 2) identify any changes in
procedures, practices or program policies; and 3) observe the courtroom operations. The
visits to the Bronx MHC also focused on data acquisition (identifying available data
sources, the parameters of those data, and the process for obtaining access). During the
October 2009 visit, researchers also met with DOHMH staff to understand discharge
planning procedures mandated by the Brad H lawsuit.10 (The Brad H lawsuit and its
relevance to this study are detailed in Chapter 3.)
The team also had intended to conduct field visits to: 1) Rikers Island to speak with
facility staff and site-based providers, and to observe protocols involving determinations
of mental illness and responses to inmates’ mental health needs; and 2) community-based
treatment providers who provided mental health services to MHC program participants.
Constrained time and resources ultimately undermined the team’s ability to perform these
activities. However, a series of phone interviews subsequently were conducted with
treatment providers who routinely served MHC participants to have them describe the
range of treatment and services offered, as well as how they interacted with the two court
programs.
To aid in identifying changes in program operations that occurred during the project’s 21month hiatus and to facilitate quick comparison of the two court sites, researchers
developed a matrix of key program characteristics for both Brooklyn and the Bronx
mental health courts. We populated the matrix based on review of initial field notes (for
the Brooklyn court site), program reports, and other materials obtained from the two
court programs. Staff in both sites reviewed the matrix, corrected any discrepancies, and
noted any key changes in program practice or policies. (The matrix of key court
characteristics is located in Appendix B.)
Stakeholder Interviews 
Stakeholder interviews with key court personnel (i.e., judges, prosecutors, and defense
attorneys), program staff (i.e., clinical directors, case managers, and intake coordinators),
and treatment providers focused on documenting current program operations, treatment
philosophy, and factors that impede or facilitate achievement of program objectives.
Semi-structured interviews with DOHMH staff focused primarily on describing business10

Since 2003, New York City has provided discharge planning services to inmates with mental illness
under the settlement terms of a class-action lawsuit, Brad H vs. The City of New York. The lawsuit argued
that given the number of inmates with mental illness who are treated by DOC, it functions as a de facto
psychiatric hospital and, as such, must provide comparable aftercare and discharge planning services to its
inmate-patients.

23

as-usual and the scope of services available to offenders with mental illness. The latter
also explored the importance and impact of the Brad H class action lawsuit in service
provision to offenders with mental illness, including discharge planning.
Courtroom Observation 
Since mental health courts are a form of problem-solving courts, they are often viewed as
an analogous application of therapeutic jurisprudence.11 Further, the theoretical model for
the study’s two mental health courts hypothesizes that active judicial monitoring helps to
achieve positive outcomes. With this in mind, the courtroom observation component of
the process evaluation was designed to achieve three central goals, to:
1. Document the day-to-day courtroom operations of the Brooklyn and Bronx
specialized mental health courts.
2. Explore the application of therapeutic jurisprudence; that is, to explore how
principles of therapeutic jurisprudence are practiced in mental health court
settings.
3. Examine similarities and differences between the mental health court models and
drug court models including how practices within mental health courts might
differ from the drug court approach.
Structured observations of judicial status hearings were conducted in both courts between
April and October 2009, with observations of two full court sessions in Brooklyn and one
full court session in the Bronx. In all, UI researchers observed in excess of 130 scheduled
court appearances (86 in Brooklyn and 46 in the Bronx).
Data collection instruments were developed (adapted from O’Keefe (2006); see
Appendix C for a copy of the study’s courtroom observation tools) to systematically
document courtroom dynamics, including the nature and tone of interactions between
courtroom actors and defendants, and the details of individual-level appearances (e.g.,
stakeholders in attendance, level of participation, decision-making process, information
sharing, demeanor of judges), as well as the disposition of each case discussed/heard
(e.g., type of appearance, compliance status, court response, judicial interaction). Our
framework for understanding courtroom dynamics within the mental health courts was

11

Therapeutic jurisprudence, a term, introduced by David Wexler, refers to ways in which the practice of
the law can be used to support and enhance beneficial outcomes beyond the immediate case disposition
(Wexler and Winick 1996). As such, therapeutic jurisprudence seeks to achieve therapeutic outcomes
through the legal system without subordinating due process and other justice values. The goal is to practice
law in a way that supports the health and well-being of those being tried in a court of law (Rottman and
Casey 1999). Two aspects of mental health courts that are theorized to promote beneficial therapeutic
outcomes: mental health treatment and ongoing judicial monitoring. The latter is hypothesized to promote
treatment adherence, thereby improving mental health outcomes and reducing criminal behavior.

24

shaped by Satel’s (1998) work conceptualizing elements of the drug court setting.12 The
structured courtroom observation data collection included the following components:
•

Length of each appearance.

•

Characteristics of the participant⎯ limited to basic demographics, and whether
the participant was incarcerated at the start of the hearing (i.e., “Yes” or “No”).

•

Appearance type⎯options included MHC status hearing, MHC
graduation/sentencing, pre-MHC or plea, non-MHC case, and no-show or nonappearance. Observers also could also select other/ unknown, and record the
nature of the appearance.

•

Courtroom participants⎯observers checked off all staff (judge, case manager,
resource coordinator, prosecutor, defense counsel, defendant, community
treatment provider, and others) who participated in each appearance, and whether
each party spoke (beyond stating their name or greeting).

•

Judicial interaction with defendant⎯observers were asked to check off all that
applied for each appearance with respect to how the judge interacted with the
defendant (possible options were: eye contact with defendant, asked non-probing
questions of defendant, asked probing questions of defendant, imparted
instructions/ advice to defendant including consequences of future compliance
and noncompliance, directed comments to audience, defendant approached bench,
spoke off-record to defendant, and touched or shook hands with defendant).

•

Defendant interaction with judge⎯observers recorded if the defendant asked
questions or made statements to the judge or displayed art/talent for the court.

•

Good report⎯observers were instructed to check “good report”13 if there was any
compliance noted during the hearing, and to record (check all) the types of
rewards conferred on participants. Options included: none, courtroom applause,
praise/recognition from the judge, decreased court appearances, phase certificate,
and graduation.

•

Bad report⎯observers were instructed to check “bad report”14 if there was any
noncompliance noted during the hearing, and to record (check all) the types of

12

Since mental health courts, in many ways, evolved from drug courts, we looked to Satel’s work for
guidance and adapted her methodology, together with instruments UI used in NIJ’s Multi-Site Adult Drug
Court Evaluation (see Volume 1, Chapter 3 of Rossman, Roman, et al. 2011) to understand mental health
courts and recorded information about the mental health court sessions and interactions during individual
defendant appearances.
13
Under the “Compliance Status” section of the courtroom observation reporting instrument, observers
could select both “Good report” if ANY compliance was recognized and “Bad report” if ANY
noncompliance was noted. Selecting both indicated “mixed” compliance by the defendant.
14
See the above note on compliance status.

25

sanctions conferred on participants. Options included: none, adjustment to
treatment plan, admonishment/recognition from judge, increased court
appearances, remand to jail, and failed MHC including sentence or other
consequence.
•

Defendant satisfaction with hearing⎯observers were instructed to select all that
applied for each appearance with respect to the defendant’s demeanor. Options
included: visibly happy/satisfied, neutral, visibly unhappy/ upset/dissatisfied, and
other.

•

DA or defense attorney reaction⎯ observers were instructed to record a brief
description of the DA’s or defense attorney’s reaction, if notable.

These structured data served to highlight differences in the courts’ procedures and
interaction with participants, and offered important context in which to interpret results
from the impact analysis.
Impact Evaluation 
The impact analysis assessed the effect of the two mental health courts on participant
criminal justice outcomes, specifically if participation in mental health court reduces
subsequent criminal justice involvement⎯namely, re-recidivism as measured by new
arrests and new convictions after admission to mental health court. A quasi-experimental
design was employed to compare the outcomes of mental health court participants with
other defendants who were mentally ill (primarily felony offenders with Axis I
designations, arrested in either the Bronx or Brooklyn, consistent with the MHCs’ target
populations) whose cases were processed as usual in the criminal justice system between
2002 and 2006.
Defining the Treatment and Comparison Groups 
Consistent with the study’s approach to conduct separate (not pooled) evaluations of the
two courts, four retrospective samples⎯a treatment group sample for each court program
and matched comparison groups for each court program⎯were drawn from
administrative records data maintained by the New York State Division of Criminal
Justice Services (NYS DCJS); New York City Department of Health and Mental Hygiene
(DOHMH) and the program databases maintained by each court program. In addition to
supplying cases for the treatment and comparison group samples, these data also
supported the impact analysis. Cases that met the following criteria were considered
eligible for the research:
•

Axis I mental health diagnosis indicative of a serious, persistent mental illness.

•

Felony offender (current offense).

•

Arrested in either the Bronx or Brooklyn between 2002 and 2006.

26

•

Detained in jail (Rikers Island) awaiting disposition.

•

Treatment group cases had to be MHC participants (individuals who were
referred, but did not participate in MHC were excluded) with valid plea date
information.

•

Comparison cases had to have been “deemed” or “designated” as eligible for Brad
H discharge planning services (i.e., sufficiently mentally ill to qualify for services
under the Brad H settlement).15

Propensity score matching techniques were used to construct equivalent comparison
groups, as discussed in Chapter 4 (Impact Analysis and Findings).
The treatment group for the Bronx impact analysis consisted of individuals who
participated in the Bronx MHC between January 1, 2002 and December 31, 2006. Of the
648 individuals who participated in the Bronx program during the reference period, 564
were matched to arrestees in jail with a diagnosed mental disorder (comparison group).
The treatment group for the Brooklyn impact analysis consisted of individuals who
participated in the Brooklyn MHC between March 1, 2002 and December 31, 2006; the
earlier reference date coincides with the inception of the Brooklyn MHC program. Of the
327 individuals who participated in the Brooklyn program since its inception, 316 met the
research criteria for inclusion in the impact analysis with 303 matched to appropriate
comparison cases (i.e., arrestees in jail with a diagnosed mental disorder).
Comparison groups for both impact analyses were drawn from a pool of approximately
5,000 offenders16 entered in the Brad H database maintained by DOHMH. The pool of
potential comparison cases consisted of individuals who were 1) arrested between
January 1, 2005 and December 31, 2006 in either Brooklyn or the Bronx and 2) either
“designated” or “deemed” as eligible for Brad H services in the DOHMH database.
Data Acquisition and File Construction 
Since the study used retrospective samples drawn from administrative records data,
informed consent was not obtained from individual offenders. To maintain the
confidentiality of the data and the anonymity of individual research subjects, UI and its
study partners devised a two-step protocol to provide redacted, but linkable analytic
datasets to UI using a vetted intermediary agency. Under the two-step protocol, the NYS
DCJS served as the intermediary by assigning unique, random, anonymous research IDs
15

“Designated” individuals were determined to be sufficiently mentally ill to qualify for services after
having gone through a series of jail-based mental health assessments; “deemed” individuals received a
baseline amount of mental health treatment in jail, but were released from custody before the full
assessment was completed.
16
The initial DOHMH data file contained records on 9,493 unique individuals, but missing data on key
variables reduced the number of viable cases to roughly 5,000.

27

to each study record, thus allowing UI to link information across databases without using
personal identifiers.
Step 1 required each partner agency (i.e., both MHC programs and DOHMH) to identify
cases for the sample from their respective databases using the parameters outlined by UI
(described below) and to submit the case identifiers only to DCJS (the submission to
DCJS contained identifiers only and no substantive program data such as mental health or
criminal justice status). DCJS then matched the individuals to their criminal history
records and assigned a unique, anonymous research ID to each record. DCJS then
appended the research ID to the identifiers-only file and returned it to the appropriate
partner agency.
Step 2 required each partner agency to extract substantive program data about sample
members, append the research ID to these records, remove all personal identifiers, and
submit the de-identified file of sample data to UI. Additionally, DCJS extracted criminal
history records for all court and comparison group sample members, including sealed
records, appended the research ID to these records, removed all personal identifiers, and
submitted the de-identified file to UI. Under this protocol, UI then linked records across
partner submissions using the anonymous research ID to create an analytic file. Figure
2.1 - Data Flow Chart provides an illustration of the two-step protocol.
Below, we briefly discuss the administrative data sources identified in Figure 2.1.
Mental Health Court Program Data 
Both the Bronx17 and Brooklyn18 MHCs maintained program databases containing basic
client-level information about program participants, and to a lesser extent, candidates
referred to their respective programs. These data informed descriptive analyses about
each court program (quantified participant characteristics and case processing statistics)
and provided the basis for construction of treatment group samples used in the impact
analysis.
For the current study, UI requested data on demographics, mental health status (current
diagnosis and history), co-occurring substance abuse, arrest and referral dates, criminal
charges, court case processing dates and milestones, case management services provided
and mental health services utilization from both court programs. Not all data were
17

Data about clients referred to the Bronx MHC between 2002 and 2006 were maintained in an Access
database that was originally developed for the “regular TASC” drug court program. This database
chronicled information on client demographics, legal representation, criminal history, substance use history
and HIV status, indictment charges, plea upon entry to the mental health court, program assignments,
warrants issued, final case disposition, and reasons for case termination (e.g., successful, unsuccessful, or
neutral case closing). Data were collected on all persons referred to the program; data entry occurred at the
time of initial screening by a TASC case manager, and the database was updated as needed throughout
TASC’s involvement in a case.
18
The Brooklyn MHC stores data as far back as March 2002 in an Access database: the Mental Health
Court Application. This database contains information on participant demographics, mental health
assessment results, court mandates, treatment compliance, and graduation status. It does not, however,
maintain post-program information about clients.

28

a Collection Plan: Evalu
uation of Crriminal Justice Interveentions for
Figurre 2.1. Data
Offen
nders With Mental Illn
ness

229

available electronically,19 and some data were not routinely collected during the study’s
reference periods (2002 through 2006). The matrix in Appendix D catalogues the types of
data supplied by each program, as well as DOHMH and the NYS DCJS, and the extent to
which comparable data were available across the sample groups.
Criminal Justice Records Data  
Criminal justice records data from the NYS DCJS were collected on all cases meeting the
research parameters. The study relied on computerized criminal history data to measure
both prior criminal involvement (age at first arrest, number of priors) and recidivism (any
arrest post-release and entry into the sample, and subsequent convictions). Criminal
justice data were collected on each subject through December 31, 200820 to allow a
minimum 24-month follow-up period for all cases in the research sample (i.e., the last
cases entered the sample in December 31, 2006).21
Mental Health Records Data 
UI collected data from the DOHMH’s discharge planning database to build appropriate
comparison groups of defendants with mental illness whose cases were adjudicated as
usual in traditional courts. The DOHMH provided data on all persons who were arrested
in Brooklyn or the Bronx during 2005 and 2006 who were sufficiently mentally ill to
receive services under the terms of the Brad H settlement (i.e., persons deemed or
designated as eligible for Brad H discharge planning services). UI received basic
demographic data, including: age at arrest, arrest date and jail admission data, release
information, mental health status (i.e., Axis I and II diagnoses, as well as measures of
functional status and the severity of illness), and mental health assessment and discharge
planning service dates.
Cost Analysis  
UI researchers had initially proposed a general analytic approach to estimate the costs
and outcomes on a per-person basis for both the intervention and comparison groups.
Although the project planned to estimate the cost of program inputs through a review of
secondary data (i.e., financial reports, budgets, invoices, progress reports, and other
documents to identify the costs of labor and services) and to supplement this information
through semi-structured interviews with program personnel to develop estimates of the
quantity of services used by each individual defendant, this proved infeasible. It became
clear during the feasibility phase (Phase I) that evaluation activities in neither site would
support this approach. Following the project’s 21-month hiatus (during which the second
19

UI explored manual data extraction from hard-copy case files to collect service utilization data, but time
and resource constraints precluded it.
20
Criminal justice records data, like all administrative data for the sample, were compiled according to the
two-step process described above; the last provided list was submitted to DCJS for processing in mid-2010,
thus allowing ample time for any official contact with the criminal justice system to have been recorded in
the criminal justice data system.
21
Most recent cases did not make it into the matched case-control sample developed for impact evaluation.
Thus, the final sample has a longer follow-up period than 24 months.

30

New York site was approved to replace the original Florida site), virtually all New York
agreements had to be renegotiated or modified to include the Bronx sites and access to
individual-level treatment data for both groups was denied, jeopardizing the proposed
cost analyses. UI continued to pursue the possibility of economic analyses that would
examine, at some level, whether mental health courts are cost-effective relative to
conventional case processing; however, as detailed in Chapter 5, we were unable to
obtain sufficient data to address this issue
Cognizant that programs in most jurisdictions share the same struggles with data as the
Brooklyn and Bronx program sites, we developed guidance to aid practitioners and
program developers in collecting and using data that can support future cost-effectiveness
analyses. This information is presented in Chapter 5.
Data Analysis: Overview 
Ultimately, robust samples were constructed for both MHC evaluation sites using
propensity score matching (PSM) techniques. For the Bronx evaluation, PSM yielded a
matched sample consisting of 1,128 cases: 564 treatment cases (Bronx MHC participants)
and 564 comparison case (offenders with mental illness drawn from DOHMH records).
For Brooklyn, a matched sample of 606 cases (303 Brooklyn MHC participants and 303
comparison cases of offenders with mental illness drawn from DOHMH records) was
generated.
Impact analyses compared recidivism outcomes of MHC participants with matched
comparison groups of offenders with mental illness processed in traditional criminal
courts. For the current study, recidivism was measured as a subsequent arrest or
conviction after being admitted to MHC. For comparison cases (i.e., arrestees in jail
serviced by DOHMH), we measured whether or not an arrestee had a subsequent arrest or
conviction after the initial mental health diagnosis, after which point they would be
exposed to the risk of reoffending. The minimum length of time observed was 30
months.22
For explanatory variables (covariates), there are four broad domains examined in the
impact analysis. First, we measured the baseline demographic characteristics of
arrestees, followed by mental health conditions and drug use history, the characteristics
of instant offense for which arrestees have been referred to mental health courts or
DOHMH, and lastly number of prior offenses. Chapter 4 discusses the analytic approach
and presents findings from the impact analyses of both MHC programs.

22

Data for our analysis were not truncated based on the amount of at-risk time. Earlier cases from MHCs
provide long survival data that could enrich our understanding about the long-term effect of treatment on
re-arrest. Instead of disregarding such information, we measured and adjusted for the time at risk in
survival analysis. It should be noted, however, that the minimum follow-up of 30 months is available for all
cases in the final sample, which is reasonably long enough to observe most failure events. The implication
of not truncating data, therefore, would have little impact on our findings.

31

CHAPTER 3. BRONX AND BROOKLYN MENTAL HEALTH COURT PROGRAMS 
AND BUSINESS‐AS‐USUAL 
The Bronx and Brooklyn Mental Health Courts are two among a growing number of
mental health courts in New York State and across the country. As O’Keefe noted (2006:
3), derived from the problem-solving court model, these courts have a common set of
goals and share common elements that include developing mechanisms to assess and
identify potential participants, providing adequate clinical information to facilitate
informed decision-making, using the court’s authority to reinforce treatment goals, and
linking participants to services in the community. Yet each employs a slightly different
approach that reflects the specific program philosophy of key stakeholders, local norms,
available resources, and the unique needs of the populations they serve.
This chapter examines each of these two mental health courts individually, including
their respective structures, philosophies, logic, policies and procedures, target
populations, and services. We begin with the Bronx Mental Health Court program and
then move to the Brooklyn Mental Health Court. (Chapter 6 presents key themes and
findings across the two courts.) The final section of this chapter explores “business-asusual,” specifically how the New York City criminal justice system processes offenders
with mental illness, and the resources available to these individuals.
Bronx Mental Health Court 
The Bronx Mental Health Court (Bronx MHC) officially started taking case referrals in
2001, although the first participants were seen during a pilot project in 1999. A
committee of 41 agencies contributed to the planning of the new mental health court, and
funding was supplied by the Bureau of Justice Assistance (BJA) Mental Health Courts
Grant Program, the New York City Council, the New York State Legislature, Department
of Probation and Correctional Alternatives (DPCA), Substance Abuse and Mental Health
Services Administration (SAMHSA), the New York City Office of the Criminal Justice
Coordinator, and the National Institute on Drug Abuse (NIDA). The initiators of the
court visited other established mental health courts and held meetings with various
stakeholders, such as defense and prosecuting attorneys, judges, Treatment
Accountability for Safer Communities (TASC) members, defendants with mental illness
diagnoses, psychiatrists, and drug treatment providers. Although the court started as a
felony court, it began accepting misdemeanor cases after grant funding was acquired for
this purpose. The court was named a BJA Mental Health Court Learning Site to assist
other jurisdictions interested in adopting the mental health court model.
Administrative Structure and Staffing 
The Bronx MHC is a collaboration comprised of criminal justice personnel (judge,
defense attorneys, and prosecuting attorneys), a clinical team (clinical director, case
managers, and psychiatrists), and coordinating staff (administrative project director). The
clinical team is a separate mental health unit of the Bronx TASC office, which is operated
by the Education and Assistance Corporation, a not-for-profit organization that operates

32

multiple social service programs in New York City. Individuals from each component of
the court described positive working relationships with the other entities. Interviewees
reported substantial open communication between the prosecution, defense, and TASC,
and characterized relationships as more collaborative than adversarial. Trust was seen as
an important factor in these relationships.
The Role of the Judge 
The MHC is a specialized court housed in the Bronx Supreme Court. The same judge
has presided over the Bronx MHC since the court’s inception. This dedicated judge is
knowledgeable about the special issues related to offender with mental illness, and helps
maintain consistency in court mandates and sentencing; there is no jury. Although the
judge has expertise with this specialized population and consequentially handles cases in
a different manner than traditional criminal case processing, the judge expressed a desire
to nonetheless maintain a formal court setting and treat mental health court defendants
(also referred to as “participants” or “clients”) like defendants in other courts.
The MHC judge’s primary functions are presiding over status hearings and sentencing
both successful and unsuccessful defendants. During status hearings, the judge monitors
compliance with the treatment mandate set forth in the initial plea agreement. For this
reason, the specialized mental health court docket also is referred to as “compliance
court.” The MHC judge is rarely involved in the initial plea negotiation into the mental
health court. Initial plea agreements are typically made in each case’s court of origin and
are handled by the judges presiding over those other courts.
The Role of the Prosecution 
All prosecuting attorneys involved in the Bronx MHC are based in the Narcotics Bureau
of the Bronx District Attorney’s (DA’s) Office. Approximately ten prosecuting attorneys
rotate to handle all Bronx MHC cases. The Chief of the Narcotics Bureau supervises all
prosecutors and makes all final prosecution and program decisions. Most of the attorneys
spend about 15 percent of their time on MHC cases, while the Deputy Chiefs and Chief
of the Narcotics Bureau spend substantially larger portions of time on these cases. When
a case is referred to the MHC from another division of the DA’s Office, the original
prosecutor may remain on the case, along with the assigned prosecutor from the
Narcotics Bureau. However, the original prosecutor rarely attends court, and conducts
few prosecutorial functions for the case. The DA’s Office has designated staff to
maintain records on substance abuse and mental health cases.
Prosecutors play a large role in the Bronx MHC, particularly in entry decisions.
Defendants in Narcotics Bureau cases are routinely screened using a seven-item
questionnaire to determine potential eligibility for the MHC and other alternatives to
incarceration. Prosecutors also make discretionary decisions on suggested sanctions for
client noncompliance, final sentencing offered after graduation, and how charges will be
handled (e.g., charges dismissed or reduced). Prosecutors have weekly meetings with
TASC to discuss defendants’ problems with compliance. If there are no particularly

33

pressing issues, prosecutors learn about the defendants’ progress through TASC’s
reporting of client progress in court.
The Role of the Defense 
Bronx MHC participants are typically represented by attorneys from indigent defense
firms, such as Legal Aid and Bronx Defenders, the 18B Assigned Counsel Plan (a local
program that contracts with private attorneys to represent indigent defendants), and
occasionally private defense attorneys. Defense attorneys are active in the Bronx MHC
process in the following ways: referring clients to the mental health court program,
explaining MHC policies and other alternatives to their clients, and serving as advocates
for clients during the monitoring period. Defense attorneys do not participate in
screening; however, they may try to negotiate eligibility with the DA’s office in certain
cases. Defense attorneys rarely attend court for routine compliance monitoring
appearances; however, they are present for court appearances that involve pleas, client
problem behavior, negotiations with prosecutors, graduations, and sentencing. While
defense attorneys do not have regularly scheduled meetings with TASC or the
prosecution, they are free to meet with either on an as-needed basis. Many of the defense
attorneys involved in the Bronx MHC receive specialized training, such as monthly
meetings about mental health issues where speakers are invited to talk and provide
training on relevant laws and policies.
The Role of the Clinical Team 
The Bronx TASC clinical team is a neutral and separate entity from the court, providing
case management services for all participants within the Bronx MHC program; a separate
TASC team handles drug court cases. At the time of this study’s fieldwork in 2009, the
MHC TASC team was composed of one clinical psychologist, two part-time consulting
psychiatrists, a supervising case manager, and 12 case managers.23 The TASC case
managers have a variety of educational backgrounds, including both bachelor- and
masters-level degrees in psychology and social work, and other degree programs
unrelated to social sciences. The primary goal of the clinical team is to engage clients in
treatment as opposed to symptom elimination, which stakeholders felt may be unrealistic
given the diagnosis and prognosis of some of the clients, and beyond the scope of what
the court could legally or ethically require.
TASC’s Clinical Director is a clinical psychologist who serves as a supervisor, performs
psychiatric evaluations, and makes final clinical eligibility decisions. The Clinical
Director and consulting psychiatrists perform comprehensive evaluations, including
psychiatric interviews, mental status exams, administration of risk assessment
23

At the time, the MHC also employed an Administrative Director and a Data Coordinator. As of June
2011, the program’s funding levels were reduced and staffing levels were lower than in 2009. The 2011
program had the following full-time positions: one clinical psychologist, one project coordinator, and six
case managers. Part-time staff included three doctoral-level psychology interns, one forensic psychiatry
fellow, and a supervising forensic psychiatrist. Three to four interns with master’s degrees assisted during
the year. Lastly, one vocational specialist and one peer specialist, each from other agencies, worked with
MHC clients.

34

instruments, and review of collateral information sources (e.g., jail medical records).
Evaluations typically last approximately one hour and are performed at the TASC office
or in the court’s holding cells. Once an evaluation is complete, the Clinical Director or
psychiatrist produces a report that describes the social history, medical and psychiatric
history, current clinical presentation of symptoms and diagnoses, assessment of risk for
danger, and recommendations on diversion decisions and types of treatment. While
TASC shares these evaluation reports with the court, neither the Clinical Director, nor
consulting psychiatrists testify in court. Once a client has undergone the biopsychosocial
assessment by a case manager and a psychiatric evaluation by the Clinical Director or a
consulting psychiatrist, the Clinical Director makes a decision on whether the client is
clinically eligible and would benefit from the Bronx MHC program. As a supervisor, the
Clinical Director also solves problems with the case management team on their current
cases and offers guidance for difficult situations such as the decompensation of clients,
difficulties with treatment providers, and ineffective medications.
The case managers are split into two types of roles: pre-placement and post-placement.
The pre-placement team serves as case managers for all Bronx MHC participants prior to
their placement in treatment. In this role, their primary tasks are to conduct psychosocial
assessments of potential participants and match accepted clients with appropriate
treatment services. Case managers typically perform two to three assessment interviews
daily. These interviews usually occur at the TASC office or inside the holding cells at
the courthouse. Potential participants then either are referred for comprehensive
psychiatric evaluations or denied entry into the Bronx MHC. Once a client has
completed the evaluation process and formally entered the Bronx MHC (by entering a
guilty plea with deferred sentencing), s/he meets with the pre-placement case manager on
a weekly basis (if they are in the community) or on court dates (if detained). At the time
of study interviews in 2009, pre-placement case managers typically had caseloads of 15
to 25 clients; recent staffing cuts have increased workloads.
In addition, three pre-placement case managers also serve as court liaisons. As court
liaisons, the case managers attend court every week to report on the progress of all the
Bronx MHC participants, including those under supervision by the post-placement team.
Interviewed case managers emphasized the importance of developing rapport with
multiple attorneys and judges (since court liaisons are involved in the pre-plea process
that occurs in the original courts).
The post-placement team provides case management services to clients who have been
matched to a treatment provider. Pre-placement case managers exchange information
about a client transitioning into the post-placement caseload through an intake memo that
summarizes information about the participant and the assigned treatment. After a
participant transitions to a post-placement case manager, they meet on a weekly basis.
Participants may be required to meet more or less often depending on their clinical status
and progress within the Bronx MHC program. TASC uses consistent appointment times
for participants, and working clients are allowed to schedule appointments for the end of
the work day. Case managers visit participants in residential treatment once or twice per
month.

35

Meetings with participants typically last about 15 minutes, although they can take up to
40 minutes if a person is in crisis. During these meetings, case managers obtain a general
status reports about participants’ treatment progress and general well-being. Case
managers also teach participants skills to work with their treatment providers and
advocate for themselves. While participants may try to use a case manager as a therapist,
the case manager’s primary role is to support successful treatment by outside agencies
and serve as a mediator between clients and treatment providers. Apart from the
traditional case management model, one case manager also runs a self-help program, and
two case managers provided an illness management psychoeducational group program in
the past. A vocational counselor also provides assistance with employment pursuits.
The TASC team may use various interventions to address individual behavior apart from
those that are court-ordered. Examples of these include verbal reinforcement of behavior,
more or less frequent TASC visits, clinical interventions, or increased treatment intensity
(e.g., more frequent treatment attendance, changing from outpatient to inpatient
treatment, or required participation in drug rehabilitation).
Another important part of the post-placement team’s job is to coordinate and collaborate
with treatment providers. Case managers’ interactions with treatment providers vary
from once per week to once per month depending on a participant’s situation. In
addition, case managers receive written progress letters from treatment providers on a
regular basis. Some interviewed members from the post-placement team stated treatment
providers, in some ways, also were considered clients.
The pre- and post-placement teams have weekly meetings to review their caseloads with
a supervisor. While the pre-placement team is focused on sharing evaluation results and
brainstorming possible treatment placements, the post-placement meetings revolve
around problem-solving issues about compliance, conflicts with treatment providers, and
other needs (e.g., housing, employment). The two teams also meet with each other
weekly to allow the court liaisons from the pre-placement team to receive information
about participant progress from the post-placement case managers. Interviewed case
managers liked the split structure and felt that it would be difficult for one case manager
to fulfill all the responsibilities of assessment, placement location, court liaison duties,
and progress monitoring for each participant in his or her caseload.
Case managers receive formal training on court liaison responsibilities and in-service
education about new laws and medications. However, most training occurs informally
through the job and from hearing about other case managers’ clients during meetings.
The Role of Treatment Providers 
Mandatory treatment is provided to participants through community-based treatment
organizations. These organizations work directly with the participants to provide mental
health services, while also remaining in regular contact with TASC case management
staff. The Bronx MHC uses a variety of different types of treatment providers, including
therapeutic communities, outpatient drug and alcohol treatment providers, inpatient

36

rehabilitation programs, detoxification programs, and temporary housing. Many of the
provider types recorded in the program’s database were substance abuse treatment
services.24 Additionally, participants are placed into therapeutic communities, and
sometimes with private therapists or psychiatrists. Due to a lack of available services in
the Bronx borough, clients may be matched to treatment providers in other parts of the
NYC metropolitan area.
Program Structure and Target Population 
Figure 3.1 illustrates how cases proceed through the Bronx MHC. In general, defendants
are referred to the program, screened for eligibility, enter the court through a formal plea
process, are matched with community-based treatment, and then participate in court
monitoring, case management, and treatment services. The duration of participation can
vary based on charge and mental illness characteristics. There is a minimum six-month
treatment mandate for misdemeanor crimes, while treatment mandates for felony crimes
typically last 18 to 24 months. The mandated length of treatment begins upon entry into
a treatment program, rather than the plea date. Since it can often take a significant
amount of time to find an appropriate and available treatment program, participants may
be under court supervision for longer periods of time than the treatment mandate. Each
stage is discussed in more detail, below.
Target Population 
The Bronx MHC was intended for felony and misdemeanor offenders (primarily
nonviolent) with serious mental illness. MHC personnel reported a mix of crime types
within the program, including both violent and nonviolent crimes at both felony and
misdemeanor levels. Stakeholders reported that very few participants were gang
members. Data obtained from the Bronx MHC database showed that 648 participants
were referred to the mental health court during the 2002 to 2006 study period. 25 As
shown in Table 3.1, 93 percent of these individuals were indicted on felony offenses.
Drug selling was the most common offense charged; substantially smaller numbers of
participants were charged with drug possession, assault, burglary, robbery, weapon, and
other offenses. MHC participants often had extensive criminal histories, as recorded by
program personnel from participants’ rap sheets. Most (86 percent) had prior arrests, and
those who had been arrested before averaged 9.3 lifetime arrests. More than threequarters (78 percent) had prior convictions, with an average of 8.7 lifetime convictions.
Many participants were concurrently involved in other cases: 60 percent had other current
open cases, 8 percent were on parole, and 5 percent were on probation.

24

This is because the MHC evolved from the Bronx’s drug court in response to the volume of offenders
with mental illness seen in drug court. The program initially adopted the drug court database to record
information on MHC cases, and used this database to provide information on 2002-2006 cases for this
research. The MHC has since adopted a new case tracking database.
25
The analysis operationally defined Bronx MHC participants as those who had a plea date recorded in the
program database.

37

Figurre 3.1. Casee Flow Diagrram of Bron
nx Mental Health Courrt

38

Table 3.1. Current Offenses and Criminal History of Bronx MHC Participants
Current Offenses and Criminal History Among Bronx MHC Participants (N=648)
Current offense
Felony
93.0%
Misdemeanor
7.0%
Incarcerated at the time of screening
58.8%
Concurrent cases
Had additional indictment charge
Current open cases
On parole at time of initial screening
On probation at time of initial screening

13.0%
60.3%
7.6%
4.8%

Criminal history
Any prior arrests
If yes, mean number of prior arrests
Any prior convictions
If yes, mean number of prior convictions
Prior Felony Offender (PFO)

85.7%
9.3
78.1%
8.7
40.3%

The court was intended to serve individuals with serious mental illnesses. However, the
program database in use during the 2002-2006 study period did not record client
diagnoses.26 These were maintained in paper files and other systems inaccessible to the
research team. As presented in Table 3.2, data obtained on a subset of 153 MHC
participants who were incarcerated in 2005-200627 suggest that Bronx MHC participants
typically were seriously and persistently mentally ill (SPMI). Nearly all (96 percent) had
received psychiatric medication while in jail, and about one in ten had been housed in a
mental observation unit (MOU). Mood disorders such as depression and bipolar disorder
were the most common Axis I diagnosis (40 percent), followed by substance-related
disorders (29 percent) and psychotic disorders such as schizophrenia (23 percent). Twothirds of Bronx MHC participants had co-occurring substance use and mental health
disorders, defined by the research team as a substance-related diagnosis on either Axis I
or Axis II.

26

Tracking forms developed by the program since 2006 record diagnostic information. Additionally, the
Bronx MHC was in the process of adopting a new and improved program database in 2010.
27
These data were obtained from the NYC DOHMH, which oversees the provision of health services to
inmates incarcerated by the NYC DOC. These data are described in Chapter 2.

39

Table 3.2. Mental Health Indicators Among a Subset of Bronx MHC Participants
who Received Jail-Based Mental Health Services from DOHMH, 2005-2006
Mental Health Indicators Among a Subset of Bronx MHC Participants who Received JailBased Mental Health Services from DOHMH, 2005-2006 (N=153)
Received medication for mental illness
96%
Housed on a mental observation unit
13%
Assessed as SPMI (Serious and Persistent Mental Illness)
75%
Mean Global Assessment of Functioning (GAF) score, on a scale of 1 (low) to
100 (high)
54
Axis I Diagnoses
Adjustment disorder
2%
Mood disorder
40%
Psychotic disorder
23%
Substance-related disorder on Axis I
29%
Axis II Diagnoses
Adjustment disorder
0%
Mood disorder
19%
Psychotic disorder
5%
Substance-related disorder on Axis II
66%
Substance use reported on either Axis I or Axis II
66%
Source: Urban Institute analysis of DOHMH data.

Bronx MHC program staff believed most clients to be dually diagnosed with substance
abuse problems, and analysis of the Bronx program data suggests a high level of cooccurring disorders among participants. As Table 3.3 demonstrates, almost all
participants in the Bronx MHC database (99 percent) reported either current or past drug
use. More than three-quarters (77 percent) had a history of alcohol abuse, while 73
percent reported a history of marijuana use. A history of hard drug use was prevalent
among Bronx participants, including cocaine (68 percent reported previous use); heroin
(59 percent), and crack cocaine (57 percent). Nearly 90 percent of participants (577 of
648 participants) reported a current primary drug at the time of screening into the MHC:
41 percent reported heroin as their primary drug; 22 percent reported crack cocaine; 15
percent marijuana; 10 percent reported cocaine, and 8 percent reported alcohol. Nearly
three-quarters (73 percent) reported prior treatment.
Demographic Profile 
Key demographic and socioeconomic characteristics of Bronx MHC participants are
presented in Table 3.4. Bronx MHC participants were, on average, 37 years old at the
time of the arrest leading to their mental health court case. The majority (62 percent)
were male. Most were from racial and ethnic minority groups: more than half (58
percent) were Hispanic and one-third (34 percent) were Black. (Note that the program
data did not distinguish between race and Hispanic ethnicity.)

40

Table 3.3. Self-Reported Substance Abuse by Bronx MHC Participants
Substance Abuse Reported by Bronx MHC Participants (N=648)
Any current or past drug use
Substance abuse history
Alcohol
Cocaine
Crack
Heroin
Marijuana
Other drugs
Primary drug reported at initial screening
Alcohol
Cocaine
Crack
Heroin
Marijuana
Age at first use
Age that client started taking primary drug
Age that client started taking secondary drug
Prior substance abuse treatment
Any prior treatment
Number of prior treatment contacts (mean)

98.8%
76.7%
68.4%
57.1%
59.3%
73.2%
43.5%
8.0%
10.2%
21.7%
41.2%
15.1%
20.9 years
19.7 years
73.2%
1.6

Participants in the Bronx MHC were typically socioeconomically disadvantaged. Nearly
all (93 percent) were unemployed, and relied on indigent defense counsel for legal
representation (94 percent). Nearly two-thirds of participants had less than a high school
education.
Most (88 percent) Bronx MHC participants lived in that borough; others reported a
residence elsewhere in New York City, and a small portion were from suburban New
York localities and from out–of-state. Bronx MHC staff reported that many participants
lacked vital family support and lived with negative influences, such as drug-using
cohabitants. Although relatively few clients (0.3 percent) were explicitly recorded as
being homeless in the program’s database, Bronx MHC program personnel estimated that
30 percent had clinical or shelter needs for housing. Staff reported that many participants
lacked stable housing, and were functionally homeless (rotated among various people’s
couches), but were not residing in shelters or on the street. This hampered individuals’
abilities to qualify for homeless assistance from the state because they technically had
alternative housing.
Veteran and HIV status were other client characteristics that could affect service delivery.
Three percent of MHC participants reported being veterans; these individuals presumably

41

were eligible for veterans’ benefits. About one in ten Bronx MHC participants were
either HIV positive or had AIDS, for which certain programs may reserve treatment slots.
Table 3.4. Demographic, Socioeconomic, and Health Characteristics of Bronx MHC
Participants
Characteristics of Bronx Mental Health Court Participants (N=648)
Mean age at arrest
Male
White, non-Hispanic
Black, non-Hispanic
Hispanic
Other race or ethnicity
Bronx resident
High school diploma or GED
Unemployed
Representation by indigent defense counsel
Veteran status
HIV Positive or Has AIDS

37.3 years
62.3%
7.3%
33.8%
58.0%
0.9%
88.2%
34.8%
92.8%
94.4%
2.5%
11.1%

Eligibility Criteria 
Defendants are eligible to participate in the MHC if they meet both clinical and legal
criteria. Potential participants must have mental health problems in order to participate.
While clients do not need to have a “severe and persistent mental illness” designation
(major Axis I disorder or substantial history of hospitalization or poor functioning), they
must have mental health problems that cannot be handled adequately in other traditional
or alternative justice venues. This often means that participants have DSM Axis I
disorders. Unlike many other mental health courts in the country, the Bronx MHC does
not exclude personality disorders if they believe the defendant can be helped through
available treatment resources. Clinicians from TASC also consider risk for future
violence in their decision-making.
On the other hand, the mental health court does not accept defendants who are unstable or
need hospitalization. If a defendant is unstable or incompetent to stand trial, the court
cannot be sure the potential participant is able to make an informed decision about
participating in the program. TASC also is not confident in its ability to secure consistent
hospital treatment due to lack of space and the defined treatment mandate (i.e., hospital
treatment can often be indefinite, which is beyond the scope of the court). TASC also will
not accept individuals if it does not believe suitable treatment can be secured.
The court accepts both felony and misdemeanor charges, excluding charges of murder,
sex offenses, and arson. Unlike the Brooklyn MHC, the Bronx court does not require that
the conviction offense be related to the individual’s mental illness. The court takes this
stance, because they feel that it is difficult to make a confident determination on this

42

matter. However, Bronx MHC staff takes into account whether treatment will reduce the
risk of violence and crime in the future.
Referral 
Cases are typically identified for referral between arraignment and the plea or trial.
Cases can be referred to the mental health court through various sources, including
prosecutors, defense attorneys, judges, family members, community providers, jail
mental health staff, probation officers, “730” competency hearings, other case
management or diversion programs, or by the defendant. Many referrals originate in the
Narcotics Bureau of the DA’s office or through the traditional TASC team that deals with
cases for substance abuse. TASC obtains approval from the DA’s Office before TASC
evaluates a client. For example, if a defense attorney suspects a mental health issue with a
client, s/he calls the Narcotics Bureau and describes the client and case. If the defense
attorney obtains approval from the DA’s Office, then either the defense or DA contacts
TASC to set up further evaluation.
The DA’s office was the chief source of referrals to the Bronx MHC, as illustrated by
Table 3.5. Mental health court participants between 2002 and 2006 were most often (46
percent) referred by the DA’s office. A sizeable number (40 percent) were referred to the
MHC by other programs. In particular, some participants who initially enrolled in the
Bronx “regular TASC” drug court program were later found to need mental health
services and so were subsequently transferred to the Bronx MHC program; these cases
were initially referred to the drug court by the DA’s office. Taking both direct referrals to
the mental health court and transfers from drug court into account, the DA’s office
initiated 76 percent of the MHC’s participant caseload (not shown in the table).
Table 3.5. Referral Sources for Bronx MHC Participants
Referral Sources for Bronx MHC Participants (N=648)
District Attorney
Drug Court, Other Program, or Self-Referral
Case transferred to MHC from drug court
Judge or Court
Defense Attorney*
Incarcerated at the time of MHC screening

45.7%
39.7%
37.4%
13.5%
1.1%
58.8%

* Defense referrals are underestimated in the program’s administrative database. Defense referrals are
often made directly to the prosecutor, who then considers the case and refers the case to TASC; these may
be recorded as prosecutor referrals, although the impetus came from the defense bar.

Roughly six in ten MHC participants had been detained at the time of referral. The time
from arrest to mental health court referral was typically five months (see Table 3.6). 28
The median time to case referral was 100 days when cases were referred directly to the
Bronx MHC program. When cases had been transferred from the drug court, the time
28

Medians are reported here because the means are skewed by some outliers with atypical case-processing
statistics.

43

from arrest to mental health screening was considerably longer (a median of 221 days) as
these clients spent 78 days (median) in the drug court first.
Table 3.6. Time from Arrest to Bronx MHC Participation
Time from Arrest to Bronx MHC Participation (N=648)
All Participants
Direct Referrals
(N=648)
(N=406)
Mean Median
Mean Median
Days from arrest to initial
TASC screening
266.0
148.5
189.4
101.0

Drug Court
Transfers (n=242)
Mean Median
397.1

221.0

Days from MH screening
to MHC participation

38.1

16.0

60.8

49.0

0.0

0.0

Days in drug court before
transfer to MHC

n/a

n/a

n/a

n/a

226.6

77.5

Days from arrest to MHC
participation

303.1

189.0

248.6

174.0

397.1

221.0

Screening 
All cases entering the DA’s Office are informally screened during intake, and any cases
with suspected mental health issues are directed to the Chief of the Narcotics Bureau for
further screening. Defendants in narcotics cases and cases directed from intake to the
Narcotics Bureau are screened with a seven-question form developed by the Bronx MHC
(see Appendix E). If there is a positive screening, the prosecutor informs the defense
attorney that the defendant may qualify for a special program.
The supervising case manager processes all referrals and assigns a pre-placement case
manager to handle the assessment process. Before the assessment can occur, case
managers need to receive medical files from Rikers Island to obtain a complete picture of
the individual’s mental health and to ensure they are on appropriate medication.
Defendants are then screened for clinical eligibility by TASC case managers who
administer a battery of measures, including a biopsychosocial assessment, clinical
interview, standardized risk assessment tools, and measures of mental health, substance
use, health, social functioning, and criminal behavior.
Specifically, TASC case managers perform a biopsychosocial assessment, which covers
areas such as demographics, family history, medical history, psychiatric history, receipt
of current services, legal issues, and risk indicators. Case managers obtain collateral
information through Rikers medical records, conversations with the initial prosecutor of
the case, and other sources, as needed. Case managers do not use a specific standardized
instrument for this assessment; however, they have guidance on which topics to cover.

44

Once this assessment is complete, the supervising case manager or Clinical Director
makes a decision on whether to proceed with an evaluation performed by a consulting
psychiatrist or the Clinical Director. The full psychiatric evaluation examines client
history, symptoms and diagnoses, risk for future violence, and issues of compliance and
appropriate treatment. The evaluation report is shared with the court, and eventually, the
treatment provider.
Entrance into the Mental Health Court 
The MHC judge makes the final approval decision based on recommendations by the
TASC team, the prosecuting attorney, and the defense attorney. The TASC team makes a
recommendation based on their clinical evaluation. Defendants cannot join the program if
they are incompetent to stand trial or if they are currently unstable. If an individual needs
to be stabilized, s/he will be sent to Bellevue Hospital (males) or Elmhurst Hospital
(females) or, if these are not available, return to Rikers Island for treatment. Language
barriers and lack of insurance do not exclude someone from participating in the Bronx
MHC. However, clients who are clinically eligible and accepted into the program may
ultimately be unable to participate if there is no available treatment that is appropriate for
the particular situation of client (e.g., Spanish-speaking provider).
The DA has discretion to refuse a case. Prosecutors take into account the complaining
witnesses’ opinion in the decision to allow a defendant to participate in the mental health
court. Reportedly, victims of crime rarely protest entry into the Bronx MHC program.
However, prosecutors have occasionally overridden a complaining witness’ request to
refuse the mental health court option. Defense attorneys may choose to have a case
dismissed rather than go through the court system if they feel the charges are unjustified.
The defendant must voluntarily agree to participate and make a guilty plea (at a higher
charge than the individual would receive otherwise) in order to be a part of the program.
This process occurs in the courtroom of the original judge for the case and is publicly
documented. As depicted in Table 3.7, most (91 percent) pleas to the mental health court
were to felony offenses, and class B felonies in particular (67 percent). Drug selling was
the prevailing charge in the majority of cases (72 percent), followed by drug possession
(8 percent). Assaults, robberies, burglaries, and weapon offenses also were observed.
More often than not, the plea charge carried one year or more of incarceration time.
A defendant makes the decision to enter the MHC program without knowing what the
required treatment will be. Program data from 2002 to 2006 showed that 91 percent of
treatment placements occurred after the plea date. However, staff reported that most
individuals would know at this point whether the recommended treatment is inpatient or
outpatient.
An individual may choose not to participate for various reasons (e.g., a person may not
want to be labeled mentally ill or s/he may try to receive a better deal in traditional court,
rather than be monitored for a lengthy amount of time and have mandatory treatment).
Defendants who decline participation or are denied access to the Bronx MHC may have
other diversion options, including the Mentally Ill/Chemically Addicted (MICA)

45

Table 3.7. Plea Charges and Sentences at Entry to the Bronx MHC29
Plea Charges and Sentences at Entry to the Bronx MHC (N=648)
Charge level of plea into mental health court
Felony
Class A
Class B
Class C
Class D
Class E
Unspecified class
Misdemeanor
Class A
Class B
Unspecified class
Violation

90.91%
0.16%
66.93%
9.56%
11.76%
2.04%
0.47%
8.93%
8.15%
0.63%
0.16%
0.16%

Charge on entry to the mental health court
Assault
Burglary
Drug possession
Drug selling
Larceny
Robbery
Weapon
Unclassified/other

3.26%
2.33%
8.23%
72.36%
2.48%
3.88%
0.47%
6.99%

Multiple plea charges on entry to the mental health court
Sentence Length, If Incarcerated on Original Plea
Probation or conditional discharge
Days
Months
1 or more years
Unspecified in database

40.9%
0.44%
5.96%
2.65%
55.41%
35.54%

program30 or drug court, if applicable. Furthermore, they remain eligible for any
business-as-usual discharge planning services provided through Rikers Island and its

29

New York State felony and misdemeanor offenses are further classified according to the level of
seriousness. Class A offenses are the most serious, carrying the longest sentences.
30
The MICA Project is a federally-funded program through BJA. MICA functions differently in various
jurisdictions, but services in the Bronx include a specially trained social worker and assistant paired with a
defense attorney to help address particular mental health issues, traumatic brain injury, or mental
retardation. Clients can be involved in both the MICA and Bronx MHC programs.

46

community-based LINK and SPAN program partners.31 Appendix F provides the results
from an analysis comparing Bronx MHC participants and nonparticipants; the analysis
examines the extent to which the two groups differ according to demographics, criminal
justice indicators (prior and current involvement), mental health status, and reason for
nonparticipation.
Court Proceedings 
While the guilty plea that initiates entry into the Bronx MHC program can occur in any
courtroom, all participants continue any future court proceedings within the Bronx
MHC’s courtroom. Participants are required to have quarterly status hearings; these may
be conducted more often in individual cases, if deemed necessary by the judge. The
Bronx MHC is managed within a mixed docket called “Compliance Court.” Other cases
seen within this docket are TASC substance abuse cases. Court sessions occur on
Tuesday and Thursday mornings.
There are various individuals involved in the court hearings. The judge, stenographer,
and clerk sit at the front of the court room. The prosecutors and court liaison case
managers sit at one table, while the defense attorney and client are at a second table. On
occasion, there are other participants such as probation or parole officers. No jury is
present. Cases generally last a few minutes, and the presiding judge chooses to manage
the court in a formal manner in order to maintain a more typical court atmosphere.
The research team observed the mental health court in session during May 2009. This
session, which staff described as typical, lasted between three and four hours. There were
46 scheduled court appearances, each lasting between one and two minutes. Researchers
observed from the jury box, taking notes on the court environment, communication
patterns, and judicial responses to both participants’ compliance and noncompliance with
court conditions.
Each court session included a variety of types of court appearances. The most common
type of appearance was a status hearing, as shown in Table 3.8. During status hearings,
the court reviewed the progress and compliance of MHC participants. A small portion of
hearings were sentencings, both successful and unsuccessful. Other types of hearings
included cases from other dockets (e.g., drug court cases), non-appearances (including
people medically unable to come, those in custody and not produced, and a few failures
to appear). One-third of the scheduled hearings were actually no-shows. Some of these
no-shows were noncompliant, whereas other offenders did not appear because they were
in detention awaiting a treatment placement. With regard to other dockets, about onequarter of the Bronx docket was drug court cases.
We observed very little, if any, orchestration of the courtroom environment in the Bronx
MHC. (The drug court literature suggests that some judges in specialized courts see a
31

Assuming they had been incarcerated pretrial and determined to be eligible for Brad H services through
the jail-based screening and assessment process. This is not applicable for other referrals who were never
incarcerated pretrial.

47

value in setting up courtroom community by requiring defendants to stay for the entire
session and ordering cases in a strategic way.) Defendants were not required to remain
for the duration of the session. Although defendants were required to arrive at the start of
the session at 10 o'clock, they could leave once their appearance was finished. There was
also no deliberate ordering of cases. Rather, the logistics of bringing defendants up to the
courtroom from the holding cells on a lower level of the building sometimes dictated
which defendants appeared because of elevator capacity constraints. There was no
assigned seating, and microphones were not used.
Table 3.8. Types of Bronx MHC Hearings Observed
Types of Bronx MHC Hearings Observed (N=46)
Status hearings
Sentencing (successful or unsuccessful)
Non-MHC
No show
Other or unknown

28%
9%
28%
33%
2%

The main courtroom actors were: the judge, the defendant, a defense attorney, a
prosecuting attorney, and the case management representative. Defense attorneys did not
appear with the defendant during routine status hearings, unless it was a particularly
significant appearance, such as program completion or if some noncompliance had been
noted. We also noted that the case management representative in court was typically not
the defendant’s assigned case manager, but instead a representative from the case
management team. In the Bronx, one person from the clinical team was assigned to cover
the court session and provide updates on the all of the cases.
Table 3.9 shows who, other than the judge, spoke at status hearings. The Bronx judge
placed a high value on maintaining a formal court atmosphere. In this court, the case
manager was typically addressed first by the judge, and asked to give a report on the
individual’s status. The judge typically did not speak with defendants during hearings,
but when he did, he spoke directly to them, not through their attorneys, addressed their
compliance with court requirements, and imparted instructions to the defendants.
Defendants did not speak much during these interactions. When they spoke, they
typically provided “yes,” “no,” or other similarly brief responses to closed-ended
questions from the judge. We also observed the use of third person language when the
judge addressed defendants. For example, when defendants completed the program
successfully, the judge often said "the court congratulates you," as opposed to personally
congratulating the defendants.
Much of the judge-participant exchange concerned monitoring compliance with court
requirements. The case management team is responsible for monitoring compliance
between scheduled court appearances (through a combination of client meetings,
urinalysis, and provider reports on clients) and circulates written progress reports to the
judge, prosecutor, and defense attorneys in advance of courtroom sessions. The case
manager in court usually gave a verbal status report on client compliance during court,

48

and we observed the judge referring to the written reports in court. Defendants rarely
spoke during these appearances, with the exception of brief responses to closed-ended
questions from the judge.
Table 3.9. Courtroom Participation During Bronx MHC Hearings
Courtroom Participation During Bronx MHC Hearings (N=46)
Other than the judge, who spoke at status hearings?
Case manager
Defense attorney
Defendant
Prosecutor
Other

69%
39%
23%
15%
15%

Judicial interactions with defendant:
Defendant addressed first
Defendant approached bench

8%
0%

The most frequent participant in Bronx MHC hearings was the case management
representative. One of the representative’s roles was to present information to the court.
The case management team prepared written progress reports in advance of the court
session and often gave a brief verbal report on the defendant status. We also observed
case managers functioning in a problem-solving role during court (e.g., troubleshooting
problems with a service provider, or dealing with court requirements during a bench
conference).
Attorneys participated less than half the time during status hearings. When attorneys
participated in status hearings, it was usually because there was some issue. In the Bronx,
defense attorneys did not appear at routine status hearings unless some noncompliance
was being discussed. Similarly, prosecuting attorneys rarely spoke during the status
hearings. However, the Bronx DA’s Office had a regularly scheduled conference with the
case management team before court to discuss the caseload. Stakeholders interviewed for
the study described a generally collaborative, rather than adversarial relationship between
prosecutors and defense attorneys once a case had been accepted to the MHC.
The majority of cases we observed were compliant. The judge typically responded to
compliant cases with praise. The Bronx court did not use any other rewards such as phase
or progress certificates. This stems from a philosophical decision that such rewards come
from a drug court tradition, and that it is not appropriate to mark treatment progress in a
mental health court setting because the court’s objective is not to cure someone of their
mental illness. Occasionally, the judge decreased court appearances in compliant cases,
but this was not done explicitly as a reward for compliance. Rather, a decrease in court
appearances or postponement was granted in response to a specific need of the defendant.
We observed that sometimes cases were simply continued without overt praise: this

49

happened in three of the seven status appearances we observed where defendants were
compliant.
We observed some defendants’ final appearances upon program completion. In the
Bronx, program completions were referred to as sentencings whether they were
successful or unsuccessful. In fact, as an observer, it took some time to figure out whether
a given sentencing represented a successful end of court supervision or an unsuccessful
one. Sentencing hearings were longer than other appearances, between five and ten
minutes. During the appearance, the old plea was withdrawn and a new plea was entered,
along with the terms of the new sentence being given by the court. Following this, the
defendant was given an opportunity to speak. This was the defendant's statement at
sentencing, and in the Bronx, was usually the only time when defendants spoke beyond
giving yes or no answers. These statements usually involved thanking the court for the
opportunity, or remarking on the difference that the court made in their lives.
The following types of noncompliance were observed in other hearings: treatment
absences or nonparticipation, positive drug tests, and failure to appear at court. All stated
noncompliance was addressed by the judge; this was typically verbal recognition by the
judge. The judge commented on the noncompliance, issued instructions or advice to the
defendant, and explained the consequences of not complying with the court's conditions.
Occasionally, we observed that the judge increased appearances to either court or to the
case management agency. Additionally, we witnessed defendants who were remanded to
jail. (More often, we saw defendants who had been remanded to jail on their previous
appearance, and were being seen again by the judge who was evaluating whether or not
to release them on recognizance at this time.) Warrants were issued if the defendants
were no-shows without a reason. In a small number of cases, we saw defendants who
failed the MHC program and were being sentenced for the original offense.
The Bronx MHC did not apply a fixed sanctioning algorithm for dealing with
noncompliance. During our stakeholder interviews, we were told that noncompliance was
dealt with on a case-by-case basis, and that frequent "second chances" were given
depending on the context of the situation. Remand is used as a last resort. Changing the
frequency of case management meetings or treatment is a more common tool for
modifying behavior.
Program data from 2002-2006 corroborate these observations, as shown in Table 3.10.
Warrants were issued in one-third (35 percent) of cases, and one-fifth (19 percent) of
individuals who ultimately completed the program successfully had been “warranted”
previously as an intermediate sanction. (Violation of conditions letters were another
means of addressing noncompliance. These were tracked in the program database
between 2002 and 2004, and were observed among successful program completers as
well.)
Clinical Treatment and Case Management 
Once clients are accepted into the Bronx MHC, the TASC pre-placement clinical team
assists participants with applications for public assistance (e.g., Medicaid) and attempts to

50

match the client with appropriate treatment. This process can often be difficult as
treatment providers may not have available slots or there may not be appropriate
treatment options (e.g., treatment for Spanish speakers). Stakeholders stated they
sometimes have to “fit square pegs into round holes” to ensure participants can receive at
least some type of treatment. About half of the participants require residential treatment;
they must be substance-free for a certain period of time before residential placements will
accept them. Stakeholders reported that it usually takes four to six weeks to secure a
placement for a client. The TASC team does not exclude facilities outside of the Bronx
borough and will match participants to treatment providers anywhere in the metropolitan
area. Roughly six in ten Bronx MHC defendants are already in custody when they are
diverted to the mental health court. If participants are detained at the time of acceptance,
they remain incarcerated until an appropriate placement is found.
Table 3.10. Intermediate Sanctions Issued in the Bronx MHC
Intermediate Sanctions Issued in the Bronx MHC (N=648)
Warrant ever issued during MHC case
Time from MHC enrollment to warrant (mean)
Time from MHC enrollment to warrant (median)
Time from warrant to administrative case termination (mean)
Time from warrant to administrative case termination (median)

35.35%
250.6 days
188 days
193.9 days
106.5 days
5.86%

Violation of Conditions letter ever presented to court during MHC case (20022004 only)
Time from MHC enrollment to violation of conditions letter (mean)
Time from MHC enrollment to violation of conditions letter (median)

189.2 days
146 days

According to program data from 2002-2006, most (81 percent), but not all of the clients
who pled into the Bronx MHC were successfully placed into treatment;32 one-fifth were
not successfully enrolled into a treatment program. Examining the program database, we
found that the most common reasons were client failure before intake, disqualification
from lack of a true mental illness, and client involvement in other open cases. More than
one-fifth of the clients who did not enroll into treatment failed to report for program
intake. One-quarter of these clients were determined to not have an underlying mental
illness after having pled into the program. Staff reported that this was possible even after
the initial psychiatric examination because symptoms of mental illness are sometimes
substance induced; symptoms of mental illness may clear once clients have abstained
from drugs or alcohol for a period of time, such as under court-monitoring or jail-based
detox. The remainder of cases not placed into treatment were for a variety of reasons,
including open cases in other jurisdictions, a lack of appropriate treatment in the
community for a given client (e.g., due to language barriers, or the severity of a client’s
illness), and client withdrawal from the program.

32

Defined by the research team as having a treatment intake date recorded in the program database.

51

As illustrated in Table 3.11, the average time from initial screening (before the plea) to a
treatment intake was roughly three months for those placed into treatment. Most clients
who were placed into treatment had their intake after the plea, but about one in ten were
placed in treatment before their plea into the MHC; some were cases that had been
transferred from the drug court, and staff explained that others could be clients who were
re-accepted to a treatment program in which they were previously enrolled. For those
who were placed after the plea, the median time to treatment intake was 41 days after the
plea into MHC.
Table 3.11. Mental Health Treatment Received by Bronx MHC Participants
Mental Health Treatment Received by Court Participants (N=648)
Percentage of participants who made intake into a treatment program

80.7%

Of those who entered a treatment program (N=523)
Number of programs participant was placed into (mean)
Treatment placement into 'alcohol rehab/inpatient alcohol'
Treatment placement into 'detox'
Treatment placement into 'outpatient drug and/or alcohol'
Treatment placement into 'substance abuse rehab/short term
Treatment placement into 'therapeutic community/IP drug'
Treatment placement into 'temporary housing'
Treatment placement into unknown program type
Time from arrest to MH treatment intake (mean)
Time from arrest to MH treatment intake (median)
Time from MH screening to treatment intake (mean)
Time from MH screening to treatment intake (median)
MH treatment services began before MHC enrollment
Time from pre-plea treatment intake to MHC enrollment (mean)
Time from pre-plea treatment intake to MHC enrollment (median)
MH treatment services began on or after MHC enrollment
Time from MHC enrollment to post-plea treatment intake (mean)
Time from MHC enrollment to post-plea treatment intake (median)
Time in MH treatment during MHC participation (mean)
Time in MH treatment during MHC participation (median)

1.7 programs
0.6%
1.9%
57.2%
0.4%
57.2%
6.5%
1.1%
370.6 days
261.5 days
100.1 days
83 days
9.4%
66.8 days
26 days
90.6%
69.9 days
41 days
450.6 days
542 days

Once a client is placed into treatment, the TASC post-placement case manager
coordinates and communicates with treatment providers to monitor treatment progress
and resolve any problems. TASC case managers rarely provide any form of treatment,
themselves, and do not typically develop formal treatment plans (treatment plans may be
in place through treatment providers).
During the treatment period, the client is required to participate in community-based
treatment in addition to maintaining routine contact with both TASC and the court. The
participant takes part in weekly appointments with TASC members; this may be reduced
to bimonthly meetings depending on the client’s progress. The participant also has

52

weekly drug tests at the TASC office, and may additionally be drug-tested through
his/her treatment provider. If a participant is in a residential placement, the TASC team
will visit the individual on a monthly basis. The majority of Bronx MHC participants take
psychotropic medications as part of their treatment regimen. Most clients pay for
treatment through SSI or Medicaid.
Participants may be placed into multiple programs as needed. A given client may require
concurrent services from multiple providers, or may have to move from one program to
another. Looking again to Table 3.11, we see that of those who were placed into
treatment, one-half (52 percent) were placed into a single treatment program, close to
one-third (31 percent) were placed into two programs, 13 percent were placed into three,
3 percent were placed into four, and 2 percent were placed into five programs. The most
common types of placement were into therapeutic communities (57 percent) and
outpatient drug or alcohol treatment programs (57 percent). Seven percent were placed
into temporary housing. Other types of substance abuse treatment placements comprised
much smaller shares. These were described in the program database as “detox” (2
percent); “alcohol rehab or inpatient alcohol” (1 percent) and “substance abuse rehab,
short term” (0.4 percent). For the most part, treatment placements were to communitybased programs, but some placements were to private psychiatrists or therapists; these
were coded in the program data as outpatient drug or alcohol treatment. Among those
who were placed in treatment, the median time in treatment was 18 months.
Program Exit 
The average time from the MHC date to case termination was 18 months during the
2002-2006 study period, regardless of case outcome (see Table 3.12). Approximately
one-half of participants completed MHC requirements successfully; three in ten failed the
program; and two in ten were classified by the program as “neutral” terminations, whose
cases were closed for other reasons unrelated to client compliance.
Individuals graduate from the Bronx MHC program once the court (TASC, judge, DA,
and defense attorney) feels that treatment plan goals are achieved. Program data from
2002 to 2006 show that one-half (52 percent) of clients in the MHC successfully
completed their treatment mandates, as shown in Table 3.12. The Bronx MHC does not
use treatment phases or stages to define success as they feel that goals must be clientcentric. They are most interested in seeing participants achieve treatment compliance and
insight, rather than reduction of symptoms. Because these graduation requirements are
somewhat nebulous, participants can be retained in the MHC beyond the typical
treatment mandate. However, most who graduate reportedly do so within the suggested
mandate period if they are complying with treatment, remain stable, and have not been
arrested or had a positive drug test. The median time to successful program completion
(i.e., from the initial plea to resentencing for the reduced charge) among 2002-2006 cases
was 21 months, of which 19 months were spent in treatment (see Table 3.13). TASC is
the guiding force behind graduation recommendations. Participants may remain in their
community-based treatment program(s) beyond the court-mandated term; this decision is
made between the client and treatment provider.

53

Table 3.12. Case Outcomes Among Bronx MHC Participants
Case Outcomes Among Bronx MHC Participants (N=646)*
Successful completion of treatment mandate
Program failure before treatment placement
Client failed before treatment placement
Client failed to report to treatment placement

52%
5%
1%
5%

Program failure after treatment placement
Client left treatment
Client rearrested, warranted, or violated conditions

26%
18%
8%

Neutral termination**
Client sentenced to incarceration prior to treatment***
Client sentenced to incarceration after beginning treatment***
Probation to supervise or other non-incarcerative sentence***
Client unable to participate in treatment due to death or medical
disability
Client withdrew application
Criminal justice system opposition (e.g., Judge, DA)***
Client ineligible due to medical reason or death
Client ineligible due to psychiatric reason
Client unable to participate due to psychiatric reason
Subsequent discovery of no true mental illness and program transfer
Other

17%
3%
1%
0%
0%
2%
2%
0%
1%
0%
5%
2%

Percentages may not sum exactly due to rounding.
* Note: two records were missing data on termination status
** The program uses the term “neutral” to describe cases terminated before program completion through no
fault of the client’s.
*** These may have resulted from other concurrent open cases.

Upon successful completion of the Bronx MHC program, individuals with felony charges
often re-plead to a lesser felony charge or a misdemeanor charge (see Table 3.14).
Misdemeanants may have charges reduced to a violation or dismissed. In most cases,
successful completers received a non-incarcerative sentence such as probation or
conditional discharge. Relatively few had their charges dismissed or were sentenced to
time already served, and fewer still, but some, were sentenced to incarceration.
Participants receive a certificate of graduation and a picture of the participant with the
certificate is placed on a bulletin board in the TASC office. In some situations, the judge
may ask that a participant return to court periodically after graduation.33 Clients who
successfully graduate from the Bronx MHC are not allowed to repeat the program if they
are re-arrested.
33

There was a time lag between the resentencing that marked successful program completion and the
administrative case closing date. The median time from resentencing to administrative case closing was 3
days, but the mean was 15 days, reflecting the influence of cases with longer time lags.

54

Table 3.13. Duration of Bronx MHC Participation
Duration of Bronx MHC Participation (N=648)
Days from MHC enrollment to administrative case termination

Mean Median
464.6
539

Days from MHC enrollment to sentencing for successful completers

626.7

644

Days from MHC sentencing (of successful completers) to administrative
closing

15.7

3

Table 3.14. Sentencing Outcomes Among Successful Bronx MHC Completers
Sentencing Outcomes Among Successful Bronx MHC Completers (N=339)
Sentence type
Dismissal
Time already served
Probation or conditional discharge
Incarceration
Unspecified

4.72
5.01
79.06
1.18
10.03

Charge level of sentence imposed on successful program completers
Dismissal
Misdemeanor
Felony
Unspecified

4.72
37.17
16.52
41.59

A participant can fail the MHC program through serious lapses in compliance, most
frequently for absconding. If individuals fail the program, they are sentenced to the jail
alternative specified in the plea agreement when they entered the program. Those who
failed the program are eligible to participate again in the future. Program data show that
about 30 percent of Bronx MHC participants were terminated unsuccessfully. Most
program failures occurred after a client stopped attending treatment. The median time
spent in treatment among program failures was 5 months, compared to 19 months among
those who completed successfully. About one-quarter of program failures were due to rearrest, having a warrant, or violation of conditions. A small share of program failures
occurred before placement into a treatment program; this was usually because the client
did not report to treatment as required.
A portion of clients—nearly 20 percent during the 2002-2006 study period—were
considered “neutral” terminations from the MHC for other reasons, such as the inability
to find suitable treatment or concurrent criminal cases in another jurisdiction. The
median time from the MHC plea date to case termination was just under four months.

55

These neutral terminations are evaluated on a case-by-case basis and subject to a range of
sentencing options including time already served, probation, or an incarceration.
Stakeholder Views of the Bronx Mental Health Court  
Every interviewed individual was satisfied or extremely proud of the Bronx MHC
program. One interviewee claimed the community has accepted the program, and
multiple stakeholders cited examples of participant appreciation, including thank-you
letters and visits from clients. However, there were some identified challenges and areas
for improvement.
Program staff estimated that 960 to 1,000 of the 6,000 felony indictments in the Bronx
each year have a serious mental illness. However, the MHC’s capacity is considerably
smaller. About 200 clients are screened for eligibility annually, and about 150 are
accepted into the program.34 A need was seen for more financial resources and a larger
clinical team.
The Bronx MHC has not had much success with misdemeanor cases. While they offer the
program to misdemeanants, some stakeholders felt that the court did not have enough
leverage to deal with misdemeanor offenders when the alternative sentence was minimal.
During the participant evaluation stage, there were reported delays and difficulties in
obtaining collateral information from hospitals and outpatient treatment providers. One
stakeholder expressed that records might not be produced, even after a judge has ordered
the records. Some also felt that the evaluation process should not end after the original
assessment, and that participants should be re-evaluated throughout the program to track
progress and identify new issues. One respondent wished the program began earlier in the
process, when individuals were identified as incompetent. This individual felt that
defendants became stable in the hospital, but then decompensated once they returned to
jail to await trial.
Many of those interviewed reported difficulties with finding placements for clients.
Common placement issues included a general lack of programs, a lack of programs
providing housing, a lack of programs well-suited for particular populations (e.g.,
Spanish-speakers, adolescent MICA clients, DSM Axis II disorders, clients requiring
residential treatment who have children), frequent turnover at provider agencies, and
programs’ denial of clients due to criminal history or violence. Furthermore, some
stakeholders expressed concerns that the quality of care was deficient in many of the
treatment facilities. Many interviewees pointed to these limitations in the public health
system as inhibiting client progress. Case managers also reported difficulties
collaborating with treatment providers. Some case managers felt that treatment providers
withheld information, thinking they were protecting clients from negative consequences
in the court system.

34

These mental health court statistics are based on the program data from 2004-2006, which reported
higher numbers than the initial years of court operations.

56

Finally, at the time of the interviews, revisions to New York’s Rockefeller Laws were
under consideration. One interviewed stakeholder felt these legislative changes had the
potential to undermine the Mental Health Court, believing that defendants would prefer
to “try their luck” with a judge who might give a more lenient sentence, rather than
commit to a lengthy treatment period. At the time of report publication, these laws were
indeed modified to remove mandatory minimum sentences; it remains to be seen whether
and how these changes will impact the problem-solving courts in New York.
Brooklyn Mental Health Court 
Established in 2002, the Brooklyn MHC was developed collaboratively by the Center for
Court Innovation (CCI) and the New York State Office of Court Administration (OCA),
in partnership with the New York State Office of Mental Health (OMH).35 As this
research study began, Brooklyn MHC was one of just five mental health courts operating
in New York State; the others included the Bronx TASC Mental Health Court Diversion
Services, Buffalo City Mental Health Court, Monroe County Mental Health Court, and
Niagara Falls Mental Health Court (O’Keefe 2006).36
The Brooklyn MHC is a post-indictment problem-solving court that handles primarily
felony offenders (roughly 80 percent).37 It links defendants with serious, persistent
mental illness (SPMI) to long-term treatment as an alternative to incarceration (either jail
or prison) and, by doing so, works to effectively address both the needs of defendants
with mental illness and the public safety concerns of the community. Stakeholders
described the goals of the Brooklyn MHC program as 1) improving the court’s ability to
identify, assess, evaluate, and monitor offenders with mental illness; 2) creating effective
linkages between the criminal justice and mental health systems; 3) engaging participants
in treatment and ensuring they are linked to high-quality services; and, perhaps of
foremost importance, 4) improving public safety by reducing recidivism among offenders
with mental illness. The underlying assumption of the Brooklyn MHC is that defendants’
criminal behaviors are the result, at least in part, of untreated or inadequately treated
mental illness. Program operations are guided by the supposition that treating a
defendant’s mental illness leads to stability, which, in turn, leads to a reduction in
criminal behavior and improved psychosocial functioning.
The following sections discuss key aspects of the Brooklyn MHC program, including
program structure, staffing and the roles of key courtroom, eligibility criteria and target
35

CCI conducted a feasibility study for a mental health court in 2000. The findings from this study were
published in Rethinking the Revolving Door (Denckla and Berman 2001). Initial funding for the Brooklyn
MHC was supplied by private foundations, a contract from the New York State Office of Mental Health, a
New York City TANF grant, and a BJA grant. Currently, the program is funded through two contracts, one
with OMH and another with the New York State Unified Court System (UCS).
36
As O’Keefe noted (2006: 3), these courts all have a common set of goals and share common elements
derived from the problem-solving court model that include developing mechanisms to assess and identify
potential participants, providing adequate clinical information to facilitate informed decision making, using
the court’s authority to reinforce treatment goals, and linking participants to services in the community
(O’Keefe 2006: 3).
37
Brooklyn MHC also takes misdemeanants that have been transferred to Supreme Court through a
Superior Court Information.

57

population, referral sources and mechanisms, the role of screening and assessment, and
case planning and treatment. Supplementing this discussion are data on program
operations and participant characteristics derived from analysis of Brooklyn MHC
records data. Descriptive analyses focus on the 327 individuals38 who participated in the
MHC between March 1, 2002 and December 31, 2006.
Administrative Structure and Staffing  
The Brooklyn MHC is a specialized court operating within the Kings County Supreme
Court. Court occurs weekly: most cases are heard Tuesday mornings, but “spillover”
cases may be heard on Thursdays. The court program is presided over by a single
designated judge. Other key personnel include an extensive clinical team (project/clinical
director, dedicated senior social worker, three forensic counselors, and consulting
psychiatrists), designated Assistant District Attorney (ADA), and a core set of defense
attorneys affiliated with local indigent services agencies (Legal Aid, Brooklyn Defender
Services). Private attorneys represent some participants.
Role of the Judge  
Part of the court’s problem-solving approach is that cases are calendared together and
heard by a dedicated judge; this allows the judge to develop expertise in dealing with that
particular type of case, as well as a deeper knowledge of each individual case. In the case
of the Brooklyn MHC, the same judge has presided over the program since its inception.
The judge’s primary functions are presiding over status hearings and sentencing both
successful and unsuccessful defendants. During status hearings, the judge monitors
compliance with the treatment mandate set forth in the initial plea agreement. He also
weighs in on decisions pertaining to the legal eligibility of referred cases, and makes the
final determination regarding which cases move forward in the program. Additionally,
the Brooklyn MHC judge regularly visits treatment providers in the community to
understand the settings in which clients receive treatment and to build rapport with both
providers and participants.
The Role of the Prosecution  
A dedicated prosecutor is assigned to the Brooklyn MHC and serves as the liaison from
the DA’s Office to the court program, making the “official” referral to the Brooklyn
MHC program. (Referrals that go directly to the court’s judge or clinical team are redirected back to the DA’s office for a determination of whether the dedicated ADA will
consider MHC participation for that defendant; the clinical team awaits word from the
ADA before beginning its evaluation). As is the case with the presiding judge, the same
ADA has worked with the Brooklyn MHC since its start. This individual screens cases,
evaluates eligibility from a legal standpoint, outlines plea agreement terms, and has the
authority to decline cases. Additionally, the ADA liaisons with the “complaining
38

Although 327 individuals participated in the Brooklyn MHC during the study period, the impact analysis
included only 316 participants – i.e., the number that could be matched to appropriate comparison cases, as
discussed in Chapter 4.

58

witness” in a case when evaluating whether to send a defendant forward to the Brooklyn
MHC; while witness agreement is not an absolute requirement, the ADA will consult
with witnesses and take the witnesses’ perspectives into account particularly in assault
cases or other violent crimes.
The Role of the Defense 
Defense attorneys from both Legal Aid and the Brooklyn Defender Services routinely
represent Brooklyn MHC clients. The Supervising Attorney for the MICA Project, an
initiative that works with individuals who are mentally ill and chemically dependent,
carries a large number of Brooklyn MHC cases and advises other attorneys from her
office assigned to MHC clients. Some candidates for the program and participants are
represented by assigned counsel from the 18B Assigned Counsel Plan (a local program
that contracts with private attorneys to represent indigent defendants) or retain private
attorneys.
In addition to representing clients throughout their mental health court involvement,
defense attorneys also provide a “court orientation” to clients considering participation in
the Brooklyn MHC. As part of this process, the defense attorney explains the terms of the
plea and the consequences of failing to comply with the court’s treatment mandate; the
alternatives to mental health court (i.e., traditional case processing); and that participation
in the Brooklyn MHC is voluntary. Defense attorneys also review the treatment plan with
their clients to ensure they know and understand the terms of their treatment plans (i.e.,
length of the treatment mandate, the jail or prison alternative if they fail to complete
MHC, and the promised disposition if the client succeeds).
The Role of the Clinical Team  
The Brooklyn MHC follows a “Clinic Court Team” design: the clinical team is part of the
court, not a separate agency, and consists of a clinical director, a senior social worker, a
resource coordinator (who is also a social worker), and three forensic coordinators; a set
of consulting psychiatrists also assists the team with psychiatric evaluations. Participants
meet with several members of the clinical team—a case manager, a social worker, and a
psychiatrist—presumably to build broader support and familiarity with the client.
Additionally, the clinical team conducts psychosocial assessments, develops
individualized treatment plans, links participants in the court to community-based mental
health treatment and related services, communicates regularly with service providers
about the participants’ progress in treatment, and advises the judge on clinical matters
related to participants’ compliance or lack of compliance with their court-mandated
treatment plans.
Psychiatric assessments are conducted by one of the team’s consulting psychiatrists.
These assessments are typically performed with clients at the courthouse in the clinical
team’s office, and take about 60 minutes. Diagnoses are made based on the psychiatric
interviews and review of the defendants’ medical and psychiatric records, if available.
Where possible, the psychiatrist also collects collateral information from family members
who accompany defendants to the interview. The psychiatric assessment serves two

59

purposes: it establishes a diagnosis and thus, eligibility for the court program; and it
provides treatment recommendations (i.e., types of treatment that would be most effective
for the individual given his/her mental health status).
Psychosocial assessments are primarily conducted by the team’s senior social worker,
although both the resource coordinator and clinical director conduct assessments, as
needed; these assessments focus on clients’ psychiatric, medical, substance abuse,
housing, employment, education, and social histories. Reports are produced and provided
to the court (judge, defense attorney, and DA) with recommendations about both program
admission and treatment. Each report forms the basis for the client’s treatment plan,
which is largely designed by the team’s clinical director.
The clinical team’s forensic coordinators each maintain a caseload numbering between
20 and 30 Brooklyn MHC participants. Key duties include coordinating with treatment
providers to support client success, monitoring participant progress, and performing
random drug tests. Forensic coordinators obtain weekly updates from the court’s
treatment providers, and meet with MHC participants prior to each court session to assess
progress and to explore any treatment issues. Periodic visits are made to treatment
providers, particularly new providers, to observe their programs. Additionally, the team’s
senior forensic coordinator leads a weekly Remand Intervention group that focuses on
clients who are struggling in the court; new in 2008, the group is designed to help
participants “get back on track” to avoid being remanded.
The team’s clinical director performs a variety of duties and has authority over a number
of critical decisions. Foremost, final approval of defendant clinical eligibility rests with
the clinical director, as do decisions about participant treatment (again, the clinical
director designs and revises participant treatment plans to address emergent needs or
developments). Additionally, the clinical director provides participant updates to the
judge prior to each court session and holds regular meetings with the clinical team to
review case progress, including which participants may be ready for advancement to the
next phase. These regular clinical team meetings are designed to build the team’s
knowledge of the collective caseload; as a result, any member of the clinical team is
prepared to assist a client if his/her assigned coordinator is unavailable. The team
meetings also provide feedback from the team members regarding individual
participants’ treatment plans.
The Role of Treatment Providers 
Treatment is provided to Brooklyn MHC participants through a network of more than
100 community-based organizations. These organizations work directly with MHC
clients to provide mental health services, while also remaining in regular contact with the
MHC clinical staff. Brooklyn MHC staff work to link participants to a variety of services,
including outpatient mental health treatment (in-patient hospital care is rare; participants
with co-occurring disorders may be placed in appropriate residential drug treatment
programs), community-based case management, supportive housing services for
participants who are homeless at entry into the program or require clinical services to
maintain housing stability, and vocational and educational services. The program will

60

place participants with treatment providers in other parts of the NYC metropolitan area,
as well as out-of-state, to ensure the most appropriate service match.
Program Structure and Operations  
Figure 3.2 illustrates how cases proceed through the Brooklyn MHC. Briefly, defendants
are referred to the Brooklyn MHC program, screened for eligibility, matched with
community-based treatment, enter the court through a formal plea process, and then
participate in court monitoring, case management, and treatment services.
As previously mentioned and discussed in more detail below, duration of participation in
the Brooklyn MHC program varies based on charge and mental illness characteristics:
•

There is a 12-month treatment mandate for misdemeanor crimes.

•

Treatment mandates for nonviolent felony crimes typically span 12 to 18 months.

•

Predicate felony offenders and first-time violent felony offenders typically have
mandates of 18 to 24 months.

Participants enter the court by agreeing to a guilty plea with a sentence comparable to
what they would have received in a traditional courtroom. Although a sentence is agreed
upon at the time of the guilty plea, formal sentencing is suspended while the defendant
participates in the court and associated treatment. Defendants typically do not take a plea
until all community-based services are in place. Because it often can take a significant
amount of time to find an appropriate and available treatment placement, participants
may wait a considerable length of time while service linkages are made, either in jail or
while released to the community (depending on their bail- or own-recognizance status.)
Each stage is discussed in more detail below.

61

oklyn Menta
al Health Co
ourt Case F low
Figurre 3.2. Broo

62

Target Population and Eligibility Criteria 
The Brooklyn MHC currently39 targets both adult40 misdemeanor and felony offenders,
including predicate felons and, in some cases, violent offenders, whose mental illness is
believed to have contributed to their current criminal justice involvement. Court planners
held the view that, “most (if not all) of the behavior that led to the court charge is due to
mental illness … hence the condition of the offender, rather than the charge, defines the
appropriateness of the court” (CCI undated); however, the court team has come to
recognize that the connection between the mental disorder and criminal behavior is often
attenuated. Participation in the MHC is voluntary.
Potential Brooklyn MHC cases must meet both the clinical and legal eligibility criteria of
the program in order to participate. Clinical eligibility rests on a diagnosis of serious and
persistent mental illness (SPMI) on Axis I41⎯such as schizophrenia, bipolar disorder,
major depression, and schizoaffective disorder⎯for which there is a known, effective
treatment. Defendants with substance disorders may be eligible as long as they have an
additional Axis I diagnosis. Individuals with personality disorders, cognitive impairment,
developmental disabilities, brain damage, and dementia are not eligible for the MHC
program unless they also are diagnosed with an eligible major mental illness. Finally,
the MHC will not accept clients unless they are stabilized and competent to stand trial.
Legal eligibility hinges on the defendant’s immediate criminal offense. Determinations
about legal eligibility are made by the Brooklyn MHC judge in conjunction with the
program’s designated ADA. The program’s formal criminal justice eligibility guidelines
are as follows:
•

Felonies. All nonviolent felonies are eligible. Felonies involving assault, robbery,
and burglary are presumed eligible, but are reviewed by the clinical team and the
ADA. Other violent felonies are presumed ineligible, but are reviewed on a caseby-case basis, if referred. Murder and rape are “charge rule outs” (i.e., excluded).

•

Misdemeanors. All offenses are eligible, but the court is not intended for
offenders who would spend only a short amount of time in jail. Therefore,
misdemeanor offenders must be willing to accept a 12-month treatment mandate
and a potential jail sentence of up to one year for failure to comply.

39

The Brooklyn MHC was initially oriented toward nonviolent felony offenders. At the urging of defense
counsel, its focus has gradually evolved to include chronic misdemeanants.
40
Although Brooklyn MHC initially only worked with individuals aged 18 or older, the court program
began to accept 16 and 17 year olds in 2003-2004, consistent with the New York State age of majority.
41
The Diagnostic and Statistical Manual of Mental Disorders (fourth edition) presents a five-fold
classification that takes into account various mental disorders, the general medical condition of the patient,
any psychosocial or environmental problems, and the level of functioning. Clinical disorders, including
major mental illnesses, substance use disorders, and developmental disorders are included in Axis I;
personality disorders are on Axis II.

63

It also is important to note that individuals may participate in Brooklyn MHC more than
once. That is, a person is still eligible for the program if they have participated in the
program in the past.
Cases deemed ineligible or individuals who decline to participate are processed as usual
with prosecution reverting to the original ADA who was assigned the case. Misdemeanor
offenses are processed by the Criminal Court, whereas felony cases go through businessas-usual Supreme Court processing. Likewise, because Brooklyn MHC is a voluntary
program, any defendant who is determined to be eligible retains the option to decline
participation and may choose to have his or her case handled through conventional (more
traditional) means.
Demographic Profile  
The Brooklyn MHC program routinely records data on participating defendants, and to a
lesser extent, on candidates referred to the MHC program. The types of data collected and
recorded in the Brooklyn MHC database include demographics, mental health and
substance abuse status, services provided through the courts, and dates of case processing
milestones.
Table 3.15 presents the demographic characteristics of the 327 individuals who
participated in Brooklyn MHC between 2002 and 2006. On average, the MHC
participants were 33 years old when arrested on the instant offense leading to their
involvement in the court. The majority were male (76 percent); Black or African
American (58 percent); and unmarried (76 percent), although another 12 percent were
separated, divorced, or widowed.
According to program records, just 40 percent of BMHC court participants had earned a
high school diploma or otherwise obtained higher educational attainment. A sizeable
number (about 20 percent) of participants were homeless at the time of program intake;
30 percent had a history of homelessness in the previous five years. These figures suggest
that Brooklyn MHC participants present with a relatively high levels of need.
Although not presented in Table 3.15, the current analysis of Brooklyn MHC participants
indicates that the majority were arrested for a felony offense (84 percent)42⎯consistent
with the program’s initial focus on felony offenders⎯and referred to the court an average
(median)43 of five months after arrest. Roughly 60 percent were incarcerated at the time
of their referral to the court.
The Brooklyn MHC is intended to serve individuals with serious mental illness, and the
consensus among program staff and partners is that they receive the “sickest of the sick”
(UI interviews April 2009). Data on Brooklyn MHC participants’ mental health status
42

Felony offenses carried an average minimum incarceration term of 2.3 years and an average maximum
term of 3.3 years.
43
Medians are reported here because the means are skewed by a small share of cases with atypical case
processing statistics.

64

and substance abuse issues are presented in Table 3.16. Mood disorders (e.g., bipolar
disorder, major depression) were the most common Axis I mental illnesses diagnosed
among Brooklyn participants (55 percent), followed by psychotic disorders such as
schizophrenia (37 percent). The remainder of participants were diagnosed with other Axis
I conditions, typically substance-related Axis I disorders (3 percent) or anxiety disorders
(3 percent). Roughly 80 percent of Brooklyn MHC participants also were diagnosed with
secondary Axis I or Axis II disorders; these were most commonly substance-related
disorders, followed by personality disorders and learning disorders. A relatively small
share of BMHC participants (less than one in ten) had organic brain impairments and
developmental disabilities.
Table 3.15. Demographic Characteristics of Brooklyn MHC Participants, 2002-2006
Demographic Characteristics of Brooklyn MHC Participants (N=327)
Age at arrest
Male
White race
Black race
Other race
Hispanic ethnicity44
Has minor children
Minor children lived with client
High school graduate or higher educational attainment
Never married
Homeless at program intake
Homeless in the last 5 years

33 years
76%
38%
58%
4%
21%
31%
20%
40%
76%
19%
30%

Roughly six in ten (62 percent) Brooklyn MHC participants had co-occurring substance
use problems, defined by the research team as a substance-related diagnosis on either
Axis I or Axis II, and one-half of the participants reported alcohol use in the six months
prior to intake. Similarly, about one-half reported drug use in the six months prior to
intake. About one-third of alcohol users (35 percent) and 45 percent of drug users
reported past treatment for their substance use. (Data not shown in a table.)
MHC participants had extensive histories of past hospitalizations: 70 percent reported
prior hospitalizations for psychiatric reasons. In the year before the arrest leading to
Brooklyn MHC involvement, roughly 40 percent visited the emergency room (ER) for
psychiatric-related reasons, while another 46 percent reported a psychiatric-related
hospitalization. Close to half (48 percent) of the MHC participants had received mental
health treatment in the year before arrest, and about one-quarter (23.9 percent) were in
treatment at the time of program intake. Also, approximately 24 percent of participants
were on medication at the time of arrest.

44

Although one-fifth of participants were Hispanic, program personnel did not cite language as a
significant barrier to client placement for treatment.

65

Table 3.16. Mental Health Characteristics of Brooklyn MHC Participants
Mental Health Characteristics of Brooklyn MHC Participants (N=327)
Axis I diagnoses
Adjustment disorder
Anxiety disorder
Attention-deficit and disruptive behavior disorders
Impulse-control disorder
Learning disorder
Mood disorder
Psychotic disorder
Substance-related disorder
Other Axis I diagnosis

0.9%
2.8%
0.3%
0.3%
0.3%
54.9%
37.1%
2.8%
0.3%

Secondary Axis I or Axis II diagnoses
Of those diagnosed with a secondary Axis I or Axis II diagnosis (n=272)
Adjustment disorder
Anxiety disorder
Attention-deficit and disruptive behavior disorders
Delirium, dementia, amnesic and other cognitive disorders
Impulse-control disorder
Learning disorder
Mental retardation
Mood disorder
Personality disorder
Substance-related disorder
Other Secondary Axis I or Axis II diagnosis

0.4%
3.7%
1.1%
0.7%
0.7%
3.3%
0.7%
1.5%
10.7%
70.2%
1.8%

Substance abuse
Any primary or secondary substance use disorder (Axis I or Axis II)

61.2%

Other mental impairments
Brain impairment
Developmental disorder
Mental health history
Psychiatric-related visits to ER in year before arrest
Visits to the ER in year before arrest
Psychiatric hospitalizations in total lifetime
Psychiatric-related visits to hospital in year before arrest
Any days in hospital in year before arrest
On medication at time of arrest
Mental health treatment during year prior to arrest
Was in mental health treatment as of program intake

6.5%
2.4%
40.4%
13.6%
70.2%
46.7%
42.9%
27.9%
47.3%
23.9%

 

66

Referral  
The Brooklyn MHC does not employ a universal screening mechanism (e.g., based on
charges or criminal history) to determine which types of cases will be sent to the court for
an eligibility determination. Instead, referrals may originate from a variety of sources,
including the Office of the DA, defense attorneys, “730” competency proceedings,45
other judges,46 and other sources within the Brooklyn court system.
As in most jurisdictions, an arrest ushers an individual into the criminal justice system.
The typical legal process in NYC is as follows: arrestees are booked by the police at the
local precinct. By law, arrestees must be arraigned in Criminal Court within 24 hours of
arrest, where they are formally charged. At arraignment, the arrestee may be released
(through case dismissal, bond, bail, or personal recognizance) or remanded to custody at
Rikers Island. Following arraignment, misdemeanor offenses are formally charged;
felony offenses are transferred to Supreme Court for indictment. A Superior Court
Information (SCI) may be used to transfer misdemeanor cases to Supreme Court.47
Most referrals to the Brooklyn MHC occur after indictment or SCI. The time from arrest
to referral is on the order of months, not weeks. As discussed in the previous section, the
MHC receives referrals from a variety of sources. While there is no universal mental
health screening process for all defendants in NYC, all admissions to Rikers Island are
screened for mental illness (e.g., those defendants who were remanded to custody after
arraignment). The purpose of this screening is to determine eligibility for discharge
planning services under the Brad H settlement (discussed more fully in the Business-As
Usual section of this chapter). Because of the differing aims of the Brad H mental health
screening process, Brooklyn MHC does not consider the results of that process in
determining program eligibility; however, the jail-based screening may help other system
actors “flag” a case as appropriate for referral to the Brooklyn MHC.
When a case is referred, it is placed on the treatment court calendar for a first appearance
(a defendant may be placed on the Brooklyn MHC calendar if the referral source believes
the person has SPMI or is at least mentally ill). If the referral source is the defense
attorney, then the defense attorney may contact the ADA ahead of the first appearance to
discuss the case. Screening of a defendant’s case does not start until after the first
appearance when all parties agree to the clinical screening.
Eligibility Screening and Program Acceptance 
Brooklyn MHC clinical staff begins the assessment process once all parties signal their
agreement to do so. The clinical assessment occurs after the client’s first MHC
appearance. It consists of a psychosocial assessment by the clinical team’s senior social
45

This process determines fitness to stand trial. Unless the DA’s office objects on criminal justice grounds,
all cases in which a defendant has been found unfit are automatically calendared in the Brooklyn MHC
once the defendant is restored to fitness.
46
Typically, this refers to the criminal court judge on whose docket the case originally appeared.
47
Indictment is specific to felony cases, which are processed in the NY Supreme Courts. Misdemeanors
are typically processed in the NY Criminal Courts.

67

worker and a psychiatric evaluation by one of the program’s consulting psychiatrist. Each
produces a written report that is provided to all parties involved (judge, defense attorney,
DA, BMHC clinical staff). The final mental health eligibility determination is made by
the Court’s Clinical Director based on the results of these assessments.
The clinical assessment includes an assessment of potential participants’ risk of violence
to evaluate whether the defendant presents too great a risk to remain in the community;48
results of the assessment are shared with key decision makers, including the judge, the
ADA, and the defense attorney. The BMHC judge makes the final decision on defendant
participation.
For those individuals deemed eligible and suitable,49 the defense counsel will work with
the defendant to confirm his/her willingness to enter the program; a key part of this
process, as noted earlier, is making sure the defendant understands the terms of the plea,
including the treatment mandate and both the positive and negative consequences of
complying with the treatment mandate.50 For eligible defendants who choose to
participate in the program, the next step is to submit a guilty plea and agree to participate
in the services specified in the individualized treatment plan; a release also is signed,
which allows the clinical team to collect and exchange client information with the court’s
treatment providers.
The individualized treatment plan typically includes some combination of mental health
treatment, substance abuse treatment, community-based case management services,
supported housing, and vocational/educational services. Tailoring treatment plans to the
needs of the individual participant is a critical element of the mental health court
approach in Brooklyn, as was emphasized by the court program’s clinical staff.
According to the MHC staff, the individualized case plan is developed prior to the time of
plea. Frequent appearances are scheduled before the judge to monitor client progress.
While defendants are in the pre-participation “candidacy” stage, they appear in court
once per month. Fully-enrolled participants appear in court every one or two weeks for
the first three months and then monthly unless otherwise specified by the judge. More
frequent court appearances may result from noncompliance or because the judge feels
extra judicial supervision will encourage compliance. Defendants also meet with case
management staff whenever they have a scheduled court appearance.
Although a sentence is agreed to at the time of the guilty plea, formal sentencing is
suspended while the defendant participates in the MHC. Defendants agree to comply with
their treatment plans for a specified period of time, typically ranging from 12 to 24
48

The Brooklyn MHC does not use a structured or actuarial risk assessment tool. The program’s social
workers and psychiatrists received training on one of the leading instruments (the HCR-20) and incorporate
risk factors into their clinical assessment.
49
Early in the study, the MHC staff noted that defendant motivation, public safety concerns, and lack of
appropriate treatment could be grounds for deeming a case ineligible.
50
This decision is typically made based on the assessed likelihood of full prosecution and a sentence to
incarceration.

68

months.51 Participants typically entered the programs in the agreed-upon treatment plan
once they submitted a guilty plea to the court. However, because of the time needed to
establish the treatment plan and secure program slots before the plea, participants were
often in contact with the court for longer than the mandated supervision period. Each
stage is discussed in more detail below.
If participants comply and successfully complete the terms of the court mandate,
misdemeanants and first-time nonviolent felons have their plea vacated and case
dismissed. Predicate felons and first-time violent felons may have their felony charges
reduced to a misdemeanor with a sentence of either conditional discharge or probation.
Conversely, participants who persistently fail to comply with the court’s mandates or
commit new offenses are sentenced to the jail or prison term agreed to at the time of the
plea. However, individual instances of noncompliance and new offenses did not
necessarily result in automatic termination from the program. The judge exercised
discretion in evaluating individual incidents within the context of a given participant’s
circumstances and history with the program, as described further in the study’s courtroom
observations.
Table 3.17 provides case processing statistics for BMHC participants. The full eligibility
and placement process for BMHC participants lasted two to three months on average
(median was 70 days) before defendants officially began their participation in the court.52
Generally, the timeline was as follows:
•

Defendants’ first, pre-plea court appearance as mental health court candidates
typically occurred on the same day or within one day of their first contact with
mental health court staff.

•

Eligibility assessment spanned about a month with a psychiatric assessment
typically conducted 11 days after initial contact with the court, the psychiatrist’s
report completed 3 days later, and assessment of other eligibility considerations
(e.g., criminal justice eligibility and cooperation of the district attorney) taking up
the remaining 2 weeks.

•

Eligibility determination to program acceptance lasted approximately another
month, during which the Brooklyn MHC clinical staff worked to finalize public
entitlements (e.g., Medicaid coverage), treatment plans, and placements with
community-based mental health service providers.

51

Early in the program, Brooklyn MHC participants signed contracts to signal their understanding and
agreement to comply with the court’s treatment mandates; as the program matured, key actors viewed the
court record as sufficiently binding.
52
Medians are reported in this section because the means were skewed by a small number of observations
with atypical case processing statistics.

69

Table 3.17. Case Processing Statistics Among Brooklyn MHC Participants
Case Processing Statistics Among Brooklyn MHC Participants (N=327)
Mean
Time from arrest to first court contact
245 days
Instant offense is a felony charge
84%
Incarcerated at time of referral
57%
Time from first court contact to MHC program start
93 days
Time from first court contact to eligibility
determination
38 days
Time from first court contact to first appearance
1 day
Time from first court contact to psychiatric
assessment
30 days
Time from psychiatric assessment to report
4 days
Time from psychiatric report to eligibility
decision
22 days
Time from eligibility decision to plea offer
55 days
Days from MHC start to case closing
521 days

Median
149 days
n/a
n/a
70 days
28 days
0 days
11 days
3 days
12 days
30 days
444 days

As previously discussed, defendants enter Brooklyn MHC by submitting a guilty plea to
the charged offense. Participants who complete the treatment mandate successfully
receive reductions in the disposition charge and associated sentence. Prospective clients
and their counsel must agree to the initial charge and sentence, the length of the courtordered treatment mandate, and the alternate charge and sentence, if any, upon successful
completion of the court’s requirements. Most participants were accepted and elected to
participate on their first referral to the mental health court, though approximately five
percent of participants had been referred once before.53
Although the final approval on whether to accept eligible cases rests with the Brooklyn
MHC judge, participation is voluntary and candidates may decline to the opportunity to
enter the treatment court.
Brooklyn Mental Health Court Proceedings 
Brooklyn MHC operates a dedicated mental health court docket (i.e., only mental health
court cases appear on the docket and are heard when court is in session). It is designed to
reduce stigma and promote a sense of community among participants.
Initially, the MHC participants appear in court weekly, then appearances drop to every
two weeks for the first three months, and then to monthly for the duration of the
treatment mandate. More frequent court appearances can be required for noncompliant
participants. Court sessions occur Tuesday mornings; “spillover” cases may be heard on
Thursdays, however, the Thursday docket may include a variety of cases, including those
53

Multiple referrals may have resulted from a defendant initially having been found mentally unfit to stand
trial, restored to fitness, then re-referred to the program.

70

on other specialized dockets, such as domestic violence or “Treatment Alternatives for
Dually Diagnosed Defendants.”
Project researchers observed two sessions of the Brooklyn MHC, first in April 2009 to
field test the courtroom observation data collection protocol and again in October 2009.
Each session lasted three to four hours. During the October 2009 session, researchers
observed 85 scheduled appearances; defendant appearances averaged one to two minutes.
Observations were conducted from the jury box with researchers using a one-page data
collection protocol to document the court environment54 and communication patterns,
and record judicial responses to both compliance and noncompliance with court
conditions (see Chapter 2 for more information on the instrument, including key data
collection components). These data were aggregated and are reported in Tables 3.18 and
3.19.
The observed session included a variety of court appearances. The most common type of
appearance was a status hearing, as shown in Table 3.18. During status hearings, the
court monitored the progress or compliance of MHC participants. Status hearings were
held weekly, biweekly, or monthly in the Brooklyn court. Roughly 10 percent of the
hearings we observed were pre-plea appearances, during which defendants were
evaluated for inclusion in the mental health court. A few of the hearings observed were
graduations, where a defendant successfully completed the MHC program. Other types of
hearings included cases from other dockets (e.g., domestic violence cases) and nonappearances (including people medically unable to come, those in custody, and a few
failures to appear). Approximately one-fifth of the scheduled hearings were actually noshows.
Table 3.18. Types of Brooklyn MHC Hearings Observed
Types of Brooklyn Mental Health Court Hearings Observed (N=85)
Pre-plea or plea
Status hearings
Graduation or sentencing
Non-MHC
No-show

13%
45%
4%
18%
20%

Main courtroom actors consisted of the judge, MHC participants, defense attorneys, the
prosecutor, and supervisory members of the clinical team (the clinical director and the
resource coordinator). The clinical director remained in the courtroom during
proceedings, but off to the side. Occasionally there were other courtroom actors,
including interpreters, family members, and the court clerk. Table 3.19 shows who, other
than the judge, spoke during these status hearings. Defendants typically appeared with
54

Main courtroom actors included the judge, MHC participants, defense attorneys, the prosecutor, the
clinical director. The judge, stenographer, and clerk sat at the front of the courtroom. The prosecutor sat at
one table; the defense attorney at the other. The clinical director remained in the courtroom during
proceedings, often sitting off to the side.

71

their attorneys at all status hearings. It should be noted that the case management
supervisors who participated in court were generally not the case managers assigned to
work with the participant for whom they were speaking in court.
Table 3.19. Courtroom Interactions Observed During Brooklyn MHC Hearings
Courtroom Interactions Observed in the Brooklyn Mental Health Court (N=85)
Other than the judge, who spoke at status hearings?
Defendant
Defense attorney
Case manager*
Prosecutor
Other
Judge-Defendant interactions
Defendant addressed first
Defendant approached bench

92%
37%
21%
8%
5%
71%
53%

* The term “case manager” is used generically here to refer to any member of the court’s clinical case
management team.

Examining the judge-participant exchange, we observed that the judge typically spoke
directly to participants, not through their attorneys, to address their compliance with court
requirements and to impart instructions. Overall, proceedings in the Brooklyn court were
often defendant-centered. The judge addressed the participant first before addressing
other courtroom actors and employed a conversational style, starting his interactions with
the defendants with open-ended queries such as, "How are things going?" The Brooklyn
judge often asked probing questions and frequently invited defendants to participate in
discussions with him at the bench. Participants typically spoke in court (about 90 percent
of the time), asking questions or making statements, and about 50 percent of the time
approached the bench to speak with the judge.
Much of the judge-participant exchange centered on monitoring compliance with court
requirements. The majority of compliance monitoring activity (regular meetings with
defendants, urinalysis, and periodic progress reports from treatment providers) took place
outside of court, and the case management team produced written reports of client
progress, which were circulated to the judge, defense, and prosecution in advance of the
court session. Additionally, the case management supervisor briefed the judge on the
caseload during the hour before the court session began. The judge was observed
referring to these written reports in court.
Compliance 
The majority of cases we observed were compliant. The judges typically responded to
compliant cases with verbal praise, but we observed that sometimes cases were simply
continued without overt praise. The Brooklyn MHC employed other rewards, as well.
The judge often invited compliant defendants to speak with him at the bench (close to 60
percent of compliant defendants spoke with the judge at the bench, regardless of overt

72

praise); noncompliant defendants were rarely invited to speak at the bench. The court also
used phase completion certificates to mark progress in the treatment mandate, and there
was courtroom applause when these certificates were awarded.
Defendants "graduated" upon successful program completion. In Brooklyn, this was
announced as graduation day. The judge awarded a certificate to the defendant, the
defendant received a small present (e.g., a box of chocolates), and the defendant shook
hands with the judge. The defendant was also given an opportunity to speak to the court.
In some observed cases, the judge further clarified the terms of the reduced sentence in
response to defendants’ questions.
Noncompliance 
The following types of noncompliance were observed in other appearances: treatment
absences or nonparticipation, positive drug tests, and failure to appear at court. All stated
noncompliance was addressed by the judge; this was typically through verbal recognition.
The judge commented on the noncompliance, issued instructions or advice to the
defendant, and explained the consequences of not complying with the court's conditions.
Occasionally, we observed that the judge increased the frequency of either court
appearances or case management meetings. Additionally, we occasionally witnessed
defendants who were remanded to jail. (More often, we saw defendants who had been
remanded to jail on their previous appearance, and were being seen again by the judge,
who was evaluating whether or not to release them on recognizance at this time.)
Warrants were issued if defendants were no-shows without acceptable reasons. The court
did not apply a fixed sanction algorithm. During our stakeholder interviews, we were told
that noncompliance was dealt with on a case-by-case basis, and that frequent "second
chances" were given. Note that the terminology "second chance" was used by both
courts, though in reality defendants were given multiple chances.
The case managers in attendance during Brooklyn court appearances were typically the
clinical team supervisors, rather than the case managers directly assigned to client cases.
During the court session, they often functioned in problem-solving roles. We observed
case management supervisors troubleshooting clients’ problems with their providers or
court requirements during bench conferences. They tended not to speak in hearings if
there was no problem. Overall, we observed that representatives of the case management
team spoke in less than half of the Brooklyn hearings. Although one of their roles is to
provide information to the court, this was done through the written reports prepared in
advance of the court session. The case management did not routinely make verbal
reports in court of defendants’ progress.
Prosecuting and defense attorneys typically participated in more traditional appearances,
such as pre-plea and sentencing (e.g., final) appearances, but participated less than half of
the time during status hearings. When attorneys participated in status hearings, it was
usually when there was some type of an issue. For example, defense attorneys were
observed advocating for their clients if some noncompliance was being discussed.
Prosecuting attorneys rarely spoke during the status hearings. Stakeholders interviewed

73

by the research team spoke of a generally collaborative, rather than adversarial spirit,
once a case had been accepted to the MHC.
Clinical Treatment and Case Management 
The Brooklyn MHC uses a Clinical Court Team design: the clinical team is part of the
court, as opposed to a separate agency, and clinical staff develops a shared knowledge of
every participant in the program through regular team case review. Once accepted into
the program, clients participate in mandated, community-based treatment for a 12-month
period (misdemeanors), 12- to 18-month period (first-time felonies), or 18- to 24-month
period (for predicate felons). The clinical team places participants in different boroughs
and states, if necessary, to facilitate access to the proper treatment; most treatment is
outpatient programming, although residential programs are also used.
Once a defendant is part of the MHC, the case management team engages in back-andforth problem-solving with the community treatment providers, designed to monitor
participant progress and trouble-shoot any potential issues the either provider or the
participant may have with the placements.
It is important to note that the court’s clinical team works to place Brooklyn MHC
participants in the same community-based programs, when possible. The rationale is that
placing participants together builds a sense of community and reduces stigma—a key
concern of the program.
Mental Health Court Participation and Outcomes 
A total of 519 individuals were referred to the Brooklyn MHC program between 2002
and 2006, of whom, 327 participated in the program. As noted earlier, cases must meet
the program’s mental health and legal eligibility requirements and agree to enter a plea in
order to participate. Not all referrals are determined to be eligible. In turn, participation in
the Brooklyn MHC is voluntary and eligible defendants may choose not to participate for
a various reasons. Appendix F provides the results from an analysis comparing Brooklyn
MHC participants and nonparticipants; the analysis examines the extent to which the two
groups differ according to demographics, criminal justice indicators (prior and current
involvement), mental health status, and reasons for nonparticipation.
For the 327 individuals who participated in the Brooklyn MHC program, participation
lasted for an average of 15 months55 from the plea date to case closing. The majority of
participants who enrolled in the mental health court completed successfully: 74 percent
graduated. In contrast, roughly 18 percent terminated unsuccessfully and just under 10
percent had their cases closed for other reasons. About one percent of cases were still
participating at the time of data collection in 2010.

55

Medians are reported in this section because the means were skewed by a small number of observations
with atypical case processing statistics.

74

Program staff reported that the majority of participants received outpatient mental health
treatment from community-based providers. However, detailed information about the
nature of the mental health treatment received (e.g., provider type, therapeutic
approaches, or prescription medications) was not present in the administrative database
available to the research team.
In addition, MHC participants received a number of specialized services through the
court, as is reflected in Table 3.20. Participants were typically referred to more than one
type of service: 40 percent were referred to two program types, 35 percent were referred
to three program types, and 7 percent were referred to four program types. Clients usually
received some form of case management services from a community-based treatment
provider; the court’s clinical team explicitly does not function in a case management
capacity once a participant is accepted into a treatment program. Overall, three-quarters
of participants were recorded as receiving case management services, including services
described as case management (53 percent), integrated or supportive case management
(39 percent), and assertive community treatment (ACT) team services (6.4 percent).
One-half of MHC participants were placed into housing as part of their court-mandated
treatment plan. This included supportive housing services (20 percent), residential
substance abuse treatment (27 percent), and mentally ill chemically addicted (MICA)
housing (10 percent). Of those who had been homeless upon referral to the court, 81
percent had been placed into some type of housing. Nearly one-half (47 percent) of the
clients who were homeless were placed into residential substance abuse programs, 12
percent were placed into shelter programs, 10 percent were residing with family
members, 2 percent were living independently, and the remainder were in a range of
supportive housing arrangements.
Nearly three-quarters of mental health court participants received substance abuse
treatment services, specifically: MICA treatment (47 percent), substance abuse residence
(27 percent), substance abuse treatment (23 percent), MICA housing (10 percent), and
adolescent drug treatment (0.3 percent).
A small portion of mental health court participants also received vocational services.
Seven percent received adult vocational or educational services and one percent received
similar services targeted to adolescents.
Program Exit 
Brooklyn MHC stakeholders look for multiple outcomes to define program success: no
re-arrests, cessation of drug use, and adherence to the treatment mandate.
To graduate, participants must successfully pass through the court’s four stages
(Adjustment; Engagement; Progress and Preparing to Graduate) and remain arrest-free.
Certificates are awarded to participants at the completion of each stage.56 Graduation
56

Note that there are no universal program criteria for moving through these four stages. Criteria for
moving from one stage to the next are determined for each individual based on their situation at program

75

results in a dismissal of charges for misdemeanor and nonviolent, first-time felony
offenders. Predicate felons and individuals who commit first-time violent felonies will
have their charges reduced to a misdemeanor plea and receive a period of probation.
Individuals who fail to complete the Brooklyn MHC are typically sentenced to jail or
prison in accordance with the original plea agreement.
Table 3.20. Specialized Treatment and Services Provided Through the Brooklyn
MHC
Additional Services* Provided Through the Mental Health Court (N=327)
Total number of program types assigned by the court (mean)

2.3 programs

Any case management services
Assertive Community Treatment (ACT) Team
Case management
Integrated Case Management (ICM) or Supportive Case Management
(SCM)

74.3%
6.4%
52.6%

Any housing
Housing
Mentally Ill Chemically Addicted (MICA) housing
Substance abuse residence

51.7%
19.6%
10.4%
27.2%

Any substance abuse treatment services
Adolescent drug treatment
Mentally Ill Chemically Addicted (MICA) housing
Mentally Ill Chemically Addicted (MICA) treatment
Substance abuse residence
Substance abuse treatment

72.8%
0.3%
10.4%
46.5%
27.2%
22.9%

38.5%

Any vocational services
Adolescent vocational/education
Vocational / Rehab / Education

7.7%
1.2%
6.7%

* Meaning services in addition to mental health treatment.
Note: Percentages may not sum exactly because a person may be assigned to multiple program types within
each category.

Table 3.21 presents program outcomes for the Brooklyn MHC’s 327 participants who
entered the program between 2002 and 2006. Nearly three-quarters graduated upon
successful completion of the treatment mandate. Close to one-fifth had their cases
terminated as unsuccessful, and the remainder had cases that were terminated for other
reasons. Four individuals who were referred to the MHC between 2002 and 2006 were
still participating in the program as of this study’s data collection in late 2010.
intake (e.g., levels of psychiatric stability, substance use, functioning, and social supports). Progression
through these stages is used to mark and celebrate progress rather than measure quantifiable achievements.

76

Table 3.21. Case Outcomes for Brooklyn MHC Participants
Case Outcomes for Brooklyn MHC Participants (N=327)
Current Participant
Graduated
Terminated
Other

1.22%
74.01%
18.35%
6.74%

Graduates of the MHC originally pled to average incarceration sentences of 2.2 to 3.1
years (median 1.3 to 2.5 years). Table 3.22 provides additional sentencing-related data
outcomes for Brooklyn MHC graduates. Nearly four in ten graduates had their charges
dismissed, and about one-half had their felony cases reduced to misdemeanor sentences.
In 21 percent of cases, there were no further supervision requirements; but in 30 percent,
there were 1- to- 3-year probation requirements. Roughly 11 percent of cases had other
case dispositions.
Table 3.22. Sentencing Outcomes Among Brooklyn MHC Graduates
Sentencing Outcomes Among Brooklyn MHC Graduates (N=242)
Charges Dismissed
Misdemeanor conviction with time served
Misdemeanor conviction with 1 year
probation
Misdemeanor conviction with 3 years
probation
Other

37%
21%
3%
27%
11%

Participants who failed to complete the MHC originally pled to average incarceration
sentences of 2.5 to 3.6 years (median 2 to 3.5). This was not a statistically significant
difference when compared to the plea offers made to program graduates, suggesting that
successful participants had comparatively serious offenses and criminal histories as those
who failed.
Longer-term criminal justice outcomes of MHC participants are addressed in later
sections of this report, specifically in Chapter 4 that covers the impact evaluation.
However, it is worth noting, here, that a very small percentage of MHC health
participants (2 percent) were subsequently re-referred to the mental health court between
2002 and 2006, after their initial case closed. These included program graduates, program
failures, and defendants who had alternate dispositions. The time from case closing to rereferral ranged from 63 to 882 days.

77

Comparison of Successful and Failed Participation in the Mental Health Court 
We conducted t-tests to understand factors that differentiated program success and failure
among Brooklyn MHC participants. The following reports on the characteristics of MHC
graduates compared to defendants who terminated unsuccessfully. Our analysis found
differences in demographic characteristics, substance abuse history, and services received
between program graduates and those who failed. Bivariate correlations were assessed
using t-tests. Unless otherwise noted, reported differences were statistically significant at
the 95 percent confidence level.
As Table 3.23 below, indicates successful completers were more likely to be White
(p=.10) and less likely to be Black (p=.08). Successful completers were also somewhat
older, with an average age at arrest of 34 compared to 31 among program failures
(p=.11). Additionally, successful completers were less likely to have been homeless at
intake; 16 percent had been homeless at program intake compared to 28 percent of
program failures (p=.08).
We did not observe any differences in the types of mental illness diagnosed among
program graduates and program failures; however, those who failed the Brooklyn MHC
were significantly more likely to have had a substance abuse diagnosis in addition to their
mental health condition. Specifically, 55 percent of program graduates had primary or
secondary substance use diagnoses, compared to 82 percent of program failures.
Program graduates and failures were similar with respect to having been charged with a
felony offense. However, program failures were more likely to have been incarcerated at
the time of referral: 72 percent of program failures had been incarcerated pretrial
compared to 52 percent of program graduates. The median time from arrest to program
referral was between 4 and 5 months (139 days) for program graduates and 5 to 6 months
(163 days) for program failures, but the difference in means was not statistically
significant. There were no appreciable differences between program graduates and
program failures in the duration of the screening and eligibility process.
Brooklyn MHC program graduates and program failures spent a comparable amount of
time in the mental health court from their plea at entry until case closing. The median
time in the mental health court was 14 to 15 months (434 days) for successful completers
and 16 to 17 months (498 days) for unsuccessful participants; the difference in means
was not statistically significant.

78

Table 3.23. Characteristics of Brooklyn MHC Graduates and Failures, 2002-2006
Characteristics of Brooklyn MHC Graduates and Failures, 2002-2006 (N=327)
Graduated
Failed
(n=242)
(n=60)
Demographics
Age at arrest
33.8 years 30.9 years
Male
72.7%
88.3%
Black race
55.0%
67.8%
White race
40.4%
28.8%
Other race
4.6%
3.4%
Hispanic ethnicity
22.1%
13.6%
Homeless at program intake
16.4%
28.1%

p-value
0.1079
0.0026
0.0753
0.1008
0.6883
0.1465
0.0775

Referral to MHC
Felony charge
Incarcerated at the time of program referral

82.6%
51.5%

88.3%
71.7%

0.2810
0.0047

Mental health
Axis I mood disorder
Axis I psychotic disorder
Substance abuse diagnosis on either Axis I or II

54.1%
36.8%
55.0%

59.3%
33.9%
81.7%

0.4740
0.6812
0.0001

215.8

319.7

0.1700

22.4

31.4

0.3411

35.7

43.3

0.1444

51.4

66.7

0.1405

519.7

547.0

0.5629

2.3
77%
46%
69%
9%

2.4
65%
70%
87%
5%

0.5144
0.0595
0.0009
0.0010
0.2771

Mental health court processing
Days from arrest to first court contact (mean)
Days from first court contact to psychiatric
assessment (mean)
Days from first court contact to eligibility decision
(mean)
Days from eligibility assessment to court start date
(mean)
Days of mental health court participation from plea to
closing (mean)
Mental health programs received
Number of program types received
Any case management services
Any housing
Any substance abuse treatment services
Any vocational services

Persons who graduated and failed the mental health court received different levels of
supportive mental health services. Given the higher rates of substance abuse among
program failures, it is not surprising that significantly more program failures participated
in substance abuse treatment, particularly residential substance abuse treatment.

79

Similarly, program failures were more likely to have been homeless before their
participation, and so were more likely to have received housing. Additionally, those who
failed were significantly less likely to have received case management services (p=.06).
Approximately 77 percent of program graduates received case management services
compared to 65 percent of program failures. This difference was driven by differences in
services described simply as “case management,” as program graduates and program
failures were similar with respect to receiving assertive community treatment (ACT)
team services, intensive case management (ICM), and supportive case management
(SCM) services.
Stakeholder Views of the Brooklyn Mental Health Court 
Brooklyn MHC stakeholders identified the long-standing stability of the court team (the
judge, clinical director, and ADA are founding members; the primary defense attorney
has been with the team many years); the court’s hands-on approach and attentiveness to
building a sense of community among participants and reducing stigma; its recognition of
the importance of individual treatment and success outcomes; and the checks and
balances of having the judge, defense, prosecution, and treatment working together as
critical elements of the Brooklyn MHC program. Stakeholders believe these elements are
critical to the program’s longevity and to achieving program outcomes, namely increased
public safety, reduced re-offending, and improved participant mental health.
Most stakeholders expressed satisfaction with current program procedures, including
assessment, decision-making, treatment mandate, case supervision, and case
management. Lack of community-based mental health treatment was the only challenge
consistently cited by stakeholders. One stakeholder provided a more nuanced view that
the challenge was in locating specialized services for clients with multiple needs; while
outpatient mental health treatment is generally available, services for clients requiring
integrated substance abuse treatment or supportive housing are considerably more
limited.
“Business‐as‐Usual” Mental Health Services Received by Criminal Defendants in 
Brooklyn and the Bronx, New York 
Generally speaking, the current evaluation used existing records and a quasi-experimental
design to measure the impact of mental health court participation as compared to
conventional criminal case processing. In this section, we discuss the mental health
services received by defendants whose cases were processed through conventional courts.
The court system as a whole neither measures, nor tracks the mental health status of
defendants, so it is impossible to precisely estimate the full population of criminal
defendants with mental illness. From our interviews with mental health court
stakeholders, we know that arrestees with mental illness include a combination of those
who have been treated in community-based settings, as well as those who have not.
However, there is no source of information that describes the characteristics of the
universe of all criminal defendants with mental illness.

80

Systematic data, however, are collected and maintained on a large subset of criminal
defendants with mental health problems. All persons incarcerated by the New York City
Department of Correction (DOC)57—typically in the jail facilities on Rikers Island—
undergo a series of screenings upon admission to identify possible mental health issues.
Persons with identified mental health issues are provided further mental health evaluation
and treatment services during their incarceration. Additionally, inmates with mental
health issues receive discharge planning services to facilitate continuity of care in the
community upon release.
This population of DOC inmates who received jail-based mental health services forms
the basis of the study's comparison group. There is no comparable source of information
on the mental health status of defendants who were not incarcerated before trial. We
describe the jail system’s admission, screening, mental health treatment, and discharge
planning processes to provide an understanding of the standard level of mental health
services typically received by defendants with mental illness whose cases were
adjudicated through conventional courts. Bear in mind, however, that defendants with
mental illness who were not incarcerated pretrial would not have received these services,
and may have received more or less care in the community. Additionally, to the extent
that mental health court defendants were incarcerated pretrial, they would have received
these services prior to the additional screening, assessment, treatment placement, and
court monitoring provided by the specialized MHC programs.
The DOC inmate population includes a large number of defendants with mental illness.
Nationwide, researchers estimate that 64 percent of jail inmates experience mental health
symptoms (James and Glaze 2006) and 14.5 percent of men and 31 percent of women are
estimated to have a serious mental illness (Steadman 2009). The DOC’s average daily
population of inmates has remained between 13,000 and 14,000 for much of the past
decade.58 According to the Council of State Governments (2005), the jail systems in
New York, Los Angeles, and Chicago house more persons with mental illness than any
psychiatric facility in the nation. The New York City Department of Health and Mental
Hygiene (DOHMH), Division of Health Care Access and Improvement, Bureau of
Correctional Health Services oversees the delivery of physical and mental health services
to inmates at City jails. DOHMH contracts out for these services and has had a contract
with the current contractor, Prison Health Services, Inc. (PHS) since 2000. PHS utilizes a
clinical staff of roughly 500 clinicians, about 82 of whom provide mental health services.
Since 2003, the City of New York (“City”) has additionally provided discharge planning
services to mentally ill inmates under the terms of the settlement in the matter of Brad H
vs. The City of New York. This class-action lawsuit argued that the DOC functions as a
de facto psychiatric hospital by virtue of the number of inmates with mental illness that it
treats and, as such, must provide comparable aftercare and discharge planning services to
its inmate-patients. Discharge planning services are provided by DOHMH’s Health Care
57

DOC admissions include criminal defendants detained after arrest but before trial, as well as offenders
sentenced to serve incarceration terms in a City jail.
58
City of New York, Department of Correction statistics reported on
http://www.nyc.gov/html/doc/html/stats/doc_stats.shtml, accessed June 16, 2011.

81

Access and Improvement, Forensic Behavioral Health Services Discharge Planning
(“Discharge Planning”) staff. Discharge Planning consists of approximately 70 staff
members, 49 of whom are line staff: 22 caseworkers who assist inmates with applications
for public entitlements, and 27 social workers who create comprehensive discharge plans
that include making appointments and referrals to care post-release. In addition, DOHMH
provides discharge planning for other, non-mentally ill inmates, including those with
substance abuse disorders and significant chronic diseases such as HIV.
Jail-based mental health and discharge planning services are often provided within the
context of relatively short jail stays. About three-quarters of jail releases are pretrial
detainees; release decisions for this population are court- and bail-based, each of which is
driven by judicial and prosecutorial decision-making. While the average length of stay
among detainees regardless of mental status has been about 50 days in recent years,59
one-half of all admissions were released within 14 days. The remaining one-quarter of
inmates are sentenced offenders. Overall, most (about 80 percent) of the Rikers Island
population is released to the community, but a minority of inmates is transferred to other
correctional authorities such as state prisons.
Program Operations 
The process by which inmates receive jail-based mental health services is summarized in
Figure 3.3, and the following description is based on the research team’s interviews with
DOHMH personnel in 2009. Staff described this as the typical experience; deviations
and exceptions are noted separately. Additionally, DOHMH provided the research team
with data on all persons who were arrested in Brooklyn or the Bronx during 2005 and
2006 who were sufficiently mentally ill to receive services under the terms of the Brad H
settlement.60 (The purpose was to form a comparison pool for the analysis of mental
health court impacts.) Our analysis of the data on this segment of DOHMH’s service
population is presented here as well.
Arrest and Admission to City Jails 
Once arrested, suspects are typically held at the precinct, transferred to a central booking
facility, and then arraigned at court. 61 At arraignment, a judge decides whether a suspect
may be released pretrial. Those who are not granted pretrial release (e.g., because they
are judged to be a flight risk) or cannot pay the bail amount are transferred to the DOC

59

City of New York, Department of Correction statistics reported on
http://www.nyc.gov/html/doc/html/stats/doc_stats.shtml, accessed June 16, 2011.
60
That is, persons who were either “deemed” or “designated” as eligible for Brad H discharge planning
services.
61
A small minority of arrestees are taken to the emergency room instead as “emotionally disturbed
persons” (EDP) and thus diverted from this process.

82

ntal Health Services Provided Durring Pretriall Detention
Figurre 3.3. Men

83

for pretrial detention. The majority of inmates under DOC control are incarcerated at the
nine facilities located on Rikers Island. There were between 102,000 and 110,000 inmate
admissions during the 2002 to 2006 study period for the mental health courts evaluation;
this figure was just below 100,000 at the time of study interviews with DOHMH
personnel.62
Referral to Mental Health Services 
All mental health service provision in City jails is referral-based. Referrals to mental
health services are usually made during the intake process, but some referrals arise from
the court during arraignment, inmate self-referral to services, and a small share (about
100 per month) result from inquiries initiated by family members or defense attorneys.
When processing a new admission to a City jail, DOC runs fingerprints, checks criminal
history, assigns security level, and administers a suicide prevention checklist developed
by the New York State Office of Mental Health, the SPSG330, to determine whether
referrals to mental health services for the inmate are necessary. Inmates subsequently
undergo full medical examinations within approximately 24 hours of arrival at a City jail.
Part of the examination includes a 22-item mental status screening conducted by a
physician or physician’s assistant, which may result in a referral to mental health
services. Copies of key screening instruments are included in Appendix E.
Mental Health Services  
Assessment and Evaluation 
Once a mental health referral is made, an initial mental health assessment is conducted by
a licensed clinical social worker, psychologist, or psychiatrist within 72 hours. This initial
assessment may be conducted sooner in urgent situations. Based on this assessment, staff
determines whether follow-up assessment or care is needed and whether an inmate should
be housed in a specialized mental observation unit (MOU) or with the general population.
Staff reported that about 90 percent of referrals are “admitted”63 to the mental health
caseload and, therefore, deemed eligible for discharge planning services under the terms
of the Brad H settlement. Inmates’ book-and-case numbers are marked internally with an
“M,” and service provision is tracked in the Mental Health Discharge Planning Citrix
database. 64
DOHMH provided the research team with 12,299 arrest-event records from this database,
representing all of its encounters with Brooklyn and Bronx arrestees in 2005 and 2006.
62

City of New York, Department of Correction statistics reported on
http://www.nyc.gov/html/doc/html/stats/doc_stats.shtml, accessed June 16, 2011.
63
The term “admitted” is used to describe someone’s entry onto the mental health caseload. This does not
mean they are hospitalized or moved to a mental observation unit.
64
This “M” designation is used internally only for DOC to identify a class member, in order to ensure that
eligible inmates are provided with Brad H-required services. This is not a functional or diagnostic
designation and should not be construed as meaningful for treatment purposes. Furthermore, the “M”
designation is used internally only, and does not appear externally as part of inmates’ book-and-case
numbers (on rap sheets or in court, for example) in order to protect the confidentiality of inmates’ health
information as required by law.

84

These represented 9,493 unique Brooklyn and Bronx defendants who were mentally ill
and served by DOHMH in jail during 2005 and 2006. A sizeable portion of these
individuals (23 percent) were arrested and received mental health services from DOHMH
more than once during the 2005 to 2006 period; in these cases, only their first arrest-event
was included in the analysis.
As shown in Table 3.24, Brooklyn and Bronx inmates with mental illness in the study
data were predominantly male (74 percent). Their average age at arrest was 36. African
Americans comprised the largest share of such inmates (59 percent); Caucasians
comprised 18 percent, while 23 percent were described as being of another race.65 Nearly
4 in 10 (38 percent) were Hispanic, and 95 percent were recorded as English speakers.
Roughly 1 in 10 (8%) had been homeless at the time of arrest.
Table 3.24. Characteristics of Brooklyn and Bronx Arrestees Who Received Mental
Health Services from DOHMH in Jail During 2005-2006
Demographic Characteristics (N=9,493)
Age at arrest (mean)
Male
Black
White
Other race
Hispanic ethnicity
English language ability
Homeless at arrest

35.5 years
74.19%
58.69%
18.27%
23.04%
38.02%
94.59%
8.00%

Note: Data on race, ethnicity, language and homelessness were available for roughly 70 percent of the sample.

Once inmates are “admitted” to the mental health services caseload, a series of follow-up
evaluations and plans must be completed within 14 days, or sooner (within 7 days) if they
are housed in a MOU. First, a psychosocial assessment is conducted by a clinical mental
health social worker under the supervision of a psychologist. A further psychiatric
assessment is conducted for those requiring psychiatric care. A Comprehensive
Treatment Plan Discharge Service Needs (CTPDSN) assessment is then written by the
clinical mental health social worker. Within seven business days of the CTPDSN, a
discharge planning social worker completes a discharge plan. The CTPDSN document
functions as a treatment plan with initial discharge planning recommendations; it includes
the mental health diagnosis, SPMI determination (i.e., whether someone has “serious and
persistent mental illness”),66 functional status, community-based treatment history,
65

Fewer than 1 percent of each were specifically described as Asian or Native American.
“Serious and persistent mental illness" (SPMI) is largely an administrative, legal, or status code and not a
diagnostic term. There are numerous definitions of SPMI in the mental health field, with different
variations of diagnostic and functional criteria. The SPMI determination is primarily for the purpose of
discharge planning, as a SPMI determination triggers certain additional rights under the Brad H stipulation.
DOHMH uses the New York State Office of Mental Health (OMH) definition for the general public, as
specified in the Brad H stipulation. The OMH SPMI definition is used in the community to determine
eligibility for Intensive Case Management (ICM) programs, Assisted Outpatient Treatment (AOT),

66

85

recommendations for treatment in jail, and recommendations for treatment upon release.
Analysis of the Brooklyn and Bronx data in 2005 and 2006 showed that arrestees were
typically67 referred to mental health services within two days of arrest and received an
initial mental health assessment on that same day, as presented in Table 3.25. It took
another nine days from the initial assessment to conduct follow-up assessments and write
a CTPDSN, including a diagnosis, SPMI designation and discharge planning
recommendations. The remaining length of stay in jail after the CTPDSN was completed
was another two months.
Table 3.25. Time to Selected Mental Health Service Milestones
Mental Health Service Milestones (N=9493)
Received initial assessment
Time from arrest to referral (median)
Time from referral to initial assessment (median)
CTPDSN completed
Time from initial assessment to CTPDSN (median)
Released from jail
Time from CTPDSN to release
Eligible, but refused Brad H discharge planning
services

96.22%
2 days
0 days
62.09%
9 days
97.96%
55 days
26.83%

Note that the extent to which mental health needs were assessed and treated in jail
depended on the length of stay. As presented in Table 3.26, the average length of stay for
Bronx and Brooklyn jail inmates with mental illness in 2005 and 2006 was between three
and four months (104 days), but there was considerable variation in the duration of
incarceration. Short stays in jail were common, with nearly one-half of inmates with
mental illness released within one month of arrest: one in five of such jail inmates (21
percent) were released in less than one week, another 10 percent were released within 14
days, while 9 percent were released within 21 days of arrest, and 5 percent were released
within 28 days. While the majority (62 percent) of inmates served by DOHMH had a
CTPDSN completed, more than one-third of the inmates who received initial assessments
were released before the full assessment process and plan were completed; the section
(below) on “alternate release and discharge planning scenarios” provides more
information on the services provided to this sub-group.
supportive housing, and other non-Brad H services. Note, however, that SPMI is neither a diagnostic
definition nor an assessment of functional status. Someone on a psychotropic medication who consistently
takes medication and can function well within the general population may be designated as SPMI, and yet
may be very different in functional capacity from someone else who also is defined as SPMI, but is not
medication-compliant.
67
The statistics given here are medians, as the means were skewed by some extreme outliers. Statistics on
the 75th percentile are similar. Seventy-five percent of mentally ill inmates were referred to mental health
services within seven days of arrest and received an initial mental health assessment on the same day. The
time from initial mental health assessment to a CTPDSN was within 12 days for 75 percent of inmates.
However, at the 75th percentile, inmates remained in jail for another 152 days after the CTPDSN was
written.

86

Table 3.26. Time Served in Jail
Time Served in Jail (N=9493)
Percentage released from jail
Time from arrest to release (mean)
Time from arrest to release (median)

97.96%
104 days
36 days

Mental Health Treatment in Jail 
DOHMH personnel estimated that on any given day approximately 3,300 inmates are on
the mental health caseload, receiving treatment or discharge planning services. Of these,
about 2,500 to 2,800 persons receive medications, while the remainder (roughly 25
percent) receive only therapy or counseling. Inmates are not forcibly medicated.
Between one-quarter to one-third of the mental health caseload is designated SPMI, and
about 900 individuals are housed in a mental observation unit (MOU). Staff reports there
are about 20,000 mental health treatment encounters per month.
Analysis of the data for Brooklyn and Bronx arrestees in 2005 and 2006, presented in
Table 3.27 showed that about one in ten (12 percent) were housed in a MOU, and the
remaining 88 percent were in the general population.68 Mental health diagnoses and
SPMI status were available for 75 percent and 85 percent, respectively, of the inmates
served by DOHMH in jail. As depicted in Table 3.28, the most common Axis I mental
health diagnoses were mood disorders (38 percent), substance-related disorders (24
percent), adjustment disorders (17 percent), psychotic disorders (13 percent) and anxiety
disorders (4 percent). Less than one percent were diagnosed as malingering or
fabricating symptoms. One half (51 percent) were assessed as SPMI.69 The average
functional status of inmates as measured by the Global Assessment of Functioning was a
score of 57 on a scale of 1 to 100, meaning that the average inmate with mental illness
had moderate symptoms and difficulties in social, occupational, or school functioning.

68

Differences between staff estimates of the mental health caseload and the study’s data analysis stem from
a number of legitimate methodological differences. First, staff reports focused on current practices at the
time of study interviews in 2009. By contrast, the data analysis is retrospective and based on a subset of
cases in 2005 and 2006. Second, staff reports reflect the stock inmate population on any given day,
whereas the study’s data collection focused on admissions during the study period. Statistics based on
stock populations tend to reflect those with a longer length of stay, which may be due to more complicated
criminal cases or more severe mental illness. Finally, staff reported on their experience with all mentally ill
inmates, regardless of case jurisdiction. By contrast, the study selected a subset of defendants in Bronx and
Brooklyn cases, who may have systematically different socioeconomic and health characteristics from the
aggregate population.
69
As noted earlier, “serious and persistent mental illness" (SPMI) is largely an administrative, legal, or
status code based on diagnostic and functional criteria, but it in itself is not a diagnostic term. It is also not,
necessarily a determination of functional status. For example, someone on a psychotropic medication who
consistently takes his or her medication, and can function well within the general population, may be
defined as SPMI (and is for the purpose of service provision under the Brad H stipulation), but may be very
different in functional capacity from someone who is not medication-compliant.

87

Table 3.27. Indicators of Mental Health Status
Indicators of Mental Health Status
(varying Ns)
Housed in MH Unit
SPMI or Likely SPMI
Mean Global Assessment of Functioning
(GAF) Score on a scale of 1 (low) to 100
(high)
Received medication in jail

12.23%
51.25%
57
83.24%

Note: The availability of different measures varied in the program database, depending on when the data were
collected in a given inmate’s incarceration. Data on placement in a mental observation unit were available for
9,380 inmates; SPMI status was available for 8,090 inmates (we report on the finalized SPMI status when
available, and the initial “Likely SPMI” assessment in the 29 percent of cases where a confirmation of SPMI
status was not available); Global Assessment of Functioning was available for 7,228 inmates; and medication
usage was available for 4,462 inmates.

Table 3.28. Mental Health Diagnoses
Mental Health Diagnoses Recorded in the Program Data
Axis I Diagnoses
(N=7,121)
Adjustment disorders
17.29%
Anxiety disorders
3.76%
Attention deficit disorders
0.87%
Cognitive disorders
0.20%
Dissociative disorders
0.03%
Eating disorders
0.01%
Mental disorder due to medical
condition
0.01%
Impulse control disorders
1.54%
Mental retardation
0.01%
Mood disorders
37.96%
Motor skills disorders
0.00%
No disorder on given Axis
0.07%
Personality disorders
0.20%
Psychotic disorders
12.96%
Sexual or gender disorders
0.03%
Sleep disorders
0.11%
Somatoform disorders
0.00%
Substance-related disorder
23.61%
Tic disorders
0.01%
Other disorder on Axis II
1.32%

Axis II Diagnoses
(N=4,791)
4.05%
4.80%
0.48%
0.13%
0.02%
0.02%
0.04%
1.65%
0.04%
14.78%
0.02%
0.04%
1.15%
5.87%
0.00%
0.08%
0.02%
63.39%
0.02%
3.40%

Inmates with mental illness housed in the general population are required to have a
minimum of monthly psychiatric visits. More extensive services are provided in the
MOU. Inmates in a MOU typically have a weekly clinical encounter and, at minimum,

88

biweekly psychiatric visits. They are also seen two to three times per day as part of
clinical rounds. Additional services include group therapy, art therapy, music therapy,
and cognitive behavioral therapy. The most extensive services are provided to the
approximately 30 inmates on suicide watch on any given day. These inmates have a
daily clinician encounter and are visited on rounds three times per day.
Staff estimated that about 80 percent of the mental health caseload has co-occurring
substance use problems. By “problems,” they did not necessarily mean dependence, but
rather that inmates engage in substance use. As shown in Table 3.29, analysis of the data
on Brooklyn and Bronx arrestees in 2005 and 2006 showed that nearly 6 in 10 (58
percent) had co-occurring substance use conditions, defined by the research team as
having a substance-related diagnosis on either Axis I or Axis II. Inmates with purely
substance-related issues are not admitted to mental health services, but instead are
provided substance abuse treatment services. There are currently five substance abuse
treatment units. Each unit houses between 40 and 50 inmates at any given time. Services
include inpatient therapeutic communities, cognitive behavioral therapy groups, as well
as court advocacy and discharge planning services.
Table 3.29. Substance Abuse Diagnoses
Substance Abuse Recorded on Either Axis I or Axis II (N=7,159)
Any substance-related diagnosis on Axis I or II
Alcohol
Cocaine
Heroin
Marijuana
Other*
Unknown substance type

57.54%
6.31%
9.00%
11.08%
3.53%
23.36%
9.60%

*Note: “Other” includes and was frequently polysubstance use.

Discharge Planning Services In Jail 
The Brad H settlement requires discharge planning for each inmate who is receiving
certain psychotropic medications in jail or received more than two mental health service
visits. Required elements of discharge planning include submitting applications for
entitlements and benefits (e.g., Medicaid, Public Assistance, Food Stamps, Medication
Grant Program), making referrals or appointments to community-based mental health
services, and providing medications upon release (a 7-day supply of medications plus
prescriptions to ensure an additional 21-day supply). Inmates who are identified as SPMI
must be given additional assistance with supportive housing applications, SSI
applications (if they are sentenced offenders as opposed to detainees), and transportation
upon release from jail; they also are referred to community-based programs that provide
assistance to SPMI adults in acquiring psychiatric care and other entitlements and
services including but not limited to mental health, drug and alcohol treatment, and
medical care.

89

Discharge planning begins soon after the initial mental health screening, but the bulk of
discharge planning is done once the CTPDSN is completed. A formal discharge plan
must be written within seven days of the CTPDSN, including the following elements:
•

The inmate’s SPMI status.

•

Housing information.

•

The status of various public assistance applications.

•

Referrals or appointments for community-based mental health services. This
must also include a justification for the choice of referral.

•

An aftercare letter to the client outlining the discharge plan. The letter includes
the name of the community provider, a contact person and their telephone and
address information (if appropriate), the inmate’s mental health diagnosis, and
medications needed upon release.

Inmates who are still incarcerated 30 days after the treatment plan is written (14 days if
housed in a MOU) undergo a treatment plan review, and the discharge plan may be
revised as needed. The treatment plan is periodically reviewed in this manner for the
duration of a person’s jail incarceration.
Release and Aftercare 
Most jail inmates are released to the community. Staff estimated that about 500 inmates
with mental illness are released every month, one-third of whom are SPMI. On the day of
release, inmates are provided discharge planning. The discharge planning staff will try to
convert service referrals to actual appointments. Inmates also are provided with “walking
medications,” (i.e., a 7-day supply of medications) and prescriptions to obtain an
additional 21-day supply of medicines. Inmates who have been identified as SPMI are
provided with van transportation to their place of residence, shelter, or program.
DOHMH also has established two programs to help released inmates with mental illness
obtain continuity of care in their home communities. The SPAN and LINK programs,
described below, are networks of community-based providers with whom the City
contracts to provide services to released inmates with mental illness.
The SPAN program was established to provide services to inmates who were released
from custody before discharge planning was completed, and who want to complete the
process once they return to the community. Brad H class members are eligible for SPAN
services for 30 days post-release. SPAN provider offices are generally located in each
borough within one-half mile of the court.70
70

As of June 2011, the SPAN provider for all five boroughs is the Bowery Residents’ Committee.

90

Persons who were identified as SPMI are additionally referred to LINK71 program
providers located in each of the boroughs. For those clients who were assessed as SPMI
in jail, and who consented to receive services, Discharge Planning staff will provide
medical records and assessments to the LINK program to provide LINK services for six
months post-release.72 Discharge planning staff also must follow up 30 days post-release
to determine whether the discharged SPMI inmates kept their appointments.
Alternate Release and Discharge Planning Scenarios 
Many inmates who qualify as part of the affected class of the Brad H action have
relatively short stays in jail. Data obtained from the DOHMH show that nearly four in ten
(38 percent) of Bronx and Brooklyn arrestees in 2005 and 2006 were released before the
CTPDSN was completed. DOHMH has a difficult task in providing the Brad H
settlement-required discharge planning process described above to individuals with
relatively short lengths of stay. DOHMH staff emphasized that release decisions are
court- and bail-based, and are independent from mental health service provision. Staff
indicated that inmates with shorter stays require a different approach to discharge
planning; at the same time, it is difficult to predict when a given inmate will be released.
Inmates released before the Brad H specified process can be completed receive
abbreviated discharge planning services in jail.
The average length of stay for this sample of Bronx and Brooklyn jail inmates with
mental illness in 2005 and 2006 was between three and four months (104 days), but there
was considerable variation in the duration of incarceration. As previously described,
short stays in jail were common, with nearly one-half of inmates with mental illness
being released within one month of arrest. In fact, one in five such jail inmates (21
percent) were released in less than one week, and another 19 percent were released within
21 days of arrest. The extent to which mental health needs were assessed and treated in
jail depended on the length of stay. This sample of Bronx and Brooklyn inmates were
typically73 referred to mental health services within two days of arrest and received initial
mental health assessments on that same day. It took another nine days from the initial
assessment to conduct follow-up assessments and write CTPDSNs, including diagnoses,
SPMI designations, and discharge planning recommendations. However, more than onethird of those who received initial assessments were released before the full assessment
71

LINK is not an acronym, but the program name is capitalized. As of June 2011, the LINK providers are:
1) Federation Employment Guidance Services, Inc. (“F.E.G.S.”), which covers Manhattan; 2) FordhamTremont Community Health Center, which is a part of St. Barnabas Hospital, and covers the Bronx; 3)
Volunteers of America–Greater New York, which covers Queens; and 4) the Education and Assistance
Corporation, which covers Brooklyn and Staten Island.
72
The original Brad H settlement specified a time period of two years for LINK services, but the period for
LINK services has since been amended to six months.
73
The statistics given here are medians, as the means were skewed by some extreme outliers. Statistics on
the 75th percentile are similar. Seventy-five percent of inmates with mental illness were referred to mental
health services within seven days of arrest, and received initial mental health assessments on the same day.
The time from initial mental health assessment to a CTPDSN was within 12 days for 75 percent of inmates.
However, at the 75th percentile, inmates remained in jail for another 152 days after the CTPDSN was
written.

91

process and plan were completed. Those who remained in jail until the CTPDSN was
completed remained in jail for another two months until release. The net result is that the
one-half of inmates with mental illness who were released within one month of arrest,
and particularly those who were released within two weeks, often received incomplete
services. For example, less than one-quarter of the inmates released within two weeks of
arrest had a CTPDSN completed compared to more than 90 percent of inmates who were
in jail for more than one month.
Individuals who are released before discharge planning can be completed are eligible for
SPAN services within 30 days of release. SPAN services include providing prescriptions;
assistance with Medicaid applications and public assistance applications; Medication
Grant Program; and transportation to home, shelter, or mental health service providers;
and escorts to appointments. DOHMH contracts with SPAN providers to assist in the
provision of discharge planning services for up to 3,500 individuals in the community.
However, DOHMH personnel reported that this is a voluntary service that few former
inmates actually request.
The most severely ill inmates are sometimes transferred to secure psychiatric wards at
Bellevue and Elmhurst hospitals. This is a relatively small portion of the mental health
caseload; approximately 50 inmates with mental illness are under the care of these
hospital wards on any given day. Additionally, inmates whose competency to stand trial
is in question are often released to a state mental facility. In these cases, discharge
planning is completed by the hospital.
Finally, a minority of inmates, roughly 20 percent, are transferred to other correctional
authorities. For example, pretrial detainees who are subsequently convicted may be
“released” to a New York State prison to serve their sentences. In these cases, medical
and mental health aftercare summaries are transferred to the new institution.
Barriers to Continuous Care in the Community 
The effectiveness of these linkages to community-based services has been a point of
concern for DOHMH and advocates alike. Relatively few released inmates remain in
prolonged, continuous care in the community. It is beyond the scope of this study to
evaluate the effectiveness of jail-based services, but we offer some observations on
potential barriers to care in the community.
Lack of client interest and motivation is one factor. Data obtained from DOHMH for
Brooklyn and Bronx arrestees in 2005 and 2006 show that 27 percent of the mental health
caseload refused discharge planning services. Receipt of discharge planning services is
voluntary. If an inmate is housed in the general population and declines services, there is
no requirement to force the inmate to accept services or to re-offer them. For inmates in
a MOU, however, DOHMH staff must re-offer services three times, twice by a social
worker and, if the inmate refuses the two re-offerings, a third time by a supervisor.
Additionally, there has to be a declaration of whether the person is "capable" of declining
the discharge plan.

92

DOHMH personnel reported that very few clients follow up with their discharge plans.
This is a concern among Discharge Planning staff. There are many potential reasons for
this lack of follow-up, including, but not limited to, a lack of interest on the part of the
former inmate. In discussions with inmates, DOHMH found that released inmates had
other, higher priorities that competed with getting mental health treatment, including
reconnecting with family members and obtaining public benefits. Additionally, some
community-based providers may experience more challenges in working with criminal
justice clients.
The uncertainly of inmates’ release dates also posed a challenge for Discharge Planning
staff. Because release dates are determined by factors beyond the control of jail-based
health providers (i.e., bail and court decisions), Discharge Planning staff reported that
they cannot know how much time they will have with an individual inmate in order to
fully process entitlement applications or make referrals to services.
Although, SPAN services were designed to assist Brad H class members who were not
incarcerated long enough to have discharge plans developed, the utilization of the service
has been low. DOHMH staff also noted challenges with successfully following up with
homeless clients, and spoke of the difficulties in locating people in the shelter system.
Clients who do access LINK services are reassessed by the LINK program, which then
makes a decision about whether they will work with a referred client. Upon reassessment,
the LINK program may decide that a person is not really SPMI, and there is then a
process to renegotiate the diagnosis. DOHMH staff and others interviewed for this study
noted that determination of SPMI status can be subjective.74 DOHMH noted that jailbased diagnoses may err on the side of overestimating the severity of illness, partly
because patients’ symptoms tend to be worse in jail (possibly due to the stress of
incarceration), and because clinicians want to ensure that clients can access needed
services.
In sum, continuity of care for jail inmates with mental illness remains a difficult issue.
While individuals receive mental health care and services in jail, follow-up in the
community remains a challenge for this population.
Collaboration and Coordination with Mental Health Courts 
The capacity of the Brooklyn and Bronx mental health courts is small relative to the total
number of defendants with mental illness. These two mental health courts serve a small
slice of the greater universe of defendants with mental illness. Cross-referencing the
population of Brooklyn and Bronx arrestees who received mental health services in jail
with the two court programs, we found that just 6 percent of jailed defendants with
mental illness had ever been referred to either the Brooklyn or Bronx mental health

74

DOHMH staff also noted that “serious and persistent mental illness" (SPMI) is largely an administrative,
legal, or status code and not a diagnostic term. There are about 22 definitions of SPMI with different
variations of diagnostic and functional criteria.

93

courts, and 4 percent of jailed defendants with mental illness subsequently participated in
one of those courts.75
Collaboration between the mental health courts and DOHMH consists mainly of the
transfer of medical and discharge planning records from the jail to the MHCs. The
Brooklyn and Bronx mental health courts sometimes request treatment records and
diagnostic information from DOHMH as part of its larger process of evaluating potential
candidates for mental health court (described earlier in this chapter). DOHMH cooperates
with these requests by providing the medical record with psychiatric and discharge
planning information.
In general, however, the court system is not always aware of an individual’s mental
health status because of the various federal, state and local privacy laws in place to
protect the confidentiality of inmates’ medical conditions and records. Those defendants
who come to the attention of the mental health courts are individually referred by their
defense counsel, prosecuting attorneys, and judges, as described earlier in this chapter.
Although the book-and-case numbers of inmates with mental illness are internally
marked with an “M”, this is done only to provide DOC information on whom they must
provide Brad H settlement-required services. No indications of inmates’ mental health
status appear on court calendars, rap sheets, or other documents available to court
personnel. Again, this is done in order to preserve the confidentiality of the medical
condition of the inmate as required by various laws.
There also is no systematic process by which DOHMH staff would know when a mental
health court client has been remanded to jail. They are reliant on the mental health courts
to inform them when someone is remanded. It stands to reason that remands from mental
health court would undergo DOC and DOHMH’s intake, screening, and assessment
processes upon admission to jail. However, the recent introduction of electronic health
records in the City jails is expected to enhance efficiency and quality of care by making
the longitudinal health record of an inmate available to all medical and mental health
providers in the jail system. Furthermore, a long term goal of DOHMH’s development of
an electronic medical records system is to ease the exchange of medical information
between jail- and community-based providers (with appropriate patient consent and
authorization), thus facilitating improved efficiency, continuity of care, and health
outcomes. DOHMH staff reported that the electronic medical records system has been
developed in line with the provisions of the recently enacted health care reform law,76
and that the agency is working to enter into agreements with local Regional Health
Information Organizations (“RHIOs”) comprised of area hospitals and community-based
health care providers.
75

These statistics were generated by matching individuals among the 9,493 inmates treated by DOHMH in
2005 to 2006 to the sample of all individuals referred to the Brooklyn and Bronx MHCs between 2002 and
2006. Of the 9,493 individuals served by DOHMH in 2005 and 2006, 218 had been referred to the
Brooklyn Mental Health Court, including 145 who participated. Similarly, of the 9,493 individuals served
by DOHMH in 2005 and 2006, 333 had been referred to the Bronx Mental Health Court, including 153
who participated.
76
The Patient Protection and Affordable Care Act of 2010.

94

How Do Mental Health Court Defendants Compare to the Broader Universe of 
Defendants With Mental Illness? 
Anecdotal evidence from mental health stakeholders suggests that MHC participants may
be systematically different from the “business-as-usual” jail population with mental
illness. Stakeholders felt that mental health court participants would be among the
defendants with more severe mental illness, predicting they would have longer lengths of
stay in jail.
We examined the subset of MHC court participants within the broader population of jail
inmates who received DOHMH services while incarcerated in order to empirically
examine these assertions. In short, our analysis corroborates stakeholder perceptions that
mental health court participants are more severely ill, but does not support the assertion
that mental health court participants served longer terms in jail. Detailed similarities and
differences are described separately for the Bronx and Brooklyn MHC programs, below,
and these findings were used to develop methods for selecting appropriate comparison
cases for the evaluation of mental health court impacts.
Bronx Mental Health Court 
We conducted a similar analysis of the Bronx mental health court. Of the 9,493 inmates
with mental illness treated at Rikers Island in 2005 and 2006, we found that 203 were
referred to the Bronx MHC in 2005-2006; ultimately, 153 of these individuals
participated in the mental health court. We conducted bivariate t-test comparisons
between this subset of Bronx MHC participants and other jail inmates with mental illness,
and found that they differed with respect to many demographic, mental health, and
substance use characteristics. These differences were statistically significant at the 95
percent confidence level.
The Bronx MHC participants who were in jail during 2005 and 2006 were more likely to
be female, Hispanic, and older compared to other jail inmates with mental illness.
Approximately four in ten of the Bronx MHC participants were female. By contrast,
about one-quarter of all the jail inmates served by DOHMH were female. More than onehalf of Bronx MHC participants were Hispanic, compared to roughly four in ten in the
entire jail population. Additionally, the Bronx program participants in jail were older,
with an average age of 39, compared to the average age of 35 among the total jail
population with mental illness.
The Bronx MHC participants who received mental health services in jail were more
severely mentally ill than other inmates. This was reflected across a range of measures
including: medication use in jail (96 percent compared to 83 percent), SPMI status (75
percent were assessed as seriously and persistently mentally ill, compared to 50 percent),
and the Global Assessment of Functioning (an average score of 54 versus 57). On other
metrics, however, Bronx MHC participants were similar to other jail inmates with mental
illness. They had a similar likelihood of pre-jail homelessness and were as likely to have
been housed on a mental observation unit while incarcerated.

95

With respect to mental health diagnosis, the Bronx MHC participants in jail in 2005-2006
were similar to other inmates in some respects, but different in others. Axis I mood
disorders were the most common disorder type in both groups, assessed in about four in
ten persons. Substance-related Axis I mental disorders (e.g., substance-induced mental
health symptoms) also were common in both groups, assessed in about one-quarter of
persons. Taking Axis II substance disorders into account as well, however, we found that
substance-related problems were more prevalent among the Bronx mental health court
participants in jail; 66 percent had primary or secondary substance use disorders (Axis I
or Axis II), as compared to 57 percent of the business-as-usual population. Axis I
psychotic disorders also were more prevalent in the Bronx MHC group (23 percent
compared to 13 percent), whereas business-as-usual inmates with mental illness were
more likely to have been diagnosed with Axis I adjustment disorders (18 percent, as
compared to 2 percent of the Bronx MHC participants). In short, the Bronx mental health
court population was more likely to have psychotic illnesses and substance-related
problems.
Those Bronx MHC participants who were in jail in 2005-2006 had a shorter length of
stay in jail compared to other inmates with mental illness, with an average of 77 days
(i.e., two to three months) as opposed to 104 days (i.e., three to four months). Despite
this shorter length of stay, there were no appreciable differences in the receipt of jailbased assessment and discharge planning services. Interestingly, inmates who were
referred to the mental health court, but did not participate, had the longest length of stay
in jail, 167 days (i.e., five to six months).
Brooklyn Mental Health Court 
Of the 9,49377 inmates with mental illness treated at City jails during 2005 and 2006, 132
were referred to the Brooklyn MHC in 2005-2006; ultimately, 77 of these participated in
the Brooklyn MHC program. We conducted bivariate analyses using T-tests and analysis
of variance to identify statistically significant differences (at the 95 percent confidence
level) between those who participated in the Brooklyn MHC and those whose cases were
adjudicated as usual.
More than one-quarter of the Brooklyn MHC participants in jail during 2005-2006 were
Caucasian, making them more likely to be White and non-Hispanic compared to other jail
inmates with mental illness. However, they were similar to other inmates with respect to
gender, age, English language ability, and homelessness.
The Brooklyn MHC participants treated at Rikers Island were more severely ill than other
inmates served by DOHMH on a number of metrics. They were more likely to be
classified as SPMI (76 percent compared to 51 percent), more likely to have received
medication in jail (93 percent vs. 83 percent), and more likely to have been housed in a
MOU (31 percent vs. 12 percent). Their average score of 54 on the Global Assessment of

77

As noted in Chapter 2, missing information reduced the pool of eligible cases for consideration in the
impact analysis to roughly 5,000.

96

Functioning also was statistically significantly lower than the average of 57 among other
inmates.
With respect to mental health diagnoses, the Brooklyn MHC participants treated at Rikers
Island were similar to other inmates in that they were most commonly (about 40 percent)
diagnosed with mood disorders while in jail. However, Brooklyn program participants
were more likely to have been diagnosed with psychotic disorders (25 percent compared
to 13 percent) and less likely to have been diagnosed with an adjustment disorder (6
percent compared to 17 percent of other jail inmates).
Roughly one-fifth of the Brooklyn MHC participants treated at Rikers Island were
diagnosed with an Axis I substance-related disorder while in jail, which was comparable
to other jail inmates with mental illness. Taking Axis II substance use into account as
well, we found that 48 percent had some co-occurring substance abuse compared to 58
percent of other inmates with mental illness. This difference, however, is of marginal
statistical significance (p=.1057).
Brooklyn MHC participants who were in jail in 2005 and 2006 were incarcerated for an
average of 111 days, which was not significantly different from the length of stay by
other inmates with mental illness (103 days). In keeping with this finding, MHC
participants were comparable to other such inmates with respect to the jail-based
assessment and discharge planning services they received. Interestingly, it was
nonparticipants in the mental health court (i.e., persons referred to the mental health court
who did not participate) who spent the longest time in jail, with an average length of stay
of 188 days.
Conclusions About Mental Health Court Participants and Other Jail Inmates With 
Mental Illness 
This analysis has identified some commonalities among MHC participants, and
differences from the broader population of jail inmates with mental illness. In brief, our
analyses indicate that MHC participants were comparable to other such jail inmates with
respect to: 1) diagnosis of mood disorders, 2) receipt of jail-based assessment and
discharge planning services, and 3) length of stay in jail—that is, MHC participants had
similar or shorter lengths of stay than other jail inmates with mental illness. Conversely,
MHC participants and individuals processed under standard conditions differed on: 1)
demographic characteristics, 2) severity of mental illness whereby MHC participants
evidenced more severe mental illness diagnoses, including the presence of psychotic
illnesses, and 3) co-occurring substance abuse issues, with MHC participants in the
Bronx (only) presenting more than Brooklyn or business-as-usual cases.

97

CHAPTER 4. IMPACT ANALYSES AND FINDINGS 
Analytic Overview 
As previously indicated, we conducted separate evaluations of the Brooklyn and Bronx
Mental Health Courts, assessing each program’s impacts on recidivism outcomes as
measured by re-arrest and re-conviction. Based on administrative records, we performed
a matched comparison analysis to determine the extent to which MHC participation
affected individual likelihoods of recidivism during a minimum 30-month follow-up
period. The impact analysis focuses on developing a plausible comparison between
arrestees who participated in a MHC program and matched arrestees in jail who were
diagnosed with mental disorders.
Each MHC identified individuals who had been referred or admitted to MHC. The New
York City (NYC) Department of Health and Mental Hygiene (DOHMH) provided data
on all arrestees in NYC who were diagnosed with mental illness. Based on official
criminal history data, analytic databases were constructed to measure a variety of factors,
such as demographic characteristics, criminal history, mental disorders, and drug use
among arrestees. It is important to note that each source of data provides different levels
of data completeness and consistency. For our impact evaluation, it was necessary to rely
on consolidated data to pool information from different sources. The final sample size
and the list of variables available for impact analyses were therefore slightly different
from those in the descriptive analyses reported in Chapter 3.
The impact analysis employs propensity score matching (PSM) methods to “match”
individuals in each MHC and its respective comparison group as closely as possible. As
an alternative to randomized experiments, PSM has widely been used with observational
data to estimate program impacts through sampling comparable treatment and control
cases from a larger pool of such cases (Rosenbaum and Rubin 1983, 1984). In this
approach, the matched comparison represents a counterfactual of what would have
happened to the treatment group had they not received treatment. The extent of program
impact on recidivism is assessed by comparing the difference in the outcome of interest
between the treatment and comparison groups.
Measures  
The impact analyses focus primarily on recidivism. Re-arrest and re-conviction were
measured as indicators of recidivism in this study. Unfortunately, other outcome
measures such as employment and mental health problems were not available for our
impact analyses. For MHC participants, we assessed whether or not an arrestee had a
subsequent arrest or conviction after being admitted to MHC. For arrestees in jail
serviced by DOHMH, we measured whether or not an arrestee had a subsequent arrest or
conviction after the initial mental health diagnosis, after which point they would be
exposed to the risk of reoffending.78
78

Some individuals in the comparison group were detained longer than others while awaiting the legally
mandated assessment at the jail. Based on detention status, our initial analyses compared different sampling

98

For explanatory variables (covariates), there are four broad domains examined in the
impact analyses. First, we measured the baseline demographic characteristics of arrestees,
including race, gender, and age at arrest. Second, mental health conditions and drug use
history were measured. We examined a series of diagnostic indicators of mental disorder
assessed by clinicians and self-reported drug use indicators. Third, we measured the
characteristics of instant offense for which arrestees were referred to mental health courts
or DOHMH. Fourth, the criminal history of study participants was examined in great
detail. Not only did we assess how many times an individual committed different types of
offenses in the past, the impact analyses also examined when they were arrested for the
first time, if they started a criminal pathway as a violent offender or drug offender, and
how many different types of offenses they ever committed. These measures reflect the
extent, severity, and character of one’s criminal history. Some of these explanatory
variables were not commonly available for both Brooklyn and Bronx MHC participants.
As such, the Bronx and Brooklyn evaluations were based on slightly different data
specifications.79
Analytic Strategies 
The most desirable feature of randomized experiments, if executed properly, is that one
can simply compare those who received treatment to those who did not, and ascribe the
difference in the outcome of interest between the two groups to treatment. In
observational data, such a comparison does not necessarily bear a causal interpretation
due largely to sample selection bias. To the extent that treatment and comparison cases
are different in their characteristics, one cannot determine whether the difference in the
outcome, if any, is due to treatment or the systematic difference in the characteristics of
study subjects.
We have thus employed propensity score matching methods to mimic the framework of
an experimental design by which to interpret the difference between the treatment and
comparison groups as a causal effect of treatment.80 Matching involves pairing treatment
protocols for the donor pool of comparison cases. Findings were consistent across different sampling
frameworks.
79
Because of this limitation, pooling data from both MHCs would restrict data to a reduced set of variables
commonly available in both MHCs. This would increase the chance of hidden bias in propensity score
matching analysis. As the primary objective of our impact analysis is to develop less equivocal, more
convincing evidence regarding the effectiveness of two MHCs, a pooled analysis is therefore not pursued in
our evaluation.
80
Another conventional approach to propensity score analysis is regression adjustment or stratification
based on propensity scores. The key difference between matching and stratification is that matching adjusts
for differences between the treated and untreated via study design, whereas stratification involves grouping
study subjects into strata based on propensity scores and executing separate analyses within each stratum.
One can think of stratification as a coarse form of matching in that the treated are matched to the untreated
as a group within each stratum. Although stratification has long been used in the literature to remove
selection bias (Cochran 1968; Rosenbaum and Robin, 1983), one of the critical disadvantages of
stratification guides against its implementation in this project. Stratification may yield tenuous results when
a stratum is disproportionately or exclusively comprised of either treated cases or untreated cases. On the
contrary, the matching approach requires that unmatched cases be removed from analysis so that the
comparison between the treated and untreated is balanced and uncompromised in terms of internal validity.

99

and comparison units that are similar in terms of their observed characteristics (Dehejia
and Wahba 1998; Rosenbaum and Rubin 1983, 1984). By “balancing” the characteristic
differences between treatment and comparison units, one can ascribe the difference in the
outcome of interest to treatment. Our impact analyses employed an extensive list of
variables, including demographic characteristics, mental health conditions, and prior
criminal history to pair treatment and comparison units. Several dozens of observed
covariates were used in a logistic regression model to estimate the probability of being
admitted to MHC.81 The predicted probability from this selection model is known as
propensity scores.
Using the propensity scores, we performed a one-to-one match between arrestees who
participated in MHC and arrestees in jail diagnosed with mental disorders.82 When the
treatment unit encountered more than one candidate for matching who have the exact
same propensity scores, one of the candidates was chosen at random. When the candidate
for matching is quite different from the treatment unit in terms of case characteristics
measured by the propensity score, the treatment unit was left unmatched and removed
from the analysis.83 Limiting the scope of analysis to those who can surely be matched
facilitates the development of an analytic framework that resembles that of randomized
controlled trials. These principles in propensity score matching can yield a higher level of
internal validity in the results of the impact analysis.

81

A list of covariates examined in data analysis is provided in Appendix G.
We have conducted the 1-to-1 match with and without replacement. With replacement, untreated cases
from the donor pool of arrestees can be used more than once for matching. If matching is implemented with
replacement, there will be, presumably, a greater number of matched pair sets and improved balance
between treatment and control cases. However, to the extent that a single control unit is repeatedly used for
matching over and over again, matching with replacement may become a problem. Because the primary
objective of our propensity score matching analysis is to construct a comparison group equivalent to the
treatment group, we present the results from the 1-to-1 match with replacement in this report. However, it
should be noted that our comparison on the results from matching with and without replacement yielded
relatively little difference. We also conducted 1-to-many matches. In principle, the 1-to-1 match would
yield the highest balance and the highest level of internal validity. As the second best match and the third
best match (and the next best match) to the treated are included in analysis, it only widens the overall
difference between the treated and untreated. The practical advantage of having multiple control units per
treated case is the increased sample size. Again, having discovered little difference between the results
from 1-to-1 and 1-to-many matches, we report the results from the 1-to-1 match in this study.
83
This can be implemented by enforcing “common support” and “caliper” in matching. In propensity score
matching, there should be a substantial overlap between the treated and untreated in terms of estimated
propensity scores. This overlap is known as “common support.” In practice, treated cases with very high
propensity or untreated cases with very low propensity tend to be outside the region of common support.
Such cases should be excluded from the comparison under most circumstances. This is one of the key
strengths of matching as it makes transparent the need for common support across the treated and untreated.
To ensure the similarity among paired cases, one can also specify a value for the maximum distance,
known as caliper, between the treated and untreated in a matching estimator. We have experimented with
various matching estimators, including nearest neighbors, kernel, and Mahalanobis matching, with varying
calipers (Leuven and Sianesi 2003). Although various matching parameters yielded slightly different results
across the models, the overall findings regarding the effectiveness of MHCs remain broadly consistent. In
this report, we present findings from the most parsimonious, less equivocal specification that is also
relatively less prone to data loss.
82

100

Results from the Bronx Mental Health Court  
Development of Comparison Groups 
The initial Bronx Mental Health Court sample was comprised of 815 persons referred to
the court. Of these, 648 arrestees participated in the program between January 1, 2002
and December 31, 2006. Individuals referred to the program who did not participate
(N=167) were not considered for inclusion in the impact analysis. Of the 648 eligible
treatment subjects, we matched 564 to arrestees in jail diagnosed with mental disorder, as
shown in Table 4.1. The matching was conducted on various measures, including
demographic characteristics of arrestees, prior criminal history, instant offense
characteristics, and drug use.84
A fraction of the MHC treatment group had the estimated chance of receiving treatment
too high to be matched to anyone in the DOHMH sample and was therefore removed
from analysis. The final sample for analysis includes 1,128 observations. Focusing on the
matched set of treatment and comparison groups provides greater confidence in the
results. Although the removal of unmatched cases could compromise the external validity
of findings, the credibility of research evidence assessing program effectiveness usually
deserves higher priority in any single evaluative research.
Table 4.1 shows descriptive statistics of key variables before and after matching. The
unmatched treated group refers to the 648 treatment subjects who participated in the
Bronx mental health court program, 564 of whom were matched to an equivalent
comparison subject. The unmatched comparison group refers to the donor pool of
arrestees in the DOHMH sample. The matched comparison group includes 564 matched
comparison subjects.
The last two columns show the t-statistic and significance level on a given variable. A
few notable observations emerging from Table 4.1 include that the treatment and
comparison groups are substantially similar to each other after matching. With rare
exceptions, the difference between the treatment and comparison groups is statistically
84

The literature on propensity score analysis offers conflicting perspectives on the issue of selecting
covariates to be included in the propensity score estimation. Although there is a general agreement that
covariates should have some theoretical bearing on treatment assignment as well as outcome measures,
differing opinions exist as to the roles that theory and statistics should play in the decision for covariate
selection. On the one hand, Rosenbaum (2002:76) notes that adding more covariates would cause no harm
especially when they are powerful predictors of treatment assignment. Similarly, Diamond and Sekhon
(2010) develop a search algorithm to iteratively improve the balance between the treated and untreated
without paying much attention to a theoretical motivation to impose any particular relationship between
covariates and outcomes. On the other hand, Pearl (2009) cautions that the “experimentalist” approach of
including covariates in the propensity score estimation based on statistical properties of the covariates may
actually amplify bias. In our approach to balancing the difference between treatment and comparison
samples, we attempted to include as many covariates as possible. Since administrative records of subjects
constitute a rather limited source of information to begin with, it did not seem reasonable to impose a
further restriction on the use of already limited data. Moreover, all the covariates examined in this study
were of theoretical relevance to recidivism, suggesting no compelling reason for exclusion.

101

indistinguishable from zero. For example, 35 percent of the Bronx treatment group
(matched) and 31 percent of the comparison group (matched) were black. This difference
of 4 percentage points has a t-statistic of 1.39, which does not reach statistical
significance. Before matching, however, the DOHMH sample (unmatched) shows a
much higher proportion of black arrestees (55 percent). Similarly, the matched treatment
and comparison groups have an average of 1.34 and 1.33 prior violent felony arrests,
respectively, while the DOHMH sample shows an average of 2.44 prior violent felony
arrests.
Table 4.1. Descriptive Statistics of Key Measures
Variable
Black

Sample
Treated
Unmatched
0.34
Matched
0.35

Comparison
0.55
0.31

t
-9.91
1.39

p>|t|
0.00
0.16

White

Unmatched
Matched

0.07
0.07

0.08
0.07

-0.54
0.00

0.59
1.00

Hispanic

Unmatched
Matched

0.58
0.57

0.35
0.61

11.17
-1.39

0.00
0.17

Male

Unmatched
Matched

0.62
0.62

0.66
0.61

-1.92
0.06

0.06
0.95

Age at Arrest

Unmatched
Matched

37.04
36.79

34.46
36.93

5.83
-0.24

0.00
0.81

Number of Prior Arrests

Unmatched
Matched

13.50
13.17

16.18
13.00

-4.25
0.23

0.00
0.82

Age at 1st Arrest

Unmatched
Matched

24.53
24.26

21.48
23.78

9.35
0.94

0.00
0.35

N of Any Violent Felony
Offenses

Unmatched
Matched

1.33
1.34

2.44
1.33

-9.49
0.13

0.00
0.90

N of Firearm-related
Offenses

Unmatched
Matched

0.19
0.21

0.43
0.20

-6.89
0.11

0.00
0.91

N of Drug-related Offenses

Unmatched
Matched

6.67
6.46

5.69
6.26

3.74
0.63

0.00
0.53

102

Table 4.1. Descriptive Statistics of Key Measures (cont.)
Variable
N of Property Offenses

Sample
Treated
Unmatched
2.88
Matched
2.78

Comparison
4.33
2.71

t
-4.36
0.23

p>|t|
0.00
0.82

N of Public Order Offenses

Unmatched
Matched

3.29
3.29

4.65
3.22

-5.96
0.29

0.00
0.77

1st Offense as Robbery

Unmatched
Matched

0.09
0.10

0.17
0.10

-4.67
0.30

0.00
0.77

1st Offense as Burglary

Unmatched
Matched

0.11
0.10

0.10
0.13

0.64
-1.41

0.52
0.16

1st Offense as Assault

Unmatched
Matched

0.09
0.09

0.12
0.09

-2.55
-0.42

0.01
0.68

1st Offense as Drug Sale

Unmatched
Matched

0.12
0.12

0.06
0.10

6.01
1.13

0.00
0.26

Alcohol Use

Unmatched
Matched

0.02
0.02

0.06
0.03

-4.35
-0.77

0.00
0.44

Cocaine Use

Unmatched
Matched

0.10
0.11

0.10
0.13

0.11
-1.18

0.91
0.24

Marijuana Use

Unmatched
Matched

0.05
0.05

0.07
0.04

-1.90
1.31

0.06
0.19

Heroin Use

Unmatched
Matched

0.16
0.16

0.11
0.13

3.91
1.28

0.00
0.20

Instant Offense: Violent
Felony

Unmatched

0.09

0.24

-8.68

0.00

Matched

0.10

0.10

-0.10

0.92

Unmatched
Matched

0.02
0.02

0.06
0.03

-4.67
-1.36

0.00
0.17

Instant Offense: Firearmrelated

103

Table 4.1. Descriptive Statistics of Key Measures (cont.)
Variable
Sample
Treated
Instant Offense: Drug-related Unmatched
0.77
Matched
0.75

Comparison
0.38
0.76

t
19.19
-0.14

p>|t|
0.00
0.89

Offense Variety Score85

Unmatched
Matched

4.72
4.65

5.36
4.76

-6.51
-0.75

0.00
0.45

Propensity Score

Unmatched
Matched

0.50
0.46

0.07
0.46

64.57
0.00

0.00
1.00

It is clear that the DOHMH sample (arrestees serviced by DOHMH) has characteristics
distinctive from those of the treatment group.86 The DOHMH arrestees tend to be
younger, have more prior arrests, started criminal careers as violent offenders, and have
engaged in different types of offenses. Table 4.1 shows that these differences between
the DOHMH and treatment groups reduced significantly after matching.
Second, there exists little difference between matched and unmatched treatment subjects.
For example, the age at arrest is 37.04 for those who participated in the Bronx MHC
program (unmatched) and 36.79 for the subset of the treatment subjects who were
matched. This implies that the consequence of removing the unmatched treatment cases
in subsequent analyses would not distort the overall characteristics of MHC participants.
Third, the estimated propensity score is substantially higher for the MHC sample (50
percent chance of receiving treatment) than DOHMH sample (7 percent chance of
receiving treatment). After matching, however, the propensity score is nearly identical for
both groups (46 percent chance of receiving treatment). The matching procedure resulted
in remarkable performance in achieving the overall balance between the treatment and
comparison groups, which is also well-illustrated in Figure 4.1.
Figure 4.1 shows the distribution of estimated propensity scores. There seem to be no
distributional differences between the treatment and comparison groups. Although the
two groups are still slightly different on a few individual covariates after matching, the
overall balance between the two groups is highly satisfactory. The distribution of logged
propensity scores estimated for the Bronx MHC and the comparison groups also yields
similar patterns. Both groups show a wide range of common grounds for comparison. To
recap, the matching procedure significantly reduced the differences in the pre-treatment
conditions. The treatment group as a whole is virtually identical to the comparison group
85

This factor is measured by the number of different types of offenses one has committed in the past.
The preexisting difference between treatment and comparison groups can be conceived as a bias. This
was measured by as a percentage of the square root of the average of the sample variances in the treated
and untreated groups (see Rosenbaum and Rubin (1985) for technical details). After matching, the extent of
standardized bias reduced from 29.8 to 3.3 – nearly a 90 percent reduction in bias.

86

104

in terms of a variety of observed characteristics of study participants, including
demographics, criminal history, and substance use.
Figure 4.1. Propensity Scores, by Bronx Mental Health Court and Comparison
Groups

.8

.6

.6

Density
.4

.8

Treatment (Bronx MHC)

0

.2

.2

.4

Propensity Score

1

Comparison

0

-8

-6

-4
-2
Logged Propensity

Comparison

0

Treatment

Before turning to the next section, it is worth emphasizing that the propensity score is the
coarsest function of all the covariates. While this approach can improve the overall
comparability between treatment and comparison groups, it is always feasible in
propensity score matching that the two groups may not balance on certain covariates. To
the extent that the propensity score is not balanced, it is important to note that the
observed covariates used to calculate the propensity score will continue to be useful in
predicting treatment assignment (Rosenbaum 2009:166). Hence, some of the key
covariates are included in later analyses to enhance the statistical rigor of the estimated
treatment effect for MHC programs. A further discussion on this issue follows in a later
section of this report.
Does Mental Health Court Reduce Recidivism? 
This section discusses the results from propensity score matching analysis. Table 4.2
reports the average treatment effect of Bronx MHC on recidivism measured by re-arrest
and re-conviction. After matching on a wide range of offender characteristics, we found
that the treatment group was less likely to experience recidivism than the comparison

105

group in our 30-month (minimum) follow-up.87 First, the re-arrest rate was 69 percent for
the treatment group and 75 percent for the comparison group. The difference of 6
percentage points is statistically significant, suggesting that MHC participation reduces
the chance of being re-arrested. In other words, had the Bronx MHC participants not been
admitted to treatment, the chance of re-arrest would have increased by 6 percentage
points.
Table 4.2. Average Treatment Effect of Mental Health Court on Recidivism
Variable
Re-arrest
Re-conviction

Bronx
Treatment
0.69
0.62

Control
0.75
0.62

Difference
-0.06
<-0.01

S.E.
0.04
0.04

T
-1.66+
-0.04

Note. + p < 0.10, * p < 0.05, ** p < 0.01
MHC participation, however, did not result in a measurable reduction in the chance of reconviction. Although its effect on re-conviction is still in the expected direction (although
rounding in Table 4.2 obscures this), there is no statistically significant difference
between the treatment and comparison groups. To better understand recidivism patterns
between the treatment and comparison groups, Table 4.3 shows the type of offense for
which study participants were rearrested. The offense categorization is approximately
based upon UCR classification. Violent crime refers to murder, non-negligent
manslaughter, forcible rape, robbery, and aggravated assault. Property crime includes
burglary, larceny, motor vehicle theft, and arson. All controlled substance offenses are
classified as drug crime. Some of the examples included in other crime are forgery,
prostitution, fraud, gambling, DWI, loitering, and disorderly conduct.
Overall, the breakdown of offense type is fairly similar between the treatment group and
the comparison group. Drug crime accounts for the majority of all recidivism offenses: 48
percent in the treatment group and 44 percent in the comparison group. The Pearson’s χ2
(= 3.92) indicates that there is no meaningful association overall between the treatment
assignment and the offense type for re-arrest (p=.27).88
87

Most recidivism studies typically rely on a 6-month, 1-year, or 2-year follow-up and universally discover
that recidivists would tend to re-offend relatively quickly after intervention. It is rather unusual that
recidivists would remain crime-free for a long time before recidivating. That said, adjusting for the at-risk
time for recidivism in this analysis might not be as critical as in other studies with a short-term follow-up
period. This view is supported by findings from survival analysis, adjusting for the exposure to risk of
recidivism. As further discussed in a later section of this report, the PSM results reported herein do not
differ much from those of survival analysis.

88

The official statistics on the three-year follow-up recidivism rates for offenders who were convicted of
felony and misdemeanor offenses in New York City and sentenced to probation afterwards indicate that the
probability of being re-arrested for a violent felony offense is similar to the probability of being re-arrested
for a drug offense in the probationer population (approximately 12-13 percent in the mid-2000s). It is clear
that the study sample examined in this evaluation is quite different from a larger offender population in
New York City in that the majority of recidivists in our study sample were re-arrested for drug crimes
(NYS DCJS 2011).

106

Table 4.3. Offense Type for Re-Arrest by Treatment and Comparison Groups

Offense Type
Violent
Property
Drug
Other
Total

Re-Arrest
Control
Bronx MHC
25
5.91%
46
10.87%
188
44.44%
164
38.77%
423
100%

29
7.46%
29
7.46%
186
47.81%
145
37.28%
389
100%

Total
54
6.65%
75
9.24%
374
46.06%
309
38.05%
812
100%

Pearson’s χ2 = 3.92 with df (3)

What Explains Recidivism in Bronx Mental Health Court?  
Although the question of program effectiveness is addressed in the above analysis, it is
advantageous to check the robustness of findings by employing further statistical
adjustments to control for residual bias that may exist among the treatment and
comparison groups. Not only does this increase efficiency (Rubin 1997), but such
analyses can afford an opportunity to examine the extent to which other factors explain
recidivism. Based on the matched case-control sample, Table 4.4 thus presents the results
from logistic regressions predicting the chance of re-arrest and re-conviction. It is
important to note that the results reported in Tables 4.2 and 4.4 rely on the same analytic
framework of a matched case-control design, but additional covariate adjustments are
applied to the later analysis. By additionally controlling for relevant covariates, the
logistic models reported in Table 4.4 complement the initial PSM estimates of program
impact.

107

Table 4.4. Logistic Regressions Predicting Re-Arrest and Re-Conviction (N=1,128)
Bronx
MHC Treatment
Black
White
Hispanic
Male
Age
Bronx (Arrest Location)
Instant Offense: Violent
Instant Offense: Property
Instant Offense: Drug
Substance Use (Hard Drugs)
Number of Prior Arrests with Violent Felony Charge
Number of Prior Arrests with Property Offense Charge
Number of Prior Arrests with Drug Offense Charge
Number of Prior Arrests with Public Disturbance Charge
Offense Variety Score
AIC
BIC
Log Likelihood
Chi-squared

Model (1)

Model (2)

Re-Arrest

Re-Conviction

Odds Ratio
0.707+
(-1.67)
1.968
(0.86)
0.778
(-0.30)
1.392
(0.42)
1.210
(0.93)
0.951**
(-4.62)
4.512**
(3.14)
0.639
(-1.36)
2.135
(1.58)
1.034
(0.13)
2.109**
(3.29)
0.980
(-0.27)
1.191**
(3.69)
1.047+
(1.82)
0.991
(-0.33)
1.020
(0.27)
1210.9
1296.4
-588.4
88.39

Odds Ratio
0.961
(-0.21)
1.747
(0.94)
0.838
(-0.27)
1.439
(0.62)
1.172
(0.81)
0.954**
(-4.74)
2.794*
(2.39)
0.765
(-0.87)
2.758*
(2.20)
1.006
(0.03)
1.616*
(2.17)
0.895+
(-1.69)
1.119**
(2.75)
1.051*
(2.15)
1.012
(0.46)
1.027
(0.42)
1402.9
1488.3
-684.4
67.76

Exponentiated coefficients; t statistics in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01

108

There are several findings that are noteworthy. First, while controlling for demographic
and case characteristics, we found that being in the Bronx MHC program would reduce
the chance of re-arrest (Model 1). Consistent with the earlier findings, the odds of being
re-arrested are approximately 29 percent lower for the treatment group than the
comparison group. The predicted probability of re-arrest is 76 percent for the comparison
group and 70 percent for the MHC treatment group. Model 2 reports the odds of being reconvicted. Similar to the earlier results, MHC participation yielded a negative, marginal
effect on the chance of reconviction, but failed to reach statistical significance. Overall,
the re-arrest and re-conviction models were in accord to each other. Most covariates also
yielded an effect that is in the same direction and of similar size between the re-arrest and
re-conviction models.
Second, with respect to demographics, neither gender nor race yielded a meaningful
effect on recidivism. However, age at arrest shows a highly significant effect on re-arrest.
Each additional year of age at arrest is associated with a reduction in the chance of rearrest by 1 percentage point.
Third, the effect of instant offense on recidivism was examined through a set of binary
indicators assessing the type of offense. The reference category was set to “other,” which
combines numerous minor offenses and public disorder offenses. The effect of violent
crime indicator, approaching statistical significance, suggests that arrestees diagnosed
with mental disorders and charged with violent offense are associated with positive
recidivism outcomes than similar arrestees charged with other offenses. By contrast,
those who were arrested for property crime are associated with negative recidivism
outcomes. Similar to the instant offense type, the number of prior arrests for property is
significantly associated with negative recidivism outcomes. The higher the number of
prior arrests for property, the more likely it is for offenders to be re-arrested. This pattern
of recidivism is also detected among those who have prior records of drug offense.
Although we found no direct support for this speculation, it is conceivable from these
observations that mental health treatment may be more equipped to address temperament,
which might instigate violence, as opposed to drug addiction or other cognitive problems,
which might be a stimulus to drug or property crime.89
Fourth, when it comes to drug use and program effectiveness, the odds of experiencing
recidivism for hard drug users (cocaine and heroin) are 2.1 times the odds of those who
do not use hard drugs for recidivism, while controlling for other factors. Put differently,
individuals diagnosed as hard drug users may be at a greater risk of recidivism, thereby
necessitating more attention in MHC.

89

We tested the interaction effect of treatment and the presence of violent crime in a separate model. The
estimated effect of the interaction term is in the expected direction, but failed to reach statistical
significance at the 0.05 level.

109

How Soon to Re‐Offend?  
So far, the results suggest that MHC participation reduces recidivism. This section further
discusses the effect of MHC treatment by examining how the treatment and comparison
groups recidivate over time. We use the Cox proportional hazards regression model to
examine how the risk of recidivism changes over time. Table 4.5 exhibits the effect of
covariates on the hazard of re-arrest (failure rates) on the matched case-control sample.
First, the parameter of primary interest is the effect of MHC participation. The estimated
hazard ratio of 0.686 in Model 1 indicates that the MHC treatment group has a 31 percent
smaller hazard of recidivism than the comparison group.90 This effect size remains stable
and statistically significant even after demographic and case characteristics are included
in the model. The estimated hazard of recidivism for the MHC treatment group is 36
percent smaller than that of the comparison group (Model 2). Again, a marginal change in
the size of treatment effects between Models 1 and 2 suggests that the matched casecontrol sample is well-balanced between the treatment and comparison groups.
Second, all other covariates yielded similar results as shown in the logistic regression
models. Each additional year of age at arrest is associated with the reduced hazard of
recidivism. The younger the age of offenders at the time of arrest, the higher the risk of
recidivism. Those who are charged with drug offenses or have prior drug charges show a
greater risk of recidivism than those do not.
Figure 4.2 plots the survivor function for the MHC treatment and comparison groups.
While controlling for all other covariates, the MHC treatment group shows a higher
survival rate than the comparison group.91 The survivor function declines fairly quickly
in the first year, but overall both groups show a gradual decay over time. Further, it is
worth pointing out that throughout the entire analysis time, the treatment group
outperforms the comparison group in terms of the estimated survival rate, and the gap
between the two groups widens as time elapses.

90

Cox proportional hazard models assume that the relative hazard of an event (i.e., re-arrest in our
evaluation) is fixed over time. In other words, if the hazard of recidivism for the MHC treatment group is
50 percent lower than that of the comparison group, this ratio should hold up at one year, at two years, or at
any point on the time scale. We tested the violations of the proportional hazard assumption on the basis of
Schoenfeld residuals (Grambsch and Therneau 1994) and found no evidence indicating that the
proportional-hazards assumption has been violated for MHC treatment.
91
The MHC treatment group had a much longer exposure to the risk of recidivism than the comparison
group. As such, we limited the analysis time to 2000 days (more than five years), by which time the last rearrest event for the comparison group was observed.

110

Table 4.5. Survival Analysis Predicting the Hazard of Re-Arrest (N=1,128)
Bronx
MHC Treatment

Model (1)
Hazard
Ratio
0.686**
(-5.24)

Black
White
Hispanic
Male
Age
Bronx (Arrest Location)
Instant Offense: Violent
Instant Offense: Property
Instant Offense: Drug
Substance Use (Hard Drugs)
Number of Prior Arrests with Violent Felony Charge
Number of Prior Arrests with Property Offense Charge
Number of Prior Arrests with Drug Offense Charge
Number of Prior Arrests with Public Disturbance Charge
Offense Variety Score
AIC
BIC
Log Likelihood
Chi-squared

10467.0
10472.1
-5232.5
27.55

Model (2)
Hazard
Ratio
0.641**
(-6.04)
0.984
(-0.04)
0.649
(-1.07)
0.910
(-0.25)
1.025
(0.31)
0.976**
(-6.00)
2.109**
(3.80)
1.159+
(1.79)
0.855
(-1.00)
1.881**
(3.61)
0.938
(-0.65)
0.944*
(-2.30)
1.035**
(5.67)
1.027**
(3.31)
0.990
(-0.98)
1.052*
(2.33)
10366.3
10446.8
-5167.2
158.3

Exponentiated coefficients; t statistics in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01

111

.2

.4

Survival
.6

.8

1

Figure 4.2. Time to Re-Arrest by Treatment Condition

0

500

1000
Time (in days)
Comparison

1500

2000

MHC Treatment

Results from the Brooklyn Mental Health Court  
Development of Comparison Groups 
The development of a comparison group for Brooklyn MHC follows the same procedures
as those used for Bronx MHC. Due to the difference in data availability and
characteristics between Bronx and Brooklyn MHCs, however, a moderately different set
of covariates were used in the impact analysis for Brooklyn MHC. In particular, Axis I
diagnoses were available for those who participated in Brooklyn MHC, as well as those
diagnosed with mental disorder in jail. We matched the Brooklyn sample to the DOHMH
sample on mental health conditions, as well as criminal history and other individual
characteristics.
Another noteworthy difference in the Brooklyn evaluation is that the donor pool from
which to develop an equivalent comparison unit is slightly different than that of the
Bronx evaluation. Due to available and complete data, the Brooklyn evaluation relies on
a slightly different set of DOHMH arrestees, as well as a moderately different
combination of covariates in the propensity score estimation. There were 316
participants in the Brooklyn MHC program eligible for the impact analysis. The
DOHMH sample for the Brooklyn impact analysis consists of 5,110 arrestees with mental
disorders from which a matched comparison is withdrawn.

112

Based on a 1-to-1 match, the propensity score matching procedure yielded 303 matched
pairs. Those unmatched cases (n=13) from the MHC sample had too high a propensity
score that could not be supported by the DOHMH sample. The average probability of
receiving treatment among the unmatched cases is 0.996. Data loss due to unmatched
cases is trivial (less than 5 percent of the MHC sample), posing no critical threat to our
analyses. However, we should note that the removed cases tend to be older and have a
longer criminal history than the matched cases.
Table 4.6 reports descriptive statistics of key measures before and after matching.
There are still a few covariates that differ significantly between the treatment and
comparison groups after matching. For example, treatment subjects are on average 1.5
years younger than comparison subjects at the time of the first arrest. There are also
slightly more subjects in the treatment group who were arrested for drug crimes (24
percent) than in the comparison group (21 percent). Despite these subtle differences, the
MHC and DOHMH samples became markedly similar to each other after matching. The
overall balance between the treatment and comparison groups is excellent.92 The average
propensity score of the treatment group (=.41) is indistinguishable from that of the
comparison group.
Also noteworthy is that there were substantial differences between the MHC and
DOHMH samples in terms of mental health conditions, but PSM achieved a significant
reduction in such differences. For instance, the DOHMH sample has a substantially
higher prevalence rate for adjustment disorder (16 percent) and a substantially lower
prevalence rate for psychotic disorder (13 percent) than the Brooklyn MHC participants
(1 percent and 37percent, respectively). These differences have been completely balanced
after matching. The proportion of the subjects diagnosed with those disorders is
practically the same for the treatment and comparison groups after matching.
These observations on the compatibility between the treatment and comparison groups
are well-supported in Figure 4.3. The boxplots of estimated propensity scores present an
almost identical overlap between the treatment and comparison groups. The distribution
of logged propensity scores also confirms that the Brooklyn and the comparison samples
share great resemblance with each other.
Despite the remarkable performance of propensity score matching, it is important to point
out a few potential shortcomings. First, similar to the Bronx evaluation, a few cases from
the Brooklyn sample had a propensity score too high to be matched. Such cases could not
be matched to a reasonably equivalent comparison unit and were therefore removed from
analysis. It is important to acknowledge that estimating the effect of treatment in a
subpopulation that would always receive treatment is not plausible through propensity
score matching (Rosenbaum 2009). Second, the propensity score is the coarsest function
of all the covariates as a whole. Although the overall performance of matching was
92

PSM reduced the extent of standardized bias across all covariates used in PSM by nearly 80 percent.
The difference between treatment and comparison groups, measured as defined in Rosenbaum and Rubin
(1985), diminished from 25.4 to 6.0 after matching.

113

exceptional, the balance between the treatment and comparison groups on a few
covariates requires further improvement. These are minor qualifications of our evaluation
that need to be recognized in the later analysis examining treatment impact.
Table 4.6 Descriptive Statistics of Key Measures
Variable
Black

Sample
Treated Comparison
T
Unmatched
0.58
0.59 -0.41
Matched
0.58
0.55 0.66

p>|t|
0.68
0.51

White

Unmatched
Matched

0.38
0.38

0.18 8.76
0.41 -0.83

0.00
0.41

Hispanic

Unmatched
Matched

0.20
0.20

0.38 -6.38
0.22 -0.40

0.00
0.69

Male

Unmatched
Matched

0.75
0.76

0.67 3.21
0.78 -0.48

0.00
0.63

Age at Arrest

Unmatched
Matched

34.98
34.80

34.78 0.33
35.40 -0.63

0.74
0.53

Number of Prior Arrests

Unmatched
Matched

22.12
20.03

16.61
18.80

5.76
0.62

0.00
0.54

Age at 1st Arrest

Unmatched
Matched

24.16
24.32

21.56 5.78
25.80 -1.62

0.00
0.11

N of Any Violent Felony
Offenses

Unmatched

4.33

2.42

Matched

3.82

N of Firearm-related Offenses

Unmatched
Matched

N of Drug-related Offenses

N of Property Offenses

3.42

10.5
2
0.94

0.35

0.68
0.65

0.42
0.56

5.00
0.82

0.00
0.41

Unmatched
Matched

6.36
5.95

5.99 0.94
6.20 -0.32

0.35
0.75

Unmatched
Matched

6.19
5.31

4.40
4.07

0.00
0.12

3.64
1.58

0.00

114

Table 4.6 Descriptive Statistics of Key Measures (cont.)
Variable
N of Public Order Offenses

Sample
Treated Comparison
T
Unmatched
7.74
4.71 8.51
Matched
6.95
7.11 -0.19

p>|t|
0.00
0.85

1st Offense as Robbery

Unmatched
Matched

0.15
0.14

0.16 -0.90
0.11 1.23

0.37
0.22

1st Offense as Burglary

Unmatched
Matched

0.10
0.10

0.10
0.10

0.07
0.00

0.94
1.00

1st Offense as Assault

Unmatched
Matched

0.24
0.23

0.12 6.30
0.25 -0.57

0.00
0.57

1st Offense as Drug Sale

Unmatched
Matched

0.04
0.04

0.06 -1.64
0.05 -0.39

0.10
0.70

AXIS I diagnosis:
Adjustment disorder

Unmatched
Matched

0.01
0.01

0.16 -7.30
0.01 -0.45

0.00
0.65

AXIS I diagnosis:
Mood disorder

Unmatched
Matched

0.55
0.56

0.39
0.51

5.53
1.06

0.00
0.29

AXIS I diagnosis:

Unmatched

0.37

0.13

0.00

Psychotic disorder

Matched

0.37

12.4
0
0.37 -0.08

0.93

AXIS I diagnosis:
Substance-related disorder

Unmatched
Matched

0.03
0.03

0.24 -8.90
0.05 -1.67

0.00
0.10

Alcohol Use

Unmatched
Matched

0.08
0.08

0.06
0.05

1.76
1.46

0.08
0.15

Cocaine Use

Unmatched
Matched

0.04
0.04

0.10 -3.37
0.04 0.21

0.00
0.83

Heroin Use

Unmatched
Matched

0.02
0.02

0.12 -5.50
0.00 1.64

0.00
0.10

Marijuana Use

Unmatched
Matched

0.07
0.07

0.06 0.23
0.08 -0.31

0.82
0.76

115

Table 4.6 Descriptive Statistics of Key Measures (cont.)
Variable
Instant Offense: Violent Felony

Sample
Treated Comparison
T
Unmatched
0.31
0.23 3.25
Matched
0.31
0.33 -0.44

p>|t|
0.00
0.66

Instant Offense: Firearm-related

Unmatched
Matched

0.07
0.07

0.06
0.06

0.84
0.67

0.40
0.50

Instant Offense: Drug-related

Unmatched
Matched

0.24
0.24

0.40 -5.68
0.21 0.97

0.00
0.33

Offense Variety Score

Unmatched
Matched

4.63
4.51

5.38 -5.50
4.25 1.17

0.00
0.24

Propensity Score

Unmatched

0.43

0.04

0.00

Matched

0.41

0.41

59.9
0
0.01

0.99

Treatment (Brooklyn MHC)

Density
.2

.6
.4

0

.2

-6
0

Propensity Score

.4

.8

1

Comparison

.6

Figure 4.3. Propensity Scores, by Brooklyn Mental Health Court and Comparison
Groups

-4
-2
Logged Propensity
Comparison

0
Treatment

116

Does Mental Health Court Reduce Recidivism? 
Table 4.7 displays the results from propensity score matching analysis. The average
recidivism rates were estimated for the Brooklyn treatment and comparison groups after
matching on a wide array of covariates, including mental health conditions, criminal
history, and demographic characteristics of study subjects. The long-term follow-up on
MHC participants indicates that the average re-arrest rates are approximately 60 percent
for the treatment group and 68 percent for the comparison group. Had the Brooklyn MHC
participants not been admitted to MHC, their re-arrest rate would have been higher by
approximately 8 percentage points. Similarly, MHC participation resulted in a reduction
of 17 percentage points in re-conviction.93 The average re-conviction rate for the MHC
treatment group is 40 percent as opposed to 56 percent for the comparison group. The
treatment group is significantly less likely to be re-convicted than the comparison group.
Table 4.7. Average Treatment Effect of Mental Health Court on Recidivism
Variable
Re-arrest
Re-conviction

Brooklyn
Treatment
Comparison Difference
S.E.
0.60
0.68
-0.08
0.40
0.56
-0.17

T
0.05
0.05

-1.51*
-3.20**

Note. + p < 0.10, * p < 0.05, ** p < 0.01
The type of offense for which recidivists were re-arrested is summarized in Table 4.8.
Findings are generally consistent with the results from the Bronx evaluation. There is no
overall association between treatment assignment and the type of offense (χ2 = 5.88,
p=0.117). The proportion of re-arrests for violent and property crimes is relatively similar
across the treatment and comparison groups (15 percent versus 17 percent and 16 percent
versus 16 percent, respectively). However, the proportion of drug crimes is larger in the
comparison sample (35 percent) than in the treatment sample (25 percent).
What Explains Recidivism in Brooklyn Mental Health Court?  
The results from logistic regressions predicting the chance of re-arrest and re-conviction
are shown in Table 4.9. By examining (controlling for) the effect of other predictors of
recidivism, these models complement the PSM analysis estimating the average treatment
effect of Brooklyn MHC. Based on the matched-case control data, the models include an
indicator for treatment assignment, demographic characteristics, county of arrest as a
geographic control, the type of instant offense, and other criminal history factors.

93

The difference was calculated with rounding.

117

Table 4.8. Offense Type for Re-Arrest by Treatment and Comparison Groups

Offense Type
Violent
Property
Drug
Other
Total

Comparison
31
15.05%
32
15.53%
73
35.44%
70
33.98%
206
100%

Re-Arrest
Brooklyn MHC
31
16.94%
29
15.85%
45
24.59%
78
42.62%
183
100%

Total
62
15.94%
61
15.68%
118
30.33%
148
38.05%
389
100%

Pearson’s χ2 = 5.88 with df (3)

Consistent with the earlier analysis, MHC participation lowers the chance of recidivism.
The odds of being re-arrested are 46 percent lower for the Brooklyn MHC treatment
groups than the comparison group. Similarly, having participated in the Brooklyn MHC
treatment lowers the chance of re-conviction. For an average study participant, the chance
of being re-arrested was 57 percent in the treatment group and 69 percent in the
comparison group. The difference of 12 percentage points is statistically significant after
additional covariate adjustments in Model 1, reaffirming the earlier results from the PSM
analysis. It is also notable that additional covariate adjustment moderately reduced the
effect of MHC participation on re-arrest. The inclusion of covariates in Model 2 appears
to be useful in refining the estimated effect of the Brooklyn MHC program.94
A few other predictors, such as prior involvement in violent crime and offense variety
score show statistical significance in predicting recidivism. These findings on criminal
history-related factors are consistent with the results from the Bronx evaluation. Violent
offenders who participated in the MHC program were much less likely to recidivate than
other offenders in the MHC program. A national recidivism study on prisoners indicates
that violent offenders tend to have a lower re-arrest rate than property or drug offenders
and a similar re-arrest rate as public-order crime such as driving while intoxicated,
weapon-related offenses, traffic offenses, and obstruction of justice (Langan and Levin
2002). Since violent offenders were considerably less likely to recidivate than any other
offenders in the Brooklyn evaluation, it seems salient to hypothesize that MHC programs

94

As discussed above, the propensity score is the coarsest function of all the covariates as a whole. It is
therefore possible that additional covariate adjustments can result in a non-negligible change even when the
overall balance between treatment and comparison groups is great.

118

Table 4.9. Logistic Regressions Predicting Re-Arrest and Re-Conviction (N=606)
Brooklyn

Model (1)

Model (2)

Re-Arrest

Re-Conviction

Black

Odds Ratio
0.540*
(-2.43)
2.207+

Odds Ratio
0.378**
(-4.31)
1.211

White

(1.85)
1.002

(0.43)
0.715

(0.00)
1.359
(0.94)
1.053
(0.18)
0.961**
(-3.49)
0.385*
(-2.17)
0.820
(-0.63)
1.205
(0.47)
1.896+
(1.66)
1.305
(0.60)
0.931*
(-1.97)
1.011
(0.55)
0.982
(-0.91)
1.048+
(1.72)
1.518**
(4.56)
663.3
738.2
-314.7
71.22

(-0.75)
1.207
(0.59)
0.791
(-0.81)
0.971**
(-2.69)
0.744
(-0.91)
0.720
(-1.09)
1.015
(0.04)
1.220
(0.60)
1.282
(0.53)
0.947+
(-1.65)
1.043*
(2.33)
0.999
(-0.05)
1.001
(0.05)
1.380**
(4.44)
748.4
823.3
-357.2
76.51

MHC Treatment

Hispanic
Male
Age
Kings (Arrest Location)
Instant Offense: Violent
Instant Offense: Property
Instant Offense: Drug
Substance Use (Hard Drugs)
Number of Prior Arrests with Violent Felony Charge
Number of Prior Arrests with Property Offense Charge
Number of Prior Arrests with Drug Offense Charge
Number of Prior Arrests with Public Disturbance Charge
Offense Variety Score
AIC
BIC
Log Likelihood
Chi-squared
Exponentiated coefficients; t statistics in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01

119

might be better suited to address criminogenic factors for violent offenders than for other
types of offenders.95
Most other covariates, including gender and age, yielded a null effect on recidivism.
Although not presented in Table 4.9, mental health conditions such as Axis I diagnosis
with adjustment disorder, mood disorder, psychotic disorder, or substance-related
disorder also resulted in a null effect on recidivism.
How Soon to Re‐Offend?  
This section presents results from survival analysis, which takes into consideration that
the amount of time at risk of recidivism varies across study subjects. We first employed
the Cox proportional hazards regression model to estimate the effect of MHC
participation on the hazard of re-arrest (failure rates). The fundamental assumption in this
approach is that the relative hazard of recidivism for MHC treatment is stable over time.
Unlike the Bronx evaluation, the matched case-control data from the Brooklyn MHC
does not satisfy this assumption. We thus specified models in such a way that relaxes the
proportional hazard assumption.96 The results from these models are reported in Table
4.10. Several observations are worth highlighting.
First, the MHC participation has a highly significant, negative effect on the hazard of rearrest. The estimated hazard ratio of 0.509 in Model 1 indicates that the MHC treatment
group has a 49 percent smaller hazard of recidivism than the comparison group. This
effect size remains stable and statistically significant even after other covariates are
introduced in the model (Model 2). This marginal change in the size of treatment effects
between Models 1 and 2 implies that the performance of PSM is robust to additional
controls.
Second, all other covariates yielded similar results as shown in the logistic regression
models. The age at arrest is associated with the reduced hazard of recidivism. For each
additional four years of age at the time of arrest, the hazard of recidivism decreases by
approximately 10 percent. The younger the age of offenders at the time of arrest, the
higher the risk of recidivism. The factors associated with an increased hazard of
recidivism are substance use, the number of prior property offenses, and the offense
variety score. Especially, the offense variety score, measured by the number of different
types of offenses one has committed in the past, moderates the effect of the MHC
participation. Although not reported in Table 4.10, the interaction term of the offense
variety score and the MHC participation is moderately positive (exp(b)=1.074, p=0.057),
indicating that MHC treatment would not be beneficial to offenders who have a variety of
prior offending experience.
95

The interaction effect of violent offense and treatment participation was tested in a variety of model
specifications. The results were consistent across different models that violent offenders who participated
in the Brooklyn MHC program were less likely to recidivate, but statistical significance was not achieved.
96
Following Lambert and Royston (2009), we used restricted cubic spline functions to model the baseline
cumulative hazards and baseline cumulative odds of survival in Cox models. These models enable
proportional hazards to be fit but can be extended to model time-dependent effects.

120

Table 4.10. Survival Analysis Predicting the Hazard of Re-Arrest (N=606)
Brooklyn
MHC Treatment

Model (1)
Hazard
Ratio
0.509**
(-5.62)

Black
White
Hispanic
Male
Age
Kings (Arrest Location)
Instant Offense: Violent
Instant Offense: Property
Instant Offense: Drug
Substance Use (Hard Drugs)
Number of Prior Arrests with Violent Felony Charge
Number of Prior Arrests with Property Offense Charge
Number of Prior Arrests with Drug Offense Charge
Number of Prior Arrests with Public Disturbance Charge
Offense Variety Score
AIC
BIC
Log Likelihood

1882.5
1935.3
-929.3

Model (2)
Hazard
Ratio
0.448**
(-6.58)
1.591
(1.53)
0.973
(-0.09)
1.255
(1.61)
0.969
(-0.22)
0.975**
(-4.78)
0.804
(-1.45)
1.115
(0.44)
0.964
(-0.27)
1.221
(1.27)
1.572**
(2.87)
0.962**
(-2.84)
1.013*
(2.20)
0.998
(-0.25)
1.010+
(1.80)
1.181**
(6.23)
1784.1
1902.9
-865.0

+ p < 0.10, * p < 0.05, ** p < 0.01

121

Figure 4.4 demonstrates the difference in the risk of recidivism between the treatment and
comparison groups over time. As mentioned above, the hazard ratio of recidivism for the
Brooklyn MHC treatment and comparison groups is not constant over time, which is
illustrated in the graph on the left. It is notable that after roughly one year of follow-up,
the difference in the hazard rate approaches zero, suggesting that the rate at which the
MHC participants perform better than their counterparts would be stable over time. The
95 percent confidence intervals also include zero approximately after one year of followup. However, during the first year of observation, especially immediately after
completion of the MHC program, the rate at which MHC participants refrain from crime
is much higher than that of the comparison group. This dynamic change in the hazard rate
of recidivism is translated into an inverted U-shaped pattern of the difference in survival
curves between the treatment and comparison groups.

0

-.003

-.002

.1

.2

Difference in Survival Curve

0
-.001

Difference in Hazard Rate

.3

.001

.4

.002

Figure 4.4. Difference between Treatment and Comparison Groups over Time

0

500
95% bounds

1000

Time(in Days)

1500

2000

Difference in Hazard Rate

0

500
95% bounds

1000

Time(in Days)

1500

2000

Difference in Survival Curves

From the regression models, it is clear that the MHC treatment group performs better
than the comparison group in terms of refraining from recidivism. The difference in
survival curves can therefore be considered as indicative of the extent of program
effectiveness over time. The wider the difference between the MHC treatment and
comparison groups, the higher the survival rate for the MHC relative to the comparison
group. The graph on the right indicates that program effectiveness would increase until
approximately three years of follow-up and decline thereafter. The MHC treatment group
would still have a higher survival rate than the comparison group after three years, but it

122

is important to acknowledge that such a crime control effect of the Brooklyn MHC would
not last persistently over three years.
Limitations 
Although PSM provides a reasonably robust analytic framework to study the causal effect
of treatment, and our results yielded fairly successful performance in achieving balance
between the treatment and comparison groups, the traditional caution against
observational, retrospective research applies here. As with all studies relying on PSM to
identify the causal impact of program participation, the robustness of our estimated
treatment effect depends on the assumptions that 1) the treated and untreated are balanced
on observed covariates and 2) there are no unobserved systematic differences between
MHC participants and their comparison groups. As shown in this chapter, PSM was able
to obtain reasonable balance on observable characteristics. However, as with all quasiexperimental program evaluation, it is possible that important unobservable
characteristics might distinguish the treatment and comparison groups and have a
meaningful impact on the likelihood of recidivism in some way that was not accounted
for in our analyses.
Appendix H presents a sensitivity analysis, exploring how such an unobservable bias
might affect results. The results show that unobservable characteristics increasing the
likelihood of participation among MHC participants have little impact on the results.
However, if unobservable characteristics decrease the likelihood (the likely type of
unobservable bias), our results may turn statistically insignificant (again, see Appendix H
for details). In an effort to avoid this unobservable bias, we restricted our focus to
examine only MHC participants with strong matches in the comparison group (those with
a propensity score within the region of common support and the caliper of 0.05). In doing
so, our analyses estimated the treatment effect with higher methodological rigor.
However, our findings cannot be generalized to offenders whose propensity to receive
treatment was too high to be included in our analyses (subjects who were most likely to
participate).
Finally, although it would be constructive to compare the results between the Bronx and
Brooklyn courts, such a comparison from our data would not be meaningful. The study
participants in the two programs differ significantly in their characteristics, including
demographics, criminal history, and drug use history. In this sense, the courts were
serving different types of individuals in different types of communities and contexts, and
therefore results are not comparable. Further, due to data limitations across the two study
sites (identical data could not be collected from the program sites), propensity score
estimation procedures were implemented on a different set of covariates for each
evaluation.97 Pooling data from the two courts would restrict the data to common
97

Even if we followed the exact same procedures and developed the exact same comparison group for both
Bronx and Brooklyn MHC, a direct comparison of the results from both evaluations would be statistically
unsupported without further adjustments. Unlike linear models, estimates from binary models (such as our
outcome model predicting recidivism) are confounded with residual variance. Therefore, a comparison of
coefficients from different samples can be misleading as residual variances differ across the samples
(Allison 1999, Williams 2009).

123

measures, which would inescapably diminish the quality of propensity score matching.
That said, one should bear in mind these limitations (the possibility of important
unobserved covariates, the restricted sub-sample used in analysis, and the differences
between the evaluations of the two courts) while interpreting the results from our
evaluations.
Summary 
In this chapter, we report the results from impact analyses, assessing the effectiveness of
both mental health courts in reducing recidivism. Propensity score matching was
implemented in the impact analyses to pair up MHC participants with compatible
nonparticipants. The re-arrest and re-conviction rates were then developed for the
treatment groups and their matched comparison groups.
Guided by conventional strategies in PSM, we have estimated propensity scores in
multiple ways with varying configurations and arrived at substantively consistent results
as to the effectiveness of MHCs. The PSM analyses reported in this study indicate that
MHC participation was effective at reducing recidivism in both Bronx and Brooklyn
although the extent of a reduction in recidivism rates differs across the programs. Those
who participated in the MHC programs had significantly lower recidivism rates than
nonparticipants by 6 to 17 percentage points.
We conclude this chapter by elaborating on the strengths and shortcomings of our impact
analyses. Some of the advantages of propensity score matching methods have already
been discussed earlier, but it is of crucial importance to note that limited data availability
and retrospective data analysis confined analytic options in developing robust estimates
for program effectiveness. The matching approach, which forced us to examine the
alternative distributions of covariates across MHC participants and nonparticipants
diagnosed with mental disorders, yielded reasonably consistent, robust performance in
terms of balancing the treatment and comparison groups. By limiting the scope of our
examination to the region of common support, our analyses estimated the treatment effect
with higher methodological rigor.
However, this also concerns one of the qualifications in our analyses that our findings
cannot be generalized to such offenders whose propensity to receive treatment was too
high to be included in our analyses. We were unable to estimate the treatment effect for
offenders with the highest propensity scores. Similarly, it is important to acknowledge
that a direct comparison between the results from Bronx and Brooklyn evaluations is not
meaningful. Above all, the study participants in both programs differ significantly in their
characteristics, including demographics, criminal history, and drug use history. Due to
data availability across the two study sites, propensity score estimation procedures were
also implemented on a different set of covariates for each evaluation.98 Further, as

98

Even if we followed the exact same procedures and developed the exact same comparison group for both
Bronx and Brooklyn MHC, a direct comparison of the results from both evaluations would be statistically
unsupported without further adjustments. Unlike linear models, estimates from binary models (such as our

124

demonstrated in the survival models, the Bronx and Brooklyn MHC treatment groups
differ in some fundamental ways that result in the different patterns of recidivism risk
over time.
Last, the robustness of our estimated treatment effect depends on the assumption that the
treated and untreated are balanced on observed covariates, and there are no unobserved
systematic differences between MHC participants and their comparison groups. If there
was an unknown bias that decreases the odds of participating in MHC programs for the
treated, our estimated treatment effect would be likely to turn statistically insignificant
(see Appendix H for sensitivity analysis). Although our estimated treatment effects for
both Bronx and Brooklyn MHCs are not sensitive to a hidden bias that affects the odds of
program participation in the opposite direction (that is, the odds of receiving MHC
treatment are greater for the treated than for the untreated), it is important to remain
cognizant of the possibility of such unknown bias.

outcome model predicting recidivism) are confounded with residual variance. Therefore, a comparison of
coefficients from different samples can be misleading as residual variances differ across the samples
(Alison 1999, Williams, 2009).

125

CHAPTER 5. COST‐BENEFIT ANALYSIS 
Cost‐Benefit Analysis as Part of the Mental Health Court Evaluation 
The UI research team had proposed to conduct a full cost-benefit analysis as part of
this evaluation. Unfortunately, this was not possible for a variety of reasons.
Generally speaking, we encountered four primary challenges: 1) each MHC program
involved a large number of agencies, 2) data systems and electronic record keeping
were not designed for the task, 3) each program worked with many mental health
treatment providers, and 4) program funding and payment methods were varied and
decentralized. Each of these challenges is reviewed in greater detail, below, after
which we discuss their implications for research and practice.
Importantly, we supplement this discussion with a detailed outline of the data, along
with the methods that we perceive are necessary to conduct rigorous cost-benefit
analyses. We hope that this information will be of use to jurisdictions that are
currently operating mental health court programs, or those that are planning to
implement these models in the future. Additionally, we hope the provision of this
information will provide a solid foundation for advancing future research on mental
health courts.
Challenges in Conducting a Cost‐Benefit Analysis 
The first challenge that we encountered in conducting the cost-benefit analyses
(CBA) of the two MHC programs evaluated in this study is that they span a range of
criminal justice agencies. An appropriate cost-benefit analysis would account for the
total resources used in processing and treating mental health court cases, as well as
comparison cases, to estimate the marginal resources used as a result of the mental
health court across all affected agencies. This presented a problem for the current
study.
The selected MHC programs had significant involvement from Prosecutor’s Offices,
Public Defender’s Offices, the Courts, Treatment Alternatives for Safe Communities
(TASC), and the Center for Court Innovation (CCI). In initial interviews with
program managers, it quickly became clear that, despite close involvement in
processing cases and dealing with issues as they arise, no single agency possessed all
relevant information about the other agencies. For example, an appropriate costbenefit analysis would account for the amount of time spent by lead prosecutors on
the cases, and also by prosecutorial staff. Unlike many drug courts the research team
has encountered (the most obvious model on which to base a mental health court
cost-benefit analysis), prosecutorial involvement in MHC appears to substantially
exceed time spent in the courtroom. While important case information (e.g., mental
health assessments, case files) is readily shared among those involved, information
on day-to-day operations⎯such as the total number of staff involved, the time spent
by each on mental health court cases, and the salaries of involved staff⎯is not
available in a centralized location. In other words, only information necessary for

126

case processing and client treatment is centralized, while much of the information
needed for a CBA is not.
Another challenge we encountered was the format in which recorded data were kept.
The electronic data systems used include some information about mental health
conditions and general information about treatment referrals. They did not, however,
include information about sanctions or treatment received. Cost-benefit analyses of
drug courts (see, for example, Downey and Roman 2011), as well as the process
evaluation performed as part of this study, indicate that treatment is likely to be the
largest cost to society of MHC programs. With information only about the types of
treatment deemed necessary (such as counseling or housing), however, it is
impossible to responsibly estimate the costs incurred as a result of treatment actually
received.
The third challenge was the sheer number of treatment providers used. Without
centralized program records on treatment received, it is possible to work directly
with the providers, themselves, to identify the intensity and length of each client’s
treatment. However, the mental health court programs’ primary goal is to ensure that
all clients get the treatment they need in a timely and consistent manner. As such, the
MHCs work with many providers with different specialties and capacities. The
Brooklyn MHC, for example, reported working with nearly 100 different providers, a
figure that is similar to the set of providers used by the Bronx MHC. This practice
makes data collection directly from treatment providers largely infeasible due to the
twin difficulties of 1) having sufficient resources for such a labor-intensive data
collection effort and 2) obtaining cooperation from so many treatment providers,
who are, themselves, under-resourced and overburdened with demands on their time.
Finally, funding for court operations and treatment came from a wide variety of
sources. If each of the MHC programs had been funded through a single grant, it
likely would have been possible to use their respective blocks of funding to roughly
estimate their programs’ operational costs, as has been done in past research (Harrell
et al. 2003). Similarly, if all of the treatment was paid by one means, it could have
served as a centralized source of treatment costs. However, this is not the reality. The
MHC operations were funded through a number of grants and funding streams, some
designated specifically for MHC funding, and some for which MHC was only one
part of funded activities. Similarly, case managers’ work with clients to ensure that
as many sources as possible are used to fund needed treatment.
As a result, it was prohibitively difficult to estimate the costs of regular MHC
operations and impossible to estimate the social costs of sanctions and treatment
(although no treatment costs were incurred to involved agencies because outside
funding was always used). In addition, it was not feasible to collect costs of
processing these cases according to traditional means (in the absence of the mental
health court) because of the large number of alternate channels the cases could have
gone through and the lack of available data.

127

Implications for Practice 
The challenges identified here should not be construed to mean that the evaluated
MHC programs are not effective or well-organized. It is simply the case that the
institutional and data collection structures are not designed to support cost-benefit
analyses, nor were they ever intended to do so. As discussed above, the courts
effectively share data about client assessments and needs. The fact that they do not
share data about the total staff hours involved in dealing with cases or the salaries of
those staff members simply reflects that that information is not crucial for serving
clients. Likewise, both courts have effective record keeping on client treatment
received and client difficulties in that treatment. This information is simply
maintained in client case files, rather than in central electronic data systems.
The key implication is that it is difficult to comprehensively access systematically
recorded information about a large number of clients. However, for staff dealing
directly with individual clients, as is necessary for the program, this is not a
limitation. Similarly, the fact that the programs use various treatment providers and
funding sources potentially constitutes service delivery strength, rather than
weakness (despite the difficulties it poses for CBA). Case managers are resourceful
in ensuring that clients are able to afford the treatment they need and able to access
the most effective services for them. Operations staff uses a variety of funding
sources to maximize the resources available to the court to serve clients.
Another important implication of this is that our understanding of MHC operations is
severely limited. Relatively little research is available about the mental health court
model to guide assumptions of staff involvement, time spent, and programmatic
operations. Program staff repeatedly stressed that they operate differently than a drug
court because the population is different and has different needs. This highlights the
dearth of practice-oriented guides for mental health courts, particularly when
contrasted with the wealth of information available on drug courts (consider, for
instance, the National Association of Drug Court Professionals). This lack of
research inhibited our ability to make informed assumptions about practice in the
wide range of agencies involved. More importantly, it inhibits information available
to jurisdictions considering starting their own mental health courts. The process
evaluation included in this evaluation is an important early step towards building a
strong knowledge base on mental health court models and operations.
Suggested Strategy for Conducting a Mental Health Court Cost‐Benefit Analysis 
In lieu of conducting a full cost-benefit analysis as part of this project, we lay out
here the approach that we would followed had we been able to complete this
component of the research as envisioned. Our objective, here, is to provide guidance
to practitioners on what data should be collected and how to use this data to conduct
future cost-benefit analyses. We focus mostly on recommendations for data
collection (with illustrations of the purpose of collecting each piece) because the
analytic methods likely will vary with the research design used for the underlying
impact evaluation. We emphasize here that cost-benefit analyses are not valuable in

128

the absence of rigorous impact evaluation. Thus, our guidance is mostly aimed at
practitioners who seek to build the capacity, particularly with regard to data
collection, that will enable them to support future cost-benefit analyses. There are a
number of domains for which data would likely be collected separately and for which
costs are usually estimated separately. We discuss each in turn.
Staff Costs 
One major program cost is staff costs. The goal is to estimate how much staff time
that would not have been used otherwise (across all agencies) is devoted to these
cases. The new time that staff spends on these cases as a result of the MHC program
is referred to as a marginal cost. Marginal costs in staff time should be collected
even if no new staff members are added as a result of the court. Even if the cases are
divided among the same number of staff, any additional time spent on the cases as a
result of the mental health court means that less time is spent on other activities,
which is a cost that should be accounted for.
Data Collection 
The easiest way to gather these data is to survey all involved staff for the amount of
time spent on the cases under the traditional system and the amount of time spent on
cases within the mental health court. Time spent on the cases includes preparation
for and attendance at hearings, periodic staff meetings, one-on-one meetings with
clients, and any other activities dealing with the cases or clients. Thus, four
quantities should be collected. For each staff member who works on relevant cases
either with or without the mental health court, the program should collect:
•
•
•
•

Average number of new cases per week.
Average number of weeks required to process each case.
Total number of hours per week spent on relevant cases.
Fully loaded (including benefits) salaries of staff members.
Analysis 

Table 5.1 provides an example, summarizing the data and methods used to calculate
staff costs. Data should be collected for columns B (new cases per week), C (average
time for each case), E (staff involvement in cases), and H (staff salaries). The other
columns are calculated from these columns.
In the example provided, the cost of processing each case through the MHC is $764.
Of this, $308 dollars per case were being spent anyway using traditional case
processing methods. Therefore, the marginal staff costs of processing each MHC
case are $764 - $308 = $456 per case.

129

Table 5.1. Sample of How Staff Costs Would Be Calculated

A. Staff
person

B. New
cases per
week

C. Weeks of
processing
per case

D. Number
of ongoing
cases at a
time

E. Hours
per week
spent on
mental
health cases

F. Hours
per week
per case

G. Hours
per case
total

H. Salary
(and wage)

I. Cost
per case

h = H/N hrs.
per week per
case

T=hxW
hrs. per case
overall

$D per yr.
($d =
D/(52x40)
per hr.)

Txd=
total cost
per case

Details of Calculations
Calculations

n new cases
per week

W weeks per
case on
average

N=nxW
cases at a
time

H hours per
week

Without Mental Health Court
Prosecutor
Defense
Attorney
Program
Manager

4 new cases

2 weeks

8 total cases

8 hrs/wk

4 new cases

2 weeks

8 total cases

16 hrs/wk

4 new cases

2 weeks

8 total cases

0 hrs/wk

1 hr/wk per
case
2 hrs/wk per
case
0 hrs/wk per
case

2 hrs/case
4 hrs/case
0 hrs/case

$120,000
($58 per hr.)
$100,000
($48 per hr.)
$70,000
($34 per hr.)

$116 per
case
$192 per
case
$0 per
case
$308 per
case

$120,000
($58 per hr.)
$100,000
($48 per hr.)
$70,000
($34 per hr.)

$232 per
case
$192 per
case
$340 per
case
$764 per
case

Total
With Mental Health Court
Prosecutor
Defense
Attorney
Program
Manager
Total

4 new cases

40 weeks

4 new cases

40 weeks

4 new cases

40 weeks

160 total
cases
160 total
cases
160 total
cases

16
16
40

0.1 hr/wk per
case
0.1 hr/wk per
case
0.25 hr/wk
per case

4 hrs/case
4 hrs/case
10 hrs/case

130

Treatment Costs 
Another important type of cost is treatment costs. Treatment costs must be accounted
for even if the court does not pay for the treatment because someone else does.
Whether treatment costs are paid by government support, private health insurance,
the clients’ families or friends, or donated by treatment providers, someone still
bears the costs of treatment and, therefore, these represent some cost to society. The
terminology “cost to society” does not ignore the fact that treatment has benefits.
The benefits are accounted for in any measured impacts (including reduced criminal
activity, reduced re-arrest, improved employment outcomes, etc.).
Data Collection 
The ideal way for treatment data to be collected is at the individual-level. This
provides a more detailed understanding of how treatment varies across individuals,
how different types of treatment complement one another, and who gets what type of
treatment. The important information is the total number of sessions of each type of
treatment each individual received. Therefore, both the length of time or duration
(e.g., number of months) during which treatment was received and the intensity of
the treatment (e.g., four sessions per month) are important to record, separately for
each treatment modality. This should be recorded even if the treatment is provided
by the MHC or court staff directly, although care should be taken to ensure that staff
time devoted to providing treatment is not double counted as a treatment cost and a
staff cost.
This information is useful programmatically, and therefore it likely makes the most
sense to keep it in an electronic version of client case files. Ideal data would reflect
how often clients attended treatment, rather than how often they were referred. Table
5.2 is one example of how the requisite treatment data could be kept.
In addition to being essential for cost-benefit or cost-effectiveness analysis, the
example in Table 5.2 provides important and detailed understanding of a client’s
experience. This client appears to have started with a regular regimen, which was
ramped up in March. In late April, the client entered residential treatment for almost
two months. When leaving residential treatment s/he returned to the previous
treatment schedule, but also was enrolled in life skills courses. During the next two
to four months, the intensity of mental health treatment was scaled back, and the
client began a job readiness course. This is a detailed understanding of the client’s
experience, from which client success and struggles can be inferred, and which can
provide invaluable understanding of the process.

131

Table 5.2. Sample of Treatment Records for Client A
Sessions or Days of Treatment per Month
Modality

Jan.

Feb.

Mar.

Apr.

May

June

July

Aug.

Sep.

Oct.

Nov.

Dec.

Total

Outpatient
individual
counseling

4

4

8

6

0

4

4

4

2

2

2

2

42
sessions

Outpatient
group
counseling

8

8

8

6

0

4

8

8

8

8

4

4

72
sessions

0

0

0

7

30

15

0

0

0

0

0

0

4

4

8

6

0

4

4

4

4

4

2

0

0

0

0

0

0

0

0

0

0

2

4

4

10
sessions

0

0

0

0

0

0

4

4

4

4

0

0

16
sessions

Residential
treatment
Drug abuse
treatment
Job
readiness
class
Life skills
class

52
days
44
sessions

Analysis 
Importantly, some of the treatment received by MHC clients would have been
received even in the absence of the mental health court. Therefore, it is critical to
estimate the marginal treatment costs, rather than the absolute costs. For this reason,
it is important to collect the same information for the comparison group. If the
treatment costs for the comparison group are not calculated in the same way as MHC
clients, there is a significant risk of overestimating the costs of mental health court.
When possible, it is preferable to use the prices of treatment (e.g., cost per session)
of the actual treatment provider(s) who delivered the client-specific services. Most of
the time, this information can be made available to MHC case managers who already
have a relationship with the treatment provider. In the absence of available prices
specific to the treatment providers used, one can approximate that information from
past published research estimates. Large data collection efforts have been undertaken
in the past specifically to estimate national prices for various treatment types.
Alternatively, individual cost-benefit analyses can be used for the less common
treatment modalities.
Once prices for each modality are obtained and the total amount of treatment
received by each client (for each modality) is obtained, these can simply be
multiplied to estimate each client’s treatment costs, as demonstrated in Table 5.3.

132

Table 5.3. Sample Treatment Costs for Client A
Modality
Outpatient
individual
counseling
Outpatient group
counseling
Residential
treatment
Drug abuse
treatment
Job readiness class
Life skills class
Total

Treatment
Received

Price

Cost
$3,360

42 sessions

$80 per session

72 sessions

$10 per session

52 days

$180 per day

44 sessions

$50 per session

10 sessions
16 sessions

$40 per session
$30 per session

$720
$9,360
$2,200
$400
$480
$16,520

Client A incurred significant treatment costs in the year of data collection ($16,520).
Clearly, this is driven by nearly two months in residential treatment. This illustrates
the importance of collecting equally accurate costs for the comparison group. For
instance, if the comparison individual (who is not a mental health court participant)
matched to Client A received just as much residential treatment, half as much group
counseling and drug abuse treatment, and no individual counseling, job readiness
class, or life skills class, the treatment costs of that individual would be $10,820. In
this case, the marginal treatment costs of mental health court participation would
only be $5,700 = $16,520 - $10,820.
Criminal Justice Costs 
Criminal justice costs are broadly defined to be all costs incurred by justice agencies
(such as the police, courts, jails, probation, and prisons). Many analysts consider
these costs the most important, since they are the potential savings that lead state and
local criminal justice agencies to invest in programs such as mental health courts.
Data Collection 
The data collection is straightforward and is often performed using official records.
Days spent in jail and prison, and days on probation, are normally reasonably
accurately recorded in existing data systems maintained by state and local
departments of corrections. Similarly, arrests⎯the main cost to police⎯are normally
accurately recorded. Like staff costs and treatment costs, an accurate estimate of the
marginal costs of MHCs requires that these data be collected for both MHC
participants and the well-defined comparison group.
The collection of these data from official agencies depends on the structure of the
research project, but is typically done by the researchers, rather than the programs.

133

Time spent in court, however, is more complicated. These data are usually not
electronically recorded. Therefore, one way that a mental health court can build
capacity to enable a cost-benefit analysis is by keeping records on either:
•

Client-level information on the number of hearings of each type that each
client had.

•

Aggregate information on the total number of clients who had each type of
hearing (and how many they had).

It is important to collect this information on different types of hearings (e.g.,
regularly scheduled MHC status updates, violation hearings, criminal incident
hearings) because these carry different costs.
Analysis 
There are three major sources to estimate the prices of these events: 1) agency
budgets, 2) past published research, and 3) original data collection. The cost per day,
month, or year of jail, probation, and prison are usually available from annual
financial reports of the relevant agencies or can be estimated using the information
contained there. With respect to arrests, it is standard practice to use results from
some of the few research projects that have estimated such quantities. Table 5.4
presents those estimates; it also demonstrates major savings are realized by
preventing serious crimes, whereas relatively petty crimes produce small savings.
Table 5.4. Commonly Used Costs of Arrest (Dollars in 2005)
Crime
Homicide
Rape
Arson
Robbery
Aggravated
assault
Burglary
Larceny/theft
Motor vehicle
theft

Cost of
Arrest
$10,614
$4,636
$1,352
$869
$354
$747
$264
$756

Source

Crime

Cohen,
1998
Cohen,
1998
Miller, et
al., 1996
Cohen,
1998
Cohen,
1998
Cohen,
1998
Cohen,
1998
Cohen,
1998

Forgery
Drugs
Gambling
DUI/DWI
Vandalism
Traffic
Disorderly
conduct
Weapon

Cost of
Arrest
$264
$25
$20
$54
$75
$57
$57
$20

Source
Cohen, 1998
Rajkumar and
French, 1997
Bierie, 2009
Miller, et al.,
1996
Austin, 1986
Austin, 1986
Austin, 1986
Bierie, 2009

Source: Bierie 2009

134

Again, court hearings are more complicated. These should be handled on a case-bycase basis. The estimated cost of a hearing depends on the length of that hearing, the
number of people involved, and the time spent by all parties preparing for that
hearing. Because most of the costs of a hearing involve the time of involved parties
(such as lawyers, judges, etc.), it is essential to ensure that these costs are not being
double counted in staff time. Whether to include them as staff costs or criminal
justice costs is largely trivial and depends on the details of data collection and
staffing structure; however, one must thoughtfully ensure that staff time is never
included as both. If the mental health court has positive impacts such as reducing rearrest, these benefits will usually manifest as reduced criminal justice costs.
Conclusions 
Here we have discussed the data collection infrastructure that mental health courts
would need to seek to set up in order to enable a cost-benefit analysis to complement
an impact evaluation. We reiterate that a cost-benefit analysis should not be
undertaken without a strong impact analysis. We have sought to illustrate how these
data would be analyzed in order to provide context to our data collection
recommendations.
The previously discussed steps can be used to generate a cost-effectiveness analysis
or a cost-benefit analysis. In the presence of a cost-benefit analysis, however, it is
important to note that what is listed above (reductions in costs to criminal justice
agencies) is only one source of benefits. Various researchers have different stances
on what other benefits to include. We believe that all social benefits should be
included just like all social costs should be included. This would require estimates (at
a minimum) of criminal acts committed, which can be estimated from arrest records
or self-reports, and ideally also would include data on government welfare receipt,
employment outcomes, hospitalizations, and any other important outcomes that carry
social costs.
Again, these considerations should be handled on a case-by-case basis. Ideally, the
program would track these outcomes at the individual-level, although we recognize
that this is often difficult or impossible for programs with limited resources. The
information discussed in the previous sections, then, is the minimum amount of data
necessary to responsibly conduct a cost-benefit or cost-effectiveness analysis. It is
our hope that this chapter will enable future research on mental health courts and
help contribute to what will eventually be a strong body of research on their
effectiveness, operational processes, and best practices.

135

CHAPTER 6.  WHAT HAVE WE LEARNED FROM THE EVALUATION OF 
CRIMINAL JUSTICE INTERVENTIONS FOR OFFENDERS WITH MENTAL 
ILLNESS? IMPLICATIONS FOR POLICY, PRACTICE, AND FUTURE RESEARCH 
Mental health courts seek to reduce recidivism among offenders with mental illness and
to improve public safety by linking offenders with serious and persistent mental illness to
long-term, community-based treatment in lieu of incarceration. Despite the steady growth
of mental health courts across the nation, research on MHC outcomes is relatively limited
and findings are mixed (Wolff et al. 2011).
With funding from the NIJ, UI researchers conducted separate process and impact
evaluations of the Bronx and Brooklyn MHCs. The process evaluation documented
MHC operations compared to business-as-usual (i.e., conventional court processing),
including the key characteristics of these problem-solving courts, the provision of mental
health services (e.g., issues surrounding diagnosis, access, availability, retention), and
overall court processing (e.g., issues surrounding competency and access to appropriate
services). The impact analysis assessed the extent to which MHC participation reduced
re-arrest and re-conviction among participants compared to other defendants with mental
disorders whose cases were processed as usual by the criminal justice system between
2002 and 2006. Propensity score matching techniques were used to construct equivalent
comparison groups.
Major findings from the process and impact evaluation are summarized below, beginning
with substantive results on program effects. Limitations of the analysis also are discussed.
The chapter ends by identifying the implications for future research.
Summary Findings 
Findings from the process evaluation indicate there are key differences in the problemsolving characteristics and orientation of the two mental health courts that could affect
participant outcomes, including:
•

Judicial Interaction and Courtroom Dynamics. Most notably, although both
courts self-identify as operating under a dedicated docket, we observed that the
Bronx MHC court docket typically included a mix of cases (close to one-third
non-MHC cases compared to 18 percent in Brooklyn, of the hearings observed).
In contrast, the Brooklyn MHC set aside the court’s Tuesday docket exclusively
for mental health court cases, but reserved Thursdays for a mix of “spillover
cases,” including drug court cases. Both courts required defendants to arrive at the
start of court and wait together as a group for their case to be called; participants
could leave, however, once their hearing was over. Unlike drug courts in which
cases are placed on the docket in specific order to facilitate program strategies
about using rewards, sanctions, and using the “courtroom as theater” (e.g., hearing
meritorious cases first to reward such individuals by permitting them to leave
court in the shortest time, or alternatively, hearing sanctioned cases first so that

136

everyone in the program will be exposed to the courtroom admonishments as a
learning device and to deter future misbehavior), dockets in both MHC programs
were not ordered in any strategic manner. The Brooklyn MHC, however,
expressed a clear rationale for calling participants to court together―namely, to
build a sense of community and reduce stigma.
Further, the Bronx MHC was marked by more formality than the Brooklyn
program and less direct participant-court interaction (e.g., about one-quarter of
Bronx defendants spoke in court compared to 92 percent of Brooklyn MHC
defendants) consistent with the program’s philosophy that the treatment arranged
by the court is the primary mode of intervention and that the courtroom, itself, is
not the central experience driving individual psychological and behavioral
changes. In contrast, the Brooklyn MHC program viewed judicial-participant
interaction as a key component of the mental health court intervention and
operated the courtroom in a visibly “client-centric” manner. These differences in
judicial philosophy and style also played out in courtroom operations. We
observed an emphasis on judge-participant interaction in the Brooklyn court
characterized by a conversational style in addressing defendants; Brooklyn
participants frequently participated in bench discussions and were engaged in
more direct contact by the judge (again, defendants spoke in court 92 percent of
the time). These observations are consistent with earlier process evaluations of the
Brooklyn MHC (O’Keefe 2006). As noted above, the more traditional courtroom
dynamics observed in the Bronx court were also characterized by limited
defendant participation during MHC hearings.
The drug court literature suggests that judge-participant interaction characterized
by direct conversation and eye contact can be a motivating factor for participants
because it conveys both care for the participant and interest in their progress (see,
for example, Volume 3 of Rossman, Roman, et al. 2011). The duration of status
hearings in both courts, however, was relatively brief, lasting between one and
two minutes, which begs the question of how meaningful the judicial status
hearing, itself, is to the participant experience. By comparison, in the recently
completed NIJ-funded Multi-Site Adult Drug Court Evaluation, Rossman and
colleagues (2011) found that drug court status hearings lasted a little longer than
three minutes on average (the range was one to eight minutes).
Additionally, it is interesting to note that the Brooklyn MHC judge sometimes
visits participants in their community-based treatment settings, which does not
appear to be typical of judicial activity in most problem-solving court programs.
•

Participation in Judicial Hearings. We also observed variation in the
involvement of other courtroom actors. Defense attorneys had a stronger presence
in the Brooklyn MHC (i.e., participated in 37 percent of hearings), while case
managers played a more prominent role in the Bronx (participated in 69 percent
of hearings, compared to 21 percent in Brooklyn) consistent with that program’s
operating philosophy. By way of comparison, recent drug court research findings

137

indicate that other than the drug court judge, participants spoke most often (92
percent) in proceedings followed by the drug court coordinator (31 percent);
defense and prosecution spoke in about one quarter of status hearings while case
managers did so for about 12 percent (Rossman, Roman, et al. 2011). This point
of comparison indicates that defense attorneys and prosecutors, as well as case
managers in these two mental health courts play a more active role in the
courtroom process than their drug court counterparts. Although a hallmark of
problem-solving courts is a non-adversarial focus, this comparison suggests a
relatively more robust collaborative approach in play at the two mental health
courts studied here. Stakeholders in both MHCs reported that the adversarial
nature of courtroom dynamics typically ends once pleas are accepted.
•

Monitoring and Testing. Status hearings are held more frequently in the
Brooklyn MHC; likewise, defendants meet with their assigned forensic
coordinator prior to each status hearing to discuss progress, address any treatment
issues, and submit to random drug tests. In the Bronx MHC, participants meet
weekly with their TASC case manager at which time drug tests are administered.
Some treatment providers also tested MHC clients for drug use.

•

Clinical Assessment. Although both court programs conduct two-part
assessments (psychosocial assessments performed by clinical staff, and
psychiatric evaluations performed by psychiatrists) to determine mental health
eligibility, the Bronx TASC staff assessments incorporated a number of structured
assessment instruments (as noted in Appendix B) in the process. While both
clinical teams meet to discuss cases, the Bronx MHC clinical team meetings were
characterized by a greater degree of mutual decision-making with regard to
treatment issues and client progress. In Brooklyn, clinical decisions were more
centralized and rested with the MHC’s clinical director. Lastly, the TASC clinical
team is housed in a separate and neutral entity from the Bronx MHC; in Brooklyn,
the clinical team is based within the court.

•

Treatment Provider Networks. Unlike most drug courts (the generic model
adapted by MHCs) that typically rely on less-than-a-handful of substance abuse
treatment providers, these two courts used extensive numbers of different
treatment providers (e.g., 100 or more) to provide both community-based and
residential treatment that met the needs of their participants. One of the courts had
a policy of not using a provider unless at least two participants simultaneously
could be enrolled in treatment; this practice was intended to ensure that
participants could have a “natural support group” of other MHC persons as they
moved through their treatment experiences.

•

Treatment Placement. Both court programs placed participants into communitybased treatment. However, in Brooklyn, the decision to accept a client was
contingent upon securing treatment. Clients did not enter a guilty plea to the
program until the clinical team had identified and “locked in” a treatment slot. As
a result, all persons accepted to the Brooklyn MHC had access to treatment. By

138

contrast, the Bronx program operated under an intent-to-treat model. Clients pled
into the program first, often before the clinical team had located a treatment
placement. The vast majority of Bronx MHC participants were successfully
placed into treatment within one to two months, but roughly one-fifth were not.
•

Referral Mechanisms. Both courts accepted clients through a variety of referral
sources, including prosecutors, defense attorneys, and other judges or court parts.
Prosecutors were often the "official" referral source into both mental health
courts, through whom defense referrals were often made. However, the two
programs differed with respect to certain systematic referral mechanisms. The
Narcotics Bureau of the Bronx DA's office routinely used a mental health
checklist to screen for potential referrals to the mental health court; this likely
contributed a greater share of participants with co-occurring disorders. Also, in
the Bronx, approximately one-third of participants had initially been enrolled in
their drug court, then transferred to mental health court, suggesting a need for
additional screening of drug court participants. In Brooklyn, by contrast, cases
referred for competency proceedings were routinely calendared to the mental
health court for consideration once a defendant was restored to fitness; this likely
added more severely mentally ill participants to the Brooklyn caseload.

•

Use of Rewards and Sanctions. Both the Bronx and Brooklyn MHCs employed
rewards and sanctions, but the Brooklyn MHC used a greater variety of rewards,
such as verbal recognition from the judge, certificates for phase advancement, and
small gifts at graduation. In contrast, the Bronx MHC did not mark treatment
progress, citing that the objective of mental health courts is not to cure
participants’ mental health.
Not only does this perspective seems to recognize that improvement in mental
health conditions is not always linear, it also seems to tacitly recognize a key
difference between participants in mental health courts and those in drug court
programs. The primary treatment issue in drug courts is substance abuse, which is
not only a health issue, but also a justice issue as substance use is illegal. By
contrast, mental illness is not, in and of itself, illegal behavior, although those
who suffer from mental illness and find themselves in a mental health court have
committed other infractions that brought them to the attention of the court.
In turn, stakeholders in MHCs reported that compliance was addressed on a caseby-case basis and that participants received frequent second chances. In contrast
to many drug courts, remand to jail was typically a last resort for the two mental
health courts in this study.

The extent to which the observed differences in judicial-participant interaction and
courtroom dynamics affect participant outcomes is unclear. Two aspects of mental health
courts are theorized to promote beneficial therapeutic outcomes: mental health treatment
and ongoing judicial monitoring. The latter is hypothesized to promote treatment
adherence, thereby improving mental health outcomes and reducing criminal behavior.

139

While Brooklyn participants fared slightly better than Bronx participants with respect to
criminal justice outcomes, both groups had significantly better outcomes than their
matched comparisons subjected to “business-as-usual,” suggesting that regular and
frequent monitoring of offenders with mental illness―rather than the type of therapeutic
courtroom model― may be the critical factors in participant success. In either case, the
outcomes from the analysis of systematic courtroom observations suggest that additional
research is warranted to explore which aspects of courtroom dynamics and interactions
have the most impact on long-term defendant outcomes.
In addition to documenting the courts’ problem-solving characteristics and orientation,
the study also catalogued the key components of each court program to describe core
operations and identified factors that impeded or facilitated program operations.
Appendix B lists the key components of the two court programs. As documented through
the process evaluation, both courts perform evaluations and assessments to determine
defendants’ mental health needs, develop plans for community-based treatment, and link
defendants to treatment providers in relatively similar ways.
While both MHC programs work with relatively extensive service provider networks,
stakeholders in each court nonetheless identified a lack of community-based treatment
options as a key challenge to program operations. As a consequence, both programs place
participants with providers in other boroughs and outside of New York State to address
treatment needs. Common placement issues included 1) a general lack of programs, 2)
too few programs providing housing accessible to criminal justice populations with
mental disorders, as well as 3) a dearth of programs to meet the special needs of other
sub-groups in the MHC programs (e.g., Spanish-speakers, adolescent MICA clients, adult
clients with dependent children). The latter was a particular challenge for the Bronx,
which served a higher concentration of Spanish-only speakers and a community where
poverty and substance abuse were more entrenched. Compounding this challenge is the
time it takes to secure open treatment slots that can accommodate defendants in need of
community-based services. Stakeholders expressed concern that clients awaiting
placement remain in jail, where they often deteriorate due either to a lack of treatment or
the stressful experience of incarceration.
Lastly, several stakeholders identified the consistent, stable participation across key
courtroom actors as a strong feature of their respective programs and a critical factor that
facilitates program operations along with the problem-solving team approach. At the time
of our study, both programs had the same judge, DA, and clinical operations (same lead
agency in Bronx, same clinical director in Brooklyn) since their programs’ inception. No
doubt this stability facilitates a shared understanding of policies, procedures, and
philosophy that also promotes continuity in approach.
Stakeholders also felt the team approach employed by problem-solving courts was
beneficial, if not critical, to effectively working with offenders with mental illness.
Compared to drug courts, however, much of the shared decision-making and substantive
interaction among criminal justice and community partners takes place early in the
mental health treatment process, largely around eligibility determination. Once a decision

140

is made to accept or decline a case, much of the team work appears to occur between the
clinical team and mental health court judge (i.e., in the form of pre-court participant
progress updates and recommendations from the clinical team). This is in contrast to the
regular drug court case staffings in which the team (which may include law enforcement
representatives, prosecutors, public defense attorneys, as well as treatment staff) gathers
to discuss client progress and weigh in on case advancement and sanctioning decisions.
Regardless, the benefits of the team approach in the MHC programs studied here may
simply be the shared sense of responsibility and commitment to these cases that mental
health court fostered across normally adversarial criminal justice actors.
Findings from the impact analysis indicate that mental health court participants are
significantly less likely to recidivate, as compared to similar offenders with mental illness
who experience business-as-usual court processing, although the extent of the impact
differs across the two programs.
•

Re-arrest. The matched case-control analysis showed that the re-arrest rate was
69 percent for the Bronx MHC participants and 75 percent for the comparison
group in the Bronx impact evaluation. The difference of six percentage points is
marginally significant, suggesting that MHC participation reduced the chance of
being re-arrested. Similarly, the re-arrest rate for Brooklyn MHC participants
was approximately 60 percent, as compared to 68 percent for the comparison
group, a significant difference at the .05 level.

•

Re-conviction. Although the effect of Bronx MHC participation was in the
expected direction, we failed to observe a meaningful difference between the
treatment and comparison groups: 62 percent were re-convicted in both groups.
In Brooklyn, MHC participation resulted in a reduction of nearly 17 percentage
points in re-conviction. The average re-conviction rate for the MHC treatment
group was 40 percent, as compared to 56 percent for the comparison group,
statistically significant at the .01 level.

These findings are consistent with the extant outcome and impact research on mental
health courts, as discussed in Chapter 1.
Additionally, individuals who recidivated were more likely to commit drug crimes than
violent or property crimes.99 The overall breakdown of offense type was nonetheless
fairly similar between the treatment and comparison groups. Based on χ2 statistic at the
0.05 level, there was no meaningful association between the treatment assignment and
offense type for re-arrest.100
99

This analysis categorized offenses in approximately the same way as UCR classification. All controlled
substance offenses are classified as drug crimes. Violent crime refers to murder, non-negligent
manslaughter, forcible rape, robbery, and aggravated assault. Property crime includes burglary, larceny,
motor vehicle theft, and arson.
100
The official statistics on the three-year follow-up recidivism rates for offenders who were convicted of
felony and misdemeanor offenses in New York City and sentenced to probation afterwards indicate that the
probability of being re-arrested for a violent felony offense is similar to the probability of being re-arrested
for a drug offense in the probationer population (approximately 12 to 13 percent in the mid-2000s). It is

141

With respect to the offender characteristics explaining recidivism, age was a significant
predictor of recidivism in both evaluations. The recidivism rate was significantly higher
for younger offenders. Other predictors of recidivism worth noting are the use of hard
drugs, the number of prior property offenses, and offense variety score. In Bronx, hard
drug users and offenders with extensive property offending histories were significantly
more likely to recidivate (odds ratios 2.1 and 1.2, respectively). In Brooklyn, those who
had engaged in a variety of offenses were more likely to recidivate than those had not.
Finally, survival analysis yielded similar findings. In both evaluations, the treatment
group had a better chance of refraining from recidivism and took longer than the
comparison group to recidivate.
Limitations  
Despite the promising findings of the impact analysis, there are several important
limitations that should be considered in interpreting these results
Outcomes Limited to Recidivism 
As discussed in earlier sections of this report, the study’s original design sought to
answer several key questions about the impact of MHCs, including the extent to which
mental health court:
•

Decreased the use of criminal justice resources (particularly the use jail or prison
beds either pre-trial or post-adjudication).

•

Increased access to, participation in, or retention in mental health treatment.

•

Reduced subsequent criminal justice involvement.

•

Proved to be cost-effective.

In the end, only one outcome―recidivism―could be examined given the limited
program and administrative records available to the study. As discussed in Chapter 2,
efforts to access post-release service use data from the New York State Department of
Health (DOH) Office of Medicaid Management, which contains service data (provider,
type of service, recipient, dates of service) for all services paid by Medicaid, were
unsuccessful. Without post-detention service information for both the treatment and
comparison groups, the impact evaluation was limited to criminal justice outcomes only.
Although both the Bronx and Brooklyn MHC programs provided client-level records to
UI, neither had sufficiently detailed information to support a more extensive descriptive
analysis of the treatment groups’ post-release services utilization. Without these data,
clear that the study sample examined in this evaluation is quite different from a larger offender population
in New York City in that the majority of recidivists in our study sample were re-arrested for drug crimes
(NYS DCJS 2011).

142

neither analyses of mental health treatment and outcomes, nor a meaningful cost analysis
could be conducted. Further, the data necessary to assess any potential reduction in the
use of criminal justice resources (the use of jail beds pre- or post-trial) attributable to
mental health court participation also could not be obtained. These limitations appear to
be directly attributable to 1) the recent economic downturn and 2) the project hiatus that
occurred when it became necessary to replace one of the two original sites intended for
study: e.g., although UI had previously negotiated an agreement to obtain local custody
data, the agency that was the data repository presumably did not have the capacity to
respond to UI’s data request when the project was finally in a position to move forward.
Limiting the assessment of the relative success of multi-faceted interventions like mental
health courts to just one outcome dimension is less than ideal. Other key research
questions surrounding important areas of interest―such as access to and engagement in
mental health treatment and other services, use of criminal justice resources, client
experiences and satisfaction with mental health court, client motivation―remain
unexplored. Future research, for example, might address participants’ levels of
satisfaction with their mental health court experiences and measure their activities while
in the mental health court program to examine how these dimensions affect program
completion, recidivism, and other psychosocial outcomes.
Analysis Limited to Program Impacts  
While this study contributes to the field by providing additional evidence that mental
health courts positively impact participants’ criminal justice outcomes, it does little to
address how or why mental health courts “work,” for whom they work best, or whether
they are more cost-effective than traditional court processing. Addressing the former fell
outside the scope of the original study. Limited data precluded the latter.
Important dimensions of mental health court performance need to be explored with
respect to how these programs function and which elements of performance are most
critical to achieving participant outcomes. These answers are particularly relevant to
policy, practice, and replication. Although the current study sought to conduct a robust
process evaluation that chronicled the key “problem-solving characteristics” of the two
court programs and to explore courtroom dynamics through structured courtroom
observation, time and resources constraints ultimately precluded a truly deep exploration
of each court program.101 While the findings from the process evaluation offer clues
about the potential influence each program may have on participant outcomes, additional
research should be considered.
Potential Sample Bias 
Project researchers employed propensity score matching (PSM), as discussed in Chapter
4, to address the threats of potential bias. Although we were guided by conventional
101

Less than one year in the study period remained when evaluation activities resumed after the original
Florida site was replaced with the second New York site, resources had been dissipated in administrative
activities never envisioned as part of the study work plan, and much of the research team’s effort had to be
devoted to obtaining official records data for the impact evaluation.

143

strategies in our use of PSM, the traditional caution against observational, retrospective
research applies here.
As with all studies relying on PSM to identify the causal impact of program participation,
the robustness of our estimated treatment effect depends on the assumptions that 1) the
treated and untreated are balanced on observed covariates and 2) there are no unobserved
systematic differences between MHC participants and their comparison groups. As
displayed in Tables 4.1 and 4.6 (found in Chapter 4), PSM was able to generate
reasonable balance on observable characteristics. However, as with all quasiexperimental program evaluations, it remains possible that important unobservable
characteristics distinguish the treatment and comparison groups, and had an important
impact on the likelihood of recidivism.
A sensitivity analysis (again, see Appendix H for results), exploring how such an
unobservable bias might affect results, showed that unobservable characteristics had little
impact on the results. In an effort to avoid this unobservable bias, we restricted our focus
to examine only MHC participants with strong matches in the comparison group (those
with a propensity score within the region of common support and the caliper of 0.05). In
doing so, our analyses estimated the treatment effect with higher methodological rigor.
However, our findings cannot be generalized to offenders whose propensity to receive
treatment was too high to be included in our analyses (participants who were most likely
to participate).
Despite these limitations, we believe that the present study uses appropriate methods to
identify the impact of mental health court on recidivism outcomes.
Motivation and Other Potential Source of Bias 
Likewise, participant motivation is not controlled for in the current analyses, which is a
critical consideration for future research (Wolff et. al 2011). More extensive, primary
data collection, including participant self-reported measures, was beyond the scope of the
current study and outside the sphere of the original award. In any case, it is unclear how
personal motivation may factor into program success (i.e., completion). For example,
were mental health court participants more motivated for treatment than nonparticipants?
Did the incentive of reduced charges provide a compelling legal motivation that overrode
any treatment reluctance as suggested by Wolff and her colleagues (2011: 5)? Were
program completers more motivated than non-completers?
Descriptive analyses for the current study found that between 20 (Bronx MHC) and 37
(Brooklyn MHC) percent of individuals referred to the two MHCs did not participate (see
Appendix F for results). Although the Bronx MHC had a lower percentage of
nonparticipants, roughly equally shares (22 to 24 percent) of each program’s
nonparticipants either refused or withdrew their application from consideration.
Likewise, 34 to 39 percent of nonparticipants did not meet the programs’ mental health
eligibility criteria―these individuals were either too ill (17 percent in the Bronx; 15
percent in Brooklyn) or did not have an acceptable mental health diagnosis (17 percent in
the Bronx; 24 percent in Brooklyn). In Brooklyn, MHC nonparticipants had more

144

extensive criminal histories, more serious instant offenses, or more complicated cases
from a criminal justice perspective. While Wolff et al. (2011) posit that felony offenders
may have more legal motivation to participate in mental health courts, this does not
appear to be the case in the Brooklyn MHC. There, roughly 99 percent of nonparticipants
had felony charges at referral. Creaming, however, might be a potential threat, in
Brooklyn where analysis suggests that nonparticipants had more severe mental health
diagnoses than participants. In the Bronx, mental health status was similar for both
groups. Additionally, in Brooklyn, referrals to the program were not accepted to
participate if program staff could not secure a treatment placement for them. The Bronx,
by contrast, used an intent-to-treat model, under which a fraction of accepted clients were
not successfully placed into treatment. Regardless, future research should explore the
considerations raised by Wolff and her colleagues.
Future evaluations of mental health courts programs also should prioritize measuring
motivation as a key factor in any type of outcome or impact evaluation.
Implications for Policymakers, Practitioners, and Researchers 
Although mental health court participants in this study had better criminal justice
outcomes than offenders with mental illness in the matched comparison groups,
recidivism is still high. As noted in Chapter 1, many researchers and advocates assert that
individuals with mental illness are trapped in a “revolving door” of the criminal justice
system, cycling in and out of correctional facilities due to their mental illness and lack of
treatment. Yet others claim that mental health has little relation to criminal behavior and
vice versa, citing the fact that the majority of individuals with mental illness do not
commit crimes. Regardless, incarcerated individuals with mental health problems have
more extensive criminal histories (James and Glaze 2006) and higher levels of criminal
activity post-release (Baillargeon, Binswanger, et al. 2009; CSG Undated; Mallik-Kane
and Visher 2008). The relatively high recidivism rates for both of the study’s treatment
groups may lend additional credence to the assertions of Skeem and her colleagues
(2009) that the majority of offenders with mental illness come in contact with the legal
system for the same reason as other non-mentally ill offenders: criminogenic needs.102 In
essence, Skeem and colleagues contend that individuals with mental illness are at higher
risk for these criminogenic needs, which would explain the disproportionately high rates
of mental illness among the incarcerated population. In support of this model, two studies
(Girard and Wormith 2004, Skeem et al. 2009) found that offenders with mental illness
score higher on the Levels of Service Inventory/Case Management Instrument (LS/CMI),
an assessment tool used to assess risk and criminogenic needs, compared to offenders
without mental illness. This suggests that mental health court participants would benefit
from the kind of cognitive behavioral programming that addresses criminogenic
(criminal) thinking. Although the Brooklyn MHC assessed for criminogenic risks and
needs, it is unclear what role cognitive behavioral therapies played in the court’s
102

Criminogenic needs are dynamic risk factors predictive of criminal activity that can be targeted in
rehabilitative treatment. For example, the “Central Eight” criminogenic needs include: substance abuse,
history of antisocial behavior, personality, cognition, peers, and circumstances regarding family/marriage,
school/work, and recreation (Andrews and Bonta 2010).

145

treatment regimen. A growing literature on evidence-based practice suggests that
cognitive behavioral therapies are critical in mitigating future offending among offenders
with high criminogenic risk-needs.
Policymaker support for and interest in criminal justice alternatives for mentally ill
offenders is strong and the number of mental health courts is growing. Although the field
has not yet produced as many studies documenting the effectiveness of mental health
courts as exist for drug courts, there is a growing body of research which consistently
provides empirical support that mental health courts are effective in reducing recidivism
and positively impacting participant functioning. The findings of this study only further
reinforce this trend. Therefore, it may well be prudent to fund additional studies that
support cross-site evaluation of multiple jurisdictions with their different policies and
practices to extend our knowledge of mental health court effectiveness.
 
Beyond outcomes, however, little research has been conducted on questions of mental
health court efficiency and cost. One study (Ridgeley et al. 2007) investigated costs for a
mental health court in Allegheny County. This study found that the jurisdiction’s mental
health court costs were similar to those of the traditional court system. The authors
speculated that it was likely that the mental health court might become less costly over
time.
Future work can build upon this promising research base. Methodological weaknesses of
individual studies (e.g., sole reliance on self-reported outcomes, lack of random
assignment, and short-term follow-up) make it difficult to reach confident conclusions.
Most outcome studies examine individual courts, which may account for conflicting
findings across studies; however, existing meta-analyses help provide overall estimates of
mental health courts’ effectiveness. Nevertheless, it is still important for researchers in
the field to expand the evidence base with strong research designs in multisite studies.
Outcome studies also should include process components so that researchers can isolate
possible causes of differing outcomes and levels of success. With modest graduation rates
in some courts (e.g., Hiday and colleagues [2005] found a little more than half of MHC
participants graduated from the court in their study), it also is important to evaluate the
relative outcomes of program graduates versus those who fail to complete the program.
While future work should continue to examine important criminal justice and mental
health outcomes, researchers also should begin to explore some additional issues, such as:
•

Cost-effectiveness of mental health courts.

•

Identification of mental health court best practices including essential program
components, in keeping with the growing emphasis on implementation of
evidence-based practices. Future research should focus on identifying precisely
which MHC policies and practices generate high performance in terms of
recidivism and improved mental health status

146

•

Development of research-driven standards to guide MHC court practices. The
drug court field has received considerably more attention than MHCs and has
matured to a state where researchers can say with a fair degree of confidence what
works best to achieve reductions in crime and drug use among substance-using
offenders in these programs. If evidenced-based standards of practice can be
identified, there is the potential to systematically introduce improvements across
current and future MHC programs by developing an accreditation program.

•

Effectiveness of mental health courts for sub-populations (e.g. first-time offenders
vs. offenders with extensive criminal histories; individuals with more or less
severe psychopathologies).

•

The relative value of various features or components of the mental health court
model, or of differing models.

•

Causes of program failure by individuals and ways to retain participants.

•

Longer term impacts.

•

Client perspectives.

•

Public opinion of mental health courts.

As noted in Chapter 1, continuing to describe and evaluate mental health courts will
assist in the improvement of existing courts and help practitioners and policymakers to
design and implement future programs with evidence-based practices. Findings from the
current study support this objective by contributing additional findings to the field
through a multi-site process and outcome evaluation of mental health courts in New York
City, using sophisticated analytic techniques to control for selection bias, the largest
methodological threat to mental health court evaluation research.
Conclusions  
Current analyses provide additional evidence that both the Bronx and Brooklyn MHC
programs positively impact participants’ criminal justice outcomes. It also identifies
characteristics of the two courts that may contribute to these outcomes. Several avenues
for future research have been identified that will address key gaps in the extant research
and ideally advance both policy and practice, in the process.

147

REFERENCES 
Abram K.M., L.A. Teplin, and G.M. McClelland. (2003). Comorbidity of Severe Psychiatric
Disorders and Substance Use Disorders Among Women in Jail. American Journal of Psychiatry.
160(5): 1007-1010.
Akers, R.L. and C. S. Sellers. (2008). Criminological Theories: Introduction, Evaluation, and
Application. New York: Oxford University Press.
Allison P. (1999). Comparing Logit and Probit Coefficients Across Groups. Sociological
Methods and Research. 28(2): 186-208.
Almquist L. and E. Dogg. (2009). Mental Health Courts: A Guide to Research-Informed Policy
and Practice. New York: Council of State Governments Justice Center.
American Psychiatric Association. (APA 2000). Diagnostic and Statistical Manual of Mental
Disorders (Revised 4th Edition). Washington, DC: American Pyschiatric Association.
America’s Law Enforcement and Mental Health Project, Pub. L. No 106-515 (2000).
Andenaes J. (1974). Punishment and Deterrence. Ann Arbor: University of Michigan Press.
Andrews D.A. and J. Bonta. (2010). The Psychology of Criminal Conduct (5th Edition). New
Providence, NJ: LexisNexis.
Anglin M.D., M.L. Brecht, and E. Maddahian. (1990). Pre-Treatment Characteristics and
Treatment Performance of Legally Coerced Versus Voluntary Methadone Maintenance
Admissions. Criminology. 27: 537-57.

Baillargeon J., I.A. Binswanger, J.V. Penn, B.A. Williams, and O.J. Murray, O.J. (2009).
Psychiatric Disorders and Repeat Incarcerations: The Revolving Prison Door. American
Journal of Psychiatry. 166: 103-109.
Belenko S. (1999). Research on Drug Courts: A Critical Review 1999 Update. National Drug
Court Institute Review. II (2): 1-58.
Berman G. and J. Feinblatt. (2005). Good Courts: The Case for Problem-Solving Justice. New
York: The New Press.
Bierie D. (2009). Cost Matters: A Randomized Experiment Comparing Recidivism Between Two
Styles of Prison.” Journal of Experimental Criminology. 5(4).
Bureau of Justice Assistance. (Undated). Mental Health Court Grantees. BJA Programs. Accessed
April 19, 2011: http://www.ojp.usdoj.gov/BJA/pdf/MHC_Grantees.pdf.
Boothroyd R., N. Poythress, A. McGaha, and J. Petrila. (2005). The Broward Mental Health
Court: Process, Outcomes and Service Utilization. International Journal of Law and Psychiatry.
26, 55-71.

148

Casey P. and D. Rottman. (2003). Problem-Solving Courts: Models and Trends. Williamsburg,
VA: National Center for State Courts. Online:
http://www.ncsconline.org/WC/Publications/COMM_SpeProProbSolvCtsPub.pdf
Casper J.D., T. Tyler, and B. Fisher. (1988). Procedural Justice in Felony Cases. Law and Society
Review. 22:483-507.
Center for Court Innovation (Undated). Brooklyn Mental Health Court Project Description. New
York: Center for Court Innovation.
Christy A., N. Poythress, R. Boothroyd, J. Petrila, and S. Mehra. (2005). Evaluating the
Efficiency and Community Safety Goals of the Broward County Mental Health Court. Behavioral
Sciences and the Law. 23: 227–243.
Cochran W. (1968). The Effectiveness of Adjustment by Subclassification in Removing Bias in
Observational Studies. Biometrics. 24: 295-314.
Collins J.J. and M. Allison. (1983). Legal Coercion and Retention in Drug Abuse Treatment.
Hosp Community Psychiatry. 34(12):1145-9.
Cosden M., J. Ellens, J. Schnell, Y. Yamini-Diouf, and M. Wolfe. (2003). Evaluation of a Mental
Health Treatment Court with Assertive Community Treatment. Behavioral Sciences and the Law.
21: 415–427.
Council of State Governments. No date listed. Fact Sheet: Mental Illness and Jail. New York
City: Council of State Governments.
Council of State Governments. (2002). Criminal Justice / Mental Health Consensus Project
Report. Lexington, KY: Council of State Governments.
Council of State Governments. (2005a). Mental Health Courts: A National Snapshot. Online:
http://www.csgeast.org/pdfs/Consensus%20Project/MHC_National%20Snapshot.pdf
Council of State Governments. (2005b). A Guide to Mental Health Court Design and
Implementation. New York, NY: Council of State Governments.
Council of State Governments. (2008). Mental Health Courts: A Primer for Policymakers and
Practitioners. Washington, DC: The Department of Justice, Bureau of Justice Assistance. Online:
http://consensusproject.org/mhcp/mhc-primer.pdf
Dehejia R.H. and S. Wahba. (1999) Causal Effects in Nonexperimental Studies: Reevaluating the
Evaluation of Training Programs. Journal of the American Statistical Association. 94(448), 1053–
1062.
De Leon G. (1988a). Legal Pressure in Therapeutic Communities. Journal of Drug Issues. 18:
625-640.
DeLeon G. (1988b). Legal Pressure in Therapeutic Communities. In C. G. Leukfield and F. M.
Tims (Eds.), Compulsory Treatment of Drug Abuse: Research and Clinical Practice (NIDA
Research Monograph 86, DHHS Publication No. ADM 88-1578, Rockville, MD: National
Institute on Drug Abuse, 160-177.

149

Denckla D. and G. Berman. (2001). Rethinking the Revolving Door: A Look at Mental Illness in
the Courts. New York: Center for Court Innovation.
Diamond, A. and Sekhon, J.S. (2010). Generic Matching for Estimating Causal Effects: A
General Multivariate Matching Method for Achieving Balance in Observational Studies. Working
Paper. Available at: http://sekhon.berkeley.edu/papers/GenMatch.pdf
Downey P.M. and J.K. Roman. (201). Chapter 9. Cost-Benefit Analyses. In S.B. Rossman, J.K.
Roman, J. Zweig, M. Rempel, and C. Lindquist (Ed). Final Report of The Multi-Site Adult Drug
Court Evaluation. Volume Four: The Impact of Drug Courts. Washington, DC: The Urban
Institute.
Elbogen E.B. and S.C. Johnson. (2009). The Intricate Link Between Violence and Mental
Disorder. Archives of General Psychiatry. 66(2): 152-161.
Elbogen E.B., R.A. Van Dorn, J.W. Swanson, M.S. Swartz, and J. Monahan. (2006). Treatment
Engagement and Violence Risk in Mental Disorders. British Journal of Psychiatry, 189: 354-360.
Farole D.J., N. Puffett, M.Rempel, and F. Byrne (2005). Applying the Problem-Solving Model
Outside of Problem-Solving Courts. Judicature. 89 (1): 40-42.
Fazel S., G. Gulati, L. Linsell, J.R. Geddes, and M. Grann. (2009). Schizophrenia and Violence:
Systematic Review and Meta-Analysis. PLoS Medicine. 6(8): 1000120.
Fisher W.H., E. Silver, and N. Wolff. (2006). Beyond Criminalization: Toward a
Criminologically Informed Framework for Mental Health Policy and Services Research.
Administration and Policy in Mental Health and Mental Health Services Research. 33: 544-557.
Folger R. (1977). Distributive and Procedural Justice: Combined Impact of ‘Voice’ and
Improvement on Experienced Inequity. Journal of Personality and Social Psychology. 35: 108119.
GAINS Center (undated). Developing a Comprehensive Plan for Mental Health and Criminal
Justice Collaboration: The Sequential Intercept Model. Online:
http://gains.prainc.com/pdfs/integrating/GAINS_Sequential_Intercept.pdf
Grambsch, P. M., and T. M. Therneau. 1994. Proportional hazards tests and diagnostics based on
weighted residuals. Biometrika 81: 515–526.
Gibbs J.P. (1975). Crime, Punishment, and Deterrence. Amsterdam, The Netherlands: Elsevier
Scientific Publishing Co.
Girard L. and J. Wormith. (2004). The Predictive Validity of the Level of Service InventoryOntario Revision on General and Violent Recidivism Among Various Offender Groups. Criminal
Justice and Behavior. 31: 150–181.
Goldkamp J.S., and C. Irons-Guynn, C. (2000). Emerging Judicial Strategies for the Mentally Ill
in the Criminal Caseload: Mental Health Courts in Fort Lauderdale, Seattle, San Bernardino, and
Anchorage. Washington, DC: US Department of Justice, Bureau of Justice Assistance.

150

Gottfredson D.C., B.W. Kearley, S.S. Najaka, and C.M. Rocha. (2007). How Drug Treatment
Courts Work: An Analysis of Mediators. Journal of Research in Crime and Delinquency, 44(1):
3-35.
Griffin P.A., H.J. Steadman, and J. Petrila. (2002). The Use of Criminal Charges and Sanctions
in Mental Health Courts. Psychiatric Services. 53: 1285-1289.
Harrell A., O. Mitchell, J. Merrill, and D. Marlowe. (2003). Evaluation of Breaking the Cycle.
Washington, DC: The Urban Institute.
Herinckx H.A., S.C. Swart, S.M. Ama, C.D. Dolezal, and S. King, S. (2005). Rearrest and
Linkage to Mental Health Services Among Clients of the Clark County Mental Health Court
Program. Psychiatric Services. 56: 853-857.
Hiday V.A., M.E. Moore, M. Lamoureaux, and J. de Magistris, J. (2005). North Carolina’s
Mental Health Court. Popular Government. Spring/Summer volume: 24-30.
Hills H.A. (2000). Creating Effective Treatment Programs for Persons With Co-Occurring
Disorders in the Justice System. Online:
http://gains.prainc.com/pdfs/disorders/Creating_Effective_TX_Prog.pdf
Hubbard R.L., M.E. Marsden, J.V. Rachal, H.J. Harwood, E.R. Cavanagh, and H.M. Ginzburg.
(1989). Drug Abuse Treatment: A National Study of Effectiveness. Chapel Hill, NC: University
of North Carolina Press.
James, D.J., and Glaze, L.E. (2006). Mental Health Problems of Prison and Jail Inmates.
Washington, DC: U.S. Department of Justice Office of Justice Programs, Bureau of Justice
Statistics.
Lamb H.R., and L.E. Weinberger. (1998). Persons with Severe Mental Illness in Jails and
Prisons: A Review. Psychiatric Services. 49: 483-492.
Lambert, P.C. & Royston, P. (2009). Further development of flexible parametric models for
survival analysis. The Stata Journal 9: 265-290.
Langan P. and D. Levin. (2002). Recidivism in Prisoners Released in 1994. Washington, DC:
Bureau of Justice Statistics.
Lawental E., A.T. McClellan, G.R. Grissom, P. Brill, and C. O’Brien. (1996). Coerced Treatment
for Substance Abuse Problems Detected Through Workplace Urine Surveillance: Is It Effective?
Journal of Substance Abuse. 8, 1: 115-128.
Leuven E. and B. Sianesi. (2003). PSMATCH2: Stata module to perform full Mahalanobis and
propensity score matching, common support graphing, and covariate imbalance testing. Online:
http://ideas.repec.org/c/boc/bocode/s432001.html.
Lind E.A. (1982). The Psychology of Courtroom Procedure, in N.L. Kerr and R.M. Bray, eds.,
The Psychology of the Courtroom. New York: Academic Press.

151

Lind E.A., R. Kanfer, and P.C. Earley. (1990). Voice, Control, and Procedural Justice:
Instrumental and Noninstrumental Concerns in Fairness Judgments. Journal of Personality and
Social Psychology. 59: 952-959.
Lind E.A. and T. Tyler. (1988). The Social Psychology of Procedural Justice. New York: Plenum
Publishing.
Mallik-Kane K. and C.A. Visher, C.A. (2008). Health and Prisoner Reentry: How Physical,
Mental, and Substance Abuse Conditions Shape the Process of Reintegration. Washington, D.C..:
The Urban Institute.
Marlowe D.B., D.S. Festinger, C. Foltz, P.A. Lee, and N.S. Patapis. (2005). Perceived
Deterrrence and Outcomes in Drug Court. Behavioral Sciences and the Law. 23(2): 183-198.
Mentally Ill Offender Treatment and Crime Reduction Act of 2003, Pub. L. No 108-414, 2003.
McNeil D. and R. Binder. (2007). Effectiveness of a Mental Health Court in Reducing Criminal
Recidivism and Violence. American Journal of Psychiatry. 164; 1395–1403.
Moore M.E., and V.A. Hiday. (2006). Mental Health Court Outcomes: A Comparison of ReArrest and Re-Arrest Severity Between Mental Health Court and Traditional Court Participants.
Law and Human Behavior. 30: 659-674.
Munetz M.R. and P.A. Griffin (2006 ). Use of the Sequential Intercept Model as an Approach to
Decriminalization of People With Serious Mental Illness. Psychiatric Services. 57 (4): 544.
NYS Division of Criminal Justice Services. (2011). Probationer Felony Re-Arrest Rates
Following Sentence to Probation, Albany. NY. Available at
http://www.dpca.state.ny.us/pdfs/probationerfelonyarrestsbkmk5apr11.pdf
Office of Justice Programs and National Association of Drug Court Professionals. (OJP/NADCP
1997). Defining Drug Courts: The Key Components. Washington, DC: U.S. Department of
Justice.
O’Keefe K. (2006). The Brooklyn Mental Health Court Evaluation: Planning, Implementation,
Courtroom Dynamics, and Participant Outcomes. NY: Center for Court Innovation.
Paternoster R. (1987). The Deterrent Effect of the Perceived Certainty and Severity of
Punishment: A Review of Evidence and Issues. Justice Quarterly. 4: 173-217.
Pearl J. (2009). On a Class of Bias-Amplifying Covariates That Endanger Effect Estimates. Tech.
Rep. R-356, University of California, Los Angeles, CA. Available at:
http://ftp.cs.ucla.edu/pub/stat ser/r356.pdf.
Petrila J., N.G. Poythress, A. McGaha, and R.A. Boothroyd. (2001). Preliminary Observations
from an Evaluation of the Broward County Mental Health Court. Court Review. 37(4): 14-22.
Pogrebin M.R. and E.D. Poole (1987). Deinstitutionalization and Increased Arrest Rates Among
the Mentally Disordered. The Journal of Psychiatry and Law. 15: 117-127.

152

Poythress N.G., J. Petrila, A. McGaha, A., and R. Boothroyd. (2002). Perceived Coercion and
Procedural Justice in the Broward Mental Health Court. International Journal of Law and
Psychiatry. 25: 517-533.
Prins S.J. and L. Draper. (2009). Improving Outcomes for People with Mental Illness Under
Community Corrections Supervision: A Guide to Research-Informed Policy and Practice. New
York, NY: Council of State Governments.
Redlich A.D., S. Hoover, A. Summers, and H.J. Steadman. (2010). Enrollment in Mental Health
Courts: Voluntariness, Knowingness, and Adjudicative Competence. Law and Human Behavior.
34(2): 91-104.
Redlich A., H. Steadman, L. Callahan, P. Robbins, R. Vessilinov, and A. Ozdogru. (2010). The
Use of Mental Health Court Appearances in Supervision. International Journal of Law and
Psychiatry. 33: 272-277.
Redlich A.D., H.J. Steadman, J. Petrila, J.,Monahan, and P.A. Griffin, P.A. (2005). The Second
Generation of Mental Health Courts. Psychology, Public Policy, and Law. 11: 527-538.
Redlich A., H.J. Steadman, J. Monahan, P.C. Robbins, and J. Petrila. (2006). Patterns of Practice
in Mental Health Courts: A National Survey. Law and Human Behavior. 30: 347-362.
Ridgeley M.S., J. Engberg, M.D. Greenberg, S. Turner, C. DeMartini, and J.W. Dembosky.
(2007). Justice Treatment and Cost: An Evaluation of the Fiscal Impact of Allegheny County
Mental Health Court (technical report). Santa Monica, CA: Rand Corporation.
Roman J.K., S.B. Rossman, and M. Rempel. (2011). Chapter 2. Review of the Literature. In S.B.
Rossman, J.K. Roman, J. Zweig, M. Rempel, and C. Lindquist (Eds). Final Report of The MultiSite Adult Drug Court Evaluation. Volume One: Study Overview and Design. Washington, DC:
The Urban Institute.
Rosenbaum P. R. (2002). Observational Studies. New York: Springer-Verlag 2nd edition.
Rosenbaum P.R. (2009). Design of Observational Studies. New York: Springer.
Rosenbaum P. R. and D.B. Rubin. (1983). The Central Role of the Propensity Score in
Observational Studies for Causal Effects. Biometrika. 70 (1), 41-55.
Rosenbaum P. R. and D.B. Rubin. (1984). Reducing Bias in Observational Studies Using
Subclassification on the Propensity Score. Journal of the American Statistical Association. 79
(387). 516-524
Rossman S.B. (2011). Chapter 1. Introduction: Study Context and Objectives. In S.B. Rossman,
J. Roman, J. Zweig, M. Rempel, and C. Lindquist (Eds). (2011). Final Report of The Multi-Site
Adult Drug Court Evaluation. Volume One: Study Overview and Design. Washington, DC: The
Urban Institute.
Rossman, S. B., J.K. Roman, J. Zweig, M. Rempel, and C. Lindquist (Eds). (2011). Final Report
of The Multi-Site Adult Drug Court Evaluation. Executive Summary and Volumes 1-4.
Washington, DC: The Urban Institute.

153

Rottman D. and P. Casey. (1999). Therapeutic Jurisprudence and the Emergence of ProblemSolving Courts. National Institute of Justice Journal. Summer,
Sarteschi C.M., M.G. Vaughn, and K. Kim. (2011). Assessing the Effectiveness of Mental Health
Courts: A Quantitative Analysis. Journal of Criminal Justice. 39: 12-20. doi:
10.1016/j.jcrimjus.2010.11.003.
Satel S. (1998). Observational Study of Courtroom Dynamics in Selected Drug Courts. National
Drug Court Institute Review. 1:43-72.
Sekhon J.S. (Forthcoming). Multivariate and Propensity Score Matching Software with
Automated Balance Optimization: The Matching package for R. Journal of Statistical Software.
Scheider M. (2001). Deterrence and the Base Rate Fallacy: An Examination of Perceived
Certainty. Justice Quarterly. 18: 63-86.
Siddall J.W. and G.L. Conway. (1988). Interactional Variables Associated with Retention and
Success in Residential Drug Treatment. International Journal of the Addictions. 23, 12: 12411254.
Skeem J., J. Eno Louden, S. Manchak, S. Vidal, and E. Haddad. (2009). Social Networks and
Social Control of Probationers with Co-Occurring Mental and Substance Abuse Problems. Law
and Human Behavior. 33, 122–135.
Skeem J.L., S. Manchak, and J.K. Peterson. (2010). Correctional Policy for Offenders with
Mental Illness: Creating a New Paradigm for Recidivism Reduction. Law and Human Behavior.
Steadman H.J., E.P. Mulvey, J. Monahan, P.C. Robbins, P.S. Appelbaum, T. Grisso, L.H. Roth,
and E. Silver, E. (1998). Violence by People Discharged from Acute Psychiatric Inpatient
Facilities and by Others in the Same Neighborhoods. Archives of General Psychiatry. 55: 393401.
Steadman H.J., F.C. Osher, P.C. Robbins, B. Case, and S. Samuels. (2009). Prevalence of Serious
Mental Illness Among Jail Inmates. Psychiatric Services. 60: 761-765.
Steadman H.J., A.D. Redlich, P.A. Griffin, J. Petrila, and J. Monahan. (2005). From Referral to
Disposition: Case Processing in Seven Mental Health Courts. Behavioral Sciences and the Law.
23: 215-226.
Steadman H.J., A. Redlich, L. Callahan, P.C. Robbins, and R. Vesselinov. (2011). Effect of
Mental Health Courts on Arrest and Jail Days. Archives of General Psychiatry. 68(2): 167-172.
Teplin, L.A. (1984). Criminalizing Mental Disorder: The Comparative Arrest Rate of the
Mentally Ill. American Psychologist. 39(7): 794-803.
Teplin, L.A. (1994). Psychiatric and Substance Abuse Disorders Among Male Urban Jail
Detainees. American Journal of Public Health. 84: 290-293.
Thibaut J. and L. Walker. (1975). Procedural Justice: A Psychological Analysis. Hillsdale, NJ:
Erlbaum.

154

Thompson M., F. Osher, and D. Tomasini-Joshi. (2008). Improving Responses to People with
Mental Illnesses: The Essential Elements of a Mental Health Court. New York: Council of State
Governments Justice Center.
Torrey E.F., A.D. Kennard, D. Eslinger, R. Lamb, and J. Pavle. (2010). More Mentally Ill
Persons are in Jails and Prisons than Hospitals: A Survey of the States. VA: National Sheriffs
Association and Treatment Advocacy Center.
Trone J. and D. Young. (1996). Bridging Drug Treatment and Criminal Justice. Vera Institute
Program Brief. New York: Vera Institute of Justice.
Trupin E. and H. Richards. (2003). Seattle's Mental Health Courts: Early Indicators of
Effectiveness. International Journal of Law and Psychiatry. 26: 33-53.
Tyler T. (1984). The Role of Perceived Injustice in Defendant’s Evaluation of Their Courtroom
Experience. Law and Society,.18: 51-74.
Tyler T. (1990). Why People Obey the Law. New Haven, CT: Yale University Press.
Tyler T. (2003). Procedural Justice, Legitimacy, and the Effective Rule of Law. Crime and
Justice. 30: 283-357.
U.S. Department of Health and Human Services (DHHS, 1999). Mental Health: A Report of the
Surgeon General—Executive Summary. Rockville, MD: U.S. Department of Health and Human
Services, Substance Abuse and Mental Health Services Administration, Center for Mental Health
Services, National Institutes of Health, National Institute of Mental Health.
http://www.surgeongeneral.gov/library/mentalhealth/home.html
Wales H.W., V.A. Hiday, and B. Ray. (2010). Procedural Justice and the Mental Health Court
Judge’s Role in Reducing Recidivism. International Journal of Law and Psychiatry. 33(4): 265271.
Watson A., P. Hanrahan, D. Luchins, and A. Lurigio. (2001). Mental Health Courts and the
Complex Issue of Mentally Ill Offenders. Psychiatric Services. 52: 477-481.
Wexler D.B. and B.J. Winick. (1996). Law in a Therapeutic Key. Durham: University of Carolina
Press.
Williams R. (2009). Using Heterogeneous Choice Models To Compare Logit and Probit
Coefficients Across Groups, Sociological Methods and Research. 37(4): 531-559
Wolff N., N. Fabrikant, and S. Belenko. (2011). Mental Health Courts and Their Selection
Processes: Modeling Variation for Consistency. New Brunswick, NJ: Center for Behavioral
Health Services and Criminal Justice Research.
Wolfgang M.E., R.M. Figlio, and T. Sellin. (1972). Delinquency in a Birth Cohort. Chicago, IL:
University of Chicago Press.

155

AppendixA–StudyDesignModifications,2005Ͳ2008(NewYorkOnly)

Study
Design Modifications,
2005-2008 (New York Only)
Appendix A. Study Design Modifications,
2005-2008
(New York Only)

Proposed, 8/2005

Approved, 1/2006

Current/ Modified Design, 8/2008

Process Component, to
document program logic,
philosophy, operations,
and processes including
coordination across
multiple systems.

Logic model/ case flow diagrams
development
Site Visits ± annual
Semi-structured interviews with program
staff
Semi-structured Interviews with
stakeholders (jail, court, and mental health)
not involved in intervention
Court observations (both BMHC and
business-as-usual)
Focus groups (both BMHC participants &
comparison group)
Review program records (inputs, activities,
outputs, and outcomes
Network Analysis of boundary spanners
DROPPED

Logic model/ case flow diagrams
development
Site Visits ± annual
Semi-structured interviews with program
staff
Semi-structured Interviews with
stakeholders (jail, court, and mental
health) not involved in intervention
Court observations (both BMHC and
business-as-usual)
DROPPED BAU element
Focus groups (both BMHC participants &
comparison group)
DROPPED
Review program records (inputs,
activities, outputs, and outcomes

Case flow diagrams development
(logic had been document by local
evaluators)
Site Visits ± 3 trips in 10 months.
Semi-structured interviews with
program staff
Semi-structured Interviews with
stakeholders (jail, court, and mental
health) not involved in intervention to
document BAU.
Court observations (MHCs only) ± 2
sessions/ court
Review program records (inputs,
activities, outputs, and outcomes

Impact Component, to
measure improvement
beyond BAU in offender
criminal justice and
service delivery
outcomes

Retrospective sample drawn from
administrative records; minimum 24-month
follow up period
o BMHC drawn from 2/02-12/05
o CG drawn from 1/05-12/05
Data Sources
o BMHC program records
o DOHMH records (MH)
o NYC DOC (local custody)
o NYS OCA (trx)
o NYS DCJS (CHI)

Retrospective sample drawn from
administrative records; minimum 24-month
follow up period (12/31/2007)
o BMHC drawn from 2/02-12/05
o CG drawn from 1/02-12/05
Data Sources
o BMHC program records
o DOHMH records (MH)
o NYC DOC (local custody)
o NYS DCJS (CHI)

Retrospective sample drawn from
administrative records; minimum 24month follow up period (12/31/2008)
o BMHC, 3/1/02-12/31/06
o BxMHC, 1/1/02 -12/31/06
o CG, 1/1/2005-12/31/06
Data Sources
o BMHC program records
o BxMHC program records
o DOHMH records (MH)
o NYS DCJS (CHI)

*DROPPED SERVICE DATA; COULD
NOT GAIN ACCESS

* COULD NOT OBTAIN NYC DOC DATA

Cost Effectiveness
Analysis, to determine
the cost and benefits of
each model as compared
to BAU

Calculate per unit cost of each program input
Calculate per person (per service) quantity
Examine criminal justice costs

Calculate per unit cost of each program
input
Calculate per person (per service) quantity
Examine criminal justice costs

Develop cost analysis primer to guide
future data collection and analyses.

Transferability, to
document legal and
administrative factors
necessary for replication
and to suggestion one
model over the other.

Program documentation of key components
(staffing requirements, training, policies and
procedures)
Review legal requirements, legislative
stipulations.

Program documentation of key components
(staffing requirements, training, policies and
procedures)
Review legal requirements, legislative
stipulations.

DROPPED from evaluation design but
key components incorporated into
process evaluation.

AppendixB–BronxandBrooklynMHCCourtCharacteristics




Evaluation of Criminal Justice Interventions with Mentally Ill Offenders
Bronx and Brooklyn Mental Health Court Program Matrix At-A-Glance

Bronx Mental Health Court

Brooklyn Mental Health Court

Year Established

x

February 2001 officially started taking cases; grew from
1999 pilot program

x

March 2002

Origins

x

Committee of 41 agencies convened; program developed
through consensus process

x

x

Eligibility: Began by accepting felony offenders; later
received grant to include misdemeanants

Developed by New York State Office of Court Administration
(OCA) and Center for Court Innovation (CCI) after publication
of white paper “The Revolving Door;” NY State Office of MH is
a collaborative partner

x

Eligibility: Began by accepting nonviolent felony offenders, but
progressed to accepting those with chronic misdemeanors; some
violent criminal offenders taken on case-by-case basis.

MHC Team

The Bronx MHC team includes:
x

Brooklyn MHC is a component of the Kings County Supreme Court;
the clinical team is based in the court.

Dedicated judge
The Brooklyn MHC team includes:

x

Coordinating staff: project director, two co-directors; medical
director also works as part-time psychiatrist

x

Clinical team: clinical director, 12 case managers, and 3
part-time psychiatrists;

x

Prosecuting attorneys: based in Narcotics Bureau of the
Bronx District Attorney’s Office; rotating Assistant D.A.s
(ADA) handle cases; Bureau chief serves as court liaison.

x

Defense counsel: rotating court-assigned indigent defense
counsel (18B attorneys, Legal Aid Society, Bronx Defenders)
or private counsel

x

Dedicated judge

x

Project director/clinical director (dedicated)

x

Clinical team: dedicated senior social worker and 3 forensic
counselors (BMHC employees), consulting psychiatrists –
rotating

x

ADA prosecutor (dedicated)

x

Defense counsel: attorneys rotate, but one Legal Aid attorney
handles the majority of MHC cases

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Bronx Mental Health Court
Target Population

Eligibility Requirements

Brooklyn Mental Health Court

x

Felony (first time and predicate) and misdemeanor offenders
with severe and persistent mental illnesses (SPMIs), as
indicated by Axis I diagnoses, or substantial history of
hospitalization or poor functioning; co-occurring substance
use is prevalent among clients

x

Felony (first time and predicate) and chronic misdemeanor
offenders, with SPMI/AXIS 1 diagnoses, where mental illness
contributed to criminal activity; assessment must indicate that
treatment may help the offender lead a crime-free life in the
community

x

93 % are felony offenders (UI analysis 2011)

x

80% are felony offenders (UI analysis 2011)

x

Misdemeanor and felony offenders, aged 16 and older, who
are eligible for jail diversion

x

Misdemeanor and felony offenders, aged18 years and older;
youth aged 16-17 are considered on case-by-case basis (began in
2003-2004)

x

Clinical eligibility determined by TASC (Axis 1 disorder,
SPMI)

x

Clinical eligibility determined by court clinical team psychiatrist
(Axis 1 disorder, SPMI - schizophrenia, bipolar disorder, major
depression, schizoaffective disorder)

x

Mental illness contributed to criminal activity; willingness to
enter treatment
Excludes murder and rape as “charge rule-outs”; other violent
offenses considered on case-by-case basis

x

Excludes: murder, sex offenses, and arson ; other violent
offenses considered on case-by-case basis

x

DA has discretion to refuse participation

x

Participation is voluntary; clients enter guilty plea to begin
program

x

Client must be stable and competent to stand trial

x

Client must not have participated in and successfully
completed MHC program in past

x

Program Duration

x

DA has discretion to deny case participation

x

Participation is voluntary; clients enter guilty plea to begin
program

x

Defendant must be stable and competent to stand trial

x

Client may have participated in MHC program in past

x

6-12 months for misdemeanor offenders, generally

x

12 months (minimum) for misdemeanor offenders

x

18-24 months for felony offenders, generally

x

12-18 months for 1st time felony offenders

x

18-24 months for 1st time violent or predicate felony offenders

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Bronx Mental Health Court
Case Flow/ Referral
Sources

x

225 participants under MHC supervision on any given day;

x

Caseflow: Cases identified pre-arraignment and before
plea/trial; referred to MHC through various sources (see
above); Project Director processes referrals and assigns preplacement case manager; individual is screened for mental
illness by TASC case manager; judge gives final approval
based on recommendation of TASC MHC Team, DA, and
defense attorney; participation is voluntary, but must enter
guilty plea; individual participates in treatment; individual
graduates when court feels treatment plan goals are achieved.

x

Referral sources include: DA (especially Narcotics Bureau
an estimated 85-90% of cases come from narcotics), defense
attorneys; criminal court; criminal court MH unit; 730
hearings on competency to stand trial; Bronx TASC (e.g.,
TASC team for drug court); felony and general arraignment
screening assessment; community providers; jail MH staff;
probation; ATI/case management programs; family; self
(clients may self-refer)

Brooklyn Mental Health Court
x

Approximately. 10 cases referred/month; 29 per quarter; ~275
graduates at 1/2009. at any point in time, there are about 135
people in treatment in the community and 20 people in Rikers
Island awaiting placement and about 30 on remand

x

As of October 2005, Brooklyn MHC had received 439 referrals,
accepted 217 cases, and graduated 76 clients (113 were active, 20
had been closed for a variety of reasons)

x

Referral sources include: DA, defense attorneys, “730”
competency proceedings (all 730 returns are referred to BMHC),
judges, other specialized courts

x

Clinical director gives final approval of mental health eligibility
based on clinical team psychosocial assessment and psychiatric
evaluation

x

Judge gives final approval for court participation based on DA,
defense attorney, and clinical team input

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Bronx Mental Health Court
Referral Process/
Screening Procedures

x

x

TASC screens individuals for clinical eligibility (biopsychosocial assessment, clinical interview, standardized risk
assessment tools, MH/Substance use/Health/Social
functioning/Criminal Justice measures); however, additional
screening is often done by referral sources (the DA’s office
pre-screens all trial and narcotics cases with a 7 question
screening form developed by BMHC, correctional referrals
often undergo full assessment). Also need Rikers Island
medical records before clinical assessment to obtain a
complete picture of illness and to ensure individuals are on
appropriate medications

Brooklyn Mental Health Court
x

Cases referred by judge, defense attorney, or prosecutor (referrals
usually occur after indictment, and median time from arrest to
referral is roughly 5 months.

x

Clinical assessment occurs after the client’s first MHC
appearance; the MHC social worker conducts the psychosocial
assessments and a consulting psychiatrist conducts the
psychiatric evaluations; MH eligibility determined by court’s
clinical director, a licensed social worker. Treatment plans
developed.

x

Assesses for risk of re-offending and violence; treatment plan is
designed to mitigate risk of violence. MHC will decline cases if
offender is deemed a community safety risks.

x

Defendant must voluntarily choose to participate in program;
give guilty plea (although all charges can reduced or dismissed
after treatment); and sign a MHC contract. Defendant participates
in treatment.

Approximately 2/3rds of clients are in custody when diverted
to the Bronx MHC (they remain in custody until placed)

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Bronx Mental Health Court
MH court x
dispositions/treatment
x

TASC DOES NOT provide treatment, only case mgmt.
services.
Use network of community agencies in greater metropolitan
area, including intermediate, residential, outpatient, and
hospital care for mental health, substance abuse, medical,
educational, and vocational needs.

x

Client generally pays for treatment through SSI or Medicaid
(treatment provider may help them sign up for these)

x

TASC Mental Health Court Unit directly contacts
community providers to link clients to appropriate services;
clients have weekly appointments at TASC office or TASC
team visits client if in inpatient care;

x

x

Individual needs to be stabilized before can be placed so may
be sent to Bellevue (males) or Elmhurst (females) or treated
at Rikers; also need to be “sober” for certain amount of time
before residential placements will accept

Brooklyn Mental Health Court
x

Clinical court team design (based within court); forensic
coordinators monitor treatment progress; conduct random drug
tests; operate family education groups and run Remand
Intervention group (new in 2008; senior forensic coordinator
only)

x

Treatment is community-based and typically includes MH
treatment, substance abuse treatment, community-based case
management, supportive housing services and vocational/
educational services. Will place participants in services outside
of borough and even in different state if necessary to ensure best
match.

x

Outpatient is primary treatment modality although residential
substance abuse treatment is available; will place in housing if
homeless.

x

About 100 treatment providers

Time from referral to placement is about 4-6 weeks

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Bronx Mental Health Court
Court Process and
Monitoring

Brooklyn Mental Health Court

x

Court is held twice/week (T/TH) as needed; mix of cases
heard

x

Dedicated court docket (Tuesday 10:00-2:00; Thursday is
spillover day)

x

Court status hearings every 3 months; court atmosphere;
Court tries to be flexible and understanding; willing to give
multiple “second chances”

x

Participants appear in court weekly in beginning, every two
weeks for first three months, and monthly for rest of duration
(more frequent court appearances can be required for
noncompliant clients).

x

Have weekly meetings with TASC (eventually reduced to
bimonthly, if doing well); TASC meets regularly with
participants in community if they are in residential treatment;

x

All non-incarcerated participants must show up at 10:00; the
program believes this creates a sense of community and reduces
stigma because participants see each other on Tuesdays and they
all have the same judge. The judge talks directly to clients, and
they usually approach the bench along with the clinical director,
defense attorney(s), and prosecution

x

Sometimes use brief period of jail time as sanction; can be
rewarded with reduced court appearances, certificates for phase
completion; other rewards.

x

Use team modality model – all staff are familiar with all clients
and meet with various clients; clients meet with assigned forensic
coordinator before each court appearances

x

Medical model: all clients meet with psychiatrist

x

Clinical team meets weekly to discuss cases

x

Weekly or biweekly drug testing (not random); many
treatment providers also do drug testing

x

TASC meets weekly to review all/problematic cases

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

Exit/Graduation criteria

Bronx Mental Health Court

Brooklyn Mental Health Court

No treatment phases or stages to define success; feel that success
could not be defined with concrete goals and instead must be
client-centric; more interested in insight and treatment
compliance than actual reduction of symptoms

Define success as no rearrests, cessation of drug use, program
participation, adherence to treatment, other subjective measures based
on client and case (individual measures of success)

When treatment plan goals are achieved; decision made by courtTASC Mental Health Court Unit, judge, DA, and defense
attorney. For certain cases, charges may be reduced or dismissed
upon successful completion.
Client receives certificate for graduation; program staff take
pictures of clients with certificates and put on bulletin board in
TASC office
If client successfully completes, they cannot work with the MHC
again in future if they receive another charge

Phased approach: 1. Adjustment (e.g., keep appointments, show up to
group)-typically awarded around 90 days, 2. Engagement (e.g., doing
something in treatment), 3. Progress, 4. Preparing to Graduate Æ
Graduation
Receive phase certificates along the way to mark progress; first
certificate is awarded at completion of Phase I, usually at first 90
days.
Participant complies with treatment mandate and does not commit
new offenses. After graduation, misdemeanants and first-time
nonviolent felony offenders have guilty pleas vacated and charges
dismissed. Predicate felons and first-time violent felony offenders
have charges reduced to misdemeanor plea, and violent offenders are
put on probation.
If participant does not graduate, s/he may have to return to jail or
prison depending on terms of plea.

Instruments/ Measures in
Use

Screening: TASC screening packet, seven-item screening, client
case information form, MHC diversion accepted/rejected form,
TASC intake memo, baseline interview (see site visit notes 1/09
for caveats to using)
Follow-up: Post-Acceptance Follow-Up Form, 6-month
interview, 12-month interview

Psychosocial assessment focuses on the client’s psychiatric, medical
substance abuse, housing, employment, education, and social history.
Psychiatric assessment conducted to obtain mental health diagnosis;
informs treatment recommendations.
Violence risk assessment; program also assesses criminogenic risk of
re-offending.

MH/Substance use/Health/Social Functioning/Criminal
Justice/insight & treatment adherence/violence/risk
behaviors/satisfaction with court program/social desirability
measures given at screening, 6 mos. out, 12 mos. out- some only
baseline, others at all 3 time points (have list of 30+ measures
used at various time periods)

Note: Information for this matrix was drawn from two sources: Bronx program materials dated December 2004, and interviews
UI conducted with the Brooklyn mental health court in October 2005. This matrix was developed in January 2009 for internal purposes. Information was updated in April of 2009 based on
new information obtained from January 2009 site visits.

AppendixC–CourtroomObservationInstruments




MHC Observation – Individual Appearance
Complete one form for each defendant appearance

Date: ___________________________________

Site: Brooklyn Bronx

Appearance start time: _________________________

Observer: ___________________________________

Defendant gender: M

Appearance type (circle one):

F

Judge: ___________________

Pre-MHC or Plea

MHC status hearing

If yes, handcuffs and/or other restraints used?

Non-MHC (describe

No-show / non-appearance

Describe:_______________________________

__________________)

Defendant incarcerated at start?

Y

N

Court participants
CHECK all parties present,
CIRCLE whether each spoke (beyond stating name or a greeting)
Also CIRCLE the person judge spoke to first.

Judge’s interaction with defendant (check all that apply)

Present

Spoke?

‰ Talked directly to D (as opposed to through attorney)

‰ Judge

Y

N

‰ Case manager

Y

N

1st

‰ Asked probing questions

‰ Resource coordinator

Y

N

1st

‰ Imparted instructions or advice

‰ District attorney

Y

N

1st

‰ Explained consequences of future compliance (e.g., phase

‰ Defense counsel

Y

N

1st

‰ Defendant

Y

N

1st

‰ Community Tx Provider

Y

N

1st

‰ Other (e.g., D family, victim)

Y

N

1st

Addressed
by judge

(Who? __________________________________)

‰ Eye contact with D (for most of the appearance)
‰ Asked non-probing Qs (Y/N or other one-word answers)

advancement, graduation, etc.)
‰ Explained consequences of future noncompliance (e.g., jail
or other legal consequences)
‰ Directed comments to audience (e.g., using the current
case as an example)
‰ Spoke off-record to D (i.e., not transcribed)
‰ Touched or shook hands with D
‰ D asked questions or made statements
‰ D displayed art/talent (e.g., told story, sang, etc.)
Other notes/impressions of the judicial interaction:
______________________________________________
______________________________________________
______________________________________________

Defendant’s overall demeanor seemed…
(circle all that apply)
Happy/satisfied
Intimidated
Angry

Forthcoming
Confused
Upset

Resentful

K:\Spc\LWinterf\MH TO\Kamala\Court_obs_appearance_v3_fmt.doc, 4/19/09

Compliance Status (circle only one):

Good report

Bad report

Circle “bad report” if any noncompliance noted.

Achievements recognized (check all that apply)

If bad report,

‰ Treatment attendance/participation

a) Noncompliance was (circle all that apply)

‰ Drug-free days: How many? _____

Treatment absence(s)

Missed court date(s)

‰ Phase advancement

Positive drug test(s)

Re-arrest

‰ Job/school event

Returned on warrant

Violated rules at treatment

‰ Eligible for graduation

Poor attitude

‰ Other (describe:______________________________)

Other (describe: _________________________________)

Any rewards administered? (check all that apply)

b) Court’s response was (check all that apply.)

‰ Courtroom applause

‰ None

‰ Shook hands with judge

‰ Adjustment to tx plan (how? ___________)

‰ Praise from judge

‰ Admonishment from judge

‰ Praise from other staff (Who? ____________________)

‰ Admonishment from other staff (Who?______________)

‰ Deceased court appearances

‰ Jail time

‰ Other (describe ______________________________)

‰ Defendant failed MHC (Indicate sentence or other
consequence _________________________________)
‰ Other (describe_______________________________)
c) Judge raised his/her voice while responding?

After the hearing,
the defendant:
‰ Was put in custody
‰ Left the courtroom
‰ Remained in court (where? ____________________)

Y

N

After the hearing,
defense counsel appeared:
Satisfied:
Upset:

not at all
not at all

somewhat
somewhat

very
very

Additional notes & impressions:

Appearance end time: ___________________________

Duration of appearance: ____________ minutes (rounded)

K:\Spc\LWinterf\MH TO\Kamala\Court_obs_appearance_v3_fmt.doc, 4/19/09

BROOKLYN -- Judge ________________
Tuesday, ____________
Observer: _________________________________
Appearance start time: __________________

COURTROOM DYNAMICS

COMPLIANCE STATUS: Good, bad, mixed?

COURT PARTICIPANTS
-CHECK all parties present.
-CIRCLE whether each spoke (beyond stating name or a
greeting) and
-CIRCLE the person the judge spoke to first.

‰ Good report—recognized ANY compliance
Details of achievements

Appearance end time: ___________________
Appearance type
‰ MHC status hearing
‰ MHC graduation / sentencing
‰ Pre-MHC or plea
‰ Non-MHC (describe ___________________)
‰ No-show or non-appearance
‰ Other/unknown (describe _______________)
Defendant gender
‰ Male
‰ Female
Defendant race
‰ White
‰ Black
‰ Other (describe______________________)
Defendant incarcerated at start?
‰ No
‰ Yes

Notes

Present
‰ Judge
‰ Case manager
‰ Resource Coord.
‰ District attorney
‰ Defense counsel
‰ Defendant
‰ Treatment Provider
‰ Other (e.g., D family,
victim, who?)

Spoke?
Y
N
Y
N
Y
N
Y
N
Y
N
Y
N
Y
N
Y
N

Addressed
1st by
Judge
1st
1st
1st
1st
1st
1st
1st

JUDGE’S INTERACTION w/ DEFENDANT
Check all that apply
‰ Eye contact w/ D (for most of appearance)
‰ Asked non-probing Qs, Y/N or other one-word
answers (talked directly to D, not through attorney)
‰ Asked probing Qs
(talked directly to D, not through attorney)
‰ Imparted instructions or advice
(e.g., consequences of compliance/noncompliance)
‰ Directed comments to audience
(e.g., using current case as example)
‰ D approached bench
‰ Spoke off-record to D (i.e., not transcribed)
‰ Touched or shook hands with D
‰ D asked questions or made statements
‰ D displayed art/talent (e.g., told story, sang)
Describe J demeanor:

Check all rewards administered
‰ None
‰ Courtroom applause
‰ Praise/recognition from Judge
‰ Decreased court appearances
‰ Phase certificate
‰ Graduation (alt. sentence? _______________)
‰ Any other notes

‰ Bad report—noted ANY noncompliance
Details of noncompliance

Check all sanctions administered
‰ None
‰ Adjustment to treatment plan
‰ Admonishment/recognition from Judge
‰ Increased court appearances
‰ Remand to jail
‰ Failed MHC (sentence? ________________)
‰ Any other notes

Defendant satisfaction with hearing
‰ Visibly happy/satisfied
‰ Neutral
‰ Visibly unhappy/upset/dissatisfied
‰ Other (describe ______________________)
DA or Defense Attorney reaction, if notable

AppendixD–AdministrativeDataVariableInventory




Administrative Data Variable Inventory
DOHMH (Comparison Group)Variables Received

Brooklyn Program Data FilesVariables Received

Bronx Program Data FilesVariables Received

X

X

X

Sex

X

X

X

Race & ethnicity

X

X

X

Age

X

X

X

X

X

Residence

X

X

Education

X

X

Marital status

X

Has children

X

Concept
VARIABLES FOR USE IN IMPACT
EVALUATION
Person identifier
DEMOGRAPHICS

SOCIOECONOMIC STATUS
Language

X

Homelessness

X

Employment & Income

X

Veteran Status

X

CRIMINAL HISTORY
X
MENTAL HEALTH STATUS
X

MH History
Current MH Status

X

X

X

X

X

X

MH TREATMENT
Services received

1

Concept
Duration of service provision

DOHMH (Comparison Group)Variables Received

Brooklyn Program Data FilesVariables Received

Bronx Program Data FilesVariables Received

X

X

X

SUBSTANCE ABUSE STATUS &
TREATMENT
Substance abuse status
Substance abuse treatment

X

X
X

X
X

HEALTH
Health insurance
X

Health status
REFERRAL, ELIGIBILITY & PROGRAM
PARTICIPATION
Referral Source

X

X

Time to referral

X

X

First referral to MHC?

X

X

Eligibility Decision

X

X

Time to eligibility decision

X

Participation Decision

X

X

Time to participation decision

X

X

Pretrial incarceration

X

X

Time from arrest to program start

X

X

Total time in MHC

X

X

CASE PROCESSING & OUTCOMES

Legal representation

X

Concurrent CJ involvement

X

Legal status

X
2

Brooklyn Program Data FilesVariables Received

Bronx Program Data FilesVariables Received

Initial charge

X

X

Initial plea offer

X

X

Concept

DOHMH (Comparison Group)Variables Received

X

Intermediate compliance outcomes
Case outcome: success or failure

X

X

Final sentence

X

X

Subsequent participation in MHC

X

X

3

AppendixE–MHCandDOHMHScreeningInstruments




MHC and DOHMH Screening Instruments

s...

Farm 3JO ADM (CC) (1100) pIge 2

INSTRUCTIONS FOR COMPLETING
SUICIDE PREVENTION SCREENING GUIDELINES-FORM 330 ADM

oI lo6tw Yo<It
COMMISS/OH Of CORRECT1OH
0II'0ct 01 MIorItII He..,

GENERAL INFORMATION
It I, reconvnended that !he form be completed in tnpIicate !of all delainH$ priot to cell aSSignment and be distributed as follows:
~Ie, second copy to medical or mental health personnel at referral, and the third copy !of use lMXOrding to facility'l procedures,

top copy in inmate',

COmmenl Column: All 'YES' response' requife note to document;
1. in!ormation about the inmate that oITiOllr leels II relevant and Important;
2. lnlormation specifically requested in questions:
3.lnlormation regarding Inmate's refusal or inability to answ er questions.
Inmate's Name:
Selt;
Date 01 birth:
MOSI Serious Charge(s):
Date:
Time;
Name of Facility·
Name 01 Screening Olliaw

Enter Inmate'slirSI sod last name and middle initial
Enter male (m) or lemale (f)
Enter month, day and year.
Enter the most serious charge or charges (no more than two (2)) from this arrest.
Enter month. day and year 10m! was completed .
Enter the time 01 day the lorm was completed ,
Enter name 01 jailor Iode·up.
PrWlt name 01 ollicer completing form
Enter NYSIO A BAC • .

P1;ychiatric Problems Dunng
Prior Incarceration: The saeening oIIicer should ask the inmate whether heJshe
anempted suiCide in the past.

"8$

INSTRUCTIONS FOR ITEMS 1·18
Ge ner.ll in st r u c tion s
ChecIIlhe appropriate YES or NO for items 1-16
If IfllonnatJon r&quired to complete tllese quesbonsls unknown to screening oIIicer, 5Ud'I inlormatlOn should be obtained by asking inmate to answer questions, However,
inmata has the rig'" to refuse to answer
If Inmate refuses to answer questions 2· 12, enter RTA (refused to answer) In the Comment Column rlIxl to each question, In addition, complete the YES or NO boMes
only If inlorma~on II known to you

If during an otherwise cooperatve Interview. inmate reluses to answer one or two questlOfl' Cileck YES in the bo_(el) next to the unanlwered question{s) and enter RTA
in the COtTIII"IBnt bo_ next to each unanswered queslIOfI
If Inmate IS unabllllO answer all quesLOns 2,12, enter UTA (unable to answer) in the Comment Column next to each question Also enter reason (e.g, not Engli$h
spealung) for not answenng these quesLOns In !he Convnent Column tIEIlCI to OuestlC)tl 2 In addillon, complete the YES or NO boxes only if infonnatton is known to you ,
O b .ervatlon of Tr an . portlng Offic er
Check YES or NO based upon the wrinentverbal report 01 the arrestingltrans~11'Ig offlCBr or upon t"e screening form completed by the arres~ng agency
If YES, notify SUpervISor

ITEM (I)

NOTe: '""' following questions and o b servatlonl Ihould not be ,.ad word l or word but ,.staled In your word.,
Pe~onal D ata Quest ion s

Familylfriends: Check NO if someone ot"er than a lawyer or bondsman would (I) be WIlling to post Inma!e's bail, (2) vi.~ inmate while he/she is Incarcer·
ated. or (3) accept a collect call from inmate
Slgnlt'tcanlloss A$k all three componenlS to this question,!oss 01 joo. loss of relationship and deat" 01 close friend or lamlly member
Worried about PfObIems Ask about such probllms as financial, medical conditIOn or flar of Ioslng;oo Check YES if Inmate at'Iswers YES to any of these
Family/significant other attempted suICIde' Significant othet is defined as someone who has an ImportantltnObOnaI fBtationship with inmate
AJcohoI or drug hiStory Check YES ,I Inmate has
pnor treatment lor alcoholldrug abuse or If prior arresls werl alOClholldl'\lg fBtated
HIStory of counseling or menial health Ivaluatoonflrealrnent Check Yes if inmate (I) has ever had psychiatric I"Iospitatiulion, (2) is currentty on psyctJo.
tropic medication, or (3) has been in ovtpatlent psychotherapy dunng past six months Note current psychotropic medication and name 01 most recent treat·
ment agency
C"eck YES if inmate expresses extreme shame as result 01 arrest or reels that arrul/detention Will cause "umlliation to sell1slgni/icant others.
Suicidal· Check YES if inmate makes suicidal statement or responds YES to direct question, ' Ale you thinking about killing yourself?' II YES,notify SUpeMsor

ITEM (2)
ITEM
ITEM
ITEM
ITEM
ITEM

(3)
(4)
(5)
(6)
(7)

"ad

(TEM (8)
ITEM (9)
ITEM (10)
ITEM (II)
ITEM (12)

PreYlOUS ane~: Check YES if inmate states he has attempted IviMe \I YES or NO. e~plore method and note scars Obtain as moch inlOtmilIion as
poss.IbIe fB method and time 01' attampt
Hopeless Check YES if inmate states leellf'Ig hopeless, Ihat he has glV1ln up. thaI he leels helpless to make hIS 1!l1 bener If YES to both ilems 10 and II ,
notify supervisor
Criminal History' Ask inmate or check files to determine II thiS IS inmate'S first Incarceration

B eh avlOf/Appea r a n ce Obse rv ation .
YES or NO must always be chec;ked lor each of these items They are observettons made by the screefllrog otricet They are not questions
ITEM (13) OtipreS$lon: Ind,caton Include behaVlOf such as cry;ng. emotlOO8l ftatness. apathy, lethargy, extreme sadness, unusually slow reactions
ITEM (1 4) Overty an~lOUs, afraid, panICked, or angry Indicators Include behaVior such as handWringing, l)ItCing, e~cessive lidgetltlg, profuse s_a~ng, cursing, physiCal
VIOlence, etc.
ITEM (15) Acting In strange manner' C"eck YES II you observe unusual behavior such as hallucltlations, severe mood swings, disorientation, withdrawal, etc.
iI Inmatl Is hearing voices telling "im to "arm himsell, make an Immedia!e re'erral 10 mental health services
ITEM (16a) Undlr Inftuence' Check YES if inmate 15 apparently intOXICated on drugs or alcohol
ITEM (16b) Incoherence, Withdrawal , or mental jttness~ Means physICal Wllhdrawalfrom substance If YES to both a & b, nollfy supervisor
CO MMENTS I IMPRES SIONS: No te any ~gulH f . .llngl

0(

Impreilion r e lulclde rllk.

SCORING
Count all checks

ItI

Column A Enter total Notify supervisor If (I) total is 6 or more, (2) any s"aded area IS checked , (3) il you leal notlffCaoon is appropriate

BOOKING OFFICER SIGNATURE AND SHIELD NUMBER
Sogn Iorm and enter shield number

DISPOSITION
Corrections Personnel ' SUpelVlsor nob!ied ' check YES or No Notification should be made prior to cell assignment.
Note if constant supervisloo Instituted
Note emergern::y/non'lmergency re'erralto medical andlor mental "ealth personnel
Medical/Mental Health Personrlli:

MedICal/mental IIealth staff should notl recommeodallOl'II and actions taken

FOtm 330 AOM ICC) (1/00)

Slale 01,..,. 'I'or\
COMMISSION Of CORRECTION
0IIice 011 t.W<'IIIII HNIIII

SUICIDE PREVENTION S CREENING GUIDELINES
I"!"""'TES "lAME

I'"

"lAME Of FACILITY

DATE Of BIRTH

I

DATE

MOST SERIOUS CHARGEtS)

I-"·"-.d...n-

"lAME Of' SCIU;ENINC OFFICFR

psychl.tric probMms durlfffl
pnorlnc.re.... lion
YES _ _ _

I

TIME

'0 ___

Chedc appropriate column 'IX each questoon
NYSIO & B&C.

,

Colum n

Column

A
YES

6
NO

Gen eral Co mm enl$lObs ef'Yations
All "YES' Responses Require
NOle to Documen t

OBSERVATIONS OF POLICEfl'RANSPORTING OFFICER

,

Poloce IX transportlog Oi'f1C8f beheves that Inmale may be a suICIde nsk
If YES , notify ,upervls or.

,-

~,-

PERSONAL DATA

3

IllInale lacks supporl of lamlly or Iflends

In

Ihe commufllly

,0,

Inmale has expenenced a l'lIOIrlCanllQss ""Ih,,, ItIe lasl SIX monlhs
M:l5S 01 JOb. Joss of relallOnshlp. dealh 01 close lamlly mem1>er)

• ,0,

,
6

,

•
,
.n,

Inmate IS very womed aboul maJOr problems other lhan legal slluabOn
senous rlllilrlClill Of lamll~ problems . a medical condItion or fear 01 loSIng JOb)

Inmale S famIly member or • .gnlficanl othlll" (spouse. palenl . close 1nIII"Id.1over)
lias attempled Of commlUed suICIde
Inmale has hIstory of drug Of alcohol abuse (Note drug and when last used)
Inmale has history 01 eounsebng Of mental health evaluallOnltreatmenl
(Nole current psychOlroplC medIcations and name 01 masl recenllrealment agency)
Inmate expresses utrerne embarrassment. Shame. Of feel,"!!s of humdlallOfl
as result of ckargelincarcerabon (conSIder Inmale·s poS4 l1on In commufllly
and shoc;klf19 nalure 01 cnme I
Inmale 1$ Ihlflklflg abc)ullolhng hImself
II YES . notify .upervl.or.
Inmale has prevIOUS suICIde allempt (Explore melhod and check lor scars)

b Anempl occurred ""lhln last mOllth

"

"
"
"
"

InlTllllte IS expreSSIng leehngs of hopelessness (nol/'lltlg to look lorward to)
TI'lIS 1$ Inmate S firstlncarcerallOn In loc kupfja ll

BEHAVIOR/APPEARANCE
InlTllllle shows s.gns 01 depressoon (e g .. cryIng. emotIOnal naltlus)
Inmale appea,s overly anxOOU1>. panICked. aha><l Of angry
Inmate IS actIng and I or Ulliong In II strange manner
(8 g . cannot locus atten~on. heafl ng Of seeIng II"IIn9S whlCl"l are no! there)

16a Inmate IS aPJ)3ret1tly IIndlll" J/Ie Influence 01 ak::ohol or dt\Igs

b If YES, ",nmate incol1e,ent. Of ShowIng s.gns Of wllhdra wlIl
If YES to bolt.. & b, notify I Upervi5Of.

Of menUlll llness?

TOTAL Col .. "," A
O/I.ce.s Comments ' ImpreSSIOns

ACTION
II total checks '" Column A are 8 Of more. or any shaded box IS checked, Of If you leel It IS neeessa.ry. nobly supervIsor and ;nSlltute constanl watch
SUpervisor NoMie<I

YES

Constant SUpervl!;lOll Instlluted

YES

Inmate Fielen"ed to MedICal I M8f11a1 Heallh
YES

"0

"0
"0

SupeNIW/·S Stgnilture
EMERGENCY

N ON.£MEIIGEN C Y

medicat

"""""

IIYES

mental heallh

SlQnatufe al'ld ShIeld Number of Scree",ng orrK;er~
MedlClllJMental Health Personnel AC1.ans. (To be completed by med.eaVMH staff)

menial heallh

AppendixF–BivariateAnalysisofMHCParticipantsandNonparticipants




Bivariate Bivariate
AnalysesAnalysis
of Mental
Health Court Participants and
of MHC Participants and Nonparticipants
Nonparticipants
Although the primary objective of the current study focused on assessing the impact of
mental health court participation on criminal justice outcomes, additional analyses 1 were
conducted on datasets from both the Bronx and Brooklyn mental health court (MHC)
programs to identify and examine potential differences between MHC participants and
nonparticipants. T-tests were performed on each dataset to examine bivariate differences
between MHC participants and individuals who were referred to the mental health court
but, for a variety of reasons, did not participate (findings of statistical significance are
reported at the .05 level). The results of these analyses are presented in this appendix.
Comparison of Participants and Nonparticipants in the Bronx Mental Health Court
A total of 815 individuals were referred to the Bronx MHC between 2002 and 2006, of
whom 648 (80%) participated. The remaining 167 (20%) did not enter a plea into the
mental health court. As Table 1 indicates, participants and nonparticipants in the Bronx
MHC differed on many demographic characteristics.2 Participants were significantly
older, more likely to be female, more likely to be Hispanic, and less likely to be Black,
when compared to nonparticipants. The time from arrest to screening was longer for
participants than it was for nonparticipants. Additionally participants were less likely than
nonparticipants to have been incarcerated pretrial.
Table 1. Comparison of Bronx MHC Participant and Nonparticipant Referrals
Comparison of Bronx MHC Participants and Nonparticipant Referrals (N=815)
Participants
Non-participants
(N=648)
(N=167)
Age at arrest*
37.3 years
35.3 years
Male**
62%
71%
White, non-Hispanic
7%
5%
Black, non-Hispanic**
34%
43%
Hispanic*
58%
50%
Other race or ethnicity
1%
2%
Incarcerated at the time of MHC screening*
59%
66%
Total number of MHC referrals per person in 200206**
1.1
1.0
Days from arrest to initial TASC screening
(mean)**
266.0
197.1
Days from arrest to initial TASC screening
(median)
148.5
107
Case transferred to MHC from drug court
37%
n/a
Days in drug court before transfer to MHC (mean)
226.6
n/a
Days in drug court before transfer to MHC
(median)
77.5
n/a
* p<.10
** p<.05
1

T-tests were performed to examine bivariate differences between BMHC participants (N=327) and the remaining 192
individuals who were referred to the Brooklyn mental health court but, for a variety of reasons, did not participate.
2
Findings reported here are statistically significant at the p<.10 level.

The program database suggests that nonparticipants in the mental health court may have
been more severely ill than participants. Although diagnostic information was not
available for 2002-VWDIIDVVLJQD³SURJUDPFRGH´WRFODVVLI\FDVHVDV³PRUH´RU
³OHVV´VHYHUHO\PHQWDOO\LOOFour in 10 participants were described as more severely ill,
whereas one-half of non-participants were described as such.
Unfortunately, data on mental health diagnoses were not captured in the program
database during the 2002-06 study period. However, a subset of individuals who were
referred to the mental health court appeared in the DOHMH data from 2005-06. This
included 153 participants and 50 individuals who were referred but did not participate.
The DOHMH data on this subset of persons referred to the Bronx mental health court
suggest that participants and nonparticipants were similar with respect to the types of
diagnoses they had. However, we observed that nonparticipants had twice the length of
stay in jail that participants did (mean of 167 days compared to 77).
Table 2. Comparison Mental Health Status in a Subset of Bronx Court Participants
and Nonparticipants Who Received Jail-based Mental Health Services
Comparison Mental Health Status in a Subset of Bronx Court Participants and
Nonparticipants Who Received Jail-based Mental Health Services
Participants
Non-participants
(n=153)
(n=50)
Assessed as SPMI
75%
64%
Received medication in jail
96%
94%
Housed in Mental Observation Unit
13%
14%
Global Assessment of Functioning Score
54
56
Axis I Adjustment Disorder
2%
7%
Axis I Mood Disorder
40%
43%
Axis I Psychotic Disorder
23%
21%
Axis I Substance-related Disorder
29%
29%
Axis II Adjustment Disorder
0%
0%
Axis II Mood Disorder
19%
22%
Axis II Psychotic Disorder
5%
5%
Axis II Substance-related Disorder
66%
65%
Substance use on either Axis I or Axis II
66%
79%
Length of stay in jail
77 days
167 days
Source: Urban Institute analysis of DOHMH data.

The top reason for nonparticipation was that clients withdrew themselves from
consideration (24%). Following withdrawals, nearly 1 in 5 nonparticipants did not have a
qualifying mental health condition, and 10 percent of nonparticipants were excluded from
the program because of criminal justice objections from either judges or prosecutors.

Table 3. Reasons for Non-participation in the Bronx MHC
Reasons for Non-participation in the Bronx MHC (N=166)
Client withdrew application
No mental illness, transferred to another program
Other
Client ineligible (pre-placement) due to psychiatric reason
Criminal justice system opposition (e.g., Judge, DA)
Client failed/ineligible during screening, before treatment placement
Client sentenced to incarceration prior to treatment*
Client left treatment after placement
Client rearrested, warranted, or committed violation
Probation to supervise or other non-incarcerative sentence
Client ineligible (pre-placement) due to medical reason or death
Client failed to report to treatment placement
Client sentenced to incarceration after treatment placement*
* This may have resulted from another case the defendant was involved in.

Comparison of Participants and Nonparticipants in the Brooklyn Mental Health Court
A total of 519 individuals3 were referred to the Brooklyn MHC between March 1, 2002
and December 31, 2006 (the parameters for the current study), of whom 327 participated
(63 percent) in the program.
Table 4 facilitates closer examination of common Brooklyn MHC referral sources by
presenting referral data for both the 519 cases referred to the court program between 2002
and 2006 and the subset of 327 individuals who participated in the Brooklyn court
program. As shown in Table 4, defense attorneys are the most common referrals source
accounting for over half (55 percent) of all cases referred to BMHC, and more than twothirds (65 percent) of program participants. In contrast, the District Attorney accounted
for just 10 percent BMHC referrals and about 7 percent of referrals were made by judges
handling other criminal dockets. About a quarter of all cases were referred through
mental competency process (i.e., 730 proceedings), and about 14 percent of BMHC
participants. A handful of cases were referred from other sources, including other
specialized problem solving courts such as drug courts. A closer look at Table 4 also

3

Statistics reported through tKHPHQWDOKHDOWKFRXUWSURJUDP¶VTXDUWHUO\UHSRUWVUHIOHFWDKLJKHUQXPEHURI
cases in the 2002-2006 period. A number of legitimate methodological differences account for the
difference. First, the mental health court program reports on cases referred to the program, whereas this
study focuses on persons referred, some of whom were referred to the court as many as four times during
the study period. Second, the mental health program records had to be matched to records in the New York
State criminal records repository in order to be included in the study. (Refer to Chapter 2 for additional
detail.) The mental health program database included records of 673 referrals to the court, but 104 records
(the majority of which did not result in court participation) did not contain sufficient identifying and arrest
date information to facilitate a match to the New York State criminal records repository. One other record
was excluded from the data provided to the study for indeterminable reasons. As a result, the study was
based on 568 case referrals to the mental health court, representing 519 unique individuals who were
referred to the mental health court between 2002 and 2006.

24%
17%
17%
14%
10%
7%
5%
2%
1%
1%
1%
1%
1%

Table 4. Referral Mechanisms to Brooklyn MHC for All Cases, 2002-2005
Referral Sources for Brooklyn MHC (N=519)
All
(n=519)
730 Return
22.5%
730 Found Fit
1.7%
Defense Attorney
54.9%
District Attorney
10.2%
Specialized Courts
1.9%
Other Judge Referrals
7.5%
Other
1.2%

Participants
(n=327)
12.5%
1.2%
65.1%
11.0%
2.1%
7.0%
0.9%

Non-Participants
(n=192)
39.6%
2.6%
37.5%
8.9%
1.6%
8.3%
1.6%

indicates that nonparticipants may also have had more serious mental illness than
Brooklyn MHC participants: almost 40 percent of nonparticipants were referred to the
Brooklyn MHC though the mental competency process compared to 13 percent of
program participants and 22 percent of all referrals.
In general, BMHC participants were similar to nonparticipants with respect to
demographic characteristics; analyses did not find any statistically significant differences
with regard to age, gender, race or ethnicity. In turn, BMHC participants and
nonparticipants were similar with respect to homelessness at the time of program referral.
However, BMHC participants and nonparticipants varied on severity of their mental
illnesses. Nonparticipants included individuals who were disqualified for having
insufficiently severe mental health conditions, as well as others whose conditions were
too severe to be suitably placed with existing community-based treatment providers.
Additionally nonparticipants were often excluded on the basis of their criminal charges,
and about one-fifth declined to participate.
As shown in Table 5, GHIHQGDQWV¶ILUVWFRQWDFWZLWKWKHBrooklyn MHC generally
occurred about 6 months after arrest (the median is 176 days). Those who ultimately
participated differed from nonparticipants with respect to the time it took to access the
mental health court. For BMC participants, the time from arrest to court contact was
about 5 months (median of 149 days), but for those who did not participate, the time to
court contact was significantly longer at 7 to 8 months (median is 222 days). As
mentioned earlier, 57 percent of BMHC participants had been incarcerated pretrial; by
contrast, a much greater share of nonparticipants, 90 percent, had been incarcerated pretrial. Additionally, nonparticipants nearly always had felony changes (99 percent),
whereas 84 percent of participants had felony charges. These differences are statistically
significant (at the .05 level); taken together, they suggest that individuals who did not
participate in the BMHC had more extensive criminal histories, more serious instant
offenses, or more complicated cases from a criminal justice perspective.

Table 5. Comparison of Brooklyn MHC Participants and Nonparticipants
Comparison of Brooklyn MHC Participants and Nonparticipants (N=519)

Time from arrest to first court appearance (mean)
Time from arrest to first court appearance (median)
Instant offense was a felony
Incarcerated at time of referral
Time from 1st court contact to eligibility determination (mean)
Time from 1st court contact to eligibility determination
(median)
--Days from 1st court contact to 1st appearance (mean)
--Percentage who had a psychiatric assessment
Percentage found eligible for MHC
Time from eligibility decision to plea offer, if found eligible
(mean)
Time from eligibility decision to plea offer, if found eligible
(median)

Participants
(n=327)
245 days
149 days
84%
57%
38 days

Nonparticipants
(n=192)
374 days
222 days
99%
90%
56 days

28 days
1 day
97%
100%

30 days
0 days
46%
8%

55 days

49 days

30 days

25 days

Once defendants had been referred to the mental health court, the overall time to
eligibility determination was between 1 and 2 months (the mean was 45 days). One-fifth
of nonparticipants either declined to proceed or were found to be ineligible immediately
after referral; another one-third declined to proceed or were found ineligible at some
point before a psychiatric assessment could be completed. Interestingly, despite these
early ineligibility decisions, the average time from initial court contact to eligibility
determination was significantly longer for nonparticipants than participants.
As Table 6 indicates, the main reasons for nonparticipation were mental health
ineligibility, criminal justice ineligibility, and client refusal. Mental health reasons were
the most common: 24 percent of nonparticipants did not have qualifying Axis I
conditions, while another 15 percent were assessed as too ill for community-based
treatment. Comparing participants with the limited set of nonparticipants who completed
psychiatric assessments, we found that nonparticipants often had substance induced
mental health symptoms. While BMHC participants and nonparticipants were just as
likely to have substance abuse problems, about one-third of assessed nonparticipants (i.e.,
those who had psychiatric assessments) were found to have substance-induced mental
health symptoms without other underlying mental health disorders. Additionally there
were defendants ZKRZHUHLQYROYHGLQ³´PHQWDOFRPSHWHQF\SURFHHGLQJVZKRPD\
not enter a guilty plea to the mental health court if found incompetent to stand trial.
Other common reasons for nonparticipation were criminal justice ineligibility and client
refusal. Close to one-third of nonparticipants were disqualified on criminal justice
JURXQGVHLWKHUWKHMXGJHRUSURVHFXWLQJDWWRUQH\IHOWWKHGHIHQGDQWV¶FKDUJHVRUFULPLQDO
history were too serious to offer an alternative to incarceration. About 20 percent of
nonparticipants refused the program. Defendant refusals typically occurred before the
psychiatric assessment, but a small share declined to participate after being found
eligible. Additionally, a handful of nonparticipants were described as uncooperative or
LQDGHTXDWHO\PRWLYDWHGWRSDUWLFLSDWHLQWUHDWPHQW)LQDOO\³RWKHU´UHDVRQVIRU
nonparticipation were noted in the program database, but without supporting details.

Table 6. Reasons for BMHC Nonparticipation
Reasons for nonparticipation among referrals to the mental health court (n=192)
Ineligible for Mental Health Reasons
39%
No Axis I MH Condition
24%
No Axis 1 Diagnosis
19%
Referred to Substance Abuse
Treatment
5%
Too unstable or severely ill
15%
Too Unstable for Community
Treatment
5%
730 Return
9%
Ineligible for Criminal Justice Reasons
29%
Rejected on Criminal Justice
Grounds
10%
Candidate Rejected by District
Attorney
19%
Client Refusal or Noncooperation
22%
Refused Before Assessment
11%
Refused Assessment
3%
Refused After Being Found Eligible
6%
Failed to Cooperate
1%
Inadequate Motivation for Treatment
2%
Other
10%
Other Kind of Eligible
Nonparticipant
2%
Other
8%

AppendixG–ListofCovariates




List of Covariates Examined in PSM Analyses
Appendix G: List of Covariates Examined in PSM Analyses
Variable
black
white
hispanic
gender
agesex
i_age
i_agesq
kings
bronx
npriorarr
age1starr
nanyvfo
nfirearm
nkidsex
nkidnonsex
ndrug
nweap
ndwi
nsolfac
nasslt
nhomi
nsex
nkidnap
nburg
nprop
nrobbery
nfraud
nmother
nfelony
nlesser
nfelcon
nconvic
ndrugsale
npublic
naddict
nbetwarr
nfelconsq

Description
1= black, non-Hispanic, 0 otherwise
1 = white, non-Hispanic, 0 otherwise
1= Hispanic, 0 otherwise
1 = male, 0 otherwise
Interaction of age and sex
Age at instant arrest
Age at instant arrest (Squared)
1 = R arrested in Kings County, 0 otherwise
1 = R arrested in Bronx County, 0 otherwise
Number of prior arrests
Age at first arrest
Number of prior arrests with any violent felony charge
Number of prior arrests with any firearm charge
Number of prior arrests with child victim sex offense charge
Number of prior arrests with child victim non-sex offense
charge
Number of prior arrests with drug crime charge
Number of prior arrests with weapons charge
Number of prior arrests with DWI charge
Number of prior arrests with solicitation or facilitation charge
Number of prior arrests with assault top charge
Number of prior arrests with homicide top charge
Number of prior arrests with sex offense top charge
Number of prior arrests with kidnapping top charge
Number of prior arrests with burglary top charge
Number of prior arrests with larceny or other theft top charge
Number of prior arrests with robbery top charge
Number of prior arrests with fraud top charge
Number of prior arrests with miscellaneous other top charge
Number of prior arrests with felony top charge
Number of prior arrests with non-felony top charge
Number of prior felony conviction
Number of prior convictions prior to plea date
Number of prior arrests with drug sale disposition
Number of prior arrests with public disorder charges
Number of prior arrests with illegal substances and gambling
charge
Number of arrests while in custody
Number of prior felony conviction (squared)

ndrugsalesq
nhomisq
nfirearmsq
robb1st
burg1st
asslt1st
drgsale1st
anyvfo1st
drug_alc
drug_coc
drug_mar
drug_her
drug_oth
drug_unk
i_attempt
i_vfo
i_gun
i_weapon
i_child
i_dwi
i_drug
i_rob
i_aggast
i_larcn
variety_all
variety_sev
recidivism1
recidivism2
axis1_adj
axis1_mood
axis1_psych
axis1_subst

Number of prior arrests with drug sale disposition (squared)
Number of prior arrests with homicide top charge (squared)
Number of prior arrests with any firearm charge (squared)
 5¶VILUVWcrime was robbery, 0 otherwise
 5¶VILUVWFULPHZDVEXUJODU\RWKHUZLVH
 5¶VILUVWFULPHZDVDVVDXOWRWKHUZLVH
 5¶VILUVWFULPHZDVGUXJVDOHRWKHUZLVH
 5¶VILUVWFULPHZDVYLROHQWIHORQ\RIIHQVe, 0 otherwise
1 = Self-reported use of alcohol, 0 otherwise
1 = Self-reported use of cocaine (either crack or powder), 0
otherwise
1 = Self-reported use of marijuana, 0 otherwise
1 = Self-reported use of heroin, 0 otherwise
1 = Self-reported use of other drug, 0 otherwise
1 = Unidentified use of drug based on AXIS I or II diagnosis,
0 otherwise
 5¶VLQVWDQWRIIHQVHLQYROYHGDQDWWHPSWFKDUJH
otherwise
 5¶VLQVWDQWRIIHQVHZDVYLROHQWIHORQ\RIIHQVHRWKHUZLVH
 5¶VLQVWDQWRIIHQVHZDVILUHDUPDUUHVWRWKHUZLVH
 5¶VLQVWDQWRIIHQVHZDVZHDSRQ-related, 0 otherwise
 5¶VLQVWDQWRIIHQVHLQYROYHGDFKLOGvictim, 0 otherwise
 5¶VLQVWDQWRIIHQVHZDV':,-related, 0 otherwise
 5¶VLQVWDQWRIIHQVHZDVGUXJ-related, 0 otherwise
 5¶VLQVWDQWRIIHQVHZDVUREEHU\RWKHUZLVH
 5¶VLQVWDQWRIIHQVHZDVDJJUDYDWHGDVVDXOW0 otherwise
 5¶VLQVWDQWRIIHQVHZDVODUFHQ\RWKHUZLVH
Number of different types of crime R has committed
Number of different types of serious crime R has committed
1 = R was re-arrested, 0 otherwise
1 = R was re-convicted, 0 otherwise
1 = R was diagnosed with adjustment disorder (AXIS I), 0
otherwise
1 = R was diagnosed with mood disorder (AXIS I), 0
otherwise
1 = R was diagnosed with psychotic disorder (AXIS I), 0
otherwise
1 = R was diagnosed with substance-related disorder (AXIS
I), 0 otherwise

axis1cat (string)

Additional codes
Adjustment disorder
Anxiety disorder
Attention-deficit and disruptive behavior
Delirium, dementia, and amnestic
Dissociative disorder
Impulse-control disorder
Learning disorder
Mental disorder due to general medical
Mental retardation
Mood disorder
No AXIS I or AXIS II disorder
Personality disorder (Axis II)
Psychotic disorder
Sexual and gender identity disorder
Sleep disorder
Substance-related disorder
Tic disorder

AppendixH–PropensityScoreMatchingandSensitivityAnalysis

Propensity
Score
Matching
Sensitivity
Analysis
Appendix H: Propensity
Score
Matching
andand
Sensitivity
Analysis
PROPENSITY SCORE MATCHING

It is a challenge to draw a plausible causal claim in observational studies because we can
never observe the outcome of a treatment and control condition at the same time for a
given individual. In other words, one cannot be in treatment and control conditions at the
same time. Propensity score matching methods are frequently used with observational
data to answer a counterfactual question; that is, what would have happened to treated
cases had they not received treatment? The primary objective of propensity score
matching is to identify a set of untreated units that resemble the treatment group in all
observable ways possible.
In this study, the propensity score, e(x), can be defined as follows:
e(x) = P(Mental Health Court Participation =1|X)
where Mental Health Court Participation indicates the treatment assignment (0, 1), and X
represents a vector of observed covariates. If we take individuals with the equivalent
propensity score and assign them separately to a treatment and control condition, the
treatment and control groups would differ only on their error term in the propensity score
equation. Given that the error term is presumably independent of the covariates, the
predicted probability of treatment assignment is independent of the outcome ± recidivism.
This analytic framework can effectively address selection bias. It is conceivable that
those who are motivated for self-improvement would be willing to participate in the
mental health court program and would also refrain from recidivism even in the absence
of treatment. By balancing such characteristics of treated and untreated individuals,
propensity score matching creates an experimental framework by which to draw causal
inferences in a reasonably rigorous fashion.
That said, propensity score matching methods require a strong assumption that treatment
assignment or outcome measures are not systematically affected by unobserved variables.
This is ultimately an untestable assumption, but warrants caution in the application of
propensity score matching. For this reason, it is of critical importance to apply propensity
score matching strictly to the extent permissible in a given dataset. In particular, while
conventionally supported in most regression-based approaches, it is not desirable in
propensity score matching to extrapolate inferences outside the range of observed data
points.
If treatment and matched comparison cases do not share much commonality, one should
remove such cases in the estimation of treatment effect. Similar to experimental designs,
the internal validity of findings is usually implied in propensity score matching to be far
more critical than the external validity of findings. These careful considerations intrinsic
to the implementation of propensity score matching necessitate sensitivity analysis on
treatment effect and a careful review on balance between the treatment and comparison

groups. In what follows, we thus report the extent that our outcome analyses are sensitive
to hidden bias in propensity score matching.
SENSITIVITY ANALYSIS

Below are the results from a sensitivity analysis of the propensity score estimates of the
effect of MHC participation on recidivism. This analysis is recommended by Rosenbaum
(2002) and can provide a sense of how reliable the PSM results are. If the probability of
receiving treatment (in this case, participation in the mental health court program) is
explained completely by observed covariates, the PSM results would be free of any
hidden bias. However, this is almost never true in any PSM studies. The sensitivity
analysis is thus to test the presence of hidden bias in the PSM results. It assesses the
potential impact of hidden bias arising from confounding variables associated with both
treatment and outcome variables.
Suppose that all the relevant covariates are included in the propensity score model. The
estimated propensity score of a treated subject and control subject with identical
characteristics should be equal. If there is an unobserved process that influences
treatment assignment, however, treated and control subjects that have identical
characteristics can have a different propensity score. As a result, the odds ratio of a
matched pair of treated and control subjects with identical characteristics will no longer
be equal to one. The larger the effect of an unobserved covariate on treatment
assignment, the larger the difference between the odds ratio and one will be. Also,
Rosenbaum (2002) demonstrates the odds ratio for matched pairs is bounded by the
function of the strength of the effect. The sensitivity test thus estimates the upper and
lower bounds of each strength level and their corresponding p-values. If the unobserved
covariates affect the treatment assignment and the outcome at a strength level greater
than the critical effect strength, the average treatment effect could potentially be
indistinguishable from zero. Further details on the procedure employed here can be found
in Rosenbaum (2002) and Apel et al. (2010).
1. Bronx
7KHUHVXOWVRIWKHHVWLPDWLRQRIµµ5RVHQEDXPERXQGV¶¶RQWKHHIIHFWRI0+&WUHDWPHQW
are shown below. Gamma coefficients Ȗ signify the odds ratio for the effect of
unobservables on the likelihood of receiving MHC treatment for individuals who actually
participated in MHC treatment versus individuals who did not participate.

Table 1. Sensitivity of Treatment Effect Estimates to Hidden Bias (Bronx MHC)

+

Gamma Q MH
1.0
1.4921
1.1
2.0345
1.2
2.5304
1.3
2.9882
1.4
3.4140
1.5
3.8121
1.6
4.1862
1.7
4.5394
1.8
4.8740
1.9
5.1921
2.0
5.4955

Re-Arrest
Re-Conviction
+
+
P MH
P MH
Q MH
Q MH
P+MH
P-MH
1.4921 0.0678 0.0678
0.7813
0.7813
0.2173 0.2173
0.9534 0.0209 0.1702
1.3672
0.1972
0.0858 0.4218
0.4613 0.0057 0.3223
1.9023
0.1733
0.0286 0.4312
0.0090 0.0014 0.4964
2.3956
0.6642
0.0083 0.2533
0.2324 0.0003 0.4081
2.8535
1.1189
0.0022 0.1316
0.6220 0.0001 0.2670
3.2808
1.5428
0.0005 0.0614
0.9867 0.0000 0.1619
3.6817
1.9397
0.0001 0.0262
1.3297 0.0000 0.0918
4.0594
2.3132
0.0000 0.0104
1.6536 0.0000 0.0491
4.4166
2.6660
0.0000 0.0038
1.9605 0.0000 0.0250
4.7556
3.0004
0.0000 0.0013
2.2522 0.0000 0.0122
5.0782
3.3183
0.0000 0.0005

Q-MH

The first row of coefficients provides test statistics and corresponding p-values for the
scenario of no hidden bias. $WȖ WKHLPSDFWRI0+&SDUWLFLSDWLRQRQUH-arrest is
marginally significant. When hidden bias decreases the odds of receiving MHC treatment
for treated cases compared to untreated cases by an additional 10% over and above the
HVWLPDWHGSURSHQVLW\VFRUHȖ 1), the estimated treatment effect is no longer statistical
significant for the re-arrest model (Q-MH=0.9534, p=0.1702). However, the negative bias
is less likely to occur and therefore less concerning than positive bias in our study. The
direction of hidden bias which raises serious concerns about the estimated treatment
effect is positive. In other words, the self-selection into MHC programs can lead to
positive outcomes even in the absence of MHC treatment.
The effect of positive bias is measured by the Q+MH test statistic. When hidden bias
LQFUHDVHVWKHRGGVRIUHFHLYLQJ0+&WUHDWPHQWIRUWUHDWHGFDVHVE\Ȗ  relative
to untreated cases, after accounting for the propensity score, the impact of MHC
participation on re-arrest and re-conviction is highly significant. It is fairly conceivable
that the use of limited administrative data in our study may not fully capture the extent
that treated cases self-selected into MHC treatment. However, the results from sensitivity
analysis suggest that our findings regarding the treatment effect of MHC programs would
be relatively tolerable to such bias.
It is also notable in the Bronx evaluation that the re-conviction model is slightly more
sensitive to hidden bias than is the re-arrest model. To the extent that unobserved
characteristics of program participants self-selected them into treatment, the results
reported in this study for the re-conviction outcome are more vulnerable to bias than is
the re-arrest outcome.

BROOKLYN EVALUATION

The table below shows the sensitivity of the PSM analysis for Brooklyn MHC. Similar to
the Bronx evaluation, the results indicate that our impact analyses are more sensitive to
negative hidden bias (P-MH). Also, the impact of MHC participation remains highly
significant even when the presence of hidden bias is highly likely in our analyses. To
sum, these results indicate that our estimated effect of MHC programs is reasonably
tolerable to positive self-selection bias, which is of main concern in most evaluation
studies.
Table 2. Sensitivity of Treatment Effect Estimates to Hidden Bias (Brooklyn MHC)
Re-Arrest
Gamma Q MH
1.0
1.4586
1.1
1.9503
1.2
2.3989
1.3
2.8129
1.4
3.1974
1.5
3.5567
1.6
3.8939
1.7
4.2119
1.8
4.5128
1.9
4.7985
2.0
5.0706
+

Re-Conviction
Q-MH
1.4586
0.9711
0.5252
0.1153
0.0686
0.4216
0.7519
1.0625
1.3555
1.6330
1.8967

+

P

MH

0.0723
0.0256
0.0082
0.0025
0.0007
0.0002
0.0000
0.0000
0.0000
0.0000
0.0000

P-MH
0.0723
0.1658
0.2997
0.4541
0.4726
0.3367
0.2260
0.1440
0.0876
0.0512
0.0289

Q+MH
Q-MH
P+MH
P-MH
1.9580
1.9580
0.0251
0.0251
2.4713
1.4501
0.0067
0.0735
2.9392
0.9852
0.0016
0.1623
3.3708
0.5579
0.0004
0.2885
3.7717
0.1625
0.0001
0.4354
4.1460
0.0179
0.0000
0.4929
4.4973
0.3618
0.0000
0.3587
4.8284
0.6851
0.0000
0.2467
5.1415
0.9899
0.0000
0.1611
5.4387
1.2785
0.0000
0.1005
5.7216
1.5525
0.0000
0.0603