The authors are affiliated with the University of Pennsylvania School of Social Policy and Practice, where Ms. Wilson is currently pursuing her Ph.D. degree in social welfare. Send correspondence to Ms. Wilson at the University of Pennsylvania School of Social Policy and Practice, 3701 Locust Walk, Philadelphia, Pennsylvania 19104 (e-mail, amyblank@comcast.net).

Each year approximately 600,000 people are released from state and federal prisons (1,2), and an additional seven million are released from local jails (3). Petersilia (1) estimated that one in six prison inmates has a mental illness, and Osher and colleagues (4) estimated that of the 11.4 million people who are committed to jail each year, 700,000 will exhibit active symptoms of mental illness.

These numbers have placed the formation of reentry services on the top of the national policy agenda. This has led to several legislative initiatives designed to fund new reentry services (5,6,7), one of which is designed specifically for people with mental illness (6). Yet scant empirical evidence is available to guide the development of reentry programs for people with mental illness. The research presented here was designed to begin to explore this area of service provision by conducting a survey that defined the various service strategies being used to assist people with mental illness who are reentering the community after a period of incarceration in jails or prisons.

Lamberti and colleagues (8) recently conducted a survey designed to assess the extent to which assertive community treatment programs are being used to prevent recidivism among people with mental illness who are involved in the criminal justice system. Because some of these assertive community treatment programs were also reentry programs, there was some overlap between their study and ours. However, Lamberti and colleagues assessed the use of a specific treatment modality, whereas our study was designed to identify and describe the range of strategies currently being used to assist people with mental illness who are reentering the community.

The central aim of our study was to use the data collected through the survey to develop a classification schema of service strategies being used to help offenders with mental illness stay in the community after release from incarceration. The research questions were: What service strategies are reentry programs using to help individuals with serious mental illness who are leaving jails and prisons stay in the community? And how might these service strategies differ on the basis of the extent of collaboration between the criminal justice and mental health systems?

The results are derived from a national survey of service strategies used to assist people with mental illness to manage their transition from incarceration to the community. Data were collected between August 2002 and January 2003. This survey had two iterative phases: the development of a sampling frame and compilation of service strategy descriptions. The procedures were reviewed and determined to be exempt by the University of Pennsylvania's institution's institutional review board.

To define the various service strategies in use, we developed a sampling frame of programs that worked with people with mental illness who were reentering the community after a period of incarceration. Reentry programs in the sampling frame met three criteria. First, the program targeted people with mental illness, defined as those having an axis I diagnosis. Second, the program provided mental health services that started while the individual was incarcerated and continued for a period of time after the individual was released into the community. Third, the program was actively providing services to clients during the data collection period. Programs designed to identify "dangerous offenders" for the purpose of further confinement were excluded from the sample.

Data collection was an iterative process of identifying and describing reentry programs that demonstrated strategies for community reintegration. Service strategies were defined as specific, programmatic mechanisms designed to maintain engagement in services and the larger community while reducing the risk of recidivism within the criminal justice system. Once programs were identified as meeting the criteria specified above, we collected data related to the structure and operation of each program. This information was used to develop descriptions that formed the basis of a typology of service strategies that were used to aid in the reentry process.

Because of the localized nature of many reentry programs, researchers began the program identification phase of the survey by geographical regions of the United States. We used conference presentations, Internet searches, published reports, key informants, and governmental agency contacts to develop an initial listing of reentry programs within each geographical region. We contacted key program staff of each identified program to determine whether the program met the criteria specified above. Snowball sampling techniques were used when talking with staff of the reentry programs, government officials, universities, and advocacy organizations to identify other reentry programs operating in their area. Program identification continued in each region until no new service strategies were able to be identified.

Researchers developed descriptions of each program initiative. The main source of information for the descriptions of the program initiatives was telephone interviews with key personnel. Descriptions also included information gathered from written materials, published articles, and conference presentations.

Defining our unit of analysis as "initiatives" rather than "programs" emerged from the recognition that some reentry programs had developed beyond the provision of a single service. These initiatives were offering an array of services that were designed to provide a continuum of interconnected services for clients reentering the community.

Data were analyzed by using a constant comparative method (9). Programmatic information was entered into a grid designed to outline programmatic elements. Initiatives were indexed by their name and geographical location. The grid included the following elements: the program's size, funding source, physical location, service structure, target population, staffing patterns, oversight agency, community integration strategies, innovative service elements, and linkages or collaborations with other service systems. Where data appeared to be sparse for some initiatives compared with others, follow-up calls were made to fill in the data so that, to the extent possible, the same data were available for each initiative. In some cases, the information that we sought was simply unavailable or undeveloped by the program initiative.

In the second stage of analysis, data in the program grid were used to develop categories of service strategies. These categories provided a comprehensive overview of the variety of service strategies in use at the time of the survey while also illustrating areas in which there was variation between the different strategies. All data in the grid were used to develop a typology of service initiatives based on organizational and policy factors. In creating iterations of the categories and the typology, we sought to come up with a way of organizing the data that could account for the variations between service strategies being used to serve this population.

We identified 58 potential program initiatives. The necessary programmatic details were available for 50 of these programs. Eight programs were dropped because we were unable to obtain the information necessary to determine whether the programs met the criteria for inclusion in the study. The typology of service strategies was developed from the 50 programs that were determined to meet the eligibility criteria for the study.

During analysis, two organizational characteristics emerged as a parsimonious and conceptually meaningful way to develop a 2×2 typology that differentiated the reentry initiatives. One factor was whether the initiative was led by the criminal justice or mental health system. The second factor was the extent to which the systems collaborated with each other to provide the services. The lead system was determined by the report of which system took leadership in initiating the program—that is, the system that found the funding, staffed the initiative, and oversaw its daily operations. The operationalization of this factor was determined from the data gathered for the grid.

The extent of collaboration was determined by the degree to which there was multisystem involvement in the initiative. That is, how was the program staffed? How was the program initiated? How was the program funded? Who had continuing oversight responsibility for the program? In operationalizing this factor, cross-system involvement in more than one area was classified as more collaboration as opposed to less collaboration.

As can be seen in Table 1, the 2×2 typology of service initiatives included four categories: behavioral health initiatives with less collaboration (eight initiatives, or 16 percent), behavioral health initiatives with more collaboration (five initiatives, or 10 percent), criminal justice initiatives with less collaboration (12 initiatives, or 24 percent), and criminal justice initiatives with more collaboration (25 initiatives, or 50 percent). A total of 13 initiatives (26 percent) were led by the behavioral health system, and 37 (74 percent) were led by the criminal justice system.

Table 1 contains programmatic details that describe the client population being served and some of the specific service strategies that were being used by the initiatives. These details include the number of programs that used assertive community treatment, which was found to be the dominant service strategy in this survey; the type of institution from which clients were being released (jail or prison); and several elements that demonstrate the degree of collaboration across systems in the staffing and funding of the initiatives.

The striking finding was that a substantial number of the initiatives surveyed were led by the criminal justice system (37 initiatives). Our data suggest that innovative leadership in this area is not shared but rather is dominated by the criminal justice system. Providing mental health services is not the main function of the criminal justice system; yet it is clear that this system is assuming more responsibility for the treatment of people with mental illness within their system.

Steadman (10) suggested that "boundary spanners" are an important component of mental health services that work with the criminal justice system. Table 1 indicates a phenomenon beyond mere spanning of the boundaries of these systems. Criminal justice personnel have clearly demonstrated leadership in the development and provision of reentry services for people with mental illness.

In addition, initiatives led by the criminal justice system tend to be more collaborative across systems than initiatives led by the behavioral health system. This may be indicative of greater resources being available for criminal justice initiatives and, thus, greater incentive for collaboration. As more funds become available for reentry services, it will be important to explore further which systems take the lead in accessing funds and the resulting extent of collaboration across systems.

The leadership that the criminal justice system has shown in gathering funds has been instrumental in increasing resources for mental health services. In many cases, these resources allow for the implementation of evidence-based practices, such as assertive community treatment. However, when the funding from grants is used up, some of the programs are discontinued whereas others are folded into other funding streams. Although the infusion of resources is helpful, it remains to be seen whether current funding mechanisms are able to build the infrastructure for a robust intersystem response to the issue of reentry for people with mental illness. Further research needs to be done to determine whether the service and funding patterns identified in this survey are sustained over time.

One of the unexplored, yet important components of this emphasis on increased collaboration between the mental health and criminal justice systems is the impact that these changes could have on the nature and context of behavioral health services. Future research needs to assess whether changes in the location, professional composition, and focus of behavioral health services fundamentally alter the services' therapeutic and rehabilitative qualities. This is especially important to consider when actively including probation and parole officers or sheriffs as members of the treatment team.

This study has identified several ways in which the location and composition of behavioral health services for offenders with mental illness have been altered by the increased collaboration between the two systems. Both this survey and the one by Lamberti and colleagues (8) found that in many cases probation and parole officers are included as active members of assertive community treatment teams. However, our study identified one instance of a more fundamental shift, in that we found at least one program that identified itself as an "assertive community treatment-like" treatment team but was run without any connection to the mental health system, meaning it was funded, staffed, and supervised through the criminal justice system. Changes such as these are likely to alter the context of rehabilitation and recovery that undergirds the evidence base for these interventions. The task of mental health services researchers is to pick up this ball, recognize the shift, and conceptualize new research on these de facto policy changes. The basic question might be if criminal justice agencies faithfully implement evidence-based mental health practices, to what extent does the justice system context change the nature of these interventions?

Lamberti and colleagues (8) asserted that the tensions inherent in combining criminal justice personnel into mental health programs can be managed effectively by using the client-centered approach to guide decisions about which approaches to use when. Unfortunately, no empirical evidence supports this approach's ability to rule out negative effects. A call for client centeredness cannot address how including law enforcement personnel as active members of therapeutic treatment teams can have an impact on the treatment team's ability to establish therapeutic rapport with individual clients. In many cases, decisions are being made within the context of the criminal justice system that are related to how behavioral health services provided within a reentry context balance their client-centered and legal-leverage approaches (8). The criminal justice system frames services from the perspective of public safety, which prioritizes issues of risk management and control (1). Within this system, behavioral health interventions are tools to manage and control offenders rather than a mechanism to promote the individual's rehabilitation or recovery from mental illness (1,11). A client-centered approach is rarely recognized in these settings, let alone prioritized as an approach to service delivery.

We found a paucity of strategies focused on mechanisms of change to reduce recidivism, improve psychiatric status, or enhance quality of life. Without innovation in therapeutic approaches along with research grounded by the same conceptual framework, organizational collaboration between these systems could likely default to the control functions of a dominant criminal justice system.

The current tone and direction of response to mental illness in the criminal justice system are governed by the criminal justice goal of public safety (1,2). Thus the objectives that garner funds for reentry services are the objectives that will shape services for this population. Just as recovery and community integration are gaining legitimacy in shaping behavioral health services for many with mental illness (12), we may also be taking steps to further isolate persons who are not fully engaged in effective mental health services in the workings of the criminal justice system. Mental health services researchers and criminal justice researchers need to draw on each other's work to understand these dynamics and to be open to how this growing area of service delivery can benefit or harm a vulnerable population.

This research was funded by grant P20-MH-068170 from the National Institute of Mental Health to the Center of Mental Health Services and Criminal Justice Research based at Rutgers, the State University of New Jersey.