Describing and Evaluating Jail Diversion Services for Persons With Serious Mental Illness

Jail diversion services consist of two broad interlocking areas of intervention. First is the diversion mechanism, or the means by which an individual is identified at some point in the arrest process and diverted into mental health services. Second is the system of integrated mental health and substance abuse services to which the client is diverted.

Broadly defined, diversion leads individuals with mental illness or substance use problems away from criminal incarceration. Diversion services may either prevent incarceration or cut it short. Conceptually, then, the definition of diversion could include many crisis services that are used to intervene after the onset of acute symptoms but before an individual has engaged in any criminal behavior, thus removing a basis for arrest. Such a broad definition would make it very difficult to differentiate crisis services from jail diversion because one could never be sure that an arrest would otherwise have occurred. Diversion could also be any planning for release from jail, because a plan for community services after release often facilitates a faster release, thus preventing extended incarceration.

What makes jail diversion unique is that the service positions itself within the criminal justice system as an immediate alternative to incarceration. A jail diversion program provides opportunities for police officers to easily redirect an individual in custody into treatment as opposed to jail. It offers jail wardens the opportunity and means to securely remove individuals from the stress of the jail environment and into a treatment environment. It provides a probation officer with an alternative to a violation hearing for a troubled client. Jail diversion provides immediate access to treatment resources by law enforcement officers as an alternative to incarceration.

At the Montgomery County Emergency Service, jail diversion is accomplished through one of several mechanisms. These mechanisms are described by their position in the arrest and detention process, postbooking and prebooking. For purposes of this discussion, booking is defined as the processing of an individual into jail custody. In some circumstances, booking is a separate process from arrest.

In Montgomery County, postbooking jail diversion occurs through regular and direct communication with county jail personnel. Individuals in the jail who may have mental health and substance abuse problems are identified by regular screenings or trained correctional officers. Screenings are conducted by the emergency service's clinicians on a weekly schedule. Correctional officers are trained to recognize signs of mental illness and refer individuals for screening. Some clients are known to the emergency service as a result of previous treatment.

For jail diversion clients identified at the jail, the service offers several alternatives. A client may be transferred to the emergency service for inpatient treatment in a secure facility. The service may negotiate a conditional release for the client with the promise that he or she will receive mental health services. Or, charges may be dropped once an offender is identified as a client of the emergency service who may benefit more from mental health treatment than from new criminal entanglement.

This array of postbooking diversion mechanisms is designed to improve access to treatment resources from the criminal justice system. First, correctional officers are trained to understand mental illness and recognize its signs in an individual. Second, emergency service staff members assist courts in structuring release plans to enable conditional release from incarceration. Third, the service provides clinical assessment services in addition to the regular health services provided at the jail.

Prebooking jail diversion occurs when access to psychiatric treatment has been provided in lieu of arrest or criminal incarceration. Identifying this diversion as prebooking presumes that an arrest was the alternative outcome of the client's encounter with law enforcement. However, not all encounters with law enforcement that result in psychiatric treatment can be construed as jail diversion. Many encounters do not result in arrest, even when illegal behavior is observed or suspected. The decision to arrest is a complicated one made by an officer at the time of an encounter (4,5,6). For purposes of evaluating jail diversion, only persons who receive psychiatric treatment as an immediate alternative to criminal incarceration should be considered diversion clients.

In addition to pre- and postbooking jail diversion, the Montgomery County Emergency Service provides coterminous jail diversion. Coterminous diversion occurs when an offender is taken into custody by the police and delivered directly into psychiatric treatment while charges are being filed. In such cases, even though an offender has been arrested and a new charge has been filed, the offender has been diverted from criminal incarceration. This type of diversion could also be considered a prebooking diversion because the individual is not processed into jail. It differs from conventional ideas of jail diversion in that the officer still has the discretion to file charges, and the client is delivered by the police officer to a secure facility. As with postbooking jail diversion, coterminous jail diversion can be complicated by existing probation and parole sentences and the possibility of a new charge for violating the terms of probation or parole.

The Montgomery County Emergency Service supports all three forms of diversion through several services and programs. The service provides police training and support to members of the 49 police forces in Montgomery County. The training focuses on developing an understanding of mental illness and substance abuse and their course and treatment and how law enforcement can respond. Emergency service staff members are available for consultation and continuing education about mental illness and substance abuse. Consequently, trained police officers can more easily differentiate behaviors arising from psychiatric and substance abuse problems from criminally motivated behaviors.

The emergency service's crisis response team is available 24 hours a day to respond to crisis situations in the community that may involve mental illness and substance abuse. The team is supported by ambulances to provide transportation directly to inpatient treatment at the Montgomery County Emergency Service if appropriate. During treatment, emergency service staff may have an opportunity to negotiate a reduction in charges or a release conditioned on the provision of needed inpatient and outpatient treatment.

Staff case managers broker services for clients and link them to services; they also follow clients into the community after release. They carry cases for varying periods of time, depending on the support that community agencies may need in helping these individuals make the transition into community mental health services. In some cases, the Montgomery County Emergency Service provides long-term forensic intensive case management with clinical support from its outpatient section.

Because of the nature of jail diversion services, several research problems arise in evaluating their effectiveness. To review these problems, the ideal of a randomized field trial is discussed, as are the problems of quasiexperimental designs. All research strategies need to address problems of operationalizing jail diversion and identifying the most appropriate sample for study.

In planning a randomized field trial of jail diversion services, an initial decision relates to the point at which randomization occurs in the arrest process. In the field, police officers could randomly assign offenders to jail or diversion programs. In this case, officers would have to be trained to recognize research inclusion and exclusion criteria and apply them reliably. Randomization procedures under similar circumstances have been easily subverted (7,8). For example, officers may omit the research procedures if they are convinced that the person should be arrested and they decide not to risk randomization to diversion. As a result, randomization is applied only to individuals who are less likely to have been arrested.

If randomization is conducted later, say at the point of booking into the jail, trained research workers would have greater opportunity to apply criteria. This procedure would also provide certainty that the individual is among the population likely to be arrested in similar circumstances. Unfortunately, it would not work as a test of prebooking diversion services that enable the offender to be sent directly to a treatment facility.

Once the point of randomization is settled, the question of measurement points becomes salient. The diversion program consists of the diversion mechanism and follow-up treatment services, such as short-term forensic case management at the Montgomery County Emergency Service. Persons in jail may remain in jail for months. Those with mental illness can be expected to remain in jail for longer periods after arrest than other jail detainees. One baseline point could be the initial contact with the criminal justice system. Using this option, time spent in jail becomes an outcome domain, with the expectation that clients who are diverted will spend significantly less time in jail than non-diverted clients.

Another baseline point is the time of release to the community. Released clients who were not diverted from jail are comparable to the diverted clients in terms of the service environment; that is, both groups would be assessed when out of jail and eligible for the same community services. However, because the former group would have spent time in jail, the groups would not be comparable in terms of time since arrest. In one case, diversion is evaluated against jail time. In the other case, release to the community with diversion services is compared with release to the community without diversion services.

If data collection points are structured so that study participants diverted from jail are to be compared with participants still in jail at the same time point, researchers will need to account for the treatment and services the participants may receive in jail. In the ideal situation, high-quality mental health services would be available in jail, and results favoring jail diversion could not be attributed to differing quality of treatment in the respective service environments. Diversion would remain the salient difference between conditions. If high-quality mental health services do not exist in jail, then researchers may have problems differentiating whether jail diversion is more effective than providing more intensive mental health services in jail.

An alternative is to target the measurement clock for both conditions at the point of re-entry to the community from incarceration. In this way, all participants are assessed in terms of their reintegration to the community.

Conceptually, these alternative research designs place value on differing sets of outcomes. Outcome measures to assess recovery from an acute episode, such as more elaborate measures of symptoms and functioning, would be a priority if the focus is on the immediate response to the arrest. However, if the focus is on adjustment and stabilization in community living, such measures as quality of life, community tenure, and quality of and satisfaction with services become the salient outcome measures.

Jail diversion services are premised on the assumption that a number of persons with mental illness or substance abuse problems get arrested for these problems. Arrest is likely to occur during an acute phase of an episode of illness. Diverted clients re-enter the community treatment system quickly. Nondiverted clients must wait for a later resolution of their legal problems, which may include stabilizing their symptoms in jail before release. Therefore, diverted clients are almost by definition released at a different phase of their acute episode than nondiverted clients. The two groups may have varying experiences with community reintegration because of differences in the acuteness of their illness and resulting differences in treatment, support services, and engagement strategies.

The problem in operationalizing the concept of jail diversion is one of assessing the extent to which arrest is the most likely alternative. One cannot assume, for example, that all police encounters with mentally ill people that result in hospitalization would have otherwise resulted in arrest, or that all illegal acts committed by mentally ill people would always result in arrest. The decision to arrest is a complicated one involving a great deal of discretion by a police officer (4,5,6). When diversion occurs at the point of arrest, researchers and evaluators need to establish that the individual who received crisis mental health services would have probably been arrested.

The likelihood of arrest can be established with police cooperation. For example, a simple check-off form could be added to admission documentation completed by a police officer at a diversion program. The form would ask "Would you have likely arrested this person if you did not have the alternative of bringing the person to this program?" The officer would check yes or no. Because the decision depends to a large extent on the officer's discretion, few options exist other than asking the officer. The alternative is to include in the study sample many crisis clients who would not have been arrested, even without the diversion program. However, such a sampling strategy would be flawed in addressing the effectiveness of jail diversion because it would include as subjects a group of clients who would not have been arrested.

One solution to the problem of distinguishing between a diverted and a nondiverted population is to use a quasiexperimental design comparing a system that provides diversion with a system that does not. We are currently in the process of such a study, comparing the Montgomery County Emergency Service with the Bucks County Correctional Mental Health Service. Such a comparison could be complicated by differences in the quality of the diversion services and the mental health services provided in jail and by differences in populations served. However, the populations and service systems of the two counties are very similar, and the Bucks County jail has a well-developed, well-regarded mental health service (3). Therefore, the variable of jail diversion has strong explanatory power for any differences found in service patterns or client outcomes. A weakness of this strategy is that it cannot be known for certain which of the Bucks County clients would have been diverted if they were in the Montgomery County system.

Another quasiexperimental strategy is to compare diverted individuals with nondiverted individuals in the same system. For example, individuals diverted from jail would be compared with individuals with mental illness or co-occurring mental illness and substance abuse in the same jail who were not diverted. This strategy is severely weakened by a high selectivity factor. Individuals who are selected for diversion from jail can be reasonably expected to differ significantly from nondiverted clients in important characteristics. If a difference in outcome in favor of diversion from jail was found, it would be very difficult to attribute this outcome to diversion.

Some researchers promote the use of propensity scores to simulate random assignment in naturalistic settings. This strategy uses a statistical model to explain differences between naturally occurring groups; a variable that is the propensity that an individual would be in one group rather than another is constructed. However, the validity of any statistical model is contingent on its theoretical strength as well as its statistical explanatory power. Models based solely on application of stepwise statistical algorithms with a large number of potential explanatory variables are too reliant on chance to be meaningful in explaining away differences between existing groups.

Another design choice is to compare the current arrest episode of clients to their own past arrest and service histories. This approach would approximate a one-group pre-post design. It assumes that the chance for recidivism and the client's receptivity to services remain constant over time, and any that differences could thus be explained by jail diversion. Research has shown that these assumptions are not supported. People with mental illness are known to have fewer acute episodes, spaced further apart, as they age. Similarly, individuals with criminal histories are known to mature out of antisocial behavior over time. Clients may also become more receptive to traditional mental health services as they age (9). Therefore, in a pre-post study of jail diversion, differences in jail recidivism, hospitalization, and services received could be explained by maturation.

Jail diversion as a policy option is consistent with the principles of community support programs. Jail diversion gives priority to treatment in a less restrictive environment. Comprehensive diversion programs are planned around service integration and continuity of care.

The slim empirical basis for developing effective jail diversion programs severely impairs effective program development in this area. The development of jail diversion programs is important because of the large number of mental health service consumers who are also involved in the criminal justice system, many of whom have co-occurring substance use disorders. A commitment among services researchers and policy makers to empirical research on all services, both in jail and out of jail, that address service options for this population can help provide the information needed to plan such programs.

The authors are grateful to their collaborators in a site demonstration of the criminal justice diversion initiative of the Substance Abuse and Mental Health Services Administration (grant U1G SM52139): Rocio Nell, M.D., William Leopold, M.S.S.A., M.B.A., and Donald Kline, B.A., of the Montgomery County Emergency Service, and Paul Woodburn, M.S., of the Bucks County Correctional Mental Health Service-Lenape Valley Foundation. They thank Robert Boruch, Ph.D., for comments and consultation.

Dr. Draine is research assistant professor and Dr. Solomon is professor in the department of psychiatry and the School of Social Work at the University of Pennsylvania, 3600 Market Street, Seventh Floor, Philadelphia, Pennsylvania 19104-2648 (e-mail, [email protected]).