People with serious mental illness (SMI) are frequently arrested for minor crimes relating to attempts to get needed shelter, food, or medical attention. Such people are more likely to be held without any charges, to be abused by other detainees, to refuse medication while in jail, and less likely to receive needed mental health services. Jail diversion programs are designed to remove SMI individuals from jail and place them in community mental health treatment as an alternative to incarceration. There are estimated to be only 52 such programs in the country, despite enthusiastic political support. One reason for this is that there is little data on the effectiveness of jail diversion programs. Although a randomized study of jail diversion would be the ideal method by which to examine the effectiveness of such programs, such a study is ethically impossible to conduct in an area where diversion services are already offered. This study is a retrospective study that follows a cohort of people with SMI who were arrested for a minor crime and documents the effects of diversion using a combination of administrative databases. There are four research hypotheses: 1) subjects who were diverted will have spent fewer days incarcerated in a one-year period after arrest than a group not enrolled in the program; 2) those diverted will have had higher utilization of inpatient and outpatient psychiatric and substance abuse services during the year after arrest than those not diverted; 3) those diverted will have spent more days in independent community living in the year after arrest; and 4) diverted clients will have incurred lower societal cost than those not diverted, as measured by total mental health, criminal justice, and other illness-related non-health cost, over the year after index arrest. Computerized medical records and administrative criminal justice data will be employed to assemble the study sample, identify diverted clients, and track days incarcerated, mental health service utilization, and costs over a one year period. Because diversion services are not randomly assigned, there are expected to be some biases across groups. Multivariate methods will be utilized to adjust for the confounding effects of diagnosis, comorbidity, criminal charges, and socio-demographic characteristics.
