Predictors of receipt of aftercare and recidivism among persons with severe mental illness: a review.

OBJECTIVE: This paper provides a comprehensive review of research predicting receipt of aftercare and recidivism (rehospitalization) among individuals with severe and persistent mental illness. METHODS: English-language studies were identified by searching PsycLIT from January 1974, the year of a previous comprehensive review of the literature on recidivism, to March 1994. A heuristic model was used to classify predictor variables into three categories: client vulnerability, community support, and system responsiveness. RESULTS: In general, variables related to system responsiveness were more consistent predictors of receipt of aftercare than variables related to either client vulnerability or community support. Community support variables were more consistent predictors of recidivism than variables in any of the other categories. Assertive community treatment and receipt of aftercare-both system responsiveness variables-were also associated with lower rates of rehospitalization. CONCLUSIONS: Conclusions based on past research should be regarded as tentative, because of several methodological and theoretical weaknesses. Multivariate statistics have rarely been used, many of the studies are based on nonrepresentative samples, and interactions between client-level variables and system responsiveness variables have rarely been studied. Much of the research has relied on archival data, and choice of variables has thus often been guided more by the data available than by theory.