Application of classification and regression tree analysis to determine drug-abusing parolees who benefit from case management services

The Transitional Case Management (TCM) study, conducted under NIDA's Criminal Justice Drug Abuse Treatment Studies (CJ-DATS) cooperative, tested whether a strengths-based case management intervention would increase participation in community substance abuse treatment, enhance access to needed services, and improve drug use and crime outcomes. 812 inmates from four sites were randomly assigned to the TCM group or to a Standard Referral (SR) group. This analysis sought to determine which characteristics of parolees who have successful outcomes. Using classification and regression tree (C&RT) analysis with data from the baseline and three-month assessment, we report on parolee characteristics associated with two measures of success over the three months since release from prison: (1) admission to residential or outpatient treatment and (2) avoiding arrest. The covariates included in the C&RT analysis were age, race, gender, employment, education, marital status, social support, primary drug, drug severity level, treatment motivation, criminal thinking scales, Brief Symptom Inventory, and previous arrests. C&RT analysis was first applied to the TCM group, and then the TCM and SR group were compared on the variables found to be significant. For the TCM group, motivation (higher) and personal irresponsibility (lower) were more likely to enter treatment. Subgroups of the TCM and the SR groups defined by the cut points for the two variables did not differ significantly on treatment admission. TCM parolees with higher motivation were less likely to be arrested. The difference on arrests between the TCM and SR subgroups defined by the motivation cut point was marginally significant (p=.065).