As might be expected, the data showed that clients who graduated from both the prison-based Key therapeutic community and the community-based Crest therapeutic community had the most positive outcomes in terms of drug-free and arrest-free status in a 6-month follow-up (Inciardi, 1996; Martin et al., 1999).
The second most successful group included those with no prison treatment who underwent work release in the Crest therapeutic community after-care unit only. They were much more likely to be drug free and arrest free than those inmates who participated in the prison treatment program but underwent work release in the community in a unit without a therapeutic community. (The comparison group was the least likely to be arrest free and drug free of the four groups.)
These rankings held in the preliminary 6 month follow-up. In a second follow-up after 18 months, the drug-free and arrest-free scores for all four groups were much lower; however, the rankings remained the same (Landry, 1997; Martin et al., 1999).
The fact that those with community-based after care only outper-formed those with prison treatment only raises questions about how much influence prison-based therapeutic communities have on positive outcomes. On the basis of these findings, one could argue that the community-based after-care program was more instrumental in reducing relapse and recidivism than the prison-based program.
Indeed, after examining the follow-up data, the Key-Crest researchers concluded that participation in a prison treatment program alone was not effective in bringing about drug-free or arrest-free status after release. In their view, successful outcomes depended on participation in the community-based after-care program (Martin et al., 1999).
This finding that prison treatment alone was considerably less important than community-based after-care was borne out by research conducted on another prison treatment program called Amity (Wexler et al., 1999). Much like Key-Crest, the Amity program combined a prison-based therapeutic community with a community-based after-care program. Hence it was possible for Amity participants to begin their recovery in the prison-based therapeutic community and continue it in a community-based therapeutic community upon release.
In the Amity program, volunteers for prison treatment were randomly assigned from a waiting list to either treatment or a no-treatment control group (Wexler et al., 1999). In this sense, the Amity program was different from the Key-Crest program as study subjects were not randomly assigned to prison treatment in the Key-Crest study (Wexler et al., 1999). Hence the Amity program controls for selection bias for those who participate in their prison program, whereas the Key-Crest program does not (see Table E.1 for study designs for both Amity and Key-Crest).