BACKGROUND AND AIMS: Illicit substance use is common among people entering prisons, as is returning to substance use after release from prison. We aimed to assess the predictive validity of the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) for returning to substance use after release from prison.

DESIGN: A longitudinal design with baseline survey conducted between 2008-2010 in the 6 weeks before expected prison release, and up to three follow-up surveys in the 6 months after release. Setting is prisons in Queensland, Australia with 1054 adults within 6 weeks of expected release from prison.

MEASUREMENTS: The ASSIST was used to assess problematic use of cannabis, methamphetamine, heroin, and other non-prescribed opioids in the 3 months before incarceration. Post-incarceration substance use was measured at 1, 3, and 6 months after release. We calculated the area under the receiver operating characteristic curve (AUROC) and the optimal ASSIST cut-off score for each substance, using Youden's index (J).

FINDINGS: Forty-one percent (n=434) of the cohort reported any substance use during follow-up: 33% (n=344) used cannabis, 20% (n=209) methamphetamine, 10% (n=109) heroin, and 9% (n=97) illicit other opioids. The optimal ASSIST cut-off score was ≥4 for heroin, methamphetamine, and cannabis, and ≥1 for other opioids. Using these cut-offs, the AUROC was highest for heroin in predicting both any use (AUROC=0.82) and weekly use (AUROC=0.88) in the past 4 weeks. AUROCs for other drugs ranged from 0.73 to 0.79.

CONCLUSIONS: The Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) shows promise as an accurate and potentially scalable tool that may be useful for predicting a return to substance use after release from prison and could inform service delivery. The substantial rates of returning to substance use after release from prison suggest that prison serves to interrupt rather than cease substance use.