Amanda E Perry, Matthew Neilson, Marrissa Martyn‐St James, Julie M Glanville, Rebecca Woodhouse, Christine Godfrey, Catherine Hewitt

The review represents one in a family of four reviews focusing on a range of different interventions for drug-using offenders. This specific review considers pharmacological interventions aimed at reducing drug use or criminal activity, or both, for illicit drug-using offenders. To assess the effectiveness of pharmacological interventions for drug-using offenders in reducing criminal activity or drug use, or both. We searched Fourteen electronic bibliographic databases up to May 2014 and five additional Web resources (between 2004 and November 2011). We contacted experts in the field for further information. We included randomised controlled trials assessing the efficacy of any pharmacological intervention a component of which is designed to reduce, eliminate or prevent relapse of drug use or criminal activity, or both, in drug-using offenders. We also report data on the cost and cost-effectiveness of interventions. We used standard methodological procedures as expected by Cochrane. Fourteen trials with 2647 participants met the inclusion criteria. The interventions included in this review report on agonistic pharmacological interventions (buprenorphine, methadone and naltrexone) compared to no intervention, other non-pharmacological treatments (e.g. counselling) and other pharmacological drugs. The methodological trial quality was poorly described, and most studies were rated as 'unclear' by the reviewers. The biggest threats to risk of bias were generated through blinding (performance and detection bias) and incomplete outcome data (attrition bias). Studies could not be combined all together because the comparisons were too different. Only subgroup analysis for type of pharmacological treatment were done. When compared to non-pharmacological, we found low quality evidence that agonist treatments are not effective in reducing drug use or criminal activity, objective results (biological) (two studies, 237 participants (RR 0.72 (95% CI 0.51 to 1.00); subjective (self-report), (three studies, 317 participants (RR 0.61 95% CI 0.31 to 1.18); self-report drug use (three studies, 510 participants (SMD: -0.62 (95% CI -0.85 to -0.39). We found low quality of evidence that antagonist treatment was not effective in reducing drug use (one study, 63 participants (RR 0.69, 95% CI 0.28 to 1.70) but we found moderate quality of evidence that they significantly reduced criminal activity (two studies, 114 participants, (RR 0.40, 95% CI 0.21 to 0.74).Findings on the effects of individual pharmacological interventions on drug use and criminal activity showed mixed results. In the comparison of methadone to buprenorphine, diamorphine and naltrexone, no significant differences were displayed for either treatment for self report dichotomous drug use (two studies, 370 participants (RR 1.04, 95% CI 0.69 to 1.55), continuous measures of drug use (one study, 81 participants, (mean difference (MD) 0.70, 95% CI -5.33 to 6.73); or criminal activity (one study, 116 participants, (RR 1.25, 95% CI 0.83 to 1.88) between methadone and buprenorphine. Similar results were found for comparisons with diamorphine with no significant differences between the drugs for self report dichotomous drug use for arrest (one study, 825 participants, (RR 1.25, 95% CI 1.03 to 1.51) or naltrexone for dichotomous measures of reincarceration (one study, 44 participants, (RR 1.10, 95% CI 0.37 to 3.26), and continuous outcome measure of crime, (MD -0.50, 95% CI -8.04 to 7.04) or self report drug use (MD 4.60, 95% CI -3.54 to 12.74). When compared to non-pharmacological treatment, agonist treatments did not seem effective in reducing drug use or criminal activity. Antagonist treatments were not effective in reducing drug use but significantly reduced criminal activity. When comparing the drugs to one another we found no significant differences between the drug comparisons (methadone versus buprenorphine, diamorphine and naltrexone) on any of the outcome measures. Caution should be taken when interpreting these findings, as the conclusions are based on a small number of trials, and generalisation of these study findings should be limited mainly to male adult offenders. Additionally, many studies were rated at high risk of bias.