Twenty-one studies (25 articles, n=1,083, range three to 305) were included in the review: nine RCTs (two unpublished) (n=383); four controlled studies (n=493); five case series (one unpublished) (n=129); two qualitative studies (n=75); and a case report (n=3). The RCTs were of moderate quality (mean MSQ score 11.3, range 8 to 14). Three of the published RCTs used blinded assessment and six had no losses to follow-up or used intention-to-treat analysis. The quality of the non-randomised studies was limited (MSQ range 4 to 8).
Randomised evidence (nine RCTs):
Six of the published RCTs reported that mindfulness meditation was associated with statistically significant (p<0.05) improvement in both substance use and other outcomes (compared to controls and/or baseline values); analysis was by intention-to-treat (two RCTs) or per protocol (four RCTs). The seventh published RCT reported statistically significant improvement (p<0.05) in the mindfulness meditation group compared to controls for medical symptom severity only.
Results tended to favour mindfulness meditation more markedly when controls received standard care or a non-matching active intervention: per protocol effect estimates from the three relevant RCTs were absolute risk reductions of 20% and 30% and an effect size of 1.0 (a moderate to large effect).
Comparisons of mindfulness meditation with behavioural interventions (two RCTs) yielded smaller non-significant absolute risk reductions of 5% (per protocol) and 12% (intention-to-treat). Both the unpublished RCTs reported statistically significant benefit in the mindfulness meditation group compared to controls for substance-use, with small (0.3) and moderate (0.6) effect sizes (both per protocol).
Other evidence:
The results of the non-randomised studies were generally positive for mindfulness meditation. These and other results (which included numbers needed to treat) were reported in the review. There were few data on adverse events.