METHODS: EBSCOhost, ProQuest and Digital Dissertation Consortium were searched using keywords. Ten randomized trials from the United Kingdom, United States and Taiwan, including 1466 participants, were identified and analysed using a random effects model. Primary outcome measures captured were: the extent of drug use, intention to use drugs and readiness to change. Each study received a high-quality score based on the Miller Quality Scoring Coding System. Moderator analyses were also conducted to examine the impacts of follow-up period, delivery setting and study design on the effectiveness of MI.
RESULTS: No statistically significant effect of MI on was found change of drug use behaviours [d = 0.05, 95% confidence interval (CI) = -0.06, 0.17, P = 0.36]. A significant effect was found on attitude change (d = 0.44, 95% CI = 0.20, 0.67, P = 0.0002). The funnel plot was asymmetrical, suggesting publication bias favouring small studies with higher effect sizes.
CONCLUSIONS: Motivational interviewing has not been found thus far to reduce adolescent use of illicit drugs. It may influence intentions to change, but evidence of publication bias weakens confidence in this conclusion.