Does Alcoholics Anonymous work? The results from a meta-analysis of controlled experiments

This article reviews the outcome (usually abstinence at 12 months) of 21 controlled studies of AA, with emphasis on methodological quality. Severe selection biases compromised all quasi-experiments. Randomized studies yielded worse results for AA than nonrandomized studies, but were biased by selection of coerced subjects. Attending conventional AA meetings was worse than no treatment or alternative treatment; residential AA-modeled treatments performed no better or worse than alternatives; and several components of AA seemed supported (recovering alcoholics as therapists, peer-led self-help therapy groups, teaching the Twelve-Step process, and doing an honest inventory).