Changes in alcohol‐related social network composition mediate the effects of AA meeting attendance on drinking following a recovery attempt in adults with alcohol use disorder

AIMS: To assess whether changes in social network drinking is a mechanism of behaviour change responsible for the benefits of attending Alcoholics Anonymous (AA) during a recovery attempt. DESIGN: An observational longitudinal cohort study investigating mechanisms of behaviour change among adults with alcohol use disorder (AUD) reporting initiation of a substantial recovery attempt. SETTING: Boston, Massachusetts, United States, and Hamilton, Ontario, Canada. PARTICIPANTS: From a larger observational cohort, participants were individuals who reported a substantive increase in AA attendance (increase of ≥1 + meetings/week) from baseline to 6 weeks (n = 71) and a propensity score-matched control group of participants who did not increase AA attendance (n = 71). Propensity score matching used demographics and baseline drinking. MEASUREMENTS: Baseline assessment and a 6-week follow-up assessment, including outcome variables: diagnostic assessment of AUD, timeline followback interview [percent drinking days (%DD) and percent heavy drinking days (%HDD)]; and exposure variables: formal egocentric social network assessment collecting egocentric social network metrics and using up to 20 network alters. FINDINGS: Compared with the control group, participants who increased AA participation statistically significantly reduced %DD [mean (M) = 5.67% (standard error of the mean, SEM = 1.81) vs 30.91% (3.59); F = 46.22, P < 0.001] and %HDD [5.21% (1.79) vs 23.32% (3.14); F = 28.34, P < 0.001] and exhibited statistically significantly improved social network indicators including reduced network drinking frequency [1.99 (0.08) vs 2.96 (0.09); F = 42.26, P < 0.001] and severity [1.68 (0.06) vs 2.34 (0.08); F = 40.51, P < 0.001]. Changes in social network drinking statistically significantly mediated the relationship between changes in AA attendance and reductions in %DD [b = -0.06 (0.02), P = 0.01] and %HDD [b = -0.05 (0.02), P = 0.04] at follow-up. CONCLUSIONS: Reduced social network drinking appears to be one mechanism of behaviour change associated with the positive effects of Alcoholics Anonymous on drinking behavior during recovery.