psychiatric, medical, social and legal) and race. Additionally, substance use treatment utilization prior to or at baseline was included in the model assessing substance use treatment utilization. Similarly, AA utilization prior to baseline was included in the model assessing AA utilization. The time-varying exposures of interest were either formal substance use treatment or whether the respondent attended AA. We additionally adjusted for time-varying ASI measures (drug, psychiatric, medical, social and legal). ASI alcohol severity over-time was not included in the models as it is highly correlated with 30-day alcohol abstinence. An exchangeable correlation structure and robust estimates of the variance were used. Stata v. 10 (Stata Corp., 2009) was used for all analyses.
