Baseline Abstinence May Moderate Substance Use Disorder Intrvention Effects

Abstract

In many evaluations of care for patients with substance use disorders (SUD), a substantial proportion of patients are abstinent from substances prior to their baseline assessment. For researchers evaluating the effect of treatments or patterns of care on follow-up abstinence, especially with non-randomized designs, this raises important conceptual and statistical questions. We discuss several ways researchers might address baseline heterogeneity in abstinence either through design or statistical approaches. For illustration, we estimate the effect of continuing care duration on follow-up abstinence in a sample of 3,181 SUD outpatients treated in 54 Veterans Health Affairs (VHA) facilities. We show that failure to separately examine individuals who are and are not abstinent at baseline overestimates the “treatment effect” on substance use outcomes for those abstinent at baseline and underestimates the treatment effect for those not abstinent at baseline. We also describe comparable findings for the effect of attendance in 12-step self-help groups.