Predictive value of non-consumption outcome measures in alcohol use disorder treatment.

Abstract

For decades, researchers have called for the systematic inclusion of non-consumption outcomes, such as alcohol-related consequences and temptation/craving, in addition to consumption outcomes (e.g. percentage of days abstinent) to evaluate alcohol use disorder (AUD) treatment efficacy. However, Food and Drug Administration (FDA) guidelines for alcohol medications development suggest that non-consumption outcomes may be insensitive to changes that occur within AUD treatment trial assessment windows, although this has never been directly tested. We aimed to measure the predictive value of diagnostically related, non-consumption measures of AUD treatment effects. Pre- to post-treatment effect sizes were examined. Logistic regression analyses were used to test the predictive value of non-consumption outcome measures that were administered in the COMBINE (Combined Pharmacotherapies and Behavioral Interventions for Alcohol Dependence) Study and Project MATCH (Matching Alcoholism Treatment to Client Heterogeneity) in detecting 12-month outcomes following treatment. Pre- to post-treatment effect sizes suggested that non-consumption measures changed during the course of treatment; effect sizes for temptation/craving, self-efficacy, consequences and consumption variables were mostly medium to large (medium effect sizes were 0.2 < dz < 0.8, large effect sizes were dz > 0.8). The regression results indicated temptation/craving and self-efficacy predicted 12-month drinking outcomes in COMBINE and MATCH, and drinking consequences predicted 12-month drinking outcomes in COMBINE. Self-report measures of temptation and alcohol craving can change as a result of alcohol dependence treatment and can predict drinking outcomes.