What predicts relapse? Prospective testing of antecedent models

Abstract

Predictors of relapse to drinking were examined in a clinical sample of 122 individuals seeking outpatient treatment for alcohol problems. Drinking status and a variety of predictor variables were measured every two months for one year following presentation for treatment. In addition to pretreatment characteristics, potential antecedents of relapse were assessed at each point within five domains: (1) the occurrence of negative life events; (2) cognitive appraisal variables including self-efficacy, alcohol expectancies, and motivation for change; (3) client coping resources; (4) craving experiences; and (5) affective/mood status. Although the occurrence of adverse life events did not predict 6-month relapse, all other domains singly accounted for significant variance in drinking outcomes. Proximal antecedents (from the prior 2-month interval) significantly and substantially improved predictive power over that achieved from pretreatment characteristics alone. When analyzed jointly, these predictors accounted for a majority of variance in 6-month relapse status. A prospective test supported Marlatt's developmental model of relapse, pointing to two client factors as optimally predictive of resumed drinking: lack of coping skills and belief in the disease model of alcoholism.