WHAT PREDICTS RELAPSE - PROSPECTIVE TESTING OF ANTECEDENT MODELS

Predictors of relapse to drinking were examined in a clinical sample o f 122 individuals seeking outpatient treatment for alcohol problems. D rinking status and a variety of predictor variables were measured ever y two months for one year following presentation for treatment. In add ition to pretreatment characteristics, potential antecedents of relaps e were assessed at each point within five domains: (1) the occurrence of negative life events; (2) cognitive appraisal variables including s elf-efficacy, alcohol expectancies, and motivation for change; (3) cli ent 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 v ariance in drinking outcomes. Proximal antecedents (from the prior a-m onth interval) significantly and substantially improved predictive pow er over that achieved from pretreatment characteristics alone. When an alyzed jointly, these predictors accounted for a majority of variance in 6-month relapse status. A prospective test supported Marlatt's deve lopmental model of relapse, pointing to two client factors as optimall y predictive a resumed drinking: lack of coping skills and belief in t he disease model of alcoholism.