Further validation of the alcohol dependence scale as an index of severity.

OBJECTIVE The Alcohol Dependence Scale (ADS) yields continuous scores purported to reflect the severity of the dependence syndrome. We evaluated the concurrent validity of the ADS as a general measure of severity and the screening accuracy of the total score and subscales to detect Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), physiological dependence. METHOD Treatment-seeking, alcohol-dependent individuals entering the Combining Pharmacotherapies and Behavioral Interventions for Alcohol Dependence (COMBINE) Study (N = 1,372; 69% men) completed the ADS, diagnostic interviews, and other measures before randomization. Analyses of variance tested differences between ADS quartiles on alcohol- related measures. Areas under the receiver operating characteristic (AUROC) curve assessed screening accuracy for DSM-IV physiological dependence (tolerance or withdrawal) or withdrawal alone. RESULTS ADS quartiles differed on age, other demographics, and prior treatment episodes. Linear correlation showed moderate to large magnitude associations with alcohol-related self-report measures and most indices of consumption. ADS quartiles differed significantly on proportion with DSM-IV physiological dependence, but AUROC curves indicated that the ADS had limited accuracy as a continuous measure to detect DSMIV physiological dependence (AUROC = .75 [95% confidence interval {CI}: .70-.79] and .67 [95% CI: .60-.74] for men and women, respectively; p = .08) or withdrawal alone (AUROC = .77 [95% CI: .74-.80] and .74 [95% CI: .69-.79] for men and women, respectively; p = .30). Screening accuracy was comparably limited based on ADS subscales reflecting psychoperceptual or psychophysical withdrawal. CONCLUSIONS The ADS reflected variation in symptom severity but did not adequately identify physiological dependence or withdrawal in treatment-seeking individuals with DSM-IV alcohol dependence.