Screening for problem drinking: does a single question work?

BACKGROUND: Brief interventions with problem drinkers have been shown to be effective, but physicians often do not ask about alcohol use. If a single question could effectively screen for problem drinking, it might facilitate intervention with problem drinkers.
METHODS: A cross-sectional study was undertaken to address the clinical utility of the question, "On any single occasion during the past 3 months, have you had more than 5 drinks containing alcohol?" Placing it between questions about tobacco and seat-belt use, we presented the three questions in writing to 1435 patients; 95.3% answered them. With a systematic sample of 101 patients who answered yes and 99 who answered no, we administered the Alcohol Use Disorders Identification Test in writing followed by two gold-standard interview instruments: (1) a calendar-based review of drinking, with at-risk drinking defined as drinking more than 4 drinks on one occasion or more than 14 drinks per week for men, and more than 3 drinks on one occasion or 7 per week for women; and (2) the alcohol questions in the Composite International Diagnostic Interview, with alcohol-use disorders defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria. We defined problem drinking as either at-risk drinking in the previous month or an alcohol-use disorder in the past 12 months.
RESULTS: The single question had a positive predictive value of 74% and negative predictive value of 88% for problem drinking, with a sensitivity of 62% and a specificity of 93%. The question's utility was similar for detecting at-risk drinking and current alcohol-use disorders; it correctly identified all 29 patients who had both.
CONCLUSIONS: A single question about alcohol can detect at-risk drinking and current alcohol-use disorders with clinically useful positive and negative predictive values.