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 m
ight facilitate intervention with problem drinkers. METHODS. A cross-s
ectional study was undertaken to address the clinical utility of the q
uestion, ''On any single occasion during the past 3 months, have you h
ad more than 5 drinks containing alcohol?'' Placing it between questio
ns about tobacco and seat-belt use, we presented the three questions i
n writing to 1435 patients; 95.3% answered them. With a systematic sam
ple of 101 patients who answered yes and 99 who answered no, we admini
stered the Alcohol Use Disorders Identification Test in writing follow
ed by two gold-standard interview instruments: (1) a calendar-based re
view 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) th
e alcohol questions in the Composite International Diagnostic Intervie
w, with alcohol-use disorders defined by the Diagnostic and Statistica
l Manual of Mental Disorders (DSM-IV) criteria. We defined problem dri
nking as either at-risk drinking in the previous month or an alcohol-u
se 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 9
3%. The question's utility was similar for detecting at-risk drinking
and current alcohol-use disorders; it correctly identified all 29 pati
ents who had both. CONCLUSIONS. A single question about alcohol can de
tect at-risk drinking and current alcohol-use disorders with clinicall
y useful positive and negative predictive values.