Background: Primary care physicians can play a unique role in recognizing a
nd treating patients with alcohol problems.
Objective: To evaluate the accuracy of screening methods for alcohol proble
ms in primary care.
Methods: We performed a search of MEDLINE for years 1966 through 1998. We i
ncluded studies that were in English, were performed in primary care, and r
eported the performance characteristics of screening methods for alcohol pr
oblems against a criterion standard. Two reviewers appraised all articles f
or methodological content and results.
Results: Thirty-eight studies were identified. Eleven screened for at-risk,
hazardous, or harmful drinking; 27 screened for alcohol abuse and dependen
ce. A variety of screening methods were evaluated. The Alcohol Use Disorder
s Identification Test(AUDIT) was most effective in identifying subjects wit
h at-risk, hazardous, or harmful drinking (sensitivity, 51%-97%; specificit
y, 78%-96%), while the CAGE questions proved superior for detecting alcohol
abuse and dependence (sensitivity, 43%-94%; specificity, 70%-97%). These 2
formal screening instruments consistently performed better than other meth
ods, including quantity-frequency questions. The studies inconsistently adh
ered to methodological standards for diagnostic test research: 3 (8%) provi
ded a full description of patient spectrum (demographics and comorbidity),
30 (79%) avoided workup bias, 12 (of 34 studies [35%]) avoided review bias,
and 21 (55%) performed an analysis in pertinent clinical subgroups.
Conclusions: Despite methodological limitations, the literature supports th
e use of formal screening instruments over other clinical measures to incre
ase the recognition of alcohol problems in primary care. Future research in
this field will benefit from increased adherence to methodological standar
ds for diagnostic tests.