Physician detection of drinking problems in patients attending a general medicine practice.

Objective: To assess the patient and physician characteristics that influence physicians' detection of problem drinking in their medical patients.

Setting: The outpatient medical clinic at an urban university teaching hospital staffed by interns and residents.

Design: Cross-sectional study of a randomly chosen subsample of consecutive patients.

Measurement: Univariate and multivariate analysis with calculated adjusted odds ratios of factors associated with physician detection of drinking problems. A problem was diagnosed according to the patient's results on the alcohol module of the Diagnostic Interview Schedule (DIS).

Results: Physicians detected 22% of 189 presumably inactive problems and 49% of 92 current problems, i.e., those that have occurred within the preceding year. Multivariate correlates of detection of active problems included male patient gender, presence of gastrointestinal complications of excessive drinking, number of concurrent medical disorders, and previous medical record reference to alcohol (p less than 0.05). Physician gender and year of training were not associated with detection.

Conclusion: Our physicians appear to rely on specific patient characteristics as well as the patient's medical record to detect drinking problems in their ambulatory patients. Their reliance upon these factors may hinder their detection of drinking problems in women patients and less seriously impaired individuals.