Identifying problem drinkers: lack of sensitivity of the two-question drinking test.

BACKGROUND AND OBJECTIVE

The two-question drinking test ("Have you ever had a drinking problem?" and "When was your last drink?") is frequently cited as an accurate method of screening for alcoholism, yet its performance has never been validated. This study evaluated its utility among patients in two primary care settings.

PATIENTS AND METHODS

The two-question drinking test was compared to the Michigan Alcohol Screening Test (MAST) as the standard criterion for the diagnosis of alcoholism. The subjects were 334 nonpregnant English-speaking patients over the age of 18 seen in two university outpatient teaching clinics.

RESULTS

The overall prevalence of alcoholism based on a MAST score > or = 5 was 18%. The sensitivity of the two-question test was 53% and its specificity 93%. The question "When was your last drink?" added little to the question "Have you ever had a drinking problem?" The latter question alone had a sensitivity of 50% and a specificity of 97%.

CONCLUSIONS

The two-question test is not sensitive enough to be used to screen for alcohol abuse among outpatients. However, because of the low false positive rate, a positive response to the question "Have you ever had a drinking problem?" may help identify alcoholic patients.