Variations on the CAGE alcohol screening questionnaire: Strengths and limitations in VA general medical patients

Background: Several variations on the CAGE alcohol screening questionnaire have been recommended. This report evaluates modifications and additions to the CAGE. Methods: Alcohol screening questionnaires were evaluated in male VA general medicine patients (n=227; mean age, 65.8). Mailed questionnaires included two scoring options for the CAGE (standard and last-year time frames), questions about quantity and frequency of drinking, two questions about episodic heavy drinking, and the question "Have you ever had a drinking problem?" Main analyses compared alcohol screening questions, at various cut-points, to a gold standard of hazardous drinking during the past year (greater than or equal to 14 drinks/week or greater than or equal to5 drinks on an occasion) and/or DSM-III-R alcohol abuse or dependence, based on standardized interviews. Results: The CAGE questionnaire with a past-year time frame was much less sensitive (0.57 vs. 0.77) but more specific (0.82 vs. 0.59) than the standard CAGE for detecting hazardous drinking during the past year and/or DSM-III-R alcohol abuse or dependence. An eight-item questionnaire that included the standard CAGE was most sensitive (0.92) but had low specificity (0.50). A single question about the frequency of drinking greater than or equal to6 drinks on an occasion, included in the eight-item questionnaire, was both relatively sensitive (0.77), and specific (0.83). Conclusion: The CAGE questionnaire with a past-year time frame was an insensitive alcohol-screening test. An eight-item augmented version of the standard CAGE was the most sensitive. A question about the frequency of drinking greater than or equal to6 drinks on an occasion performed better than the standard CAGE, which made it the optimal brief screening test for at-risk drinking.