1024 — Effectiveness of AUDIT-C as a Screening Test for Alcohol Misuse in Three Race/Ethnic Groups
Frank D (VA Puget Sound Health Care System) , Debenedetti A
(VA Puget Sound Health Care System), Volk RJ
(Baylor College of Medicine), Williams EC
(VA Puget Sound Health Care System, University of Washington), Kivlahan DR
(VA Puget Sound Health Care System, University of Washington), Bradley K
(VA Puget Sound Health Care System, University of Washington)
The AUDIT-C, a brief validated screen for risky drinking and/or alcohol abuse and dependence, was implemented in VA clinics nationwide in 2004. The AUDIT-C was validated in predominantly White populations but its performance in different racial/ethnic groups is unclear. This study evaluated the validity of the AUDIT-C among primary care patients from the predominant race/ethnic subgroups within the U.S.: White, African-American, and Hispanic.
This cross-sectional study was set in a University-based family practice. Recruited patients self-identified as White, African-American, and Hispanic and completed the AUDIT-C and a diagnostic interview to determine the reference standard (alcohol misuse). Alcohol misuse was defined as current DSM-IV alcohol abuse or dependence and/or risky drinking in the past year (men: >14 drinks/week or >5 drinks/occasion; women: >7 drinks/week or >4 drinks/occasion). Analyses compared the AUDIT-C to interview criteria for alcohol misuse, and included Areas under Receiver Operating Characteristic Curves (AuROCs) and 95% confidence intervals, sensitivity, and specificity stratified by race/ethnicity and gender. “Optimal” cut-points for the AUDIT-C maximized sensitivity and specificity (nearest left upper corner of ROC curve,) in men and women. Sensitivities and specificities of optimal cut-points in the three race/ethnic groups were compared using two-sample tests of proportions.
Of 1,445 eligible outpatients, 1,292 (90%) participated: 504 White, 458 African-American, and 333 Hispanic patients. AuROCs were >0.85 in all 3 groups, with no significant differences across racial/ethnic groups in men (p=0.43) or women (p=0.12). However, at optimal cut-points, there were statistically significant differences in the sensitivities of the AUDIT-C. The sensitivity was significantly higher in Hispanic women (85%) than African-American (67%; p=0.03) or White (70%; 0.04) women at a cut-point of 3. In men, the sensitivity was significantly higher in Whites (95%) than African Americans (76%; p=0.01), and non-significantly higher than Hispanics (85%; p=0.11) at a cut-point of 4.
Although the overall performance of the AUDIT-C was excellent in all three racial/ethnic groups as reflected by high AuROCs, there were significant differences in the AUDIT-C’s performance at optimal cut-points.
This is the first and largest study to validate the AUDIT-C in a racially and ethnically diverse population.