Selection of screening items for alcohol abuse and alcohol dependence among Mexicans and Mexican Americans in the emergency department.

OBJECTIVE: This article reports on the selection of screening items to detect Mexican or Mexican-American patients in the emergency department (ED) who have alcohol problems and could benefit from an intervention or a referral for treatment. Items are tested against the Rapid Alcohol Problems Screen (RAPS), which has been optimized from standard screening instruments and has outperformed these instruments. METHOD: The performance of individual items from standard screening instruments (CAGE, TWEAK, AUDIT, TRAUMA and BMAST) against International Classification of Diseases, Tenth Revision and Diagnostic and Statistical Manual, Fourth Revision criteria for alcohol abuse and dependence was evaluated in a merged probability sample (N = 869; 72% men) of 537 ED patients from three hospitals in Pachuca, Mexico, and 332 Mexican-American ED patients in Santa Clara County, CA. Logistic regression and tree-classification models were used for item selection. RESULTS: We found a prevalence of 15% for alcohol dependence and a prevalence of 28% for alcohol abuse or dependence in the merged sample. The RAPS items did not perform as well in terms of sensitivity (93%) as the optimal five items identified in these analyses (sensitivity = 98%) for alcohol dependence, but did demonstrate better specificity (79%) than the optimal five items (65%), which is an important consideration in a time of cost containment. Both sets of items showed better sensitivity and positive predictive value but similar Receiver Operating Characteristic values for respondents in the high acculturation group compared to those at other levels of acculturation. Differences in positive predictive value across all subgroups tended to increase at increased cutpoints, especially for the RAPS. CONCLUSIONS: These analyses suggest that the RAPS performs favorably compared to those items optimized, in this sample of Hispanic ED patients. Based on comparative item performance in these analyses, the RAPS may hold promise as a useful tool for screening for alcohol dependence, but requires further evaluation as a stand-alone instrument in comparison with other standard screening instruments.