Quantity-frequency measures of alcohol consumption: beverage-specific vs global questions

Estimates of alcohol consumption (ounces of absolute alcohol per day, AA) based on beverage-specific and global quantity-frequency (QF) questions were compared in a survey representative of the adult drinking population in New York State. Beverage-specific AA estimates were higher than global (0.72, 95% confidence intervals = 0.68, 0.76) compared to 0.49 (95% confidence intervals = 0.47, 0.51), although estimates were highly correlated (r = 0.75). Discrepancies between beverage specific and global AA estimates increased as the number of beverages and the amount drunk increased. Sociodemographic characteristics were not significantly related to differences between beverage-specific and global AA estimates after adjusting for the amount drunk; however, drinking patterns did influence the differences. It was concluded that beverage-specific QF questions are probably more valid measures of alcohol consumption than global QF questions, but that the global questions provide useful information. Parallel analyses of variant global QF questions employed in the first US Health and Nutrition Examination Survey obtained similar results.