Accuracy of Estimated Blood Alcohol Concentration Norms From College Student Drinking Survey Data: Verification Using Matched Late-Night Breath Measurements

Abstract

This study assesses the accuracy of estimated blood alcohol concentration (eBAC) from survey data in a college student sample using six computational methods from the literature. Corrections for heavy-drinker metabolic rate and preabsorptive error are added to the estimating formula. Late-night interception of 2,282 students returning to residence halls every night of the week was used in a double-blind breath test and survey about typical and current-evening alcohol use and body characteristics. Measured breath alcohol concentration (BrAC) was used to assess the accuracy of the computational models. The Seidl method yielded the best performance, with further reduction in error observed when a heavy-drinker metabolic rate correction was applied. In situations when an intercept occurred shortly after alcohol consumption, a correction for alcohol absorption kinetics further reduced error. Despite significant remaining error of estimates for a substantial minority of individuals, using the refined calculation of an eBAC permitted computation of low risk/low impairment (<.05 g/dl) and high risk/high impairment (≥.08 g/dl) prevalences that were virtually identical to the corresponding prevalences revealed using the data from the BrAC measurements. In late-night intercepts of college students, an eBAC can be calculated from survey data with only modest accuracy for individuals. This measure can quite accurately measure norms, however, with regard to actual BAC levels in circumstances in which students are returning from activities that might include heavy or high-risk drinking.