"Nurse, I only had a couple of beers": validity of self-reported drinking before serious vehicular injury.

BACKGROUND: Of the more than 40000 Americans killed each year in vehicular crashes, 40% are involved in alcohol-related collisions. Although self-reported alcohol use has become an anchor for alcohol intervention after traffic crashes, clinicians are often skeptical about the truthfulness of self-reporting. OBJECTIVE: To determine the validity of self-reported alcohol consumption of vehicular occupants hospitalized for a serious, alcohol-related injury. METHODS: Non-alcohol-dependent subjects 18 years and older who were injured in motor vehicle crashes were interviewed. The self-reported number of standard drinks, time that drinking commenced, sex, and weight were used to calculate estimated blood alcohol concentration. This value was compared with the blood alcohol concentration measured at admission. RESULTS: Of the 181 subjects, 60% provided sufficient data to calculate the estimated concentration. Seven men with admission concentrations of 10 mg/dL or more denied drinking. Among the 113 subjects with estimated concentrations who acknowledged drinking (excluding the 7 who denied drinking), the mean concentration at admission was 158.67 mg/dL, and mean estimated concentration was 83.81 mg/dL. According to multiple regression analyses, weight and number of drinks accounted for 3% of the variance in alcohol concentration at admission for women (R=0.174, F2,40=0.623, P=.54) and for 29% of the variance in men (R=0.543, F2,128=26.71, P< .001). CONCLUSIONS: Most persons who drink before vehicular injury acknowledge drinking. Self-reported data from men generally reflect the overall trend of alcohol consumption but with systematic underreporting. Reports from women are less predictable.