Objective:The primary objective of the study was to determine if drinking patterns on the days immediately prior to an alcohol-related motor vehicle crash (ARMVC) were significantly different than drinking patterns in the weeks prior to the crash.

Methods:Following ARMVC, 187 hospitalized non-alcohol dependent young-adults (43 females, 144 males) were enrolled. Mean age was 29.03 years, mean blood alcohol level was 165.18 mg/dL, and mean injury severity score was 10.50. When alcohol-free, subjects were interviewed by nurse clinicians to determine the quantity/frequency of alcohol consumption during the 28 days prior to the crash. Subjects reported the number of standard drinks using the Timeline Followback procedure. Total drinks/day were determined, with day 1 considered 4 weeks prior to the crash and day 28 the day of the crash. A random-intercepts general linear mixed model (GLMM) was used to test the effect of several covariates (segment 1 [days 1-26], segment 2 [days 27-28], age, sex, race, holiday/non-holiday period, driver/passenger status, and weekend/weekday crash) on the amount of standard drinks/day.

Results:There was no significant interaction among the covariates. The only significant predictors of drinks/day were segment 2 (b = .322, p < .0001) and gender (b = -.221, p = .016). The positive, statistically significant slope for segment 2 indicated an increase in consumption of drinks/day in the two-day period prior to the ARMVC and the negative slope for gender indicated greater consumption of drinks/day for men than women.

Conclusion: Persons injured in an ARMVC had a significant increase in alcohol consumption on the day before and the day of vehicular crashes (days 27 and 28) as compared to the first 26 days in the 28-day period preceding the crash. When non-alcohol-dependent subjects are counseled to reduce their risk of traffic crashes, they should be alerted that when their patterns of drinking change, they are at higher risk than usual for a crash.