Effects of alcohol on the geographic variation of driver fatalities in motor vehicle crashes.
OBJECTIVE: To determine whether the increased risk of dying in a rural vs nonrural motor vehicle crash (MVC) can be attributed to driver demographics, crash characteristics, or police-reported alcohol use. METHODS: A retrospective cohort study was conducted, comparing all rural (116,242) and a 20% random sample of nonrural (104,197) Michigan drivers involved in an MVC during 1994-1996. Data consisted of all police-reported traffic crashes on public roadways. A logistic regression model was created, using survival as the dependent variable and gender, age, crash characteristics, and rural or nonrural county as independent variables. Driver alcohol use, as reported by the investigating officer, was introduced into the model, and the effect was analyzed. RESULTS: Nonsurvivors represented 0.2% of the total; 99.8% were survivors. Police-reported alcohol use was reported for 3.9% of drivers. Drivers in rural MVCs were more likely to be male, to be more than 50 years of age, to have been drinking alcohol, and to have more severe vehicle deformation as a result of the MVC. The relative risk (RR) for MVC nonsurvivors was 1.69 [95% confidence interval (CI) = 1.3 to 2.1] times higher for drivers in rural than nonrural counties. After adjusting for demographic and crash characteristics, the RR was 1.56 (95% CI = 1.2 to 1.9). Controlling for alcohol and its interactions decreased the RR to 1.26 (95% CI = 0.6 to 2.4), a nonsignificant difference between rural and nonrural MVC mortalities. CONCLUSIONS: Alcohol use by drivers in Michigan was a significant contributor for nonsurvivors of rural crashes. Efforts to decrease rural MVC mortality must address alcohol use.