Multiple Imputation of Missing Blood Alcohol Concentration (BAC) Values in FARS

Alcohol involvement is a major contributing factor in the occurrence of motor vehicle traffic crashes. According to NHTSA’s preliminary estimate for 2002, alcohol was involved in about 42 percent of all motor vehicle crashes where there was a fatality. The most direct measure of a driver’s or nonoccupant’s alcohol involvement is a BAC test result reported in NHTSA’s Fatality Analysis Reporting System (FARS). These results are based on a variety of sources like breath-tests administered by police or a toxicology test from the Medical Examiner’s Office. BAC is the grams of alcohol in a deciliter of blood and can have a plausible value between 0 and 0.94. However, in FARS, BAC results are not known for many of the persons involved in the fatal crashes. The significant number of missing BAC values (about 58 percent in 2001) greatly inhibits the ability to report the extent of alcohol involvement, to identify groups for targeting campaigns to reduce impaired driving, and to evaluate the effectiveness of existing impaired-driving programs.