The fraction of traffic fatalities attributable to alcohol

The vast literature on alcohol's effect on traffic safety does not contain even a moderately satisfactory answer to one of the most basic questions, namely "What is the fraction of all traffic fatalities attributable to alcohol use?" A published estimate of 23.7% based on an erroneous calculation has been widely quoted. This paper combines 1987 Fatal Accident Reporting System (FARS) data from 26 states that recorded blood alcohol concentrations for over 84% of fatally injured drivers with published estimates on how alcohol affects crash risk. By categorizing all traffic fatalities as either nonoccupants of vehicles, or occupants killed in single-vehicle, two-vehicle or three-or-more-vehicle crashes, and developing calculation procedures appropriate for each category, the fraction of all fatalities due to alcohol was inferred. The main finding was that eliminating alcohol would reduce traffic fatalities by (47 +/- 4)%. It was also concluded that alcohol use changes from 1982 to 1987 have reduced traffic fatalities by 12% (6,400 fatalities), which helps explain the absence of the fatality increase predicted because of a buoyant economy. Reducing the fraction of fatalities due to alcohol from the 1987 value of 47% to 42% (say) would reduce all traffic fatalities by 8%.