Traffic Tech
Technology Transfer Series
Number 222       May 2000


As of September 1999, 31 states defined driving with a blood alcohol concentration (BAC) above .10 percent as a crime per se, and another 17 states plus the District of Columbia set their per se limit at .08 percent. Alcohol-related fatalities have been declin-ing for nearly two decades, both in absolute num-bers and as a proportion of all fatalities. Despite this, there were still 15,935 alcohol-related traffic fatalities in the United States that accounted for nearly 38 percent of total traffic fatalities in 1998.

Based on extensive research over several decades, we now have overwhelming evidence showing that even BACs as low as .02 percent impair driving-related skills. One line of evidence grows out of laboratory research with dosed subjects. Confirming evidence comes from field research that compares the BACs of crash-involved with non-crash-involved drivers to determine the relative risk of crash involvement.

Two types of relative risk studies have been conducted. Classical studies used a procedure in which data from non-crash-involved drivers were collected at the same times and locations as the reference crash had occurred. This procedure ensures that the only potential difference between the crash and non-crash driver would be the presence or absence of alcohol. An alternative survey procedure compared crash-involved drivers identified in NHTSA's Fatality Analysis Reporting System (FARS) with data from the National Roadside Survey. The National Roadside Survey collected interviews and breath samples of 6,000 representative drivers. While this procedure loses some of the precision of specific crash sites, it gains reliability because it uses larger numbers and a broader representation of the country as a whole.

Clearly, many drivers are involved in crashes through no fault of their own but because of the mistakes of others. It is important to consider responsibility in selecting the crash-involved drivers. Usually this is accomplished by including only drivers in single vehicle crashes.

Westat and the Pacific Institute used recent data to refine estimates for alcohol-related relative risk of driver involvement in fatal crashes by age and sex as a function of BAC. To obtain these estimates, they combined data from FARS with exposure data from the joint NHTSA and the Insurance Institute for Highway Safety's 1996 National Roadside Survey (see Traffic Tech No. 152). Using logistic regression techniques they derived age and sex specific relative risk functions.

Relative Fatality Risk for Drinking Drivers by
Age and Sex in Single-Vehicle Crashes


The relative risk of being killed as a driver in a single vehicle crash at .08 BAC was found to be 13 times that at .00 BAC for drivers aged 21 to 34. Each .02 percentage point increase in the BAC of a driver more than doubled the risk of receiving a fatal injury in a single vehicle crash among male drivers 16-20, and nearly doubled the comparable risk for other driver groups. In part, contrary to the findings from the 1991 study that this one updates, for the 16-20 age group, females had a lower relative risk of receiving a fatal injury than males at every BAC. This is an important finding because of the increasing nighttime presence of young female drinking drivers observed in the 1996 Roadside Survey.

At the midpoint of the .08 -.10 percent BAC range, the relative risk of a fatal single vehicle crash injury varied between 11.4 (drivers 35 and older) and 51.9 (male drivers, 16-20). This similar pattern was also found overall (for any involvement in a fatal motor vehicle crash), although with smaller magnitudes.