It is a well-established fact that drunk driving plays a major role in fatal crashes. Research has demonstrated that alcohol in a driver’s bloodstream greatly impairs the driver’s ability to operate a vehicle safely. This report presents data obtained from the Fatality Analysis Reporting System (FARS) and analysed using a procedure to estimate blood alcohol content (BAC) levels for drivers and nonoccupants involved in fatal crashes. The report describes the magnitude of the drunk driving problem in the United States, highlights the circumstances under which fatal crashes are frequently associated with alcohol, and shows recent trends in alcohol involvement in fatal crashes. FARS contains data on all fatal traffic crashes from each of the states. The data include the results of chemical blood alcohol tests of drivers involved in fatal crashes when they are available. BAC tests form the basis for the statistics reported here; however, for a variety of technical, practical, and economic reasons, no state reports BAC values for all the drivers and nonoccupants involved in fatal crashes. The missing data rate ranges from a few percent in some states to nearly complete absence of testing in others. Although the nationwide BAC reporting rate has risen over the past 17 years from about 54 percent for fatally injured drivers and 16 percent for surviving drivers to 62 percent and 25 percent, respectively, there are still too many unknown BAC values to ignore. Alcohol involvement for drivers with unknown BAC values must therefore be estimated before valid statistics on the role of alcohol in fatal crashes can be determined. The same is true for nonoccupants. Several methods have been used in the past to estimate BAC values for drivers who were not tested. Each method has substantial limitations. To overcome many of the limitations and, in particular, to estimate BAC values for surviving drivers, the National Centre for Statistics and Analysis (NCSA) has developed a method based on discriminant analysis to estimate BAC values for all drivers involved in fatal crashes. The method is documented completely in a 1986 report from the National Highway Traffic Safety Administration (NHTSA). Briefly, the method estimates unknown BACs from the known BAC data for drivers with similar characteristics (such as sex, time of the crash, police alcohol indication, and vehicle type). This method was used to produce all the statistics in this report. BAC test results range from 0.00 grams per deciliter (g/dl) to more than 0.30. The numbers represent the amount of alcohol, by weight (grams), per amount of blood, by volume (deciliters). In practice, BAC test results measure the percentage of alcohol contained in the blood. For the purposes of this report, it is impractical to treat BAC as a continuous variable. Instead, BAC values are classified into three groups that tell the story of drunk driving in a concise and directly accessible way: - the 0.00 group of drivers (sober drivers), whose blood contains no alcohol - the 0.01 to 0.09 group of drivers, whose blood contains some alcohol but less than 0.10 percent (the legal threshold for intoxication in many states) - the 0.10+ group of drivers (intoxicated or drunk drivers), whose BAC is at or above the usual level of legal intoxication. Alcohol involvement is shown in the tables of this report by listing either the percentages of drivers in each of the three groups or the percentage of drivers in the high BAC (0.10+) group only, together with the total number of crashes or drivers as appropriate. Because some data are missing or unknown, totals from the disaggregated tables may not add up to the aggregated totals in summary tables. For example, the total of daytime and night-time crashes is less than the total of all crashes, because the crash times for a few crashes are not known. The procedure used throughout this report produces estimates, not exact counts. The possible error of the estimates is not known precisely, but extensive validation tests suggest that the error of any one estimate is relatively small and, more importantly, does not appreciably affect comparisons such as those in the trends section. In addition, it is necessary to emphasise that none of the tabulations presented here can be interpreted as implying a direct causal relationship between alcohol use and any other attribute of fatal crashes. Inferences concerning causality can only be made on the basis of additional information that is independent of the FARS data. Alcohol involvement in motor vehicle crashes is customarily reported for crashes or for the persons involved in crashes. For persons, the BAC status of each active participant (driver, pedestrian, or pedalcyclist) in the crash is reported individually. For crashes, the entire crash is classified at the highest BAC level of any active participant. In crashes in which individual BACs are known, the crash is given a count of 1 at the appropriate BAC level. Thus, a 0.00 crash is one in which all drivers and nonoccupants were sober; a 0.01-0.09 crash is one in which at least one driver or nonoccupant had a BAC level between 0.01 and 0.09 but none had a higher BAC level; and a 0.10+ crash is one in which at least one driver or nonoccupant was intoxicated. For crashes in which not all individual BACs are known, the count of 1 is distributed among the three BAC levels according to the probability distributions for alcohol involvement of each active participant. In crashes with only one active participant, the crash-level BAC distribution will be identical to that of the one participant. Where two or more persons were actively involved, joint probabilities are calculated from the individual BAC probability distributions to arrive at the crash-level BAC distribution. (A) http://www.nhtsa.dot.gov/people/ncsa/Reports/809-104.pdf
Samenvatting