This report presents the results of a study of alcohol testing and reporting methods and rates for drivers involved in fatal traffic crashes. It identifies the best practices currently in use and the major barriers to increasing testing, as determined from detailed studies of ten States. The report concludes with suggestions for implementing these best practices.

Background

Drivers impaired by alcohol have been and continue to be a major highway safety problem. In 2002, the National Highway Traffic Safety Administration (NHTSA) estimates that 14,662 drivers involved in fatal crashes -- over one-quarter of all drivers in fatal crashes -- had a positive Blood Alcohol Concentration (BAC), and 12,344, or 21 percent of all drivers in fatal crashes, had a BAC level at or above the typical per se limit of 0.08. Extensive efforts by many persons and organizations have reduced impaired driving substantially over the past 20 years. These efforts continue, since every fatality caused by an impaired driver is preventable.

Accurate data on impaired driving are critical to this effort. Accurate data are the best and most objective way to measure the size of the impaired driving problem, describe its characteristics, evaluate trends, explore potential countermeasures, and evaluate the effects of laws and programs. Accurate data are needed at both national and State levels.

BAC Data in FARS

The most comprehensive data on impaired driving come from drivers involved in fatal crashes, both fatally injured and surviving drivers. NHTSA has collected these data in its Fatality Analysis Reporting System (FARS) since 1975. The most accurate data come from direct measurements of a driver’s BAC from a blood or breath test. Lacking a BAC, inferences about a driver’s alcohol use are derived from an investigating law enforcement officer’s observations and other information. The quality of these observations varies substantially. Even the best may not estimate BAC levels accurately.

In 1982, NHTSA began publishing both National and State estimates of alcohol involvement in fatal crashes. In 1982, 54 percent of the fatally injured drivers and 16 percent of the surviving drivers in fatal crashes had a BAC test result recorded in FARS. Together, only 33 percent of all drivers in fatal crashes had a known BAC. For the remaining two-thirds, NHTSA used a complex imputation methodology to estimate the likelihood that each driver’s BAC fell into each of three categories 0.00, 0.01-0.09, and 0.10 and above (Klein, 1986).

Twenty years later, in 2002, BAC data have improved somewhat, with results in the FARS annual report data file recorded for 65 percent of fatally injured drivers, 25 percent of surviving drivers, and 43 percent of all drivers in fatal crashes. Final data for
2002, to be released in 2004, may increase these figures slightly. But this still leaves well over half the drivers with no objective BAC data. In 2002, NHTSA introduced a new imputation method to estimate the missing data at any BAC level (NCSA, 2003a; Subramanian, 2002). However, estimated data still may be inaccurate. State-level estimates are quite accurate for States with high levels of known BAC results but may be far less accurate for States with substantial missing BAC data.

States do differ substantially. In the 2002 annual report file, BAC levels are known for 85 percent or more of fatally injured drivers in 11 States, but for less than 50 percent in another 11 States. They are known for 60 percent or more of surviving drivers in 7 States, but for less than 15 percent in 16 States. Tables 1 and 2, from NHTSA’s 2002 State Alcohol Estimates (NCSA, 2003b), present the most recent BAC data by State.

Study Goals and Plan

This study’s goals were to identify the best practices for, and the barriers and problems that hinder, obtaining BAC data for drivers involved in fatal traffic crashes, and to provide recommendations for States that wish to improve their alcohol testing and reporting. The study’s principal activity was to study BAC testing and reporting procedures in detail in ten States. A steering committee of representatives of national organizations involved in BAC testing and reporting helped guide the project.

Outline of Report

Chapter II describes the study methods. Chapter III presents results on how States decide whom to test, through laws, policies, and practices. Chapter IV describes process issues from the point at which a blood or breath sample is taken, through determining the BAC, to recording the BAC in FARS. Chapter V presents best practice recommendations and Chapter VI discusses potential strategies for implementing these recommendations. The Appendices document State laws affecting BAC testing and the ten State case studies.

Table 1. BAC Testing and Reporting Rates by State, FARS 2002 (click here to view)

Table 2. BAC Testing and Reporting Rates by State, Sorted by Percent Known, FARS 2002 (click here to view)

FARS Data Files

FARS produces a data file for the previous year’s crashes in approximately June. NHTSA uses this file for its annual reports and Fact Sheets. FARS continues to accept crash data for at least a full year after the crash date, and produces the final annual file approximately 18 months after the calendar year’s end (Traffic Safety Facts 2001, DOT HS 809 484, p. 3.) In particular, BAC data may be added during this time.

Tables 3 and 4 compare BAC testing and reporting data from the 2001 annual report and final files. Data for about one-third of the States were unchanged from the annual report to the final file. Most other States added, or occasionally subtracted, a small number of drivers and tests. A few States had substantial changes: California added 209 known BACs of fatally injured drivers; Ohio added 136; Georgia added 131; Pennsylvania added 118. Georgia added 182 known BACs of surviving drivers while Pennsylvania added 71 and Texas added 68. Nationwide, the percent tested increased from the annual report file to the final file by 4.1 percent for fatally injured drivers and 1.1 percent for surviving drivers while the percent known increased by 4.5 percent for fatally injured drivers and 1.6 percent for surviving drivers.

NHTSA’s goal is to obtain as complete BAC reporting as possible in the annual report file, because this file not only provides the first complete look at the previous year’s alcohol-related crashes, but also will be the file used for all analyses for at least a year. All data presented in this report come from annual report files unless explicitly noted otherwise.

Table 3. BAC Testing and Reporting in Annual Report and Final Files, FARS 2001 Fatally Injured Drivers (click here to view)

Table 4. BAC Testing and Reporting in Annual Report and Final Files, FARS 2001 Surviving Drivers (click here to view)