Alcohol, Drugs, and Crash Responsibility: Explaining the Grand Rapids Dip

Introduction: The “Grand Rapid Dip” was first reported by Borkenstein et al. (1974) and replicated in several other studies (e.g., Blomberg et al., 2005). These crash-control studies reported that, relative to .00 BAC drivers, crash risk declined at BACs of .01 to .03. This dip seems to suggest that drivers at very low BAC levels have a reduced risk of crashing compared to drivers with .00 BACs. Allsop (1966); Hurst (1973); and Hurst, Harte, and Frith (1994) argued that such a dip was an artifact of the confounding effects of uncontrolled covariates. This effort examines that possibility by determining the association between alcohol and crash responsibility at different BAC levels, in particular the proportion of drivers responsible for a crash at BAC=.00 and at .00<BAC<.05.

Methods: Stemming from a collaboration between PIRE and the CA DMV, we merged fatal crashes in the 1987-2009 California Statewide Integrated Traffic Records System (SWITRS) and the Fatality Analysis Reporting System (FARS). This merger allowed an examination of crash responsibility (present in the SWITRS) and BAC information (present in the FARS). Data merging was based on the following variables: crash date (year, month, day), crash time (hour, minute), driver age, driver sex, and vehicle model year. The matching was done at the party (driver) level. To be included, a crash had to involve exactly two vehicles (drivers); have one at-fault driver and one not-at-fault driver; and have both drivers identified as validly licensed, Suspended/Revoked, or unlicensed. Only passenger cars or pickup trucks were included. Analyses were conducted to examine if potential cofactors (drugs, inattention) could explain the Grand Rapids Dip.

Results: As expected, alcohol was found to be associated with crash responsibility: as BAC increases, so does the percentage of drivers responsible for the crash (from 44% at BAC=.00, to 83% at BAC≥.08). However, that increase was not linear: our findings reproduced the “Grand Rapids dip”, with only 36% of the .00<BAC<.05 drivers being responsible for the crash. When we added the contribution of drugs to this association, we found a U-shaped curve, with drugs more prevalent among sober (BAC=.00) and heavy drinker (BAC≥.08) drivers. For inattention-related crashes though, the pattern was less clear.

Conclusions: Although not conclusive, our findings seem to suggest that the Grand Rapids dip is caused, at least in part, by drug-driving being responsible for a larger proportion of crashes at BAC=.00 than at .00<BAC<.05.