Methods for investigating crash risk: Comparing case–control with responsibility analysis

Abstract Objective: There are 2 primary methods for establishing relative risk: case–control studies, in which crash and matched control data are collected separately, and responsibility analysis, which exploits a single existing crash database by using nonresponsible drivers as an “induced exposure” control group (which is less expensive and therefore more feasible for examining the large number of substances that can impact driving behavior). Though both approaches are scientifically sound and methodologically valid, each approach has its own inherent obstacles to overcome. In this article, we examine in detail how different criteria for the development of control cases influence the accuracy of crash risk estimates for drivers with positive blood alcohol concentrations (BACs). Methods: We applied responsibility analysis to crash-involved drivers in a recent crash case–control study, thereby providing 2 sets of control cases: Those from responsibility analysis and those from the case control study. Results: Case–control and responsibility analysis crash risk curves did not differ significantly, indicating that both systems generate valid estimates of the relative crash risk of drivers on the road. Conclusions: The results suggest that when researchers are faced with finance or time constraints that make case–control studies infeasible, responsibility analysis should be considered a viable alternate methodological approach.