A method is presented for estimating the risk of driver involvement in injury crashes for case-control data where control drivers have reliable measures of BAC (blood alcohol concentration) and other driver characteristics, but crash-involved drivers do not have BAC measures. The usual estimates of risk associated with BAC derived from case-control studies depend on the provision of driver BAC for virtually all case (crash-involved drivers) and control (drivers on the road) samples. The method described here makes use of the case-control study design but requires only the total numbers of case drivers within groups of drivers defined by variables common to both samples. The risk associated with BAC of New Zealand driver night-time involvement in injury crashes is estimated to illustrate the application of this method. (Author/publisher)
Abstract