The aim of this study was to estimate crash risk for precrash factors in run‐off road crashes in curves by using a novel methodology for combining different types of available databases. When describing precrash situations, in‐depth or statistical crash databases are frequently used. However, without exposure data, risk for crashes cannot be estimated. Exposure could be estimated from naturalistic driving study (NDS) data or national statistics, but each data source has its weakness with regard to details and representativeness.
Using one statistical crash database and one NDS dataset, a matched case‐control methodology was applied to run‐off road crashes in curves (n=367) and a set of controls consisting of curve driving events. Precrash factors were harmonised and the contribution to crash risk was estimated using a conditional logistic regression model. The results showed that the risk increased with longer travel times, slippery road conditions and driving late at night and a significantly lower crash risk was found for drivers travelling together with a passenger.
The methodology presented offers a new perspective on how available datasets can be used to estimate the contribution to crash risk for precrash factors on a more detailed level than was made in previous studies.