Unraveling the relationship between trauma types and traffic crash characteristics: an error component logit approach

On-site decisions regarding the identification of critical injuries have an important role in the survival rate of people injured in road crashes. This study focuses on modelling the probability of a wide range of injury types with crash, vehicle and personrs characteristics, with separate analysis for car occupants and VRUs. The considered injury types are combinations of primary and secondary injuries occurring in the following body parts: head, neck, thorax, spine, upper extremities, lower extremities and other. The employed model is the mixed-logit model due to its ability to represent heteroscedasticity and cross-nested correlations across combinations of injury types. Police and hospital data on from Funen, Denmark, during the years 2002 to 2008, serves for the analysis. Results show that injury types, and in particular severe injuries such as head-neck, spine and thorax, are associated with safety gear use, crash configuration, infrastructure characteristics, environmental and light conditions, in addition to individual characteristics.