An analysis of factors influencing accidents on road bridges in Norway

•A negative binomial regression model for 6824 road bridges.•Traffic volume is most important factor influencing accidents.•In-depth study of 31 fatal accidents.•Curves, slippery road and lack of guardrail contributing factors. Factors explaining systematic variation in the number of injury accidents on road bridges in Norway during 2010–2016 have been identified by means of negative binomial regression models. A total of 6824 bridges recording in total 1368 accidents were included. Although almost 90% of the bridges recorded zero accidents, there is no evidence of an excessive number of zeros, often referred to as “zero-inflation”. Traffic volume, stated as AADT (Annual Average Daily Traffic), was found to be the single most important factor influencing the number of accidents. It explained nearly 72% of the systematic variation in the number of accidents. The number of accidents increased less than proportionately with traffic volume, meaning that accidents per million vehicle kilometres declined with increasing traffic volume. Long bridges were found to be safer than short bridges and recently built bridges were safer than older bridges. Based on in-depth studies, a more detailed analysis of factors associated with fatal accidents was performed.