Traffic accidents impart both economic and social costs upon communities around the world, hence the desire for accident rates to be reduced. For this reduction to occur, the factors influencing the occurrence of accidents must be understood. The role of congestion in modifying accident risk has been widely studied, but consensus has not been reached, with conflicting results leaving open questions. An inverse relationship between accidents and congestion would imply a benefit of congested conditions for road safety, posing a difficult situation for traffic management. This paper assesses articles that reveal the shape of the relationship between traffic accidents and congestion. We find a positive linear response to dominate the literature. However, studies with higher numbers of statistical units tend to show a U-shaped relationship. This suggests an important role of high spatio-temporal traffic data in understanding factors causing accidents and identifying the combination of real-time conditions which may lead to increased accident risk. Modern advancements in traffic measurement systems provide the ability for real-time alleviation of accident-prone conditions before they can fully develop.
Keywords: Bluetooth; congestion; real-time traffic data; traffic accidents; traffic volume.