Exploring determinants of bicycle fatalities in urban mixed-traffic road segments: A Bayesian microscopic study

Road Traffic Injuries (RTIs) are one of the leading causes of death and disability worldwide. More than half of this burden falls on the Vulnerable Road Users (VRUs), which includes pedestrians, cyclists, and motorised two- and three-wheeler riders. In developing economies, bicycles are primarily used by the low-income population for commuting to work. Although the frequency of bicycle fatalities is lower, the rate per unit exposure is very high, making the safety investigation of bicycle riders important. This study investigates the factors that contribute to bicycle fatalities in heterogeneous traffic conditions. A microscopic crash investigation study is performed at the level of the road segment in Delhi, India. The preliminary outcomes suggest that cars and Heavy Commercial Vehicles (HCVs) are the two main road user categories that cause bicycle fatalities. The fatalities mainly occur in the early morning hours. Count data models are also developed that underscore the effect of various physical and operational road characteristics on bicycle fatalities. The outcome suggests that the volume of bicycles (β ln (Cycle) = 0.340) and HCVs (β ln (HCV) = 0.320) increases the likelihood of bicycle fatality, while the volume of buses (β ln (Bus) = −0.390) decreases the fatality risk. The presence of a curb buffer on high-speed roads (β = 0.670) also increases the probability of bicycle fatalities. Zonal-level attributes are also included to investigate the effect of zonal land use and built environment characteristics on bicycle safety. The outcomes suggest that recreational land use (β = 1.390) increases the probability of bicycle fatalities. Inclusion of zonal-level attributes significantly improves model performance. Furthermore, the outcomes validate the concept of Safety-in-Numbers (SiN) for bicycle safety (β ln (Cycle) < 1) but suggest that the variability in the effect of bicycle volume on bicycle safety warrants further exploration. Lastly, the policy implications of the research outcomes are also discussed. • The study examines the factors that influence bicycle fatalities in a heterogeneous traffic condition. • The preliminary investigation suggests that cars and heavy commercial vehicles are the major impacting vehicle categories. • Inclusion of zonal-level attributes significantly improves model performance. • The volume of bicycles and HCVs increases the likelihood of bicycle fatality, while the volume of buses it. • Safety-in-Numbers (SiN) for bicycle safety got validated, but the outcomes suggest significant variability in the impact.