An overdispersed count regression model for analyzing road accident fatalities and injuries in Bangladesh

Abstract

Road traffic accidents (RTAs) continue to be a major global public health issue, particularly in countries with developing road safety infrastructure like Bangladesh, where the road traffic fatality rate remains alarmingly high. This study aims to examine the relationship between road traffic accident outcomes—fatalities and injuries—and multiple contributing factors, including driver behavior, vehicle characteristics, environmental conditions, and road infrastructure. Using a comprehensive dataset of 64,050 police-reported road traffic accidents in Bangladesh (2006–2015), we apply a Negative Binomial Regression (NBR) model to account for overdispersed count data. Our results highlight that driver-related factors such as seatbelt use and age, vehicle factors such as fitness certification, and environmental conditions such as weather and road geometry significantly influence both fatalities and injuries. Notably, roads without dividers and in rural areas were found to be particularly hazardous. The study underscores the need for targeted road safety interventions, such as improved enforcement of seatbelt use, infrastructure upgrades (e.g., dividers, lighting), and more transparent vehicle fitness monitoring. By integrating driver, vehicle, environmental, and infrastructural variables, this study provides a comprehensive understanding of road traffic accident severity in Bangladesh, offering data-driven insights to inform evidence-based policymaking and infrastructure planning aimed at reducing road traffic injuries and fatalities.