Factors influencing duration between non-fatal and fatal crashes among high-risk drivers

A small fraction of drivers accounts for a disproportionately large share of the world's most serious road crashes, yet remarkably little is known about how quickly these high-risk individuals progress from one crash to the next. This temporal blind spot limits the ability of policymakers and enforcement agencies to intervene at the moments when drivers are most vulnerable. To address this gap, this study investigates the determinants of the time interval between a prior non-fatal crash and a subsequent fatal crash among crash-prone drivers in the United States. Using 5 years of Fatality Analysis Reporting System (FARS) data (2015- 2019), a sample of 36,144 fatal crash-involved drivers with recorded prior non-fatal crashes was analysed through a random-parameter hazard-based duration model. The findings demonstrate that fatal crash recurrence is not random but reflects distinct behavioural and demographic risk signatures. Male drivers, young drivers (16-24 years), Hispanic drivers and individuals with extensive violation histories exhibited significantly shorter intervals between non-fatal and fatal incidents. Prior speeding violations, license suspensions, other non-driving under the influence (DUI) infractions and involvement in multiple previous crashes sharply accelerated the progression to a fatal crash. Characteristics of the fatal crash itself, including drug impairment, speeding, invalid licensing and single-vehicle involvement, were also associated with compressed temporal intervals. The incorporation of random parameters reveals further heterogeneity, particularly in the influence of prior speeding violations. By and large, the temporal structure of crash recurrence observed in the study aligns with the long-standing theory of crash proneness.