Temporal instability of injury severity determinants in chain-reaction crashes: A multi-period Maximum Abbreviated Injury Scale-based analysis

Chain-reaction crashes (CRCs) result in higher severity compared to single-vehicle and two-vehicle crashes. However, previous research on CRCs is insufficient and overlooks substantial considerations. This study addresses the existing gap by examining factors influencing CRC severity and their temporal evolution. A novel approach is introduced to address the limitations of the traditional attitude in measuring injury outcomes of multi-vehicle crashes. The proposed method quantifies total injury severity using MAIS scores and categorizes them into ordered levels through K-means clustering. Ordered probit and random parameter ordered probit models are applied to the observations, and temporal stability tests are conducted to assess variations in model parameter estimates over time. As a case study, CRC data from Pennsylvania (2010–2023) is analyzed across four time periods. The CRC events are first categorized into distinct scenarios, with these scenarios introduced as new explanatory variables. The results reveal a temporal evolution of CRC risk factors. Female driver involvement exhibited diminishing effects, whereas elderly drivers, impaired driving, speeding, and older vehicles emerged or reemerged as significant severity-intensifying factors, particularly during the COVID-19 period. Heavy striking vehicles and a specific interaction effect consistently escalated injury severity. These findings underscore the need for adaptive safety policies.