A matched case cohort study on occupants injury severities in underride crashes.

OBJECTIVE: Underride crashes have been consistently associated with catastrophic occupant injuries. However, quantitative evidence comparing injury severity between underride and non-underride frontal crashes remains limited. This study aimed to investigate injury patterns and mechanisms associated with underride crashes using real-world crash data. METHODS: This retrospective observational study utilized the Korean In-Depth Accident Study (KIDAS) database from January 2011 to December 2024. To reduce confounding between underride and non-underride crashes, a 1:2 matching approach was performed using exact matching for vehicle type and seatbelt use, followed by nearest-neighbor propensity score matching (PSM) based on collision deformation classification (CDC) extent with a caliper of 0.2. After PSM, generalized estimating equations (GEEs) models were used to compare the general characteristics between the two groups, including seat position, crash object, seatbelt use, and frontal and knee airbag deployment, while accounting for correlations within matched sets. Fisher's exact test was applied when a GEE model failed to converge due to extreme data sparsity. Injury outcomes were evaluated using GEE logistic regression models at MAIS/AIS 2, 3, and 4+ severity levels and were reported as adjusted odds ratios (aORs) with 95% confidence intervals (CI). CDC-based vehicle deformation patterns were examined to interpret underride-related injury mechanisms. RESULTS: In this study, 80 underride occupants were matched to 151 non-underride occupants. Whole-body MAIS 2, MAIS 3, MAIS 4+, and fatality were not significantly associated with underride status. In injury-record-based Abbreviated Injury Scale (AIS) models, underride crashes were associated with significantly higher odds of AIS 3 head injuries (aOR: 3.292; 95% CI: 1.331-8.146; p = 0.010). Lower extremity injuries showed significantly lower odds at AIS 2 (aOR: 0.325; 95% CI: 0.141-0.749; p = 0.008) and AIS 3 (aOR: 0.175; 95% CI: 0.058-0.526; p = 0.002). No statistically significant associations were observed for the estimable AIS 4+ outcomes. CONCLUSIONS: Underride crashes were associated with higher odds of AIS 3 head injuries and lower odds of AIS 2 and AIS 3 lower extremity injuries compared with non-underride crashes, whereas whole-body injury severity and fatality were not significantly associated with underride status. These findings are consistent with body region-specific injury patterns related to impact loading toward the upper occupant compartment. Future investigations should consider vertical impact height compatibility and rear impact guard (RIG) performance in relation to these injury patterns.