Runlin Li, Rui Liu, Mingxu Yin, Qin Peng
In this defined sample of CISS crashes, pickup index-vehicle occupants had lower adjusted hospitalized-or-fatal odds and model-standardized risk than car index-vehicle occupants under released weights, but sensitivity to modified weights increased uncertainty. The severity, temporal, selected-partner, and whole-crash analyses did not establish overall car-pickup compatibility. These associations estimate neither crash-involvement risk nor causal vehicle effects and do not support a general safety ranking.
OBJECTIVES: To compare Maximum Abbreviated Injury Scale (MAIS) severity and hospitalized-or-fatal injury among adult front-row occupants of passenger cars and pickups in 2017-2024 Crash Investigation Sampling System (CISS) crashes. The car or pickup, termed the index vehicle, had vehicle-level CRASHCAT = 5, officially labeled "Intersecting Paths (Vehicle Damage)," which describes trajectories rather than roadway location. We examined calendar time, selected-partner outcomes, and whole-crash outcomes.
METHODS: Public CISS files were pooled, and survey-weighted models used the released case weight (CASEWGT), strata, and primary sampling units. Outcomes were MAIS 1, MAIS 2, MAIS 3, MAIS 4 + F (MAIS 4 or higher or recorded fatality; F denotes fatality), and hospitalized-or-fatal injury. Sensitivity analyses used crash-grouped covariance and modified weights that capped and redistributed weights above an author-defined threshold. Time models used 2017-2023 data. Selected-partner analyses assessed whether any eligible front-row adult in the partner was hospitalized or died; whole-crash analyses combined eligible front-row adults across included vehicles.
RESULTS: Unweighted counts were 4,220 occupants in 2,620 crashes; the hospitalized-or-fatal model included 3,196 occupants and 299 events. Compared with passenger-car occupants, pickup occupants had 61.0% lower hospitalized-or-fatal odds (odds ratio [OR], 0.390; 95% confidence interval [CI]: 0.158, 0.963). Model-standardized risks were 4.09% (95% CI: 2.78%, 5.97%) for car and 1.73% (95% CI: 0.79%, 3.72%) for pickup, a 2.36 percentage-point lower risk (risk difference, -2.36 percentage points; 95% CI: -4.30 to -0.42); both differences were statistically significant. Under modified weights, the estimated odds remained 52.2% lower (OR, 0.478; 95% CI: 0.226, 1.011), although the difference was not statistically significant. The estimated odds for MAIS 1-3 were also lower for pickup occupants, but those differences were not statistically significant. MAIS 4 + F relied on three pickup events from three primary sampling units. We found no clear evidence that the hospitalized-or-fatal difference between pickup and car index-vehicle occupants narrowed during 2017-2023. Selected-partner estimates were imprecise, and whole-crash results were descriptive.
CONCLUSIONS: In this defined sample of CISS crashes, pickup index-vehicle occupants had lower adjusted hospitalized-or-fatal odds and model-standardized risk than car index-vehicle occupants under released weights, but sensitivity to modified weights increased uncertainty. The severity, temporal, selected-partner, and whole-crash analyses did not establish overall car-pickup compatibility. These associations estimate neither crash-involvement risk nor causal vehicle effects and do not support a general safety ranking.