Cabin heat exposure and pediatric vehicular heatstroke: a case-crossover study of 970 U.S. deaths, 1998–2025

Abstract Background Pediatric vehicular heatstroke (PVH) is the leading non-crash vehicle-related cause of death among U.S. children: roughly 37 die each year, the majority under five, mostly on days above 27 $${}^\circ$$ C (80 $${}^\circ$$ F). A closed cabin reaches lethal interior temperatures within 30–90 minutes, driven by solar radiation rather than ambient air, yet the heat–mortality relationship has never been characterized at the incident level, and the two state policy responses (caregiver criminal liability and Good Samaritan civil immunity for rescuers) have not been evaluated against that benchmark. Methods I assembled the near-complete media-tracked universe of 970 U.S. PVH deaths under age five (1998–2025), geocoded and linked to ERA5 reanalysis temperature and solar-radiation fields. A case-crossover design matched each death to location-specific non-event days, identifying the heat effect within each incident and absorbing all fixed child, vehicle, and place confounders. I compared apparent maximum temperature with a cabin-temperature index adding standardized solar radiation, and evaluated the two statutes across staggered state adoption using cluster-bootstrap and randomization inference, alongside a state-year analysis of CDC WONDER death-certificate counts. Reporting follows STROBE. Results The cabin-temperature index predicted death-day risk better than ambient temperature (odds ratio 1.62 vs. 1.48 per standard deviation; $$t = 9.51$$ ). In physical units, odds rose 11% per $${}^\circ$$ C and 11% per MJ/m $${}^2.$$ Holding apparent temperature fixed, moving from the tenth to the ninetieth percentile of radiation multiplied death-day odds by 4.8. Caregiver criminal liability showed no detectable effect in any coding or subset, precisely estimated in the pooled sample. Good Samaritan immunity was directionally protective only on the gained-access subset ( $$b = -0.81,$$ $$t = -2.59$$ ); placebo subsets (forgotten, knowingly left) were null. The estimate was design-sensitive and absent in aggregate death-certificate data. Conclusion The proximate exposure that kills children in closed vehicles is the solar-loaded cabin interior, supporting prevention keyed to cabin thermal load rather than ambient heat advisories alone. Criminal statutes show no detectable mortality benefit, and any benefit of bystander-rescue immunity is confined to the minority of cases where a bystander could intervene; death-only data can characterize exposure but bound what can be inferred about these statutes.