Roof strength and injury risk in rollover crashes

Previous studies have reported that roof strength as measured in FMVSS 216 is unrelated to occupant injury risk in actual rollovers. However, those studies did not control for important confounding factors. The present study reevaluated roof strength and injury risk using a single class of vehicles with similar driver and vehicle usage characteristics in the expectation that some potential confounders will be controlled by this focused analysis. Alternative roof strength measures also were examined in case strength-to-weight ratio (SWR) at 5 inches, the measure in FMVSS 216, is not the best measure.

Eleven midsize SUV roof designs were crushed according to FMVSS 216 protocol. Force was measured continuously, and the maximum force and the amount of energy used to achieve crush of 2, 5, and 10 inches were recorded; these six measures of roof strength were evaluated both directly and after normalization for vehicle weight. After controlling for other potentially confounding variables, logistic regression models estimated the effect of each of the twelve roof strength measures on risk of fatal or incapacitating injuries to drivers in single-vehicle rollover crashes. Crash data were obtained from fourteen states.

All roof strength measures correlated inversely, and similarly, with risk of fatal/incapacitating injury in single-vehicle rollovers. Across all states, 12 percent of drivers sustained fatal/incapacitating injuries, but individual state rates varied from 6 to 24 percent. Among available variables, only state substantially affected the estimated effect of roof strength; it was kept in the final models with roof strength, driver age, and static stability factor (SSF). Driver age was retained because it had a significant effect on injury outcome and SSF was retained for theoretical reasons. This model estimated that a one-unit increase in strength-to-weight ratio (SWR) within 5 inches of crush was associated with 24 percent lower risk of fatal/incapacitating injury (95% CI: 15-33). Ejection was lower by 41 percent (95% CI: 30-51), and injury risk among nonejected occupants was lower by 16 percent (95% CI: 3-29).

The FMVSS 216 measure of roof strength related strongly to occupant protection in real-world rollover crashes in this study, which restricted analysis to vehicles with reasonably homogeneous driver and usage characteristics and accounted for state variation in injury rates. Further research should determine whether these results generalize to other vehicle types with different drivers and rollover risk. Research should also focus on factors other than roof strength, because even large increases in roof strength would leave many rollover crash deaths unaffected.