Lower-Body Injury Rates in Full-Engagement Frontal Impacts: Field Data and Logistic Models

Lower-body injury data for adults in real-world frontal impacts in the National Automotive Sampling System (NASS) were collected, analyzed, and modeled via statistical methods. Two levels of lower-body injury were considered: maximum serious-to-fatal (MAIS3+) and moderate-to-fatal (MAIS2+). In the analysis, we observed that a substantial fraction of all lower-body injured occupants had no recorded floor/toe pan intrusion: 47% of all MAIS3+ injured occupants; 69% of all MAIS2+ injured occupants. In the statistical modeling, we developed binary logistic regression models to fit the MAIS3+ and MAIS 2+ injury data. The statistically significant variables (p ≤ 0.05) were the speed change of the crash, postcrash floor/toe pan intrusion, level of restraint, occupant age, and occupant gender. The two resulting models demonstrated different levels of fidelity: the Goodman-Kruskal Gamma statistic of association was 0.84 for the MAIS3+ model and 0.60 for the MAIS2+ model (where 1.0 is perfect association). However, the two statistical models identically predicted the total number of lower-body injured occupants in NASS. These analyses and statistical models provided insight into the effect of various crash variables on lower-body risks.