Considerations of “Combined Probability of Injury” in the Next-Generation USA Frontal NCAP

Abstract

Objective: The numerical basis for assigning star ratings in the next-generation USA New Car Assessment Program (NCAP) for frontal impacts was assessed. That basis, the Combined Probability of Injury, or CPI, is the probability of an occupant sustaining an injury to any of the specified body regions. For an NCAP test, a CPI value is computed by (a) using risk curves to convert body-region responses from a test dummy into body-region risks and (b) using a theoretical, overarching CPI equation to convert those separate body-region risks into a single CPI value. Though the general concept of applying a CPI equation to assign star ratings has existed since 1994, there will be numerous changes to the 2011 frontal NCAP: there will be two additional body regions (n = 4 vs. 2), the injury probabilities will be evaluated for lower-severity (more likely) injury levels, and some of the occupant responses will change. These changes could yield more disperse CPIs that could yield more disperse ratings. However, the reasons for this increased dispersion should be consistent with real-world findings. Related assessments were the topic of this two-part study, focused on drivers. Methods: In Part 1, the CPI equation was assessed without applying risk curves. Specifically, field injury probabilities for the four body regions were used as inputs to the CPI equation, and the resulting equation-produced CPIs were compared with the field CPIs. In Part 2, subject to analyses of test dummy responses from recent NCAP tests, the effect of risk curve choice on CPIs was assessed. Specifically, dispersion statistics were compared for CPIs based on various underlying risk curves applied to data from 2001–2005 model year vehicles (n = 183). Results and Conclusions: From Part 1, the theoretical CPI equation for four body regions demonstrated acceptable fidelity when provided field injury rates (R2= 0.92), with the equation-based CPIs being approximately 12 percent lower than those of ideal correlation. From Part 2, the 2011 NCAP protocol (i.e., application of a four-body-region CPI equation whose inputs were from risk curves) generally increased both the CPIs and their dispersion relative to the current NCAP protocol. However, the CPIs generally increased due to an emphasis on neck injury—an emphasis not observed in real-world crashes. Subject to alternative risk curves for the neck and chest, again there was increased dispersion of the CPIs, but the unrealistic emphasis on the neck was eliminated. However, risk estimates for the knee/thigh/hip (KTH) for NCAP-type events remained understated and did not fall within the confidence bands of the field data. Accordingly, KTH risk estimation is an area for future research.