1National Highway Transportation Safety Administration, Washington, DC
Candidate curves are developed, using cadaveric data, that express the probability of skull fracture as a function of HIC. The statistical method uses the concept of thresholds that are both right and left censored. This is done using SAS.
Some simple measures of goodness of fit are applied to each of these two-parameter models to select a "best" one. The lognormal and the Weibull are close and both are good by these criteria.
A reduction in HIC from 1,000 to 800 would result in an estimated reduction of 21.7 percent in the risk of skull fracture.