A ranking system is established whereby injury codes in the National Automotive Sampling System (NASS) are ordered by survivability based on actual mortality rates. Special provisions are made for cases in which injuries are coded as "not further specified" and "severity unknown." Once the ranking system is established, an injury analysis is carried out in which NASS crash victims are characterized by their two highest-ranking injuries. Then, each victim's probability of survival is estimated using a new "primary/secondary" fatality prediction procedure. When deviance statistics are considered, the new procedure predicts fatalities better than the Injury Severity Score, a commonly applied metric that is based on the Abbreviated Injury Scale. Ultimately, the new rankings - which single out specific injuries - provide a means to improve benefits analyses used to support crash injury research.
Abstract