deaths due to injury (Trunkey and Lewis, 1991). These figures translate into nationwide annual estimates of approximately 20,000 to 25,000 lives saved (Champion and Teter, 1988). Nonetheless, studies of preventable deaths are beset by methodological limitations (Cales and Trunkey, 1985; Roy, 1987; Mendeloff and Cayten, 1991; MacKenzie et al., 1992). The purpose of this section is not to review comprehensively the peer-reviewed, published literature on patient outcomes with trauma systems, but rather to point to some illustrative studies, including more recent studies employing refined methodologies.
One noteworthy study of preventable mortality assessed the implementation of a regional trauma system in San Diego County, California (Shackford et al., 1986). The study found that after the implementation of a regional trauma system, the proportion of preventable fatalities fell from 13.6 to 2.7 percent. Such studies have been instrumental in stimulating the wider adoption of trauma systems.
Other approaches that are more objective than studies of preventable mortality have been developed to assess the benefits of trauma systems implementation. The Trauma and Injury Severity Score (TRISS) offers a means of predicting patient mortality based on injury severity, age, and revised trauma score (blood pressure, respiratory rate, and Glasgow Coma Scale for brain injury) (Boyd et al., 1987). With TRISS, the actual death rate in a hospital or trauma center can be compared to the predicted death rate from a large national data set of seriously injured trauma center patients voluntarily submitted to the Major Trauma Outcome Study (Champion et al., 1990). Champion and colleagues (1992) employed TRISS to assess longitudinally a reduction in trauma deaths in a center undergoing improvements from 1977 to 1982. They found an average of 13.4 more survivors per 100 seriously injured patients treated per year over the course of the improvements. Similarly, Stewart and colleagues (1995) found with TRISS that designation of a Canadian hospital as a trauma center led to a reduction in unexpected deaths from motor vehicle crashes from 8.8 percent before its designation to 3.6 percent after designation.
A newer method of evaluating outcomes related to trauma system implementation relies on population-based registries. The first study to capitalize on a comprehensive statewide population-based registry was performed by Mullins and coworkers (1994). They analyzed mortality outcomes among 70,350 patients who were hospitalized with injuries before and after institution of a trauma system in the Portland, Oregon metropolitan area. They found that the adjusted rate of mortality at Level I trauma centers was lowered by one-third compared to the pre-trauma system rate (the adjusted odds ratio for death declined from 1.00 before system establishment to 0.65 afterward). The impact was significant enough to be detected as an overall decline in the regional injury death rates, according to their analyses of two vital statistics databases. They also determined that the Portland metropolitan trauma system, through its prehospital triage criteria, was successful at shifting more seriously injured patients to trauma centers. Similar results were found when the authors broadened their analysis to cover five categories of injury across the entire state of Oregon before and after the implementation of a statewide trauma