Proposed Objective and Approach: To determine and evaluate components of Emergency Medical Services (EMS) and trauma systems that contribute to improved outcomes for acutely injured persons, using public data and survival-time statistical models. Importance: Injuries are a substantial cause of death and disability, some of which may be prevented by effective EMS and trauma systems. However, evaluation of these systems is complicated by the concentration of medical resources in urban areas, and possibly unavoidable delays in initiating therapy due to geographic disparities. To overcome this methodological problem, we propose using survival-time models that explicitly adjust for the length of time between injury, intervention, and death. Data sources are now available that allow us to apply these innovative methods to make appropriate comparisons among EMS systems, their specific components, and other potentially modifiable factors.
Specific Aims and Hypotheses: 1) To construct survival curves for subjects after traffic crashes or intentional injuries, using public data and standard methods of survival-time analysis;2) To determine the overall effects of EMS and hospital intervention on these curves and models;and 3) After controlling for identifiable fixed factors, to determine how much the effects of EMS and hospital intervention vary among specific trauma systems, and use this residual variability as a measure of individual system performance. Study Design: We will use survival-time modeling techniques to analyze existing data from fatal traffic crashes, homicides, and suicides. Data on time to death following injury will be obtained from the National Vital Statistics System (death certificates), National Highway Traffic Safety Administration Fatality Analysis Reporting System (NHTSA/FARS, a census of fatal crashes), and National Violent Death Reporting System (a census of homicides and suicides). Corresponding populations at risk for death following similar injuries will be identified using FARS (which includes the other persons involved in a fatal crash), the NHTSA State Data System (police crash reports), or State Inpatient Databases (hospitalizations). The onset of EMS or hospital treatment will each be added to standard proportional hazards survival-time models as a time-varying covariate (0 before, 1 after), and we expect to show a decrease in the hazard with EMS and a further decrease with hospital treatment. This will confirm the importance of minimizing response time and prehospital time, and help explain the observed association between rural location and mortality after injury. Finally, each state will be incorporated as a cofactor in the models, allowing us to estimate the residual effect of each specific trauma system, after accounting for the disparities in their geographic settings. These models will be further refined by including the known structure of EMS and trauma systems in the corresponding regions and estimating the effects of specific components. Setting and Participants: Our study population will include injured persons during the two most recently available years from the 25 states that have data in at least four of the five public databases cited above (or the National EMS Information System currently under development). Interventions: None. Outcome Measures: The primary outcomes are survival and time elapsed from injury to death. We will also estimate the effects of EMS and hospital treatment, general characteristics of EMS and trauma systems, and effects of specific regional systems, using coefficients derived from survival-time models.
This project presents a new opportunity to assemble data already available and apply innovative statistical methods to evaluate trauma systems. This will enable us to distinguish between the impact of level of trauma system development and other factors affecting the disparity in rural and urban trauma mortality.
|Clark, David E; Doolittle, Peter C; Winchell, Robert J et al. (2014) The effect of hospital care on early survival after penetrating trauma. Inj Epidemiol 1:24|
|Clark, David E; Hannan, Edward L (2013) Inverse propensity weighting to adjust for bias in fatal crash samples. Accid Anal Prev 50:1244-51|
|Clark, David E; Winchell, Robert J; Betensky, Rebecca A (2013) Estimating the effect of emergency care on early survival after traffic crashes. Accid Anal Prev 60:141-7|
|Clark, David E; Qian, Jing; Sihler, Kristen C et al. (2012) The distribution of survival times after injury. World J Surg 36:1562-70|
|Clark, David E; Qian, Jing; Winchell, Robert J et al. (2012) Hazard regression models of early mortality in trauma centers. J Am Coll Surg 215:841-9|