G. Eckenrode1, E. J. Kaufman1, D. N. Holena2, C. C. Branas3, M. Narayan1, R. Winchell1 1Weill Cornell Medical College,Surgery,New York, NY, USA 2University Of Pennsylvania,Department Of Surgery,Philadelphia, PA, USA 3Columbia University,Mailman School Of Public Health,New York, NY, USA
Introduction:
Trauma center (TC) care has been shown to reduce mortality after injury by approximately 25%. The American College of Surgeons Committee on Trauma establishes criteria for TC designation, but little is known about differences among TCs. We hypothesized that the injured population would differ among centers, yielding fields of emphasis, and perhaps expertise, among U.S. TCs.
Methods:
We used 2014 data from the National Trauma Data Bank to analyze level I and II TCs. TCs were excluded if they had < 100 adult patients. For each TC, we calculated the proportion of adult patients falling into each of 10 categories of mechanism (penetrating injury, motor vehicle crash, pedestrian and bicyclist crash), injury type (splenic injury, blunt multisystem injury, fractures requiring fixation, traumatic brain injury, severe injury), and population (geriatric trauma and geriatric hip fractures (age ≥ 75)). These categories were drawn from the areas of interest determined by the Trauma Quality Improvement Program. We included an 11th variable for total number of traumas. TCs were classified as “high” in an area of interest if the proportion of that center’s patients falling into the area in question was in the top quartile of all TCs. We performed a cluster analysis to identify the relationship among these features. We computed a matrix of Jaccard measures of similarity between these variables. An average linkage function then clustered these variables and yielded a dendrogram of the results. The length of the bars shown on the dendrogram x-axis represents the degree of association among characteristics, with lower values indicating greater similarity among variables.
Results:
The 653 included TCs varied widely in the proportion of their patients falling into each area of interest. Median total traumas were 1,033 (interquartile range 535-1606). Median percent severe trauma (ISS ≥ 25) was 4.4% (2.4-6.6%). Median percent penetrating was 4.9% (3.1-7.6%). Median percent geriatric was 24.3% (16.2-32.5%). We observed several clusters of trauma center characteristics, as shown in the Figure. Centers with high proportions of geriatric trauma also tended to fall in the top quartile for geriatric hip fractures. Centers high in motor vehicle trauma tended to have high proportions of severe injuries, splenic injuries, and blunt multisystem injuries. Centers with high proportions of penetrating trauma also had high proportions of bicycle and pedestrian injuries.
Conclusion:
We identified clustering patterns amongst trauma center characteristics that suggest there are several distinct types of trauma centers in the United States. Further investigation is needed to identify variation in outcomes amongst trauma center types.
