D. Ang1, S. Kurek1, M. McKenney1, E. Barquist1, E. Barquist1, S. Norwood1, B. Kimbrell1, D. Villarreal1, H. Liu1, M. Ziglar2, J. Hurst1 1University Of South Florida College Of Medicine,Tampa, FL, USA 2Hospital Corporation Of America,Nashville, TENNESSEE, USA
Introduction: Improving clinical outcomes of trauma patients is a challenging problem at a statewide level, particularly if data from the State’s registry is not publically available. Promotion of optimal care throughout the State is not possible unless clinical benchmarks are available for comparison. Using publically available administrative data from the State’s Department of Health and the Agency for Healthcare Research and Quality (AHRQ) Patient Safety Indicators (PSI), we sought to create a statewide method for benchmarking trauma mortality while also identifying a pattern of unique complications that have an independent influence on mortality.
Methods: Data for this study was obtained from State’s Agency for Health Care Administration (AHCA). Adult trauma patients were identified as having ICD-9 codes defined by the State. Stepwise logistic regression was used in order to create a parsimonious and predictive inpatient expected mortality model. The expected value of PSIs was created in a similar method using a multivariate model provided by the AHRQ. Case mix adjusted mortality results were reported as observed to expected ratios (O/E).
Results:There were 37,793 trauma patients evaluated during the study period. The overall fit of the expected mortality model was very strong at a c-statistic of 0.891. Eleven out of 25 trauma centers had O/E ratios less than one, or better than expected. Six statewide PSIs had O/E ratios higher than expected. The PSI which had the strongest influence on trauma mortality for the State was PSI# 4 or death among surgical Inpatients with serious treatable complications. Mortality could be further sub-stratified by complications at the hospital level.
Conclusion:This method offers an adjusted benchmarking method which screens at risk trauma centers in the State for higher than expected mortality. Stratifying mortality based on Patient Safety Indicators may identify areas of needed improvement at a statewide level.