Dane Scantling1, Justin Hatchimonji2, Elinore J Kaufman3, Aria Xiong4, Peter Yang5, Jason D Christie6, Patrick M Reilly7, Daniel N Holena8. 1. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: dane.scantling@pennmedicine.upenn.edu. 2. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: justin.hatchimonji@uphs.upenn.edu. 3. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: elinore.kaufman@pennmedicine.upenn.edu. 4. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: aria.xiong@pennmedicine.upenn.edu. 5. The University of Pennsylvania, Perelman School of Medicine, Department of Biostatistics, Epidemiology and Informatics, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: weiyang@pennmedicine.upenn.edu. 6. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: jason.christie@pennmedicine.upenn.edu. 7. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: patrick.reilly@uphs.upenn.edu. 8. The University of Pennsylvania, Division of Traumatology, Critical Care and Emergency, Philadelphia, PA, United States. Electronic address: daniel.holena@uphs.upenn.edu.
Abstract
INTRODUCTION: Cardiac complications (CC) after injury are rare but contribute disproportionately to mortality. Variability in rates of CC and failure to rescue (FTR) after CC (FTR-C) within trauma systems may suggest opportunities for improvement, but we have not yet demonstrated the ability to identify high and low performers. We examined center-level rates of CC and FTR-C in a mature trauma system with the hypothesis that high-performing centers for each of these outcomes could be identified. METHODS: Using a statewide trauma registry from 2007-2015, we developed multivariable logistic regression models on CC and FTR-C including patient demographics, physiology, comorbidity, and injury data. Predicted probabilities of each outcome were summed to generate expected event rates, which were compared to observed event rates to generate centerlevel observed-to-expected (O:E) ratios. We measured internal consistency between CC and FTR-C for centers using Cronbach's alpha. RESULTS: Cardiac complications occurred in 5,079/278,042 (1.8%; center-level range: 0.9-3.8%) of included patients (median age 55 (IQR 34-76), 84% Caucasian, 60% male, 92% blunt, median ISS 9 (IQR5-16)). Death after CC occurred in 982/5,097 patients for an FTR-C rate of 19.3% (center-level range: 7.8-30.4%). 10/27 centers were high-performers (95% confidence interval for O:E ratio <1) for CC and 2/27 centers were high-performers for FTR-C, but internal consistency between these metrics was poor (alpha = 0.31). CONCLUSION: Rates of CC and FTR-C vary significantly between hospitals in mature trauma systems but high-performing centers can be identified. Inconsistent performance between metrics suggests unknown institutional factors underlie performance for CC and FTR.
INTRODUCTION:Cardiac complications (CC) after injury are rare but contribute disproportionately to mortality. Variability in rates of CC and failure to rescue (FTR) after CC (FTR-C) within trauma systems may suggest opportunities for improvement, but we have not yet demonstrated the ability to identify high and low performers. We examined center-level rates of CC and FTR-C in a mature trauma system with the hypothesis that high-performing centers for each of these outcomes could be identified. METHODS: Using a statewide trauma registry from 2007-2015, we developed multivariable logistic regression models on CC and FTR-C including patient demographics, physiology, comorbidity, and injury data. Predicted probabilities of each outcome were summed to generate expected event rates, which were compared to observed event rates to generate centerlevel observed-to-expected (O:E) ratios. We measured internal consistency between CC and FTR-C for centers using Cronbach's alpha. RESULTS:Cardiac complications occurred in 5,079/278,042 (1.8%; center-level range: 0.9-3.8%) of included patients (median age 55 (IQR 34-76), 84% Caucasian, 60% male, 92% blunt, median ISS 9 (IQR5-16)). Death after CC occurred in 982/5,097 patients for an FTR-C rate of 19.3% (center-level range: 7.8-30.4%). 10/27 centers were high-performers (95% confidence interval for O:E ratio <1) for CC and 2/27 centers were high-performers for FTR-C, but internal consistency between these metrics was poor (alpha = 0.31). CONCLUSION: Rates of CC and FTR-C vary significantly between hospitals in mature trauma systems but high-performing centers can be identified. Inconsistent performance between metrics suggests unknown institutional factors underlie performance for CC and FTR.