Evaluation of independent predictors of in-hospital mortality in patients with severe trauma

Introduction/Objective The aim of this study was to determine independent predictors and the best trauma scoring system (REMS, RTS, GSC, SOFA, APPACHE II) of in-hospital mortality in patients with severe trauma at the Department of Emergency, Emergency Center, Clinical Center of Serbia, Belgrade. Methods Longitudinal study included 208 consecutive patients with severe trauma. In order to determine independent survival contributors, univariate and multivariate Cox regression analyses were performed. The power of above-mentioned scoring systems (measured at admission to the Emergency center) to predict mortality was compared using the area under the curve (AUC). Results There were 208 patients (159 male, 49 female), with the average age of 47.3 +/- 20.7 years. Majority of patients were initially intubated (86.1%) on admission to the emergency department, and 59.6% patients were sedated before intubation. After finishing of diagnostic procedures, 17 patients were additionally intubated, and, at that time, 94.2% patients were on mechanic ventilation. The majority of patients was traumatized in a car crash (33.2%), followed by falls from height (26.4%) and as pedestrians (22.6%). Patients had an average of 24.7 +/- 21.2 days spent in intensive care unit. The overall case-fatality ratio was 17/208 (8.2%). In Cox regression analysis only elevated heart rate (HR = 1.03, p = 0.012) and decreased arterial oxygen saturation (SpO(2)) (HR = 0.91, p = 0.033) singled out as independent contributors to in-hospital mortality of patients with severe trauma. REMS (AUC 0.72 +/- 0.64) and SOFA (AUC 0.716 +/- 0.067) scores were found fair and similar predictor of in-hospital mortality, while APACHE II (AUC 0.614 +/- 0.062) and RTS (0.396 +/- 0.068) were poor predictors. Conclusion Results of this study showed an important role of REMS, which appears to provide balance between the predictive ability and the practical application, and components of REMS in prediction of outcome in patients with severe trauma and that HR and SpO(2) are independent predictors of in-hospital mortality.