Designing a primary model for estimating adjusted death according to severity of injury in three specialized trauma centers

Abstract

Background: Trauma is a major health problem in the world. There are different scoring scales for predicting deaths due to trauma in the world, but despite the high number of deaths due to trauma in Iran, there is no local scale to predict predictable deaths due to trauma in Iran. This study aimed to design a primary model for estimating adjusted death, according to the severity of injury in three specialized trauma centers. Materials and Methods: This study was a prospective study. First, univariate logistic regression was used to identify variables that have a significant relationship with trauma death. After identifying variables with a significant relationship with death due to trauma, multiple binary logistic regression was used for modeling. Variables that were significant in the univariate logistic regression stage at the level of significance (p-value) of 0.1 were included in the models and modeling was performed at the significance level of 0.05. Result: A total of 929 patients participated in this study. Finally, 91 patients (9.79 %) died by one-month follow-up. The AUC value for the Triage Model (TM) model on the stage of model making, external validity evaluation and among all patients was 0.797, 0.850 and 0.801, respectively. AUC value in the modeling phase, external validity review and in all patients for GCS and SpO2 (GCSO2) model equal to 0.875, 0.960 and 0.940, respectively, for GAP and SpO2 (GAPO2) model respectively 0.876, 0.950 and 0.943, for Trauma Prognostic Scale with SpO2 (TPS-O2) equal to 0.947, 0.946 and 0.976 and for Trauma Prognostic Scale with FIO2 (TPS-FIO2) they were 0.946, 0.998 and 0.981. These models had external validity. Conclusion: It can be concluded that the physiological model GAPO2 can be used for triage of trauma patients in the emergency department of Iranian hospitals and the combined model TPS-O2 and TPS-FIO2 can be used for the most accurate evaluation of patients. We recommend using the TPS-O2 model for non-intubated patients and the TPS-FIO2 model for intubated patients. Since this study is only a preliminary study, it is necessary to conduct more extensive studies in the country to finally create a national model or models for predicting predictable deaths due to trauma in Iran. Keywords: National model, Trauma, Prognostic, Death, Validity