Risk factors of Traumatic Myocardial Contusion and Establishment of Nomogram Prediction Model

Abstract

Abstract Objectives To clarify the risk factors of traumatic myocardial contusion (MC) and to build an MC prediction model and assess its clinical application values. Methods The clinical data of 370 rib fracture patients treated upon emergency call in the Department of Thoracic Surgery at Jiangsu Shengze Hospital Affiliated to Nanjing Medical University between January 2017 and December 2019 were retrospectively analyzed. Of them, 159 patients were diagnosed as MC. All patients were fully randomly divided at a 7:3 ratio to a training cohort (264 cases, 106 MC cases, 158 NMC cases) and a validation cohort (106 cases, 53 MC cases, 53 NMC cases). The related risk factors of MC in the training cohort were identified via univariate Logistics regression, then the optimal independent risk factors were screened out using LASSO regression and multivariate Logistics regression. A Nomogram model for MC prediction in the training cohort was built with the selected independent risk factors. The receiver's operating characteristic (ROC) curves and calibration curves in the two cohorts were plotted and used to analyze the prediction efficacy of the Nomogram model, and the clinical application value of the model was assessed using decision curve analysis (DCA) and clinical impact curves (CIC). Results The multivariate Logistics regression and LASSO regression analysis showed there were five independent risk factors of MC, including the upper chest anterolateral segment (UAL), the middle chest proximal spinal segment (MSS), sternal fracture (SF), Pneumothorax and aspartic transaminase (AST). The ROC curves showed the Nomogram model based on the C index had discrimination of 0.838 (95%CI, 0.790–0.886) and 0.846 (95%CI, 0.770–0.921) in the training cohort and the validation cohort respectively. The calibration curves showed there was high predictive precision between the actual probability and predicted probability in both cohorts. DCA showed at threshold probability > 0.1, the Nomogram model had significant clinical net benefits in both cohorts. CIC showed at the threshold probability > 0.6, the predicted number of positive patients was basically consistent with the actual number. Conclusions SF, UAL, MSS, Pneumothorax and AST are the independent risk factors and predictors of MC in rib fracture patients. The Nomogram model based on the 5 independent risk factors has high discrimination, calibration and clinical net benefits, and shows extensive prospects for clinical application in basic hospitals.