Predicting Adverse Events in Blunt Chest Trauma: A Novel Nomogram Integrating Vitals, Hemogram, and Comorbidities.

Abstract

Blunt chest trauma (BCT) is common and frequently associated with adverse complications. Beyond merely impeding regular respiration, adverse events (AEs) such as hemothorax or pneumothorax can hinder the patient's recovery. Herein, we aim to validate potential predictive factors for AEs among adults with BCT who were admitted concurrently through the dataset focusing on the limited information available upon their arrival at the emergency department (ED). Seventeen variables-including patients' demographics, comorbidities, and vital signs/hemogram data upon arrival at the ED-were investigated. A penalized logistic regression model was applied to the derivation cohort and validated in a subgroup using the same dataset (80%:20%). In addition, we employed the least absolute shrinkage and selection operator (LASSO) logistic regression to develop a nomogram, which enhances the accuracy of estimating individual probabilities for AEs after admission for BCT. Our retrospective review encompassed 3,668 adult patients between 2017 and 2021, and the incidence of AEs was 15.6% (572 out of 3,668). Penalized logistic regression was conducted both without and with the hemogram data (Model 1 and Model 2), yielding relatively satisfactory results (R²: 0.271 vs. 0.291; area under the curve: 0.784 vs. 0.797, respectively). Despite the model's relatively high predictive value in the derivation cohort, the validation data still maintained an acceptable accuracy of 0.7456 and 0.7049, respectively. Employing our penalized logistic regression analysis, the recently formulated nomogram exhibited proficiency in predicting AEs following BCT. This effectiveness was achieved by integrating vital signs, hemogram data, and comorbidities recorded upon their arrival at the ED.