Development and performance evaluation of a clinical prediction model for sepsis risk in burn patients.

Sepsis is a common and severe complication in burn patients and remains one of the leading causes of mortality. This retrospective study aimed to develop a predictive model for the risk of in-hospital sepsis among burn patients treated at Guangzhou Red Cross Hospital between January 2022 and January 2024, with the goal of improving clinical outcomes through early prevention based on risk stratification. A total of 302 eligible patients were randomly divided into training and validation cohorts in a 7:3 ratio for model development and validation, respectively. Predictive factors were initially selected using LASSO regression, followed by logistic regression analysis to establish the prediction model and construct a nomogram. The final model incorporated 4 independent predictors: burn area (odds ratio [OR] = 1.043, 95% confidence interval [CI]: 1.026-1.062/1%), hemoglobin (OR = 0.968, 95% CI: 0.954-0.980/1 g/L), diabetes (OR = 10.91, 95% CI: 2.563-56.62), and potassium (OR = 3.091, 95% CI: 1.635-6.064/1 mmol/L). The areas under the receiver operating characteristic curve were 0.875 and 0.861 for the training and validation cohorts, with Youden indexes of 0.634 and 0.600, respectively. The calibration curve and decision curve analysis demonstrated good predictive accuracy and clinical utility of the model. These findings suggest that our developed model exhibits robust predictive performance for the risk of in-hospital sepsis in burn patients, and early prevention strategies based on risk stratification may potentially improve clinical outcomes.