Construction of risk prediction model for death in patients with extremely severe burns

Li Yun; Zhang Ting; Ning Jinbin; Zhao Xu; Dong Yumin; Zhou Qin

doi:10.11986/j.issn.1673-873X.2024.12.18

Abstract：Objective To analyze the influencing factors of death in patients with extremely severe burns and to construct a risk prediction model.Methods The clinical data of 108 patients with extremely severe burns admitted from January 2021 to March 2024 were collected.According to the 90-day survival status after admission,the patients were divided into sur-vival group and death group.The demographic data,total body surface area(TBSA),sequential organ failure assessment score(SOFA)and platelet count at admission and at 3 days and 7 days after admission were compared between groups.Multivariate Logistic regression analysis was used to screen the independent influencing factors of death in patients with ex-tremely severe burns,and nomogram was adopted to construct the death prediction model of patients with extremely severe burns.Results The proportion of TBSA≥70%(84.21%vs 39.33%),proportion of third-degree burn area≥30%(68.42%vs 26.97%),mechanical ventilation rate(84.21%vs 35.96%)and proportion of SOFA≥2 points at 3 days after ad-mission(47.37%vs 25.84%)were higher in death group than those in survival group(P<0.05)while the proportion of normal platelet count at 7 days after admission(26.32%vs 67.42%)was lower than that in survival group(P<0.05).Lo-gistic regression forward stepping analysis showed that third-degree burn area≥30%,mechanical ventilation and SOFA score≥2 points at 3 days after admission were independent risk factors for death in patients with extremely severe burns(OR=5.342,5.298,22.794,P<0.05).Based on the above three independent risk factors,a risk nomogram early warn-ing model for death of patients with extremely severe burn was constructed,and the internal verification of the nomogram model was carried out by repeated sampling with Bootstrap method,and the calibration curve was close to the ideal curve,indicating good prediction accuracy.Receiver operating characteristic(ROC)curve showed that the area under cure(AUC)was 0.883(95%CI:0.784-0.982,P<0.05),indicating good discrimination.Conclusion Third-degree burn area,me-chanical ventilation and SOFA score at 3 days after admission are the influencing factors of death in patients with extremely severe burns.The prediction model constructed by the three factors is beneficial to evaluating the risk of death in patients.