Construction and validation of a predictive model for patients with sepsis after congenital heart disease surgery

LUO Yu-ting; CHEN Chun-ling; GUO Yu-xiong

doi:10.19538/j.ek2024020609

Abstract：Objective To investigate the predictive factors of death in patients with sepsis after congenital heart disease surgery,and establish a nomogram prediction model.Methods The clinical data of children who developed sepsis after cardiopulmonary bypass thoracotomy at Guangdong Provincial People's Hospital between January 2012 and December 2021 were collected.Univariate and multivariate logistic regression analysis was conducted to identify the risk factors associated with mortality in children with sepsis after congenital heart disease.Based on the results of the analysis,a nomogram prediction model was developed and the Bootstrap resampling method was used for internal verification.Results A total of 157 children were included in the study,with 135 survivors and 22 deaths.Univariate analysis and multivariate logistic regression analysis revealed several independent risk factors for death in children with sepsis after congenital heart disease surgery.These factors included the total time of postoperative fever(OR=1.084,95％CI 1.022-1.151),asoactive-inotropic score on the third day after surgery(OR=1.162,95％CI 1.070-1.263),and secondary thoracotomy(OR=6.033,95％CI 1.906-19.098).Based on the above-mentioned three indepe-ndent risk factors,a nomogram prediction model was developed with an area under curve of 0.909(95％CI:0.852-0.966).The identification C index of the Bootstrap resampling method was 0.9139(95％CI 0.9100-0.9178).The Hosmer-Lemeshow goodness of fit test demonstrated that the nomogram model accurately predicted the risk of death in children with sepsis after surgery(x2=7.617,P=0.472).Conclusion The duration of postoperative fever,vasoactive-inotropic score on the third day after surgery,and secondary thoracotomy are independent risk factors for death of children with sepsis after congenital heart disease surgery.Based on the risk factors,the nomogram prediction model is successfully constructed and verified to predict the survival outcome of sepsis patients with congenital heart disease after surgery.