| Objective:Sepsis is a major cause of death in children.Early identification of children with sepsis at high risk of death plays an important role in clinical decision-making and prognosis.In this study,we analyze the risk factors of death in children with sepsis and establish a Nomogram model to predict the risk of death within 28 days in children with sepsis,which may help clinicians to evaluate the risk of early death in time and take active treatment measures to improve the cure rate.Methods:Clinical data of children with sepsis under 14 years old admitted to the Children’s Medical Research Center of the First Affiliated Hospital of Gannan Medical University from January 1st,2018 to November 30 th,2022 were retrospectively collected.According to the time of admission,the patients were divided into the modeling group and the verification group,including 421 children with sepsis from January 1st,2018 to October 31 st,2021 as the modeling group,and195 children with sepsis from November 1st,2021 to November 30 th,2022 as the verification group.Children with sepsis were divided into survival group and death group according to the death event within 28 days.Univariate and multivariate analysis were used to screen out the risk factors for 28-day mortality in children with sepsis,and R software was used to construct the mortality risk model of children with sepsis.Conduct internal and external validation on the modeling group data and validation group data respectively.The discrimination was evaluated by the area under curve(AUC)of the receiver operating characteristic curve(ROC).Hosmer-Lemeshow test and calibration chart was used to evaluate calibration.The clinical applicability of the model was evaluated by decision curve analysis(DCA).Results:(1)A total of 616 children with sepsis were enrolled in this study,including 421 children in the modeling group,10.7% of whom died within 28 days,and 195 children in the validation group,8.2% of whom died within 28 days.(2)There were 555 septic children in the overall survival group and 61 in the total death group.A total of 37 variables were analyzed in the two groups.The study showed that there were no statistically significant differences in neutrophil to lymphocyte ratio,serum sodium,neutrophil count,body temperature,neutrophil to prealbumin ratio,C-reactive protein and platelet crit between the two groups(P >0.05),while other variables were statistically significant(P < 0.05).(3)According to the outcome events,the samples in the modeling group were divided into the survival group(n=376)and the death group(n=45).Age,heart rate,prealbumin,platelet count,serum calcium and lactate dehydrogenase were screened out as independent risk factors by univariate analysis and multivariate Logistic regression analysis.At the same time,a Nomogram model was established to predict the risk of death in children with sepsis within 28 days.(4)Nomogram models showed good discrimination [AUC=0.891,95%confidence interval(95%CI),0.836-0.946],calibration(Hosmer-Lemeshow test,P=0.493),and clinical applicability.(5)The validation group model also had good discrimination(AUC=0.823,95%CI 0.699-0.946),calibration(Hosmer-Lemeshow test,P=0.118),and clinical applicability.Conclusion:(1)This study found that high age group,increased heart rate,decreased levels of platelet,serum calcium and prealbumin,and increased levels of lactate dehydrogenase would increase the risk of death in children with sepsis.In other words,age,heart rate,prealbumin,platelet count,serum calcium and lactate dehydrogenase were independent risk factors for death in children with sepsis within28 days.(2)Based on 6 independent risk factors,we established a Nomogram model for predicting 28-day mortality in children with sepsis to evaluate short-term prognosis.(3)After internal and external verification,the mortality risk model in this study had good discrimination,calibration and clinical applicability,which could provide certain reference significance for clinical decision-making. |
PDF Full Text Request