Construction of a risk prediction model for Gram-positive bacterial sepsis in neonates
Objective To develop a risk prediction model for Gram-positive bacterial sepsis in neonates and to pre-dict the incidence risk of Gram-positive bacterial sepsis in newborns as early as possible.Methods The outcome vari-able was whether Gram-positive bacterial sepsis occurred or not.Data were collected from 149 neonates diagnosed with Gram-positive bacterial sepsis,including 26 clinical indicators(prenatal factors,vital signs and symptoms before infec-tion,and blood indices at the time of infection).The least absolute shrinkage and selection operator(Lasso)regression model and multivariate Logistic regression were used to identify risk factors for Gram-positive bacterial sepsis in neonates.Based on these selected risk factors,a risk prediction model was constructed and visualized as a nomogram.The calibra-tion,discrimination,and clinical utility of the nomogram were assessed using calibration curves,receiver operating charac-teristic(ROC)curves,and decision curve analysis(DCA).Internal validation of the nomogram was performed using the Bootstrap method.The 149 children with sepsis were categorized into low-risk(total score 0-73 points),medium-risk(to-tal score 74-83 points),high-risk(total score 84-95 points),and very high-risk(total score>95 points)groups based on the quartiles of the total score in the nomogram.The risk of developing Gram-positive bacterial sepsis among these groups was evaluated using one-way ANOVA to validate the predictive performance of the nomogram.Results Thirteen predic-tive variables were identified through Lasso regression,including neonatal respiration,gastric retention or abdominal dis-tension,blood glucose levels,body weight changes within 3 days before onset,gestational age,mode of delivery,maternal diabetes during pregnancy,late pregnancy infection,day of onset,duration of central venous catheterization before onset,duration of mechanical ventilation before onset,platelet count,and C-reactive protein(all P<0.05).Multivariate Logistic regression further identified five risk factors:rapid respiration,significant body weight changes within 3 days before onset(>80 g),shorter days of life at onset(<14 days),prolonged central venous catheterization before onset(15-30 days),and extended mechanical ventilation before onset(≥7 days)(all P<0.05).Based on these five risk factors,a nomogram was constructed.The nomogram demonstrated good calibration,with the area under the ROC curve of 0.77,sensitivity of 69.5%,and specificity of 77.8%.The decision curve showed that the nomogram was beneficial for predicting Gram-posi-tive bacterial sepsis within a threshold probability range of 18%-88%.Bootstrap validation indicated an accuracy of 69.5%and a consistency of 43.9%.The medium-risk,high-risk,and very high-risk groups showed significantly higher risk of de-veloping sepsis compared with that of the low-risk group,with OR values of 5.85,36.563,and 105.625,respectively(all P<0.05).Conclusion The nomogram,based on rapid respiration,significant body weight changes within 3 days before onset,prolonged central venous catheterization,extended mechanical ventilation,and shorter days of life at onset,effectively predicts the risk of Gram-positive bacterial sepsis in neonates,which has good calibration,discrimination,clini-cal applicability,and stability.
国家科技期刊平台
NSTL
万方数据
