极低/超低出生体质量儿晚发型败血症风险预测模型的构建与验证

Construction of a risk prediction model for late-onset sepsis in infants with very/extremely low birth weight

目的:构建并验证极低/超低出生体质量儿(very/extremely low birth weight infants,VLBWI/ELBWI)发生晚发型败血症(late-onset sepsis,LOS)的风险预测模型。方法:选取2019年1月-2021年12月南京医科大学附属妇产医院新生儿重症监护室收治的579例VLBWI/ELBWI作为研究对象,其中发生LOS 93例为LOS组,未发生LOS 486例为对照组,分析两组发生LOS的影响因素,将单因素中P<0.05的因素纳入Logistic回归分析,建立风险预测模型,并应用ROC曲线检验模型的预测效果。结果:最终纳入出生体质量、外周静脉置管时间、禁食时间、早发型败血症(early-onset sepsis,EOS)发生史共4个因素构建出VLBWI/ELBWI发生LOS的风险预测模型,模型公式为Y=-2.349-0.003×出生体质量(g)+0.095×外周静脉置管时间(d)+0.198×禁食时间(d)-3.075×EOS发生史。预测模型ROC曲线下面积为0.860,约登指数最大值为0.597,预测敏感度为0.817,特异度为0.780,最佳临界值为0.159。结论:构建的模型预测性能良好,适用于临床VLBWI/ELBWI中LOS的筛查,可为医护人员尽早采取预防性措施提供理论依据。

Objective:To investigate the risk prediction model of late-onset sepsis(LOS)in very/extremely low birth weight infants(VLBWI/ELBWI).Methods:579 infants admitted to the Neonatal Intensive Care Unit of Women′s Hospital of Nanjing Medical University from January 2019 to December 2021 were selected,of which 93 infants with LOS occurrence were set as LOS group and 486 infants without LOS occurrence were set as control group.The influencing factors of the occurrence of LOS in two groups were analyzed.Factors with P<0.05 were included in Logistic regression analysis,risk prediction model was established,and ROC curve was used to test the prediction effect of the model.Results:Four factors including birth weight,duration of peripheral venous catheterization,fasting duration and history of early-onset sepsis(EOS)were included to construct the risk prediction model of LOS occurrence in VLBWI/ELBWI.The model formula was Y=-2.349-0.003×birth weight(g)+0.095×duration of peripheral venous catheter(d)+0.198×fasting duration(d)-3.075×EOS occurrence history.The area under ROC curve of the prediction model was 0.860,the maximum of the Jorden index was 0.597,the prediction sensitivity was 0.817,the specificity was 0.780,and the optimal critical value was 0.159.Conclusion:The constructed model has good prediction performance and can predict the occurrence of LOS in VLBWI/ELBWI.It can provide theoretical basis for medical staff to take early interventions as soon as possible.