个性化预测新生儿坏死性小肠结肠炎发病风险的列线图模型构建

目的 探究个性化预测新生儿坏死性小肠结肠炎(NEC)发病风险的列线图模型构建,为新生儿NEC的防治提供科学依据。方法 回顾性分析2017年10月—2019年12月海南现代妇女儿童医院与外院联合收集的1 173例新生儿临床资料,根据是否发生NEC分为NEC组(n=46)与非NEC组(n=1 127),采用Logistic回归模型分析影响NEC发生的独立危险因素,基于筛选出的独立危险因素利用R软件构建NEC发生风险的列线图模型,应用ROC曲线下面积检验模型预测效果并进行拟合优度验证。结果 新生儿NEC发生率为3.92%(46/1 173);多因素Logistic回归分析显示,胎龄＜32周(OR=3.186)、出生体重＜1.5 kg(OR=2.520)、合并感染性休克(OR=2.517)、合并败血症(OR=2.566)、合并妊娠期糖尿病(OR=1.973)、人工喂养(OR=2.267)为影响新生儿NEC发生的独立危险因素(P<0.05),产前使用地塞米松为保护性因素(OR=0.475,P＜0.05);该预测模型ROC曲线下面积为0.741;Hosmer-Lemeshow 拟合优度检验χ²=7.859,P=0.447;绘制列线图的校准曲线为斜率接近1的直线。结论 个性化预测新生儿NEC发病风险的列线图模型具有良好的区分度与准确度,可有效评估新生儿NEC的发生概率,能够为新生儿NEC的防治提供一定指导价值。

Construction of nomogram model for personalized prediction of neonatal necrotizing enterocolitis

Objective To explore the construction of nomogram model for personalized prediction of neonatal necrotizing enterocolitis (NEC) risk,in order to provide scientific reference for the treatment and prevention of neonatal NEC. Methods The clinical data of 1 173 newborns delivered in Hainan Modern Women's and Children's Hospital and the outer hospital from October 2017 to December 2019 were retrospectively collected and analyzed,and were divided into NEC group(n=46) and non-NEC group(n=1 127). Logistic regression model was used to analyze the independent risk factors of NEC. Based on the selected independent risk factors,the nomogram model of NEC risk was constructed by R software,the area under ROC curve was used to test the prediction effect of the model,and the fit goodness was verified. Results The incidence of neonatal NEC was 3.92% (46/1 173). Multivariate Logistic regression analysis showed that gestational age ＜32 weeks(OR=3.186),birth weight ＜1.5 kg (OR=2.250),septic shock(OR=2.517),septicemia(OR=2.566),gestational diabetes mellitus(OR=1.973) and artificial feeding(OR=2.267) were independent risk factors of neonatal NEC(P<0.05),and prenatal use of dexamethasone was a protective factor (OR=0.475,P＜0.05). The area under ROC curve in this prediction model was 0.741. Hosmer-Lemeshow goodness-of-fit test showed good fit (χ²=7.859,P=0.447). The calibration curve of nomogram was a straight line with slope close to 1. Conclusions The nomogram model for personalized predicting the risk of neonatal NEC has good discrimination and accuracy. It can effectively evaluate the occurrence probability of neonatal NEC,and can provide certain guidance value for the prevention and treatment of neonatal NEC.