新生儿医院感染危险因素的风险预测模型构建

Construction of risk prediction model for neonatal nosocomial infection risk factors

目的分析新生儿医院感染的危险因素并构建风险预测模型。方法收集中国医科大学附属盛京医院2023年新生儿病房住院治疗患儿病历数据,根据是否发生医院感染分为感染组和非感染组,采用单因素分析比较两组间相关因素的差异,采用Logistic多因素回归分析筛选新生儿医院感染独立危险因素并构建风险预测模型,用受试者工作特征曲线分析模型的预测价值。结果共纳入3487例新生儿,其中男1881例,女1606例;中位胎龄35(32,38)周,中位出生体重2470(1760,3200)g;感染组158例,非感染组3329例。新生儿医院感染的主要病种为新生儿败血症(142例,85.5%)、中心静脉导管相关感染(13例,7.8%)、呼吸道感染(8例,4.8%),主要病原菌为肺炎克雷伯菌(50株,29.8%)、大肠埃希菌(26株,15.5%)、表皮葡萄球菌(25株,14.9%)。新生儿医院感染的独立危险因素为低出生体重(OR=0.999,95%CI 0.999~1.000,P=0.039)、中心静脉导管置入(OR=2.122,95%CI 1.359~3.315,P=0.001)、气管插管(OR=3.241,95%CI 1.811~5.798,P<0.001)、进行肠外营养(OR=11.087,95%CI 3.696-33.258,P<0.001)。以此为基础构建的风险预测模型回归方程为Logit(P)=-47.407-0.001×出生体重+0.752×中心静脉导管置入+1.176×气管插管+2.406×肠外营养,受试者工作特征曲线下面积为0.870,灵敏度为89.2%,特异度为76.0%。结论低出生体重、中心静脉导管置入、气管插管、进行肠外营养4个独立危险因素构建的风险预测模型,对新生儿医院感染具有较好的预测性,有一定的临床应用价值。

ObjectiveTo analyze the risk factors of neonatal nosocomial infection and establish a risk prediction model.MethodsThe medical records of patients in the neonatal ward of Shengjing Hospital of China Medical University in 2023 were collected and divided into infection group and non-infection group according to whether nosocomial infection occurred.Univariate analysis was used to compare the differences of related factors between two groups.Multivariate Logistic regression analysis was used to screen the independent risk factors of neonatal nosocomial infection and construct the risk prediction model.The predictive value of the model was analyzed by receiver operating characteristic curve.ResultsA total of 3487 neonates were included in the study,including 1881 males and 1606 females.The median gestational age was 35(32,38)weeks,and the median birth weight was 2470(1760,3200)g.There were 158 cases in infection group and 3329 cases in non-infection group.The main types of nosocomial infection were neonatal septicemia(142 cases,85.5%),central venous catheter-related infection(13 cases,7.8%)and respiratory tract infection(eight cases,4.8%).The main pathogens were Klebsiella pneumoniae(50 strains,29.8%),Escherichia coli(26 strains,15.5%),and Staphylococcus epidermidis(25 strains,14.9%).The independent risk factors of neonatal nosocomial infection were low birth weight(OR=0.999,95%CI 0.999-1.000,P=0.039),central venous catheter placement(OR=2.122,95%CI 1.359-3.315,P=0.001),tracheal intubation(OR=3.241,95%CI 1.811-5.798,P<0.001),and parenteral nutrition(OR=11.087,95%CI 3.696-33.258,P<0.001).The regression equation of the established risk prediction model was Logit(P)=-47.407-0.001×birth weight+0.752×central venous catheter placement+1.176×tracheal intubation+2.406×parenteral nutrition.The area under the receiver operating characteristic curve was 0.870,the sensitivity was 89.2%,and the specificity was 76.0%,respectively.ConclusionThe risk prediction model of low birth weight,central venous catheter placement,tracheal intubation and parenteral nutrition is a good predictor of neonatal nosocomial infection and has certain clinical application value.