新生儿呼吸机相关性肺炎发生的影响因素及列线图预测风险模型构建

Analysis of influencing factors for the occurrence of ventilator-associated pneumonia in newborns and construction of a column chart risk prediction model

目的探讨新生儿呼吸机相关性肺炎(VAP)发生的影响因素,并构建预测VAP的风险模型。方法回顾性分析2020年7月至2023年8月在本院行机械通气(MV)的391例新生儿的资料,根据是否发生VAP分成VAP组和对照组。对新生儿的一般资料及诊疗资料行单因素分析,利用Logistic回归分析筛查出VAP发生的相关因素,通过R软件中的rms包构建预测VAP的风险模型,并分别以ROC曲线下面积(AUC)、校准曲线检验模型的区分度和准确度。结果391例新生儿中,发生VAP 83例,VAP发生率为21.23%;83例VAP检出病原菌98株,其中革兰氏阳性菌26株(占26.53%),革兰氏阴性菌72株(占73.47%),检出率排在前3名的病菌为肺炎克雷伯菌(占29.59%)、大肠埃希菌(占17.35%)、阴沟肠杆菌(占13.27%);VAP组是否为早产儿、是否为低体重儿、出生时1min Apgar评分、白蛋白含量、插管次数、MV时间、吸痰方式、是否预防性应用抗生素和对照组比较有统计学差异(P<0.05);Logistic回归分析显示,新生儿VAP发生的危险因素主要有早产儿、低体重儿、出生时1min Apgar评分≤7分、插管次数≥2次、机械通气时间≥5d、开放式吸痰、未预防性应用抗生素等7项(P<0.05);模型验证显示,AUC为0.825(95%CI:0.762~0.888),校准曲线拟合较好,拟合优度H-L检验χ^(2)=6.713,P=0.584。结论根据是否为早产儿、是否为低体重儿、出生时1min Apgar评分、插管次数、机械通气时间、吸痰方式、是否预防性应用抗生素构建的列线图风险模型可对新生儿VAP发生起到良好的预测作用。

Objective To explore the influencing factors of neonatal ventilator-associated pneumonia(VAP)and construct a risk model for predicting VAP.Methods A retrospective analysis was conducted on the clinical data of 391 newborns who underwent mechanical ventilation(MV)in our hospital from July 2020 to August 2023.They were divided into the VAP group and the control group based on whether VAP occurred.The general information and diagnosis and treatment data of newborns were analyzed using single factor analysis.Logistic regression analysis was applied to screen the factors related to the occurrence of VAP,RMS package in R software was applied to construct a risk model for predicting VAP,area under the ROC curve(AUC)and calibration curve were applied to verify the model′s discrimination and accuracy.Results Among 391 newborns,83 cases developed VAP,with a VAP incidence of 21.23%.98 strains of pathogenic bacteria were detected in 83 cases of VAP,including 26 Gram positive bacteria(26.53%)and 72 Gram negative bacteria(73.47%).The top three pathogens with detection rates were Klebsiella pneumoniae(29.59%),Escherichia coli(17.35%),and Enterobacter cloacae(13.27%).The VAP group showed statistical differences compared to the control group in terms of whether they were premature infant,whether they were low birth weight infant,Apgar score at 1-minute of birth,albumin content,intubation frequency,MV time,sputum aspiration method,and whether prophylactic use of antibiotics(P<0.05).Logistic regression analysis showed that the main risk factors for the occurrence of VAP in newborns included premature infants,low birth weight infants,Apgar score≤7 points at 1 minute of birth,intubation frequency≥2 times,mechanical ventilation time≥5 days,open sputum aspiration,and prophylactic use of antibiotics(P<0.05).Model validation showed that the AUC was 0.825(95%CI:0.762~0.888),and the calibration curve fitted well,the goodness of fit H-L test wasχ^(2)=6.713,and P=0.584.Conclusion A column chart risk model constructed based on premature infant,low birth weight infant,Apgar score at 1-minute of birth,intubation frequency,mechanical ventilation time,sputum suction method,and prophylactic use of antibiotics can play a good predictive role in the occurrence of VAP in newborns.