儿童肺炎支原体坏死性肺炎的危险因素及预测模型  

Risk factors and the prediction model of necrotizing pneumonia in children with Mycoplasma pneumoniae pneumonia

作　　者：罗娟 陈鹏[2] 郭宏溪 丁娟娟 Luo Juan;Chen Peng;Guo Hongxi;Ding Juanjuan(Department of Children's Medical Ward,Wuhan Children's Hospital(Wuhan Maternal and Child Healthcare Hospital),Tongji Medical College,Huazhong University of Science&Technology,Wuhan 430015,China;Department of Children's Respiratory Medicine,Wuhan Children's Hospital(Wuhan Maternal and Child Healthcare Hospital),Tongji Medical College,Huazhong University of Science&Technology,Wuhan 430015,China;Department of General Surgery,Wuhan Children's Hospital(Wuhan Maternal and Child Healthcare Hospital),Tongji Medical College,Huazhong University of Science&Technology,Wuhan 430015,China)

机构地区：[1]华中科技大学同济医学院附属武汉儿童医院(武汉市妇幼保健院)综合内科病区,武汉430015 [2]华中科技大学同济医学院附属武汉儿童医院(武汉市妇幼保健院)呼吸内科,武汉430015 [3]华中科技大学同济医学院附属武汉儿童医院(武汉市妇幼保健院)普外科,武汉430015

出　　处：《中华实用儿科临床杂志》2025年第3期187-193,共7页Chinese Journal of Applied Clinical Pediatrics

基　　金：武汉儿童医院临床研究人才培养项目(2023WHET-TFJH)。

摘　　要：目的分析儿童肺炎支原体肺炎(MPP)中坏死性肺炎(NP)的早期危险因素,并建立预测模型。方法病例对照研究。回顾性分析2021年1月至2024年5月在华中科技大学同济医学院附属武汉儿童医院住院的MPP患儿的临床资料。根据是否发生肺坏死将患儿分为坏死性肺炎组(NP组)和非坏死性肺炎组(NNP组),NP组共纳入62例患儿,按照最近邻匹配法1∶2比例进行倾向性评分匹配(卡钳值为0.02),124例患儿被纳入NNP组。采用LASSO回归选择最优因素,采用多因素Logistic回归分析建立临床预测模型,随后对模型进行内部及外部验证。采用受试者工作特征(ROC)曲线和校准曲线评估预测模型的预测能力和校准度。采用临床决策曲线分析(DCA)来评价其临床预测价值。结果LASSO回归分析显示,白细胞、中性粒细胞百分比、C-反应蛋白、降钙素原、D-二聚体、铁蛋白、热程和肺实变是MPP患儿发生NP的影响因素(P<0.05)。ROC曲线分析结果显示,本研究建立的预测肺炎支原体坏死性肺炎(MPNP)的预测模型在训练集的曲线下面积(AUC)为0.838(95%CI:0.765~0.911,P<0.001),在验证集的AUC为0.834(95%CI:0.755~0.913,P<0.001),外部验证集AUC为0.924(95%CI:0.902~0.981,P<0.001)。采用Bootstrap重复采样1000次进行内部验证,校准曲线显示,该模型具有良好的一致性。临床DCA表明,该模型具有良好的临床应用价值。结论白细胞、C-反应蛋白、D-二聚体、铁蛋白、热程、肺实变对MPNP早期预测有较好的价值。Objective:To analyze the early risk factors of necrotizing pneumonia(NP)in children with Mycoplasma pneumoniae pneumonia(MPP)and construct a clinical prediction model.Methods:In this case-control study,the clinical data of MPP patients who were hospitalized at Wuhan Children′s Hospital,Tongji Medical College,Huazhong University of Science&Technology,from January 2021 to May 2024 were retrospectively analyzed.According to whether NP occurred,the children were divided into the NP group and the non-NP(NNP)group.A total of 62 and 124 children were included in the NP and NNP groups after nearest neighbor matching at a ratio of 1∶2(with a caliper value of 0.02),respectively.LASSO regression was used to select the optimal factors,and the multivariate Logistic regression analysis was used to establish a clinical prediction model.Internal and external validation of the prediction model was then conducted.The receiver-operating characteristic(ROC)curve and calibration curve were used to evaluate the predictive ability and calibration of the prediction model.The clinical decision curve analysis(DCA)was used to evaluate its clinical predictive value.Results:The LASSO regression analysis showed that white blood cells(WBC),neutrophil percentage,C-reactive protein(CRP),procalcitonin,D-dimer,ferritin,fever duration,and lung consolidation were factors influencing the occurrence of NP in children with MPP(P<0.05).The ROC analysis showed that the area under the curve(AUC)of the prediction model was 0.838(95%CI:0.765-0.911,P<0.001)in the training set,0.834(95%CI:0.755-0.913,P<0.001)in the validation set,and 0.924(95%CI:0.902-0.981,P<0.001)in the external validation set.Bootstrap was used for repeated sampling for 1000 times for internal validation,and the calibration curve showed that the model had good early consistency.The clinical DCA showed that the model had good clinical application value.Conclusions:WBC,CRP,D-dimer,ferritin,fever duration and lung consolidation have good value for the early prediction of MPNP in children.

关 键 词：儿童 肺炎支原体 坏死性肺炎 危险因素 列线图模型 

分 类 号：R72[医药卫生—儿科]