儿童大环内酯类药物无反应性肺炎支原体肺炎预测模型构建  

A predictive model for macrolide unresponsive Mycoplasma pneumoniae pneumonia in children

作　　者：饶睿 李志鑫[2] 贾忠莉 李松 宋丽瑶[2] 董文斌 RAO Rui;LI Zhixin;JIA Zhongli;LI Song;SONG Liyao;DONG Wenbin(Department of Pediatrics,the Affiliated Hospital of Southwest Medical University,Luzhou,Sichuan 646000,China;Department of Pediatrics,the People's Hospital of Leshan,Leshan,Sichuan 614000,China)

机构地区：[1]西南医科大学附属医院儿科,四川泸州646000 [2]乐山市人民医院儿科,四川乐山614000

出　　处：《中国热带医学》2024年第7期783-790,共8页China Tropical Medicine

摘　　要：目的探讨儿童大环内酯类药物无反应性肺炎支原体肺炎(macrolide-unresponsive mycoplasma pneumoniae pneumonia,MUMPP)的危险因素,并构建MUMPP的预测模型。方法选择2023年3月1日—2023年12月1日乐山市人民医院儿科收治的符合纳入标准的肺炎支原体肺炎患儿为研究对象,按照对大环内酯类药物反应情况将患儿分为有反应组和无反应组。收集患儿一般资料、实验室检查及影像学资料进行比较,采用logistic回归分析影响肺炎支原体肺炎对大环内酯类药物反应的危险因素;R语言建立预测风险的列线图;受试者操作特征曲线下面积(area under curve,AUC)评价预测模型区分度;Hosmer-Leneshow检验拟合优度、校准曲线分析;决策曲线分析评价模型的临床净收益。结果共纳入224例患儿,将研究对象随机抽取156例(70%)作为训练集,68例(30%)作为验证集。logistic回归分析显示,有胸腔积液(OR=6.986,95%CI:1.362~35.847)、入院前最高体温(OR=3.095,95%CI:1.487~6.439)、中性粒细胞计数(OR=1.294,95%CI:1.103~1.519)、C反应蛋白(OR=1.030,95%CI:1.002~1.058)、降钙素原(OR=2.899,95%CI:1.353~6.214)是MUMPP的独立危险因素(均P<0.05)。使用R软件建立列线图,训练集和验证集的Hosmer-Lemeshow检验结果为χ^(2)=4.018,4.657(均P>0.05),提示模型拟合度良好。训练集和验证集的AUC分别为0.825(95%CI:0.755~0.894)、0.828(95%CI:0.729~0.928),提示模型区分度好。校准曲线分析提示模型预测效能良好,决策曲线分析提示预测模型临床应用价值较高。结论入院前最高体温、中性粒细胞计数、C反应蛋白和降钙素原升高及发生胸腔积液是MUMPP的独立危险因素,根据以上自变量构建的儿童MUMPP预测模型有较高的预测效能和临床应用价值。Objective To explore the risk factors for macrolide unresponsive mycoplasma pneumoniae pneumonia(MUMPP)in children and to develop a model for predicting the risk of MUMPP.Methods Children with mycoplasma pneumoniae pneumonia admitted to the Pediatric Department of Leshan People′s Hospital who met the inclusion criteria from March 1,2023,to December 1,2023,were retrospectively selected and divided into the responsive group and unresponsive group according to their reactions to macrolides.General patient data,laboratory tests,and imaging findings were collected and compared.Logistic regression analysis was used to analyze the risk factors of the Macrolide unresponsive mycoplasma pneumoniae pneumonia,and R language(R4.2.3)to establish the nomogram model.The goodness of fit,discriminative ability,calibration,and clinical utility of the model were assessed using the Hosmer-Lemeshow goodness of fit test,receiver operating characteristic(ROC)curve analysis,calibration curve analysis,and decision curve analysis,respectively.Results A total of 224 patients were included in the analysis of children.Among them,156(70%)were randomly selected as the training set,and the remaining 68 cases(30%)were used as the validation set.Logistic regression analysis revealed that pleural effusion(OR=6.986,95%CI 1.362-35.847),highest temperature before admission(OR=3.095,95%CI 1.487-6.439),neutrophil count(OR=1.294,95%CI:1.103-1.519),C-reactive protein(OR=1.030,95%CI 1.002-1.058),and procalcitonin(OR=2.899,95%CI:1.353-6.214)were independent risk factors for MUMPP in children(all P<0.05).A nomogram was established using R software.The Hosmer-Lemeshow goodness of fit tests for the training set and the validation set wereχ^(2)=4.018 andχ^(2)=4.657(all P>0.05),indicating a good fit of the model.The AUC values for the training set and validation were 0.825(95%CI:0.755-0.894)and 0.828(95%CI 0.729-0.928),respectively,suggesting good discriminative ability of the model.Calibration curve analysis suggested that the model had good predictive perf

关 键 词：肺炎支原体肺炎 大环内酯类无反应性肺炎支原体肺炎 预测模型 列线图 

分 类 号：R563.1[医药卫生—呼吸系统]