803例肺炎患者实验室指标特征及重症影响因素分析  

Analysis of laboratory index characteristics and influencing factors of severe illness in 803 patients with pneumonia

作　　者：张恒恒 崔晓静 陈玲玮 黄亮玲 陈杰 林文森 黄焕绍 邝兆威 符宏建 田育佼 郭主声 ZHANG Hengheng;CUI Xiaojing;CHEN Lingwei;HUANG Liangling;CHEN Jie;LIN Wensen;HUANG Huanshao;KUANG Zhaowei;FU Hongjian;TIAN Yujiao;GUO Zhusheng(Department of Laboratory Medicine,the First Dongguan Affiliated Hospital of Guangdong Medical University,Dongguan,Guangdong 523710,China;Department of Laboratory Medicine,the Southeast General Hospital of Dongguan,Dongguan,Guangdong 523710,China;The First Clinical College of Guangdong Medical University,Zhanjiang,Guangdong 524000,China)

机构地区：[1]广东医科大学附属东莞第一医院检验医学科,广东东莞523710 [2]东莞市东南部中心医院检验医学科,广东东莞523710 [3]广东医科大学第一临床学院,广东湛江524000

出　　处：《热带医学杂志》2024年第12期1711-1716,1746,共7页Journal of Tropical Medicine

基　　金：广东省基础与应用基础研究基金自然科学基金项目(2023A1515010938);2023年东莞市发展科技项目面上项目(20231800905052)。

摘　　要：目的探讨普通肺炎患者发展为重症肺炎的影响因素,以降低重症肺炎的发生率和死亡率。方法回顾性收集2021年11月-2024年5月在广东医科大学附属东莞第一医院住院的803例肺炎患者的临床资料,按照标准分为普通组(n=747)和重症组(n=56),统计分析患者的人口学特征以及实验室检测指标,并且构建二元Logistic回归模型和决策树模型,比较二元Logistic回归和决策树的预测能力。结果患者总体年龄分布差异有统计学意义(χ^(2)=81.033,P<0.001),重症组70岁及以上占比明显较高(37.5%),普通组中占比较低(5.6%)。重症组白细胞介素-6(IL⁃6)、D2聚体、降钙素原(PCT)、C-反应蛋白(CRP)、血清淀粉样蛋白A(SAA)、中性粒细胞、红细胞沉降率、可溶性生长刺激表达基因2蛋白(sST2)、是否好转方面显著高于普通组,差异均有统计学意义(P均<0.001)。二元Logistic回归结果显示,年龄、性别、D2聚体、PCT和红细胞沉降率是发生重症肺炎的危险因素(P均<0.05)。将一般资料中差异有统计学意义的变量纳入决策树模型中,最终得到一个包含4层、11个节点和3个预测因子的模型,包括PCT、D2聚体和IL⁃6;第1层的分裂特征为PCT,当PCT≤4.2 ng/mL时,患者被预测为普通肺炎的概率高达93%;第2层基于PCT进行分裂,当PCT≤1.4 ng/mL时,患者被预测为普通肺炎的概率进一步上升至99%;而当PCT>1.4 ng/mL时,患者被预测为普通肺炎的概率下降至65%,同时重症肺炎的概率则为35%;第3层在PCT>1.4 ng/mL的情况下,引入了D2聚体作为分裂特征;当D2聚体≤2.7μg/mL时,患者被预测为普通肺炎的概率为91%,而重症肺炎的概率仅为9%;最后一层的分裂特征为IL⁃6,该层是建立在PCT>4.2 ng/mL的前提下进行的,当IL⁃6>105 pg/mL时,患者发生重症肺炎的概率高达98%。根据普通肺炎转为重症肺炎的二元Logistic回归模型和决策树模型的受试者工作特征曲线结果显示,二元Logistic回归�Objective To explore the influencing factors of common pneumonia patients developing severe pneumonia in order to reduce the incidence and mortality of severe pneumonia.Methods The clinical data of 803 hospitalized patients with pneumonia in the First Dongguan Affiliated Hospital of Guangdong Medical University from November 2021 to May 2024 were retrospectively collected and divided into common group(n=747)and severe group(n=56)according to the standards.The demographic characteristics and laboratory test indicators of the patients were statistically analyzed,and a binary Logistic regression model and a decision tree model was constructed to compare the predictive abilities of binary Logistic regression and decision tree.Results There was a statistically significant difference in the overall age distribution of patients(χ^(2)=81.033,P<0.001).The proportion of patients aged 70 and above was significantly higher in the severe group(37.5%),while the proportion in the ordinary group was lower(5.6%).Interleukin⁃6(IL⁃6),D2 polymer,procalcitonin(PCT),C⁃reactive protein(CRP),serum amyloid A(SAA),neutrophils,erythrocyte sedimentation rate,soluble growth⁃stimulating expression gene 2 protein(sST2),and whether it was improved or not in the severe group were significantly higher than those in the common group(all P<0.001).The binary Logistic regression results showed that age,gender,D2 dimer,PCT,and erythrocyte sedimentation rate were risk for severe pneumonia(all P<0.05).Variables with statistically significant differences in the general information were incorporated into the decision tree model,and finally a model containing 4 layers,11 nodes and 3 predictors was obtained,including PCT,D2 polymer and IL⁃6;the splitting characteristics of the first layer was PCT.When PCT≤4.2 ng/mL,the probability of the patient being predicted to have ordinary pneumonia was as high as 93%;the second layer was split based on PCT.When PCT≤1.4 ng/mL,the probability of the patient being predicted to have ordinary pneumonia was f

关 键 词：重症肺炎 病原体 降钙素原 D2聚体 白细胞介素-6 

分 类 号：R563.1[医药卫生—呼吸系统]