基于危急值构建慢性阻塞性肺疾病急性加重患者死亡风险的预测模型  

Constructing a prediction model of death risk of patients with acute exacerbation of chronic obstructive pulmonary disease based on critical values

作　　者：宋宁 吴先红 陈浩[3] 张明川 蔡明春 SONG Ning;WU Xianhong;CHEN Hao;ZHANG Mingchuan;CAI Mingchun(The First Clinical College of Chongqing Medical University,Chongqing 400016,China;Department of Orthopaedics Tongliang Hospital,the First Affiliated Hospital of Chongqing Medical University,Chongqing 402560,China;Department of General Medicine,Tongliang Hospital,the First Affiliated Hospital of Chongqing Medical University,Chongqing 402560,China;Department of Respiratory and critical Diseases,Tongliang Hospital,the First Affiliated Hospital of Chongqing Medical University,Chongqing 402560,China;Medical Record Department of Tongliang Hospital,the First Affiliated Hospital of Chongqing Medical University 402560,China)

机构地区：[1]重庆医科大学第一临床学院,400016 [2]重庆医科大学附属第一医院铜梁医院骨科,402560 [3]重庆医科大学附属第一医院铜梁医院全科医学科,402560 [4]重庆医科大学附属第一医院铜梁医院呼吸与危重症科,402560 [5]重庆医科大学附属第一医院铜梁医院病案科,402560

出　　处：《重庆医学》2023年第12期1806-1811,共6页Chongqing medicine

基　　金：重庆市科卫联合医学科研项目(2022MSXM034)。

摘　　要：目的从危急值和合并疾病等临床资料中探索慢性阻塞性肺疾病急性加重(AECOPD)患者死亡的相关危险因素,构建预测死亡风险的列线图模型,并评估其应用效能。方法回顾性分析2019年1月至2021年12月于重庆医科大学附属第一医院铜梁医院呼吸与危重症科住院治疗的626例AECOPD患者的临床资料,按7∶3的比例将研究对象分为训练组(440例)与验证组(186例)。采用Lasso回归进行数据降维,筛选最佳预测因子。结合多因素logistic回归分析得到独立危险预测因子,运用R软件构建列线图,受试者工作特征曲线(ROC曲线)下面积(AUC)评价模型的区分度,采用Calibration校准曲线评估模型预测概率与实际结果的一致性,利用决策分析曲线(DCA)评估模型的临床实用性。结果血气分析危急值项目:pH、氧分压(PaO_(2))、二氧化碳分压(PaCO_(2))及合并疾病:Ⅰ型呼吸衰竭、肺性脑病、肺结核、低蛋白血症,年龄校正Charlson合并症指数为AECOPD患者死亡的独立风险因素(P<0.05)。将以上8个变量纳入构建列线图预测模型,模型在训练组(AUC=0.759)和验证组(AUC=0.753)显示出良好的区分性,两组Calibration校准曲线接近于对角线,具有较好的准确性。DCA阈值为0~82%,显示了正的净效益,具有良好的临床实用性。结论基于危急值项目及合并疾病构建的AECOPD死亡预测模型有助于更好地辨识危急重症患者并及时为其提供更加精准的干预和治疗,降低死亡率。Objective To explore the related risk factors of death in patients with acute exacerbation of chronic obstructive pulmonary disease(AECOPD)from the clinical data such as critical value and concomitant diseases,to construct a nomogram to predict the risk of death,and to evaluate its effectiveness.Methods The clinical data of 626 AECOPD patients hospitalized in the Department of Respiratory and Critical Diseases,Tongliang Hospital,the first affiliated Hospital of Chongqing Medical University from January 2019 to December 2021 were analyzed retrospectively.The subjects were randomly divided into training group(n=440)and verification group(n=186)according to 7∶3 proportion.the least absolute shrinkage and selection operator method(LASSO)was used to reduce the dimension of the data and select the best predictors.The independent risk predictors were obtained by multivariate logistic regression analysis,The nomogram was constructed by R software,and the discrimination of the model was evaluated by area under the receiver operating characteristic curve.The calibration curve was used to evaluate the consistency between the prediction probability and the actual results,and the decision curve analysis(DCA)was used to evaluate the clinical practicability of the model.Results The critical value items of blood gas analysis:pH,PaO_(2),PaCO_(2) and Concomitant diseases:typeⅠrespiratory failure,pulmonary encephalopathy,pulmonary tuberculosis,hypoproteinemia,and age-adjusted Charlson comorbidity index were independent risk factors for death in patients with AECOPD.The above 11 variables were incorporated into the nomogram,and the model showed good differentiation in the training group(AUC=0.759)and verification group(AUC=0.753).The two groups of the calibration curve were close to the diagonal,with good accuracy.The DCA threshold shows a positive net benefit between 0 and 82%,suggesting that the model is clinically practical.Conclusion The AECOPD mortality prediction model based on critical value items and concomitant diseases

关 键 词：慢性阻塞性肺疾病急性加重 死亡风险 危急值 合并疾病 预测模型 

分 类 号：R563.9[医药卫生—呼吸系统]