个体化预测长期气管切开肺部感染风险的列线图模型建立  

Establishment of nomogram model for individualized risk prediction of pulmonary infection in long-term tracheotomy patients

作　　者：李贤英 丁克元 王捷[1] LI Xian-ying;DING Ke-yuan;WANG Jie(The Fifth People's Hospital of Wuhu City,Wuhu,Anhui 24l000,China)

机构地区：[1]芜湖市第五人民医院呼吸内科,安徽芜湖241000

出　　处：《中华医院感染学杂志》2023年第10期1500-1503,共4页Chinese Journal of Nosocomiology

基　　金：安徽省科技厅项目(1607a0202055)。

摘　　要：目的建立长期气管切开患者肺部感染风险个体化预测的列线图模型,并评估模型的预测价值。方法回顾性分析2018年1月-2022年1月芜湖市第五人民医院收治的75例长期气管切开并发肺部感染患者及同期230例长期气管切开无肺部感染患者临床资料,行单因素及多因素Logistic回归分析,筛选长期气管切开患者肺部感染独立危险因素,利用R软件建立预测长期气管切开并发肺部感染风险的列线图模型,采用Boot-strap法验证模型,受试者工作特征(ROC)曲线分析列线图模型预测肺部感染的价值。结果多因素Logistic回归分析显示,合并糖尿病、原发性疾病为肺部疾病、吸烟史、意识障碍、低蛋白血症、气管切开前7 d内经常误吸、气道湿化不足/过度是长期气管切开患者并发肺部感染的危险因素。根据筛选的危险因素绘制列线图模型,R软件计算结果显示,列线图模型一致性指数(C-index)为0.811,预测能力良好。ROC曲线结果提示该列线图模型在预测长期气管切开患者肺部感染中的价值较高,曲线下面积为0.862(95%CI:0.792~0.932)灵敏度为85.0%,特异度为81.7%。结论基于糖尿病、原发性疾病为肺部疾病、吸烟史、意识障碍、低蛋白血症、气管切开前7d内经常误吸、气道湿化不足/过度7项独立危险因素建立的列线图模型可有效预测长期气管切开患者肺部感染发生风险。OBJECTIVE To establish the nomogram model for individualized risk prediction of pulmonary infection in patients undergoing long-term tracheotomy and assess the predictive value of the model.METHODS A total of 75 long-term tracheotomy patients who were complicated with pulmonary infection and treated in the Fifth People′s Hospital from Jan 2018 to Jan 2022 were enrolled in the study,meanwhile,230 long-term tracheotomy patients who were not complicated with pulmonary infection were also recruited as the research subjects,the clinical data of the patients were retrospectively analyzed.Univariate analysis and multivariate logistic regression analysis were performed,the independent risk factors for the pulmonary infection in the long-term tracheotomy patients were screened out,the nomogram model for risk prediction of pulmonary infection in long-term tracheotomy patients was established based on R software,and the value of the nomogram model in prediction of pulmonary infection was analyzed by Boot-strap verification model and receiver operating characteristic(ROC)curve.RESULTS Multivariate logistic regression analysis showed that complication with diabetes mellitus,primary pulmonary disease,smoking history,disturbance of consciousness,hypoproteinemia,frequent aspiration within 7 days before tracheotomy and inadequate/excessive airway humidification were the risk factors for pulmonary infection in the long-term tracheotomy patients.The nomogram was drawn based on the screened risk factors,and the result of R software calculation showed that the consistency index(C-index)of the nomogram model was 0.811,and it has favorable predictive ability.ROC curve analysis indicated that the nomogram model had high value in prediction of pulmonary infection in the patients undergoing long-term tracheotomy,with the area under curve 0.862(95%CI:0.792-0.932),the sensitivity 85.0%,the specificity 81.7%.CONCLUSION The nomogram model that is established based on the 7 independent risk factors diabetes mellitus,primary pulmonary disease

关 键 词：个体化预测 气管切开 肺部感染 列线图模型 危险因素 

分 类 号：R63[医药卫生—外科学]