摘要
目的探讨重症肺炎(SP)患者住院结局的影响因素,构建并评估预测模型。方法回顾性分析2023年1-9月在徐州医科大学附属医院住院的SP患者临床资料,将2023年1-5月住院的SP患者作为训练集,将2023年6-9月住院的患者作为验证集。根据住院结局将患者分为改善组和未改善组。通过多因素Logistic回归分析筛选SP患者住院结局的影响因素并构建预测模型,采用受试者工作特征(ROC)曲线、校准曲线评价模型并通过验证集验证模型,同时与CURB-65评分、A-DROP评分和序贯器官功能衰竭(SOFA)评分比较,评价预测模型的效能。使用X-tile确定预测模型的最佳截断值对SP患者进行风险分层。结果多因素Logistic回归分析显示,痰培养结果阳性、D-二聚体(D-D)水平升高、组织纤溶酶原激活物-纤溶酶原激活物抑制剂-1复合物(t-PAIC)水平升高是SP患者住院未改善的独立危险因素(P<0.05)。SP住院结局预测模型:预测指数(PI)=2.167×X痰培养结果+0.273×X D-D+0.078×X t-PAIC,其中痰培养结果阳性=1、阴性=0。将预测模型应用于训练集和验证集进行H-L检验,均P>0.05。在训练集中预测模型预测SP患者住院未改善的曲线下面积(AUC)为0.842(95%CI:0.778~0.905)。在验证集中,预测模型、CURB-65评分、A-DROP评分、SOFA评分预测SP患者住院结局未改善的AUC分别为0.774(95%CI:0.688~0.861)、0.620(95%CI:0.521~0.719)、0.650(95%CI:0.553~0.747)、0.684(95%CI:0.590~0.779),预测模型的AUC与CURB-65评分、A-DROP评分的AUC比较,差异均有统计学意义(Z=2.656,P=0.008;Z=2.026,P=0.043),与SOFA评分的AUC比较,差异无统计学意义(Z=1.502,P=0.133)。基于模型将SP患者分为低、中、高危组,住院改善率依次为81.25%(26/32)、46.51%(20/43)、25.53%(12/47)。结论痰培养结果阳性、D-D水平升高和t-PAIC水平升高是SP患者住院未改善的独立危险因素,该研究构建的模型具有更高的临床应用价值,可补充现有评分系统的�
Objective To explore the influencing factors for hospitalization outcomes in the patients with severe pneumonia(SP),and to construct and evaluation predictive model.Methods A retrospective analysis was conducted on the clinical data of SP patients admitted to the Affiliated Hospital of Xuzhou Medical University from January to September 2023.The SP patients hospitalized from January to May 2023 served as the training set,and the SP patients hospitalized from June to September 2023 as the validation set.The patients were divide into the improved group and an non-improved group based on hospitalization outcomes.The influencing factors were analyzed and screened through multivariate Logistic regression analysis and a predictive model was constructed.The model was evaluated by using the receiver operating characteristic(ROC)curve and calibration curve and validated through the validation set.Meanwhile which was compared with the CURB-65 score,A-DROP score and Sequential Organ Failure Assessment(SOFA)score for evaluating the efficiency of the predictive model.X-tile was used to determine the optimal cut-off value of the model for conducting the risk stratification of SP patients.Results The multivariate Logistic regression analysis showed that the positive sputum culture results,elevated levels of D-dimer(D-D),and elevated levels of tissue plasminogen activator inhibitor-1 complex(t-PAIC)were the independent risk factors for hospitalization without improvement in SP patients(P<0.05).The prediction model for SP hospitalization outcome was:prediction index(PI)=2.167×X sputum culture result+0.273×X D-D+0.078×X t-PAIC,where the sputum culture result positive=1 and negative=0.The H-L test P-values for applying the model to both the training and validation sets were>0.05.In the training set,the area under the curve(AUC)of prediction model for predicting non-improvement in SP patients with hospitalization was 0.842(95%CI:0.778-0.905);in the validation set,AUC of the predictive model,CURB-65 score,A-DROP score and SOFA sco
作者
陈燕
郭毅
孙静芳
徐银海
CHEN Yan;GUO Yi;SUN Jingfang;XU Yinhai(College of Medical Technology,Xuzhou Medical University,Xuzhou,Jiangsu 221004,China;Department of Clinical Laboratory,Affiliated Hospital of Xuzhou Medical University,Xuzhou,Jiangsu 221004,China;Department of Clinical Laboratory,Second Affiliated Hospital of Xuzhou Medical University,Xuzhou,Jiangsu 221006,China)
出处
《检验医学与临床》
CAS
2024年第12期1665-1670,1675,共7页
Laboratory Medicine and Clinic
基金
中国博士后科学基金项目(2023M732975)。
关键词
重症肺炎
住院结局
危险因素
预测模型
D-二聚体
severe pneumonia
hospitalization outcomes
risk factors
predictive model
D-dimer
