围术期食管癌术后吻合口瘘的危险因素及风险预测模型构建与验证  

Risk factors for anastomotic leakage after perioperative esophageal cancer surgery and construction and verification of risk prediction model

作　　者：时祥音 白璟 吕丛 张冰 窦晓雅 SHI Xiang-yin;BAI Jing;LV Cong;ZHANG Bing;DOU Xiao-ya(Department of Esophagus Surgery,Nanyang Central Hospital,Nanyang 473000,Henan,CHINA;Department of Urology,Nanyang Central Hospital,Nanyang 473000,Henan,CHINA;Outpatient Office,Nanyang Central Hospital,Nanyang 473000,Henan,CHINA;Department of Infectious Diseases 4,Nanyang Central Hospital,Nanyang 473000,Henan,CHINA)

机构地区：[1]南阳市中心医院胸外科食管病区,河南南阳473000 [2]南阳市中心医院泌尿外科,河南南阳473000 [3]南阳市中心医院门诊办公室,河南南阳473000 [4]南阳市中心医院感染科,河南南阳473000

出　　处：《海南医学》2024年第7期1026-1030,共5页Hainan Medical Journal

基　　金：河南省卫生和计划生育委员会(编号:2018-YJ-057-R06/06)。

摘　　要：目的探讨围术期食管癌(EC)术后食管胃吻合口瘘(EGAL)的危险因素及风险预测模型构建与验证。方法回顾性分析2021年4月至2023年4月于南阳市中心医院行EC手术的176例患者的临床资料,根据术后是否发生EGAL分为EGAL组(n=30)和非EGAL组(n=146)。收集两组患者的临床资料并进行单因素及多因素Logistic回归分析,以确定EC患者发生EGAL的独立危险因素,并将其纳入Rstudio 4.2.1软件以构建预测EC患者发生EGAL的Nomogram模型;绘制受试者工作特征(ROC)曲线和决策曲线(DCA),采用H-L拟合优度检验与校准曲线评估模型一致性,并进行内部验证。结果EGAL组患者的吸烟史、高血压、糖尿病、吻合方式为手工、手术时间、N分期为N2、术后白蛋白<30 g/L及术后肺部感染率明显高于非EGAL组,差异均有统计学意义(P<0.05);经多因素Logistic回归分析结果显示,高血压、糖尿病、手工吻合方式、术后白蛋白<30 g/L、术后肺部感染是EC患者术后发生EGAL的独立影响因素(P<0.05);基于Logistic回归分析构建Nomogram,Nomogram模型纳入了高血压、糖尿病、吻合方式、术后白蛋白及术后肺部感染5个因素;Nomogram模型预测EC患者术后发生EGAL的ROC曲线下面积(AUC)为0.914,95%CI为0.867~0.961(P<0.05),灵敏度为90.00%,特异度为79.50%;验证模型的AUC为0.875(P<0.001,95%CI:0.806~0.944),灵敏度为73.30%,特异性为90.40%。且模型校正曲线总体趋势接近理想曲线,表明该模型拟合度良好。结论高血压、糖尿病、吻合方式、术后白蛋白、术后肺部感染均是EC患者术后发生EGAL的危险因素,根据危险因素构建的Nomogram预测模型具有较好的预测效能。Objective To explore the risk factors of esophagogastric anastomotic leakage(EGAL)after perioperative esophageal cancer(EC)and the construction and verification of risk prediction model.Methods The clinical data of 176 patients who underwent EC surgery in Nanyang Central Hospital from April 2021 to April 2023 were retrospectively analyzed.According to whether EGAL occurred after operation,they were divided into EGAL group(n=30)and non-EGAL group(n=146).The clinical data of the two groups of patients were collected.Univariate and multivariate logistic regression analysis was performed to determine the independent risk factors for EGAL in EC patients,which were included in Rstudio 4.2.1 software to construct a Nomogram model for predicting EGAL in EC patients.Receiver operating characteristic(ROC)curve and decision curve(DCA)were drawn.The consistency of the model was evaluated by H-L goodness of fit test and calibration curve,and internal verification was carried out.Results The smoking history,hypertension,diabetes,manual anastomosis,operation time,N stage N2,postoperative albumin<30 g/L,and postoperative pulmonary infection of patients in the EGAL group were significantly higher than those in the non-EGAL group(P<0.05).Multivariate Logistic regression analysis showed that hypertension,diabetes mellitus,manual anastomosis,postoperative albumin<30 g/L,and postoperative pulmonary infection were the independent influencing factors for postoperative development of EC patients(P<0.05).Five factors including hypertension,diabetes,anastomotic pattern,postoperative albumin,and postoperative pulmonary infection were included in the Nomogram model.The area under ROC curve(AUC)and 95%Cl in the prediction of postoperative EC by Nomogram model were 0.914 and 0.867-0.961(P<0.05),with the sensitivity of 90.00%and the specificity of 79.50%.The area under ROC curve of the validated model was 0.875(P<0.001,95%CI:0.806-0.944),with the sensitivity of 73.30%and specificity of 90.40%.The general trend of the model correction curve was

关 键 词：食管癌 食管胃吻合口瘘 危险因素 预测模型 

分 类 号：R473.73[医药卫生—护理学]