个体化预测低温等离子扁桃体切除术后出血疗效的列线图模型构建

Construction of a nomogram model for individualized prediction of postoperative hemorrhage after coblation assist tonsillectomy

目的 分析低温等离子扁桃体切除术后出血的影响因素,并构建列线图模型。方法 选取在淮安市第二人民医院接受低温等离子扁桃体切除术治疗的患者645例,根据术后是否出血将其分为出血组(n=45例)和未出血组(n=600例)。收集患者资料,分析影响患者术后出血的因素(单因素和多因素Logistic回归分析),构建列线图模型(使用R软件)并对模型进行验证。结果 单因素分析显示,患者年龄、扁桃体包埋程度、既往出血史、术后进食不当及术后继发感染是影响低温等离子扁桃体切除术后出血的因素(P均<0.05),而与性别、手术方式、烟酒嗜好史、术者资历、手术季节、凝血酶原时间(prothrombin time,PT)、活化部分凝血活酶时间(activated partial thromboplastin time,APTT)、血小板(PLT)、白细胞(WBC)及红细胞(RBC)无关(P>0.05);多因素Logistic回归分析显示,年龄<18岁、既往出血史及术后继发感染为影响低温等离子扁桃体切除术后出血的独立危险因素(P<0.05);使用R软件根据多因素Logistic回归分析结果构建列线图模型,并以Bootstrap法重复1000次绘制H-L拟合度曲线,结果显示χ2=7.025,P=4.236;绘制ROC曲线评估模型的区分度,结果显示ROC曲线下面积为0.715(95%CI:0.677-0.751),敏感性为62.22%,特异性为81.80%。结论 年龄<18岁、既往出血史及术后继发感染为影响低温等离子扁桃体切除术后出血的独立危险因素,根据结果构建的列线图模型具有较高的区分度及有效性,可作为临床进行个体化预测的有效工具。

OBJECTIVE To analyze the influencing factors of postoperative bleeding after coblation assisted tonsillectomy,and to construct a nomogram model.METHODS A total of 645 patients who underwent coblation assisted tonsillectomy in our hospital were chosen in this study,and they were divided into bleeding group(45 cases)and non-bleeding group(600 cases)according to whether there was bleeding or not after operation.The clinical data of the patients were collected,the factors affecting postoperative bleeding were analyzed(univariate and multivariate Logistic regression analysis),and a nomogram model was constructed(using R software)and verified.RESULTS Univariate analysis showed that age,the degree of tonsil enlargepment,past bleeding history,improper postoperative eating,and secondary postoperative infection were the factors affecting postoperative bleeding after coblation assisted tonsillectomy(P<0.05),but gender,surgical method,history of tobacco and alcohol use,surgeon's qualifications,operation season,PT,APTT,PLT,WBC and RBC were not related to the bleeding after operation(P>0.05).Multivariate Logistic regression analysis showed that age<18 years old,past bleeding history,secondary postoperative infection were independent risk factors for postoperative bleeding after coblation assisted tonsillectomy(P<0.05);R software was used to construct a nomogram model according to the multivariate Logistic regression analysis results,and the Bootstrap method was repeated 1000 times to draw the H-L fitting curve,and the results showed that x2=7.025,P=4.236.The ROC curve was drawn to evaluate the discrimination of the model,the results showed that the area under the ROC curve was 0.715(95%CI:0.677-0.751),and the sensitivity and specificity were 62.22% and 81.80%,respectively.CONCLUSION Age<18 years old,past bleeding history,and secondary postoperative infection are independent risk factors for postoperative bleeding after coblation assisted tonsillectomy.The nomogram model constructed according to the results has a high degree of discrimination and validity,and can be used as an effective tool for individualized clinical prediction.