扁桃体切除术后患儿继发性出血风险因素分析及预测模型构建  

Analysis of risk factors for secondary haemorrhage after tonsillectomy in children and construction of a prediction model

作　　者：李云雁 黎奥 韩朝晖[3] 燕志强 LI Yunyan;LI Ao;HAN Zhaohui;YAN Zhiqiang(Graduate School of Xuzhou Medical University(Department of Otolaryngology,Funing People's Hospital),Yancheng,Jiangsu,221004,China;Department of Otolaryngology Head and Neck Surgery,Gulou Hospital Affiliated to Nanjing University Medical School,Xuzhou Medical University,Nanjing,Jiangsu,210008,China;Department of Otolaryngology,Funing People's Hospital,Yancheng,Jiangsu,224400,China;Department of Otolaryngology,Affiliated Huaihai Hospital of Xuzhou Medical University/71st Army Group Military Hospital,Xuzhou,Jiangsu,221004,China)

机构地区：[1]徐州医科大学研究生院(阜宁县人民医院耳鼻咽喉科),江苏盐城221004 [2]徐州医科大学鼓楼临床学院,南京大学医学院附属鼓楼医院耳鼻咽喉头颈外科,江苏南京210008 [3]阜宁县人民医院耳鼻咽喉科,江苏盐城224400 [4]徐州医科大学附属淮海医院/陆军第七十一集团军医院耳鼻咽喉头颈外科,江苏徐州221004

出　　处：《中国耳鼻咽喉头颈外科》2023年第8期519-523,共5页Chinese Archives of Otolaryngology-Head and Neck Surgery

基　　金：中国博士后科研基金第68批面上项目(2020M681561)。

摘　　要：目的分析扁桃体切除术后患儿继发性出血风险因素,并探讨构建风险预测模型的预测效能。方法回顾性分析2015年1月~2020年12月阜宁县人民医院收治的338例接受扁桃体切除术治疗患儿临床资料,依据术后24 h后有无发生继发性出血分为出血组和未出血组,比较两组临床资料、手术资料及围术期相关资料。采用二元Logistic回归模型分析影响扁桃体切除术后继发性出血的相关因素,依据筛选的风险因子构建风险预测模型,对模型进行拟合度及预测效能评价。结果本次入选的338例患儿术后继发性出血发生率为10.95%(37/338),其中出血程度Ⅰ级35例(10.36%)、Ⅱ级2例(0.59%),无Ⅲ~Ⅳ级。所有出血情况均发生于术后2周内,其中术后6天发生9例,占总出血例数的比例最高(24.32%)。与未出血组比较,出血组患儿年龄、术后疼痛评分较高,监护人初中及以下学历、扁桃体炎发作频次≥3次/年、扁桃体全部切除、术后感染、术后进食不当的占比较高,扁桃体轻度包埋、术者工作经验≥6年、抗生素使用时间>3 d占比较低,差异比较有统计学意义(P<0.05)。二元多因素Logistic回归模型分析结果显示,监护人初中及以下学历、扁桃体全部切除、术后感染、术后疼痛评分高、术后进食不当是影响术后继发性出血的独立危险因素(P<0.05)。构建预测扁桃体切除术后患儿继发性出血的Logistic回归方程为Logit(P)=-0.125+0.315×监护人初中及以下学历+0.287×扁桃体全部切除+0.812×术后感染+0.418×术后疼痛评分+0.133×术后进食不当(χ^(2)=24.192,P<0.001)。模型验证结果显示模型校准曲线与理想曲线拟合良好(HosmerLemeshowχ^(2)=1.987,P=0.223);受试者工作特征(ROC)曲线进行外部验证结果显示模型预测术后继发性出血的AUC为0.893(95%CI:0.855~0.924,P<0.05),灵敏度89.19%、特异度80.40%、约登指数0.696。结论监护人初中及以下学历、扁桃体全部切OBJECTIVE To analyse the risk factors for secondary bleeding after tonsillectomy in children and to investigate the predictive efficacy of the constructed risk prediction model.METHODS Retrospective analysis of the clinical data of 338 children admitted to our hospital from January 2015 to December 2020 who underwent tonsillectomy.They were divided into the bleeding group and the non-bleeding group based on the occurrence of secondary bleeding 24 h after operation.The clinical data,surgical data and perioperative related data of the two groups were compared.A binary logistic regression model was used to analyze the factors affecting secondary bleeding after tonsillectomy,and a risk prediction model was constructed on the basis of the screened risk factors,and the model was evaluated for fit and predictive efficacy.RESULTS The incidence of secondary hemorrhage in the 338 children enrolled in this study was 10.95%(37/338).Among them,35 cases(10.36%)had grade I bleeding,2 cases(0.59%)had grade II bleeding,and there were no cases of grade III to IV bleeding.All bleeding occurred within 2 weeks after operation,with 9 cases occurring 6 days after operation,which accounted for the highest percentage(24.32%)of the total number of bleeding cases.Compared with the non-bleeding group,the bleeding group had higher age,higher postoperative pain score,higher percentage of guardians with junior high school education or below,frequency of tonsillitis episodes≥3 times/year,total tonsillectomy,postoperative infection,postoperative inappropriate eating,and lower percentage of mildly encapsulated tonsils,≥6 years of experience of the operator,duration of antibiotic use>3 d(P<0.05).The results of the binary multifactorial logistic regression model analysis showed that guardians with junior high school or lower education,total tonsillectomy,postoperative infection,high postoperative pain score,and inappropriate postoperative eating were the independent risk factors affecting secondary bleeding(P<0.05).The logistic regression equat

关 键 词：扁桃体切除术 危险因素 手术后出血 预测模型 

分 类 号：R766.9[医药卫生—耳鼻咽喉科]