胰十二指肠切除术后胰瘘风险预测模型的系统评价  

Risk prediction model for pancreatic fistula after pancreaticoduodenectomy:a systematic review

作　　者：谢若男 王国蓉 李成香 蒋欣 冷英杰 聂茁苗 XIE Ruonan;WANG Guorong;LI Chengxiang;JIANG Xin;LENG Yingjie;NIE Zhuomiao(School of Nursing,Chengdu University of Traditional Chinese Medicine,Chengdu 610075,China;Department of Nursing,West China School of Public Health/West China Fourth Hospital of Sichuan University,Chengdu 610042,China)

机构地区：[1]成都中医药大学护理学院,成都610075 [2]四川大学华西公共卫生学院/华西第四医院护理部,成都610042

出　　处：《数理医药学杂志》2025年第2期133-144,共12页Journal of Mathematical Medicine

摘　　要：目的系统评价胰十二指肠切除术后胰瘘风险预测模型,为临床选择风险评估工具提供借鉴。方法计算机检索PubMed、Embase、the Cochrane Library、Web of Science、中国知网、万方、维普、中国生物医学文献服务系统(SinoMed)发表的相关文献,检索时间为2016年1月1日至2024年4月17日。采用PROBAST偏倚风险评估工具对纳入研究进行评估,采用Stata 16.0软件对风险预测模型中包含的常见预测因子进行Meta分析。结果共纳入27项研究。所有研究均报告了模型区分度,14项研究报告了校准。纳入研究的构建模型受试者工作特征曲线下面积(area under the curve,AUC)为0.620~0.970,内部验证AUC为0.620~0.915,外部验证模型AUC为0.744~0.849。纳入模型最常见预测因子为胰腺质地、体重指数(body mass index,BMI)、胰管直径、术后第1天腹腔引流液淀粉酶水平、术前白蛋白水平及术中出血量。所有研究的偏倚风险均为高风险,纳入研究的适用性良好。结论胰十二指肠术后胰瘘风险预测模型的建模质量较差,模型的临床应用有待验证,未来研究可考虑对现有模型进行外部验证和重新校准,或参照方法学指南构建新的预测模型并进行验证。Objective To systematically evaluate the risk prediction model of pancreatic fistula after pancreaticoduodenectomy,and to provide a reference for clinical selection of risk assessment tools.Methods The relevant literature published in PubMed,Embase,the Cochrane Library,Web of Science,CNKI,WanFang Data,VIP and SinoMed from January 1,2016 to April 17,2024 were searched.The PROBAST risk of bias assessment tool was used to assess the included studies,and Stata 16.0 software was used to perform Meta-analysis of common predictors included in the risk prediction model.Results A total of 27 studies were included.All studies reported model discrimination,14 studies reported calibration.The area under the curve(AUC)of the receiver operator characteristic curve of the constructed model included in the study ranged from 0.620 to 0.970,the AUC of the internal validation model ranged from 0.620 to 0.915,and the AUC of the external validation model ranged from 0.744 to 0.849.The most common predictors of the included models were pancreatic texture,body mass index(BMI),pancreatic duct diameter,abdominal drainage amylase level on the 1st postoperative day,preoperative albumin level,and intraoperative bleeding loss.The risk of bias was high in all studies,and the applicability of the included studies was good.Conclusion The modeling quality of the risk prediction model of the pancreatic fistula after pancreaticoduodenectomy was poor,and the clinical application of the model need to be validated.In future studies,external validation and recalibration of the existing model can be considered,or a new prediction model can be constructed and verified by referring to methodological guidelines.

关 键 词：胰十二指肠切除术 胰瘘 风险 预测模型 系统评价 

分 类 号：R657.5[医药卫生—外科学]