2型糖尿病患者低血糖风险预测模型的系统评价  

A systematic review of hypoglycemia risk prediction models in patients with type 2 diabetes mellitus

作　　者：詹立睿 张紫薇 宋萍 杨雨菡 曾冬阳[1] Zhan Lirui;Zhang Ziwei;Song Ping;Yang Yuhan;Zeng Dongyang(International College of Nursing,Hainan Medical College,Haikou 570100,China;Department of Academic Affairs,First Affiliated Hospital of Hainan Medical College,Haikou 570100,China)

机构地区：[1]海南医学院国际护理学院,海口570100 [2]海南医学院第一附属医院教务科,海口570100

出　　处：《中华糖尿病杂志》2023年第3期244-251,共8页CHINESE JOURNAL OF DIABETES MELLITUS

基　　金：海南省自然科学基金(822RC692);海南省教育厅科研基金(Hnky2022ZD-14)。

摘　　要：目的系统评价2型糖尿病(T2DM)患者低血糖风险预测模型。方法检索中国期刊全文数据库(CNKI)、万方数据知识服务平台、美国国立医学图书馆数据库(PubMed)、Cochrane循证医学数据库(Cochrane Library)、医学文摘数据库(EMbase)及Web of Science数据库从建库至2022年7月前发表的T2DM患者低血糖风险预测模型相关文献。由研究者独立筛选文献,并提取文献中涉及模型的曲线下面积(AUC)及其95%CI、校准方法和预测因子,使用预测模型研究的偏倚风险评估工具(PROBAST)对模型进行质量评价。使用Revman5.3软件对模型中预测因子的预测价值进行Meta分析。结果共纳入9篇文献,包含12个模型,其中11个模型的AUC>0.7,7个模型同时报告了AUC的95%CI,7个模型进行了模型校准。PROBAST结果显示,纳入的9篇文献中,有1篇为低偏倚风险,其余8篇均为高偏倚风险;在模型适用性中,仅有1篇为低适用性。Meta分析结果显示,胰岛素的使用(OR=6.11,95%CI 5.41~6.91)、体重指数(OR=2.69,95%CI 1.75~5.10)、糖尿病病程(OR=3.39,95%CI 2.37~4.85)、既往低血糖史(OR=9.73,95%CI 8.72~10.85)、磺脲类药物的使用(OR=1.30,95%CI 1.30~1.31)是预测模型中位列前5的预测因子。结论T2DM患者低血糖风险预测模型尚存在不足,未来预测模型的建立可重点关注胰岛素的使用、体重指数、糖尿病病程、既往低血糖史、磺脲类药物的使用等预测因子。Objective To systematically evaluate hypoglycemic risk prediction models in patients with type 2 diabetes mellitus(T2DM).Methods The Cochrane Library,PubMed,Embase,Web of Science,China national knowledge infrastructure(CNKI),and Wan Fang Databases were searched to collect the studies on T2DM hypoglycemia risk prediction model from inception to July 2022.The investigators independently screened the literatures and extracted the area under the curve(AUC)and its 95%CI,calibration method and predictors of the models involved in the literature using the prediction model risk of bias assessment tool(PROBAST)for quality evaluation.Meta-analysis of the predictive value of the predictors in the model was performed using Revman 5.3.Results A total of 9 literatures containing 12 models were included,11 of which had AUC>0.7,7 models reported a 95%CI of AUC,and 7 models performed model calibration.PROBAST results showed that 1 of the 9 included literatures was at low risk of bias and the remaining 8 were at high risk of bias.Among the model applicability,only 1 was at low applicability.Meta-analysis showed that insulin use(OR=6.11,95%CI 5.41-6.91),body mass index(OR=2.69,95%CI 1.75-5.10),duration of diabetes(OR=3.39,95%CI 2.37-4.85),previous history of hypoglycemia(OR=9.73,95%CI 8.72-10.85),and sulfonylurea use(OR=1.30,95%CI 1.30-1.31)were the top 5 predictors in the prediction model.Conclusions The prediction model of hypoglycemic risk in T2DM patients was still inadequate,and the future prediction model should focus on such predictors as insulin use,body mass index,duration of diabetes,previous history of hypoglycemia,and use of sulfonylureas.

关 键 词：糖尿病 2型 低血糖 风险预测 模型 系统评价 

分 类 号：R587.1[医药卫生—内分泌]