Background: This study aims to compare the efficacy of oral anti-diabetes drugs in glycemic control in treating type 2 diabetes patients through network meta-analysis with an automated workflow. Methods: Systematic searching and screening were conducted for the RCTs on the drugs for monotherapy of type 2 diabetes by pre-specified search strategy and selection criteria. Searched databases included PubMed and Cochrane Library. Eligible RCTs were included for data extraction and quality assessment according to Jadad scale and Cochrane risk of bias assessment tool. Mean differences (MD) of the outcomes were analyzed with a random-effects model. Direct and indirect comparisons were conducted to compute the efficacy ranking based on glycated haemoglobin (HbA1c) and Fasting plasma glucose (FPG). Sensitivity analysis based on RCT quality and sample sizes were used to test the robustness of the results. Results: A total of 56 RCTs including 16066 patients were included for network meta-analysis. Quality assessment found the included RCTs low in quality. Preliminary analysis results showed that metformin and repaglinide were the top among efficacy ranking based on HbA1c and FPG. Sensitivity analysis indicated that the results were not robust enough. The automated workflow was effective in updating network meta-analysis. Conclusion: Metformin and repaglinide may have the best efficacy in lowering the HbA1c and FPG levels.

背景：本研究的目的是通过网络荟萃分析评价不同药物治疗成年二型糖尿病控制 血糖的疗效。并设计一个自动化的网络荟萃分析框架进行数据分析和结果更新。 方法：根据设定的检索策略和筛选标准对选定药物单独治疗成年二型糖尿病病人 的随机对照试验（RCT）进行系统检索和筛选，检索数据库为 PubMed 和 Cochrane Library。符合筛选标准的 RCTs 纳入数据提取环节，并利用 Jadad 评分表和 Cochrane 风险评价工具对 RCTs 的质量进行评价。随机效应模型和平均差 MD 值 用于药物两两对比的荟萃分析。直接比较和间接比较用于评价药物基于 HbA1c 和 FPG 的疗效。基于 RCT 质量和样本量的敏感性分析用于评估分析结果的稳定 性，分析结果通过网络荟萃分析框架进行定期更新。 结果：共有 56 篇随机对照试验包含 16066 个病人纳入网络荟萃分析。质量评价 显示纳入的随机对照试验质量偏低。初步分析结果显示 metformin 和 repaglinide 位于 HbA1c 和 FPG 疗效排序的的前两位。敏感性分析显示由于纳入 RCTs 数量 较小，目前药物排序结果还不够稳定。网络荟萃分析框架能够高效实现网络荟萃 分析并对结果进行更新。 结论：通过网络荟萃分析得出 metformin 和 repaglinide 可能是基于 HbA1c 和 FPG 疗效最好的药物。