Fig 1.
We searched the MEDLINE and EMBASE, the Cochrane Central Register of Controlled Trials, and the ClinicalTrials.gov registry for articles published through March 2016 (S1 Table). The OADs targeted in our comparison were metformin, sulfonylureas, thiazolidinedione (TZD), dipeptidyl peptidase 4 (DPP4) inhibitors, sodium glucose cotransporter 2 (SGLT2) inhibitors, alpha-glucosidase inhibitors, and meglitinide. *When searching the ClinicalTrials.gov registry, we did not use the outcome keywords. As a result, we reviewed all registered clinical trials including unpublished reports about OADs that met our comparison criteria. We included not only pre-specified cardiovascular outcome but also cardiovascular events reported as severe adverse events (SAEs).
Fig 2.
Network of oral antidiabetic drugs comparison of all-cause mortality for the network meta-analysis.
Each circle node represents a drug included in the analysis and the size of circle is proportional to the number of patients randomly assigned to the drug. Each line corresponds to direct comparison between drugs and the width of line is proportional to the number of trials comparing each pair of treatments. TZD: thiazolidinedione. DPP4: dipeptidyl peptidase-4. SGLT2: sodium glucose cotransporter-2.
Fig 3.
Network and pairwise meta-analyses for all-cause mortality of oral antidiabetic drugs.
Traditional pairwise (upper right side) and network (lower left side) meta-analytic results are depicted for the all-cause mortality. Outcome of meta-analysis is expressed as relative risks (RRs) (95% confidence intervals) in the case of pairwise meta-analysis and (95% credible intervals) in the case of network meta-analysis. For the pairwise meta-analysis, RRs of less than 1 indicate that the drug located in the column is safer. For the network meta-analysis, RRs of less than 1 indicate that the drug located in the row is safer. Bold results indicate statistical significance. The analyses of all-cause mortality risk included data from 73 RCTs but the sum of total studies is 75 because two trials split in four. One is A Diabetes Outcome Progression Trial (ADOPT) [29], three-arm (metformin, glyburide, and rosiglitazone) study, which split in three. The other is Rosiglitazone Evaluated for Cardiovascular Outcomes and regulation of Glycaemia in Diabetes (RECORD) [31] which split in two after searching reports as separated by metformin or sulfonylurea at ClinicalTrials.gov website. TZD: thiazolidinedione. DPP4: dipeptidyl peptidase-4. SGLT2: sodium glucose cotransporter-2.
Fig 4.
Network and pairwise meta-analyses for cardiovascular-related mortality of oral antidiabetic drugs.
Traditional pairwise (upper right side) and network (lower left side) meta-analytic results are depicted for the cardiovascular-related mortality. Outcome of meta-analysis is expressed as relative risks (RRs) (95% confidence intervals) in the case of pairwise meta-analysis and (95% credible intervals) in the case of network meta-analysis. For the pairwise meta-analysis, RRs of less than 1 indicate that the drug located in the column is safer. For the network meta-analysis, RRs of less than 1 indicate that the drug located in the row is safer. Bold results indicate statistical significance. The analyses of all-cause mortality risk included data from 29 RCTs but the sum of total studies is 32 because two trials split in four. One is A Diabetes Outcome Progression Trial (ADOPT) [29], three-arm (metformin, glyburide, and rosiglitazone) study, which split in three. The other is Rosiglitazone Evaluated for Cardiovascular Outcomes and regulation of Glycaemia in Diabetes (RECORD) [31] which split in two after searching reports as separated by metformin or sulfonylurea at ClinicalTrials.gov website. TZD: thiazolidinedione. DPP4: dipeptidyl peptidase-4. SGLT2: sodium glucose cotransporter-2.
Fig 5.
Network and pairwise meta-analyses for acute coronary syndrome of oral antidiabetic drugs.
The analyses of acute coronary syndrome risk included data from acute coronary syndrome, acute myocardial infarction or unstable angina counted as severe adverse events. Traditional pairwise (upper right side) and network (lower left side) meta-analytic results are depicted for the cardiovascular-related mortality. Outcome of meta-analysis is expressed as relative risks (RRs) (95% confidence intervals) in the case of pairwise meta-analysis and (95% credible intervals) in the case of network meta-analysis. For the pairwise meta-analysis, RRs of less than 1 indicate that the drug located in the column is safer. For the network meta-analysis, RRs of less than 1 indicate that the drug located in the row is safer. Bold results indicate statistical significance. The analyses included data from 51 RCTs but the sum of total studies is 52 because two trials split in two. Rosiglitazone Evaluated for Cardiovascular Outcomes and regulation of Glycaemia in Diabetes (RECORD) [31] split in two after searching reports as separated by metformin or sulfonylurea at ClinicalTrials.gov website. TZD: thiazolidinedione. DPP4: dipeptidyl peptidase-4. SGLT2: sodium glucose cotransporter-2.
Fig 6.
Network and pairwise meta-analyses for myocardial infarction of oral antidiabetic drugs.
The analyses of myocardial infarction risk included data from not only published article as non-fatal myocardial infarction but unpublished reports as acute myocardial infarction or myocardial infarction as severe adverse events. Traditional pairwise (upper right side) and network (lower left side) meta-analytic results are depicted for the cardiovascular-related mortality. Outcome of meta-analysis is expressed as relative risks (RRs) (95% confidence intervals) in the case of pairwise meta-analysis and (95% credible intervals) in the case of network meta-analysis. For the pairwise meta-analysis, RRs of less than 1 indicate that the drug located in the column is safer. For the network meta-analysis, RRs of less than 1 indicate that the drug located in the row is safer. Bold results indicate statistical significance. The analyses of all-cause mortality risk included data from 70 RCTs but the sum of total studies is 73 because two trials split in four. One is A Diabetes Outcome Progression Trial (ADOPT) [29], three-arm (metformin, glyburide, and rosiglitazone) study, which split in three. The other is Rosiglitazone Evaluated for Cardiovascular Outcomes and regulation of Glycaemia in Diabetes (RECORD) [31] which split in two after searching reports as separated by metformin or sulfonylurea at ClinicalTrials.gov website. TZD: thiazolidinedione. DPP4: dipeptidyl peptidase-4. SGLT2: sodium glucose cotransporter-2.