Concerns About Meta-analysis of Glucose Control and Cardiovascular Disease in Type 2 Diabetes

Potential Conflicts of Interest: None disclosed.

IN RESPONSE:

We thank Drs. Yudkin and Richter for their interest in our meta-analysis. The decision to analyze the subgroups of early and more recent trials was an a priori determination based on important differences in patient populations, targeted hemoglobin A_1c levels, and treatment regimens. Specifically, the early UKPDS trials included only patients with newly diagnosed diabetes, whereas the more recent ACCORD (Action to Control Cardiovascular Risk in Diabetes), ADVANCE (Action in Diabetes and Vascular Disease), and VADT (Veterans Affairs Diabetes Trial) studies included patients with prevalent diabetes. The intensive glucose control target in the UKPDS was similar to the conventional glucose control target in the more recent trials. Finally, the UKPDS used diet as the primary method of conventional treatment, whereas the more recent trials used hypoglycemic agents. We included UKPDS 34 in our meta-analysis because the 342 patients assigned to metformin for intensive treatment were completely independent of the 2729 patients assigned to sulfonylureas or insulin for intensive treatment in UKPDS 33 (1). Results from different treatment groups of clinical trials that share a common control group are sometimes combined, although ideally our estimates should have accounted for underlying correlation between the effect size, which they did not (2, 3). The UKPDS 34 control group comprises a relatively small proportion of the UKPDS 33 control group (approximately 36%), and no meta-analytic methods exist to account for correlations in binary end points of trials that use partially shared control groups. We believe that the UKPDS 34 provides important information on the effects of intensive glucose control on cardiovascular disease, and any correlation it may have with the UKPDS 33 due to a partially shared control group does not warrant its exclusion from the meta-analysis.