Abstract

The study published by Nissen¹ is a meta-analysis of randomized, controlled trials of 24 weeks or longer duration in which rosiglitazone was compared with placebo or other therapies. The analysis included phase 2, 3, or 4 trials for which data was available regarding outcomes of myocardial infarction (MI) and death from cardiovascular (CV) causes. Out of 116 studies screened, 42 met the inclusion criteria and were included in this meta-analysis. All but 5 of the trials were less than 52 weeks in duration, with 23 of them lasting 26 weeks or less. Included were subjects without diabetes: a large study of subjects with prediabetes (DREAM study), 2 with psoriasis and 1 with Alzheimer disease. A major limitation of this meta-analysis was the use of summary trial-level data (rather than patient-level data), which made it impossible to conduct time-to-event analysis (ie, how soon an MI occurs after starting study drug with earlier events providing higher contribution to the overall risk) or to evaluate the time course of risks. In addition, there was no standard method to identify or validate (adjudicate) CV outcomes in these trials, and there were fewer events than expected for what is considered a high-risk population. This may be related to the fact that glycemic control (and not CV morbidity or mortality) was the primary outcome chosen for most of these trials.