A retrospective evaluation of congestive heart failure and myocardial ischemia events in 14,237 patients with type 2 diabetes mellitus enrolled in 42 short-term, double-blind, randomized clinical studies with rosiglitazone

Pharmacoepidemiol Drug Saf. 2008 Aug;17(8):769-81. doi: 10.1002/pds.1615.


Purpose: Retrospectively investigate potential associations between rosiglitazone and congestive heart failure (CHF) and, separately, events of myocardial ischemia.

Methods: Data from 14 237 individuals in 42 short-term, double-blind, randomized studies of rosiglitazone versus placebo or active diabetes medications were analyzed across seven treatment comparisons using an exact logistic regression model, adjusted for number of major cardiovascular risk factors and duration of exposure.

Results: CHF incidence ranged 0-1.27% (SAEs) and 0.12-2.42% (all AEs) with rosiglitazone versus 0.07-0.75% (SAEs) and 0.25-1.36% (all AEs) with control. Higher odds ratios (95%CI) were observed for CHF SAEs with sulfonylurea- and insulin-containing combinations: rosiglitazone monotherapy versus placebo, 0.25 (<0.01-4.75); rosiglitazone monotherapy versus sulfonylurea/metformin monotherapy, 0.23 (<0.01-2.14); sulfonylurea + rosiglitazone versus sulfonylurea monotherapy, 0.95 (0.01-75.20); metformin + rosiglitazone versus metformin monotherapy, 0.60 (0.00-8.28); metformin + rosiglitazone versus metformin + sulfonylurea, 1.04 (0.39-2.86); sulfonylurea + metformin + rosiglitazone versus sulfonylurea + metformin, 3.15 (0.35-150.52); insulin + rosiglitazone versus insulin monotherapy, 1.63 (0.52-6.01). Myocardial ischemia incidence ranged 0.75-1.40% (SAEs) and 1.49-2.77% (all AEs) with rosiglitazone versus 0.21-2.04% (SAEs) and 0.56-2.38% (all AEs) with control. Each comparison had an OR >1, with wide confidence intervals generally including unity. With data pooling, more events of myocardial ischemia were observed with rosiglitazone (2.00%) versus control (1.53%) (HR 1.30, 95%CI 1.004-1.69).

Conclusions: CHF incidence may be greater when rosiglitazone is combined with sulfonylureas or insulin. When data were pooled, more events of myocardial ischemia were observed with rosiglitazone versus control. Final results from RECORD will allow a more rigorous evaluation of the cardiovascular safety profile.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Diabetes Mellitus, Type 2 / complications*
  • Double-Blind Method
  • Female
  • Heart Failure / chemically induced*
  • Humans
  • Hypoglycemic Agents / adverse effects*
  • Male
  • Middle Aged
  • Myocardial Ischemia / chemically induced*
  • Odds Ratio
  • Randomized Controlled Trials as Topic
  • Retrospective Studies
  • Rosiglitazone
  • Thiazolidinediones / adverse effects*
  • Time Factors


  • Hypoglycemic Agents
  • Thiazolidinediones
  • Rosiglitazone