Risk of myocardial infarction associated with selective COX-2 inhibitors: Meta-analysis of randomised controlled trials

Purpose To evaluate the risk of myocardial infarction (MI) associated with the use of selective cyclooxygenase-2 (COX-2) inhibitors (coxibs). Methods Systematic review and meta-analysis of randomised controlled trials (RCTs) using a fixed-effect model to estimate the odds ratios (ORs) for risk of MI associated with coxibs compared against placebo, non-steroidal anti-inflammatory drugs (NSAIDs) and other coxibs. Results Fifty-five trials (99 087 patients) were included in the meta-analysis. The overall pooled OR for MI risk for any coxib compared against placebo was 1.46 (95%CI: 1.02, 2.09). We found celecoxib, rofecoxib, etoricoxib, valdecoxib and lumiracoxib were associated with higher Ml risks compared against placebo. The pooled OR for any coxib compared against other NSAIDs was 1.45 (95%CI: 1.09,1.93). Rofecoxib had a significantly higher risk of MI than naproxen (OR: 5.39; 95%: 2.08, 14.02) and valdecoxib had lower MI risk than diclofenac (OR: 0.14, 95%CI: 0.03, 0.73). There were no significant differences identified in the risk of MI from the available head-to-head comparisons of coxibs. Conclusions Coxibs were associated with increased risks of MI when compared against placebo or non-selective NSAIDs. Differences in MI risk were also apparent between comparisons of individual NSAIDs. Future work should consider using individual patient data (IPD) meta-analysis to explore differences in Ml risk between different subgroups of patients. Copyright (c) 2007 John Wiley & Sons, Ltd.