Context: Evidence that rofecoxib increases the risk of myocardial infarction has led to scrutiny of other nonsteroidal anti-inflammatory drugs (NSAIDs). Regulatory agencies have provided variable advice regarding the cardiovascular risks with older nonselective NSAIDs.

Objective: To undertake a systematic review and meta-analysis of controlled observational studies to compare the risks of serious cardiovascular events with individual NSAIDs and cyclooxygenase 2 inhibitors.

Data sources: Searches were conducted of electronic databases (1985-2006), scientific meeting proceedings, epidemiological research Web sites, and bibliographies of eligible studies.

Study selection: Eligible studies were of case-control or cohort design, reported on cardiovascular events (predominantly myocardial infarction) with cyclooxygenase 2 inhibitor, NSAID use, or both with nonuse/remote use of the drugs as the reference exposure. Of 7086 potentially eligible titles, 17 case-control and 6 cohort studies were included. Thirteen studies reported on cyclooxygenase 2 inhibitors, 23 on NSAIDs, and 13 on both groups of drugs.

Data extraction: Two people independently extracted data and assessed study quality with disagreements resolved by consensus.

Data synthesis: Data were combined using a random-effects model. A dose-related risk was evident with rofecoxib, summary relative risk with 25 mg/d or less, 1.33 (95% confidence interval [CI], 1.00-1.79) and 2.19 (95% CI, 1.64-2.91) with more than 25 mg/d. The risk was elevated during the first month of treatment. Celecoxib was not associated with an elevated risk of vascular occlusion, summary relative risk 1.06 (95% CI, 0.91-1.23). Among older nonselective drugs, diclofenac had the highest risk with a summary relative risk of 1.40 (95% CI, 1.16-1.70). The other drugs had summary relative risks close to 1: naproxen, 0.97 (95% CI, 0.87-1.07); piroxicam, 1.06 (95% CI, 0.70-1.59); and ibuprofen, 1.07 (95% CI, 0.97-1.18).

Conclusions: This review confirms the findings from randomized trials regarding the risk of cardiovascular events with rofecoxib and suggests that celecoxib in commonly used doses may not increase the risk, contradicts claims of a protective effect of naproxen, and raises serious questions about the safety of diclofenac, an older drug.