Rofecoxib (Vioxx) was recently withdrawn from the market because of increased cardiovascular events in participants taking this drug as part of a trial of prevention of colorectal polyps. An earlier study, the Vioxx Gastrointestinal Outcomes Research trial, showed a fivefold increase in myocardial infarction and included comparisons with naproxen (Aleve). This and other similar reports were attributed to a cardioprotective effect of naproxen rather than a cardiotoxic effect of rofecoxib. Because an estimated 80 million patients have taken rofecoxib, Jüni and colleagues reviewed all randomized controlled trials (RCTs) to establish the magnitude of the risk.

Searches of electronic databases, citation indexes, conference proceedings, and information from experts and U.S. Food and Drug Administration panels were used to identify all RCTs comparing rofecoxib with other nonsteroidal anti-inflammatory drugs (NSAIDs) or placebo for chronic musculoskeletal conditions in adults. All studies were reviewed for quality by two independent reviewers and data were entered into the meta-analysis.

From the 383 reports originally identified, 63 (including 18 RCTs) were included in the study. Of the 18 RCTs, which included 25,273 patients, 12 were concerned with osteoarthritis, five with rheumatoid arthritis, and one with low back pain. The duration of the trials ranged from four weeks to longer than one year. A placebo group was used in 14 trials. Because five trials were extended to report on patients after reallocation of treatment, 22 comparisons were available for the analysis. All of the RCTs were sponsored by the drug manufacturer. The median incidence of myocardial infarction in control groups was 1.45 per 1,000 patient-years. Twelve myocardial infarcts occurred in the control group, compared with 52 in patients taking rofecoxib. The relative risk varied from 1.04 to 2.93, depending on the agent being used for comparison, but it was not significantly altered by the dose of rofecoxib or the duration of the trial (see accompanying table). The relative risk of serious cardiovascular event was 1.55. Relative risk of stroke was 1.02.

The authors conclude that rofecoxib is associated with an increased risk of myocardial infarction in short- and long-term trials. They challenge statements that no excess risk occurred until after 18 months of therapy, and that the risk is dose dependent. They also found no evidence of a significant cardioprotective effect of naproxen. The authors emphasize that few trials included patients with significant history of cardiovascular disease, which is not consistent with patients encountered in normal clinical practice. The relative risk of rofecoxib could be significantly higher in such patients.