Rofecoxib is a relatively new cyclooxygenase-2 (COX-2) inhibitor. Because it does not inhibit the COX-1 isoform, rofecoxib is thought to avoid some of the gastrointestinal (GI) side effects occurring with traditional nonsteroidal anti-inflammatory drugs (NSAIDs) in the treatment of osteoarthritis. Langman and associates conducted this study to compare the incidence of GI tract perforations, ulcerations and bleeding in patients taking rofecoxib and in patients taking NSAIDs.

Eight double-blind controlled trials served as the data sources for this study. The comparison treatments included ibuprofen, placebo, diclofenac and nabumetone. All studies recorded the occurrence of perforation, ulceration, bleeding or other adverse events during treatment or within 14 days of the last ingestion of the study agent.

Of the 5,435 patients included in these studies, 3,357 took rofecoxib, 1,564 took NSAIDs and 514 took placebo. About 10 percent of the patients had a history of GI perforation, ulceration or bleeding. Adverse events caused discontinuation of the study medication in 9.4 percent of the rofecoxib groups and 10.7 percent of the NSAID groups. Of these, 3.5 percent of rofecoxib patients stopped the drug because of GI adverse events, as did 4.8 percent of patients receiving NSAIDs. There were significantly fewer instances of GI dyspepsia in patients receiving rofecoxib compared with patients receiving NSAIDs. Up to six months into treatment, there were significantly fewer confirmed GI perforations, ulcerations and instances of bleeding in patients taking rofecoxib than in patients taking NSAIDs. After six months, the incidence rates converged. This study had the power to determine only whether there were differences between rofecoxib and traditional NSAIDs in general (as opposed to specific NSAIDs). However, the authors conclude that rofecoxib is less likely than traditional NSAIDs to cause GI perforations, ulcerations or bleeding when taken by patients with osteoarthritis.

In an accompanying editorial, Peterson and Cryer point out that this study has an adequate sample size to detect small differences between rofecoxib and NSAIDs. However, it seems that the benefit of rofecoxib, if any, is to be found in its lower rate of adverse events, not in an increased ability to treat pain or inflammation. It would cost nearly $400,000 per 500 low-risk patients to prevent only one ulcer-related complication. However, only about 40 high-risk patients would need to be treated with a COX-2 inhibitor to prevent an ulcer-related complication. High risk is defined as being at least 75 years of age with a history of ulcer or GI tract bleeding. Giving these patients a COX-2 inhibitor would also be more cost effective than using an NSAID plus misoprostol or a proton pump inhibitor (see the accompanying table). Patients at low risk for ulcer complications should still be given a generic NSAID instead of a COX-2 inhibitor.