BACKGROUND: The adverse gastrointestinal effects of older nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) are well documented. These effects are attributed to the drugs’ inhibition of cyclooxygenase-1 (COX-1), which is found in the gastrointestinal tract. The newer NSAIDs that selectively inhibit cyclooxygenase-2 (COX-2) have been shown to cause fewer endoscopically proven gastrointestinal erosions/ulcerations than nonselective NSAIDs that inhibit both COX-1 and COX-2. This disease-oriented evidence leads to concern about whether selective NSAIDs are really safer in the clinical setting than nonselective NSAIDs. This study attempts to answer the following question: Does rofecoxib, a selective COX-2 NSAID, lead to fewer clinically significant gastrointestinal events than naproxen, a nonselective NSAID?
POPULATION STUDIED: Patients were aged 50 years or older with rheumatoid arthritis (or 40 years or older with rheumatoid arthritis and taking long-term glucocorticoids) and were expected to be taking an NSAID for at least 1 year. Patients were recruited from 301 centers in 22 countries.
STUDY DESIGN AND VALIDITY: This was a randomized double-blinded controlled trial. In general it was a well-designed study with no major flaws. The randomization process and the steps taken to conceal the treatment assignment were not described in detail. A total of 8076 patients were initially assigned to either 50 mg of rofecoxib once daily or 500 mg of naproxen twice daily. Patients were not allowed to take other NSAIDs but could take antacids and H2-receptor antagonists. Median follow-up was 9.0 months. Patients were followed until the study ended even if they stopped taking the study medications. Only 71.1% of the patients continued to take their assigned medication until the end of the study. Rates of discontinuation (29.3% for rofecoxib, 28.5% for naproxen), discontinuation for adverse events (16.4% and 16.1%, respectively), and discontinuation for lack of efficacy (6.3% and 6.5%, respectively) were similar between the groups. A committee using preset criteria and blinded to the treatment assessed the end points.
OUTCOMES MEASURED: Two outcomes were measured: the total number of gastrointestinal events (gastroduodenal perforation or obstruction, upper gastrointestinal bleeding, and symptomatic gastroduodenal ulcer) and the number of complicated events (perforation, obstruction, and severe upper gastrointestinal bleeding). Also, the effectiveness of medication in reducing disease activity was measured using the Global Assessment of Disease Activity questionnaire. Finally, all episodes of confirmed and unconfirmed gastrointestinal bleeding were analyzed.
RESULTS: The risk of a gastrointestinal event was lower for patients receiving rofecoxib for the year of the study period (2.1 vs 4.5 events per 100 patient years, number needed to treat [NNT]=42). The risk of a complicated event was also lower for patients receiving rofecoxib (0.6 vs 1.6 events per 100 patient years, NNT=100). Overall mortality between the 2 groups was similar. Both drugs were equally effective in relieving the symptoms of rheumatoid arthritis. Of note, the rate of myocardial infarction was lower in the rofecoxib groups (0.1%) than in the naproxen group (0.4%); however, the mortality rates from cardiovascular causes were similar between the 2 groups. The difference in the rate of myocardial infarctions has been attributed to the antiplatelet effect (related to COX-1) of traditional NSAIDs, which the selective NSAIDs do not possess.
The risk of gastrointestinal events is lower with rofecoxib than with naproxen in patients treated continuously for 1 year with standard doses. The absolute difference between these 2 agents is small and should be weighed against the increased cost of rofecoxib. A recent study comparing celecoxib with ibuprofen and diclofenac showed similar results.1 Of note in the previous study was that patients taking aspirin for cardiovascular prophylaxis and celecoxib had the same rate of adverse gastrointestinal events as those taking ibuprofen or diclofenac. Even a small amount of aspirin seems to negate the advantage of the selective NSAID. A selective NSAID may be a better choice for patients at high risk for adverse gastrointestinal events and who need long-term treatment, while a nonselective NSAID is probably the better choice for patients who are at low risk and require short-term therapy.