The annual incidence of clinically important gastrointestinal complications resulting from therapy with nonsteroidal anti-inflammatory drugs (NSAIDs) approaches 2 percent in patients with arthritis. Endoscopic studies of such patients suggest that the incidence of NSAID–induced gastroduodenal ulcers ranges from 15 to 30 percent. Cyclooxygenase-2 (COX-2) inhibitors appear to have fewer gastrointestinal side effects than NSAIDs. Hawkey and associates conducted a comparative study of the gastrointestinal effects of the COX-2 inhibitor rofecoxib, ibuprofen and placebo in patients with osteoarthritis.

The randomized double-blind study included 775 patients who were assigned to one of four treatment groups: 800 mg of ibuprofen three times daily (193 patients), 25 mg of rofecoxib daily (195 patients), 50 mg of rofecoxib daily (193 patients) and placebo (194 patients). The duration of therapy was 16 to 24 weeks. Baseline endoscopic examination was performed in all patients after they had discontinued their usual anti-inflammatory medications. They were reevaluated endoscopically at weeks 6, 12 and 24 of the study. The final outcome measure was the presence or absence of gastroduodenal ulcers at the 12th and 24th weeks of therapy.

Significantly fewer patients who received rofecoxib developed ulcers compared with those who received ibuprofen. The cumulative incidence of ulcers 3 mm or greater in size after 24 weeks of therapy was 9.9 percent in patients who received 25 mg of rofecoxib and 12.4 percent in those who received 50 mg of the COX-2 inhibitor. In contrast, in the ibuprofen group, the incidence of ulcers 3 mm or greater in size was 46.8 percent after 24 weeks of therapy. The incidence of ulcers in patients who received 25 mg of rofecoxib was comparable to that in the placebo group.

The authors conclude that rofecoxib in a dosage two to four times greater than the therapeutically effective dosage causes fewer ulcers than ibuprofen. The effects on the gastroduodenal mucosa of a daily dosage of 25 mg (the highest dosage recommended for treatment of osteoarthritis) were similar to those associated with placebo.