COX-2 inhibitors are as effective as non-selective NSAIDs and are well tolerated. Naproxen reduces prostaglandin synthesis in the gastric mucosa by 70% whereas rofecoxib has no effect. Endoscopic studies have shown that rofecoxib is associated with significantly less risk of upper gastrointestinal complications than are the non-selective drugs in patients with osteoarthritis. Similar findings have been reported in patients with rheumatoid arthritis. Now a large international study has produced similar results.

At 48 centres in 18 countries a total of 660 patients with rheumatoid arthritis were randomised to rofecoxib 50 mg once daily, naproxen 500 mg twice daily, or placebo for 12 weeks. Endoscopy was performed at baseline and at 6 and 12 weeks. Injectable or oral steroid treatment (equivalent to prednisone 10 mg or less daily) was allowed. The incidence of gastroduodenal ulcers 3 mm or less in diameter by 12 weeks was significantly higher on naproxen (25.5%) than on rofecoxib (6.8%) or placebo (2.9%). For naproxen v rofecoxib p<0.001, naproxen v placebo P<0.001, rofecoxib v placebo p = 0.066). Similar results were obtained using ulcers of 5 mm or more diameter or number of gastroduodenal erosions as end points. The cumulative incidence of ulcers of 3 mm or greater diameter at 12 weeks in H pylori negative patients was 3.6% (placebo), 3.9% (rofecoxib), and 33.3% (naproxen). For H pylori positive patients the corresponding figures were 2.3%, 8.3%, and 17.9%. Adverse events were reported in 61.1%, 62.1%, and 66.4% (no significant differences) and drug related adverse events in 25.8%, 35.6%, and 39.5% (both active drugs significantly worse than placebo). Serious adverse events occurred in 2.7%, 1.8%, and 4.1% (no significant differences). Epigastric discomfort and discontinuation because of an adverse event were both significantly more frequent in the naproxen group but there were no significant group differences in the incidence of dyspepsia or heartburn. Perforation, ulcer, or bleeding (PUBs) occurred in 1/221 (placebo), 2/219 (rofecoxib), and 4/220 (naproxen). There was a small increase in hypertension (2.3% vs 6.4% vs 0.9%) with rofecoxib but it usually had only minor clinical significance and no patient withdrew because of it.

Rofecoxib 50 mg daily (twice the usual dose) caused less upper gastrointestinal upset than naproxen 500 mg twice daily.