This article has a correction

Please see: Gut 2004;53:915Gut 2003;52:1800

Gut 52:820-826 doi:10.1136/gut.52.6.820
  • Stomach

Incidence of gastroduodenal ulcers in patients with rheumatoid arthritis after 12 weeks of rofecoxib, naproxen, or placebo: a multicentre, randomised, double blind study

  1. C J Hawkey1,
  2. L Laine2,
  3. T Simon3,
  4. H Quan3,
  5. S Shingo3,
  6. J Evans3,
  7. on behalf of the Rofecoxib Rheumatoid Arthritis Endoscopy Study Group
  1. 1University Hospital, Nottingham, United Kingdom
  2. 2USC School of Medicine, Los Angeles, CA, USA
  3. 3Merck Research Laboratories, West Point, PA, USA
  1. Correspondence to:
    Professor C J Hawkey, School of Medical and Surgical Sciences, Division of Gastroenterology, University Hospital, Nottingham NG7 2UH, UK;
  • Accepted 17 December 2002


Background: Previous studies in patients with osteoarthritis have suggested that the selective cyclooxygenase (COX)-2 inhibitor rofecoxib results in less gastrointestinal damage than non-selective non-steroidal antiinflammatory drugs (NSAIDs). This study compared the incidence of endoscopically detected gastroduodenal ulcers in rheumatoid arthritis patients treated with rofecoxib or a non-selective NSAID.

Methods: In this multicentre, randomised, double blind, 12 week study, patients with rheumatoid arthritis were allocated to rofecoxib 50 mg once daily (n=219), naproxen 500 mg twice daily (n=220), or placebo (n=221). Endoscopy was performed at baseline and at six and 12 weeks. Lifetable analysis and log rank tests were used to analyse the incidence of gastroduodenal ulcers ≥3 mm. Gastric or duodenal ulcers ≥5 mm and erosions were also evaluated as secondary end points. Tolerability was assessed by adverse events.

Results: The cumulative incidence of ulcers ≥3 mm at 12 weeks was significantly higher in patients on naproxen (25.5%) than in patients receiving rofecoxib (6.8%; difference 18.7% (95% confidence interval (CI) 11.7%, 25.7%); p<0.001) or placebo (2.9%; difference 22.6% (95% CI 16.1%, 29.1%); p<0.001). The difference between rofecoxib (6.8%) and placebo (2.9%) did not reach statistical significance (p=0.066). Results were similar for ulcers ≥5 mm and for mean changes from baseline in the number of gastroduodenal erosions. The overall incidence of clinical adverse events was similar among treatment groups (61% of patients on placebo, 62% in patients on rofecoxib, and 66% in patients on naproxen).

Conclusions: Rofecoxib 50 mg daily (twice the dose recommended for this patient population) resulted in a lower incidence of endoscopically detected gastroduodenal ulcers and erosions than treatment with naproxen 500 mg twice daily.