The use of nonsteroidal anti-inflammatory drugs (NSAIDs) for treatment of arthritis is well established, but the side effect of gastrointestinal (GI) bleeding causes significant morbidity and mortality. Up to 2,500 deaths in Britain are attributed to NSAID-induced GI bleeding annually. The cyclooxygenase-2 (COX-2) enzyme-inhibiting drugs were developed to provide symptom relief without GI effects. Deeks and colleagues systematically reviewed all available evidence to compare the effectiveness and GI tolerability of a leading COX-2 drug, celecoxib, with NSAIDs in the treatment of arthritis.

The authors sought all published and unpublished trials through electronic searches plus contacts with drug manufacturers and registries of trials. The analysis included all randomized, double-blind controlled trials using licensed dosages of celecoxib for at least 12 weeks to treat active rheumatoid arthritis or osteoarthritis. Trials against placebo were included, as well as those comparing celecoxib with NSAIDs.

Clinical outcomes were assessed using the Western Ontario and McMaster universities osteoarthritis index for pain, stiffness, and physical function, and the American College of Rheumatology (ACR-20) index for pain, tenderness, and swelling of joints. Tolerability was assessed based on the rates of study withdrawal because of adverse effects, the incidence of ulcers in those studies using routine endoscopy, and reports of symptomatic GI ulcers, bleeding, perforation, and obstruction. Of the 17 trials identified, nine met inclusion criteria and were of high quality.

Clinical improvement was significantly better in patients treated with celecoxib than in those treated with placebo. Compared with NSAIDs (naproxen, in a dosage of 500 mg twice daily, and diclofenac, in a dosage of 75 mg twice daily), celecoxib showed non-significantly better outcomes in ACR-20 and other clinical outcome measures in patients with rheumatoid arthritis. In patients with osteoarthritis, there were no significant differences between those treated with NSAIDs and those treated with celecoxib; both were superior to placebo in relieving clinical symptoms.

After 12 weeks of treatment, 56 percent of placebo-treated patients dropped out, compared with 39 percent of celecoxib at 200 mg per day and naproxen at 1,000 mg per day; 32 percent for celecoxib at 400 mg per day; and 26 percent for diclofenac at 150 mg per day. Patients taking placebo were more likely to withdraw because of lack of improvement in symptoms than because of adverse events.

When compared with placebo, celecoxib was significantly associated with withdrawal caused by adverse effects and GI effects (relative risk, 1.49 and 1.68, respectively). Overall, celecoxib and NSAIDs did not differ in the number of withdrawals for all adverse events, but celecoxib had significantly fewer withdrawals for GI events, especially dyspepsia and abdominal pain. The number needed to treat for benefit at three months was 35.

Ulcers occurred three times more often in patients treated with celecoxib than in those treated with placebo, but the incidence of ulcers in patients treated with celecoxib was 71 percent lower than in patients treated with NSAIDs. At three months, the number needed to treat for benefit was six. This reduction also was evident in a study using endoscopy at six months. The rate of serious GI complications (i.e., bleeding, perforation, or obstruction) in patients taking celecoxib was nearly one half that of patients using NSAIDs, but this difference was not statistically significant.

The authors conclude that celecoxib is as effective as NSAIDs in the symptomatic treatment of both rheumatoid arthritis and osteoarthritis and is associated with significantly fewer GI-related adverse outcomes.

editor's note: Although this study and a Canadian study of elderly patients (Mamdani M, et al. Observational study of upper gastrointestinal haemorrhage in elderly patients given selective cyclooxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs. BMJ September 21, 2002;325:624–9) published in the same issue add more evidence that COX-2 inhibitors offer greater GI safety than NSAIDs, an accompanying editorial (Jones R. Efficacy and safety of COX 2 inhibitors. [Editorial]. BMJ September 21, 2002;325:607–8) points out that the appropriate use of these agents remains somewhat controversial. Studies have been criticized for design flaws and undue influence from drug manufacturers. In addition, one member of this class of drugs (rofecoxib) has been associated with increased cardiovascular events. The stakes in this controversy are very high. As the baby boomer generation ages, the market for safe, effective antiarthritic medications is enormous. Neither the British meta-analysis nor the Canadian study considers costs, yet in the United States, cost-effectiveness is a serious factor. None of us wants to place patients at risk because of cost, but even COX-2 agents have side effects. We risk compliance problems or issues of patients electing not to purchase other drugs, in addition to financially draining our patients, their families, and health care systems when we prescribe an expensive agent. As in other areas, good communication and the ability to negotiate the very best therapeutic decision in a complex and ambiguous situation are the mark of a true family physician. How can we communicate this necessity to our students and residents?—a.d.w.