Skip navigation

Celecoxib for OA and RA


Clinical bottom line

Celecoxib was as effective as NSAIDs, and had a modest effect in reducing serious GI adverse events.


JJ Deeks et al. Efficacy, tolerability and upper gastrointestinal safety of celecoxib for treatment of osteoarthritis and rheumatoid arthritis: systematic review of randomised controlled trials. BMJ 2002 325: 619


This review included randomised and double blind trials, and for efficacy had to use licensed doses for at least 12 weeks and be compared with placebo or standard dose NSAID in patients with OA and RA. For safety, doses higher than the licensed dose were included. Because of the scope of the review, many different outcomes were reported. All sponsored trials completed by May 25 2000 were included.


There were reports of 17 trials. Eight were excluded because of a follow up of less than 12 weeks. The nine included trials had 15,187 patients. All were properly randomised and blind. Withdrawal rates after 12 weeks of treatment are shown in Table 1:

Table 1: Withdrawal at 12 weeks


Withdrawals (%)



Celecoxib 200 mg


Celecoxib 400 mg


Naproxen 100 mg


Diclofenac 150 mg


Efficacy in RA

Celecoxib was better than placebo for ACR 20 (NNT 8 with 95% CI 5.7 to 13; 33% of patients had an ACR 20 response at 12 weeks in all trials), and for painful and swollen joints. There was no difference in efficacy between celecoxib and NSAID.

Efficacy in OA

Celecoxib was better than placebo, but no different from NSAID for pain scores, stiffness, physical functioning or the WOMAC composite score.

Endoscopic ulcers

Celecoxib was no different from placebo for endoscopic ulcers at 12 weeks, but was much better than NSAID (Figure 1).

Figure 1: Endoscopic ulcers

Endoscopic ulcers and aspirin

Concomitant prophylactic aspirin increased the rate of endoscopic ulcers with celecoxib, but not to the high levels seen with NSAIDs (Figure 2).

Figure 2: Endoscopic ulcers with and without prophylactic aspirin

Adverse event withdrawals

Compared with placebo, more patients with celecoxib withdrew because of adverse events and GI adverse events, but not more abdominal pain, diarrhoea dyspepsia nausea or vomiting (Table 2).

Compared with NSAID, fewer patients with celecoxib withdrew because of GI adverse events abdominal pain and dyspepsia (Table 2).

Table 2: Adverse events (NNH at 12 weeks)

Adverse event

Celecoxib vs placebo

Celecoxib vs NSAID

Any event

39 (23 to 124)

no difference

Any GI event

83 (44 to 671)

-33 (-23 to -55)

Abdominal pain

no difference

-92 (-57 to -243)


no difference

no difference


no difference

-186 (-85 to 934)


no difference

no difference


no difference

no difference

Perforations, ulcers and bleeds

Serious upper gastrointestinal events were reported from the CLASS study, in which there were 11/3987 with celecoxib and 20/3981 with NSAID. This difference was not significant.

When patients withdrawn from the study for safety reason and found, on endoscopy to have ulcers were include, the numbers were 30/3987 and 49/3981. The difference here did reach statistical significance with a relative risk of 0.6 (0.4 to 0.96). The NNT to prevent one serious upper gastrointestinal event plus ulcers was 209 (109 to 2306).


The CLASS study, with the main information of serious upper GI events, has been heavily criticised because it reported six month data, rather than information for the full period of the trials. Many (1,739) patients were taking aspirin, which may have complicated interpretation, because there was greater reduction in serious GI events in those not taking aspirin.

Overall, celecoxib was as effective as NSAIDs, and had a modest effect in reducing serious GI adverse events.