Benefits and harms of celecoxib for treating people with rheumatoid arthritis

Review question

We aimed to assess the benefits (improvement in pain, stiffness, physical function) and harms (gut and heart problems) of celecoxib compared with other similar drugs or a fake drug (placebo) for adults with rheumatoid arthritis.


Rheumatoid arthritis is an auto-immune disease that causes the lining of joints to become inflamed making them painful, stiff and swollen. The small joints of hands and feet are usually affected first. Presently, there is no cure; treatment aims to achieve remission or delay disease progression to improve mobility and reduce pain, swelling and stiffness. Nonsteroidal anti-inflammatory drugs (NSAIDs) are used to treat people with rheumatoid arthritis. Celecoxib is a selective NSAID which may help to relieve symptoms of rheumatoid arthritis.

Search date

We searched for evidence up to May 18, 2017.

Study characteristics

We included eight studies published between 1998 and 2014 that involved 3988 adults (average age 54 years), most of whom were women (73%). Participants had rheumatoid arthritis for an average of 9.2 years.

Studies compared celecoxib with another treatment; 1786 participants received celecoxib and 2202 received either placebo or a traditional NSAID (tNSAID).

Five studies were supported or funded by the pharmaceutical industry.

Key results

When compared with placebo, of every 100 people who received celecoxib, 15 had symptom improvement after 4 to 12 weeks. People who took celecoxib rated their pain 11 points lower (on a scale of 0 to 100) after 12 weeks.

Results were inconclusive about improvements in physical function and numbers of people who developed gastroduodenal ulcers over 3 mm diameter when we compared celecoxib and placebo. Evidence was also inconclusive about harms that appear shortly after use of the drugs. None of the studies that compared celecoxib and placebo reported heart attacks or strokes.

Results about improvement in pain and physical function were inconclusive when celecoxib was compared with tNSAIDs. A small improvement was found in relation to a scoring system (ACR20) used by doctors to assess people with rheumatoid arthritis. Of every 100 people who took celecoxib: 13 fewer developed gastroduodenal ulcers over 3 mm diameter; and 7 fewer withdrew from studies compared with people receiving tNSAIDs.

Celecoxib may improve rheumatoid arthritis symptoms and alleviate pain more than placebo, but probably provides little or no difference in physical function improvement.

Quality of evidence

Evidence was rated as moderate or low quality due to methodological shortcomings and few observed events indicating harms.

Authors' conclusions: 

Celecoxib may improve clinical symptoms, alleviate pain and contribute to little or no difference in physical function compared with placebo. Celecoxib was associated with fewer numbers of participant withdrawals. Results for incidence of gastroduodenal ulcers (≥ 3 mm) and short-term serious adverse events were uncertain; however, there were few reported events for either.

Celecoxib may slightly improve clinical symptoms compared with tNSAIDs. Results for reduced pain and improved physical function were uncertain. Particpants taking celecoxib had lower incidence of gastroduodenal ulcers (≥ 3 mm) and there were fewer withdrawals from trials. Results for cardiovascular events and short-term serious adverse events were also uncertain.

Uncertainty about the rate of cardiovascular events between celecoxib and tNSAIDs could be due to risk of bias; another factor is that these were small, short-term trials. It has been reported previously that both celecoxib and tNSAIDs increase cardiovascular event rates. Our confidence in results about harms is therefore low. Larger head-to-head clinical trials comparing celecoxib to other tNSAIDs is needed to better inform clinical practice.

Read the full abstract...

Rheumatoid arthritis is a systemic auto-immune disorder that causes widespread and persistent inflammation of the synovial lining of joints and tendon sheaths. Presently, there is no cure for rheumatoid arthritis and treatment focuses on managing symptoms such as pain, stiffness and mobility, with the aim of achieving stable remission and improving mobility. Celecoxib is a selective non-steroidal anti-inflammatory drug (NSAID) used for treatment of people with rheumatoid arthritis.


To assess the benefits and harms of celecoxib in people with rheumatoid arthritis.

Search strategy: 

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase and clinical trials registers ( and the World Health Organization trials portal) to May 18, 2017. We also searched the reference and citation lists of included studies.

Selection criteria: 

We included prospective randomized controlled trials (RCTs) that compared oral celecoxib (200 mg and 400 mg daily) versus no intervention, placebo or a traditional NSAID (tNSAID) in people with confirmed rheumatoid arthritis, of any age and either sex. We excluded studies with fewer than 50 participants in each arm or had durations of fewer than four weeks treatment.

Data collection and analysis: 

We used standard methodological procedures expected by The Cochrane Collaboration.

Main results: 

We included eight RCTs with durations of 4 to 24 weeks, published between 1998 and 2014 that involved a total of 3988 adults (mean age = 54 years), most of whom were women (73%). Participants had rheumatoid arthritis for an average of 9.2 years. All studies were assessed at high or unclear risk of bias in at least one domain. Overall, evidence was assessed as moderate-to-low quality. Five studies were funded by pharmaceutical companies.

Celecoxib versus placebo

We included two studies (N = 873) in which participants received 200 mg daily or 400 mg daily or placebo. Participants who received celecoxib showed significant clinical improvement compared with those receiving placebo (15% absolute improvement; 95% CI 7% to 25%; RR 1.53, 95% CI 1.25 to 1.86; number needed to treat to benefit (NNTB) = 7, 95% CI 5 to 13; 2 studies, 873 participants; moderate to low quality evidence).

Participants who received celecoxib reported less pain than placebo-treated people (11% absolute improvement; 95% CI 8% to 14%; NNTB = 4, 95% CI 3 to 6; 1 study, 706 participants) but results were inconclusive for improvement in physical function (MD -0.10, 95% CI 0.29 to 0.10; 1 study, 706 participants).

In the celecoxib group, 15/293 participants developed ulcers, compared with 4/99 in the placebo group (Peto OR 1.26, 95% CI 0.44 to 3.63; 1 study, 392 participants; low quality evidence). Nine (of 475) participants in the celecoxib group developed short-term serious adverse events, compared with five (of 231) in the placebo group (Peto OR 0.87 (0.28 to 2.69; 1 study, 706 participants; low quality evidence).

There were fewer withdrawals among people who received celecoxib (163/475) compared with placebo (130/231) (22% absolute change; 95% CI 16% to 27%; RR 0.61, 95% CI 0.52 to 0.72; 1 study, 706 participants).

Cardiovascular events (myocardial infarction, stroke) were not reported. However, regulatory agencies warn of increased cardiovascular event risk associated with celecoxib.

Celecoxib versus tNSAIDs

Seven studies (N = 2930) compared celecoxib and tNSAIDs (amtolmetin guacyl, diclofenac, ibuprofen, meloxicam, nabumetone, naproxen, pelubiprofen); one study included comparisons of both placebo and tNSAIDs (N = 1149).

There was a small improvement, which may not be clinically significant, in numbers of participants achieving ACR20 criteria response in the celecoxib group compared to tNSAIDs (4% absolute improvement; 95% CI 0% less improvement to 8% more improvement; RR 1.10, 95% CI 0.99 to 1.23; 4 studies, 1981 participants). There was a lack of evidence of difference between participants in the celecoxib and tNSAID groups in terms of pain or physical function. Results were assessed at moderate-to-low quality evidence (downgraded due to risk of bias and inconsistency).

People who received celecoxib had a lower incidence of gastroduodenal ulcers ≥ 3 mm (34/870) compared with those who received tNSAIDs (116/698). This corresponded to 12% absolute change (95% CI 11% to 13%; RR 0.22, 95% CI 0.15 to 0.32; 5 studies, 1568 participants; moderate quality evidence). There were 7% fewer withdrawals among people who received celecoxib (95% CI 4% to 9%; RR 0.73, 95% CI 0.62 to 0.86; 6 studies, 2639 participants).

Results were inconclusive for short-term serious adverse events and cardiovascular events (low quality evidence). There were 17/918 serious adverse events in people taking celecoxib compared to 42/1236 among people who received placebo (Peto OR 0.71; 95% CI 0.39 to 1.28; 5 studies, 2154 participants). Cardiovascular events were reported in both celecoxib and placebo groups in one study (149 participants).