Celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: A touch of CLASS?

Original Citation

Gastrointestinal toxicity with celecoxib vs. nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study: A randomized controlled trial. Celecoxib Long-term Arthritis Safety Study. Silverstein FE, Faich G, Goldstein JL, Simon LS, Pincus T, Whelton A et al.

Overall Study Question

The primary objective of this trial was to compare the annualized incidence of upper GI ulcer complications (gastric or duodenal perforation, gastric outlet obstruction or upper GI bleeding) in osteoarthritis (OA) or rheumatoid arthritis (RA) adult patients treated with celecoxib (at dosages 2-4-fold higher than recommended by the FDA) or an NSAID (diclofenac or ibuprofen).

Are the Results of the Study Valid?

Was assignment of patients randomized?

Yes.  Patients were randomly assigned to receive celecoxib 400mg twice daily; ibuprofen 800 mg 3 times daily; or diclofenac 75 mg twice daily.

Were all patients who entered the trial properly accounted for and attributed at its conclusion?

Yes.  Of the 8,059 patients randomized, 3,987 received celecoxib and 3,981 received NSAID treatment (1,985 ibuprofen and 1,996 received diclofenac).  Approximately  equal numbers (44 and 47 respectively) of patients did not receive treatment as assigned.  When compared to the celecoxib arm, more patients withdrew in the NSAID treatment arms as a result of adverse events (20.6% vs. 18.4%, p=0.01) or lack of therapeutic effect (14.8% vs. 12.6%, p=0.005).  No patients were lost to follow up.  Analysis of the primary outcome was by intention to treat such that all patients who had received at least one dose of the assigned study medication were included.

Were patients, their clinicians, and study personnel ‘blind’ to treatment?

Yes.  All treatment regimens were double-blinded and double-dummy.  Unblinding occurred in three instances, although none were related to a study outcome.

Were the groups similar at the start of the trial?

Yes.  Baseline characteristics between the celecoxib and NSAID groups (combined) did not differ significantly.  All measured potential confounding variables were equally balanced across groups.

Aside from the experimental intervention, were the groups treated equally?

Yes.  All patients were treated equally from both diagnostic and therapeutic  perspectives. Outcomes were adjudicated by a blinded GI events committee not  involved in the care of the patient.

Overall, are the results of the study valid?


What were the Results?

How large was the treatment effect?

The annualized incidence of upper GI ulcer complications  was 0.76%  in the celecoxib group as compared to 1.45% in the NSAID groups (p=0.09).  This  translates into an ARR of 0.69% and a NNT of 145 patients (CI 60 to -1098) treated for one  year to avoid one upper GI ulcer complication.

Secondary analyses revealed that the annualized incidence of upper GI ulcer  complications plus symptomatic ulcers was 2.08% in the celecoxib arm as compared to 3.54% in the NSAID arms (ARR 1.46%, NNT 69 patients treated for one year to avoid one event, p=0.02).  In patients on concurrent low dose ASA, the potential reduction in GI toxicity of celecoxib as compared to the NSAID was not evident.  However, in non-ASA users, the annualized incidence of upper GI ulcer complications was significantly lower with celecoxib as compared to the NSAID groups (ARR  0.83%, NNT to avoid one upper GI ulcer complication for one year was 121,  p=0.04).

How precise was the estimate of the treatment effect?

For the primary outcome (upper GI ulcer complications) the 95% CI for the RR crossed one indicating the possibility of a null result and thus, an imprecise estimate of the effect.  The 95% CI for the RR for the secondary outcome variables were fairly wide, but none crossed one.

Will the Results Help Me in Caring for My Patients?

Can the results be applied to my patient care?

Yes.  This study is the first to show that there likely is a reduction in  clinically important GI events in patients treated with COX-2 specific  inhibitors.  However, at best, this effect is modest and may nullified by the presence of low dose ASA.

Were all clinically important outcomes considered?

Yes. The outcomes considered were clinically important.

Are the likely treatment benefits worth the potential harms and costs?

This question is really what needs to be addressed with further  research.  It is unclear if the modest reduction in GI events with celecoxib translates into a reduction in overall treatment costs that will compensate for the higher acquisition cost of the drug.


It would appear that the incidence of upper GI ulcer complications plus symptomatic ulcer is lower in patients treated with celecoxib than those on regular non-specific COX inhibitors.  In addition, it would appear that the annualized incidence of upper GI complications is reduced through the use of celecoxib in patients not receiving low dose ASA. However, both of these outcomes were secondary in the CLASS trial.  The primary outcome, the annualized incidence of upper GI complications in all study entrants, was not significantly different between the study arms although there was a trend  for a modest reduction in events in the celecoxib arm.  Since the use of  concomitant anti-ulcer drugs was prohibited, it is unclear how the  differences would manifest in real world use of these agents.  In addition, the cost-effectiveness of celecoxib as compared to traditional non-specific COX inhibitors has not been established.  Considering the widespread use of these agents, further research is warranted to determine both their economic impact and the outcomes achieved when used outside of a clinical trial.


1.  Marra CA, Esdaile JM, Sun H, Anis AH. The cost of COX inhibitors: How selective should we be? J  Rheumatol 2000 (in press)).

Latest Posts