Selective COX-2 inhibitors generally have a more favorable gastrointestinal risk profile than traditional, non-selective nonsteroidal anti-inflammatory drugs (NSAIDs). Recommendations for long-term anti-inflammatory treatment suggest adding a proton pump inhibitor (PPI) when prescribing a traditional NSAID to reduce upper gastrointestinal risks (Am J Gastroenterol 2008 Nov;103(11):2908). A new randomized trial with 4,484 patients with osteoarthritis or rheumatoid arthritis reports that a COX-2 inhibitor alone (celecoxib) may reduce some adverse gastrointestinal events compared to the combination of a traditional NSAID and a PPI.
Patients > 60 years old (or > 18 years old with previous gastroduodenal ulcer) were randomized to celecoxib vs. diclofenac plus omeprazole for 6 months. All patients were free of H. pylori and cardiovascular disease at baseline. The primary outcome was the rate of "clinically significant upper and lower gastrointestinal events," including hemorrhage or perforation, gastric-outlet obstruction, or a decrease > 20 g/L in hemoglobin or ≥ 10% in hematocrit without an identified non-gastrointestinal source (defined by the authors as "clinically significant gastrointestinal anemia"). The celecoxib group had a significantly lower rate of the primary outcome (0.9% vs. 3.8%, p < 0.0001, NNT 35). This difference was due to differences in the rates of "clinically significant gastrointestinal anemia." It was not clear if the anemia resulted in clinical symptoms or was identified solely by monitoring for the study, so this is considered a surrogate outcome (level 3 [lacking direct] evidence). There were no events of gastric-outlet obstruction or perforation in either group.
Celecoxib was also associated with lower rates of moderate to severe abdominal symptoms (16% vs. 19%, p = 0.03, NNT 34) and early withdrawal for gastrointestinal effects (6% vs. 8%, p = 0.0006, NNT 50). These outcomes were considered level 2 [mid-level] evidence because of unclear reporting for a substantial number of patients who discontinued medication for reasons classified as "other" (Lancet 2010 Jul 17;376(9736):173).
A significant caveat to this study is that patients on antiplatelet or anticoagulant drugs were excluded. Previous trials have shown no reduction in gastrointestinal toxicity for selective COX-2 inhibitors compared to traditional NSAIDS in patients treated with low-dose aspirin or warfarin (Gastroenterology 2004 Aug;127(2):395, Arch Intern Med 2005 Jan 24;165(2):189).
