OBJECTIVE: The aim of this study was to assess the rate of upper gastrointestinal (UGI) ulcer complications (bleeding, perforation, or gastric outlet obstruction) associated with celecoxib, a specific COX-2 inhibitor, compared with the rate associated with nonspecific, nonsteroidal anti-inflammatory drugs (NSAIDs). METHODS: A pooled analysis was conducted of 14 multicenter, double-blind, randomized, controlled trials (RCTs) and a separate analysis of one long-term open label trial that assessed the efficacy and safety of celecoxib for symptomatic treatment of arthritis. The RCTs enrolled 11,008 patients with osteoarthritis or rheumatoid arthritis treated for 2-24 wk; the long-term open label trial enrolled 5,155 patients receiving celecoxib for a maximum of 2 yr. In the RCTs, patients were randomly assigned to receive placebo (n = 1,864; 208 patient-years), celecoxib 25-400 mg b.i.d. (n = 6,376; 1,020 patient-years), or a comparator NSAID (n = 2,768; 535 patient-years); NSAIDs were naproxen 500 mg b.i.d., diclofenac 50 or 75 mg b.i.d., or ibuprofen 800 mg t.i.d.). In the long-term, open-label trial, patients received celecoxib 100-400 mg b.i.d. for up to 2 yr (n = 5,155; 5,002 patient-years). The principal outcome measure of this analysis was development of a UGI ulcer complication, which was prospectively defined as bleeding, perforation, or gastric outlet obstruction. Ulcer complications were assessed and adjudicated by persons blinded to the patient's treatment assignment or the study in which the patient participated. RESULTS: In the RCTs, UGI ulcer complications occurred in no placebo patients (0 of 1,864 patients), in 2 of 6,376 celecoxib patients (0.03%), and in 9 of 2,768 patients receiving an NSAID (0.33%), corresponding to annual incidences of 0.20% for celecoxib (p > 0.05 vs placebo) and 1.68% for NSAIDs (p = 0.002 vs celecoxib and placebo). In the long-term open-label trial, nine UGI ulcer complications occurred, for an incidence of 0.17% and an annualized incidence of 0.18%. CONCLUSIONS: The incidence of UGI ulcer complications associated with celecoxib was 8-fold lower than with nonspecific NSAIDs. The incidence of ulcer complications observed in celecoxib-treated patients was similar to that in patients receiving placebo in the RCTs, and to that in non-NSAID users reported in the literature. (C) 2000 by Am. Cell. of Gastroenterology.
