Reference | Study design | Population | Main outcomes | Main results | Relative risk, coxib compared with NSAID (95% CI) |
---|---|---|---|---|---|
Large randomised trials | |||||
Bombardier et al. N Engl J Med 2000 343: 1520–1528 [17] | Randomised trial powered for PUB outcome, comparing 50 mg rofecoxib with 1000 mg naproxen daily | Patients with RA, at least 50 years n = 8,076 | Confirmed clinical upper GI events (perforation, bleeding symptomatic ulcer) | 177 events, 53 complicated Confirmed events at 2.1/100 pt years with rofecoxib, 4.5 with naproxen Complicated 0.6 and 1.4 per 100 pt years | 0.5 (0.3 to 0.6) 0.4 (0.2 to 0.8) |
Silverstein et al. JAMA 2000 284: 1247–1255 [18] | Randomised trial powered for PUB outcome, comparing 800 mg celecoxib with 2400 mg ibuprofen and 150 mg diclofenac daily | Patients with OA or RA, ≥ 18 years n = 8,059 | Confirmed upper GI ulcers and complication (bleeding, perforation, obstruction) | 83 events including symptomatic ulcers, 35 complicated Confirmed events at 2.1/100 pt years with celecoxib, 3.5 with NSAID Complicated at 0.8 and 1.5 per 100 pt years | 0.6 (0.4 to 0.9) 0.5 (0.3 to 1.1) |
Schnitzer et al. Lancet 2004 364: 665–674 [19] | Randomised trial powered for PUB outcome, comparing 400 mg lumiracoxib with 2400 mg ibuprofen and 1000 mg naproxen daily | Patients with OA, at least 50 years n = 18,325 | Confirmed upper GI ulcers and complication (bleeding, perforation, obstruction) | 283 events, 112 complicated Confirmed events at 1.0% with lumiracoxib, 1.5% with NSAID Complicated events at 0.3% with lumiracoxib, 0.9% with NSAID | 0.7 (0.5 to 0.8) 0.3 (0.2 to 0.5) |
Meta-analyses of randomised trials | |||||
Langman et al. JAMA 1999 282: 1929–1933 [20] | Presecified meta-analysis of eight randomised trials of rofecoxib versus NSAIDs | OA patients, mean age 63 years n = 5,435 | Confirmed clinical upper GI events (perforation, bleed, ulcer) | 35 confirmed complicated events Complicated events at 1.3/100 pt years with rofecoxib, 2.6 with NSAID | 0.5 (0.3 to 1.0) |
Goldstein et al. Am J Gastroenterol 2000 95: 1681–1690 [21] | Meta-analysis of 14 randomised trials of celecoxib versus NSAIDs | OA or RA patients, mean age 60 years n = 11,008 | Confirmed clinical upper GI events (perforation, bleed, ulcer) | 11 confirmed complicated events Complicated events at 0.2/100 pt years with celecoxib, 1.7 with NSAID | 0.2 (0.1 to 0.5) |
Edwards et al. Pain 2004 111: 286–296 [22] | Meta-analysis of nine randomised trials of valdecoxib versus NSAIDs | OA or RA patients, n = 5,726 | Clinically significant upper GI bleed | 10 confirmed complicated events Complicated event rate 0.1% with valdecoxib, 0.4% with NSAID | 0.2 (0.04 to 0.8) |
Goldstein et al. Aliment Pharmacol Ther 2004 20: 527–538 [23] | Meta-analysis of eight randomised trials of valdecoxib versus NSAID | OA or RA patients, mean age 58 years n = 7,434 | Confirmed clinical upper GI events (perforation, bleed, ulcer) | 88 symptomatic ulcers, 19 complicated Symptomatic + complicated 0.8% with valdecoxib, 3.3% with NSAID Complicated 0.2% with valdecoxib, 0.5% with NSAID | 0.3 (0.2 to 0.4) 0.4 (0.1 to 0.9) |
Hooper et al. BMJ 2004 329: 948–952 [24] | Meta-analysis of 17 randomised trials of coxibs versus NSAIDs | n = 25,564 | Variety of outcomes reported, including serious gastrointestinal complications, and symptomatic ulcers | 114 serious gastrointestinal complications, 0.36% with coxib, 0.73% with NSAID 288 symptomatic ulcers, 0.8% with coxib, 1.8% with NSAID | 0.5 (0.4 to 0.8) 0.5 (0.4 to 0.6) |
Moore et al. Arth Res Ther 2005 7:R644–R655 [25] | Meta-analysis of 31 randomised trials of celecoxib versus NSAIDs | OA or RA n = 39,605 (31,171 in analysis of ulcers and bleeds) | Variety of outcomes reported including clinical ulcers and bleeds | 184 clinical ulcers or bleeds, 0.4% with celecoxib, 0.9% with NSAID | 0.6 (0.5 to 0.8) |
Large observational studies | |||||
Mamdani et al. BMJ 2002 325: 624–630 [26] | Observational cohort study | Users of NSAID, coxib, or non users. Total population about 144,000 | Hospital admission for upper gastrointestinal bleeding | 82 events with controls, 17 with NSAID, 75 with coxib Rofecoxib, but not celecoxib had significantly greater association with bleeding than controls | Celecoxib compared with NSAID 0.2 (0.1 to 0.4) Rofecoxib compared with NSAID 0.5 (0.3 to 1.0) |
MacDonald et al. Gut 2003 52: 1265–1270 [27] | Retrospective cohort analysis | Users of NSAID, coxibs, and non users. Total 26,000 incident cases of upper gastrointestinal haemorrhage | Hospital admission for upper gastrointestinal bleeding in high risk patients | 2,875 events on NSAID, 4 on coxib Adjusted relative risk | 0.4 (0.1 to 1.0) |
Norgard et al. Aliment Pharmacol Ther 2004 19: 817–825 [28] | Population based case-control study | Users of NSAID, coxibs, and non users. 780 incident cases in patients with high risk of gastrointestinal bleeding | Hospital admission for upper gastrointestinal bleeding | 35 patients had been exposed to coxib (4.5%) 97 patients had been exposed to NSAID (12%) Rofecoxib, but not celecoxib had significantly greater association with bleeding than controls | 0.4 (0.3 to 0.5) |