To estimate the risk of upper gastrointestinal complications associated with use of cyclooxygenase-2 (COX-2) selective (celecoxib and rofecoxib) and individual nonselective nonsteroidal antiinflammatory drugs (NSAIDs) compared with nonuse of these drugs.
Nested case-control study.
Administrative health care databases of Saskatchewan, Canada.
Among a population of men and women aged 20-89 years who were covered by public health insurance with prescription drug benefits between November 15, 1999, and December 31, 2001, 726 case patients with first hospitalization for upper gastrointestinal complications (with validation of cases through review of hospital medical records) were confirmed from 1,054,532 person-years of follow-up, and 20,002 control patients were randomly selected from all eligible controls, frequency matched to cases on their index date (+/- 3 mo).
Logistic regression was used to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between upper gastrointestinal complications and use of NSAIDs. Current rofecoxib and naproxen users had the highest risk for upper gastrointestinal complications with adjusted ORs of 3.6 (95% CI 2.2-5.7) and 3.4 (95% CI 1.8-6.7), respectively. No association was found between the risk of upper gastrointestinal complications and use of celecoxib (OR 1.1, 95% CI 0.7-1.8) or the use of diclofenac plus misoprostol (OR 0.7, 95% CI 0.3-1.8). A dose-response relationship was observed for rofecoxib and naproxen with ORs for high dose of 5.2 (95% CI 2.5-10.6) and 5.1 (95% CI 2.1-12.3), respectively. Short-term users of celecoxib and naproxen had a higher risk than long-term users, whereas among users of rofecoxib the risk was higher among long-term than short-term users.
These findings support the variability of individual NSAIDs in the elevated risk of upper gastrointestinal complications. Our results suggest that the risk for rofecoxib is similar to that for naproxen. Celecoxib users appear to have a similar risk for upper gastrointestinal complications as nonusers; however, the risk may be increased at the start of treatment with celecoxib.