Gastrointestinal Safety of Nonsteroidal Antiinflammatory Drugs and Selective Cyclooxygenase-2 Inhibitors in Patients on Warfarin

BACKGROUND: The interaction between warfarin and nonsteroidal antiinflammatory drugs (NSAIDs) is well known. However, warfarin and NSAIDs are still commonly prescribed together. Selective cyclooxygenase-2 (COX-2) inhibitors, a newer class of NSAID, offer potential advantages over the nonselective NSAIDs in patients treated with warfarin. OBJECTIVE: To study the rates of hospitalization for gastrointestinal (GI) bleeding events in 3 groups of patients: those taking warfarin only, those taking warfarin plus a nonselective NSAID, and those taking warfarin plus a selective COX-2 inhibitor. METHODS: This was a retrospective cohort analysis in a large nonprofit health maintenance organization. All warfarin users from January 1, 2000, to December 31, 2005, were eligible for inclusion in the study. Eligible patients were grouped by their exposure time to warfarin only, warfarin plus nonselective NSAIDs, or warfarin plus selective COX-2 inhibitor. The study endpoint was hospitalization for a GI bleed. Patients were matched using a propensity scoring methodology. A multivariate Cox proportional hazards model was used to estimate the hazard ratio for GI bleeding between patient cohorts, controlling for age, sex, baseline medical conditions, prior history of GI bleeding, and prescription drug use. RESULTS: The eligible population consisted of 35,548 patients undergoing 46,214 courses of warfarin therapy. The adjusted hazard ratio for hospital-associated GI bleeding in the warfarin plus nonselective NSAID group versus warfarin alone was 3.58 (95% CI 2.31 to 5.55; p < 0.01) and for warfarin plus selective COX-2 inhibitor versus warfarin alone was 1.71 (95% CI 0.60 to 4.84; p = 0.31). For nonselective NSAIDs plus warfarin versus selective COX-2 inhibitor plus warfarin, the adjusted hazard ratio was 3.69 (95% CI 1.42 to 9.60; p = 0.01). CONCLUSIONS: In general, nonselective NSAIDs and selective COX-2 inhibitors should be avoided in patients taking warfarin. In situations where patients require NSAIDs and cannot be managed using other therapies, our results suggest that selective COX-2 inhibitors are associated with fewer hospitalizations for GI bleeding.