OBJECTIVES: To determine the association of gastrointestinal (GI) bleed over one year in a cohort of Medicare/Medicaid dual eligibles (Duals) taking warfarin alone or warfarin & non-steroidal anti-inflammatory drugs (NSAIDS) and determine the annual health care costs for patients with GI bleed versus no GI Bleed. METHODS: Data were obtained from Arizona State Medicaid Agency from January 1, 2005 through December 31, 2005.  Inclusion criteria consisted of Duals ³65 and <90 years of age with at least two claims for warfarin.  Duals were grouped as taking warfarin alone or warfarin plus NSAID.  Factors examines included age, sex, race/ethnicity membership, rural or urban residence, with adjustment for total number of prescription claims and comorbidities.  Analysis was performed utilizing a logistic regression model to determine odds of GI bleed when taking warfarin compared to warfarin plus NSAID. RESULTS: A total of 2916 patients met inclusion/exclusion criteria: 2538 taking warfarin alone and 378 taking warfarin plus NSAID.  Males comprised 771 (30.4%) of warfarin and 88 (23.3%) of warfarin plus NSAID groups.  Mean age (SD) was 76.6 (6.8) and 74.0 (6.1) in the warfarin and warfarin plus NSAID groups respectively.  There were 126 (5.0%) GI bleed events in the warfarin and 29 (7.7%) events in the warfarin plus NSAID groups (p=0.02).  The odds ratio for GI bleed was 1.51 (0.98-2.33).  Mean total annual health care costs (SD) were $13,232 (15,476) in the warfarin and $10,195 (12,780) in the warfarin and NSAID groups (p<0.01). CONCLUSIONS: This analysis did not identify a difference in the odds of having a GI bleed when on warfarin or warfarin plus NSAIDS.  Unexpectedly, there were significantly lower costs for the warfarin plus NSAIDs group.  These results should be interpreted with caution due to the small sample size and limitations of retrospective analysis.