OBJECTIVES: Gastrointestinal (GI) complications are the most-common serious adverse reactions associated with nonsteroidal anti-inflammatory drugs (NSAIDs). We quantified the effect of specific NSAIDs on the rate of GI hospitalizations among older people living in long-term care. DESIGN: Retrospective cohort study. SETTING: All Medicare/Medicaid certified nursing homes in four states (Maine, Minnesota, New York, and South Dakota). PARTICIPANTS: We identified 125,516 newly admitted residents from a database of all residents (1992–1996) of all Medicare/Medicaid certified nursing homes in four states. Using the federally mandated Minimum Data Set, which includes information on all drugs received (prescription and over-the-counter), we identified patients who received at least one prescription for aspirin (n = 19,101) or NSAIDs (n = 9,777). The control population consisted of all institutionalized persons who did not receive these drugs. MEASUREMENTS: From Health Care Financing Administration inpatient claims, we identified the first hospitalization for GI perforation, ulcer, or hemorrhage that occurred during the year of follow up (ICD9-CM discharge codes: 531–534, 578). Cox proportional hazards models provided adjusted estimates of rate ratios. RESULTS: NSAID exposure increased the GI-event-related hospitalization rate in both men (rate ratios (RR) = 2.64; 95% confidence interval (CI) = 1.17–5.99) and women (RR = 3.23; 95% CI = 1.85–5.65). The rate of GI hospitalizations for both men and women taking sulindac, naproxen, or indomethacin was higher than for nonusers. The risk of GI-event-related hospitalizations was greatest among women exposed to diflunisal (RR = 6.08; 95% CI = 2.27–16.26) or oxaprozin (RR = 6.03; 95% CI = 2.49–14.58). CONCLUSIONS: Despite the high background rate of GI events, most NSAIDs increased the risk of GI hospitalization. Careful attention to choice of agent and dosing is needed in prescribin... [ABSTRACT FROM AUTHOR]