Determinants and sequelae associated with utilization of acetaminophen versus traditional nonsteroidal antiinflammatory drugs in an elderly population.

Acetaminophen is recommended as initial therapy for patients with arthritis, particularly those at increased risk of nonsteroidal antiinflammatory drug (NSAID)-induced gastrointestinal (GI) side effects. However, higher doses of acetaminophen inhibit prostaglandin synthesis and have been associated with GI events. This study was undertaken to compare the observed and adjusted rates of GI events (hospitalizations, ulcers, dyspepsia, GI prophylaxis) occurring with higher versus lower doses of acetaminophen. This was a retrospective cohort study of subjects ages >or=65 years who received a prescription for acetaminophen or NSAID between 1994 and 1996. Pharmaceutical and medical records were reviewed for 1 year of historical data prior to the index prescription of acetaminophen or non-aspirin NSAID. Risk factors for GI events were identified based on the historical data. To further control for bias, patients were categorized by propensity score (the likelihood of receiving acetaminophen, given defined risk factor values). Records were then reviewed for the duration of the index prescription or 30 days, whichever was less, to generate data on the occurrence of GI events. Determinants of acetaminophen utilization were identified using logistic regression, and rates of GI events for each therapy were examined using Poisson regression analyses, controlling for duration of exposure, individual risk factors, and propensity scores. The study included 26,978 patients in the NSAID cohort and 21,207 in the acetaminophen cohort. Determinants of acetaminophen utilization compared with NSAIDs (odds ratio [95% confidence interval]) included recent hospitalization (8.6 [7.7-9.5]), concomitant anticoagulation therapy (3.2 [2.7-3.8]), age >85 years (2.3 [2.1-2.4]), and history of prior GI events, especially those requiring hospitalization (14.6 [11.7-18.7]). Unadjusted rates of GI hospitalization, ulcer, and dyspepsia were higher for patients in the acetaminophen cohort than for those in the NSAID cohort. The occurrence of GI events in acetaminophen-treated patients was dose dependent, with rate ratios (compared with high-dose NSAIDs and adjusted for risk susceptibility) ranging from 0.6 (95% confidence interval 0.5-0.7) for <or=650 mg/day to 1.0 (0.9-1.1) for >3,250 mg/day. In this cohort, acetaminophen utilization is more common in patients at higher risk of GI events. After adjustment for risk susceptibility, patients receiving higher doses of acetaminophen have higher rates of GI events compared with those receiving lower doses.