OBJECTIVE: Nonsteroidal antiinflammatory drug (NSAID) related gastrointestinal (GI) and renal adverse events are commonly reported. Although published guidelines recommend periodic laboratory monitoring, NSAID safety practices of physicians have not been investigated at a population level. We examined the associations of physician specialty and patient characteristics with NSAID safety practices. METHODS: Using administrative data and medical record review from a regional managed care organization, we studied a retrospective cohort of 373 frequent NSAID users (> or = 3 consecutive NSAID prescriptions and > or = 1 month of continuous NSAID use and followup). NSAID safety measures included: complete blood count (CBC) testing, creatinine testing, use of GI cytoprotective agents, and lack of simultaneous prescriptions for different NSAID (NSAID overlap). RESULTS: The mean duration of cumulative NSAID use was 14.4 +/- 7.7 months/patient, patient age was 62.0 +/- 11.4 years, and 63% were women. About two-thirds of patients received CBC (238, 63.8%) and creatinine monitoring (263, 70.5%), one-third (120, 32.2%) were prescribed cytoprotective agents, and one-fourth (97, 26%) had at least one NSAID overlap. After multivariable adjustments, concomitant use of disease-modifying antirheumatic drugs (OR 2.5, 95% CI 1.1-5.8), longer NSAID exposure (OR 1.3, 95% CI 1.1-1.4), and a greater number of physician visits/year (OR 1.1, 95% CI 1.0-1.2) were significantly associated with receipt of a CBC. A history of hypertension (OR 2.0, 95% CI 1.2-3.2), longer NSAID exposure (OR 1.3, 95% CI 1.2-1.4), and more physician visits/year (OR 1.1, 95% CI 1.0-1.2) were significantly associated with serum creatinine testing. Rheumatologists, and to a lesser extent internists, trended toward more NSAID toxicity monitoring than family/general practitioners. However, family/general practitioners and internists were more likely to monitor creatinine than rheumatologists among patients with renal risk factors. CONCLUSION: While rheumatologists and internists trended toward more CBC and creatinine testing, visit frequency, duration of NSAID use, and comorbidities were the factors most consistently associated with safety monitoring.