OBJECTIVES: To explore influences of patient and clinical characteristics on the cost-effectiveness of selective COX-2 inhibitors and traditional NSAIDs in patients with osteoarthritis. METHODS: A published cost-effectiveness model (Arthritis Rheum 2003;49:283–92) was modified to incorporate important clinical characteristics that can influence the risk of gastrointestinal (GI) complications such as age, gender, history of GI bleed, and low-dose aspirin intake. The modified model used data from a large clinical trial (TARGET) to estimate base GI event rates. Differences between NSAIDs were modeled from observed rates of GI events and adverse effects after adjusting for differences in population characteristics across three clinical trials (TARGET, CLASS and VIGOR). Other enhancements included modeling: 1) serious hepatic, renal, and skin adverse events (AEs); 2) proton pump inhibitor use after dyspepsia, while taking an NSAID; and 3) multiple occurrences of myocardial infarction (MI) (as opposed to one per patient). Health state utilities for AEs were assigned a value equal to that for the hospitalized surgical management of a complicated GI event. For MI, a 5% discount factor was used to reduce the patient's utility score. Patients switching to acetaminophen because of an AE can experience reduced analgesic effect compared with NSAIDs; therefore utilities were discounted by 20%. RESULTS: The modified model produced lower estimates of LYs and QALYs (approximately 0.05 and 0.08 less, respectively) compared with the original model which could be clinically meaningful in a 5-year model. Patient and clinical characteristics that defined low GI-risk subgroup versus high GI-risk group produced differences in LYs and QALYs of up to 1 LY and 0.7 QALY. CONCLUSION: Effectiveness can vary considerably across patients with varying clinical characteristics. Therefore, the cost-effectiveness of treatment in any population should consider the heterogeneity of patients. This model provides flexible means to compare cost-effectiveness of treatment for patients with osteoarthritis.