Objective: To compare disease-related medical care and productivity costs, and utilities, in 482 patients with rheumatoid arthritis randomised to receive leflunomide, methotrexate or placebo during a 12-month period. Design and setting: Prospective pharmacoeconomic analysis of a 1-year randomised double-blind trial set in North America. Perspective: Societal and the Ontario Ministry of Health. Methods: Information on healthcare resources, out-of-pocket expenses, loss of working time and time spent on chores, related to the disease or the medication, were collected at 4-week intervals and at study discontinuation. Rating scale and standard gamble (SG) utilities (0 = worse; 100 = best) were collected at baseline and at 6 and 12 months or study exit. Medical care costs in Canadian dollars ($Can) were calculated using Ontario reimbursement schedules. US patients' expenses were converted to $Can using 1995 purchasing power parity. Lost wages were calculated by age and gender according to 1995 Canadian wage data. All costs were adjusted to 1999 $Can and arithmetic mean costs were compared using the nonparametric bootstrap. Analysis of covariance was performed to compare utilities between groups. Results: Mean (standard deviation) rating scale values and SG utilities, respectively, for leflunomide, methotrexate and placebo were 67.7 (18.0), 64.8 (18.1) and 57.5 (9.2), and 80.2 (22.1), 83.2 (18.0) and 77.0 (20.5). Both leflunomide and methotrexate had higher rating scale values (p < 0.05) compared with placebo; SG utilities were significantly different between methotrexate and placebo (p < 0.05). Annualised total rheumatoid arthritisb- or drug-related costs for leflunomide, methotrexate and placebo, respectively, were $Can1761, $Can1280 and $Can1324, and medical care costs were $Can753, $Can620 and $Can167 (all costs exclude drug acquisition and monitoring costs). Annual drug acquisition/ routine monitoring costs were estimated, respectively, at $Can3853/$Can483 for leflunomide and $Can258/$Can599 for methotrexate. Differences between overall costs (excluding drug acquisition and monitoring costs) and medical care costs were not statistically significant. The costs of treating patients with leflunomide were significantly higher than for methotrexate when drug acquisition and monitoring costs were included (p < 0.0001) Conclusions: No statistically significant differences in utilities could be found between leflunomide or methotrexate. When drug monitoring and acquisition costs are excluded, leflunomide has an otherwise similar economic profile compared with methotrexate, the current gold standard. The acquisition cost of leflunomide is a driving factor in increasing the costs of therapy. These higher costs need to be assessed relative to the therapeutic value of leflunomide.
