Background: The positive results of a randomised clinical trial of rivastigmine in patients with dementia associated with Parkinson's disease have been published recently. Patient-level healthcare utilisation data were also collected, and this report is the economic evaluation based on these data. Objective: To determine the cost effectiveness of rivastigmine 3-12 mg/day in patients in whom mild to moderate dementia developed at least 2 years after they received a clinical diagnosis of Parkinson's disease. Methods: A cost-effectiveness analysis was performed by applying Canadian and UK cost weights (year 2004 values) to healthcare utilisation data collected prospectively during a randomised, double-blind, multinational, 24-week trial of rivastigmine 3-12 mg/day (n = 362) versus placebo (n = 179). Patients were >=50 years of age, had a Mini-Mental State Examination (MMSE) score of between 20 and 24 and had contact with a responsible caregiver at least 3 days a week. Quality-adjusted survival time, transformed from MMSE scores, was the measure of effectiveness. Caregiver costs included paid and unpaid time, and direct costs included concomitant medications, outpatient care, hospitalisations, long-term care and study medications. Analysis was conducted from a societal perspective with a time horizon of 24 weeks. Results: Consistent with the improvement in clinical outcomes, there was an observed increase in quality-adjusted survival time in the rivastigmine arm of 2.81 quality-adjusted life-days (two-sided p-value 0.13 [90% CI -0.243, 5.86]). Using Canadian price weights, there was an observed increase in cost in the rivastigmine arm of $Can55.76 (two-sided p-value 0.98 [90% CI -3431, 3543]), with a resulting incremental cost-effectiveness ratio of $Can7429 per QALY. Using UK price weights, there was an observed decrease in cost in the rivastigmine arm of Lstg 26.18 (two-sided p-value 0.99 [90% CI -2407, 2355]). Conclusion: Although no between-treatment differences in cost were seen, the small sample size, highly variable cost distributions and short time horizon prevent us from making strong conclusions with regard to the effect of rivastigmine on total costs and, by inference, on cost effectiveness.
