Cost effectiveness of memantine in Alzheimer's disease: an analysis based on a probabilistic Markov model from a UK perspective.

Background: Clinical trials with memantine, an uncompetitive moderate-affinity NMDA antagonist, have shown improved clinical outcomes, increased independence and a trend towards delayed institutionalisation in patients with moderately severe-to-severe Alzheimer's disease. In a randomised double-blind, placebo-controlled, 28-week study conducted in the US, reductions in resource utilisation and total healthcare costs were noted with memantine relative to placebo. While these findings suggest that, compared with placebo, memantine provides cost savings, further analyses may help to quantify potential economic gains over a longer treatment period.

Objective: To evaluate the cost effectiveness of memantine therapy compared with no pharmacological treatment in patients with moderately severe-to-severe Alzheimer's disease over a 2-year period.

Methods: A Markov model was constructed to simulate patient progression through a series of health states related to severity, dependency (determined by patient scores on the Alzheimer's Disease Cooperative Study-Activities of Daily Living [ADCS-ADL] inventory and residential status ('institutionalisation') with a time horizon of 2 years (each 6-month Markov cycle was repeated four times). Transition probabilities from one health state to another 6 months later were mainly derived from a 28-week, randomised, double-blind, placebo-controlled clinical trial. Inputs related to epidemiological and cost data were derived from a UK longitudinal epidemiological study, while data on quality-adjusted life-years (QALYs) were derived from a Danish longitudinal study. To ensure conservative estimates from the model, the base case analysis assumed drug effectiveness was limited to 12 months. Monte Carlo simulations were performed for each state parameter following definition of a priori distributions for the main variables of the model. Sensitivity analyses included worst case scenario in which memantine was effective for 6 months and one-way sensitivity analyses on key parameters. Finally, a subgroup analysis was performed to determine which patients were most likely to benefit from memantine. Informal care was not included in this model as the costs were considered from National Health Service and Personal Social Services perspective.

Results: The base case analysis found that, compared with no treatment, memantine was associated with lower costs and greater clinical effectiveness in terms of years of independence, years in the community and QALYs. Sensitivity analyses supported these findings. For each category of Alzheimer's disease patient examined, treatment with memantine was a cost-effective strategy. The greatest economic gain of memantine treatment was in independent patients with a Mini-Mental State Examination score of > or =10.

Conclusion: This model suggests that memantine treatment is cost effective and provides cost savings compared with no pharmacological treatment. These benefits appear to result from prolonged patient independence and delayed institutionalisation for moderately severe and severe Alzheimer's disease patients on memantine compared with no pharmacological treatment.