OBJECTIVES: Recent drug developments in Alzheimer's disease (AD) have increased attention for decision-analytic models to assess their value. Transparency and credibility of decision-models are key for their results to be used by decision-makers. Relatively short-term trial evidence requires assumptions on long-term treatment effect, which drives uncertainty. We aim to develop an open-source AD decision-model framework for the health-economic evaluation of emerging AD treatments.
METHODS: A previously developed Markov model (MMSE-based states: mild cognitive impairment (MCI), mild, moderate, severe dementia and death) was rebuilt in R and internally validated by replicating its published outcomes (maximum relative difference<0.6%). The model was updated with new features including treatment discontinuation, European country/region-specific mortality, costs and utilities, and probabilistic analyses on natural disease history parameters. A hypothetical treatment with 25% reduction in 70-year old early AD patients (progression rate from MCI to mild dementia and mild to moderate dementia) with 10% treatment discontinuation and 5% waning per year, maximally treated for 7 years was compared to standard care. We calculated the maximum price at which treatment would be cost-effective and performed sensitivity analyses on net health benefits. The model is freely available via IPECAD (www.ipecad.org/open-source-model), including coding comments and a deterministic spreadsheet.
RESULTS: Early AD treatment was cost-effective at €4444, €3170, €4692, €5385, €3943 price per year per person in British Isles and in East, North, South and West Europe respectively (WTP=€40,000/QALY). Net health benefit was mainly sensitive to treatment costs, treatment duration, treatment effect on dementia onset and starting age.
CONCLUSIONS: We developed an accessible open-source model framework for health-economic evaluation of early treatment in AD. The sensitivity to natural progression and age reflects the importance of predicting treatment response for new treatments. Planned features include budget impact module, online interface and individualized progression to allow evaluation of individualized treatment per country.
