The potential cost‐effectiveness of roflumilast drug treatment in mild cognitive impairment

Abstract

AbstractBackgroundEvidence is needed on the value of new (pharmacological) health technologies in their early stage of development, to support stakeholders (industry, regulators, payers, and health care providers) in decision‐making linked to their further development and introduction in routine care. Early health‐technology assessment supports stakeholders by providing insight into the size and uncertainty of the expected benefits new technologies may have. This can facilitate early guidance on positioning in clinical practice and research recommendations. Our objective was to provide a pragmatic example of the potential cost‐effectiveness of roflumilast treatment in addition to usual care versus usual care alone if prevention of progression to dementia is in part realized in a population of patients diagnosed with a mild cognitive impairment (MCI) at a memory clinic.MethodThe IPECAD open‐source decision‐analytic model was used to simulate the natural disease progression of MCI and dementia and estimate the quality‐adjusted life years (QALYs) and care costs, comparing roflumilast treatment added to usual care with usual care alone. In this model roflumilast proof of concept evidence on a visual 30‐word verbal learning test (0.39 z‐score) was translated using a published risk equation model to a relative risk of 0.93 for developing dementia. The treatment was assumed symptomatic with no effect on mortality.ResultMore lifeyears were spent in MCI in the roflumilast strategy compared with usual care, with 0.2 increment lifeyears (95%CI: 0.0 to 0.6) (see figure 1), and 0.04 incremental lifetime QALYs (95%CI: 0.00 to 0.11) and k€1.3 adjusted societal costs (95%CI: ‐6.3 to 5.5) (see figure 2). This resulted in an incremental cost‐effectiveness ratio of k€34 per QALY (see table 1). Roflumilast was estimated to have a 37% probability of being cost‐effective (net health benefit 95% interval ‐0.27 to 0.42), with uncertainty mainly driven by extrapolating the word learning surrogate outcome to the patient‐relevant outcome of dementia onset, and the efficacy estimate (see figure 3).ConclusionThe potential cost‐effectiveness of roflumilast treatment is uncertain, driven by uncertainty in treatment efficacy and extrapolation to postponing dementia onset. We believe the results indicate a potential for roflumilast and provide several recommendations to consider investing in its future research.