Cost-Effectiveness Modelling of Multiple System Atrophy for Early Health Technology Assessment.

BACKGROUND AND OBJECTIVES: Multiple system atrophy is a neurodegenerative and rapidly progressing disease. Although symptoms often overlap, diagnostic criteria define two motor phenotypes: cerebellar and Parkinsonian. Little is known about available health-economic parameters for multiple system atrophy, and it remains unclear if the current available data are sufficient to construct an early health technology assessment. This assessment considered (i) if data gaps can be minimised or cost-effectiveness modelling can be facilitated by conducting literature searches for disease analogues, (ii) if an early health technology assessment is feasible with the currently available data, and (iii) the potential for developing user-friendly model interfaces in rare diseases. METHODS: Literature searches were conducted for economic evaluations and health-economic parameters (including costs, utilities, natural history of disease) for multiple system atrophy. The searches for disease analogues focussed on economic evaluations and showed a widespread use of functional or disability scales to inform model structures; we therefore used the unified multiple system atrophy rating scale part IV (global disability scale) to inform the model structure. Natural history of disease data were used to inform utilities and transition probabilities. Analyses of a hypothetical intervention were conducted from the perspective of the United Kingdom National Health Service with a cost-effectiveness threshold of £30,000 per quality-adjusted life year. The cost-effectiveness analysis was presented using a user-friendly interface in R Shiny, which provides stakeholders with an opportunity to explore results. RESULTS: Cost-effectiveness analyses were not identified for multiple system atrophy; however, two cost-of-illness, 20 quality-of-life, and nine natural history of disease studies were identified. Health-economic parameters were estimated from registry data, such as utilities and transition probabilities. A cost-effectiveness analysis was constructed for a hypothetical intervention despite evidence shortcomings, for example, aggregated cost data. An R Shiny app with a user-friendly interface was developed for stakeholders to change inputs and evaluate results. DISCUSSION: Literature searches for disease analogues proved useful for early modelling in multiple system atrophy, for example, to inform model structures, but surrogate data could only be used when sufficient granularity was available (e.g., micro costing). These findings suggest that early conceptual modelling can help identify data gaps and guide future evidence generation, such as selecting appropriate endpoints for natural history of disease studies. Cost-effectiveness results can be highly uncertain for rare diseases at these early stages and should be considered as part of an iterative modelling framework.