OBJECTIVES: In 2023, the Institute for Clinical and Economic Review – the organization (ICER-O), an influential US health technology assessment (HTA) body, published a relapsing-remitting multiple sclerosis report, concluding that dimethyl fumarate was cost-effective and that four monoclonal antibodies it assessed were not. It also assessed cost reductions needed to make the monoclonal antibodies cost-effective. We examine whether dimethyl fumarate was the appropriate comparator for all treatments across different willingness-to-pay (WTP) thresholds.

METHODS: We reviewed this model according to accepted long-standing published methods for health economic evaluation, focusing on the calculation of appropriate Incremental Cost-effectiveness Ratios (ICERs). We recalculated ICERs and Net Monetary Benefit (NMB) for various WTP scenarios to determine whether/when ICERs versus dimethyl fumarate were appropriate.

RESULTS: ICER-O’s model excluded many treatment alternatives, but, conditional on that not affecting results, its qualitative cost-effectiveness conclusion is correct, but some associated calculations are not. Using appropriate methods - the elimination of all but two treatments in the model due to dominance (strong or weak) and the magnitude of the sole relevant ICER versus any conventional willingness-to-pay threshold – results in the same conclusion. However, at relatively high – but sometimes observed in the US – levels of WTP (above $292,000 in this case), dimethyl fumarate was not the appropriate comparator, in terms of having the highest NMB. Thus, ICER-O’s price/cost reductions that other drugs would require to be cost-effective were inappropriately calculated and were lower than required.

CONCLUSIONS: While the WTP scenarios where these concerns occurred may not have strong policy relevance, as the US’s premier HTA body, ICER-O carries a responsibility to carefully adhere to accepted methods in economic evaluation – or to justify any exceptions. The comparator appropriate for ICERs and associated price calculations may change across various model scenarios and thus should be selected with those scenario parameters in mind.