OBJECTIVES: Based on “similar” effectiveness, the economics of treating relapsing-remitting multiple sclerosis with fingolimod and natalizumab were studied with a cost-minimization analysis (CMA) in 2012 based on Swedish data. Fingolimod was cost-saving compared with natalizumab. However, studies have shown a difference in efficacy favoring natalizumab, leading to a possibility that, while more costly, it may be cost-effective. This study uses PSA to estimate cost-effectiveness and associated decision uncertainty for the CMA and CEA.

METHODS: A later publication, Vollmer (2018) indicated efficacy differences in “disease activity.” We used those count data in a PSA. For a cost difference, we used the original 2 year cost difference mean and set the standard error equal to it (85,128 SEK). We then calculated the incremental net monetary benefit in each PSA iteration to estimate decision uncertainty. We did the same, but setting the effectiveness difference to zero, reflecting the original CMA.

RESULTS: A previously published study suggested a WTP of 500,000 SEK (approximately $US 55000) for relapse avoidance. PSA results were similar to a deterministic analysis, suggesting fingolimod is cost effective at WTP values up to 660,000 SEK. At 500,000 SEK, the probability that fingolimod was cost-effective was 0.59, indicating significant uncertainty. In a CMA-based PSA, this probability was 0.84.

CONCLUSIONS: The original deterministic CMA claimed fingolimod as cost-saving. Our CMA-based PSA is in agreement, with minimal decision uncertainty. Allowing for effectiveness differences based on natalizumab data suggesting greater effectiveness still shows fingolimod’s cost-effectiveness but with significant decision uncertainty. The CEA conclusion is consistent with the CMA, but there is higher (and significant) decision uncertainty, suggesting the potential value of additional data collection. CMA has been increasingly faulted in its assumptions of equal effectiveness and our analysis supports the notion that its results may be misleading.