OBJECTIVES: Based on an assessment of “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. In this analysis 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.

METHODS: A later publication, Vollmer (2018), indicated a range of statistically significant odds ratios for a “composite activity measure” that favored natalizumab over the evaluated two-year period (1.86 to 2.90). We employed those estimates to perform several CEA’s, using the CMA’s cost data for 2 years and, for the baseline analysis, the lowest odds ratio of 1.86 (also expressed as differential absolute “disease activity” of 22.2 and 34.7 percent for natalizumab and fingolimod). Efficacy for ICER calculations was expressed as avoiding an additional case of “disease activity.” Sensitivity analysis were done for the other odds ratios.

RESULTS: The base case ICER indicated natalizumab as cost effective for WTP values greater than 681,024 SEK (approximately US$ 75000, currently). At the highest estimated odds ratio (2.90), the ICER drops to 368,519 SEK (approx. US$ 42,000, currently). A previously published study suggested a WTP of 500,000 SEK (approximately $US 55,000) for relapse avoidance.

CONCLUSIONS: The original CMA claimed that fingolimod is cost-saving vs. natalizumab. Our more appropriate CEA shows that this conclusion may hold even when natalizumab’s apparent greater effectiveness is taken into account. However, this depends on the WTP for avoiding an instance of “disease activity,” a non-standard outcome without an established value. Utilizing differential efficacy estimates, regardless of their lack of statistical differences, can be crucial in determining economically optimal treatments; CMA has been increasingly faulted and our analysis supports the notion that its results may be misleading.