2021 CMSC Annual Meeting

Assessing the Impact of Fingolimod Adherence on Relapse and Costs Using Marginal Structural Models


Background: Disease modifying therapies (DMTs) for relapsing MS are most effective when patients adhere to them as prescribed. Published studies using real-world data show that DMT adherence is associated with improved clinical and economic outcomes. However, these studies may be subject to simultaneity bias when adherence and outcomes are measured over the same time period. Marginal structure models (MSM) mitigate this type of time-dependent bias. Objectives: Estimate the impact of fingolimod adherence on the odds of relapse and costs during a 12-month period using MSM to account for simultaneity bias. Methods: This retrospective administrative claims data analysis examined patients initiating fingolimod from January 1, 2012, to May 10, 2018, indexed on the first fingolimod claim date. Commercial and Medicare Advantage (MA) patients with continuous medical and pharmacy benefits 6 months before the index date (baseline period), 6 months starting on the index date (initiation period), and 12 months after the initiation period (post-initiation period), were included. Patients were required to be naïve to all DMTs (no claim in baseline) have an MS diagnosis code in the baseline, initiation, or post-initiation periods, and ?1 fingolimod claim after the index date. Adherence was measured as proportion of days covered ?0.8. Outcomes were measured in the initiation period and each quarter of the post-initiation period and included relapse, all-cause total costs, all-cause medical costs, and all-cause total costs excluding fingolimod. MSMs assessed the impact of adherence in a given period on outcomes in the next quarter of the post-initiation period, adjusting for other time-dependent and time-independent confounders. Results: A total of 694 (562 commercial and 132 MA) patients met eligibility criteria. Fingolimod adherence was 0.87 in the initiation period and decreased to 0.65 in the last quarter of the post-initiation period. Adherence was a statistically significant predictor of lower odds of relapse (odds ratio [OR] 0.526, P=0.009), lower all-cause medical costs (cost ratio [CR] 0.612, P<0.001), and lower all-cause total costs excluding fingolimod (CR 0.543, P<0.001). As expected, adherence was associated with higher all-cause total (medical plus pharmacy) costs (CR 1.647, P<0.001). Conclusions: MSM applied to a sample of DMT naive patients initiating fingolimod estimated that higher medication adherence reduced the risk of relapse and lowered all-cause medical costs as well as total all-cause costs excluding fingolimod.