Patterns of discontinuation of Epilepsy Specific Anti-Seizure Medications in Medicare Beneficiaries
Shuo (Mila) Sun1, Rafaella Caze de Medeiros2, Julianne Brooks2, Madhav Sankaranarayanan1, Sebastien Haneuse1, Lidia M. V. R. Moura2
1Department of Biostatistics, Harvard T.H. Chan School of Public Health, 2Department of Neurology, Massachusetts General Hospital, Harvard Medical School
Objective:
We analyzed patterns of discontinuation of epilepsy specific anti-seizure medications (ESMs) in a national sample of older adults after acute ischemic stroke (AIS) discharge.
Background:
The lack of guidelines for managing ESM use in post-stroke patients poses a challenge for providers and may contribute to differences in drug selection and treatment duration.
Design/Methods:
We retrospectively analyzed a 20% sample of U.S. Medicare beneficiaries aged 65 and over hospitalized for first AIS between 2009-2021, who were discharged and initiated ESM treatment within 30 days. We excluded patients without continuous Part D Medicare coverage and those prescribed ESMs before discharge. Death and non-persistence (no refill for over 90 days) were censoring factors. We used an adjusted proportion of days covered (PDC) measure, accounting for prescription overlap, hospital readmissions, and non-persistency to analyze medication discontinuation patterns during the first year after discharge among ESMs initiators (the 3 most commonly prescribed). We applied cumulative PDCs and trajectory clustering to characterize adherence behavior.
Results:
In this cohort of 1,961 ESM initiators, 1,708 patients met the inclusion criteria, including a total of 1,607 for Levetiracetam, 64 for Lamotrigine, and 50 for Carbamazepine. Over half patients had cumulative PDCs below 0.8 in the first year: Levetiracetam (63%), Lamotrigine (59%), and Carbamazepine (64%). With 0.8 as a threshold, 43% discontinued Levetiracetam within two months, and 26% for Lamotrigine and 53% for Carbamazepine (p=0.03). Longitudinal clustering model identified four distinct medication discontinuation patterns, with the following proportions for the three medications: fast discontinuation [<1month] (34%, 34% and 42%), early discontinuation [2-6 months] (17%, 17%, 10%), late discontinuation [7-12 months] (13%, 13%, 12%), and long-term adherent [12 months or more] (36%, 36%, 36%). The differences across the medications were substantial (p<0.01).
Conclusions:
Medicare beneficiaries who survive an acute ischemic stroke and are discharged home exhibit varying patterns in drug choice and treatment duration trajectories.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.