The Epilepsy‐Cog study: Methods to establish a harmonized study of late onset epilepsy in a metacohort of six population‐based cohorts in the United States

OBJECTIVE: With the expected demographic shift toward those ≥65 years of age in the United States, late onset epilepsy (LOE) poses a significant public health issue, yet it has been historically understudied. We are undertaking an effort in the Epilepsy-Cog study to pool individual participant data from six United States-based prospective cohort studies. In this paper, we outline the process for ascertaining epilepsy, harmonizing, and pooling individual participant data across the six cohorts. METHODS: The Epilepsy-Cog study includes individual participant data from six United States-based longitudinal cohort studies: Atherosclerosis Risk in Communities Study, Cardiovascular Health Study, Multi-Ethnic Study of Atherosclerosis, Northern Manhattan Study (NOMAS), Reasons for Geographic and Racial Differences in Stroke, and Washington Heights/Inwood Columbia Aging Project. In all cohorts except NOMAS, prevalent and incident epilepsy were ascertained using Medicare claims-based algorithms. In NOMAS, epilepsy cases were identified through cohort-based reporting and medical record review. To perform cross-cohort harmonization of variables, we used the lowest common denominator approach, assigning response categories or value levels in common across all cohorts. RESULTS: From a total of 68 544 participants across six cohorts, 43 753 participants met eligibility criteria for Epilepsy-Cog. Among them, we identified 551 (1.3%) participants with prevalent epilepsy and 1500 (3.4%) participants with incident epilepsy. We have harmonized demographic characteristics, health behaviors, vascular risk factors (VRFs), one genetic variable, medication use, subjective health status measures, incident events, and cause-of-death variables. SIGNIFICANCE: The Epilepsy-Cog pooled cohort of 43 753 participants with and without epilepsy, combined with harmonized demographic, VRF, and event data, offers a unique resource to yield new insights into LOE.