Older adults represent the fastest-growing population of individuals with epilepsy with an incidence that peaks after age 65. Patients with late-onset epilepsy (LOE) have a multitude of risk factors for accelerated cognitive and brain aging, including vascular and metabolic risk factors. Despite this, there are few studies investigating the cognitive profiles of older adults with LOE, a neglected area in aging research. We examine the cognitive profiles of older adults with LOE and determine the contribution of demographic and vascular risk factors to impairment.

Participants were part of the Atherosclerosis Risk in Communities Study (ARIC) and the incidence of epilepsy was identified using ARIC hospitalization records and Centers for Medicare and Medicaid Services claims data from 1991 to 2015. Approximately 1.8% of the participants with sufficient Medicare coverage data were classified as having LOE (LOE n=281; Non-LOE n=9808). Vascular, lifestyle, and cognitive data were obtained from the ARIC Neurocognitive Study (ARIC-NCS) which consisted of three visits since 2011. Participants with ARIC-NCS visits completed after the onset of seizures were included in the final sample. Non-LOE participants with normal cognition (Black: n=603 and White: n=2543 participants independently) were used to generate z-scores across tests of language, memory, executive function, and processing speed/attention. Impairment was defined as <1.5 standard deviations below the mean of the normative sample. Stepwise regressions were conducted to examine the contribution of demographic (age, race, sex, education) and vascular risk factors (hypertension, diabetes, hyperlipidemia, obesity, smoking) to cognitive performance.

Average age of first seizure of all LOE participants (n=281) was 76.23 (SD=6.24), 55.9% female, 30.7% Black/African American, and the majority had either a college (28.1%) or high school degree (26%). Fifty-six LOE participants had ARIC-NCS visits after the onset of seizures (average age=79.84, SD=5.17, 57.1% female, 32.1% Black). Approximately 67.9% of the sample had at least one vascular risk factor with 81.5% having hypertension, 37% diabetes, 26.4% hyperlipidemia, 20.4% obesity (BMI>30), and 4.5% current smoker. The most frequently impaired domains were language (naming=29.7%; animal fluency=20%; letter fluency=30%) and memory (prose immediate recall=18.4%; prose delayed recall=44.7%; word delayed recall=19.4%). Higher education was associated with better naming (b=0.801, p=0.040). Female sex (b=-0.799, p=0.017) and lower education levels (b=0.418, p=0.050) were associated with poorer immediate prose recall. Older age was associated with poorer delayed prose recall (b=-0.191, p=0.036). Hypertension was associated with worse digit span backward (b=-0.942, p=0.002).

In older adults with LOE, language and memory were the most commonly impaired cognitive domains, similar to studies in early onset epilepsy. Vascular risk factors were prevalent among LOE and hypertension was associated with worse working memory. Further, important demographic factors (sex, education, and age) were associated with the extent of cognitive impairment. Characterizing cognitive profiles in LOE and determining the contribution of demographic and vascular factors to impairment could help to identify patients at risk for future cognitive decline and/or the development of LOE itself, as well as interventions aimed at reducing the risk of further decline.