Smoking, Diabetes, and Stroke Linked to Higher Risk of Seizures That Begin After Age 65

Epilepsy is often thought of as a childhood condition, yet new-onset seizures become more common after age 65, and the cause is frequently unclear. Because stroke, small-vessel disease, and dementia share vascular roots, researchers have suspected that late-onset epilepsy (LOE) may also reflect vascular injury to the brain. This study asked which vascular risk factors (VRFs) are associated with developing LOE, using one of the largest pooled datasets assembled for the question.

The Epilepsy-Cog study is an individual participant data meta-analysis of six US population-based prospective cohorts: ARIC, CHS, MESA, REGARDS, WHICAP, and NOMAS. Five were linked to Medicare claims to identify epilepsy. Follow-up began once a participant was 65 or older and had 2 years of continuous fee-for-service enrollment. Claims coverage spanned 1991 to 2020 across cohorts. NOMAS was not Medicare-linked and used cohort-based case ascertainment, with follow-up from entry (1993–2008) through 2024.

Of 43,753 participants, 551 with prevalent epilepsy and 945 with no follow-up were excluded, leaving 42,257 people. Their mean age was 71.4 years and 56.0% were female. Over a mean follow-up of 9.3 years, 1,500 participants (3.5%) developed LOE, an incidence of 3.8 per 1,000 person-years. Associations were estimated with multivariable Cox proportional hazards models.

In fully adjusted models, several factors were associated with higher LOE risk. Prior stroke carried the largest effect (HR 1.99; 95% CI 1.63–2.43). Diabetes (HR 1.42; 1.25–1.62), current smoking (HR 1.39; 1.18–1.64), chronic kidney disease (HR 1.29; 1.13–1.48), and hypertension (HR 1.17; 1.04–1.31) were also linked to risk. The APOE ε4 allele showed a dose-response pattern: one copy gave an HR of 1.22 (1.07–1.39) and two copies an HR of 1.84 (1.36–2.50).

The associations were similar across age, sex, and race and ethnicity subgroups, which suggests the risk pattern is broadly applicable to older US adults. The authors conclude that LOE may reflect the underlying burden of brain vascular disease. If so, the modifiable factors identified here (smoking, blood pressure, diabetes, and kidney health) overlap with established stroke and dementia prevention targets. Seizure risk may be one more reason to manage them. The APOE ε4 finding also hints at overlap with Alzheimer's-related pathways.

Several caveats apply. This is observational research, so the results show association, not causation, and no study here tests whether treating these risk factors lowers LOE risk. Medicare-based case ascertainment requires fee-for-service enrollment, and NOMAS used a different method, so case-finding was not uniform across cohorts. The supplied text contained only the abstract, so detailed limitations, sensitivity analyses, and conflict-of-interest disclosures could not be reviewed here.