LAUSR

Background People with epilepsy (PWE) are reported to have higher risk of dementia. However, the magnitude of this association, whether it varies according to dementia subtypes and whether there are factors that modify the association is uncertain. We investigated the apparent association in a large population-based retrospective cohort study using routinely-collected linked health data from hospitalisation, mortality records and primary care consultations. Methods We used linked health data from the Secure Anonymised Information Linkage (SAIL) databank to follow-up Welsh residents for whom linked primary care data were available from their 60th birthday to estimate dementia rate and associated risk factors. Disease (dementia), exposure (epilepsy) and covariates (medication, smoking, stroke and diabetes) were classified using previously validated code lists. We studied rate of dementia and dementia subtypes in people with and without epilepsy using (stratified) Kaplan-Meier plots and flexible parametric proportional hazard analyses. The study population comprised 563,808 people of whom 19,807 (4%) had indications of epilepsy in the linked health data. 13,454 (68%) of PWE and 49,439 (9%) of people without epilepsy had at least one record for a prescription of an antiepileptic drug (AED). Compared to people without epilepsy, PWE had lower survival (median survival to age 79 years compared to 84), higher smoking risks (74% compared to 66%) and higher stroke risks (20% compared to 7%) before or during follow-up. Results Between ages 60 and 70 years, 6% of PWE and 1% of people without epilepsy had a diagnosis of dementia; corresponding figures between ages 60 and 80 years were 23% and 8%. The difference in dementia rate between those with and without epilepsy was larger for vascular dementia than for Alzheimer’s disease. The increased rate for PWE was modified by a history of stroke, smoking and, to a lesser effect, diabetes. PWE who were first diagnosed before age 25 years had a lower dementia rate than those diagnosed later in life. Compared to PWE not exposed, those exposed to sodium valproate were at higher risk of dementia (crude HR: 1.7; 95% CI: 1.5–1.9) while those exposed to a group of enzyme-inducing AED were at similar risk (crude HR: 1.1, 95% CI: 1.0–1.3). Conclusion At least some of the increased risk of dementia in PWE can be attributed to increased vascular risk factors in PWE causing vascular dementia. Given the widespread use of sodium valproate in PWE, the association of the drug with higher dementia risk is concerning.