Differences in Survival Associated With the First Antiseizure Medication in People With Dementia and Epilepsy

Epilepsy is an increasingly recognized comorbidity in neurocognitive disorders. Although experts recommend newer antiseizure medications (ASMs), whether there are real-world benefits of selecting particular ASMs for epilepsy in dementia remains unknown. Our objective was to examine whether different ASMs are associated with differences in survival in persons with epilepsy and dementia.

A cohort study was conducted using Swedish national registers. We included individuals with dementia who received ASM treatment after a diagnosis of epilepsy after January 1, 2006. Data were analyzed until December 2023. Dispensed ASMs (Anatomical Therapeutic Code N03) were used to identify treatment; patients were categorized as starting one of the 4 most common ASMs (carbamazepine, levetiracetam, lamotrigine, valproate) or other ASMs. Recurrent prescriptions were used to determine treatment duration. Associations between all-cause death and ASMs were assessed by Cox proportional hazards regression. We also assessed causes of death and the risk of cardiovascular death.

In Sweden, between 2006 and 2023, we included 5,764 individuals (2,811 men [48.8%] and 2,953 women [51.2%]) using their first ASM after a diagnosis of epilepsy and dementia. Carbamazepine (n = 1,578) was the most common ASM before 2015 and levetiracetam (n = 2098) most common thereafter. In Kaplan-Meier analyses and Cox regression models adjusting for age, sex, year of ASM start, and comorbidities, valproate (n = 746) was associated with increased adjusted hazard ratio (aHR) of death (1.34, 95% CI 1.20-1.48), in contrast to lamotrigine (n = 922, aHR: 0.84, 95% CI 0.75-0.93) and levetiracetam (n = 2098, aHR: 0.93, 95% CI 0.85-1.03). The risks were similar in analyses using restricted mean survival time, propensity score-matched sets of participants, balancing weights in regression models, and cases with epilepsy onset in existing dementia. Cardiovascular causes of death were more common among users of valproate or carbamazepine than among users of lamotrigine and levetiracetam. Compared with carbamazepine, valproate was associated with increased risk of cardiovascular death (aHR: 1.30; 95% CI 1.11-1.52) while lamotrigine (aHR 0.79; 95% CI 0.66-0.94) was associated with a reduced risk.

In this population-wide cohort study, use of valproate was associated with the highest risk of death in persons with epilepsy and dementia. Lamotrigine and, in some models, levetiracetam were associated with better survival than both valproate and carbamazepine. This provides real-world support for existing expert guidelines.