Late-onset unprovoked seizures and subsequent risk of dementia in older adults: A target trial emulation.

BACKGROUND: Late-onset unprovoked seizures of unknown etiology (LOSU) are increasingly recognized in older adults, but their prognostic implications for cognitive outcomes remain uncertain. Prior studies suggest an association with dementia, although the direction of this relationship has remained unclear. To evaluate the association between LOSU and subsequent dementia-related outcomes in older adults using a target trial emulation framework to structure an observational exposure comparison. METHODS: We conducted a retrospective cohort study using the TriNetX Global Collaborative Network. Adults aged ≥65 years with cardiometabolic or vascular comorbidity and incident LOSU were compared with a propensity score-matched cohort with syncope and no seizure diagnosis. Propensity score matching was performed 1:1 using 28 baseline covariates. The primary outcome was incident all-cause dementia. Secondary outcomes included Alzheimer's disease (AD), vascular dementia (VaD), mild cognitive impairment (MCI), non-AD dementia, and all-cause mortality. Cox proportional hazards models were used to estimate hazard ratios (HRs) over 5 years of follow-up. RESULTS: After matching, 40,401 patients were included in each group. Over 5 years, all-cause dementia occurred in 14.5% of patients with LOSU and 10.0% of those in the comparator cohort (HR, 1.55; 95% CI, 1.49-1.61). LOSU was also associated with higher risks of AD (HR, 1.44; 95% CI, 1.31-1.58), VaD (HR, 2.07; 95% CI, 1.85-2.31), MCI (HR, 1.42; 95% CI, 1.32-1.54), non-AD dementia (HR, 1.64; 95% CI, 1.56-1.72), and all-cause mortality (HR, 1.64; 95% CI, 1.58-1.71). Results were consistent across subgroup and sensitivity analyses, including an extended 10-year follow-up window (HR, 1.50; 95% CI, 1.45-1.56), a source population defined by a relaxed cardiometabolic eligibility requirement (HR, 1.61; 95% CI, 1.54-1.67), a propensity score model excluding the cardiometabolic conditions (HR, 1.63; 95% CI, 1.56-1.70), and lagged models. CONCLUSIONS: In this observational cohort study structured using a target trial emulation framework, LOSU was associated with higher subsequent risks of dementia-related outcomes and mortality. These findings suggest that late-life unprovoked seizures without an identified etiology may identify older adults at increased risk of neurodegenerative or cerebrovascular outcomes, supporting closer cognitive follow-up in this population. Dementia subtypes were assigned from diagnostic codes without biomarker confirmation, and subtype-specific estimates should be interpreted with caution.