The association between atrial fibrillation (AF) and dementia remains debated, and data from Mediterranean populations are scarce. We aimed to evaluate the relationship between AF and dementia in a large Catalan cohort, with a particular focus on identifying subgroups with the strongest associations.

We conducted a population-based observational study including individuals aged ≥45 years without prior diagnosis of dementia. Data were obtained from the System for the Development of Research in Primary Care, which covers more than 80% of the Catalan population. Dementia was identified using ICD-10-CM codes and prescription records. Early-onset dementia (EOD) was defined as diagnosis before age 65 years. Participants were followed up from 2009 to 2021. Cox regression models were used to identify predictors of incident dementia.

A total of 2 520 839 individuals were included (mean follow-up: 13 years), of whom 79820 (3.25%) had AF at baseline. On univariate analysis, AF was a strong predictor of dementia (HR, 3.39; P<.001). However, after multivariable adjustment, the association was modest (HR, 1.04; P<.001). AF was a stronger independent predictor in individuals aged <70 years (HR, 1.21; P<.001), whereas the association was not significant in those aged ≥70. The strongest association was observed for EOD (HR, 1.36; P<.001). All associations remained consistent in analyses restricted to individuals without prior stroke.

In a large Mediterranean cohort, AF was independently associated with a modest increase in dementia risk, regardless of stroke. This association was stronger in younger individuals and in EOD. These findings highlight the need for studies assessing the impact of early identification and management of AF on dementia incidence in younger populations.