Introduction Atrial fibrillation (AF) is associated with dementia. Anticoagulation may modify this relationship, but it is unclear if this is due to stroke reduction alone.
Methods Age- and sex-matched individuals from the U.K. Clinical Practice Research Datalink (2008–2016) with and without an incident diagnosis of AF were followed for a new dementia diagnosis. We estimated adjusted hazard ratios (aHRs) for incident dementia diagnosis in the AF cohort, overall and stratified by anticoagulation status, using the matched non-AF cohorts as reference. We performed a sensitivity analysis excluding individuals with stroke/transient ischaemic attack (TIA) before the observation period.
Results Over 193,082 person-years (mean follow-up 25.7 ± 0.1 months), 347/15,276 AF (2.3%) and 1,085/76,096 non-AF (1.4%) were newly diagnosed with dementia (aHR, 1.31, 95% confidence interval, 1.15–1.49). The AF group had more co-morbidity and higher rates of dementia, both with and without anticoagulation, than non-AF. When those with history of stroke/ TIA before the observation period were excluded and those with incident stroke/TIA during the observation period were censored, AF individuals not on anticoagulation had significantly higher rates of dementia compared with non-AF, aHR 1.30 (1.06–1.58).
Conclusion Our findings support the hypothesis that AF is a distinct risk factor for dementia, independent of stroke/TIA and other vascular risk factors. In those without stroke/TIA, risk of dementia is increased only in those who are not on anticoagulation, suggesting anticoagulation is protective presumably through reduction of sub-clinical embolic events. Further prospective research is needed to better ascertain the role of anticoagulation amongst targeted therapeutic strategies to reduce cognitive decline in AF.