Association of chocolate consumption with neurological and cardiovascular outcomes in atrial fibrillation: data from two Swiss atrial fibrillation cohort studies (Swiss-AF and BEAT-AF)

AIM: To assess the associations of chocolate consumption with neurocognitive function, brain lesions on magnetic resonance imaging (MRI), and cardiovascular outcome in patients with atrial fibrillation (AF). METHODS: We analysed data from patients of two prospective multicentre Swiss atrial fibrillation cohort studies (Swiss-AF) and (BEAT-AF). Assessments of MRI findings and neurocognitive function were performed only in the Swiss-AF population (in 1727 of 2415 patients [71.5%] with a complete data set), as patients enrolled in BEAT-AF were not systematically evaluated for these outcomes. Otherwise, the two cohorts had an equivalent set of clinical assessments. Clinical outcome analysis was performed in 3931 patients of both cohorts. Chocolate consumption was assessed by questionnaire. Patients were categorised as no/low chocolate consumption (No/Low-Ch) ≤1 servings/week, moderate chocolate consumption (Mod-Ch) >1–6 servings/week, and high chocolate consumption (High-Ch) >6 servings/week, respectively. Brain lesions were evaluated by MRI. Assessment of cognitive function was performed by neurocognitive functional testing and included global cognition measurement with a cognitive construct score. Cerebral MRI and cognition were evaluated at baseline. Cross-sectional associations between chocolate consumption and MRI findings were analysed by multivariate logistic regression models and associations with neurocognitive function by multivariate linear regression models. Clinical outcome events during follow-up were recorded and assessed by a clinical event committee. The associations between chocolate consumption and clinical outcomes were evaluated by Cox regression models. The median follow-up time was 6 years. RESULTS: Chocolate consumption was not associated with prevalence or volume of vascular brain lesions on MRI, nor major adverse cardiac events (ischaemic stroke, myocardial infarction, cardiovascular death). However, No/Low-Ch was independently associated with a lower cognitive construct score compared to Mod-Ch (No/Low-Ch vs. Mod-Ch: coeff. –0.05, 95% CI –0.10–0), whereas other neurocognitive function tests were not independently associated with chocolate consumption categories. In addition, there was a higher risk of heart failure hospitalisation (No/Low-Ch vs. Mod-Ch: HR 1.24, 95% CI 1.01–1.52) and of all-cause mortality (No/Low-Ch vs. Mod-Ch: HR 1.29, 95% CI 1.06–1.58) in No/Low-Ch compared to Mod-Ch. No significant associations with the evaluated