Atrial Fibrillation and Coronary Events: Competing Risks and Bidirectional Mediation in a Longitudinal Swedish Cohort

BACKGROUND: The prevalence of atrial fibrillation (AF) and coronary events (CEs) overlap, yet their differential associations with risk factors and reciprocal mediation remain poorly characterized. METHODS: In a large Swedish cohort study (baseline: 1991–1996), subjects without preexisting AF, CEs, stroke, or heart failure were analyzed. Associations between baseline risk factors (age, sex, body mass index or waist circumference, smoking, alcohol intake, systolic blood pressure, diabetes, apoA1 (apolipoprotein A1) and apoB (apolipoprotein B) levels, leukocyte counts, education, physical activity, and medications) and incident AF or CEs were assessed using multivariable-adjusted Cox models, with association strengths compared using competing risk analysis. For each risk factor, we performed multivariable-adjusted mediation models for survival data separately, first with AF as a time-varying mediator for CEs, then reciprocally with CEs for AF, to estimate bidirectional mediation pathways. RESULTS: Among 25 963 subjects (aged 58.0±7.60 years, 62.0% women), 5447 incident AF and 3462 incident CEs occurred (1125 overlapped cases) over a median follow-up of 24.6 and 24.9 years, respectively. Various traditional risk factors were predominantly associated with CEs, whereas certain ones, particularly adiposity indices, demonstrated stronger associations with AF. Notably, higher apoA1 levels were associated with higher AF risk but lower CEs risk, whereas higher apoB showed opposite associations ( P for equal associations