Atrial fibrillation development in the heart failure population from nationwide British linked electronic health records

Aims: Atrial fibrillation (AF) is a frequent comorbidity in heart failure (HF). We analysed factors associated with new-onset atrial fibrillation in patients with heart failure using linked real-world UK data from primary and secondary care, along with findings from genome-wide association studies. Methods and results: Among 163 174 participants with a diagnosis of HF (January 1998 to May 2016) from Clinical Practice Research Datalink (CPRD) and Hospital Episodes Statistics (HES), 111 595 participants had no previous history of AF (mean age 76.3 ± 12.6; 50.3% women; 95.8% white ethnicity). Multivariate weighted Cox regression was used to identify predictors for new-onset AF. Linkage disequilibrium score regression was performed to assess the strength of the genetic correlation between AF and identified predictors. During follow-up (median 1.33 years, IQR 0.15–4.18), the incidence rate for AF was 2.8% at 30 days, 9.9% at 1 year, 18.0% at 3 years, and 24.9% at 5 years after HF diagnosis after HF diagnosis. Female sex (HR = 0.79, 95% CI 0.71–0.88), age (HR = 1.04, 95% CI 1.04–1.04), white ethnicity (HR = 1.30, 95% CI 1.06–1.59), social deprivation (HR = 1.20, 95% CI 1.01–1.42), BMI (HR = 1.01, 95% CI 1.00–1.02), gentle physical activity (HR = 0.84, 95% CI 0.72–0.97), hypertension (HR = 1.15, 95% CI 1.03–1.29), chronic kidney disease (HR = 1.15, 95% CI 1.06–1.24), chronic obstructive pulmonary disease (COPD) (HR = 1.10, 95% CI 1.01–1.19) and valvular heart disease (HR = 1.17, 95% CI 1.06–1.29) were associated with new-onset AF. Angiotensin-converting enzyme inhibitors were associated with lower AF incidence (HR = 0.88, 95% CI 0.80–0.96), and the magnitude of effect was dependent on the duration of administration. Linkage disequilibrium score regression showed important genetic correlation between AF and HF (rg = 0.57, P = 2.30 × 10^-59) and reduced, but still significant, overlap between AF and BMI (rg = 0.19, P = 6.18 × 10^-20), systolic and diastolic blood pressure, smoking, and COPD (P values ranging from <10^-4 to <0.05). Conclusions: Incident AF in the HF population is high, with good genetic correlation for the two conditions. Identified predictors for new-onset AF might be helpful to improve management of HF patients and AF prevention.