Development and validation of the HARMS2-AF lifestyle risk score to predict incident AF

Abstract

Abstract Background Lifestyle risk factors (RFs) are a modifiable target in atrial fibrillation (AF) management. However, the relative contribution of individual lifestyle RFs to AF incidence has not been described. Purpose Development and validation of a novel AF-lifestyle risk score to determine AF risk in the general population. Methods The UK Biobank (UKB) is a large prospective cohort with outcomes measured >10 years. In the UKB, we performed regression analysis of AF lifestyle RFs which were then evaluated in a multivariable model and a weighted score was developed. Next, the risk score was externally validated in the Framingham Heart Study (FHS) population. Kaplan-Meier estimates ascertained the 10-year risk of AF development. Results In the UKB, AF incidence was 5.3% among 302,926 participants, with a median time to AF 7.3 years (IQR 4.3–9.8). Hypertension, sleep apnoea, male sex, age, obesity (BMI>30 kg/m2), alcohol and smoking were predictive variables (all p<0.001); physical inactivity (OR 1.02,95% CI 0.97–1.10, p=0.3), diabetes (OR 0.98,95% CI 0.91–1.06, p=0.2) and BMI 27–30 kg/m2 (OR 1.02, 95% CI 0.97–1.07, p=0.424) were not significant. The HARMS2-AF score (Figure 1) had similar predictive performance (AUC=0.782, LogLoss 0.178, Brier Score 0.046) to the unweighted regression model (AUC 0.808) in the UKB. Validation in the FHS (AF incidence 6.7% of 7206 participants) maintained excellent predictive performance with an AUC of 0.747 (95% CI 0.724–0.769, Figure 2). A higher HARMS2-AF score (>5 points) was associated with a heightened 10-year AF risk (score 5–9: OR 9.35, score 10–14: OR 33.34). Conclusions The HARMS2-AF score is a novel lifestyle risk score which may help identify individuals at risk of AF and assists in general population screening. Funding Acknowledgement Type of funding sources: None.