Abstract 12125: The CHARGE-AF Risk Score Performs Better at Predicting Risk of Atrial Fibrillation than the CHA2DS2-VASc Risk Score in the Framingham Heart Study

Introduction: Atrial fibrillation (AF) is a common arrhythmia that affects more than 30 million individuals worldwide and confers serious complications, such as stroke, heart failure, dementia and death. The CHARGE-AF risk score was developed in 2013 to predict risk of AF in the general population. It was derived from the Framingham Heart Study, the Cardiovascular Health Study and the Atherosclerosis Risk In Communities study and validated in the Age, Gene/Environment Susceptibility Study and the Rotterdam Study, with more than 26,000 individuals of European, European-American, and African-American ancestry in total. The CHA2DS2-VASc risk score was derived in 2010 to predict risk of stroke in individuals already diagnosed with AF. CHA2DS2-VASc has not been validated for prediction of AF, but it has been used for this purpose in recent studies. Hypothesis: We hypothesized that the CHARGE-AF risk score would perform better than the CHA2DS2-VASc risk score for predicting risk of AF in a community-based cohort. Methods: Individuals from the Framingham Heart Study aged 45-94 years, without missing covariates or prevalent AF were included. For both risk scores, we performed Cox proportional hazards regression to assess the predicted risk of AF. Model fit was assessed using the likelihood ratio X2 statistic. We evaluated discrimination and calibration of each model using the C-statistic and Hosmer-Lemeshow X2 statistic. Results: We included 9722 observations (mean age 63.9±10.6 years, 56% women) from 4548 unique individuals: 752 (16.5%) developed incident AF and 793 (17.4%) died. The mean CHARGE-AF score was 12±1.2 and mean CHA2DS2-VASc score 2.0±1.5. The CHARGE-AF risk score was associated with a 115% increased risk of AF (95% CI, 99-131%; p<0.0001) per unit increase, corresponding numbers for CHA2DS2-VASc were 43% (95% CI, 37-51%; p<0.0001). CHARGE-AF had better model fit than CHA2DS2-VASc (Wald X2 = 403 vs. 209, both with 1 df), improved discrimination (C-statistic=0.75, 95% CI, 0.73-0.76 vs. C-statistic=0.71, 95% CI, 0.69-0.73) and better calibration (X2=5.6, p=0.69 vs. X2=28.5, p<0.0001). Conclusions: The CHARGE-AF risk score performed better than the CHA2DS2-VASc risk score at predicting AF in a European-American ancestry cohort.