External validation of H2ARDD score for the prediction of heart failure events in patients with atrial fibrillation

Abstract Background Heart failure is an important consequence in patients with atrial fibrillation (AF) which is associated with worse prognosis. The H2ARDD score, calculated from 5 clinical risk factors, was reported as a predictor of heart failure events in patients with AF. However, this score has not been externally validated. Purpose The purpose of this study was to evaluate and validate the usefulness of the H2ARDD score for the prediction of heart failure events in AF patients. Methods We used prospective data of patients with AF followed up from 2007 to 2017 in our institute. Patients with active cancer were excluded according to the previous report. H2ARDD score was calculated as follows; history of heart disease=2 points, anemia=1 point, renal dysfunction=1 point, diabetes =1 point, diuretic use=1 point (range from 0 to 6 points). Outcome of interest was defined as heart failure events including new-onset heart failure and death with heart failure. Heart failure was ascertained based on the Framingham criteria. Univariable and multivariable Cox-proportional hazards model were used to assess the risk of heart failure events. Heart failure events-free survival was estimated with Kaplan-Meier methods, and the predictive accuracy of the H2ARDD score for the prediction of heart failure events was measured by the area under the receiver operating characteristic (ROC) curve. Results Of 562 AF patients, 522 (age 69±10 year–old, 64.9%men) met study criteria. Patients who had a history of heart disease was 185 (35%), diabetes mellitus was 135 (26%), anemia was 54 (10%), renal dysfunction was 221 (43%), and diuretic use was 193 (37%). The mean H2ARDD score was 1.88±1.57. Of all study patients, 84 (16.2%) developed heart failure events during a mean follow–up of 54±42 months. Patients who developed heart failure events in 1 year was 24 (4.6%). In multivariable Cox–proportional hazards model, H2ARDD score was shown as an significant predictor for heart failure events (hazard ratio: 1.56, 95% confidence interval: 1.36 - 1.79, P