Validation of ARIC heart failure risk score in the Asian population: The CORE-Thailand cohort study

Abstract Background The Atherosclerotic Risk in Communities (ARIC) heart failure (HF) score was developed from population in the United States to predict HF. However, it has not been validated in Asian population. Purpose To validate the ARIC HF score and develop the new ARIC-CORE score to predict HF in Asian population in the Cohort Of patients with high Risk for cardiovascular Events (CORE-Thailand). Methods The CORE-Thailand registry was a prospective study that enrolled Thai patients with high atherosclerotic risk. Patients were followed for the occurrence of cardiovascular events including HF. The ARIC HF score comprised nine variables (age, race, heart rate, systolic blood pressure, BP-lowering medication use, diabetes, coronary heart disease, smoking status, and body mass index) to predict the occurrence of HF. Original coefficients of the full ARIC HF model were used to calculate the original ARIC HF score. The new ARIC-CORE score was developed by re-estimating the coefficients of all ARIC score variables in the study population using ridge regression analysis. The polynomial function was applied to the continuous variables with non-linear association with HF, which included heart rate, systolic blood pressure, and body mass index. Discrimination of the models was determined using C-index and calibration was assessed by plotting the observed outcomes versus the predicted probabilities of the outcomes. Net reclassification index (NRI) was used to compare the prediction performance between the ARIC HF and the ARIC-CORE risk score. Clinical utility of the scores were assessed with decision curve analysis. Results From a total of 8919 patients who had no history of HF, 199 (2.23%) developed HF. The ARIC HF score provided a good discrimination at C-statistics of 0.71 (95% confidence interval (CI) 0.67-0.75) with good calibration. (Figure 1 A). The ARIC-CORE risk score had a C-statistics of 0.75 (95%CI;0.71-0.76) with good calibration. (Figure 1B) Using the ARIC-CORE was associated with an improved classification of HF risk (NRI 0.32, 95%CI:0.18-0.44) compared to the ARIC HF score. The net benefit of using the ARIC-CORE score was higher than the original ARIC HF score in the decision curve analysis (Figure 2). Conclusion The ARIC HF score had acceptable discrimination and calibration in the CORE cohort. Interestingly, the ARIC-CORE score performed better in classifying HF risk and could be a suitable model to be used in Asian populations.Calibration plotDecision curve analysis