Abstract 14014: Inclusion of Chronic Kidney Disease Measures Improves Heart Failure Risk Prediction: The Atherosclerosis Risk in Communities (ARIC) Study
Neel Koyawala, Justin B Echouffo, Sui Zhang, VIJAY NAMBI, Morgan Grams, Kuni Matsushita, Roger S Blumenthal, Janani Rangaswami, Christie M Ballantyne, Josef Coresh, Chiadi E Ndumele- Physiology (medical)
- Cardiology and Cardiovascular Medicine
Introduction: Although chronic kidney disease (CKD) is closely linked to hypertension and diabetes, and is an independent risk factor for heart failure (HF), most HF risk prediction algorithms do not include CKD measures as covariates.
Hypothesis: We hypothesized that the addition of estimated glomerular function (eGFR), using creatinine (Cr) and cystatin C (CysC), and urine albumin to creatinine ratio (UACR) to the ARIC HF Risk Score will improve risk discrimination for HF.
Methods: We performed a prospective analysis of participants at ARIC Visit 4 (1996-1998) to assess changes in risk prediction after incorporating CKD measures. eGFR was estimated by both Cr and by Cr-CysC using CKD-EPI 2021 race-free equations. Using eGFR and UACR, we categorized CKD into four risk progression levels per the Kidney Disease Improving Global Outcomes (KDIGO) heat map. We assessed changes in 10-year risk discrimination of the ARIC HF Risk Score with the addition of eGFR-Cr, eGFR-Cr-CysC, and with KDIGO CKD categories using eGFR-Cr+UACR and eGFR-Cr-CysC+UACR. Analyses were performed in the overall population and stratified by hypertension and diabetes.
Results: Among 10,277 participants (mean age 63.2 years, 56% women, 21.4% Black, 15.5% diabetes, 40.8% hypertension), 845 people developed HF over 10 years. Risk discrimination progressively improved by adding eGFR-Cr, eGFR-Cr-CysC, and KDIGO categories with eGFR-Cr+UACR and eGFR-Cr-CysC+UACR to the ARIC HF Risk Score (C-statistic base model, 0.783 [95% CI, 0.768, 0.798] versus 0.796 [95% CI, 0.781, 0.810] with KIDIGO categories using eGFR-Cr-CysC+UACR). Greater improvements in risk discrimination were observed among those with diabetes and hypertension than in those without these conditions (Table).
Conclusion: Incorporating eGFR and UACR in prediction algorithms improves HF risk discrimination and can help inform HF screening and prevention efforts, particularly among individuals with diabetes and hypertension.