Abstract 13420: Association of Digital ECG Biomarkers With Heart Failure Status

Abstract

Introduction: Electrocardiography (ECG) may be useful as an adjunct method for evaluating heart failure (HF) severity, but few studies have systematically examined the electrocardiographic parameters associated with heart failure volume status. This may include effects of hypervolemia on thoracic impedance, autonomic function, and subendocardial ischemia. We hypothesized that ECG parameters related to these mechanisms co-vary over time with brain natriuretic peptide, a measure of HF severity. Methods: A retrospective cohort was obtained from the Emory Clinical Data Warehouse for patients who had >2 ECGs performed >3 days apart and repeated BNP measures within 24 hours of each ECG. We quantified the change in HF status by calculating the differences in BNP value between each ECG acquisition time point. A total of 37 metrics were extracted from digitized ECGs as primary predictors and evaluated using multiple linear regression. Results: We evaluated 10,376 pairs of ECGs from 7,171 unique patients. The mean age was 66.9 years, 48% were women, and 51.4% were black . The median absolute change in BNP over time was 178 pg/mL (IQR 52-532), and the median sampling interval was 106 days (IQR 27-357). We evaluated interval, duration, and axis metrics from the P, QRS, and T vectors from the raw digital ECG and found 8 co-varied with BNP changes, including heart rate, P wave axis, QRS amplitude, QRS duration, QRS axis, ST segment, T wave amplitude, and frontal T wave axis (see table). Conclusions: We found several ECG metrics that quantify autonomic activity, depolarization, and repolarization associated with changes in HF status as measured by BNP. These findings may help to understand the downstream effects of hypervolemia on cardiovascular risk and also understand the potential role for certain ECG metrics in helping determine volume status, especially in low-resource settings not offering BNP.