Standard ECG features including PR interval, corrected QT, and T axis were independently associated with higher risks of incident heart failure hospitalization in patients with CKD (p<0.001).
Cohort (n=11,409)
Do standard ECG features or heart rate variability parameters predict de novo HF hospitalization in individuals with CKD?
Standard ECG features, including PR interval, corrected QT, and T axis, independently predict de novo heart failure hospitalization in patients with chronic kidney disease, whereas heart rate variability does not.
p-value: p=<0.001
Background: Tools for predicting heart failure (HF) in CKD patients remain limited. We aimed to study whether standard ECG features or heart rate variability parameters predict de novo HF hospitalization in individuals with CKD. Methods: Utilizing a large NYU ECG database linked with electronic health records (2012-2021), we analyzed a cohort of patients with pre-existing CKD . Besides standard ECG features, we extracted heart rate variability (measures the time between consecutive heart beats in milliseconds) features from the ECGs as predictors. The index ECG was the first ECG performed after the index eGFR date (baseline) and was required to be done prior to initiation of dialysis, end-stage kidney disease (ESKD), or transplant. The primary outcome was time to index HF hospitalization (≥30 days after the index ECG) based on discharge ICD-10 codes. LASSO-penalized Cox regression was used to identify predictors. Sensitivity analyses used Fine-Gray competing risk models for death and ESKD. Results: Among 11,409 individuals (median age: 72; ∼50% male) with a median of 976 days, 880 individuals (8%) experienced an index HF hospitalization. Models incorporating ECG and clinical parameters had excellent discrimination (C-statistic 0.76 in the training set and 0.73 in the validation set). Among ECG features, the PR interval, corrected QT, and T axis were independently associated with higher risks of index HF hospitalization ≥30 days after the index ECG in both primary models (p<0.001 for all) and in models accounting for competing risks (p<0.01 for all). History of arrhythmia (hazard ratio (HR, 1.60, 95% CI: 1.36-1.88), valvular disease (HR1.51, 95% CI: 1.27-1.81), and diabetes (HR 1.41, 95% CI: 1.22-1.65) were the strongest clinical predictors. HRV parameters were not independently associated with HF. Conclusions: Although ECG-derived HRV indices were not independently associated with risk of HF, several standard ECG features are associated with HF hospitalization in CKD.
Soomro et al. (Tue,) conducted a cohort in Chronic Kidney Disease (n=11,409). Standard ECG features and heart rate variability parameters was evaluated on time to index HF hospitalization (≥30 days after the index ECG) (p=<0.001). Standard ECG features including PR interval, corrected QT, and T axis were independently associated with higher risks of incident heart failure hospitalization in patients with CKD (p<0.001).