Validity and training sensitivity of heart rate variability thresholds in patients with coronary artery disease

Abstract Background Accurate exercise-intensity determination is crucial in patients with coronary artery disease (CAD) to balance training effectiveness and clinical safety. Given the limitations of traditional thresholds, heart rate variability (HRV) thresholds derived from the short-term scaling exponent of detrended fluctuation analysis (DFA a1) have emerged as a practical alternative to ventilatory thresholds (VT1, VT2). Specifically, DFA a1 values of 0.75 and 0.50 have been proposed to indicate first (HRVT1) and second (HRVT2) HRV thresholds, respectively. Their application may be particularly relevant in cardiac rehabilitation, where repeated cardiopulmonary exercise tests (CPETs) are often unfeasible. However, evidence on their validity and sensitivity to training-induced adaptations in CAD remains limited. Purpose To examine (i) the agreement between HRVTs and VTs, and (ii) whether HRVTs reflect chronic changes following 12 weeks of aerobic training in patients with CAD. Methods Nineteen patients with stable CAD (15 men, 4 women) completed an incremental cycling CPET before and after a 12-week supervised aerobic-training program (moderate/high intensity, 3 sessions/week). VT1 and VT2 were determined using a mixed gas-exchange method, while HRVT1 and HRVT2 were identified at DFA a1 values of 0.75 and 0.50, respectively. Thresholds were expressed in heart rate (HR), power output (PO), and oxygen uptake (VO2). Agreement across methods and the relationship between pre–post changes were assessed using paired t-tests, mean differences, Pearson correlations, intraclass correlation coefficients (ICC), and Bland–Altman analysis. Results Significant differences were found for all VT–HRVT comparisons except HRVT2–VT2 in VO2. HRVT1–VT1 correlations were r = 0.25 for HR (p = 0.30; ICC = 0.61), r = 0.72 for PO (p 0.05; ICC = 0.86), and r = 0.62 for VO2 (p 0.05; ICC = 0.75). HRVT2–VT2 correlations were r = 0.35 for HR (p = 0.15; ICC = 0.67), r = 0.74 for PO (p 0.05; ICC = 0.89), and r = 0.64 for VO2 (p 0.05; ICC = 0.84). Bland–Altman analyses showed smaller mean differences between HRVT2 and VT2 than between HRVT1 and VT1. Sensitivity analyses revealed weak correlations between training-induced changes, with only ΔHRVT2-ΔVT2 in PO (r = 0.48, p = 0.15) and VO2 (r = 0.50, p = 0.12) showing low-to-moderate but non-significant associations. Conclusion Although HR values showed weaker and more inconsistent agreement, HRVTs demonstrated good agreement with VTs in PO and VO2 values, supporting their use as a practical alternative for exercise-intensity prescription in cardiac rehabilitation, particularly when gas-exchange testing is unavailable. However, their limited ability to track pre–post changes indicates insufficient sensitivity to detect training-induced adaptations in this CAD cohort. Further research with larger samples is warranted to clarify the potential of HRVTs for longitudinal monitoring in cardiac rehabilitation.