Non-linear RR interval metrics in canine atrial fibrillation and their relationship to 24-h mean heart rate

In both dogs and humans with atrial fibrillation (AF), rate control is monitored using the 24-h Holter mean heart rate (meanHR), which reflects only the central tendency of the RR interval distribution. Naturally occurring canine AF provides a translational model because of its comparable AV nodal physiology and autonomic regulation.

Fifty client-owned dogs with confirmed AF underwent 24-h Holter monitoring. Ten non-linear RR metrics were evaluated: coefficient of variation of RR (CVRR), Poincaré SD1, SD2, SD1/SD2 ratio, |ΔRR|, turning-point ratio, DFA-α, sample entropy, and multiscale entropy (MScEn). The Jensen-gap index (D) was also calculated; rate dependence was assessed using Spearman correlation and the within-quintile IQR ratio.

The metrics were partitioned into rate-coupled (CVRR, SD1, SD2, |ΔRR|, DFA-α; |ρ| = 0.64-0.91, R ² = 0.42-0.75) and rate-independent (SD1/SD2 ratio, turning-point ratio, MScEn; |ρ| < 0.14, R ² < 0.01) classes. All 50 dogs showed a negative Jensen gap (D: mean -13.4 bpm, range -29.9 to -5.8; ρ with CVRR = -0.934), with no discontinuity observed at 125 bpm. Five dogs were reclassified as non-rate-controlled based on medianHR. Within the rate-controlled group, CVRR ranged from 0.285 to 0.748. Rate-independent metrics differed between pre- and postdrug dogs (SD1/SD2 p = 0.020; MScEn p = 0.032). The 125-bpm threshold used to assess rate control in dogs dichotomizes a continuous relationship and collapses heterogeneous RR distributions, limitations that have direct parallels in human AF rate-control monitoring. Future studies in both canine and human AF should incorporate rate-independent metrics alongside meanHR and specify the HR computation method.