How to Interpret Heart Rate Variability (HRV) Indices from the ECG in Dogs and Cats

Interpreting heart rate variability (HRV) from the ECG in dogs and cats means translating the beat-to-beat oscillation of the RR intervals (NN intervals, normal to one another) into indices that reflect the heart's autonomic modulation. These indices fall into three families: the time domain (SDNN, SDANN, RMSSD, pNN50), the frequency domain (the LF and HF bands and the LF/HF ratio) and geometric or nonlinear methods (the Poincare plot, the histogram and the triangular index). In healthy dogs, respiratory sinus arrhythmia is so pronounced that very high HRV values are physiological, not pathological: the starting point for any reading is always the species, the animal's state and signal quality, never an isolated number borrowed from another field. The clinical decision always remains with the veterinarian.

Before calculating: signal quality, ectopy and artifacts

HRV is only reliable over a series of truly normal intervals (NN) originating in the sinus node. For this reason, the step with the greatest clinical weight comes before any index: reviewing the tracing and removing atrial or ventricular premature complexes, escape beats, pauses and motion artifacts. A single ectopic beat or an artifact mistaken for a QRS produces a spurious RR interval that artificially inflates indices sensitive to successive differences, such as RMSSD and pNN50. In short-term HRV reference studies in healthy dogs, automatic inspection followed by manual review to eliminate atrial and ventricular premature complexes was an explicit step before calculation, precisely because the healthy dog already has large physiological variability that must be separated from noise.

In practice, this requires an adequate recording: a stable lead, good electrode contact and an acquisition time matched to the desired index. Short in-clinic windows are useful for some parameters, but robust prognostic stratification in heart disease usually comes from 24-hour Holter recordings, which capture the full cycle of sleep, rest and activity. Without this signal hygiene, the rest of the interpretation is compromised.

Time domain: SDNN, SDANN, RMSSD and pNN50

Time-domain indices describe, in milliseconds, how much the NN intervals vary. SDNN (the standard deviation of all NN intervals) is the most global index: it reflects total variability, with both sympathetic and parasympathetic contributions, and is strongly dependent on the recording length. SDANN, the standard deviation of the means of NN over short segments (typically 5 minutes across the 24 hours), captures slow oscillations and circadian cycles, that is, the long-term structure of variability.

RMSSD (the square root of the mean of the squared differences between successive intervals) and pNN50 (the proportion of successive intervals that differ by more than 50 ms), in turn, capture the fast, beat-to-beat variability. Both predominantly reflect short-term parasympathetic (vagal) tone, and that is exactly why they are markedly high in the healthy dog: vagally mediated respiratory sinus arrhythmia shortens the interval during inspiration and lengthens it during expiration, raising RMSSD and pNN50 to levels that, outside the species context, would sound exaggerated. In healthy dog series, RMSSD values in the hundreds of milliseconds and pNN50 above 70% have been described as normal. The clinically useful reading is the trend: a fall in these indices, especially the vagal ones, tracks the loss of autonomic modulation in the progression of cardiac and systemic disease.

Frequency domain: LF, HF bands and the LF/HF ratio

Spectral analysis decomposes the interval series into frequency components. The high-frequency (HF) band corresponds largely to respiratory sinus arrhythmia and is a robust marker of parasympathetic activity. The low-frequency (LF) band is under mixed sympathetic and parasympathetic influence, modulated by the baroreflex. There is also the very-low-frequency (VLF) band, whose physiological interpretation is less well defined. In healthy dogs, strong vagal modulation translates into high HF and, consequently, low LF/HF ratios, which reinforces that vagal predominance is the species' expected baseline state.

The LF/HF ratio is often presented as an index of sympathovagal balance, but it deserves a cautious reading in animals. The premise that LF represents purely sympathetic tone is a simplification inherited from human physiology and disputed even there; in dogs and cats, with their intense respiratory sinus arrhythmia and variable breathing, the separation between bands and their normalization (LFnorm, HFnorm) depend heavily on recording conditions. Feline studies have shown that pharmacological maneuvers shift the balance with a fall in HFnorm and a rise in LFnorm consistent with vagal withdrawal and sympathetic enhancement, which validates the direction of the signal but does not license treating LF/HF as an absolute, standalone measure of sympathetic tone. Use it as a trend within the same animal and the same protocol, not as a universal number.

Geometric and nonlinear methods: Poincare, histogram and triangular index

Geometric methods summarize the distribution of intervals. The NN interval histogram and the triangular index (the histogram area divided by the peak height) provide a measure of global variability that is little affected by isolated artifacts, a practical advantage in long recordings. The Poincare (or Lorenz) plot graphs each interval against the next, forming a cloud whose width (SD1) reflects the fast, vagally driven variability and whose length (SD2) reflects long-term variability; the shape of the cloud is interpreted visually beyond the measurements.

In the dog, the Poincare plot reveals something distinctive: the sinus rhythm is not merely variable, it is nonlinear. Veterinary cardiology work has shown that dogs, unlike the human pattern, display interval distributions with clusters, branches and zones with a scarcity of beats, and that these nonlinear patterns depend on parasympathetic modulation and are abolished by atropine. This has a direct clinical consequence: the diagnostic value lies in the shape of the plot, not only in a numeric index. In cases of loss of variability with sympathetic predominance, described in dogs with severe systemic disease, the Poincare plot contracts and loses its branched structure, in agreement with the reduced time- and frequency-domain indices. This is why geometric methods complement, and do not replace, the time and frequency domains.

Bringing HRV into clinical routine, with prudence

HRV is not a diagnosis in itself: it is a window into autonomic modulation that gains value when read serially, in the same patient, under standardized conditions. Reductions in global HRV and in vagal indices have been associated with progression and worse prognosis in canine heart disease, but the cutoffs vary with breed, body size, age, emotional state and acquisition protocol, and still lack broad standardization in veterinary medicine. Patient behavior matters: stress, pain and excitement in the clinic acutely reduce HRV and can mask the underlying picture.

The practical recommendation is to integrate HRV with the rest of the cardiology workup, the morphological ECG, physical examination, blood pressure and echocardiogram, and never to view it in isolation. Connected ECG platforms, such as the INpulse ecosystem with the INcardio X and INcardio Agile electrocardiographs integrated into INpulse One and INcloud, can make it easier to acquire a clean signal and to follow the same indices longitudinally over time. Even so, the interpretation of HRV indices and any decision derived from them are always the responsibility of the veterinarian.