Statistical and Spectral Assessment

The technology for standard statistical and spectral assessment of HRV components has not advanced much since it was reviewed by the European Society of Cardiology (ESC)/North American Society of Pacing and Electrophysiology (NASPE) Task Force in the mid-1990s. Nevertheless, even when populations of cardiac patients treated according to contemporary standards are investigated, simple global HRV measurements are not losing their predictive power. Recent studies have shown that global HRV, that is, the expression of overall variability of RR intervals with no distinction between different (e.g., slower and faster) modulators, maintains its predictive value. In the field of prophylactic ICD implantation, it has been reported that global HRV measures were among the factors that most powerfully differentiated between patients with and without appropriate ICD shocks.¹¹ Consistent with what we have already described, the authors of this study observed appropriate ICD therapy in less than 20% of patients who had the device implanted; therefore they interpreted the HRV difference between patients who did and did not use the ICD appropriately as a possible route to improving the selection of device recipients.¹¹

Spectral analysis of HRV, that is, the distinction between modulators of cardiac periodicity operating at different frequencies, was originally seen mainly as a tool serving physiological studies. The spectral assessment of HRV is indeed linked to the physiology of the autonomic nervous system. Vagal afferent stimulation leads to reflex excitation of vagal efferent activity and inhibition of sympathetic efferent activity. Opposite reflex effects are mediated by the stimulation of sympathetic afferent activity. Efferent vagal activity also appears to be under “tonic” restraint by cardiac afferent sympathetic activity. Efferent sympathetic and vagal activities directed to the sinus node are characterized by discharge largely synchronous with each cardiac cycle that can be modulated by central and peripheral oscillators. These oscillators generate rhythmic fluctuations in efferent neural discharge that manifest as short- and long-term oscillation in the heart period. Efferent vagal activity is a major contributor to the so-called high-frequency component of HRV, that is, cardiac cycle modulations appearing at frequencies between 0.15 and 0.4 Hz (modulating waves of between approximately 2.5 and 7 seconds). Slightly more controversial is the interpretation of the low-frequency component, that is, slower modulations at frequencies between 0.05 and 0.15 Hz (modulating waves between approximately 7 and 20 seconds). These modulations are considered by some as markers of sympathetic modulation and by others as parameters that include both sympathetic and vagal influences.

Results of the standard spectral analysis of HRV also preserve their predictive power in contemporarily treated patients. They have recently been reported to improve the prediction of cardiac mortality in heart failure patients independently of other clinical risk characteristics including reduced LVEF and advanced age.¹²

Nonlinear Analysis

The association of spectral HRV components with different branches of the autonomic nervous system is based on well-established provocative experiments. Nevertheless, it is known that strictly engineering approaches to analysis of the series of cardiac periodograms cannot reflect all possible autonomic reflexes, especially during their dynamic interplay when underlying autonomic regulations are constantly changing. To address such characteristics of cardiac autonomic status, nonlinear approaches to the HRV analysis are frequently advocated. Different nonlinear characteristics have been proposed,13–15 but compared with the spectral HRV components, the physiological models underlying these characteristics are somewhat less well defined. Still, multiple reports have appeared showing the advantages of using nonlinear HRV indices for the purposes of cardiac risk prediction.¹⁶ Improvement in risk assessment by nonlinear HRV analysis has been reported not only in cardiac patients¹⁷ but also in other clinically well-defined populations.¹⁸

Nonlinear measures not only refine conventional approaches to HRV assessment. Even in terms of risk prediction, nonlinear HRV measures are seen to independently complement rather than supersede conventional HRV parameters.¹⁹ Although the corresponding physiological background is still to be researched, some of these measures actually seem to stratify patient groups with very different risk profiles from those seen in conventionally assessed overall HRV.

Some aspects of the relationship between conventional and nonlinear HRV indices are still poorly understood. Specifically, seminal studies of HRV use for the purposes of risk stratification indicate that the distinction between modulating frequencies of HRV components was of little importance for risk assessment. Early studies suggested that most of the predictive power was linked to very slow variations in cardiac periodicity that appeared not to be directly linked to any distinct physiological regulatory processes. It was believed that most of the risk prediction power derived from nonperiodic responses to the environment and the consequent circadian pattern of heart rate. However, more recent observations not only suggest that improvement in risk prediction can be made by concentrating on specific modulating processes (e.g., low-frequency modulations) but also show that different components (e.g., some of the nonlinear characteristics in comparison with conventional global HRV) select high-risk patient groups with substantially different clinical characteristics including responses to antiarrhythmic interventions.²⁰

Deceleration Capacity

The major recent advance in HRV methods for the purpose of cardiac risk prediction is probably related to the concept of the so-called deceleration capacity.²¹ This concept is based on so-called phase-rectified signal averaging, which is a universal technique capable of processing different medical and biological signals. It defines moments of interest in a “trigger” signal (e.g., beat-to-beat decelerations of cardiac periods) and averages surroundings of the moments of interest in an analyzed signal, which might be the same as the trigger signal or different recordings (in the deceleration capacity assessment, both signals are the same, that is, cardiac periodograms). It has been suggested that deceleration capacity technology is capable of distinguishing the vagal components of cardiac autonomic regulations irrespective of the lack of the stationary nature of analyzed data (which is always difficult to maintain in long-term recordings).^21,22 Initial observations showing that deceleration capacity is a more powerful risk predictor than conventional HRV measures²¹ have been independently confirmed.^16,23

The technology of phase-rectified signal averaging has many favorable properties and has been subject to further methodologic studies. Among other properties, it has been reported that short-term deceleration capacity is more stable and more reproducible than conventional HRV indices.²⁴ Also, it appears that deceleration capacity is capable of providing useful prognostic information from mid- and short-term recordings (e.g., 30-min Holter data), which makes assessment of cardiac risk much more practical and feasible in the standard clinical setting compared with conventional HRV indices, which are well known to perform best if assessed from nominal 24-h recordings that include day/night differences in the circadian pattern.