Although it appears to beat at a fairly regular pace, the speed with which the heart beats is in fact quite variable. Over the course of a minute, heart rate variability can spike and fall numerous times. This variability reflects changes in autonomic nervous system activity and can be affected by respiration, movement, and psychological states among other factors. Heart Rate Variability (HRV) provides us with a measure of these normal naturally occurring changes in heart rate.
To understand how and why HRV reflects autonomic activity, let’s first take a look at the heart and how the heart works.
THE HEART
How does the heart work?
The heart is a remarkable feat of engineering. It works endlessly, tirelessly every minute of every day of your life. It beats about 70 times per minute, which works out to be about 100,000 times per day, pumping more than 7000 liters of blood through your body in that same day. Yet it is compact and light, about the size of your fist and weighing about 200-425 grams.
The heart consists of 4 chambers. The two chambers on the right side of the heart, the right atrium and right ventricle, receive de-oxygenated blood returning from the body and pumps it into the lungs for oxygenation. Oxygenated blood from the lungs is returned to the left side of the heart, to the left atrium and left ventricle, where it is pumped out to all living tissue in the body.
Figure 1. The human heart
Blood returning from the body enters the right atrium via the superior and inferior vena cavae and then moves into the right ventricle where it is pumped out to the right and left lungs via the pulmonary artery. Blood returning from the lungs enters the heart via the pulmonary veins into the left atrium, and moves into the left ventricle where it is pumped out to the body via the aorta arteries.
Electrical conduction in the heart
To move blood through the body, the heart must contract. A coordinated system of electrical conduction in the heart enables it to do just that.
Figure 2. Electrical conduction system of the heart.
Each normal heartbeat begins with an electrical impulse arising at the sinoatrial node (SA node), which sits at the top of the right atrium. The SA node is commonly referred to as the heart’s natural pacemaker. The signal propagates from the SA node down nerve fibers to the atrioventricular node (AV node), where it is delayed slightly. This phase of the heart contraction cycle is known as atrial systole, and is characterized by the contraction of the atria causing ejection of blood into the ventricles (the lower chambers of the heart). The delay at the AV node enables blood to fully eject from the atria into the ventricles[1]. After the delay, electrical signals propagate through the bundle of HIS and down through the Purkinje fibers surrounding the ventricles, thereby causing the ventricles to depolarize and contract. This phase of the heartbeat cycle is known as ventricular systole. Following this phase, the ventricles enter a phase of repolarization wherein they relax and the resting state is restored, ready for the next cycle. This final phase is known as ventricular diastole.
HEART RATE VARIABILITY (HRV)
What is heart rate variability?
Heart rate is typically quantified using a single number, beats per minute. For example, the normal resting heart rate for a healthy adult is 60 to 100 beats per minute. This single metric implies that the rate at which the heart beats is fairly constant, at least over the minute to which the metric applies. On the contrary, the rate at which the heart beats changes often, sometimes speeding up, and sometimes slowing down. The phenomenon by which oscillations are observed in heart rate over time is referred to as heart rate variability (HRV). This variability is normal and in fact relatively lower levels of HRV have been linked to several negative health outcomes. If you’ve noticed the slowing of your heart rate during expiration, then you have experienced this natural variability that occurs all the time.
What causes the variability?
Heart rate is determined by a wide range of factors including respiration, physical exercise, blood pressure feedback mechanisms involving baroreflex sensor signals, thermoregulation, the renin-angiotensin system, circadian rhythms, and a host of psychological factors (Stein et al, 1994). The autonomic nervous system (ANS), which is the part of the peripheral nervous system that operates primarily below conscious awareness to control a wide range of bodily functions, is the primary mechanism by which most of these factors operate to influence heart rate. There are two branches of the ANS, the sympathetic nervous system (SNS) and the parasympathetic nervous system (PNS). The heart rate estimated at any given time predominantly represents the net effect of these two branches of the ANS, which continuously produce variations. Sympathetic branch activation leads to cardio-acceleration while parasympathetic (PNS) activation results in an inhibition of SA node firing via the vagus nerve, resulting in cardio-deceleration. SNS activity leads to a smaller interval between the beats and PNS activity leads to a longer interval between beats (Thayer and wager, 2011).
The SNS and PNS operate at different speeds and this fact can be exploiting in helping us in our assessment of HRV. SNS activity promotes a quickening of heart rate but its response is slow, on the order of 1-3 seconds. By contrast, PNS activity slows heart rate by inhibiting the signals that prompt beats. Its response is considerably faster, at 0.2-0.6 seconds. We will return why this difference in speed is important when we discuss how to measure HRV.
It is important to realize that even without any ANS influences the heart would continue to beat as it has an intrinsic pacemaker. Left to its own, the heart would beat away at about 100 beats per minute. Resting heart rate is typically lower than this, in the range of about 60 to 100 beats per minute. How is this so? It is because of the influence of the PNS by way of the vagus nerve. At rest, the PNS influence is greater than SNS influence and the result is a slowed heart rate. During physical exercise or when experiencing psychological stress, sympathetic activity increases while parasympathetic influence decreases, the result being an increase in heart rate. In fact, in animal models, ablation of vagal nerve (PNS) influences results in faster resting heart rates than prior to the ablation, indicating that PNS tone dominates heart rate.
The variability therefore comes about as a result of PNS influences. It is for this reason that HRV is a useful measure of the balance of SNS to PNS influences. As a result, HRV has been found to reflect cardiorespiratory control and has proven a valuable indicator of SNS and PNS activity.
We next turn to how to measure HRV.
HOW TO MEASURE HRV?
To measure HRV, information regarding beat-to-beat intervals is needed. Two ways by which this information can be obtained is via electrocardiogram (ECG) or via photoplethysmogram (PPG). An ECG measures electrical activity in the heart, which is typically accomplished by way of electrodes affixed to the body, although there are ambulatory devices such as the Zephyr BioHarness that are simpler to use. Photoplethysmography senses small changes in blood volume in subcutaneous tissues by shining a light through the skin and measuring the variations in light absorption using a small receptor typically worn on the finger or ear. ECG data provide more complete data and therefore we will focus on obtaining HRV metrics from ECG data.
What is an electrocardiogram?
An electrocardiogram (ECG) noninvasively detects the small signals at the surface of the skin produced by the polarization and depolarization of the heart muscle, and then visualizes this activity as a waveform.
Fig. 3. Example of ECG.
This waveform can reveal a great deal about the activity and health of the heart and of the autonomic nervous system, including rate and variability of heartbeats, presence of cardiac damage, and the effects of drugs. The figure below shows an example of an ECG trace. Each large spike represents a heartbeat.
Figure 4. Typical placement of electrodes in a commonly used 12-lead ECG.
Cardiac electrical impulses are typically picked up via a multi-lead system in which electrodes are affixed to the chest in the region around the heart as well as on the right and left arms and legs. Each of the electrodes provides a different “view” of the electrical activity in the heart and together they can be used to infer specific areas of damage in the heart and many other conditions. For HRV analyses, only a single ECG trace is necessary and this is important as it permits the possibility of easier to use and more ambulatory options. In fact, ambulatory devices such as the Zephyr BioHarness system provide the equivalent of a 1-lead ECG.
How can heart rate variability be extracted from ECG?
An ECG detects electrical impulses as they travel through the heart with each beat and plots these impulses as a distinctive waveform. An ECG waveform consists of series of specific characteristics, each of which reflects a particular phase of the cardiac cycle. Each waveform can be decomposed into several segments (see figure).
The P wave reflects initiation of an impulse from the SA node and activation of the atria, during which time the atria contracts and ejects blood from the atria into the ventricles.
Figure 5. PQRST complex
The flat PR segment represents the short interval during which the signal is delayed at the AV node, which allows blood to fully eject into the ventricles. The QRS complex represents the quick impulse that propagates from the AV node down through the intra-atrial septum, through the bundle of HIS and through the Purkinje fibers, causing the ventricles to contract. The ST segment represents the interval between ventricular depolarization and repolarization. The T wave represents the repolarization of the heart. While these different components are all important indicators of cardiac function and can be used by clinicians to diagnose cardiovascular dysfunction and disease, for HRV analyses, the key task of interest in ECG waveform is the identification of the QRS complexes.
The sequence of time intervals between heart beats represents the raw data needed to calculate HRV. The relative influence of the sympathetic and parasympathetic branches determines HR. Increases in SNS activity relative to PNS activity increases HR and therefore the heart beats become closer together in time. By contrast, increases in PNS activity relative to SNS activity decrease (inhibits) HR, making heartbeats grow farther apart.
As indicated earlier, sympathetic and parasympathetic branch activity occurs at different speeds, with the parasympathetic system operating considerably faster than the sympathetic system. We can take advantage of this fact in decomposing the relative contributions of each branch by performing a frequency analysis on heart rate time series data.
A series of beat-to-beat (i.e., R-R) intervals over some period of time, say 5 minutes, forms a time series. The power spectrum of a time series represents the amount of variability (i.e., fluctuations) observed at every frequency. Relatively high variability in a time series should thus show up as high power at higher frequencies relative to lower frequencies. We delve more deeply into the topic of frequency analyses later on in the discussion of frequency domain HRV analyses.
The SNS works relatively slowly, operating at the timescale of seconds, whereas PNS influences operate on the timescale of milliseconds. PNS but not SNS signals are therefore the only influences capable of inducing rapid changes in beat-to-beat intervals. The rapid fluctuations provoked by the PNS therefore enable it to influence HR over the entire frequency range of the HR power spectrum, whereas SNS influences tend to be confined to the slower frequency bands, lower than 0.15 Hz. Therefore the higher frequency bands are considered to be associated with PNS activity, whereas lower frequency bands are generally regarded as representing a mixture of both SNS and PNS influences.
Differing contributions from SNS and PNS rhythms result in the modulation of intervals between R to R peaks (i.e., R-R intervals) in the ECG at specific and distinct frequencies. SNS influences are associated with fluctuations within the low frequency range (.04 – 0.15 Hz) and PNS activity is associated with fluctuations in the higher frequency range (0.15 – 0.4 Hz). This decomposition into distinct frequency ranges allows the ability to quantify the separate SNS and PNS contributions.
There are primarily two ways to quantify HRV: 1) time domain and 2) frequency domain measures.
Time domain measures
Time domain measures quantify variability in terms of heart rate at any given point in time or the intervals between successive beats (Task Force, 1996). The inter-beat intervals in an HR time series will vary over the course of time. Sometimes they are shorter and sometimes longer. Time domain measures quantify this variability. A large number of time domain indices can be found in the literature. The simplest is the standard deviation of the N-N interval (SDNN). Because variance is equal to the total power of the spectral analysis, SDNN indexes all the components responsible for variability in the time series under consideration. It is important to realize that the total variance of a time series will increase as the series lengthens. As a result, SDNN depends on the length of the time series, making it inappropriate to use for comparison purposes, if they are based on different time series lengths. An alternative is the standard deviation of the average NN interval (SDANN), which is typically computed over short periods, typically 5 mins and used to estimate changes in heart rate over longer periods of time.
HRV measures computed based variations in successive intervals include the square root of the mean squared differences of successive NN intervals (RMSSD), the number of interval differences of successive NN intervals greater than 50 ms (NN50), and the proportion derived by dividing NN50 by the total number of NN intervals (pNNSO). All these measurements have been shown to reflect variations in the high frequency band of the power spectrum and are themselves highly correlated (Task Force, 1996). Because PNS influences extend along all frequencies but SNS influences are restricted to lower frequencies, these indices therefore mainly reflect PNS activity.
Frequency domain measures
While time domain measures index the amount of variability in R-R intervals measured in units of time (seconds), frequency domain measures index how variable the fluctuations are at various frequencies in an R-R time series.
The power spectrum of a time series represents the frequency content within that time series and the power contained in every frequency. Thus if a selected R-R series has significant variability, it is expected that there would be a greater contribution at higher frequencies relative to the lower frequencies that are associated with lower variability. A mathematical transformation is used to convert R-R data into a power spectral density (PSD) representation. This enables us to visualize the R-R signal as being broken down its constituent frequency components and quantifies the relative power of these components. Thus, the PSD enables the determination of the amount of variance or power as a function of frequency ranges. The HF signal band (.15 and .4 Hz) measures the effect of PNS-mediated RSA on heart rate. The LF signal band (.04 to .15) reflects both SNS and PNS activity as well as fluctuations relating to the regulation of blood pressure, and vasomotor tone. Three frequency bands are typically used for HRV analyses: high frequency (HF), low frequency (LF), and very low frequency (VLF). Therefore, HF power primarily measures the variability prompted by respiration. HF power correlates highly with the time-domain measure RMSSD.
Fig 6. Power density frequency diagram
It has been posited that the LF band marks sympathetic activity and that the LF/HF ratio indexes the relative contribution of each ANS branch in the regulation of HR (Malliani, Pagani, Lombardi, & Ceruttti, 1991). While HF clearly reflects PNS activity, controversy has surrounded the question of whether LF indexes SNS tone. This interpretation has mostly been based on findings from studies involving orthostatic tilt, which are typically associated with an increase in LF and a decrease in HF fluctuations, thereby producing an increase in the LF/HF ratio, with magnitude of tilt correlated with extent of degree of tilt. More recent findings suggestion, however, that LF represents both sympathetic and vagal elements (Berntson et al., 1997; Japundzic, Grichois, Zitoun, Laude, & Elghozi, 1990; Randall, Brown, Raisch, Yingling, & Randall, 1991; Task Force, 1996). In a recent review by Reyes Del Paso et al (2013) they examined the literature on the topic and concluded that the HRV power spectrum is dominated by SNS activity. In particular they found the following: 1) vagal blockade reduces LF power by 90%, but SNS blockade has little effect, 2) experimental manipulations do not increase LF but rather very often reduce it, 3) drugs that induce changes in sympathetic tone do not affect LF in a manner consistent with LF as a specific indicator of LF activity, 4) LF power is not correlated with valid indicators of sympathetic cardiac control, and 5) computation of the LF/HF ratio is based on an assumption of LF/HF reciprocity where HF goes down when LF goes up, an assumption not born out in the current literature. In short, it appears that the entire HRV spectrum, including the LF band, is almost entirely determined by PNS vagal activity. Indeed Reyes Del Paso et al (2013) concluded that “the suitability of HRV analysis is restricted to the estimation of parasympathetic influences on HR, whereas further interpretations of spectral components as extra- vagal have to be regarded as misleading.” (p. 484). These authors also make the point that although the HRV power spectrum is dominated by PNS influences, the different frequency bands likely reflect distinct physiological mechanisms. While HF fluctuations reflect respiratory influences, LF fluctuations provide information about baroreflex regulatory activity, whereas VLF power is associated with kidney function and thermoregulation (Berntson et al., 1997; Task Force, 1996; Taylor et al., 1998; van Roon et al., 2004).
Applications of HRV analyses
As a reliable non-invasive indicator of autonomic activity, HRV has proven to be remarkably useful across a wide range of domains and purposes. Indeed to cover all the applications of HRV would require a large article in itself. Here, I will provide only a hint of the different ways in which HRV can be used:
- Biofeedback
- Information about one’s current HRV is used as part of a treatment. For example, an ear-based photoplethysmographic sensor can convey info on heart rate cycles to an iPhone app where they can be visualized while an individual practices a relaxation exercise.
- Diagnosis
- HRV values in response to diagnose many different conditions, especially those in which autonomic nervous system dysfunction has been found to play a role. The ANS is vital in the performance of the organs and HRV provides a tool to identify the problems in the system.
- Prognosis
- HRV can be used to predict future physiological and psychological problems.Numerous studies have shown that low HRV is associated with increased risk of several adverse outcomes including all-cause mortality. For example, HRV values can be involved in pre-operative investigations of ANS function, which has been shown to predict many conditions including post-operative cardiac-related morbidity and mortality.
- Explanation
- HRV increases our understanding of physiological, physiological, and cognitive and various other mechanisms.
- Outcome measurement
- HRV can be used as an indicator of overall cardiac health. Because HRV reflects autonomic function, it can be used to assess any treatment in which autonomic dysfunction might play a role.
I will explore applications of HRV in a separate article. Please watch out for it.
Notes
[1] The AV node has an additional important feature. It responds more slowly the more frequently it is stimulated. This property of the AV node protects the ventricles from rapid atrial rhythms.
License Attributions
Fig 1: “Diagram of the human heart (cropped)” by This file is lacking author information. – Own work. Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons – http://commons.wikimedia.org/wiki/File:Diagram_of_the_human_heart_(cropped).svg
#mediaviewer/File:Diagram_of_the_human_heart_(cropped).svg
Fig 2: By Blausen Medical Communications, Inc. (Donated via OTRS, see ticket for details) [CC-BY-3.0 (http://creativecommons.org/licenses/by/3.0)%5D, via Wikimedia Commons
Fig 3: http://d2ik8svv2swg1p.cloudfront.net/ekgtracings/70.gif
Fig 4: From http://en.wikipedia.org/wiki/Electrocardiography#mediaviewer/File:ECGcolor.svg
Fig 5: By Derivative: Hazmat2 Original: Hank van Helvete (This file was derived from: EKG Komplex.svg) [CC-BY-SA-3.0 (http://creativecommons.org/licenses/by-sa/3.0)%5D, via Wikimedia Commons
Fig 6: Vivonoetics Vivosense HRV Manual
Information Intelligence 
.svg){kind=link}
