Heart Rate Variability Analysis

It is believed that Heart Rate Variability (HRV) will become as common as pulse, blood pressure or temperature in patient charts in the near future. In the last ten years more than 2000 published articles have been written about HRV. HRV has been used as a screening tool in many disease processes. Various medical disciplines are looking at HRV. In diabetes and heart disease it has been proven to be predictive of the likelihood of future events. In 1996, a special task force was formed between the US and European Physiological associations to outline current finds on HRV and set specific standards on using HRV in medical science and future practice. Since then a steady stream of new information and value continues to come out of HRV research.

It all started in 1966 when a variation in the beat-to-beat intervals between heartbeats was noticed. Initially all recording devices were averaging heart rate data stream trying to get rid of any rapid HR fluctuations. Then there were very specific patterns in such fluctuations were noticed that had links to certain conditions way before any clinical symptoms appeared.

The origin of heartbeat is located in a sino-atrial (SA) node of the heart, where a group of specialized cells continuously generates an electrical impulse spreading all over the heart muscle through specialized pathways and creating process of heart muscle contraction well synchronized between both atriums and ventricles. The SA node generates such impulses about 100-120 times per minute at rest. However in healthy individual resting heart rate (HR) would never be that high. This is due to continuous control of the autonomic nervous system (ANS) over the output of SA node activity, which net regulatory effect gives real HR. In healthy subject at rest it is ranging between 50 and 70 beats per minute.

The autonomic nervous system is a part of the nervous system that non-voluntarily controls all organs and systems of the body. As the other part of nervous system ANS has its central (nuclei located in brain stem) and peripheral components (afferent and efferent fibers and peripheral ganglia) accessing all internal organs. There are two branches of the autonomic nervous system - sympathetic and parasympathetic (vagal) nervous systems that always work as antagonists in their effect on target organs.

For most organs including heart the sympathetic nervous system stimulates organ's functioning. An increase in sympathetic stimulation causes increase in HR, stroke volume, systemic vasoconstriction, etc. The heart response time to sympatheticstimulation is relatively slow. It takes about 5 seconds to increase HR after actual onset of sympathetic stimulation and almost 30 seconds to reach its peak steady level.

In contrast, the parasympathetic nervous system inhibits functioning of those organs. An increase in parasympathetic stimulation causes decrease in HR, stroke volume, systemic vasodilatation, etc. The heart response time to parasympathetic stimulation is almost instantaneous. Depending on actual phase of heart cycle it takes just 1 or 2 heartbeats before heart slows down to its minimum proportional to the level of stimulation.

At rest both sympathetic and parasympathetic systems are active with parasympathetic dominance. The actual balance between them is constantly changing in attempt to achieve optimum considering all internal and external stimuli.

There are various factors affecting autonomic regulation of the heart, including but not limited to respiration, thermoregulation, humoral regulation (rennin-angiotensin system), blood pressure, cardiac output, etc. One of the most important factors is blood pressure. There are special baroreceptive cells in the hear and large blood vessels that sense blood pressure level and send afferent stimulation to central structures of the ANS that control HR and blood vessel tonus primarily through sympathetic and somewhat parasympathetic systems forming continuous feedback dedicated to maintain systemic blood pressure. This mechanism is also called baroreflex, which increases HR when blood pressure decreases and vice versa. This mechanism is also targeted to maintain optimal cardiac output.

The heart rate variability analysis is a powerful tool in assessment of the autonomic function. It is accurate, reliable, reproducible, yet simple to measure and process. The source information for HRV is a continuous beat-by-beat measurement of interbeat intervals. The electrocardiograph (ECG or EKG) is considered as the best way to measure interbeat intervals. ECG is an electrical signal measured with special conductive electrodes placed on chest around heart area or limbs. It reflects minute changes in electrical field generated by heart muscle cells originating from its SA node. ECG signal has a very specific and robust waveform simple to detect and analyze. Because of that cardiac rhythm derived from ECG is the best way to detect not only true sinus rhythm but all types of ectopic heartbeats, which must be excluded from consideration of HRV analysis.

The other approach to measure cardiac intervals is a measurement of pulse wave. It is less invasive and simple method of measurement based on photoplethysmograph. PPG is a signal reflecting changes in a blood flow detected when infrared light is emitted towards microcirculatory blood vessels. Depending on blood flow volume certain portion of that light is absorbed letting other part to pass or b reflected. An optical sensor detects a quantity of light passed (or reflected from) the blood flow producing a waveform identifying pulse wave. Such waveform can also be processed to derive beat-by-beat interbeat intervals. Although PPG gives the summary information reflecting both cardiac and blood vessel components of HRV, some research studies showed a significantly high correlation between interbeat interval data measured by both ECG and PPG in short-term steady-state recordings.

One of the important issues when measuring either ECG or PPG is the absence of abnormal heartbeat used in interval detection. Only heartbeats originated in SA node can be processed to obtain HRV data. Whether there are ectopic heartbeats (PVCs or other types of extrasystolic heartbeats) or various movement artifacts on ECG (or PPG) considered as heartbeats, they must be excluded from consideration. There are various statistically-based algorithms of detection of such abnormal heartbeats that minimize chances to get contaminated HR recordings. Nevertheless, for the sake of accuracy in HRV analysis it is important to be able to visually verify all heartbeats automatically found, remove abnormal ones and include missing.

Short-term HRV analysis requires much shorter recordings - typically 5-min long. However such recordings are assumed to be done at steady-state physiological condition and should be properly standardized to produce comparable data. Typically such measurements should be done in either supine or comfortably sitting relaxed position, limiting body movements, conversations, any mental activities.

According to the standards set forth by the Task Force of the European Society of Cardiology and North American Society of Pacing and Electrophysiology in 1996, there are two methods of analysis of HRV data: time- and frequency-domain analysis. In either method, the interbeat intervals should be properly calculated and any abnormal heartbeats found.

Time-domain measures are the simplest parameters to be calculated. Before such calculation all abnormal heartbeats and artifacts must be removed from consideration. The following time-domain parameters can be calculated for both long-term and short-term recordings: Mean HR, SDNN and RMS-SD. Some extra parameters can be calculated specifically for long-term recordings. The time-domain parameters are associated mostly with overall variability of HR over the time of recording, except RMS-SD, which is associated with fast (parasympathetic) variability.

Frequency-domain measures pertain to HR variability at certain frequency ranges associated with specific physiological processes. Before frequency-domain analysis is performed, all abnormal heartbeats and artifacts must be detected and removed, then cardiotachogram (sequence of RR intervals) must be resampled to make it as if it is a regularly sampled signal. A standard spectral analysis routine is applied to such modified recording and the following parameters evaluated on 5-min time interval: Total Power (TP), High Frequency (HF), Low Frequency (LF) and Very Low Frequency (VLF). When long-term data is evaluated an additional frequency band is derived - Ultra Low Frequency.

The HF power spectrum is evaluated in the range from 0.15 to 0.4 Hz. This band reflects parasympathetic (vagal) tone and fluctuations caused by spontaneous respiration known as respiratory sinus arrhythmia.

The LF power spectrum is evaluated in the range from 0.04 to 0.15 Hz. This band can reflect both sympathetic and parasympathetic tone.

The VLF power spectrum is evaluated in the range from 0.0033 to 0.04 Hz. The physiological meaning of this band is most disputable. With longer recordings it is considered representing sympathetic tone as well as slower humoral and thermoregulatory effects. There are some findings that in shorter recordings VLF has fair representation of various negative emotions, worries, rumination etc.

The TP is a net effect of all possible physiological mechanisms contributing in HR variability that can be detected in 5-min recordings, however sympathetic tone is considered as a primary contributor.

The LF/HF Ratio is used to indicate balance between sympathetic and parasympathetic tone. A decrease in this score might indicate either increase in parasympathetic or decrease in sympathetic tone. It must be considered together with absolute values of both LF and HF to determine what factor contributes in autonomic disbalance.

The frequency domain analysis is traditionally performed by means of Fast Fourier Transformation (FFT). This method is simple in calculation but for fair representation of all frequency-domain HRV scores at least 5-min data should be collected. FFT assumes that time series represents a steady-state process. Because of that all data recordings should be conducted at highly stable standardized conditions, when no other factors other than current autonomic tone contributes in HRV. One of the most serious disadvantages of that is its insensitivity to rapid transitory processes, which often possess very valuable information about how physiology or certain pathological processes behave dynamically.

Some most recent studies implemented an alternative way to estimate power spectrum of HRV. It is based on autoregression methods. One of its major advantages is that it doesn't require to have analyzed data series to be in steady state. Thus any HRV data can be analyzed and fair HRV information still derived. Such analysis can be also performed on relatively shorter time intervals (less than 5 minutes) without missing meaningful HRV information. Finally this method is sensitive to rapid changes in HR properly showing tiny changes in autonomic balance. The drawback of this approach is a necessity to perform massive calculations to find best order of autoregression model.

From clinical perspective it is important not only to evaluate all HRV scores but be able to assess such HRV data, whether they are normal or not and how to interpret such data. It is known that HRV scores are age-dependent. Most of scores decrease with age. For better HRV data assessment special sets of reference ranges for each HRV parameter were created. Such ranges are based on statistics derived from HR data measured in a number of healthy individuals of different ages under standardized conditions. Such norms are considered as a reference point and cannot be used for any diagnostic purpose.

It is found that lowered HRV is associated with aging, decreased autonomic activity, hormonal tonus, specific types of autonomic neuropathies (e.g. diabetic neuropathy) and increased risk of sudden cardiac death after acute MI.

Other research indicated that depression, panic disorders, anxiety have negative impact on autonomic function, typically causing depletion of parasympathetic tonus. On the other hand an increased sympathetic tonus is associated with lowered threshold of ventricular fibrillation. These two factors could explain why such autonomic imbalance caused by significant mental and emotional stress increases risk of acute MI followed by sudden cardiac death.

Besides that there are multiple studies indicating that HRV is quite useful as a way to quantitatively measure physiological changes caused by various interventions both pharmacological and non-pharmacological during treatment of many pathological conditions having significant manifestation of lowered HRV.

Most known existing HRV measurement tools available at this time have capability to measure short-term HR data and provide their further static assessment. Each assessment provides information on what happened at certain point of time without any consideration of the dynamics of physiological changes reflected in HRV.

We have recently developed a new approach of HRV evaluation that is performed repetitively and continuously over longer periods of time. A standard 5-min time window is used for such HRV evaluation which is continuously updated every half a second. A robust method of real-time HR data evaluation and artifact detection is implemented to ensure reliability of HRV data. Use of new autoregression method of power spectrum calculation allows for fast and reliable early detection of transitory processes caused by various interventions. This tool is especially effective when various effects are examined. As an example, when performing tests for specific allergies or toxins various substances have to be tested with regard to what kind of autonomic response they create. Having ability to detect sudden changes in autonomic balance gives an opportunity to reliably detect such agents that cause such allergic reactions by means of affecting autonomic balance. When monitoring HRV changes multiple time markers can be posted along with data recorded. When HRV data is reviewed multiple comparisons of the data associated with those time markers can be performed.