Everything starts with the time between consecutive heartbeats – the RR intervals – measured in milliseconds. From a series of RR intervals, HRV can be calculated in two main ways.
Time domain
Time-domain analysis uses conventional statistical methods to describe the variation in the time between consecutive normal heartbeats:
| Measure | What it is | What it mainly reflects |
|---|---|---|
| RMSSD (ms) | How much each interval differs from the previous one | The brake (parasympathetic). Most used in watches and apps |
| ln RMSSD | The natural logarithm of RMSSD | Same as RMSSD, but easier to follow over time |
| SDNN (ms) | The standard deviation of all intervals | Total variability. Mostly used on 24-hour recordings |
| pNN50 (%) | The share of intervals differing by more than 50 ms from the previous one | The brake (parasympathetic) |
Frequency domain
Spectral analysis has been used as a tool in non-medical sciences such as economics, physics and meteorology for decades, but it was first introduced into medicine in the early 1980s. It is based on the equations created by the French mathematician Joseph Fourier.
Spectral analysis estimates the variation of the time between consecutive normal heartbeats within time periods of varying length. The degree of variation within a defined frequency range is termed power, measured in ms²:
- High frequency (HF, 0.15–0.40 Hz): periods lasting 2.5–7 seconds, roughly in step with breathing. Only the parasympathetic nervous system is capable of changing the heart rate within such short periods. HF power therefore reflects parasympathetic activity.
- Low frequency (LF, 0.04–0.15 Hz): periods lasting 7–25 seconds. Both the parasympathetic and the sympathetic nervous system can mediate heart rate changes within these longer periods.
- Total power (0.0033–0.40 Hz) is the total variation across all period lengths. Very low and ultra low frequency power (below 0.04 Hz) account for about 95 % of total power, but their physiological correlates are unknown.
Joseph Fourier (1768–1830) was a French mathematician and expert in Egyptology. In 1798 he accompanied Napoleon on his expedition to Egypt, and until 1801 he was engaged in extensive research on Egyptian antiquities. In thankfulness, Napoleon made him a baron in 1809. In 1822 he published «Théorie analytique de la chaleur» (The Analytical Theory of Heat). The equations presented there, known as Fourier series, have since been applied in a variety of sciences like physics and economics, and lately medicine.
The LF/HF ratio
Because the LF band is influenced by both branches, the LF/HF ratio is not a clean measure of the balance between them, although it is often presented that way. By exclusively using relative parameters – the LF/HF ratio and normalised units – important information may be lost.
For example, the parasympathetic markers decrease in two dramatically different situations. During parasympathetic blockade with atropine, the parasympathetic modulation of the heart rate declines. But if vagal activity is constantly stimulated, for instance during drug-induced blood pressure increases, the parasympathetic markers also decrease: the continuously increased vagal tone lowers the heart rate, but without heart rate fluctuations. In both situations the LF/HF ratio would change in the same direction, even though parasympathetic activity is reduced in the first and increased in the second. Absolute values should therefore always be presented alongside relative ones.
If you breathe slowly – fewer than nine breaths per minute – the breathing oscillations themselves also fall into the LF band (see HRV, breathing and meditation).
For short measurements at home, such as a morning reading of a few minutes, HF, RMSSD or ln RMSSD are the most reliable measures.
Baroreflex sensitivity
While HRV measures the tonic control of heart rate, baroreflex sensitivity (BRS) measures the dynamic control. BRS can be defined as the change in RR interval following a change in systolic blood pressure. A BRS of 10 ms/mmHg indicates that a rise of 1 mmHg in systolic pressure will induce 10 ms of RR-interval lengthening.
Normal values in young healthy people are approximately 15–20 ms/mmHg. BRS decreases with increasing age and is lower in people with high blood pressure. Low BRS after a heart attack is a powerful independent marker of increased risk of malignant arrhythmia and sudden death. BRS requires beat-to-beat blood pressure, so it cannot be calculated from a chest strap or watch alone.
ECG or optical sensors?
HRV – especially measures of fast oscillations – depends on precise recordings. Two different techniques are used:
- ECG records the heart’s electrical signals. Every contraction of the ventricles is preceded by an electrical impulse (the QRS complex). Its peak is a sharp spike called the R wave. This allows the time between consecutive beats (the RR interval) to be determined with an error of just a few milliseconds. Chest straps such as the Polar H10 and Movesense record ECG this way.
- Photoplethysmography (PPG), optical measurement, shines light through the skin; the light is reflected by red blood cells and picked up by sensors to calculate heart rate. These are the green lights on the underside of smartwatches and smart rings. The pulse wave is gentle and rounded and arrives slightly after the heartbeat, so the measurements are less precise.
| ECG chest strap | Optical (watch, ring) | |
|---|---|---|
| Signal | Electrical, sharp R wave | Pulse wave in the blood |
| Beat-to-beat precision | Very good | Good at rest, worse with movement |
| During exercise | Good | Often imprecise |
| Ectopic beats | Can be detected and removed | Hard to distinguish from noise |
| Comfort | Strap around the chest | Barely noticeable, can be worn all day |
| Best for | Precise readings, training, morning readings | Trends over time, sleep |
In my doctoral work around 2000 we recorded heart rate both by ECG and with an optical finger device (Finapres). Already then, ECG was clearly superior: it makes it easier to identify and exclude artefacts and arrhythmias, and the more accurate recognition of the R wave makes the RR intervals more precise. The finger recordings showed a larger spread.
A 2023 study compared HRV from a smartwatch (photoplethysmography) with high-resolution ECG in patients with cardiovascular disease and healthy controls. Agreement was good for measures of total and low-frequency variability, but poorer for beat-to-beat measures, especially in the patients. Those are exactly the measures – RMSSD and HF – that say most about parasympathetic activity.
In short: optical sensors are good enough to follow trends at rest and during sleep. If you need precise values, or measurements during activity, ECG is best.
References
- Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation 1996; 93: 1043–65.
- Sevre K. Autonomic cardiovascular regulation in healthy subjects and patients with cardiovascular disease. Doctoral thesis. University of Oslo, 2001.
- Sevre K, Rostrup M. Undersøkelser av hjertefrekvensvariabilitet og baroreflekssensitivitet [Heart rate variability and baroreflex sensitivity]. Tidsskr Nor Lægeforen 2001; 121: 3059–64.
- Billman GE. The LF/HF ratio does not accurately measure cardiac sympatho-vagal balance. Front Physiol 2013; 4: 26.
- Theurl F, Schreinlechner M, Sappler N, et al. Smartwatch-derived heart rate variability: a head-to-head comparison with the gold standard in cardiovascular disease. Eur Heart J Digit Health 2023; 4: 155–64.