Heart rate variability during exercise in cycling

HRV (Heart Rate Variability) can be used in cycling to check readiness to perform heavy training, or even to predict training adaptations. Nevertheless, it is usually measured at rest. Can we use exercise values in some way? It turns out that yes. HRV during exercise close to the first lactate / ventilatory threshold practically disappears and remains very low. This can be used as a way of estimating the first (sometimes called aerobic) threshold. Although this method is not as accurate as lab tests, it may be a non-invasive alternative for them.

HRV, or Heart Rate Variability, is the variability between successive heartbeats (Task Force, 1996). Intuitively, it seems that our heart beats at regular intervals. However, in reality, this is practically never the case. Its successive beats differ by fractions of a second, and the time intervals between them are on the one hand a source of variability, but on the other, they contain a lot of valuable information.

Paradoxically, a higher HRV means that our heart beats less steadily, but its high values can even be a sign of health (Singh et al., 2018b).

The current level of heart rate is largely controlled (although this isn’t the only factor) by two parts of the autonomic nervous system (Aubert et al. 2003, Billman, 2011):

• parasympathetic;
• sympathetic.

Parasympathetic activity decreases heart rate. It is more active in situations that are often referred to as “rest and digest”, as this phenomenon often occurs in situations where energy is conserved and restored.

The sympathetic part, on the other hand, is more active, for example in situations involving physical or mental stress, and its activity allows to cope with situations considered by the body as threatening, hence it is often said that it is active during “fight or flight” situations. Its activity, in turn, causes an increase in heart rate (Singh, 2018a).

Unfortunately, the relationships between the two systems are not linear, but rather really complex (Shaffer & Ginsberg, 2017). For example, an increase in heart rate at a given moment may well be caused by a decrease in parasympathetic activity with the same sympathetic activity or a decrease in sympathetic activity, but with the same parasympathetic one.

The changes in the activity of these two branches of the autonomic nervous system work independently of each other. We do not always deal with a situation where the activity of one system decreases, this is automatically followed by a decrease of the other part of ANS or the other way around.

HRV is a non-invasive method of measuring the activity of both parts of the autonomic nervous system, but it should be remembered that it is not a direct measurement (Billman, 2011). 

But in practice, HRV measures can only accurately reflect parasympathetic activity (e.g. HF or RMSSD) (de Geus, 2019), while indicators that were considered to reflect sympathetic activity (e.g. LF), are not effective (Michael et al., 2017).

Therefore based on HRV we can only estimate the activity of the parasympathetic system and determine whether it is more or less active. But as I mentioned earlier it does not mean, that if parasympathetic HRV indices drop, then automatically the sympathetic activity will increase.

Is HRV lower during exercise?

After the start of exercise, our heart rate increases significantly, while HRV is much lower than at rest. This is because parasympathetic activity is gradually diminishing.

When exercise intensity increases, we will observe a further increase in heart rate and a decrease in HRV due to even lower parasympathetic activity, combined with increased sympathetic one (Hautala, 2009).

As it turns out, in the classic test to exhaustion, in which the generated power increases by a constant value (e.g. 30 W), at regular intervals (e.g. 3 minutes), HRV gradually decreases, so that at about 50 -60% VO2max practically disappear and remain at a constant, very low level (Tulppo et al., 1998).

The point at which the HRV (or rather its indices) reaches this constant low value occurs at a similar intensity of exercise as the first lactate or ventilatory threshold.

Although it is difficult to say whether both phenomena are caused by the same processes, stabilization of HRV seems to occur at the same intensity as the first threshold, and therefore may be a good approximation of it (Garcia-Tabar et al., 2013).

This is quite important information, as there are not many methods that would allow us to estimate power associated with the first threshold in a non-invasive manner. The most accurate way is, of course, to perform a fitness test in a lab, but its frequent repetition is inconvenient and highly expensive.

HRV aerobic threshold

There are many methods by which we can estimate the first (aerobic) threshold from HRV during exercise. Some of them are based on time-domain indices (RMSSD, SD1) (Dourado and Guerra, 2013). In this case, determining the threshold is not very complicated.

Other methods, are based on more complex calculations related to frequency-domain HRV (HF) indices, but allow both thresholds to be calculated (Cottin et al., 2006). 

Another approach is to determine the threshold (again only the first), based on the nonlinear method of HRV analysis (DFA – Detrended Fluctuation Analysis), which is derived from the theory of mathematical chaos (Rogers et al., 2021). 

To estimate the first threshold, you could use an index DFA – alpha 1. Perhaps here the calculation process itself is not the simplest one, but we can deal with it using free software (which I will mention later). The index itself takes values from about 0.5 to 1.5.

This method makes it possible to find dependencies in the seemingly random behavior of the heart. If its values are around 1, it means that we are dealing with a mixture of random and orderly behavior.

When its value approaches 1.5, it means an orderly operation of the system, while on the other hand, its values are closer to 0.5, we deal with completely random behavior.

When determining the first metabolic threshold, some authors have assumed the DFA – alpha 1 value of 0.75, which is usually the level of its occurrence. Although they have achieved really good results (Rogers, 2021), this approach is quite innovative, and there is not so much data published so far. 

It is hard to say at the moment whether the value of the index of 0.75 for all people will be equivalent to the first threshold. In my case, this method was quite off from other HRV thresholds, but I must admit that I didn’t perform a lab test for comparison.

Some also tried to relate the assumptions of the Dmax method (based on lactate measurements) for HRV analysis, but the results weren’t impressive (Nascimento, 2019).

However, in the next part, I will only focus on the first method, even though it only allows the first threshold to be determined.

The method based on the analysis of frequency-domain indices does allow for estimation of both thresholds, but there is some caveat to it.

Methods allowing determination of both thresholds were usually compared with a second ventilatory threshold, otherwise known as RCP – Respiratory Compensation Point, or simply VT2. 

Indeed, HRV methods gave a good approximation of its value, but it turns out that, RCP / VT2 systematically overestimates the real value of the second metabolic threshold (Galán-Rioja, 2020).

The second metabolic threshold is the limit below which our body maintains a stable metabolic state, but when exceeded, lactate reaches very high values, and VO2 slowly rises to reach its maximum value (Jamnick, 2019).

In practice, there are many methods of determining it (FTP is one of them), some are less and the others more accurate. 

In the past, Maximal Lactate Steady State (MLSS) was considered the gold standard for estimating the second metabolic threshold, but that point of view has changed over the years.

Now it is believed that Critical Power is probably the best estimate of the second metabolic threshold (Jones et al., 2019), but of course, it isn’t without limitations.

RCP / VT2 usually overestimates the MLSS and Critical Power (even though these parameters are often highly correlated) (Galán-Rioja, 2020), and thus the method that reflects the second the ventilatory threshold, may not be a good basis for defining the intensity zones.

Due to the above-mentioned facts and the relative simplicity of the determination, later in the text, I will present the process of estimating the first metabolic threshold using RMSSD or SD1.

Determining the HRV threshold on my example

I will use my own example to illustrate the entire process of estimating the power (or heart rate) value associated with the first metabolic threshold. The first step in the process is performing a step test, but we also need a telemetry belt that will record the intervals between successive heartbeats (so-called R-R intervals, the example of that belt could be Polar H10).

In the step test, we simply increase power by a constant value (e.g. 30 W), at equal time intervals (e.g. 3 minutes) until exhaustion. Nevertheless for HRV threshold estimation step tests don’t need to be maximal (which I will describe later in the text).

Sales and his team used a protocol where the power was increased by 15 W every 3 minutes. The initial power, in this case, was 15 W, but the participants were quite untrained, which was a reason for such a choice.

I chose the variant in which every 3 minutes, the power was increased by 30 W, starting from 70 W, but in my case, I performed this test until exhaustion.

Regarding the initial power value, I personally arbitrarily assumed the value of 70 W, as I was sure that it would be low enough for me. Nevertheless, I think adopting a really low power of 30-40% FTP should work well (as the intensity should be quite low). We need to have a couple of steps below the power that we think is associated with the first threshold, to allow its estimation. If we start from a power that is higher than our threshold, we simply won’t be able to detect it, but on the other hand, it seems that it is not a big deal when we have more steps that are clearly below the first threshold.

It is also worth mentioning that the test should not be preceded by any warm-up.

After the test is completed, we can start the analysis. We can do it using the Kubios software, which is completely free for personal use. However, please refer to the instruction manual and/or videos provided by the manufacturer to learn the basic functions and to avoid errors in use.

A very important element of the HRV analysis is the correction of errors that appear during the measurement. They are caused for example by premature or omitted heartbeats that can completely skew the values of the obtained parameters.

Of course, the above-mentioned software can detect and correct them, but we must remember that sometimes it may happen that there will be a lot of “noise” in our data, which makes it impossible to accurately determine the threshold. 

In addition, a related issue is the selection of the threshold for correcting artifacts, which most authors set on “medium”.

We will be interested in two indicators:

• RMSSD;
• SD1.

RMSSD (root mean square of successive differences) is an index of parasympathetic activity, and we calculate it by taking the square of the sum of differences between successive RR intervals (i.e. time intervals between successive heartbeats) divided by the number of RR intervals minus 1 (Shaffer and Ginsberg, 2017).

SD1 is also an index of parasympathetic activity which is actually very close to the RMSSD (Michael et al., 2017). It is obtained from the Poincaré plot, where a given RR interval is plotted as a function of the previous one.

We can use one of them to estimate power or heart rate at the first threshold, but later in this article, I’ll describe the approach using two methods.

We need to determine the values of the selected HRV metric (RMSSD or SD1) at the last minute of each 3 minutes of the step segment of the test. Then the obtained values should be plotted with the power or time obtained during the test.

Determining the HRV threshold is based on a visual analysis of the graph obtained in this way. We need to identify the point at which HRV values stabilize at a low, constant level despite the increase in exercise intensity. It’s the point when the curve starts to flatten.

In my case, the analyzed indicator was RMSSD, and after comparing it with power, I got the following graph:

The HRV threshold (a reflection of the first metabolic threshold) is found at the time of the flattening of the HRV values obtained. In my case, it would be 190 W. However, I think that in this particular case it would be better to take the value from the earlier segment.

This is because the RMSSD between the power of 160 and 190 W differed only slightly (2.35 vs. 1.99 ms), and also obtained a very low value here, which may indicate that I already exceeded the first metabolic threshold at 160 W (or I achieved intensity exactly at first threshold).

In a later section, I also describe why this procedure seems appropriate to me, even though most authors have not followed it.

If anyone is interested in it, I am sending a link to Strava from the test: https://www.strava.com/activities/4819301369

Can we determine the HRV threshold without measuring the power?

It turns out that yes, but in this case, we will need a simple turbo trainer (e.g. magnetic) and a telemetry belt that allows you to measure HRV. As far as I know, most smart trainers have a built-in power meter anyway, so this probably doesn’t apply to them.

Outdoors, speed isn’t a very good measure of intensity (due to changing external conditions – weather, topography, etc.). However, on the trainer, we can control the external conditions, so the speed may prove useful here. So we can mount a speedometer on the rear wheel.

As long as we always choose the same pressure in the rear tire (I would suggest inflating it before each training) and we always use the same load on the shifter of the trainer (in the case of magnetic solutions), we will obtain quite stable external conditions.

Speed as such will be a similar measure of external intensity as power. We could call heart rate internal intensity, as it is a measure of our body’s response to a given effort. In turn, the external intensity tells us how much work we put into the effort. So in this case external intensity would be a speed.

Using speed, we could also perform a step test. Every 3 minutes we would increase it by, for example, 2 km / h. However, we would then need to determine a start speed that corresponds to a sufficiently low intensity of the effort to be able to derive an HRV threshold.

In turn, for the HRV threshold estimation, we take average heart rate values in the last minute of the step of the test, at which we identified the HRV threshold.

The biggest problem with this approach is the fact that the actual load of a magnetic trainer changes over time as the tire gets hotter. We often get more resistance when the tire is still quite cold, so the same speed would mean different power at the beginning, and the end of the exercise.

HRV threshold in relation to lactate / ventilatory threshold

We already know how to determine the HRV threshold using RMSSD / SD1. However, the question remains, will the obtained value coincide with the laboratory methods?

In the previously cited work by Sales et al. (2011), 10 people leading a sedentary lifestyle, and 9 people struggling with diabetes, performed the maximum test to exhaustion.

During the experiment, the first metabolic threshold was determined both by the analysis of exhaled gases and ventilation (i.e., it was the ventilatory threshold), using the concentration of lactate in the blood (thus it was the lactate threshold), and by analyzing the RMSSD and SD1 during exercise.

As already mentioned, the test protocol consisted of 15 W “steps” incremented every 3 minutes (starting at 15 W) until exhaustion.

It turned out that both methods were a good approximation of both the lactate and ventilatory threshold, and the correlations between the power associated with HRV thresholds and ones determined by laboratory methods were really strong for people who were sedentary (r = 0.91 to 0.98) while in people struggling with the disease, they were still at a good, but lower level (r = 0.68 to 0.87).

The authors concluded that HRV thresholds are a good non-invasive way to estimate values at the first metabolic threshold.

HRV threshold in case of professional cyclists

The results of the previous work were promising, but we can not necessarily directly relate them to people who actively participate in sports. For this reason, it would be worth checking whether this method will also work well among active cyclists.

Unfortunately, I haven’t come across many sources that have verified the effectiveness of the HRV threshold among active cyclists. Yes, there were a lot of publications on running or even skiing, but I managed to find only one that directly related to cycling (although I might simply not find the others).

Garcia-Tabar and his team (2013) verified the effectiveness of the HRV threshold for professional cyclists. They compared its value to the first lactate threshold. 

For HRV threshold determination they didn’t use a visual, but several variants of different mathematical approaches of determination, which are more objective, but objectivity does not always have to translate into better effectiveness (Candido et al., 2015 [after:] Soares -Caldeira et al., 2020).

Calculations were based on the SD1 indicator and used three different approaches:

• value by 1 ms higher than the lowest value observed during the step test;
• an SD1 value below which this index varied by less than 0.5 ms between the two steps;
• 0.5 ms higher than the first below 2.5 ms.

Nevertheless, later it turned out that only the first above-mentioned approach turned out to be a good method. 

Despite a decent (but nothing crazy) correlation between the methods (r = 0.68 to 0.88), in individual cases, it sometimes turned out that this approach either significantly inflated or lowered the lactate threshold value.

However, one should be aware of what could have resulted in such observations. In part, this could have been the analysis of the threshold itself, but also the protocol used. 

During the test until exhaustion, the power was increased by as much as 58 W / 3 min, which could result in lower accuracy both in the case of determining the first lactate threshold and the HRV threshold.

The other reason could be using mathematical approaches for determining HRV threshold. As I said earlier maybe they are more objective, but in this case maybe it doesn’t translate into their effectiveness.

Therefore, it seems to me that this does not necessarily have to be evidence of the weak effectiveness of the method itself, but not necessarily of the whole concept. I think more research on well-trained cyclists should be performed to help draw clearer conclusions.

HRV threshold issues that we need to be aware of

A large number of potential calculation methods

In addition to the aforementioned division of the methods of determining the HRV threshold due to time, frequency, and non-linear metrics, also within these subgroups, we come across various approaches to calculating HRV threshold.

For example, taking into account the previously described methods, using time-domain indices (eg. RMSSD and SD1), their results can be interpreted differently. We can make a visual analysis of indicators or use a mathematical method.

An example of a mathematical approach was the work on professional cyclists, while other authors use other assumptions than those presented there, e.g. a constant value of SD1 equal to 3 ms, and the threshold is defined as the first step of the test that falls below this value (Duarte et al., 2014).

In addition, an alternative method is to use another time-domain metric – SDNN, which is simply the standard deviation of all time intervals between heartbeats (Soares-Caldeira et al., 2020).

Which approach is the best? I have no idea. I think it would be interesting to check most of the available methods and analyze the effectiveness of each of them, just like N. Jamnick (2018) did, but in the case of the second lactate threshold.

How to solve the problem?

If we do not really know which method is the best, we can adopt a strategy similar to that presented by S. Gaskill and his team (2001). It refers to the determination of the first ventilatory threshold, but it seems to me that it can also be used in this case.

These authors combined three different methods of determining the threshold, and it turned out that together they gave much better results than when used separately. Also here we can perform analysis using both SD1 and RMSSD.

I think that it is better to estimate the HRV threshold (as I did) by taking the power value related to one segment before HRV stabilization. Returning to my example, the threshold established in this way is 160 W. Referring to the classic recommendations, it would be 190 W.

Nevertheless, perhaps the real flattening of the graph was at 175 W, not 190 W, and we can’t tell from the step test. We can say that my HRV threshold is somewhere between 160 and 190 W, but I think it is better to take an earlier, more conservative value.

Using both SD1 and RMSSD, we create two plots and determine two thresholds. If they occur at the same power, we just take this value, otherwise, we take the average of two values and we assume that first threshold power.

Measurement error

The HRV threshold is a good method of determining the first metabolic threshold, but not a direct one, so we must take into account a measurement error in this case.

Often used in experiments comparing two different methods of measurement, the statistical analysis of Bland and Altman (1986) is used. It allows, among others, to determine the maximum error we can expect when deriving one value from another.

Unfortunately, I have not come across any papers that would present this value in the results. There have been such studies, but for various reasons, it was not possible to adapt this information, either due to the use of a different methodology or due to the studied population (eg. Cunha, 2014).

In fact, in the previously cited study on professional cyclists, this information was available, but due to the problems with the protocol and analytical methods used, we may not necessarily use it for our needs (Garcia-Tabar, 2013).

From this information, we could draw quite an important hint, as knowing the maximum probable error, we would be able to define the “uncertainty range” or “grey zone” around the obtained power value at the HRV threshold.

If we move too close to this range, the probability that we are actually working over the real value of the threshold will increase.

In general, it can be well illustrated by the example of low-intensity training. This type of training often is performed below the power or heart rate at the first threshold.

Perhaps (although this is an example value, I have no idea if this is the case), the maximum error associated with the HRV threshold (determined according to the previously described methods), relative to the first lactate / ventilatory threshold is ± 20 W. 

If we assume that we want to do such training under the first threshold, but we would get a power 15 W lower than its value, we are completely unsure whether we were training under or over the real value of the lactate/ventilatory threshold.

However, if we kept a sufficiently large distance, e.g. 30 W from the estimated value, there would be a small chance that we are working over a real first threshold (remember that the HRV threshold is only an approximation), and thus we could say that the training was actually performed following our assumptions.

Solution

If it is not possible to determine the aforementioned “uncertainty range”, we can simply assume that during training performed below the first metabolic threshold, we would use a power lower by the value of one step before the HRV threshold occurred (eq. -30 W below determined power at HRV threshold). By doing this we would probably get the intensity that is below the first threshold determined in a lab (but of course, there’s no guarantee for that).

On the other hand, if we want to do medium intensity training between both thresholds, we need to take power at least one step greater, than HRV threshold (eg. +30 W). That way we increase dramatically the probability that we really train with the intensity over the first threshold

Referring to the example of my test, I can assume that I should do the low-intensity training sessions at 130 W and below, while those with medium intensity, at least 190 W, or above.

The above-mentioned approach isn’t perfect, but it should do the job.

Influence of cadence on HRV

It turns out that during exercise, cadence may have a large impact on the observed HRV (Blain et al., 2009). This is because the cyclic movements of the lower limbs while cycling affect the work of the heart during exercise.

For this reason, the use of free chosen cadence during the step test may theoretically have an impact on the results obtained. Indeed, in most studies on the HRV threshold, a fixed cadence (e.g., 70 rpm) was used.

Unfortunately, I haven’t found any work that would check whether it can really have an impact on the determination of the threshold. 

On the other hand, this situation creates artificial conditions, because in training we usually choose the cadence spontaneously, so I am not sure if the translation of the results of such a test will reflect the training situation later.

I will say right away that I do not know how to solve this problem. I am not able to say whether, despite the effect of cadence on HRV during exercise, it will have any implications for the effectiveness of determining the HRV threshold.

But personally, I did not control the cadence in my test.

A visual method of threshold determination 

There are mathematical methods of determining the threshold, but as it turns out, they are not necessarily better than the visual methods. For me, the threshold identification was rather easy, but in some cases, it may be a bit less clear.

Solution

It seems to me that part of the problem can be solved by using two determination methods simultaneously (SD1 + RMSSD). Then, when one case is completely unclear, we will be able to identify the threshold by the other method.

Usually in scientific research, the subjective nature of such a method is solved by determining them by two, and sometimes three different experienced people.

Unfortunately, when we determine the HRV threshold for the first time, we do not have much experience in doing that. Nevertheless, you can always contact me to consult the results.

However, I must admit that I do not consider myself an expert on this matter, but two heads are always better than one.

We can use this approach both with heart rate, and power HRV threshold.

Estimation of metabolic thresholds and determination of intensity zones using the HRV threshold and FTP / FTHR

Taking into account the body’s physiological response to exercise, we can define three (or possibly four) intensity zones (Jamnick, 2019). The basis for their determination are metabolic thresholds, as they separate the body’s different reactions to exercise.

The first metabolic zone is below the first metabolic threshold. With this type of exercise, blood lactate levels remain at resting levels, and VO2 (or oxygen utilization) is relatively stable. 

Regeneration of HRV, i.e. how fast its return to pre-exercise values at this intensity, will occur very quickly (Seiler et al., 2007).

It is believed that in this zone, we should do low-intensity training.

After crossing the first metabolic threshold, we enter the second zone. Here, lactate will exceed its resting values but still will reach a stable state. In the case of VO2, we will observe the so-called “slow component”, i.e. its slow increase (a phenomenon similar to heart rate drift), although theoretically, it should be at a constant level. The slow component won’t drive VO2 to its maximal values in this case.

Once the first metabolic threshold is exceeded, HRV recovery will be significantly delayed in comparison to exercise below it (Stanley et al., 2013).

Here we do medium-intensity workouts, and this intensity is often called tempo or sweet spot.

The third metabolic zone is effort above the second threshold. Here, the blood lactate concentration loses a stable state and rises to very high values, while the VO2 “slow component” drives it to VO2max.

In the third metabolic zone, we do the so-called HIIT training using shorter or longer intervals.

HRV recovery after exercise will also be significantly delayed in this case. It would seem that it will be even slower than in the case of training in the second zone.

Indeed, many studies show that HRV regeneration is largely dependent on the intensity of exercise, regardless of its length (Kaikkonen, 2015). So the higher the intensity, the longer the HVR will return to normal values.

Nevertheless, the relationships between the intensity and duration of training appear to be a bit more complicated than that (Reichert and Picanço, 2014; Bechke et al., 2020).

When doing interval training in the second metabolic zone, it may not always be the case that HRV will recover faster than when training in the third metabolic zone. Both workouts can have the same effect on HRV recovery, as it will depend not only on the intensity but also the length of the intervals or the number of repetitions.

It seems that in an extreme case (if we would ride long enough) training in the first metabolic zone, could theoretically lead to the same long recovery of HRV as training above the first or second metabolic threshold.

However, we can generally assume that after exercise in the first metabolic zone, HRV will recover very quickly, and when the first threshold is exceeded, this process will be delayed. Training in the second compared to the third metabolic zone, on the other hand, may have the same effect on HRV after exercise in both cases (Seiler et al., 2007).

There is still a final fourth zone. It will include all kinds of short-term sprint efforts. Here, neither lactate nor VO2 will reach maximum levels, possibly because the exercise is too short.

As was the case with training in the second and third metabolic zones, HRV recovery after this type of exercise will also be delayed (Buchheit et al., 2008).

Why is exercise intensity determination in relation to metabolic thresholds so important?

Generally, using the ready-made intensity zone systems (I mean the ones that are not individualized), we are not sure if they coincide with the physiological reactions of our body. 

For example, as early as 1988, in a study by E. Coyle et al., It was observed that training at the same% VO2max produced significantly different physiological responses in different participants of the experiment.

Some of them were working below the lactate threshold, and some working above it, so it wasn’t the same intensity for them.

Similar observations were also made in the case of training based on HRmax (Meyer et al., 1999), comparing it with one of the methods of determining the second lactate threshold (IAT – Individual Anaerobic Threshold).

However, determining the intensity from% VO2max or% HRmax has one thing in common – both refer to the maximum values. If the second metabolic threshold can be found at different levels of HRmax or VO2max, it is better to build a zone system on a threshold rather than a maximum value. 

Probably, for this reason, the popular zone system proposed by A. Coggan is based on FTP and FTHR.

However, this does not solve all problems. FTP is a fairly good approximation of the power associated with the second metabolic threshold, so we can estimate the boundary between the second and third metabolic zones. 

However, based on this zones system we are unable to tell what percentage of FTP a person will get at the first threshold.

This value can certainly differ from person to person (Garcia-Tabar & Gorostiaga, 2018). If we want for example do the low intensity training below the first threshold, assuming some fixed FTP percentages can lead to a situation when some people will be over a real threshold value and some will be below.

This means that for some it will be still low intensity, and for others, it will be medium intensity. 

Moreover, when exercising above or below the first threshold, the body’s reactions are significantly different, and thus such an effort is a different training stimulus for two different people.

If it is a different training stimulus for two different people, then also the reaction of their organisms (and probably adaptations), will be different, despite theoretically the same training intensity.

The problem with predefined intensity zones is mainly because we think that a given training session will be a specific signal for our body, but it does not necessarily have to be that way. The problem is that in terms of metabolism, it can be a completely different load on our body than we assumed.

As a result, our training could have completely unpredictable effects, because it will be something completely different than what we’ve planned.

In addition to the purely theoretical effectiveness of a zone system based on metabolic thresholds, there is also evidence that it is indeed more effective than a standard zone approach

This is illustrated by the study by A. Wolpern et al. (2015). Although it refers to people who previously led a sedentary lifestyle, the results are quite interesting. The study participants were divided into two groups:

• one training based on% HRR, i.e. heart rate reserve;
• the second training according to zones based on ventilatory thresholds.

They found that the group training based on thresholds as a whole had a better average result, but more importantly, all participants achieved a significant increase in their VO2max. 

In the second group, many people improved their VO2max to such an insignificant degree that the increase did not exceed the measurement error of VO2max.

These differences were observed despite the use of an identical training program. We can see that in this case, individualizing the intensity of the effort resulted in more predictable training adaptations.

Although all the people in the group training according to thresholds improved their results, some still to a lesser extent, and the others to a greater extent. 

For this reason, the appropriate selection of intensity did not prevent the heterogeneity of training responses, so it is not the only factor responsible for this phenomenon.

Nevertheless, as shown by the results of the study, the appropriate selection of exercise intensity is one of the important steps in improving the effectiveness of our training.

How to estimate the metabolic zones based on the HRV and CP or FTP

The first step that we should take to estimate the threshold values and define the boundaries of the intensity zones is to perform the FTP / FHR test. 

These values will approximate the intensity associated with the second metabolic threshold.

The classic protocol for determining FTP consists of (Allen, Coggan, & McGregor, 2019):

• 20 minutes of low-intensity effort;
• three 1-min high cadence accelerations (100 rpm);
• 5 min, low intensity driving
• 5 minutes of maximal effort;
• 10 minutes of low intensity driving;
• 20 minutes of maximal effort;
• 10-15 minutes cool down.

But actually, it is better to do a Critical Power testing, than to estimate the power at the second threshold by FTP. I described briefly how to do that in the post related to Garmin’s training plans, in the section called the “Incorrect selection of exercise intensity”.

After performing the FTP test or CP tests and calculating its value, we could move to the second stage, i.e. the test step to determine the HRV threshold. Knowing the CP/FTP value, we would not have to do it to exhaustion, because when we reach power at FTP or CP we definitely crossed the first threshold.

When determining the HRV threshold by heart rate, we also don’t have to perform the step test to exhaustion, but like in the case of the power approach, we can stop the test when we achieve FTHR.

Therefore, shortly after the FTP test or CP tests (e.g. next week), we could perform the step test. The initial power value could be defined as 30-40% FTP or CP (but it is better to round it to full values, e.g. if it comes out 75 W, then take the initial power of 80 W), because it should be really low intensity.

Then, every 3 minutes, we should increase the power by 30 W (but before that, we do not warm up, we go straight to the step test). Nevertheless, as mentioned before, we do not perform such a test until exhaustion, but at the moment when we reach or exceed our FTP or CP value during a given segment, we can stop the test.

Then we calculate the SD1 and RMSSD values, in the last minute of each of the 3 min segments, and then we plot obtained values.

Using a combined approach (SD1 + RMSSD), we identify one step before the HRV plot flattens, and either take one value (when both methods point to the same segment) or take the mean.

With the power values related to the HRV and FTP or CP, we can estimate our metabolic zones. The first of them should be below the first metabolic threshold, and because the HRV threshold is only its estimated value, we take a power value of one step before HRV threshold (i.e. -30 W, but here from the obtained threshold value, not the flattening itself ). 

If we want to train in the second metabolic zone, we need to take power at least one step greater from HRV threshold (eg. +30 W). But on the other hand, we don’t want to cross the FTP or CP.

As with the first threshold, we shouldn’t train too close to CP or FTP, because that way we don’t really know if we crossed a real line between stable and unstable metabolic state (because of measurement error). But for me, it’s also hard to say how big that “grey zone” should be.

From examples that I know, if we cross Critical Power only by a couple of watts, we should get a nonstable response in lactate and VO2, but when we want to train in the second metabolic zone it is better to take a larger margin (I think at least 15-20 W, bellow CP), because from what I’ve seen CP can overestimate the point where the physiological reactions changes, but it rarely underestimates it.

In practice, I use for example 90% of CP for 20 min intervals, and 85% for 30+ min intervals.

On the other hand FTP from 20 min test in most cases is lower than CP. Because of that if think that you can train a little bit closer to its value than in the case of CP (in the case of the second zone).

On the other hand, if 20 min FTP could be a little bit lower than CP, if we want to train in the third zone (above the real second threshold), it is a good idea to do such training at least at 105% FTP.

We have also a fourth zone, but actually, I don’t think that we need to pace such short sprint-type efforts by power. We can just go as hard as we can.

Above mentioned recommendations are just rough guidelines, but I think that they definitely can help.

Metabolic zones for FTHR

In the case of using heart rate, we can determine the boundaries between zones, by using heart rate at the HRV threshold, and FTHR.

FTHR, like FTP, is an approximation of the intensity associated with the second metabolic threshold. But is it a good approximation? Actually, it’s hard to say.

McGehee and his team (2005) compared several non-invasive methods for determining heart rate and threshold speed among runners. One of them was the 30-minute FTHR test. It was the maximal 30-minute exercise test, and the FTHR was the average heart rate in the last 20 minutes of exercise.

Indeed, compared to the 4 mmol threshold approach, the studied methods appeared to be effective. The problem is that the 4 mmol approach is completely ineffective itself.

The second method for determining the FTHR is to take 95% of the heart rate from a 20-minute maximum exercise test (Friel, 2018). Joe Friel in the latest edition of the cited book also states that if we perform a 30-minute test, we simply consider the entire effort, not the last 30 minutes.

Which method is better, I have no idea.

Even when we don’t really if FTHR is a good method for estimating the heart rate at the second threshold, we can still use it.

To set our individualized training zones, we take heart rate at HRV threshold as a boundary between the first and second zone, and FTHR as a boundary between the second and third zone.

Of course, it is a good idea to establish some kind of “grey zone” around thresholds, to account for the measurement error.

Because of that training in the first metabolic zone could be performed with the heart rate corresponding to the last minute of one step, before the one we qualified as the HRV threshold (again determined by the value on the threshold, not necessarily a flattening on the chart).

For the second metabolic zone, we can take the middle value of the heart rate between the HRV threshold and the FTHR.

In the case of training in the third metabolic zone, I’m not really convinced, that using heart rate would be a good approach.

Can we use HRV during exercise in another way?

I have to admit that using HRV either during rest or regeneration after completion of exercise is much more exploited in the literature, than using it during exertion. For this reason, I did not know many of its uses other than the HRV threshold.

Nevertheless, HRV when exercising at the same absolute intensity can determine a person’s level of fitness. For example, at a constant power of 200 W, people who have a higher HRV during exercise also are characterized by better performance (Hottenrott and Hoos, 2017).

However, this fact should not surprise us, because, with a better performance level of a given person, he or she would be exercising at lower relative intensity, so for that reason, HRV could be lower.

For example, if a person has a first metabolic threshold at 150W, and we compare him or her with a person with a 250 W threshold, it is obvious that at 200W, the better performer will have (a) lower HRV. If you still remember, after crossing the first metabolic threshold, HRV drops to virtually zero and remains at very low values, so in this particular case, a slightly less trained person will have a very low HRV.

Moreover, it has been shown in one study that the power associated with the HRV threshold may increase with training (Fronchetti et al., 2007), which is related to the previously described phenomenon.

It also means that potentially HRV at a predetermined power can be an indicator of improvement. If our training is going in the right direction, after some time we will notice that our HRV at a given, constant power value will decrease.

Similarly submaximal heart rate at constant power changes with training. It can be said that a decrease in submaximal heart rate for a given power or speed is one of the basic adaptations to endurance training (Rankinen et al., 2012).

It’s actually quite intuitive. The better we are, the lower the heart rate will be for a given generated power on the bike.

Hence, some authors have proposed submaximal tests, based on an analysis of the behavior of the submaximal heart rate at absolute exercise intensity (Lamberts et al., 2011, Buchheit et al., 2012).

Although the publication of Buchheit et al. (2012) was about young footballers and not endurance athletes, an interesting phenomenon was observed in it. The decrease in the heart rate of the players, when running at a preset speed (9 km / h), was the best predictor of the increase in performance in the test to exhaustion on the treadmill. The decrease in submaximal heart rate was therefore associated with the improvement of the aerobic capacity of the players.

Perhaps HRV measured during exercise can provide us with similar information as heart rate – the higher the HRV, the better the performance. Nevertheless, it would be interesting to find out which of these two indicators would be more effective as a performance indicators

Unfortunately, I haven’t come across any more potential applications, but I am sure there is a lot of potentially useful information hidden in the HRV values obtained during exercise, but at the moment it is difficult to say how it could be used.

Summary:

• with increasing exercise intensity, HRV reaches lower and lower values, and at the intensity of about 50-60% VO2max, it stabilizes at a very low level (almost disappears);
• it appears that this stabilization of the HRV occurs at a similar intensity to the first metabolic threshold;
• due to this phenomenon, using the HRV analysis during the step test, we can estimate power or heart rate related to the first metabolic threshold;
• thanks to this, in a non-invasive way, we can determine the approximate boundaries of metabolic zones, but we must be aware that this is not a direct method and is associated with some kind of error;
• appropriate selection of intensity, based on metabolic thresholds, is one of the steps to improve the effectiveness of our training program;
• HRV during exercise can be used as a performance metric and tracker.

Bibliography:

1. Electrophysiology, Task Force of the European Society of Cardiology the North American Society of Pacing. “Heart rate variability: standards of measurement, physiological interpretation, and clinical use.” Circulation 93.5 (1996): 1043-1065.
2. Singh, N., Moneghetti, KJ, Christle, JW, Hadley, D., Plews, D., & Froelicher, V. (2018a). Heart Rate Variability: An Old Metric with New Meaning in the Era of using mHealth Technologies for Health and Exercise Training Guidance. Part One: Physiology and Methods. Arrhythmia & electrophysiology review , 7 (3), 193–198.
3. Singh, N., Moneghetti, KJ, Christle, JW, Hadley, D., Froelicher, V., & Plews, D. (2018b). Heart rate variability: an old metric with new meaning in the era of using mhealth technologies for health and exercise training guidance. part two: prognosis and training. Arrhythmia & electrophysiology review , 7 (4), 247.
4. Aubert, AE, Seps, B., & Beckers, F. (2003). Heart rate variability in athletes. Sports medicine , 33 (12), 889-919.
5. Billman, GE (2011). Heart rate variability – a historical perspective. Frontiers in physiology , 2 , 86.
6. Shaffer, F., & Ginsberg, JP (2017). An overview of heart rate variability metrics and norms. Frontiers in public health, 5, 258.
7. de Geus, EJ, Gianaros, PJ, Brindle, RC, Jennings, JR, & Berntson, GG (2019). Should heart rate variability be “corrected” for heart rate? Biological, quantitative, and interpretive considerations. Psychophysiology , 56 (2), e13287.
8. Michael, S., Graham, KS, & Davis, GM (2017). Cardiac autonomic responses during exercise and post-exercise recovery using heart rate variability and systolic time intervals — a review. Frontiers in physiology , 8 , 301.
9. Hautala, AJ, Kiviniemi, AM, & Tulppo, MP (2009). Individual responses to aerobic exercise: the role of the autonomic nervous system. Neuroscience & Biobehavioral Reviews, 33 (2), 107-115.
10. Tulppo, MP, Makikallio, TH, Seppänen, T., Laukkanen, RT, & Huikuri, HV (1998). Vagal modulation of heart rate during exercise: effects of age and physical fitness. American Journal of Physiology-Heart and Circulatory Physiology , 274 (2), H424-H429.
11. Garcia-Tabar, I., Sánchez-Medina, L., Aramendi, JF, Ruesta, M., Ibañez, J., & Gorostiaga, EM (2013). Heart rate variability thresholds predict lactate thresholds in professional world-class road cyclists. J Exerc Physiol Online, 16, 38-50.
12. Dourado, VZ, & Guerra, RLF (2013). Reliability and validity of heart rate variability threshold assessment during an incremental shuttle-walk test in middle-aged and older adults. Brazilian Journal of Medical and Biological Research, 46 (2), 194-199.
13. Cottin, F., Leprêtre, PM, Lopes, P., Papelier, Y., Médigue, C., & Billat, V. (2006). Assessment of ventilatory thresholds from heart rate variability in well-trained subjects during cycling. International journal of sports medicine, 27 (12), 959-967.
14. Rogers, B., Giles, D., Draper, N., Hoos, O., & Gronwald, T. (2021). A new detection method defining the aerobic threshold for endurance exercise and training prescription based on fractal correlation properties of heart rate variability. Frontiers in Physiology , 11 , 1806.
15. Nascimento, EMF, Antunes, D., do Nascimento Salvador, PC, Borszcz, FK, & de Lucas, RD (2019). Applicability of Dmax Method on Heart Rate Variability to Estimate the Lactate Thresholds in Male Runners. Journal of Sports Medicine , 2019 .
16. Galán-Rioja, M. Á., González-Mohíno, F., Poole, DC, & González-Ravé, JM (2020). Relative proximity of critical power and metabolic / ventilatory thresholds: systematic review and meta-analysis. Sports Medicine, 1-13.
17. Jamnick, N. (2019). An examination of current methods to prescribe exercise intensity: validity of different approaches and effects on cell signalling events associated with mitochondrial biogenesis (Doctoral dissertation, Victoria University).
18. Jones, AM, Burnley, M., Black, MI, Poole, DC, & Vanhatalo, A. (2019). The maximal metabolic steady state: redefining the ‘gold standard’. Physiological reports, 7 (10), e14098.
19. Galán-Rioja, M. Á., González-Mohíno, F., Poole, D. C., & González-Ravé, J. M. (2020). Relative Proximity of Critical Power and Metabolic/Ventilatory Thresholds: Systematic Review and Meta-Analysis. Sports medicine (Auckland, N.Z.), 50(10), 1771–1783. https://doi.org/10.1007/s40279-020-01314-8
20. Billat, VL, Sirvent, P., Py, G., Koralsztein, JP, & Mercier, J. (2003). The concept of maximal lactate steady state. Sports medicine, 33 (6), 407-426.
21. Borszcz, FK, Tramontin, AF, & Costa, VP (2019). Is the functional threshold power interchangeable with the maximal lactate steady state in trained cyclists ?. International journal of sports physiology and performance, 14 (8), 1029-1035.
22. Sales, MM, Campbell, CSG, Morais, PK, Ernesto, C., Soares-Caldeira, LF, Russo, P.,… & Simões, HG (2011). Noninvasive method to estimate anaerobic threshold in individuals with type 2 diabetes. Diabetology & Metabolic Syndrome, 3 (1), 1-8.
23. Bland, JM, & Altman, D. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. The lancet, 327 (8476), 307-310.
24. Cunha, FA, Montenegro, RA, Midgley, AW, Vasconcellos, F., Soares, PP, & Farinatti, P. (2014). Influence of exercise modality on agreement between gas exchange and heart rate variability thresholds. Brazilian Journal of Medical and Biological Research , 47 (8), 706-714.
25. Hottenrott, K., & Hoos, O. (2017). Heart rate variability analysis in exercise physiology. ECG time series analysis: Engineering to medicine, 245-257.
26. Fronchetti, L., Nakamura, FY, De-Oliveira, FR, Lima-Silva, AE, & De Lima, JR (2007). EFFECTS OF HIGH-INTENSITY INTERVAL TRAINING ON HEART RATE VARIABILITY DURING EXERCISE. Journal of Exercise Physiology Online , 10 (4).
27. Candido, N., Okuno, NM, da Silva, CC, Machado, FA, & Nakamura, FY (2015). Reliability of the heart rate variability threshold using visual inspection and Dmax methods. International journal of sports medicine, 36 (13), 1076-1080. [After:] Soares-Caldeira, LF, da Silva, CC, Chierotti, P., de Souza Dias, N., & Nakamura, FY (2020). Influence of aerobic fitness on the correspondence between heart rate variability and ventilatory threshold. Revista Brasileira de Educação Física e Esporte , 34 (4), 555-566.
28. Duarte, TS, Ferreira Júnior, AJ, Rezende, EC, Pertence, LC, Bouzas Marins, JC, & de Lima, JRP (2014). Association between Heart Rate Variability Threshold and Post-Exercise Vagal Reactivation. Journal of Exercise Physiology Online .
29. Jamnick, NA, Botella, J., Pyne, DB, & Bishop, DJ (2018). Manipulating graded exercise test variables affects the validity of the lactate threshold and V˙ O 2 peak. PloS one, 13 (7), e0199794.
30. Gaskill, SE, Ruby, BC, Walker, AJ, Sanchez, OA, SERFASS, RC, & LEON, AS (2001). Validity and reliability of combining three methods to determine ventilatory threshold. Medicine & Science in Sports & Exercise, 33 (11), 1841-1848.
31. Blain, G., Meste, O., Blain, A., & Bermon, S. (2009). Time-frequency analysis of heart rate variability reveals cardiolocomotor coupling during dynamic cycling exercise in humans. American Journal of Physiology-Heart and Circulatory Physiology , 296 (5), H1651-H1659.
32. Seiler, S., Haugen, O., & Kuffel, E. (2007). Autonomic recovery after exercise in trained athletes: intensity and duration effects. Medicine & Science in Sports & Exercise, 39 (8), 1366-1373.
33. Stanley, J., Peake, JM, & Buchheit, M. (2013). Cardiac parasympathetic reactivation following exercise: implications for training prescription. Sports medicine , 43 (12), 1259-1277.
34. Kaikkonen, P. (2015). Post-exercise heart rate variability: a new approach to evaluation of exercise-induced physiological training load. Studies in sport, physical education and health , (224).
35. Reichert, F., & Picanço, L. (2014). Effects of different training protocols on the heart rate variability of trained cyclists. Journal of Science and Cycling, 3 (2), 79.
36. Bechke, E., Kliszczewicz, B., McLester, C., Tillman, M., Esco, M., & Lopez, R. (2020). An examination of single day vs. multi-day heart rate variability and its relationship to heart rate recovery following maximal aerobic exercise in females. Scientific Reports , 10 (1), 1-7.
37. Buchheit, M., Millet, GP, Parisy, A., Pourchez, S., Laursen, PB, & Ahmaidi, S. (2008). Supramaximal training and postexercise parasympathetic reactivation in adolescents. Medicine and science in sports and exercise , 40 (2), 362-371.
38. Coyle, EF, Coggan, AR, Hopper, MK, & Walters, TJ (1988). Determinants of endurance in well-trained cyclists. Journal of Applied Physiology, 64 (6), 2622-2630.
39. Meyer, T., Gabriel, H., & Kindermann, WILFRIED (1999). Is determination of exercise intensities as percentages of VO2max or HRmax adequate ?. Medicine & Science in Sports & Exercise, 31 (9), 1342-1345.
40. Garcia-Tabar, I., & Gorostiaga, EM (2018). A “Blood Relationship” Between the Overlooked Minimum Lactate Equivalent and Maximal Lactate Steady State in Trained Runners. Back to the Old Days ?. Frontiers in physiology, 9, 1034.
41. Wolpern, AE, Burgos, DJ, Janot, JM, & Dalleck, LC (2015). Is a threshold-based model a superior method to the relative percent concept for establishing individual exercise intensity? a randomized controlled trial. BMC sports science, medicine and rehabilitation, 7 (1), 16.
42. Allen H., Coggan AR, McGregor S. (2019). Training and racing with a power meter. VeloPress;
43. Inglis, EC, Iannetta, D., Passfield, L., & Murias, JM (2020). Maximal Lactate Steady State Versus the 20-Minute Functional Threshold Power Test in Well-Trained Individuals: “Watts” the Big Deal ?. International Journal of Sports Physiology and Performance, 15 (4), 541-547.
44. McGehee, JC, Tanner, CJ, & Houmard, JA (2005). ACOMPARISON OF METHODS FOR ESTIMATING THE LACTATE THRESHOLD. The Journal of Strength & Conditioning Research, 19 (3), 553-558.
45. Friel, J. (2018). The cyclist’s training bible: the world’s most comprehensive training guide. VeloPress.
46. Rankinen, T., Sung, YJ, Sarzynski, MA, Rice, TK, Rao, DC, & Bouchard, C. (2012). Heritability of submaximal exercise heart rate response to exercise training is accounted for by nine SNPs. Journal of Applied Physiology, 112 (5), 892-897.
47. Lamberts, RP, Swart, J., Noakes, TD, & Lambert, MI (2011). A novel submaximal cycle test to monitor fatigue and predict cycling performance. British Journal of Sports Medicine, 45 (10), 797-804.
48. Buchheit, M., Simpson, MB, Al Haddad, H., Bourdon, PC, & Mendez-Villanueva, A. (2012). Monitoring changes in physical performance with heart rate measures in young soccer players. European journal of applied physiology, 112 (2), 711-723.