Training guided by heart rate variability

In recent weeks I have focused on the role of my Polar RS800cx in alerting me to the possibility of atrial fibrillation.   While this has been an interesting and perhaps important diversion, my primary purpose in obtaining a Heart Rate Monitor with the capacity to measure heart rate variability (HRV) had been to explore its utility for monitoring my training, and in particular, for the identification of over-reaching and overtraining.   In this post, I will return to my original purpose, and address the question whether or not it has proven useful.  But first of all, it is worth reviewing the published literature on the subject.

HRV at rest

The most consistent body of evidence comes for studies that employ measurement of HRV at rest.  Several different processes contribute to fluctuations in heart rate. One of the most important is the fluctuations known as respiratory sinus arrhythmia (RSA).  Heart rate accelerates as you breath in and decelerates as you breath out. This fluctuation is a one of the main contributors to high frequency fluctuations in HR (frequency range 0.15 – 0.4 Hz).  At rest, RSA is governed by the parasympathetic division of the autonomic nervous system.  It serves to ensure that the flow of blood via the pulmonary artery from heart to the lungs in maximal when the lungs are full of air.  You can easily appreciate this variation in heart rate by manual palpation of the pulse at your wrist as you breathe in and out slowly.   The magnitude of this fluctuation provides a good measure of the activity of the parasympathetic nervous system, which has the overall role of regulating bodily processes concerned with rest and recovery.  Hence, the magnitude of RSA at rest is a fairly good indicator of how well recovered you are from a heavy training session.   

In general, heavy training causes a short term reduction in high frequency HRV but as recovery proceeds there tends to be a compensatory rebound reflecting increased fitness.   If HRV is still reduced the morning after an intense training session, yet another intense session is likely to produce yet further reduction the following morning, but eventually there is a somewhat paradoxical switch from parasympathetic under-activity to parasympathetic over-activity – perhaps the body’s self-defense strategy in the face of persistent stress   By this stage, parasympathetic over-reaching is on the way to becoming the seriously debilitating condition know as parasympathetic over-training, characterized by persistent fatigue, low mood and reduced performance.

In principle, monitoring high frequency HRV each morning can be used to determine whether or not you are sufficiently well recovered to undertake another heavy training session.  The practicality of this was first demonstrated in a study by Kiviniemi and colleagues from Oulu in Finland [1].  They used morning measurements of high frequency HRV to adjust the training of a group of male recreational athletes over a period of 4 weeks.  They compared increases in fitness, assessed by measurement of VO2max and also maximal running speed during a graded maximal exercise test on a treadmill, in 9 young men who trained according to HRV guidance, with the increases in fitness in a matched group of men who undertook a fixed training schedule comprising 4 high intensity sessions, 2 low intensity session and one rest day per week.  

In the HRV guided group, HRV was measured using a Polar s110 monitor, during 3 minutes of sitting followed by 5 minutes of standing at home each morning.  Because parasympathetic effects in fit athletes can be saturated when sitting, HRV during the 5 minutes of standing was employed to guide training. The calculation was done automatically using Polar software, and included a quite elaborate procedure for removal of artifact.  The full details of the schedule for adjusting training according to HRV were complex, but the crucial issue was that it high frequency HRV measurement was below a threshold level set at one standard deviation below the mean of the measurements in the  preceding 10 days, low intensity training or rest was selected for that day.    

To check on the reliability of the calculation of HRV, the data was subsequently examined in the laboratory using a more sophisticated procedure based on an 18th order autoregressive spectral model.  One the 10 subjects was dropped from the study because the laboratory check did not agree with the automatic computation performed at home.  In view of the frequency of artifacts in HRM data, this is perhaps not entirely surprising, though in my own experience, artifacts during resting or standing recordings are extremely rare if the belt has been properly moistened.  Maybe there was some other malfunction, or perhaps the subject had an abnormality of heart rhythm.  Whatever, the explanation, the fact that 1 in 10 subjects was dropped is an indication that automated measurement of HRV is not entirely free of technical hitches.

The outcome of the study was that the HRV guided group showed increase in both VO2max and also in maximum running speed in the treadmill test (from 15.5 to 16.4 Km/Hr).  The comparison group who followed the standard training protocol also showed improvement, with maximum running speed in the test increasing from 15.1 to 15.7 Km/Hr.  The increase in maximum running speed was significantly greater in the HRV guided group.  The difference between groups in in increase in VO2 max was not statistically significant.

Thus, this study provides encouraging evidence of benefits of HRV guided training. However, before recommending that serious athletes should invest in a HRM with the capability of measuring HRV, several points should be noted.  Most important was the fact that the standard training protocol included  4 intense sessions per week, in two back-to back  pairs of sessions.  Many coaches and athletes would regard back-to-back intense sessions as creating too great a risk of over-training.  The intense session involved warm-up followed by  30 minute tempo run at 85% HRmax, which I would regard as only moderately high intensity.  Nonetheless, the fact that in the HRV guide group the average number intense sessions was only three per week indicate that that appreciable  over-reaching occurred at least once per week on average, confirming what many sensible coaches and athletes know: back-to-back intense sessions are risky.   This would be  expected to impede the development of the benefits of training.  Furthermore, it is probable that at least some of the standard group were in a state of over-reaching which might have impaired performance on the test day.  In the example provided in the manuscript, illustrating the training of one of the  individuals  in the standard training group, the individual did show evidence of impaired HRV on the test day.   Finally, in view of the non-trivial the cost of a HRM capable of measuring HRV, it would be interesting to know if a resting pulse measurement, or maybe a measure of the orthostatic rise in pulse on standing from sitting, would have provide adequate information to guide training.

A replication

More recently Kiviniemi and colleagues have reported a similar study [2] in a group of 21 men and 22 women, with only a moderate level of preceding activity.  In this study, HRV was assessed similarly, during 3 minutes of standing.  The measurement employed was SD1 – the spread across the 45 degree line in the Poincare plot (as illustrated in my post of 17th July 2009).  SD1 is a little easier to compute than high frequency power, but is closely related to high frequency power.  The training program was for eight weeks. The standard protocol specified at least 3 high intensity sessions per week (the average number achieved was 3.3 /week) .  In the males, the outcome was very similar to the earlier study: improvement in maximum output during testing on a cycle ergometer was significantly greater in the HRV guided group than the standard training group, though the increase in VO2 max did not differ significantly between groups.  In the women, there was no significant  difference between groups in the improvement in fitness, but the HRV guided group did fewer intense sessions.  Again, it is clear that the HRV guided protocol detected over-reaching at least once per week, confirming that back-to-back intense sessions is undesirable.

 In conclusion, the two studies by Kiviniemi and colleagues confirm that HRV guided training produces greater improvement in fitness than standard training that includes excessively frequent  high intensity sessions.  The question of whether similar improvement in training could be achieved more cheaply by manual recording of morning heart rate, or by more sensible coaching, remains unanswered.

My own experience

Last summer, I was quite ill for several weeks, and as I recovered I faced the need to get fit enough to race a half-marathon in about 7 weeks time.  I cautiously increased training intensity to include two moderately intense sessions per week (eg 5K at 80-85% HRmax) and a longish run.  However I had not allowed for the degree to which I had been debilitated by my illness, and after about four weeks I began to feel quite marked fatigue.  Measurement of HRV suggested that I was suffering from over-reaching.  However simple measurement of the orthostatic rise in HR provided even clearer evidence that I was suffering from quite serious parasympathetic over-reaching.  I usually have an increase in HR of about 8-10 BPM on standing from resting, but this orthostatic increase was abolished entirely as the fatigue developed, as shown in the figure. 

Orthostatic increase in heart rate showing nadir on 18th August. red arrows indicate long-run days; blue arrows indicate rest days

I was able to implement adjustments to my training schedule that resulted in relief of the fatigue after about 2 weeks.  In fact I was not completely recovered, as I will reveal in my next post, but my aerobic fitness was enough to run the half-marathon.  Unfortunately due to inadequate number of long runs my leg muscles were not well conditioned and  I tore one of my hip adductors after about 10 Km.  I limped painfully with a stride that was very short and with virtually no airborne time, for the final 11 km.   Despite this mishap, my experience is that monitoring for signs of over-reaching is indeed worthwhile, though my experience does not answer the question of whether or not it is worth investing in a HRM capable of measuring HRV.

But what about measurement of HRV while running?  That is a far more complex question, which I will address in my next post*.

*Added note: discussion of HRV whle running deferred until the post after next



[1] Antti Kiviniemi, Arto Hautala, hannu Kinnunen & Mikko Tulppo (2007) Endurance training guided individually by daily heart rate variability measurements. Eur J Appl Physiol. 101(6):743-751.

[2] Kiviniemi AM, Hautala AJ, Kinnunen H, Nissilä J, Virtanen P, Karjalainen J, Tulppo MP . (2010) Daily exercise prescription based on Heart Rate Variability among men and women. Med Sci Sports Exerc. 42(7):1355-63


11 Responses to “Training guided by heart rate variability”

  1. John Thompson Says:

    Have you tried the following gaget?

    I have not – yet. One of my (cycle) club mates suggested it and I found your web site when researching HRV and its uses.

    • canute1 Says:

      I have not used Ithlete , though I have had several very informative email discussions with Simon Wegeriff, the cyclist (and signal processing expert) who designed the device.

      Ithlete is designed for measuring high frequency heart rate variability (HF HRV) during a one minute period when resting, under similar circumstances each day. This information is very similar to the information used by the athletes in the studies Kiviniemi which I described in my blog post . Ithlete, Suunto t6 and the Polar RS800 all provide the data required for the calculation of HF HRV. Ithlete is much more economical than the others, and in general, it is simpler to use.

      However with Suunto T6 and the Polar RS800 you can examine the data in much more detail and perform different types of measurements

      Because I sometimes have an irregular heart beat, I find it useful to be able to examine the data in greater detail that would be possible with an Ithlete. Also, my own personal; experience is that when suffering from serious fatigue, it is useful to be able to perform additional measurements such as the orthostatic test. I am also interested in examining HRV while exercising . So my opinion is that for an athlete who wants to examine heart rate data in various different ways, it is worthwhile to buy the more expensive Suunto T6 or Polar RS800. However, if you want a simple, economical device, I think the ithlete is probably good value.

  2. soiltosustenance Says:

    First, thanks so much for your blog. If you noticed any type of spike in your stats yesterday, that was me crawling through every post regarding HRV. Your insights into your experience will provide me with a great deal of assistance as I dig into my own r-r data (just purchased a Suunto t6d).

    re: Orthostatic heart rate – Are you saying that your resting heart rate was elevated? Or that your standing heart rate failed to kick up its normal 8-10 beats during the period in the graph when you were fatigued?


    …Tim Huntley

  3. canute1 Says:

    Tim, thanks for your comment. I am pleased you have found the site interesting. I usually get about 100 page views a day. It was 248 yesterday so I suspect that reflected your interest – thanks for that interest. The all time record for page views per day is almost 600, but I suspect that a larger than usual number of readers contributed to that total.

    With regard to your query about the low orthostatic test in mid August, there was a weak trend for resting HR to be slightly higher than average around that time; but the main contribution to the loss of orhtostatic effect was a lower value when standing. On 18th Aug when the orhostatic increase (from sitting to standing) was abolished entrely, sitting HR was 50. On standing it rose to 60 by 20 seconds but fell back to 49 by 30 sec and remained around 49-50 for the next few minutes. In the preceding few weeks, sitting HR had been in the range 47-50 with an avergae of 49. On standing it usualy rose immediatley to around 60-65 and then settled at aound 58-60 for the subsequent few minutes. I regard the difference between sitting and the stable plateau from 30 sec to 2 minutes after standing as the most reliable measure of the orthostatic effect

  4. soiltosustenance Says:

    Thanks so much for the quick reply. Do you know if that is typically the measurement that is impacted by overtraining/fatigue (standing heart rate – plateau)?

    Also, I am curious as to what your diet is like? I don’t recall seeing anything about how you eat in any of the posts that I read.


    • canute1 Says:

      The parasympathetic form of overtraining is characterised by increased parasympathetic activity. On standing from a sitting position, the rise in HR is largely due to withdrawal of parasympathetic input. In the over-trained state, parasympathetic activity is excessive and the parasympathetic withdrawal is weaker. Hence both the initial rise of HR in the first 20 sec and the plateau that follows tend to be weaker. My own experience is that the plateau value provides a more consistent measurement. The initial rise also tends to be weaker but this effect is less consistent.
      Some people suggest that a lower HR during light exercise is also a good marker but I have not examined this in a systematic manner.
      I do not follow a strict diet, and am fairly omnivorous, but tend to eat mainly vegetables, fruit, cereals & rice, fish, chicken, cheese, a modest amount of red meat, and nuts. I consume only small amounts of sugary or fatty foods and avoid needlessly processed foods such a white bread. I drink mainly tea or water, a modest amount of red wine and an occasional beer. The principles that guide my choice are that it is especially important for an older person to protect muscle. Hence during regular training it is important to consume enough carbohydrate to keep glycogen stores topped up and enough protein to rebuild any muscle that gets damaged. I am very sceptical about long runs after an overnight fast because of the risk of muscle catabolism. I also aim to keep omega 3 fatty acid levels topped up (from oily fish) to minimise degradation of nerve cells; and to take in a moderate amount of foods containing antioxidants to limit oxidative damage to tissues. I do not take any supplements.

  5. soiltosustenance Says:

    Thank you. I had a big light bulb come on after that explanation. So rather than being in a state that looks more like chronic stress (stronger sympathetic control), your parasympathetic was being protective by keeping you from overexerting yourself (keeping your heart rate low).

    I am easing myself back into some mid-to-upper range aerobic activity after years of neglect. I am 44, 6’1″, and 165 lbs; however anything resembling fitness ends with there. I have always had what I would consider an elevated resting heart rate, and now armed with some basic information about orthostatic heart rate and HRV (plus a new Suunto t6d), I believe I have confirmed how much work I need to do. Here is some rough data for the past 3 mornings (these are 2 minute readings within a 3 minute window – immediately after waking and seated):

    SD1 (ms) SD2 (ms) Mean HR RMSSD (ms)
    10-Feb 28.9 129.9 74.4 40.8
    11-Feb 31.2 136.4 73.8 43.9
    12-Feb 34.1 141.6 75.8 48

    Orthostatic differences for those same three days were 19, 24, and 30.

    I am afraid that my Poincare plots look more like long, skinny cigarettes vs. nice fat eggs…

    Again, thanks for all of the self-experimentation you have done (and written about).

    All the best,


    • canute1 Says:

      The important comparison are between your own measurements over time. However, for what it is worth, here is a typical example of my own data from last year.
      Sitting: sd1 42; sd2 69, RMSSD 60; HR 47
      Standing: sd1 31; sd2 80; RMSSD 44; HR 56
      (Orthostatic rise 9)
      In recent months, I have simply been recording RMSSD for 2 min while standing in a relaxed state. Typical values are in the range 50-110 ms, and standing HR is typically 50-53. Both the standing RMSSD and standing HR indicate a modest improvement in fitness since last year

  6. soiltosustenance Says:

    Thanks very much. I plan to track my numbers every day for the next few months to see how things change.

  7. food supplement drinks Says:

    food supplement drinks…

    […]Training guided by heart rate variability « Canute’s Efficient Running Site[…]…

  8. Deedee Wiley Says:

    Under FDA regulations at 21 CFR part 111, all domestic and foreign companies that manufacture, package, label or hold dietary supplement, including those involved with testing, quality control, and dietary supplement distribution in the U.S., must comply with the Dietary Supplement Current Good Manufacturing Practices (CGMPs) for quality control.*’*:

    Check ya later

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: