Monitoring stress, recovery and fitness

The gains from training effort vary greatly both within and between individuals.  Genes and life-time training history play a large part in determining differences between individuals in response to training.  However, genes cannot feasibly be altered and life-time history takes a life-time to create.  But there are also factors that we can control in the present that have the potential to influence how much benefit we get from training.

Perhaps the most important controllable factor is the balance between stress and recovery.   Training creates stress – there is microscopic damage to muscle fibres and an increase in catabolic hormones that break down the tissues of the body.  During recovery, local repair processes such as the engagement of satellite cells in muscle aided by systemic anabolic hormones such a growth hormone and vascular endothelial growth factor which promote the development of new capillaries, and the synthesis of mitochondrial enzymes, make the body stronger and fitter.  However, if the training stress was too large or too sustained, the body fails to adapt and might even suffer irreversible degradation – the over-training syndrome.

Autonomic indices of stress and recovery

Because the ways in which the body adapts to stress are complex there is no single physiological measurement that allows us to monitor stress and recovery.  However one of the most reliable measures is the balance between the two divisions of the autonomic nervous system: the sympathetic system that employs adrenaline as the signalling molecule to mobilise the body in response to stress; and the parasympathetic system that employs the signalling molecule acetylcholine to mediate recovery.   Because increase in sympathetic activity increases heart rate while increase in parasympathetic activity increases beat-to-beat variability in heart rate (high frequency HRV) we can monitor the balance between stress and recovery by measuring heart rate and/or  high frequency HRV.

However, there are three major complications that make this tricky.

  • First of all, the balance between the sympathetic and parasympathetic systems is affected many transient circumstances, such as the occurrence of loud noise or time of day, that have only minor relevance to the balance between stress and recovery on the time scale relevant to repairing the body.
  • Secondly, the way in which the autonomic system is regulated is not simple.  Although initially stress causes an increase in sympathetic activity, it the stress is too great or too sustained, the body reacts by a compensatory increase in parasympathetic activity – the rabbit in the headlights situation.
  • Thirdly, as we get fitter, the amount of parasympathetic activity in any standardised restful situation increases, while the amount of sympathetic activation required to generate a particular power output decreases.  This third factor provides additional useful information: increased high frequency HRV at rest or decreased HR at a given level of power output provides a useful measure of increased fitness, but there is a catch.  Changing fitness level produces a shifting baseline if we wish to use sympathetic/parasympathetic balance to assess the current balance between rest and recovery, while transient stress such as sleep disturbance, can interfere with the use of either resting HRV or HR in response to load as a measure of fitness.

One way of dealing with these problems is to perform regular measurements, perhaps daily, under very carefully standardised conditions – such as immediately after getting up in the morning in a quiet setting while breathing at a predetermined rate – as recommended for HRV measurements using the Ithlete, developed by Simon Wegerif.  I have followed this procedure over a period of many months, using my Polar RS800CX to provide the required information about beat to beat variation in heart rate, rather than the Ithlete.   However, at least with my varying daily schedule and sleep patterns, apparently random fluctuations tend to overwhelm the systematic changes of interest.  Furthermore, I found it difficult to distinguish the increases in HRV that denote healthy recovery from those heralding the onset of over-training.   While I cannot rule out the possibility that Simon’s Ithlete algorithm is more successful than my own interpretation of the data, the inconsistency of some of the variation in HRV led me to conclude that additional data acquire under a range of circumstances might be more informative.

The role of the central governor in setting heart rate

As you get fitter HR at a given power output falls. There is a fall in HR relative to maximum heart rate, due to increased efficiency of the delivery of oxygen to muscles, but also an appreciable fall in maximum heart rate.  Maximum heart rate is usually identified as the rate beyond which there is no further increase despite increasing work load when power output is increased in a graded manner.  Interestingly, it is sometimes possible to achieving a rate greater than this maximum by an initial brief, intense burst of activity at a power output exceeding that required for peak HR (as in a Wingate test), and then allowing power output to decrease gradually.  This suggests that maximum heart rate is not fixed but is set by a ‘central governor’.    I do not think that the ability to trick the governor into allowing a higher HR is a useful procedure, but the realisation that HR at high power output might be regulated by a governor led me to reconsider an observation that I reported in my post of 12th September 2009 and have repeated on several occasions since: namely that following a period of excessive training, a major subjective effort required to push HR even to the level of the second ventilatory threshold (beyond which acidity accumulates very rapidly creating a very powerful respiratory drive). In that post of 12th September 2009 I had described the experience of overwhelming subjective difficulty in maintaining a high power output on the elliptical cross trainer despite a lower heart rate than usual at that power output, at a time when I was suffering from over-training, following over-enthusiastic recommencement of training after illness.   It felt as if a powerful restraining influence was preventing my heart from beating too hard.

This suggested that the most sensitive way to assess impending over-training might include not only resting measurements of HR and HRV but also measurements across a range of work-loads spanning the full aerobic range.  Although in principle, such measurement might be performed either running or on the elliptical cross trainer, on account of the fact that I am limited mainly to training on the elliptical when my arthritis is troublesome, I initially developed a protocol for the elliptical cross trainer.   I have subsequently developed a version of the test for use to measure HR when running at various paces.  The running version works similarly, but because I have acquired data on the elliptical at the beginning and end of my recent period of base-building, I present the elliptical version in this post and will describe the running version in my next post.

Design criteria

The criteria that I applied in developing the protocol were:

1)      It should proved a comprehensive set of measures of autonomic function across a range of work-loads from rest to near the second ventilatory threshold (identifiable in practice by the need to increase respiratory rate to around 80/minute, and corresponding approximately to the anaerobic threshold).

2)      It should be feasible to include it within a normal warm up, requiring less than a total of 20  minutes and demanding only a short period of high effort.

3)      The key measurements should be based on heart rate, though during development of the test I used Polar RS800CX to record the R-R interval between consecutive heart beats, thereby allowing comparison of the predictions based on HR with those based on HRV.

I modelled the major features on the protocol on the sub-maximal cycling test developed by Lamberts and Lambert, in which heart rate is measured at three different work-loads.

The test protocol

The key component of the test is an assessment of heart rate at three work-loads spanning the aerobic range: a load in the lower part of the aerobic zone, with heart rate around 60% of maximum heart rate; a medium load with heart rate around 75% of maximum, and a high load with heart rate around 85-90% of maximum.  The low and medium loads are maintained for 4 minutes each, during which time heart rate reaches a stable plateau, while the high load is maintained for only three minutes.   At the high load, the HR does not reach a plateau.  Achieving a plateau at this work-load takes about 10-15 minutes and would make the test unnecessarily stressful.

The three work-loads were achieved at resistance settings 4, 8 and 12 on the elliptical.  I maintained a constant cadence of 80 cycles/min (160 steps per minute) giving a power output of 62, 135 and 199 watts at resistance settings of 4, 8 and 12 respectively. I would anticipate that a younger and stronger person would require higher resistance settings and greater power output to achieve the target heart rates for the three levels.  In addition to recording heart rate, I also record the subjective effort level at each of the three levels on a scale from 1 to 20 where 1 is very easy (typical of relaxed walking, 3 is slight effort (eg faster walking or very easy jogging), 8 is comfortable running, while 16 requires determined effort to sustain and 20 is the hardest effort I could imagine.

To provide a more comprehensive picture of the autonomic responses the test also includes measurement of resting HR and HRV while standing in a relaxed state for one minute before the start of the test, and measurement of heart rate recovery (HRR) while standing in  a relaxed state for one minute at the end of the high load phase.  Between the initial minute of relaxed standing I do a  very gentle warm-up for 2 minutes, at around 55% HRmax, so as to exclude that the initial period of erratic HR (apparently due to somewhat erratic opening of capillaries) from the low-load test period.  After the second period of standing, I do 4 minutes of gentle cool down at the same work load as the initial warm-up to assess the stress accumulated during the test, and to make a transition to subsequent activity.

The time course of the HR during the test is shown in figure 1.  R4, R8 and R12 denote the three resistance settings required to produce the target heart rate for the three different levels of the test.  The red trace is a recording in the week immediately preceding a 10 week block of base-building with an average weekly training volume equivalent to 50 miles of running. The blue trace is a recording at the end of this training block. Comparison of the red and blue traces reveals a similar proportional decrease in HR at each load.  This is also shown the bar chart in figure 2.  The height of the bars represents the heart rate at each of the stages of the test.  There is a reduction in heart rate of approximately 5% at each of the three test loads.  This fairly uniform reduction in heart rate across the aerobic range should be compared with the fairly uniform reduction of about 10%  in heart beats/Km across the entire aerobic range achieved by Hadd’s protégé Joe following 16 weeks of base-building at a weekly training volume that was mostly above 80 miles per week.  Bearing in mind that I am an elderly runner with limited capacity to withstand hard training, I am pleased that I achieved a roughly similar amount of improvement relative to training volume to that achieved by Joe.  I will describe this base-building in more detail in a later post.

Fig 1: beat by beat records of heart rate during tests on 13th March, in the week before 10 weeks of systematic base-building (red trace) and on 1st June, at the end of this block of base-building (blue trace). R4, R8 and R12 denote the elliptical resistance setting require to achieve the heart rate in the three target ranges.  Note the large heart rate variability when standing.  The sporadic spikes at around 4 min in the red trace are probably premature atrial contractions.

Fig 1: beat by beat records of heart rate during tests on 13th March, in the week before 10 weeks of systematic base-building (red trace) and on 1st June, at the end of this block of base-building (blue trace). R4, R8 and R12 denote the elliptical resistance setting require to achieve the heart rate in the three target ranges. Note the large heart rate variability when standing. The sporadic spikes at around 4 min in the red trace are probably premature atrial contractions.

The traces in figure 1 show beat by beat variation, and appear more erratic that the 5 second averages produced by most heart rate monitors.  The high variability during relaxed standing at the start indicates large high frequency variability in heart rate, indicative of a high level of parasympathetic activity. Similarly the rapid increase in variability after about 50 seconds of standing during the recovery period activity demonstrates the return of parasympathetic activity.  The sporadic sharp spikes at around 4 minutes in the record for 13th March (red trace) are probably premature atrial contractions, which are fairly common in an elderly runner.

Fig 2: Effects of base-building: bars indicating the near-stable hear rate achieved in each phase of the test, at the beginning (red bars) and end (red bars) of 10 weeks of base-building.  WU denotes warm-up; CD denotes cool-down.  Due to instability as capillaries open during the first 90 seconds of the warm-up, the value in the final 15 seconds is shown.  The other values are the average over the final 1 minute of the relevant phase of the test.

Fig 2: Effects of base-building: bars indicating the near-stable hear rate achieved in each phase of the test, at the beginning (red bars) and end (red bars) of 10 weeks of base-building. WU denotes warm-up; CD denotes cool-down. Due to instability as capillaries open during the first 90 seconds of the warm-up, the value in the final 15 seconds is shown. The other values are the average over the final 1 minute of the relevant phase of the test.

The detection of over-training

At the beginning of the year I had suffered an episode of mild arthritis.  As the joint pain began to resolve in late January I was eager to build up the volume of training on the elliptical cross trainer.  However, in my eagerness I built up the training volume too rapidly, and at the end of the month, I was becoming exhausted.  Figure 3 depicts the heart rate at each test phase during tests performed on 27 and 29th of January and again on 1st Feb.

Fig 3: The early signs of over-training.  From 27th to 29th January heart rate at each phase of the test dropped rapidly by almost 5% while the subjective effort at the high load level (R12) increased dramatically.  The decrease was greatest at R12. After 2 days rest, HR in 1st Feb had returned to levels similar to those recorded on 27th January.

Fig 3: The early signs of over-training. From 27th to 29th January heart rate at each phase of the test dropped rapidly by almost 5% while the subjective effort at the high load level (R12) increased dramatically. The decrease was greatest at R12. After 2 days rest, HR in 1st Feb had returned to levels similar to those recorded on 27th January.

The crucial informative feature was my rating of subjective effort at the high load.  During all 5 tests in the preceding two weeks in mid-January, I had rated the effort at the high load as 14 on a scale from 1 to 20.  On 27th Jan I also assigned a rating of 14 but two days later, on 29th, there was a clearly perceptible change in the effort required.  It required a determined effort to complete the three minutes and I rated the effort at 16.   This increase in effort was accompanied by a sudden drop on HR of almost 5% at each of the three loads.  I recognise that this was the onset of an over-trained state and rested for two days. By 1st February, the heart rate reading were again back to the levels recorded on 26th Jan, and I found it quite easy to complete the three minutes at the high load.  I assigned an effort rating of 13.

Thus, the comparison of 27th and 29th January reveals a drop in HR very similar in magnitude to that achieved by 10 weeks of systematic base-building a few months later.  The crucial factors are that this reduction of almost 5% occurred over a 2 day period and was accompanied by a very easily perceived increase in subjective effort.  However the problem resolved rapidly with two days of rest indicating that it was only the early phase of over-training.

The effect of moderate stress

During the systematic base-building phase starting in March, I built up training load in a well controlled manner and did not experience any dramatic sudden reductions in heart rate at a given load.  However other aspects of my life were less easy to control.  In mid-May experienced a demanding period at work.  On 13th and 15th May I had only a very short night’s sleep.  I did continue training according to my plan, but in a sub-maximal test on 16th May, I noted that the HR at the various levels of the test was 3-5% higher than it had been two weeks earlier.  It is probably significant that this effect was most noticeable at the low load and also in the standing HR.  By the beginning of June, the stressful time at work was behind me and the heart rate at each level was back to the value at the beginning of May (though slightly lower at the R4 and R12 levels, perhaps indicating a small gain in fitness due to training).  Thus, the immediate effect of stress was an increase in HR that reached its peak following a night of seriously curtailed sleep, but it is also noteworthy that the gains in fitness over the entire month were only marginal.

Fig 4: the effect of a stressful week at work with reduced sleep, in mid-May.  HR increased by about 3-5%, with the greatest increases in the standing and low load phases of the test after a busy previous day and short night’s sleep preceding the test on 16th May.  The overall gain in fitness in May was only marginal, as indicate by the small reductions in HR at R4 and R12 on 1st June relative to 4th May.

Fig 4: the effect of a stressful week at work with reduced sleep, in mid-May. HR increased by about 3-5%, with the greatest increases in the standing and low load phases of the test, after a busy day at work and a short night’s sleep preceding the test on 16th May. The overall gain in fitness in May was only marginal, as indicated by the small reductions in HR at R4 and R12 on 1st June relative to 4th May.

Standing heart rate and HRV

Standing HR shows similar pattern of variation with training (fig 2), over-training (fig 3) and stress (fig 4) as the HR at each of the three test levels.  However, in general standing HR is less reliable than during exercise because it is more sensitive to current stress levels.  In fact the standing and R4 scores alone can provide a very quick and easy test of current stress level, but detection of the early over-training is more reliable when the full range of scores is obtained.

On days when standing HR was low, high frequency HRV was usually high, confirming strong parasympathetic activity and generally denoting good recovery.  When standing HR was less than 48 beats/min, the root mean square of deviations (RMSSD) was typically greater than 70 milliseconds, whereas at times of stress, standing HR exceeded 50 beats/min, and RMSSD was usually much lower, sometimes as low as 41-43 milliseconds.  The lowest recorded standing HR during base-building was 45 and this was accompanied by an extremely high RMSSD of 163 milliseconds.   Thus RMSSD provides a rough guide to the level of recovery, though the extremity of the swings in RMSSD made it an unreliable measure.  On 29th Jan, when I was showing signs of over-training RMSSD was 94, a moderately high value indicating substantial parasympathetic activity, but the definitive evidence of over-training was provided by the low HR and high effort at the high load (R12 level).

Heart rate recovery (HRR)

Heart rate recovery (HRR) tended to increase steadily as I became fitter.  The drop in HR during 1 minute when standing in a relaxed state immediately after cessation of the high load epoch was typically 50 beats/min in January and February as I began the build up training volume. It had increased to 55 beats/min by the time I commenced regular base-building in mid-March and was in the low 60’s by the completion of the block of base-building in early June.  It was atypically low (45 beat/min decrease over 1 min) when I showed sign of over-training on 29th Jan and atypically high (71) when I exhibited signs of stress on 16th May. Thus HRR augments the picture provided by the changes in HR at the high work load (R12) but shows less consistency.


While the HR at the three work-loads provides the most reliable information, the picture is consolidated by the overall pattern, including standing HR, HRR and HRV.  The use of this sub-maximal test gave me the confidence to build up my training load at a fairly rapid rate, while nonetheless avoiding serious over-training or injury.  However it was also noteworthy that the data indicated that I achieved only a small increase in aerobic fitness during the month in which overall stress level was quite high, suggesting that I might have benefitted more if I had taken more rest during that month.

Addendum added 28 June 2013. 

In his comment below, Ewen suggests that large amount of data generated by this test appears a bit over-whelming. He reports that he himself simply measures resting HR.  As I noted in my response to him, I consider that resting HR is certainly a worthwhile measure, but the reason I have spent time designing this more complicated test is that I found that a single measures such as resting HR or resting high frequency HRV are too sensitive to various different transient circumstances, making it difficult to interpret the results reliably.  To some extent one can deal with this by performing a single measurement such as resting HR or HRV under constant circumstances every day, but with my variable life routine this doesn’t work well enough.  So in the test described in this post, I have adopted the approach of performing multiple different measurements within a more complex test. It is still necessary to standardise the circumstances, but the overall picture provided by multiple measurements makes it possible to obtain a more reliable estimate of changes in both fitness and recovery.  However, it does produce a bewildering amount of information.  While I find this large amount of information quite interesting I accept that it might appear a bit over-whelming.  So maybe for routine use, one should focus on one key measurement and only examine the other variables when the key measurement is out of line with expectation.

I think that the single most informative measurement is the heart rate at a medium work load.   Provided the load is consistent, this measurement is the one that is most consistently related to overall recovery and fitness. If this HR is decreasing slowly over a period of many weeks during a block of training, all is well.  If there is a rapid jump upwards from recent values, it probably indicates transient stress. This can be confirmed by noting an increase in standing HR (and perhaps a decrease in HF HRV if you have a monitor that records beat by beat variation). It is no cause for alarm, but it is worth identifying and dealing with the stress.    A sudden drop in HR at medium load is the important marker.  If this is accompanied by additional effort that is most noticeable at the  higher workload, it is probable that you are in the early phase of over-training and need to adjust training accordingly.


19 Responses to “Monitoring stress, recovery and fitness”

  1. Ewen Says:

    Wow. That’s a lot of data to absorb. As I’ve no way of measuring HRV, I may get the ithlete monitor and app. I could probably work out a repeatable procedure using the Concept2 erg, as ‘conditions’ would be the same on an indoor machine, rather than running.

    Phil Maffetone mentions how max HR changes in his book (depending on fitness) – reducing as one becomes aerobically fitter. That’s what I’ve found too — it’s easiest to get a high HR reading after I’ve had time off from running – for example, I recorded 159 in a 2.5k race (the second run back from my 24 days off) — and I didn’t sprint all out at the end of that race! It was easy to hit 159 and I didn’t feel stressed doing it. I’ve regarded my current max HR as 161. When I was fit last summer I would usually only get 157 or so as a maximum, and that was racing flat out with a sustained sprint finish.

    Anyway, interested in how the running method works. I should continue to monitor resting HR, even though it’s steadily reducing as I’m returning to fitness.

    • canute1 Says:

      Thanks for your comment. Resting HR is certainly a worthwhile measure, but the reason I have spent time designing this more complicated test is that I found that single measures such as resting HR or resting high frequency HRV are very sensitive to various different transient circumstances and hence difficult to interpret reliably. I will add an addendum to my post to explain in more detail why I have developed a test employing multiple different measures, and also discuss how in routine use it is probably best to focus on one key measure and only examine the other measures in detail when the key measurement is out of line with expectation.

  2. Helen House Says:

    What a great post! You are so diligent in your search for the sweet spot of training/ good inflammation.
    I was wondering what your verdict on HRV would show. It sounds like a bit too high noise to signal ratio.
    What I also notice is after all your careful HR measurements your score of 16 on the how-tough-was-that record did flag up the need to rest.
    Well, maybe that’s enough? Do you ever get a high score on that subjective scale, but chose to ignore it because the HR data conflicts?

    • canute1 Says:

      Thanks for your comment, and for raising the possibility that subjective assessment of degree of difficulty alone might provide a good indicator of when it is necessary to rest. I have scored effort level as high as 16 or 17/20 on only three occasions. On all three occasions, HR during the three different effort levels within the test was lower than usual, confirming the early phase of over-training. On two of the three occasions I rested the next day and on the third occasions I did only a very easy elliptical session. On all three occasions, 1 rest day or easy day led to a substantial recovery. So as you surmised, a sudden marked increase in effort level for a given work load is a good indicator of the need for extra rest. However I feel less confidence in my assessment of subjective effort than in an objective recording of HR, so I do think that it is worth having additional evidence to back up the decision such as when to take an extra rest day. Nonetheless, I accept that provided you are well attuned to effort levels required for particular work-loads, it is possible to make a fairly reliable estimate of when an extra rest day is required simply be ‘listening to the body.’

  3. EternalFury Says:

    I think this is a great step in the right direction!
    I have been tinkering with various metrics as well and I also came to HRrest, HRR and HRV as key markers of the training/recovery cycle.

    However, I have great difficulty using RPE for anything. I have a tendency to perceive anything slower than HM pace as being “easy” and anything faster than 1K pace as being “hard”. I just don’t have the granularity of perception a 20-point scale requires.

    Others have expressed similar concerns before:
    Br J Sports Med 1999 33: 336-339
    K L Lamb, R G Eston and D Corns
    Reliability of ratings of perceived exertion during progressive treadmill exercise.

    In your scheme, RPE is key to detecting the early signs of overtraining, so it’s important to get this right.

    Yet, I wonder what I could use instead.
    If the Central Governor is at work, “it” probably uses physiological mechanisms that leave a biological trace. For instance, if at the same work output, HR is low and RPE is high, then maybe blood glucose concentration is also coincidentally low. (just an example, a similar case could be made for Epinephrine, if it were easy to measure)

    For HRR, I use the average HR measured over the last 30 seconds of a 75-second period of complete immobility.
    60 seconds is the usual way of measuring HRR, but there is no real standard, and I find the average HR over the last 30 seconds of the 75 second-period eliminates variations due to the vagaries of reaching the state of complete immobility.
    But honestly, it only matters to be consistent in the way the measurements are taken.

    For HRV, I have been using ithlete, and I can confirm the ratings using that tool are as erratic as you would expect. I found it hard to use the HRV index to modulate my training. Too many factors appear to influence the measurements and many in ways that weigh as significantly as the type of training you did the day prior.

    I will look forward to your protocol focused on “running”.

    I also wonder how one would go about distinguishing improvements in running economy from improvements in cardiovascular fitness. That probably cannot be done with this type of tests.

    I feel that you have done yourself and all your readers a great favor by taking the time to put together such a tool.

    • canute1 Says:


      Thanks for those comments. I agree that reliance on subjective assessment of effort for the identification of over-training is not ideal, though I would emphasize that maintaining a heart rate just below the anaerobic threshold does becoming perceptibly more demanding as over-training develops. Other features provide supporting evidence. An unexpected decrease in the HR at the higher levels is a strong pointer towards over-training.

  4. Re-appraisal: the benefits and damage produced by cortisol | Canute's Efficient Running Site Says:

    […] had used submaximal tests throughout the summer to ensure that I was just one step short of over-training, as indicated by […]

  5. Seth Leon Says:

    Hello Canute,

    Yes, it is me again 🙂

    I came across a study I found very interesting and potentially quite useful for employing a (very) sub-maximal running protocol for monitoring fitness/fatigue. In this study of 400 fit adults aged 18 to 40:

    Pat Vehrs and James George from BYU derive a regression equation for estimating vo2 max from a self-selected running pace that produces steady HR early in a run. The eqaution has an R=0.91 with vo2 max. Here is the equation.

    VO 2 max (mL • kg − 1 • min − 1 ): VO 2 max (mL • kg − 1 • min − 1 ) = 58.687+(7.520 × Gender; 0 = woman and 1 = man)+(4.334 × mph) − (0.211 × kg) − (0.148 × HR) − (0.107 × Age), where mph = self-selected jogging speed, HR = steady- state HR (bpm) while jogging at level grade at the self-selected jogging speed, kg = body mass, and age = age in years.

    I really like that the pace is self-selected as this allows for a structured incorporation of perceived exertion. I am not sure that the Vo2 max estimates hold for ages over 40 (although putting in a recent run of mine it looks to work out pretty well. As long as the relation between running speed and hr are generally stable at the selected exertion level when not in a over-training or fatigued state within an individual then using this simple formula and protocol should produce decent estimates of fitness/fatigue levels across time. All that is needed is to run with HR monitor and speed tracking watch and to start runs of on the flat while paying attention to a specific consistent perceived exertion level while finding a steady HR.

    I think I will use this rather than HRV recovery to track alongside the HRV data I am tracking. As you mentioned the HRV recovery seems unstable in the first few minutes post exercise. It would be troublesome for me to wait and collect it at long time frames, and this sub-maximal running measure looks to probably be a better quick fitness/fatigue indicator as well.

    If you are interested, and don’t have access to the paper feel free to ask me additional questions.

    • canute1 Says:

      Thanks, that is interesting. I have access to the paper

      It provides very similar information to that provided by the mid-aerobic phase of my three level submaximal (running) test. However, I use (pace/HR) as the measure of fitness whereas Vehr and colleagues use (constant + pace – HR) as their measure of VO2max. Vehr’s ‘constant’ is modified by gender, weight and age. Over the course of a few months these quantities are fairly constant, though I think that allowing for variation in weight is likely to produce a useful increase in precision.
      Although I have no independent estimate of my VO2 max, I will see how well the estimation of VO2 max obtained by applying Vehr’s formula to my submaximal tests correlates with my own estimate of my aerobic fitness.

      You have provided me with some interesting things to look at in my own data. At present I am fairly busy. I hope to do a post on nutrition next week-end and then after that I will turn my attention to the issues your comments have raised. I will be very interested to hear about your own experiences with the Vehr test.

      • Seth Leon Says:

        Good summary,

        What I specifically like about the Vehr approach is that you are running to a perceived exertion, not to a specified speed or HR. So both speed and HR can vary. I tend to have an ‘easy’, & a ‘comfortable’ perceived pace that I can subjectively find pretty well. If the formulas works correctly then the fitness estimate on a given day in theory when fresh early in a run should be similar at either pace. Also pace and HR may each have have different slopes (or standardized regression betas) relative to fitness estimation which direct (pace/HR) would not capture.

        I look forward to your nutrition post. I have been paying close attention to that aspect for quite awhile.

    • canute1 Says:

      I agree that the issue of perceived exertion is very important.

      First I should say that in the mathematical expressions I used in my reponse to your first comment above, I omitted several additional ‘constants’ for simplicity. These ‘constants ‘ are not strict constants but vary according to various circumstances including stress level. Nonetheless, both Vehr’s equation and my implicit equation both reflect the fact that for a given pace, low HR indicates greater fitness, while at a given HR, higher pace indicates greater fitness. My equation assumes a different mathematical form of the relationship that would affect the predictions over a large range of pace or HR, but over a small range, the variation is linear (as is assumed by Vehr). I deliberately specify that HR is measured at a near constant pace in each of the three stages, so both my test (in either the low or mid-aerobic stage) and Vehr’s test should yield virtually identical sensitivity to fitness change.

      However the crucial issue is the fact that stress level affects the ‘constants’, During over-reaching, HR will be higher for a given pace. Therefore you can identify the development of over-reaching by noting when HR increases at a given pace despite an increase in training , using either Vehr’s test or my test.

      But in the case of serious over-training, HR actually decreases at a given pace. The challenge is how best to identify the difference between improved fitness and over-training. In part this is based on the fact that the decrease in HR at a given pace is more dramatic when over-training occurs, but even more importantly, over-training is revealed by the fact that the effort to achieve a given pace is increased. With Vehr’s procedure, you identify serious over-training by noting that your pace at a given effort is much lower. In my test, you note that a higher effort is required to achieve a given pace (or power output). Furthermore, I deliberately included three levels of pace including low, mid and upper aerobic range. In my experience, the increased subjective effort is most easily detected in the upper aerobic zone.

      So in conclusion, I think the two tests are likely to have a similar sensitivity to changing fitness in the absence of over-training. Both will be sensitive to over-reaching. But I suspect my test might give a more sensitive indication of the onset of serious over-training because the increased effort is likely to be most noticeable in the upper aerobic zone. Provided there is no likelihood of serious over-training, the upper aerobic part of my test colud be omitted, and in this situation, my test would yield very similar information to Vehr’s test. Vehr’s test is a bit simpler to do and could easliy be incoprtated into the warm up of every training session. In both tests, awareness of effort is crucial if serious overtraining is to be identified.

    • canute1 Says:


      I applied Vehr’s equation to some of my HR data during both very easy low aerobic running and mid aerobic running at constant HR with the following conclusions:

      1) At both paces it seriously over-estimates my VO2 max, giving values around 54 for data from very easy running and 56 from mid aerobic running, whereas true VO2 max at the time was about 43 ml/min/Kg. I suspect the discrepancy is due to the fact that I almost certainly have a larger than average cardiac stroke volume, so HR is lower than anticipated by Vehr’s equation, though maybe Vehr’s equation also requires a difference adjustment for age in the older age range.

      2) Vehr’s equation gives consistently higher values of VO2max when applied to mid aerobic running compared with easy jogging.

      3) It does show an increase in VO2 max of about 2 ml/min/Kg during a period when my fitness increased, and this is probably a reasonable though approximate estimate of the magnitude of change.

      4) The percentage change in VO2max is less than the % change in HR/pace but that is to be expected. In effect Vehr’s test, my sub-maximal running test, and a simple measure of beats/Km all assess change in HR at a similar low speed, as fitness improves and therefore all will be similarly sensitive to change in fitness.

      5) The relative advantages/disadvantages of the Vehr’s test, my test and assessment of beats/Km are in regard to sensitivity to over-reaching and to serious over-training. I think all three are similarly sensitive to over-reaching as this produces increased HR at similar pace despite continued training. I think it is likely my test will prove to be most sensitive to the distinction between reduced HR due to increased fitness and reduced HR due to serious over-training because that distinction is probably easiest to make by estimating effort at a given pace in upper aerobic zone. Both Vehr’s test and beats/Km are very easy to incorporate into warm-up for every session, whereas my full test protocol takes longer and is more complex to perform.

      I will be interested to hear of your experiences with Vehr’s test

  6. Seth Leon Says:


    Interesting and insightful comments as usual. I wouldn’t expect the Vehr formula to work well as a V02 estimate in older populations as it was calibrated on a younger group. The oldest participant was 40.

    It is also a very different measure to obtain a rough Vo2 (fitness) estimate for comparison across individuals, then it is to evaluate changes within an individual.

    As you indicate, I am not sure if an easy running measure will be sensitive enough to differentiate changes due over-reaching vs those due to over-training.

    If a HR measure during exercise is to be used to inform that days workout I think it needs to be extracted from an easy pace warm-up. This is why I would like to track the measure daily in comparison with training loads and HRV data.

    I expect there might be some U-shape relationship going on. I would expect fitness readiness on a given day to be more acutely affected negatively by the fatigue of training then by fitness improvements which accrue more gradually. It will probably take an accumulation of a good deal of data to find a meaning pattern assuming one exists.

    I had a crazy week traveling to Nor Cal, driving long hours and providing support for a family member who recently lost her spouse. My training was interrupted in what should have been a peak week (a month out from my marathon), and I lost my garmin watch leaving my backpack in SF. Anyway I did collect some data and will get back to posting weekly updates soon.

  7. Seth Leon Says:

    Above I meant to say:
    ‘As you indicate, I am not sure if an easy running measure will be sensitive enough to differentiate changes due over-reaching vs those due to gains from training.’

  8. canute1 Says:

    It is true that Vehr’s test was not normalised for age across a range extending into late middle age, though I am not sure that age should affect the results that much. Daniels formula for computing VO2 based on velocity does not depend on age. This is consistent with evidence that running efficiency does not deteriorate much with age. Similarly the conversion of % HRmax to % VO2 max does not depend on age. So the estimate of VO2 max based on pace and HR depends only on age insofar as to HRmax decreases with age. The rate of decrease in HRmax with age is similar in early middle age and late middle age.

    When I used heart rate and speed data from my submaximal test in September last year to estimate VO2max, the speed and HR recorded in the low aerobic zone gave an estimated VO2 max of 45.9, while the data for the mid-aerobic zone gave a VO2 max estimate of 45.5. My Half marathon performance a week later corresponded to a VO2 max of 44 (based on the assumption of optimum performance, but in fact I performed a little worse than expected on basis of the previous race performances, so my true VO2max was probably slightly higher than 44). Overall, I consider the VO2 max estimated from my sub-maximal test was a remarkably accurate estimate of true VO2max. On the other hand, applying Vehr’s equation to the same data yielded VO2 max values of 55.2 and 55.9 from the low and mid-aerobic data respectively.

    I am sorry to hear that you have had a stressful time recently. Nonetheless, good luck with your marathon.

  9. Seth Leon Says:

    Thanks again Canute,

    I have after reading your comments I have read the various Lamberts papers, and some others. It does seem clear that there is considerably less variation in heart-rate during exercise at higher intensities then at easy paces. Thus the Lambert recomendations regarding the use of HRR from higher intensities. I initially used the Vehrs formula on my data at quite low intensities and the Vo2 estimate seemed reasonable. As with you however looking at somewhat faster paces the formula vastly overestimates my V02. Not sure how he was able to obtain such a high R on so many subjects. In any event, I don’t think the formula will be useful in the way I had hoped.

    On the bright side I have also seen some recent separate papers lead respectively by Plews, Bucheit, & Le Muer each with data supporting the use of weekly RMSSD averages (rather than daily values) in monitoring fitness change. I found the third paper especially enlightening:

    ‘Evidence of Parasympathetic Hyperactivity in Functionally Overreached Athletes’

    This paper appears to shed light on a number of the contradictions I have been struggling with. The researchers collected HRV, HR, and HR during sub-maximal and maximal tests as well as baseline Vo2 and performance outcomes. HRV data was collected both supine and standing, and a functionally-overreaching group (competitive triathletes) was compared to a control.

    The study found that using weekly RMSSD averages they were able to link HRV findings to changes related to the overload period & a one week taper. I find it very interesting that for the functionally overeached group weekly averaged standing HRV actually increases during the overload period when performance is compromised, & then starts to decrease during a one week taper after which that group improves performance (supercompensation). This makes sense given my initial intuitions of my standing HRV scores in relation to fatigue. If you can access that paper you might want to check it out in full. The HR during exercise measures also concur with Lamberts findings as HR at the higher exercise intensities has the strongest relationships with changes across the overload and taper periods.

    • canute1 Says:

      That looks an interesting paper. I have not yet obtained access to the full manuscript . It is certainly plausible that a weekly average of RMSSD might provide more reliable estimate of autonomic state. The observation that HRV increases during the overload period is a little surprising, though I have occasionally observed something similar myself. Occasionally I have observed standing RMSSD over 180 (about double a typical value) during a period of heavy training, yet there was no other evidence suggesting serious over-training. Thus it does appear that in some circumstances parasympathetic excess is might indicate functional over-reaching rather than serious over-training.

  10. Seth Leon Says:

    Hi Canute,

    I have wondered about those very high RMSSD #’s you have mentioned in past posts. Using ithlete that never happened for me. I recall an occasional outlier using Suunto and the kubios software but nothing anywhere near that high for RMSSD?

    The trends for the functionally over-reaching triathletes in the study were pretty clear. Performance went down in the 3 week overload period as did HR during exercise at various intensities (especially higher intensities ), while HRV (standing weekly average in particular) went up and HR at rest went down. Trends reverse after 1 week taper and this over-reached group performs well after taper.

    I am now thinking that I will track one day measure of HR at relatively hi intensity after an easy or rest day, along with weekly average HRV (standing & resting) and perhaps some measure of weekly training load and perceived fatigue. The daily graphs will then just fill in details, and this should make for an interesting analysis across a training block.

  11. Seth Leon Says:

    AS you indicate, what the paper doesn’t enlighten is at what point does functional over-reach pass the threshold to over-training. Certainly seeing the over-reaching indications in a non-overload section of training periodation would be a sign to back off.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

%d bloggers like this: