Heel-striking and a brief history of the modern running shoe

Recently, Rick asked me to act as judge in a debate with a friend, who works in a store that sells running shoes, about heel-striking versus mid-foot landing.  

At first sight, it does seem rather amazing that the manufacturers of  running shoes continue to emphasize the virtues of cushioning and stabilization (to reduce pronation – the ‘natural’ tendency to roll from outside edge of the foot towards the medial edge as the longitudinal arch absorbs the energy of footfall)) five decades after Gordon Pirie trenchantly pointed out to Adi Dassler, the founder of Adidas, the reasons why the leading principle in running shoe design should be  ‘less is more’.   But the story has some interesting twists and turns.

Pirie argued that the arch of the human foot is well designed to absorb the stress of footfall provided the runner lands on the forefoot.  In chapter 3 of his book ‘Running Fast and Injury Free’ Pirie cites two observations to support his argument.  The first was the set of video recordings of 100 elite athletes at the 1972 Montreal Olympics, by Bill Toomey (winner of the decathlon gold medal in Mexico City in 1968).  According to Pirie, all 100 elite athletes filmed by Toomey were fore-foot strikers.  The second observation was more anecdotal: Pirie himself ran more recorded miles than any other human being (around 216,000 miles in 40 years) and suffered minimal injuries.  He attributes this to his forefoot running style. 

 

Zatopek and Dassler shoes

However, more recently video analyses reveal that a large number of elite and sub-elite are heel strikers.  What has changed?  I think the seeds were sown two decades before Montreal.  In Helsinki in 1952, Emil Zatopek won gold medals in the 5,000m, 10,000m and marathon, wearing Dassler shoes.  As far as I know, the shoes worn by Zatopek in Helsinki were in fact rather light-weight, though he is reputed to have trained in army boots.  However the more relevant fact is that at around that time, Dassler added the famous three stripes to Adidas shoes to stabilize the mid-foot.  As far I can see, that was the point at which engineering and marketing formed an alliance and abandoned the ‘less is more’ principle.  Fueled by Zatopek’s achievement, Adidas rapidly came to dominate the market.  Ultimately, the engineering led to more cushioned soles and marketing managers persuaded runners that cushioning and stability were crucial.  With heavy cushioning, it was no longer essential to land in a way that absorbed the energy of impact in the longitudinal arch of the foot, and eventually, many runners accepted heel striking as the norm.

In recent times, several schools of thought (most notably Pose and Chi) have resurrected Pirie’s ideas about efficient running, and there has been a resurgence of interest in minimalist shoes.  Nike, which grew from the foundation provided by Bill Bowerman’s famous waffle iron technique for fabricating a durable sole, and went on the eclipse Adidas, have recently capitalized on the minimalist trend with the Nike Frees.  Nonetheless, Nike are currently putting a lot of resources into promoting the Lunarglide, a lightweight shoe designed to combine cushioning and stability, and are targeting their marketing at female athletes.  Whatever the merits of the engineering, marketing has now made it almost impossible to draw any useful conclusions about how it is best to run from observations of elite and sub-elite athletes.

However, neither can we draw reliable conclusions from idealized accounts of ‘primitive’ tribesmen who are reported to achieve phenomenal long distance feats running barefoot or in rudimentary shoes.   Running a 10K in less than 27 minutes, or a marathon in just over two hours, are quite different from pursuing a wild animal for hour after hour across the African savanna or the North American prairies.  Drawing on arguments based on the evolution of the human foot to guide us about the most efficient way to run competitively might not be the best way to settle the question of how to run fast on road or track.

     

Short time on stance is crucial

One thing is fairly clear.  The fastest runners spend a short time on stance.  Studies by Peter Weyand and colleagues at Harvard University have demonstrated convincingly that the feature that distinguishes the fast runners from slower runners is a short time on stance (Journal of Applied Physiology, volume 89, pp 1991-1999, 2001).  A short time on stance necessarily entails a very strong push against the ground, resulting in powerful upwards propulsion, a long stride and relatively high cadence.

Schools of efficient running such as Pose also emphasize a short time on stance.  However, the theory of Pose promulgated by Dr Nicholas Romanov rather misleadingly  implies that the runner becomes airborne due to un-weighting of the foot as a result of gravitational torque, and a hamstring contraction that pulls the foot from the ground.  I believe that it is impossible to become airborne by this means.  A runner who spends 20% of the gait cycle on stance must necessarily exert an average downwards force on the ground that is 5 times body weight. 

Unfortunately, I do not know of any force-plate data that confirms that this is the case for a Pose runner.  I was a little disappointed when I attended a weekend Pose course with Dr Romanov, at Loughborough University (the home of Sport Science in the UK), and none of the Pose experts present showed any inclination to arrange a force-plate recording session.  Nonetheless, the Law of Conservation of Momentum requires that the impulse generated by ground reaction force must balance the downwards impulse generated by gravity acting on body weight, and hence the force exerted by the foot on the ground averaged over the entire gait cycle must be equal to body weight. 

A short time on stance not only ensures a powerful push against the ground, but also necessitates landing only a short distance in front of the centre of gravity, with the foot traveling backwards relative to the body’s centre of gravity at footfall.   This is most easily achieved with a forefoot or mid-foot landing.  Thus simple mechanical principles support Pirie’s argument for a forefoot landing.  However, it would be foolish to under-estimate the forces involved. 

 

Risks of minimalist shoes and forefoot landing

I was interested to note that Dallas Pose coach and stalwart of the PoseTech forum, Jack Becker, suffered a metatarsal stress fracture about two years ago.  While I have a great respect for Jack’s thoughtfulness, and I am personally grateful for advice that he once gave me regarding choice of shoes, I am inclined to think that his enthusiasm for minimalist Puma H-street shoes may have contributed to his stress fracture.  It is an interesting side-issue to note that Puma was founded Rudolf Dassler, brother of Adi – perhaps Rudolf took more note of Gordon Pirie’s opinions.  On balance, I am a little cautious about minimalist shoes, but certainly believe that cushioned heels, and heel striking, are undesirable.  It might be argued that it is better to train the intrinsic muscles of the feet to distribute the load along the arches of the foot rather than to allow these muscles to atrophy within heavily cushioned shoes.

There have been very few studies that have directly compared the benefits and risks of fore-foot, mid-foot and heel striking.  Perhaps the best known is the study by Arendse and colleagues from Tim Noakes’ laboratory in Capetown (Medicine & Science in Sports & Exercise: Volume 36,  pp 272-277, 2004).  The fore-foot landing group was instructed by Nicholas Romanov.  The main finding reported in the published paper was significantly decreased stress on the knee joint in the fore-foot runners compared with the heel-strikers.   However, forces around the ankle were noted to be higher, and Ross Tucker, who assisted Dr Romanov, reports on the Science of Sport blog that calf and Achilles problems were common in the fore-foot group.

 (http://scienceofsport.blogspot.com/2007/09/running-technique-part-ii-scientific.html )

In fact since the publication of the Arendse study, many Pose coaches have reduced the previous emphasis on a ball-of-the foot landing with marked plantar flexion of the ankle.  At least some Pose coaches acknowledge that the heel should be allowed to touch the ground lightly, to relieve the strain on the plantar fascia and Achilles tendon.

 

Conclusion

A short time on stance is essential if you want to run really fast, and this is most easily achieved with a forefoot or mid-foot landing. However the ground reaction forces are necessarily large, and landing on the ball of the foot with ankle plantar flexed places a great strain on the feet, ankles and calf muscles.  At least during long races, it is probably best to let the heel lightly touch the ground, to minimize risk of injury to the plantar fascia and Achilles tendon and perhaps even, risk of metatarsal stress fracture due to bone fatigue resulting from repetitive impact.

Heart Rate Variability maps the road to recovery

I am on the road to recovery from the debilitating illness that had incapacitated me for 4 weeks. The two charts below show the Poincare plots of R-R intervals recorded using my Polar RS800cx during 5 minutes of relaxed deep breathing while sitting, on 12th July (3 days after the resolution of symptoms) and on 18th July (9 days after resolution of symptoms).

The much greater scatter of points on 18th July demonstrates that I am now far less stressed. On 12th July mean heart rate while sitting was 60 bpm while it had decreased to 54 bpm by 18th. Even more dramatically, the overall standard deviation, which provides an indication of the overall amount of variability in heart rate, had increased from 40.1 milliseconds to 82.9 milliseconds. The standard deviation in the direction at right angles to the 45 degree line (which provides an indication of the amount of parasympathetic activity) increased from 23.8 milliseconds to 56.4 milliseconds. These numbers reveals that the variability of my heart rate had more than doubled over the six day period. The increase is due to an increase on both parasympathetic activity (which is associated with relaxation and recovery) and also an increase in variability of sympathetic activity.

I had done some easy running on 10th and 11th July, and since 12th, I have done a further 3 sessions, on each occasion running 5Km in the lower aerobic zone.

Overall, these observations are very encouraging. However, I have lost a great deal of fitness during the four weeks of illness. Today, during an easy 5Km run, I recorded 722 heart beats per Km, whereas I was recording values around 650-680 beats per Km when running at a similar pace before I became ill. Although the degree of variability is similar to that before I became ill, my average heart rate of 54 bpm when sitting today was still somewhat higher than the 46 bpm recorded prior to my illness. I think these data should be interpreted as evidence that my stress levels are back to near the pre-illness levels, but my aerobic fitness is substantially reduced. Nonetheless, the recovery of my heart rate variability suggests that I am now sufficiently recovered to resume regular training.

A profligate purchase and an interesting stress test

After much deliberation about buying a heart rate monitor capable of recording Heart Rate Variability (HRV), I have decided to be profligate, and have bought a Polar RX800cx.  I had been vacillating between a Polar RS800cx and a Suunto t6r.  For the purpose of measuring HRV, the Polar RS800cx and the Suunto t6r have very similar capability.  Both devices record R-R interval (the interval between the ventricular contraction of consecutive heart beats), and either can be used to provide the raw data required by the Firstbeat software which I have discussed on several recent postings.  This software computes a stress/ recovery index which promises to provide a sensitive indicator of impending over-training.  This stress/recovery index is based on estimate of the balance between activity of the parasympathetic nervous system (which promotes recovery) and the sympathetic nervous system (which promotes fight or flight).

When it came to the final decision about which device to buy, I was swayed by the fact that RS800cx comes with a stride sensor that appears to provide a reliable measure of cadence.  I am very interested in assessing my cadence when running, but a discussion of the reasons for my interest will have to wait for later posting as my first objective was to see how useful R-R data might be for measuring stress levels.

 

The Polar Own Optimizer

Although I have focused mainly on First beat software in recent weeks, the software that comes with the Polar RS800cx does include a utility called Own Optimizer, which assess stress levels on wakening in the morning.  Unfortunately, the information provided in the manual provides little scientific justification for the Own Optimizer.  As far as I can gather, Own Optimizer is based largely on the changes in both heart rate and HRV in response to rising from sitting to standing.  In principle such measurements might provide a sensitive measure of the degree of withdrawal of parasympathetic nervous activity and increase in sympathetic activity associated with the mild challenge of standing-up. Therefore, I will be curious to experiment with Own Optimizer, but the output of Own Optimizer is difficult to interpret unless one has baseline data recorded while in a relaxed state. There is little point in me trying out Own Optimizer until I am fully recovered from my recent episode of illness.  Nonetheless, I hope eventually to compare Own Optimizer with the Firstbeat stress/recovery index.

 

Back to the Poincare scatter plot

Meanwhile, I exported the R-R data from my new RS800cx and subjected it to the same Poincare analysis which I had presented in my post on 26^th June.  The Poincare analysis requires the production of a scatter plot which can be produced easliy using software such as Excel.  Interpreting  the scatter plot taxes the grey cells, but the effort is probably worthwhile.

The Poincare analysis is based on a scatter plot in which each heart beat is plotted on the x-y plane with the x-coordinate equal to the interval between the preceding beat and the beat of interest, while y coordinate is equal to the interval between the beat of interest and the following beat.   A heart beat for which the preceding inter-beat interval is equal to the subsequent inter-beat interval must lie on a line that represents the equation x=y.  This line slopes upwards and to the right at 45 degrees, as shown in the figure below.  Strong parasympathetic activity is associated with large beat by beat variability in heart rate, so consecutive interbeat intervals will differ substantially in duration and the  points representing the heart beats  will lie at a substantial distance for the 45 degree line. 

Conversely, if inter-beat interval varies slowly (over a time scale long compared with the average inter-beat interval) then each heart beat will be represented by a point near the 45 degree line, though the slow variation will cause the location of the points representing the heart beats to wander along the 45 degree line.  Slow fluctuations are largely due to sympathetic nervous activity, though there is evidence that parasympathetic activity can also contribute to slow variations – however for the present purpose, let us assume that sympathetic nervous activity is mainly responsible for slow variation, resulting in a spread of points along the 45 degree line, while parasympathetic activity is responsible for rapid variation that causes a spread of points away from the 45 degree line.

 As I described on 26^th June, if there is a good balance between sympathetic and parasympathetic activity, the scatter plot should produce a cluster of points that looks like a swarm of bees the spreads out both along the 45 degree line and away from the 45 degree line. If we draw an ellipse that captures most of the points, this ellipse will be almost circular.  Conversely, if the person is stressed, so that there is an excess of sympathetic activity, the scatter plot will produce a cigar shaped cluster extending along the 45 degree line.  In the data which I had recorded using my ‘home-made’ ECG device before my recent illness, the scatter plot had exhibited a large spread away from the 45 degree line in addition to a large spread along the line.  The ellipse that embraced most of the points was pleasingly round.

 Here is a comparison of the scatter-plot for data recorded today (using my new RS800cx)  with data recorded before my recent illness.  Both sets of data were recorded while sitting in a relaxed state, breathing slowly and deeply.  The pre-illness recording was in fact made at around 6pm at the end of a day at work, while today’s recording was made in the early afternoon after a relaxing Sunday morning.  Thus, if all other circumstances were the same, today’s plot might have been expected to show an even more favorable balance between parasympathetic activity and sympathetic activity.  However, all other circumstances were not the same.  Today, I am in the fourth day of convalescence after a peculiar and debilitating illness that lasted almost 4 weeks.

 

 

Poincare scatter plot pre- and post illness

 

There are two striking differences.  First, the cluster of pink dots presenting today’s data has moved down and to the left, indicating that the average inter-beat interval was shorter.  A shorter inter-beat interval corresponds to a faster heart rate.  Before my illness, my average heart rate was around 46 beats per minute.  Today’s value was 59 beats per minute. Secondly, the ellipse that embraces the majority of the points is thinner – somewhat more like a cigar.  Today’s data reveals substantially reduced heart rate variability, especially a reduction of the parasympathetic activity that produces rapid changes in heart rate and scatters the points far from the 45 degree line.

 

Occasional parasympathetic surges

There are a few points that are outside the ellipse that embraces the general trend.  Several of these points represent heart beats for which a preceding short inter-beat interval was followed by a longer than usual inter-beat interval (causing a large displacement above the 45 degree line).  In fact these points represent a sudden drop in heart rate from around 58-62 beats per minute to 52-55 beats per minute. It appears that I was experiencing occasional surges of parasympathetic activity.  These surges were also apparent when I examined an even longer recording, so I suspect that they are not just random chance events but in fact represent some fairly consistent pattern of autonomic nervous activity.  I do not have an explanation for those occasional sudden surges of what appears to be parasympathetic activity.

 

The main conclusion

However, the main conclusion is very clear.  My recent illness has left me in a quite stressed state – at least compared to my relaxed pre-illness condition.  In fact, although today’s recording reveals a marked deterioration, it is not too bad for a 63 year old, so there is no reason for me to be too alarmed.   However, it would probably be unwise for me to resume vigorous training until the scatter plot of data recorded in a resting state returns to something more like the widely dispersed shape exhibited in my pre-illness data.

The apparent sensitivity of the Poincare scatter-plot to stress level makes me wonder how useful if might prove to be as a measure of training stress.  Once I have recovered fully, it will be interesting to compare the scatter plot following a hard training session (‘over-reaching’) compared with that following easier training sessions.  Maybe the Poincare scatter plot of data record during relaxed deep breathing will be as informative as either Polar’s Own Optimizer or the stress/recovery index computed by Firstbeat software.  Whatever the relative merits of the different ways of assessing autonomic imbalance from HRV data, it appears that HRV is a sensitive indicator of one’s internal milieu.

Snakes and ladders again

The good news is that I have recovered from the peculiar illness that has troubled me for almost four weeks.  During the past three days my symptoms have been diminishing rapidly with each passing day.  Two days ago, I did a gentle Pilates session and yesterday I went for a short, easy cycle ride, without adverse effects.  By this evening the only remaining traces of the illness were mild constriction of my upper airways (peak flow 270 litres/min compared with my usual 520-550 litres/min) and very mild diffuse musculo-skeletal aches.  I decided it was time to return to running.  Before I set out I tested my heart rate versus power on the elliptical cross trainer.  To avoid stressing myself, I did not go beyond 200 watts.  As you can see from the chart, there has been a fairly dramatic deterioration in my fitness.  Heart rate today about 11 beats higher at each level of power output, compared with 6^th June.

 

I then went for an easy 3 Km run in the woods.  After a warm up at gradually increasing pace, I timed myself over the middle Km, which I ran at a comfortable pace in the mid-aerobic zone.  I was pleased to find that this comfortable pace was 5:05 per Km and average heart rate was 133.  I then slackened the pace to a jog for the final Km.  Although I felt tired at the end, it was great to be out running in the woods again. 

The chart of heart rate v power provides graphic evidence of how rapidly fitness is lost, but in view of the severity of my illness, I was expecting at least this much deterioration, so I was neither surprised nor disappointed.  The interesting question is how quickly can I recover to my level of fitness in early June.  It took 4 weeks to lose this much fitness.  I hope I can recover to my pre-illness level by the end of summer, but  it would probably be counter-productive to push myself too hard.  I anticipate that I am going to need a few weeks of convalescence.

Getting the balance right

I have been blogging a little more frequently in the past week or so because I have been ill, and therefore not running.  I have been exploring issues related to over-training and heart rate variability on my blog because one possibility is that I had become ill because of over training.  At this stage I am puzzled.

A peculiar illness

First an outline of the illness: it started over three weeks ago with an exacerbation of my long standing inflammatory arthritis; then became an acute fever with a temperature of  101-104 degrees F. for a few days; then what appeared to be chicken pox with  fairly typical skin vesicles, and  most distressingly, severe mouth ulceration. I have largely been living on ice-cream and cool fluids for over a week.  This morning, I woke at 3:30 am with a new crop of painful vesicles in my mouth, a painful throat and a mild asthma attack.  At this stage I am a little worried that the problem will extend more deeply into my lungs, though as I sit at my desk typing this I do not feel seriously ill.  Maybe it’s just chicken pox and a few incidental problems, though since I had chicken pox as a child, it’s all a bit mysterious.

Could this peculiar illness be a consequence of over-training?  At this stage I think it is very unlikely.  As I have been discussing in recent postings, there is good evidence for regarding the balance between sympathetic and parasympathetic nervous activity as a useful measure of recovery from training.  As I posted last week, my heart rate variability before I became ill indicated a good balance between parasympathetic and sympathetic activity. Since I have been ill, I have not recorded heart rate variability but simple tests such as the orthostatic test (change in heart rate on standing) indicate there has been a small shift from parasympathetic to sympathetic activity.  This is only to be expected with the degree of illness I have experienced.  Nonetheless, I can still get my heart rate down to the low 50’s by relaxed rhythmic breathing, so the evidence suggests I still have reasonably good parasympathetic control. If maintaining a good balance between parasympathetic and sympathetic activity is a sign of good recovery from training, I have been recovering well.

Could my illness be due to immune suppression produced by the steroid inhaler I use for my asthma?  My doctor thinks that is very unlikely. So the situation remains a mystery,   At this stage I do not know when I will get back to running.

 

The overall balance sheet

Though at the moment I am not well, it is also important to keep in perspective the overall balance sheet with regard to my health since I recommenced running.  Let’s start with the negative side of the balance sheet.  The one definite deterioration has been in my asthma.  Although I have suffered mild asthma since childhood, I had never needed treatment until a year or so ago.  It is possible that getting cold air into my lungs when running has exacerbated that problem. My arthritis presents a different story.  It had also been a mild problem since childhood but had started to become more of a problem as I approached middle age. In particular my right knee and the metatarsophalangeal joints in my feet were starting to be troublesome.  But contrary to expectations, those problems have greatly improved since I started running. The recent flare-up of arthritis was I fact very minor.  Most importantly for my general health, I think the decrease in my resting heart rate, from around 60 a few years ago down to the mid 40’s, is an indication that my heart is much healthier as a result of running.  So I think the balance sheet is positive.

 

Maintaining the balance

Meanwhile, I still continue to be fascinated by the question of how best to monitor training so as to maximize both my health and my running performance.  Even though there is little to suggest recent over-training, my experience of the past couple of years has demonstrated that I am now less able to cope with heavy training than forty years ago.  Maybe that is an inevitable consequence of aging.  But if I accept that, it becomes all the more important to develop good strategies for optimizing training level.

My overall conclusion is that training vigorously is almost certainly the best way to remain healthy into old age, but finding a good way to judge just how vigorously to train is the challenge.   I am also inclined to think that for an athlete of any age, the challenge is similar.  Finding the optimum balance between stress and recovery is likely to be the recipe not only for achieving for good general health but also for maintaining the consistency of training necessary to achieve one’s peak performance.  I just hope I can get back to running soon, though I might have to wait a year or two to achieve my M60 peak performances.

Do Firstbeat offer more in 2009 than Forbes and Ursula Carlile in 1959?

In recent postings I have been exploring the possibility that measurements of Heart rate Variability (HRV) might provide a useful way of detecting over-training and of adjusting training load to achieve optimum outcome. In response to a query from Ewen, on 30^th June I had looked at the question of which commercially available heart rate monitor might provide the most useful measurement of HRV.  I had attempted to compare the merits of the products offered by Polar and by Suunto.  In fact, this was a frustrating task because the material presented on the websites of the two companies does not provide enough details of the principles underlying their procedures for using HRV to monitor training load and over-training, nor adequate evidence that using their devices leads to improved outcomes of training.

However I had concluded that the most promising current approach uses software developed by the Finnish company Firstbeat Technologies.  Firstbeat software can read the data from either the Suunto t6 or the Polar RS800. Suunto have incorporated Firstbeat software into their own Training Manager software and have committed themselves more heavily to the utilization of Firstbeat software.  Some  of the measurments such as Training Effect (see below) can be read directly from Suunto HRMs  such as the t3,  t4 and t6r, during the training session.  Nonetheless, the primary input required for Firstbeat software is a record of inter-beat interval in a series of heart beats recorded under whatever circumstances one is interested in, whether than be during rest or exercise, and this data can be provided by either the Suunto t6 or Polar RS800, though the full analysis cannot be performed until after the training session is over. 

Therefore in deciding between Suunto and Polar for the purpose of measuring HRV, the major issues is likely to be reliability of the recorded data, the size of the data store and the ease with which it can be read by Firstbeat software.  I have not looked into any of these questions, though issues such as susceptibility to interference arising from power lines or nearby HRM’s worn by other athletes are addressed in a comparison of Suunto and Polar devices by PC Coach

(http://www.pccoach.com/newsletters/Nov05/speedist5.htm ) 

Note added 5 July 2009:   If you are interested in comparing other practical aspects of the Polar PS800CX and the Suunto t6r, such as convenience for use during a triathlon, or the utility of the software provided by Suunto and Polar for planning of your training sessions, Jan Musil provides an excellent comparison at:

http://runtotri.blogspot.com/2009/01/polar-rs800cx-or-suunto-t6c-that-is.html

As far as I can see, despite differences in detail, both companies provide technically sound equipment.    However my present interest is in the Firstbeat software. 

KIHU and Firstbeat

Firstbeat is a spin-off company created by members of KIHU – Research Institute for Olympic Sports, located in Jyväskylä, Finland.  KIHU researchers have conducted a number of very informative studies of HRV over the past decade.  The senior investigator in many of these investigations is Heikki Rusko, a well known exercise physiologist.  By examination of both the scientific publications produced by KIHU and by reading the material presented on the Firstbeat technologies website, it is possible to get a reasonable understanding of what the Firstbeat software has to offer, though not quite as clear a picture as I would like to have.

 

In my opinion the peer reviewed publications from KIHU provide only moderate, but nonetheless tantalizing, support for the proposal that HRV and related measures would provide a reliable estimate of training stress.  The material presented on the website provides more information about what computations the Firstbeat software performs, but many details are missing, and the quality of some of the crucial scientific evidence falls below that I would expect in a peer reviewed scientific publication.  Nonetheless, some of the material on the Firstbeat website (especially the downloadable white papers) appears to have been written by exercise physiologists rather than marketing personnel.  Maybe this makes it harder to read but ultimately, more worthwhile. So in this posting I will present a personal overview of what I regard as the most relevant outputs from the Firstbeat software

The three outputs most relevant to the scientific assessment of athletic training are:

EPOC: an estimate of oxygen debt acquired during a training session.  This provides a measure of the stress on the cardio-respiratory system resulting from the session;

Training Effect: an estimate of the potential benefit (or in some instances, degree of  overreaching)  from a training session;

Recovery index:  a measure of autonomic nervous system balance that is potentially a useful indicator of over-training.

It should be noted that Firstbeat produce three main software packages, each of which is specialized for different users:

Firstbeat ATHLETE (FBA), which calculates EPOC, Training effect and provides guidance on planning a training program.  However FBA does not provide an analysis of recovery.  While it does provide guidance that should minimize the risk of over or under-training, the planning is based on the estimated Training Effect of recent sessions rather than on a direct measurement of the degree of recovery immediately prior to the next session.   

Firstbeat SPORTS, which calculates EPOC, training effect and also provides a detailed analysis of stress and recovery.  This includes a recovery analysis based on overnight recording, and charts of stress and recovery throughout the day.  This analysis takes account of cumulative stress not only of recent training sessions, but also of other life events. This software is designed for sports professionals

Firstbeat HEALTH, which provides detailed stress and recovery analysis. It is designed for use in studies of occupational health and therefore is not oriented towards the management of an athlete’s training program.

EPOC 

It is well known that exercise creates an oxygen debt such that oxygen uptake over a period of minutes or hours after cessation of the exercise is increased compared with baseline.  This is known as Excess Post Exercise Oxygen consumption (EPOC).  This can be measured directly in a laboratory using equipment to measure respiration.  KIHU researchers report that the oxygen debt can be predicted reliably from heart rate and respiration rate (calculated from HRV) measured during the exercise.   The EPOC measurement produced by the Firstbeat software is the predicted EPOC based on heart rate recording during exercise.  As oxygen debt only accumulates when exercise is of at least moderate intensity, EPOC generally increases with time during moderate or heavy exercise but tends to decrease with time during periods of light exercise following heavier exercise.

How good is the evidence that EPOC is a good estimate of the total accumulated stress applied to the cardiovascular system?  Firstbeat show that EPOC is strongly correlated with lactate accumulation, which is quite plausible.  However, at low workloads, such as the levels proposed by Philip Maffetone during the base-building phase, lactate accumulation is minimal, and therefore one might expect that EPOC will be small, yet some training effect on the heart would be anticipated. 

The data provided by Firstbeat in their white paper on EPOC demonstrate that at a workload corresponding to 40% of VO2max, EPOC scarcely rises at all after the first 30 minutes.  Does this indicate that long slow runs do not produce stress (and hence useful training effects) on the cardio-respiratory system?   As far as I can see this issue remains unresolved,  and I would like to see more evidence.

Furthermore the calculation of predicted EPOC requires a prior estimate of HRmax and if this is inaccurate, the calculation of EPOC will be inaccurate.

Training Effect

This is an estimate of the training benefit derived from a given training session and is computed from the peak EPOC achieved during the session  taking account of the athlete’s current activity level.  Activity level is scored on a 10 point scale largely based on current weekly training load.  The underlying principle is that an athlete with a high current activity level will require a higher value of EPOC to achieve a comparable training benefit compared with a less active athlete.   This is perfectly reasonable, but on account of the crudeness of the estimate of current fitness provided by the activity level scale, I suspect that the Training Effect value is imprecise.  Training Effect itself is quantified on a scale from 1 to 5 where 1 indicates a minor training effect and 5 or more indicates over-reaching. 

A good coach (or even a thoughtful athlete) might perhaps be able to estimate the value of a training session just as well on the basis of experience, but many of us are not wise judges of how hard we are training, and we do not all have access to a wise and experienced coach.  In principle, keeping a log of Training Effect values for the week’s sessions would probably be a more sensible guide to how effectively we are training that keeping a log of total miles or Km run  each week, yet many of us are inclined to draw psychological support from our weekly mileage total. 

However, just as a log of weekly mileage has its limitations as a measure of training, so does a log of Training Effect.  Not only is it likely to be an imprecise estimate of effects on the heart,  but in addition, Training Effect does not take adequate account of other organs of the body, especially the musculo-skeletal system.  It is highly likely that long slow distance training strengthens bones, tendons and ligaments (and conversely that running excessively long distances creates risk of musculo-skeletal injury) yet the computation of Training Effect appears to under-estimate either the benefits or risks of long slow distance running.  Furthermore, the calculation of Training Effect does not take account of the benefits derived from strength or flexibility exercises.   So it might be a little more useful that a log of weekly mileage, but it does not reflect all of the important benefits (and risks) from a  training program, and it would be foolish to plan a program guided only by measurments of Training Effect.

Recovery Analysis

The potentially most useful analysis provided by the Firstbeat software is the recovery analysis based on an overnight estimate of stress and recovery.  The measurement is based on recorded heart beat during a four hour period of sleep.  A quantity known as the recovery index is computed from heart rate, heart rate variability (HRV) and estimated breathing rate derived from HRV. 

Firstbeat do not say exactly how this computation is performed, but a white paper presented on the website states that the procedure  was derived by fitting a mathematical model to a large amount of data collected in many studies.  In this context, the mathematical model is an equation that estimates the balance between sympathetic and parasympathetic nervous activity using physiological information such as HRV.   The principle of deriving an equation that predicts a physiological variable of interest from the values of related physiological variables is well established in exercise physiology.  For example Daniels’ famous equation for predicting VO2 from speed and duration of a run is a quadratic equation that was derived by determining the coefficients of the quadratic equation that gave the best fit to the observed data in a large number of individuals.  It is likely that Firstbeat have used a different type of equation, based on more complex mathematics, for the prediction, but the principle of fitting observed data to a mathematical model is likely to be similar. 

Potentially the biggest limitation of this type of approach is that even if the prediction works well for the average value in a group, it might not be accurate for the individual.  This issue is most clearly illustrated by the linear equations that have been proposed to predict HRmax from age. In some instances, the prediction is quite inaccurate.   The relationship between HRmax and age is a notorious example of an unreliable mathematical model of physiological data.  Nonetheless, I would like to see more data demonstrating the reliability of the model used by Firstbeat to estimate the recovery index.

Firstbeat quite rightly point out that the absolute value of the  recovery index is not very meaningful.  What is required is a measure of change from baseline within the individual.  However they do not present clear evidence for the consistency of changes from baseline within individuals who develop over-training syndrome.   The most relevant supportive data comes for a study by Hynynen and colleagues comparing over-trained with non over-trained  athletes (  Hynynen, E., Uusitalo, A., Konttinen, N. & Rusko H. (2006). Heart rate variability during night sleep and after awakening in overtrained athletes. Medicine and Science in Sports and Exercise 38(2): 313-317).

However the data relevant to the recovery index was not included in the published peer reviewed paper reporting the study but is only available in a white paper on the Firstbeat website.   Ironically, the published article actually concluded that overnight HRV did not distinguish between the over-trained and non-over-trained athletes, whereas measurement of change in HRV on rising did.   In contrast, the table of data presented in the Recovery white paper on the Firstbeat website does indicate that the stress/recovery index computed from overnight values by the Firstbeat software distinguished between the well recovered and poorly recovered state in 7 athletes.  Unfortunately, the white paper provides no indication of how this data was selected. 

I am intrigued by the possible utility of the recovery index and would be very interested to try this out myself.  However for the time being I will persist with my own amateur system described in my post on 26th June.  My system provides me with the ability to study the shape of the ECG  T wave as well as HRV.  In my analysis, the most informative estimate  of balance between sympathetic and parasymatheic activity is provided by the Poincare analysis of HRV.  This analysis assesses the ratio of high frequency variation (presumed closely associated with recuperative parasympathetic activity) to low frequency variation (predominantly determined by sympathetic activity)  by comparing the length of the two axes of the ellipse which I presented in my posting on 26th June.  Unfortunately, the interpretation of the Poincare analysis is not quite as simple as described in my posting on 26th June.  It is possible that the computations done by Firstbeat software are more reliable.  On the other hand, the evidence presented by Firstbeat is scarcely any more convincing than the data on ECG T waves presented by Forbes and Ursula Carlile to the Australian  Sports Medicine Association  in 1959.  As I described in my post on 30^th June, the report by Forbes and Ursula Carlile demonstrated that in selected cases, flattening of the T waves corresponds very closely to deterioration in performance due to over-training.  The presentation of data on individual subjects or small groups of subjects can look very impressive, but what I would like to see is evidence showing how well the Firstbeat procedure works for an unselected sample of athletes.

Conclusion

The available evidence does suggest that HRV measurements can provide a useful assessment the quality of training and might detect over-training.  I think that for any athlete who can afford the cost, and is prepared to interpret the data thoughtfully, a Suunto t6 with Firstbeat software (or maybe a Polar RS800 with Firstbeat software) would be a worthwhile investment.  However I am disappointed that half a century after the thought provoking presentation by Forbes and Ursula Carlile to the Australian Sports Medicine Association in 1959, it is still difficult to find publicly accessible data that would allow an objective evaluation of the reliability of measurements of autonomic nervous system function for the purpose of detecting over-training.