What is a training load?

Training: pushing adaptations in limiting systems Most runners training for a marathon follow some form of standard plan containing a mix of running speeds and durations with a gradual build towards race day. The plans are constructed to induce adaptations within different systems that might limit marathon performance. These plans also contain elements that enable further training to occur - i.e. the concept of training to train. Whilst most runners are aware of the standard adaptations that are required: muscle mass, glycogen content, mitochondrial density and enzyme expression, capillary density, 'fibre-type', circulatory capacity (stroke volume etc), tendon and bone strength, mental capability, 'endurance' there is often little understanding as to just how specific or non-specific the types of training are in terms of the adaptations they produce. The result is that there is sometimes an impression that particular types of training can produce disproportionately large benefits. There are three of those types that have developed cult followings within the community: tempo running, hills/hard intervals and the long run. It is worth - very quickly - looking at some of the evidence supporting a benefit of one type compared to another.

High Intensity Interval Training (HIT)

Having just returned from Mara Yamauchi's marathon talk, where she mentioned all-out 10s hill efforts, I was reminded of a paper I had read some years ago by Little et al. (2010) which will do for this example. Their test was of a milder form of HIT training. In the paper they examined the effect of about ten one minute high, but not maximal, efforts with just over a minutes recovery between them every other day. They found similar adaptations using their 'practical model of low-volume high intensity training' as are found with all-out HIT. Most studies on HIT generally state that the adaptations in skeletal muscle that occur resemble what is obtained by long or endurance running. Thus, there is evidence - at least from muscle - that a continuum of training loads exists that can produce similar adaptations. All-out HIT, mild-HIT and long runs have 'much' the same effect on skeletal muscle (if the exercise program can be tolerated) but consuming different amounts of time. HIT is without doubt time efficient. However, there are some major differences between the two extremes. First HIT is associated with remarkably little total energy consumption - and almost all of it will be in the form of carbohydrates stored in the muscle. Long runs consume a lot of energy and a large proportion comes from fat. Together with the energy difference comes a large difference in thermal load and the consequential effects on the training of the skin and osmoregulatory systems. So, whilst skeletal muscle can be trained over a wide range of intensities, other systems require more specific loads to force adaptations.

Characteristics of a training load

FITT is an oft used acronym for describing a training load: frequency, intensity, time and type. Mara used the abbreviated acronym FIT in her talk. These characteristics have different units (e.g. frequency might be, per day; intensity could be characterised as grade-adjusted pace, time is simply time, and type might be represented by muscle groups). What is hard to achieve is an understanding of how one of these domains interacts with another with regards to the likely effects on performance. Frequency and time might be conflated if one simply assumes that it is the total amount of activity that counts. So session frequency multiplied by the session time might represent a measure of training benefit. However, intensity is more difficult to combine with time. The efficacy of HIT suggests a non-linear or power relationship with very high intensities being worth considerably more than the simple multiplication of speed and distance would suggest. Indeed, athletes generally feel that this is the case - the hard work associated with fast running instinctively feels more worthwhile than a longer slow run.

The consensus view on the benefits of specific training types

There are two Midgley papers, that I came across some time ago, that suggest there is little evidence for believing that there are specific intensity or distance domains that can produce disproportionately large performance benefits. The 2006 review considered maximal oxygen uptake. The abstract is relatively informative and tends to the conclusion that there are no sufficiently well controlled studies to differentiate between exercising at 70-80% or 95-100% of maximal aerobic effort. Importantly the review acknowledges that whilst exercising at maximal aerobic effort is effective, lower intensity training can be equally effective  - at least in moderately trained individuals - because it can be done for longer. For the well trained runner higher intensity work may be necessary - but the evidence supporting that is limited. The 2007 review considers race performance and the advice that can be given to coaches and runners. It contains - in the abstract - the following statement; "Scientists should be cautious when giving training recommendations to runners and coaches based on the limited available scientific knowledge." I agree with this - our knowledge is indeed limited and there is a tendency within the literature for Sport Scientists to attempt to apply a particular set of training loads to cohorts. The idea is simple and obvious; Is intensity A better than intensity B? But, the experiment is fraught with all of the problems associated with human nature, the placebo effect, sub-conscious corrective actions and intensity matching. For instance imagine that intensity A is better tolerated than intensity B but less effective at producing performance benefits. If the tolerance is ignored one might conclude the intensity B is better. But, if the subjects had been allowed to do more of intensity A - since it was easier - then it is possible that it may have been equally or even more effective. The devil really does lie in the detail. 

It seems rather better to study athletes who are attempting to become fast and to extract training characteristics that are associated with enhanced performance. One of the problems with doing this is making sense of the diverse dataset. Each athlete will tend to perform a different mix of FITT sessions. However, if one can begin to draw some form of general relationship between the FITT sessions and performance it might well be possible to identify the most important training elements. Whilst the benefit of doing this is that one can develop a 'recipe' for success, it does not necessarily reveal all of the different ways that an athlete might be able to train to become successful. Furthermore, it is likely that there will be a degree of 'noise' within such an analysis since the number of individuals required per parameter included within the model rises rapidly if one is to avoid what is known as over-fitting.

Fortunately, there have been many studies that have attempted to predict both VO₂max and race performance from both anthropometric (i.e. body characteristics) and training-based indices. I will consider one of those next.

Next: Predicting marathon performance from training data