Polarized Training and Athletic Performance

A Hill Climb

Photo from Tavo Mx on Pinterest

[Part 2 of 3 in a series on whether I should give sprint workouts another try.]

Is Polarized Training Better for Improving Athletic Performance?

In my last post I wrote that I had seen no evidence that I would be either more or less healthy if I substituted sprint workouts for some of my moderate intensity cycling. Thus, from a health perspective, I should do sprint workouts or moderate intensity rides entirely determined by what I prefer. Although health is my most important reason for cycling, having fun is also a very important reason and I think I would have more fun if I could ride a bit faster because that would allow me to do more rides with my friends who, as it happens, are faster than I am. Might sprint workouts make me faster?

Polarized training is training done mostly at very low or very high intensity and minimizes riding at moderate intensity. In the context of this series of posts, a polarized training advocate would argue that I should replace some of my riding at moderate intensity with a mixture of sprint workouts and rides even easier than my moderate rides. Vermeire et al., Journal of Strength and Conditioning Research: September 06, 2019 is a paper about the effectiveness of polarized training at improving athletic performance. I am not an expert on exercise physiology so I don't know the names of the reputable journals in the field, but I do recognize the name of the publisher of this journal, Wolters Kluwer, allowing me to tentatively conclude this is a legitimate scientific publication. The journal's claim to be the Official Journal of the National Strength and Conditioning Association suggests it may be much better than that. On the other hand, a big barrier to my reviewing this paper is that to see the full paper I would have had to purchase the article, all I can see without paying is the abstract. Fortunately, there are things we can confidently learn from the abstract. This study looks as 11 recreational cyclists, not a large number. According to the Bicycling article about this paper, these cyclists "were just beginning spring training. All the riders were preparing for a hill climb in the Alps or Pyrenees." At the beginning of the study, athletes were tested for power output at aerobic threshold (a level of effort than can be maintained for many hours), power output at anaerobic threshold (a level of effort than can be maintained for 30 to 60 minutes) and the maximum power that can be generated going all out, an intensity that can be maintained for less than a minute. These athletes then trained on their own for 12 weeks following whatever training plan they liked, but during that training, the amount of time they spent exercising and their level of effort was recorded. At the end of the 12 weeks, the total training load (a combination of time and intensity) was calculated using several versions of the widely used TRIMP method and, in addition, the percent time they spent:
1) below the aerobic threshold,
2) between the aerobic and anaerobic threshold, and
3) above the aerobic threshold
...were determined. The power tests were repeated, and statistics used to determine if there were any correlations between how the cyclists trained and how much their power measures increased.

The first thing I note about this study is that it is both an observational study and a biomarker study*, and thus, as detailed below, has the limitations of both. The second thing I note is that the questions they chose to ask are not the ones I would choose. The first question they ask is if more training load (as determined by one of the TRIMP calculations) is better. Given the assertion that 65% of competitive cyclists train too hard`, I would assume not and don't consider this a particularly interesting question. The second question would, in principle, be more interesting, and that is does the percent of time spent training at different intensity levels matter? This is the polarized training question: is it better to train mostly at high and low intensity (polarized training), or mostly at moderate intensity? Because this is an observational study, that is, they allowed the athletes to train however they liked, their study contains two barriers to answering this question. First, the range of different training programs chosen by the subjects of the study might not cover the range of exercise plans that ought to be considered. What if none of these 11 particular athletes train in a particularly effective way? The second problem, common to observational studies, is that it is impossible to distinguish between cause and effect. Are the best athletes best because of how they train, or do they chose to train the way they do because of their natural athletic ability? It is very possible that athletes who are more gifted find intense training more enjoyable, but that the reason they perform better is not because of this intense training but due to their natural ability. Finally, there is the issue that this is a biomarker study, and more generally, that an exercise plan can only be judged with reference to its goal. The best training plan to win the Tour de France will be very different than the best training plan to win a gold metal in the Olympic 1000 meter individual sprint.

These reservations aside, what were the results of this study? Taking the results the authors reported in the abstract at face value, none of the TRIMP scores correlated with improvement in any of their three power measurements. However, when they fit an equation containing percent of training done at the three intensity levels to improvement in their power measurements, the power at anaerobic threshold was weakly but positively correlated (more is better) with training below aerobic threshold (easy training), was inversely correlated (less is better) with training between aerobic and anaerobic threshold (moderate intensity training) and strongly correlated (more is better) with training above anaerobic threshold (high intensity training, e.g. sprint workouts.) There were no correlations observed with power at aerobic threshold or maximum power. Thus, overall conclusion of this paper is that polarized training is better than moderate intensity training for increasing power at anaerobic threshold. At best, this is a conclusion limited to one parameter, power at anaerobic threshold, and it is left up to the reader to determine the application of that to their real world goals. Unfortunately, the other explanation for their data is that the best athletes who improve most from training prefer polarized training, that it is their ability, not how they train, that is responsible for their better results. Finally, this paper only includes 11 participants and they were free to choose their own training plans so may not have provided the best set of different amounts of high intensity training. That said, I liked the experimental design of looking at the correlation between amount of time spent at different intensities to various measures of fitness. I would be very interested in seeing a similar study done with many more participants and with a more structured set of training plans.

Does power at anaerobic threshold matter to me? Interestingly, after having given this considerable thought, I honestly don't know. I can't ride at anaerobic threshold for much more than 30 minutes and the rides I want to be faster on are metric centuries (62 miles long) and take me more like 300 minutes to complete so it might seem that power at anaerobic threshold would not be all that important. However, these centuries are not a single uniform effort. Rather, their effort varies over the course of the ride. My goal is to improve my ability to stay with my faster friends and maybe if I had more power for short bursts at anaerobic threshold I could stay with them under situations where I am now getting dropped, situations like heading up a hill or when the pace temporarily picks up, but who knows, and that's the problem with biomarker studies. Even if I were persuaded that polarized training would increase my power at anaerobic threshold (the biomarker) there is no way to know for sure if my ability to stay with my friends on a metric century ride will be improved.

I don't want to be negative about this paper, its limitations are due to the difficulties of doing exercise research, difficulties about which I have previously blogged. Sure, its results are far from definitive, but its experimental design is a clever response to the difficulties of these kinds of studies and I think that fairness demands recognizing that it is one more brick in the wall supporting polarized training, albeit a small one. Rather than criticizing this paper I would rather think about what it would take to get past the difficulties that plague this and other studies of exercise, and that is what I will do in the third and final installment in this series. Stay tuned.

---

* An observational study is one in which participants are not randomly assigned to different protocols (e.g. training plans) but rather chose their own training plan. They are observed, and those who chose one kind of plan are compared to those who chose a different kind of plan. The problem with observational studies is that the two groups are not "matched", those who chose one plan might be different on average than those to chose the other, affecting the results. At first glance this might seem like nit picking, but it is not, especially in exercise studies. People who are naturally more athletic will find it easier and more comfortable to exercise harder and will be overrepresented in the hard exercise groups. How much of their better performance is due to their harder training, and how much their athletic ability? On the other hand, in many cases an observational study is sometimes the only study one can do, an interventional study, where participants are randomly assigned to different protocols is just not possible for any of a variety of reasons. Observational studies ought not be dismissed, but they should be treated with more caution than an interventional study. 

   A biomarker study is one where what is measured is not the thing in which one is interested (e.g. the ability to race up a hill) but something easier to measure that is believed to be related, (e.g. power at anaerobic threshold.) The problem with biomarker studies is that the biomarker might not, in fact, be related to the thing in which one is interested.

` From an article by Coach John Hughes: "My friend and fellow cycling coach Neal Henderson says that 65% of the athletes he sees train too much, 25% train too little and 10% get it right – the pros who are paid to perform."

^ Quote by Thumper, in the Disney film "Bambi".