Saturday, February 22, 2020

Reconsidering Sprints



[Part 1 of 3 in a series on whether I should give sprint workouts another try.]

I started looking at published training plans in about 2012, and from the beginning, I have been skeptical about the value of  sprint workouts (aka brisk training rides aka interval training aka high intensity interval training aka HIIT.) To understand why, consider one example from my recent past. At the end of last season (November 2019) my cycling ability was as good as it has been since I moved to California two and a half years ago. During the 8 to 10 months prior to that I did no sprint workouts. Just before that, about 9 to 10 months ago, my ability was as low as it had been since the move, and it was my impression that my low ability was the result of cumulative fatigue caused by the sprint workouts I had been doing back then. Why, given my many similar experiences, do I still think about giving sprint workouts one more try? Because the exercise community (e.g. the coach-authors*, including the one coach-author I still follow, Coach John Hughes) keep pushing them. But should I be ignoring this advice, as some of my readers have suggest, because it is meant for young racers, not for old fitness and recreation riders like me? Hughes is my favorite coach-author because he focuses on a wider range of cyclists than most, including older and less competitive cyclists like me. Recently, he published two articles (this one and that one) on roadbikerider about the value of interval training for two of his clients, 76 and 62 years old. To be fair, these are stronger. more competitive cyclists than I, but in his book "Anti-Aging: 12 Ways You Can Slow the Aging Process" Hughes also recommends high intensity training for older riders exercising for health. In that book, he claims that a recent scientific paper demonstrated that "intensity training increased the number of mitochondria where energy is produced but neither moderate cardio or strength training increased the number of mitochondria." I have previously noted that I was eager to review the paper from whence Hughes got this assertion and this post is that review. In summary, and with apologies to Coach Hughes, his statement is not supported by the paper he cites and explaining why gives me an opportunity to discuss some important principles useful for understanding a study like this.

Is High Intensity Interval Training Healthier than Moderate Intensity Continuous Training?


The paper I am discussing in this post is Robinson et al., 2017, Cell Metabolism 25, 581–592. The publisher of this paper is Cell Press, one I recognize from my years as a scientist as a highly prestigious publisher, and the study was done at the similarly prestigious Mayo Clinic, so I consider it likely to be a high quality paper. After reviewing it, I have no significant quarrels with the paper itself, but only with how Hughes interprets it. This paper does not attempt to compare the benefits of one exercise program to another, that is, it does not try to answer the question posed in the heading for this section. Rather, this paper is primarily about the changes in gene and protein expression underlying training effects. Because the primary goal of the paper was not to compare the benefits different exercise protocols, the experiments in the paper were not designed to that end. However, this paper does contain data that might speak to the issue, so we can look at it with that in mind, but need to do so with caution.

Two broad categories I use to classify scientific studies are:
  1. Observational versus Interventional studies. Interventional are better, but not always possible.
  2. Biomarker versus Actual Endpoint studies. Actual Endpoint are better, but not always possible.
This paper is interventional rather than observational (that is, it randomly assigns participants to different exercise protocols), so avoids the problems of an observational study, but it is a biomarker study and has the problems associated with that. Specifically, in this case, what we care about (the actual endpoint) is health, but it is almost impossible to measure that directly in an interventional study, so we study something we can measure that we believe is linked to health. The biomarkers used in this study are:
  • VO2peak (a measure very similar to VO2max)
  • Insulin Responsiveness (relevant to adult onset diabetes, for example)
  • Muscle Mitochondrial Activity
  • Fat Free Mass (Muscle Mass)
  • Leg Strength
All five of these biomarkers are generally believed to be good indicators of health so in the interests of brevity I am going to assume that the results of this study are, in fact, relevant to health.

The overall design of this study was to compare four exercise protocols across two groups of participants, a young group (18-30 years old) and an older group (65-80 years old). None of the participants were exercising regularly before the study. This is a common but unfortunate precondition of many exercise studies. I will talk both about why this is common and why it is unfortunate in a later part of this series but will ignore it for this post. This is a fairly small study, each of the eight groups contains only 7 to 11 participants making the statistics of comparison between these groups pretty weak. As a result, only the largest effects can be seen and there is lots of possibility for both false positive and false negative results.

The four exercise protocols were a no exercise control group (SED for sedentary), a High Intensity Interval Training group (HIIT), a Resistance Training group (RT, e.g. weight training), and something they called combined training (CT) which was a mix of resistance and aerobic training. I am going to ignore the details of RT and the resistance part of CT. HIIT involved 5 sessions a week of aerobic exercise. The high intensity part was done 3 times a week and consisted of 4 intervals, each done for 4 minutes at an intensity of > 90% of VO2peak, with 3 minutes recovery between intervals. Converting 90% VO2peak to my training intensity levels, this would correspond to the border between Zone 5b and 5c. For me, this would be a very hard workout. The remaining 2 days a week were 45 minutes at a VO2peak of 70%, roughly equivalent to my MAF tests, an easy workout. The aerobic part of the CT protocol consisted of five days a week of 30 minutes at a VO2peak of 70%, like my MAF test only shorter. Because these protocols involve different intensities, to compare them would require a way to quantitate intensity, and there is no consensus in either the exercise or medical communities as to how to do that. (I have previously blogged about this issue.) That said, based on my intuition and experience I strongly believe that the HIIT protocol involves significantly more aerobic exercise than the CT protocol. Thus, there are three differences between HIIT and CT: the total amount of aerobic exercise, the distribution of that exercise across intensity levels, and the fact that CT includes resistance in addition to aerobic training. If the HIIT protocol were found to be better than the CT and RT protocols, it would not be possible to know if that was due to the greater amount of aerobic exercise in the HIIT protocol, the the higher intensity in the HIIT protocol, or the absence of resistance training in the HIIT protocol. These are very serious and realistic concerns, but having noted them, I will move on.

A lesson offered to virtually all scientists-in-training is "absence of proof is not proof of absence." This is particularly true in a statistical context. Traditionally, the scientific community has used a P value of 0.05 as the cutoff for statistical significance. This corresponds to a 1 in 20 chance that the difference observed is due to chance, or that you are 95% sure the difference is real. If you compare two things and the P value is less than 0.05, then you say the difference is significant. If the P value is greater than 0.05, then you say, not that there is no difference, but that the difference is not significant; that you have not demonstrated a difference. A difference may or may not exist, you simply cannot tell from the experiment you did. Because of the small number of subjects, this is a big issue for this study. Consider the assertion made by Hughes, that HIIT improved mitochondrial activity while CT did not. Here is the data from the paper that speaks to that assertion:



Increase in Muscle Mitochondrial Activity Resulting from Exercise


The dotted line at 0 is placed at the mitochondrial activity at the beginning of the study. If the value is the same as that after 12 weeks of study, then no improvement was observed. Each of the filled in squares or open circles represents the average improvement for the 7-11 participants after the 12 weeks of the protocol shown at the bottom of the graph. The result for each of the participants will be different, and that is how we estimate the variability in the results. Because of that variability, we are not certain that the average of these few participants gives us the exact value of how much each protocol improves mitochondrial activity, and this is shown by the lines above and below the average which give us the range in which we think the answer probably lies (with 95% certainty.) Those groups where the authors are at least 95% sure that mitochondrial activity improved are marked with asterisks. (More asterisks means less chance that the difference is due to chance, that we are more than 95% sure.) First note the SED group, the control group that didn't exercise at all. For both young and older participants, their average mitochondrial activity is lower than before the start of the 12 weeks, but that the difference is not statistically significant, we are not 95% sure that difference is real. In fact, we expect that it is not real, there is no reason for it to change in the absence of exercise. If we only look at the averages, then all three exercise protocols increased mitochondrial activity for both younger and older participants. However, in only three cases can we be 95% sure that is not just due to chance. Maybe RT helps, maybe not. Maybe older participants benefit from CT, maybe not. But now, compare each group not to where they started (the dotted line) but to each other. For example, compare the older HIIT and older CT groups. Although the authors of the study do not compare these two groups, we can estimate the difference by eye, and to my eye, the difference between them is not statistically significant. The difference that is observed is that the comparison between the mitochondrial activity in the HIIT group before and after they participated in the exercise protocol is statistically different, but that the difference in the CT group is not statistically different. I suspect that this is the basis for Hughes' assertion that HIIT improves mitochondrial activity and CT does not. If so, that is an incorrect conclusion from the data. The correct conclusion is that we have significance evidence that the HIIT protocol improves mitochondrial activity, but in the case of the CT group, it might improve mitochondrial activity or it might not, this experiment was not able to answer the question. In short, we cannot say that the HIIT protocol is better at improving mitochondrial activity in older participants than the CT protocol. Based on that, there is no justification for adding a sprint workout to my training.

Let's consider a second example:


Increase in VO2peak (maximum oxygen uptake) Resulting from Exercise

This graph looks at increases in VO2peak with exercise. VO2peak and the very similar VO2max are perhaps the most common biomarker used as a stand-in for the health benefits of exercise. These measure the maximum amount of oxygen a subject can use when exercising as hard as possible. Starting again with SED, the control group, we see the surprising yet statistically significant result that younger participants had their VO2peak decrease over 12 weeks. Despite the fact that this is statistically significant, absent some explanation for this result, I would be inclined to believe that is due to chance. After all, 95% sure is not 100% sure. When we look at the HIIT protocol, we see that younger participants seem to benefit more than older. To me, that says that sprint workouts might have been fine when I was young, but now that I am old, they might not be so helpful. In this case, the difference was so striking the authors calculated the probability that this difference was due to chance, and got a P value of 0.0037, meaning that they were more than 99% sure this didn't happen by chance. For young participants, the difference between HIIT and RT is even more striking; interval training (HIIT) is, with high probability, better than weight training (RT) for improving cardiovascular fitness. In fact, there is no evidence that RT helps cardiovascular fitness at all. (This is the result that would be predicted by both the scientific and exercise communities.) When we compare the benefits of HIIT and CT for younger participants, it looks to my eye to be on the very edge of significance, that we might be 95% sure that HIIT is better at improving CT for young participants. However, when we compare older participants, the average improvement for CT is greater than for HIIT, but but to my eye this difference is not statistically significant. That is, CT and HIIT are, as best we can tell, equally good at improving the cardiovascular fitness of older participants.

Conclusions


My conclusion after reading this paper is that it provides no evidence that sprint workouts improve the health of an older rider like me. Once again, absence of proof is not proof of absence, but if anything, the evidence in this paper suggests that sprint workouts provide no more or less benefit beyond what moderate intensity workouts provide, I should do sprint or moderate intensity workouts as I prefer. Why might I choose to do sprint workouts? First, because they take less time. I hate riding on a trainer, and when doing so, might prefer sprint workouts to get off the trainer as quickly as possible. Second, just because sprint workouts appear to offer me no health benefits doesn't mean that they might not help me become a better recreational cyclists. In my next post, I will ask if sprint workouts might be better than moderate intensity riding at helping me to increase my average speed on a ride, to help me keep up with my faster friends.



* I coined the term coach-author to help highlight the difference between working one on one with a coach and from taking advice from a book or article written by a coach. In the first case, I would refer to the coach as a coach. In the second case, I refer to the coach who is the author of the training book as a "coach-author". The reason this is important is if I were working with a good coach one on one, they would see how I responded to different training plans and would adjust their recommendations accordingly. In contrast, once a coach puts a training plan into a book or article, it becomes one size fits all, it might work for me or it might not.






No comments:

Post a Comment