One of the most exciting and heavily debated topics in exercise science today is something that most physiologists and coaches hadn’t even heard of 10 to 15 years ago: distribution models. Simply, they are how an endurance athlete divides their training time between easy, moderate, and hard training. We talk all the time on the show about polarized training and sweet spot training, which are two popular distribution models. But there’s also pyramidal, high-intensity, and low-intensity training models.
To avoid confusion, distribution models and periodization are not the same thing. Periodization is about how you vary your training through the season. You can be polarized or sweet spot in your distribution and still use a linear or block periodization approach. And if that sounds a little confusing, these are the things that keep your coach up late at night figuring out your plan.
In the last 10 years, a large number of studies have been published comparing different distribution models, but they often have a small number of subjects and only last a few weeks. It’s been hard to get enough data to truly compare the approaches.
Our guest, Dr. Michael Rosenblat, working with Dr. Stephen Seiler, decided to take on this challenge. They created an ambitious plan to collect data from multiple distribution model studies and pool that data into a network meta-analysis to see what conclusions could be drawn from a much larger sample size. Dr. Rosenblat works with the Sylvan Adams Sports Institute and has taught both exercise science and statistics. As Dr. Seiler said, he was the right person for this ambitious job.
On this episode, we talk with him about the different distribution models, how they put together their study, what they found, the challenges they faced, and whether the proper study of distribution models is beyond the scope of the lab. Throughout, Dr. Rosenblat shares his recommendations based on the science. Joining Dr. Rosenblat, we also hear from Dr. Seiler, including his thoughts on why they did the study and how he felt about the results.
RELATED: Fast Talk Episode 210—Nerd Lab: Eccentric Damage and a Heated Polarized Debate
One final note. After we were done with the episode, Dr. Rosenblat did a deep dive with us explaining the biggest issues in meta-analyses nowadays. Particularly, what to look for in order to determine if it’s worth your time. If you’re interested, check out our bonus recording from the episode.
So with that, put on your best statistical thinking cap and let’s make you fast!
References:
- Burnley, M., Bearden, S. E., & Jones, A. M. (2022). Polarized Training is Not Optimal for Endurance Athletes. Medicine & Science in Sports & Exercise, Publish Ahead of Print. Retrieved from https://doi.org/10.1249/mss.0000000000002869
- Foster, C., Casado, A., Esteve-Lanao, J., Haugen, T., & Seiler, S. (2022). Polarized Training is Optimal for Endurance Athletes. Medicine & Science in Sports & Exercise, Publish Ahead of Print. Retrieved from https://doi.org/10.1249/mss.0000000000002871
- Rosenblat, M. A., Watt, J. A., Arnold, J. I., Treff, G., Sandbakk, Ø. B., Esteve-Lanao, J., … Seiler, S. (2025). Which Training Intensity Distribution Intervention will Produce the Greatest Improvements in Maximal Oxygen Uptake and Time-Trial Performance in Endurance Athletes? A Systematic Review and Network Meta-analysis of Individual Participant Data. Sports Medicine, 1–19. Retrieved from https://doi.org/10.1007/s40279-024-02149-3
- Seiler, S. (2010). What is Best Practice for Training Intensity and Duration Distribution in Endurance Athletes? International Journal of Sports Physiology and Performance, 276–291.
- Spragg, J., Leo, P., & Swart, J. (2022). The relationship between training characteristics and durability in professional cyclists across a competitive season. European Journal of Sport Science, 1–17. Retrieved from https://doi.org/10.1080/17461391.2022.2049886
- Stöggl, T. L., & Sperlich, B. (2015). The training intensity distribution among well-trained and elite endurance athletes. Frontiers in Physiology, 6, 295. Retrieved from https://doi.org/10.3389/fphys.2015.00295
- Treff, G., Winkert, K., Sareban, M., Steinacker, J. M., & Sperlich, B. (2019). The Polarization-Index: A Simple Calculation to Distinguish Polarized From Non-polarized Training Intensity Distributions. Frontiers in Physiology, 10, 707. Retrieved from https://doi.org/10.3389/fphys.2019.00707
Episode Transcript
Trevor Connor 00:00
Trevor, hello and welcome to fast talk. Your source for the science of endurance performance. I’m your host. Trevor Connor, here with Dr Griffin McMath, one of the most exciting and heavily debated topics in exercise science today is something that most physiologists and coaches have even heard of 10 or 15 years ago. Distribution models simply, they are how an athlete divides their training time between easy, moderate and hard training. We talk all the time on the show about polarized training and sweet spot training. Well, they’re just two popular distribution models, but there’s also pyramidal, high intensity and low intensity. And quickly, to avoid confusion, distribution models and periodization are not the same thing. Periodization is about how you vary your training through the season. You can be polarized or sweet spot in your distribution and still use a linear or Block Periodization approach. Now, that sounds a little confusing. These are the things that keep your coach up late at night, figuring out your plan. In the last 10 years, a large number of studies have been published comparing different distribution models, but they ought to have a small number of subjects and only lasts a few weeks, it’s been hard to get enough data to truly compare the approaches our guest today, Dr Michael Rosenblatt, working with Dr Steven Seiler, decided to take on this challenge. They created an ambitious plan to collect data from multiple distribution model studies and pool that data into network meta analysis to see what conclusions could be drawn from a much larger sample size. Dr rose millat works with the Sylvan Adams Sports Institute and has taught both exercise science and statistics. As Dr Seiler said, he was the right person for this ambitious job on today’s episode, we’ll talk with him about the different distribution models, how they put together their study, what they found, the challenges they faced, and whether the proper study of distribution models is really beyond the scope of the lab. Throughout, Dr Rosenblatt will share his recommendations based on the science. Joining Dr Rosenblatt will, of course, also hear from Dr Seiler, including his thoughts on why they did the study and how he felt about the results. One final note, after we are done with the episode, Dr Rosenblatt did a deep dive with us, explaining the biggest issues in meta analysis nowadays, in particular, what to look for in order today, I shouldn’t read this one. Fortunately, we were still recording. So if you’re interested, go to our show notes for this episode. We have a link to the recording there, and with that, put on your best statistical thinking cap, and let’s make you fast. Well, welcome Dr Rosenblatt to the show. This is the first time we’ve had you here. Looking forward to talking with you. Thanks for joining us.
Dr. Michael Rosenblat 02:27
Yeah, thanks for having me. It’s great to join you guys today. So you reached
Trevor Connor 02:30
out to us because you worked on a meta analysis with Dr Seiler, comparing different distribution models. So really excited to hear this, because it’s a heavily debated topic right now. So you guys have kind of dived into this, and definitely want to get into what you found, how you did the study. But before we get there, just for some of our newer listeners, can you explain to us what we mean by distribution models? Sure,
Dr. Michael Rosenblat 02:57
- And even before we jump into the models themselves, I think it’s important to describe how we would divide training intensity there’s, of course, there’s these different training intensity models. The one that’s commonly used in training intensity distribution literature is a three zone model, and it’s divided up into these different physiological zones. So you have your first zone, which is below your first lactate or ventilatory threshold, and you have your middle zone, which is between your first and your second threshold. The third zone is above that second threshold, or your maximal metabolic steady state. I guess you know, we’d say, well, athletes would tend to incorporate all three of those zones within their training. And historically, what we found is that maybe athletes may do one type of distribution over another. And so a common model, and actually probably the most popular model, would be a polarized training intensity distribution. That’s where the majority of training would be below or below the first threshold, or in Zone One, about 80% maybe 75% and then maybe the next greatest amount would be in zone three, maybe 15 or 20% and then a very small amount, if any, in zone two. And so that’s probably the most popular or common training intensity distribution. I think another one that’s commonly discussed as well would be a threshold model, where actually majority of that training would be in that zone two, which is completely the opposite of what you would see in a polarized model. Yeah.
Trevor Connor 04:27
And so that can be a little confusing to people, because in the scientific literature, threshold is the range between that first threshold and second threshold, where most people think of it as that just that second threshold, anaerobic threshold, or lactate threshold, and that’s where you would time trial at. But that whole range is actually the threshold range. But the more common term, if you’re out talking to your cycling buddies, everybody refers to that as sweet spot training. Yeah,
Dr. Michael Rosenblat 04:53
that’s definitely correct. And depending on the duration of an event, you kind of end up somewhere within that second. In zone in
Trevor Connor 05:01
all due respect to Frank but one of the things that you said right at the beginning of your study is the threshold model tends to come out not as well as the others in the literature. Yeah, actually,
Dr. Michael Rosenblat 05:11
I think that was my first publication on the topic. Was comparing polarized versus threshold training intensity distribution, and found that polarized training model tended to produce better outcomes in time trial performance or racing performance. You know
Trevor Connor 05:29
that said, I have seen some really good athletes do the sweet spot approach and actually do quite well, but we’ll get into that in a minute. So let’s dive a little into your study. So let’s just hit the very basics here. Why did you and Dr Seiler decide to do this study?
Dr. Michael Rosenblat 05:47
It’s a very interesting question, because I tell people that, you know, it’s such a major study, and in terms of how much work and what went into doing this, but we kind of came up with it on a whim, which isn’t quite what a lot of people would have expected, and it was actually near the end of our the first study that I worked on with him, which came out a year ago, where we compared low intensity training alone versus interval training and low intensity training combined. And the results from that study were quite interesting, and it showed that we see improvements in vo two Max when we add high intensity training, but maybe not a change in time trial performance. And around this time, Dr Seiler was asked to work on another project, and he was including some of my original work on training intensity distribution, and he said, Maybe we should do, look at doing an additional analysis, because there’s so much new literature out there, and so that would be a huge project. So kind of toying with the idea. And so he continued on with his project. I thought, you know, you know, maybe we should look into this, but not just do a regular meta analysis, but maybe do a network meta analysis, which is looking at all the different types, or including as many types of interventions as we could. But then maybe even take it a step further and see if we can get all the Indian the individual participant data that’s ever been published on the topic. So kind of thinking about that, I reached out to Dr Seiler again and said, Okay, well, I think there’s a shot that we could put this together. Let’s try to go ahead with something like this. And just kind of decided to move forward. Yeah, let’s
Trevor Connor 07:19
pause for a minute and hear Dr Seiler thoughts and just how ambitious a study was and why he chose to work with Dr Rosenblatt.
Dr. Stephen Seiler 07:27
My involvement in that was obviously that I know a lot of the people in the whole ecosystem, and so I was involved in recruiting them and saying, Hey, let’s pull together. Let’s give our data, you know, and Michael and I had talked about this is doing this kind of, you know, a big database that could grow over time. Because, you know, we know that a lot of exercise physiology studies are underpowered. These training studies are traditionally underpowered. So for me, that was the magic here was to achieve that cooperation among a bunch of different sports scientists that would actually give their raw data to someone and say, yep, go ahead, use it, you know. And so Michael is an outstanding first, he’s a physical therapist. He’s an outstanding clinician. He’s excellent methodologically, he’s as rigorous as they get when it comes to going through all the appropriate steps on the systematic review and the appropriate data use. I didn’t agree with him on all interpretations and on, you know, I would have even maybe limited because there were some of the sample sizes I felt were too small to even include, whereas he said, well, let’s just put it out there. And so we agreed to disagree on certain things, but the overall thing that we both agree 100% on was that this was a really good exercise in cooperation, and I think this is where sports science has to go if we’re gonna make progress on some of these issues like individualization, like understanding clustering of different kinds of athlete that respond differently to different kinds of stimuli and so forth. So I think that was, for me, the most exciting aspect of it. And obviously, yeah, the high intent, you know, that we see some differences that maybe for lower trained or less trained, that the pyramidal is better than the polarized. That makes sense, you know, in terms of stress, what they tolerate.
Trevor Connor 09:29
I was actually pretty impressed by that because so a meta analysis is where you take the results from multiple studies and you basically, and there’s a lot of different ways you can do it. You pool that data together into a larger group of data that hopefully gives you better indicators of whatever it is you’re trying to study in that particular meta analysis. And I was really impressed that you not only collected the studies, you actually reached out to the authors of all these studies and said, Hey, can you send us the original data? Yeah.
Dr. Michael Rosenblat 09:59
And we didn’t realize that we were going to do an IPD or individual participant data meta analysis initially, and actually, I didn’t know if we were even going to do a network meta analysis, which I can explain as well because of the methods for that. Or we start getting a little bit complex in terms of the statistical approaches. But when I reached out to a few people, they decided they Yeah, they’re willing to share their data, and then one after another after another was willing to share their data, and it was very exciting. Didn’t just ask for their outcome data and maybe their participant characteristics, but I also asked for their training characteristics data to be able to even dive even further into the analysis. And
Trevor Connor 10:36
so you kind of hinted at this. So tell us what a network meta analysis is, and we don’t need to go into the gory details. I know meta analysis get complicated. I don’t
Griffin McMath 10:46
know. I kind of want to see how deep this rabbit hole goes. It’s not often you can be on the call with a stats nerds who proudly want to dive in. So Trevor might get up the rail, but I’m fascinated.
Dr. Michael Rosenblat 10:57
You know, it’s interesting. I only became somewhat of a methodologist and biostatistician because COVID occurred. I was actually in the lab doing looking at oxygen uptake kinetics for my PhD and interval training, and that’s what and COVID happened, and then kind of led into evidence synthesis. So if we were talking about a meta analysis, most commonly, it’d be a pair wise analysis, which is where you just look at Group A versus group B, or intervention a versus intervention B, and you’re just trying to compare the results and pool and synthesize the results of as many studies that would meet the eligibility criteria. Whereas a network meta analysis does several things. One, you’re able to compare more than than two interventions, so you can see which intervention is optimal, which is very interesting. So you can have a polarized model, a threshold model, and then we’ll talk about a pyramidal model, and maybe other distributions as well. At least in this case, you can also rank the interventions based on which is most effective. But even more interestingly, to further increase the power to detect a significant effect, basically, to increase sample size, you can do indirect comparison. So for instance, if you have study that compares intervention A versus B and B versus C, but you don’t have a versus C because you have this connection within this network, you’re able to indirectly determine what A versus C would look like. It’s quite an interesting way to compare interventions. And if
Trevor Connor 12:21
anybody goes and looks at the study, and of course, we’ll put the reference in the show notes, you have these great diagrams looking at the different distribution models with connecting lines and showing the strength of each one. That just shows the network relationship between all these different models, which was fun to look at. Yeah. What’s really
Dr. Michael Rosenblat 12:38
cool about it, too, is there’s another analysis that I ran, which at the time, I didn’t think was going to be such a important analysis, but we’re able to compare different ways of looking at the different groups, and so you can see how those networks change based on the type of analysis that we did.
Griffin McMath 12:55
So Michael, before we dive into the results, you talked about the study, why you did it, how you pulled the data together, and you know, before we reveal the juice to this hard squeeze here, why would an athlete or a coach actually care about this? And that’s not to add insult here, but we’re enjoying this because we want to get into the details. I want to ask you about your methodology, and I find it fascinating. But how can this information that you’re about to share, the results, be framed for the athlete and coach listener? How can we frame this so the results you’re about to share actually mean something to what they do every day? So
Dr. Michael Rosenblat 13:31
I think it’s one thing. So I was a coach, and I still say I’m a triathlete coach, but I haven’t coached in several years because of all the work I’ve been doing, and I was a recreational triathlete as well. But it’s one thing to say, Well, do we already, we already know this. We’ve been doing this for years. But the issue is, is, unless we’re actually comparing one intervention to another, we don’t really know for sure is, is this intervention that I’ve been applying truly effective? And so you say, Well, that would just, you know, include a randomized trial for that, assuming that we’re able to conduct a randomized trial that that would assess what it is that we’re looking for. The issue in sports science is, all coaches and athletes would agree, especially as performance increases, you’re less likely to allow someone to manipulate your training because you have certain goals, and so typically, in a study, you might have anywhere from 10 to 20 participants. In some studies, you might have up to 40, if we’re very lucky, and the sample size is very small, so we can’t really determine if the results that we’re looking at is truly going to be important and if it’s going to help our athletes. So by doing this type of a study, a meta analysis, especially if we really do very meticulous design, then we’ll be able to get a much better understanding of the results and try to make it as accurate as possible. So that way, for knowledge translation purposes and implementation purposes, we can say, well, here’s these results. Am. Able to directly implement these for my athletes, and how robust are they, and are they trustworthy? And so by conducting something like a network meta analysis, we’re able to pull as much data as possible to try to feel more confident in the decisions that we’re going to make. So
Griffin McMath 15:14
Michael, I think what I just heard you say is that coaches who listen to our podcast and this episode with you are therefore more trustworthy and more credit more credible, right? This, we can immediately make this it’s not a fallacy to say that, right?
Dr. Michael Rosenblat 15:29
Well, in terms of saying is the results from this study as accurate and as trustworthy as possible, I’d say, based on all the literature that’s out there, I did certainly would say, I feel very confident to say, would a study like this provide the most accurate type of results? And I’d say, probably yes, there’s no studies that have all of this data that’s out there that’s all pooled together with such type with such an analysis. So I feel very confident in these results. And I’ve certainly been asked this question several times. So you’re not the first person to have asked me this.
Griffin McMath 16:02
No, I love it. You answered it very seriously, and Trevor knew immediately that I was being sassy about making it about our podcast, but I appreciate you thinking. I think that’s really helpful. So now on to the results.
Trevor Connor 16:14
Yeah, so let’s dive into that. You’ve talked a little bit about polarize, you talked about threshold, but there were multiple different distribution models that you looked at. So let’s just make sure we cover them all. There was also pyramidal. Let’s explain that, and what, how it’s different from polarized. And I believe you had high intensity and low intensity. What were your distribution models? Yeah,
Dr. Michael Rosenblat 16:32
so the pure middle model, that’s where, of course, you’d still have the majority of time spent in Zone One. The next would be in zone two, and then the third would be zone three. So that’s where you’d see a difference between that and that polarized, more time would be spent in zone two. A high intensity model would be where the majority of time would be spent in zone three, and then the next would be anywhere in Zone One or zone two. More commonly, if you’re going to spend a lot of time in zone three, there’d be much more time spent in Zone One. But we we were okay with either one, zone one or zone two, based on how we defined it in the study. And then zone the low intensity distribution was just 100% time spent in Zone One. So there’s those are the five different distributions. Interestingly, the most common distribution is a polarized model. So we compared poll to whatever other distributions were in the literature,
Trevor Connor 17:21
and so tell us what you found, what were the main results? So
Dr. Michael Rosenblat 17:25
it’s a very important thing to describe the results in terms of, is there a difference, no difference, or is there a statistically significant difference or no statistically significant difference? And I’m only giving a little. I’m only, I only want to describe this initially, because there’s been some misinterpretation of the results. And I think it’s really important to say that from the get go, when we compare and we look at the network meta analysis, there appear to be no difference across any of the interventions for both vo two Max as well as time trial performance. So the two outcomes that we looked at were vo two and time trial, and we ran two types of analyzes. We did something called an intention to treat analysis and a per protocol analysis. Intention to treat in this study, we called it, just as the groups were originally allocated. So wherever the participants were, they stayed in those groups, whereas the per protocol analysis, because we were able to collect all the individual participant data, including the training data, we were able to reallocate participants into the respective groups for which they actually completed the intervention. So we know athletes, not just in training studies, but also as coaches, we’ll see that athletes may not exactly do what we program. And so I was able to redistribute athletes based on what they actually did. And so for both for Time Trial performance as well as for vo two Max, we had two different analyzes based on their original groups and their reallocated groups. And interestingly, both for time trial and for vo two Max, there was no difference, at least right from the get go. But when we look at if we really look at those results for polarized versus pyramidal, there was no absolute difference at all. However, for the other groups, when we looked at the comparisons, there was no statistically significant difference. The reason why I say that is the magnitude of the effect, meaning, well, it shows like, oh, there’s some degree of a difference. Could have gone in one direction or another, but there was no real meaning to that because the sample sizes were so small. And so when I say the sample sizes are small, we can’t get a true effect here. We don’t know, is it really one intervention versus the other intervention that’s beneficial, and how large of an effect is it? Is it a small benefit? Is it a very large benefit? And so when we compare all the interventions other than polarized versus pyramidal, we can’t really comment on that, and that certainly leads to some future research thing. Well, maybe there’s some things we need to look at, but because the sample size was large enough for the polarized versus pyramidal groups, we were able to do some subgroup analyzes. And so this is where we pulled out those two intervention groups and ran what I call a pairwise analysis, just. Comparing the two groups, and still, there was no statistically significant difference, as well as no difference between those groups. However, there was a large degree, I shouldn’t say a large degree. There’s actually a small degree of statistical heterogeneity, meaning that there’s something interesting about how we pool these studies together. Something wasn’t quite right. Should we have actually pooled all these studies together. So we ran all these different analyzes based on study design, participant characteristics, differences across interventions, even differences in how the outcomes were done. So basically in time trial duration, how they collected the data, etc. And what we found after looking at all of these different characteristics, or study characteristics, was that performance level actually influenced how individuals responded to an intervention, but specifically for vo two Max, so the different types of athletes we had were recreational and competitive athletes, and we found that a polarized distribution tended to benefit competitive athletes, whereas a pyramidal model tended to benefit recreational athletes. And that was quite interesting. It wasn’t again, we weren’t going to look for this. We were trying to determine why there might be this degree of heterogeneity, and explore, well, where there’s differences in these studies. And by doing so, we found something quite interesting, and you
Trevor Connor 21:22
did take it a step further, which I agree with completely, to say that when you’re dealing with really recreational athletes, almost any approach, they’re going to see benefits from where, when you get to the elite, you’re really only seeing benefits or improvements in vo two Max when they were polarized, certainly.
Dr. Michael Rosenblat 21:37
And it’s interesting because, you know, with recreational athletes, and I’m very cautious to say this, but almost if you do anything within reason and an appropriate volume, meaning not too much, you’re going to see improvements, because they’re really starting from from such a low level of fitness across the board. Of course, I’m generalizing depending on the type of recreational athlete, but with competitive athletes, there’s certain limitations, or certain, I’d say, limitations that they already, that they’ve reached in terms of their ability to see improvements. And so there’s only certain things that we can see or that we can do to improve competitive athletes. Well, you
Trevor Connor 22:15
are a coach, and I will say this as a coach, it’s one of the struggles you have with relatively new athletes, because when somebody comes off the couch and just starts training, as you said, almost anything, they’re going to get fitter because they’re starting at such a low baseline. And I have had the problem many times as a coach, where you start working with an athlete, you try to get them to a better approach and the resistance, because the way they were coaching was working so well. And the other thing is, when you start from the couch and start doing training as a brand new athlete, you improve quite rapidly, where, as you know pros takes a lot of work and a lot of time to see even small improvements. So you can often have an inexperienced athlete who employs a very bad training plan and is very resistant to get away from it, because it worked. It worked when they were coming off of the couch. And it’s
Dr. Michael Rosenblat 23:08
interesting, because I’ve spoken with Dr Seiler quite a bit about training and optimizing training versus providing as little stimulus as you need to maximize the result, and it made me think quite a bit about one of my previous studies, which found, you know, how can you optimize interval training to maximize performance? And what my results in that study showed that maybe doing something like five by five minute intervals would produce the best adaptive response for Time Trial performance. However, if you know five by five minute intervals is quite a bit of work. And so you’d say, well, is that what somebody who’s you know, just coming off the couch or recreational athlete should do? It’s like, Well, I wouldn’t necessarily say that’s the case. I mean, if just putting out some degree of stimulus at an intensity should be sufficient, at least, to cause an adaptive response. Before we move
Trevor Connor 24:01
on from the results, let’s hear Dr Seiler thoughts on why they didn’t find bigger differences between the approaches.
Dr. Stephen Seiler 24:07
Well, I think it’s kind of consistent with where my head has gone in understanding this whole process that, you know, we can even go back to the signaling part, that it follows a bow tie architecture, meaning that there’s these multiple pathways, metabolic pathways, into a central not which is PGC one alpha. So you can turn on the PGC one alpha, and that upregulation that happens through the ATP to amp ratio, through reactive oxygen and nitrogen species through certain aspects of mechanical stress, through calcium concentration in the cell, there’s a number of metabolic processes that function as signals, and then they coalesce at a central regulator that then up regulates and turns on a whole slew of. Adaptive effects, including mitochondrial increases, capillary density changes and so forth. So there’s this beautiful and exquisite kind of bow tie architecture that we see in biology a lot. So that doesn’t surprise me that we can achieve these adaptations with different intensity and duration combinations, and then. So if we know that, then that means, okay, there’s different ways to get the signal. Which of these ways creates the least stress. So that’s where I’m at in my understanding of things. Is that I wish I’d maybe use different terminology two and a half decades ago, but the idea is the same, that we have to manage intensity distribution by doing so we’re managing stress.
Griffin McMath 25:49
You know, you’re talking about recreational versus competitive. If an athlete is listening to this and they’re wondering where they line up, because they might think, Well, I’m recreational, but I mean, doing at least 15 hours a week, I just don’t happen to like, what’s the delineation they can make if they’re listening wondering, Where do I fit? That’s
Dr. Michael Rosenblat 26:08
actually a good question, because it was a discussion that I had with Dr Seiler, if we were to call it performance level versus training status, because training status would be, well, what’s the volume of training that you’re doing? And I’ll certainly say you can be doing five hours a week of very good quality training, or 30 hours a week of terrible quality training. You may not see the adaptive responsive to someone who’s only doing five hours. So the volume in which you’re doing, of course, I’m just arbitrarily coming up with those numbers, but the volume in which you’re doing certainly is related to your competitive level, at least, if you look at retrospective analyzes of studies, we’ll see competitive athletes tend to do more training. May also have time to do it, but they can also handle that type of volume. But what we found was that competitive athletes, or at least the way that we described them, were that if they competed at some degree of high performance level, so like tier three and above, if they were university or college level athletes, provincial, state, national, international levels, like anywhere along the board that there was some degree of competitiveness. And then if they weren’t competing at one of those levels, we then considered them recreational athletes. Now, of course, I’m very particular about how I do my studies, and I want to really make sure that I’ve done everything as accurately as I could. So I say, well, is that even an appropriate way to separate these groups? Because that was what the authors of the original articles, how they classified their athletes. So what I did was, I want to see, is there an actual difference between these these categories, so post hoc or after I ran the analyzes, what I did was I I assessed to see if there’s a difference in baseline vo two Max between the two groups. Now normally what people would do is, initially, they’d say, Well, your VO two Max is 50, or these, these individuals are close to 50, these are close to 70. We’re going to just divide them based on their vo two but then you’re missing a whole bunch of other criteria. So not necessarily the best way to divide individuals. But what I did was we wanted to say, well, at baseline, was there a difference in vo two between these individuals, as that was the only measurement that we had. And we found that, on average, recreational athletes had a vo two max of approximately 55 mils per kg and or per kilogram. And competitive athletes, I think they were close to 65 males, approximately. I can’t remember the exact number, and the only thing that we found too, is that there was actually more female athletes that were recreational as compared to competitive. So I wanted to again, be even more sure that I’m separating these individuals correctly, so I ran a separate analysis to see, well, is it a lower vo two specifically because of the sex of the athletes? And then it actually didn’t influence the results. So it was an interesting way to find out after that, yeah, these groups were actually significantly different based on certain criteria.
Griffin McMath 28:55
Thanks for explaining that. I think this is a great segue to actually talk about any potential issues or shortcomings or what else came up with your study that may have impacted the results.
Trevor Connor 29:09
Yeah, and this is a meta analysis, there are always issues when you are taking studies that have very different study designs and trying to pool their data together. So you did have a fairly long list of here’s the things that you encountered.
Dr. Michael Rosenblat 29:22
Yeah, and it’s interesting, because the most important thing to do once you’ve pooled studies together, and I would argue that 99% of meta analyzes in sports science do not do this, is they look at the degree of statistical heterogeneity. And all that means is, when we look at the results, are those results somewhat similar, and was it appropriate to pool them together? And a good example would say, let’s say training duration. If somebody trains for two weeks versus 10 weeks, you’d say, well, obviously the person who trains for 10 weeks, at least, it sounds intuitive that the person who trains for 10 weeks should see greater improvement, but if they both had, let’s say a. Two mil improvement, then we say, well, maybe in this case, obviously, I’m just giving an example, that maybe training duration didn’t matter, because there’s no difference in those results. And so when we look at these results, there was no statistical heterogeneity. So if we basically, we measure that by saying, 0% would be nothing, 100% would be a high degree of statistical heterogeneity, and then only in that one subgroup analysis, when we compare pole versus peer, we found that performance was some degree. Did play some degree of an influence. But once we separated by performance level, there was no statistical heterogeneity. We still ran all the sub all those analysis based on the types of study designs, based on the duration of the studies, all the participant characteristics you can imagine, and there was no difference in there was no degree of statistical heterogeneity. So it’s very interesting to find these results, because normally it would be very high, above 50% 60% the other thing that I’d say is one of the things that was very different in this studies would be that the way in which they determined or calculated the different thresholds. So how they would divide people into these different groups? We’re all different across studies. Some studies you’d lap use lactic threshold. Some use ventilatory threshold. Some people use lactate but they would be very they use arbitrary cutoffs, like four millimoles for their second threshold, which is somewhat of an older way of thinking about the second threshold, boy, there
Trevor Connor 31:21
are people die on the sword for that one, four millivolts as your threshold. Everybody doesn’t matter.
Dr. Michael Rosenblat 31:30
They don’t realize, you know, the system, the human system, is somewhat continuous here, and it doesn’t necessarily coincide with one number. I think, you know, that was quite interesting. AND to combine all those studies, and that’s that had these differences in how those thresholds were determined, because to be quite honest with you, I thought that’s where we were going to see a major limitation, and we didn’t. Because, again, there was no statistical heterogeneity, saying that there’s still significant shortcomings to this study. And I’d say the largest one was sample size, yeah, even though we really tried to increase the sample size and this, this study included all the experimental and quasi experimental studies that exist on the feet in the field that met the eligibility criteria, we were able to collect all the individual participant tab. And so that would be the first shortcoming, I would say, is that sample size, we still didn’t have enough sample size for the other training intensity distribution model, so we can’t really comment on that. And the one other, I guess, major shortcoming, I’d say, would be how we reallocated participants. We used heart rate. And for those of you who know the training intensity distribution literature, or even just exercise science literature, we know that when you exercise at a certain intensity. So let’s say, for instance, you’re doing interval training, your heart rate is going to go up as soon as you stop, and maybe you’re now in Zone One instead of zone three. Well, your heart rate’s still elevated. So even though you’re exercising in Zone One, it’s going to look as though you’re still in zone three, or maybe as that time comes down into zone two, so it might start shifting the models. And so that’s certainly a limitation that we discussed, and it’s important to consider. But I will say there’s some very interesting results that somewhat conflict with that. Though I can’t explain why. I can just say something conflicts with it, but it doesn’t make sense as to why. The first thing was in one of the studies, the distribution was 8010 10, and so that didn’t really follow a real distribution. So we’d say, Well, should we include that? When we actually distributed based on heart rate or reallocated, it actually shifted from like 8515 and so it shifted to a more polarized model, which is the opposite of what you may have thought. Saying that, of course, other models may have shifted from a polarized to a more pure middle. So we did. We did see things shift in a way that we predicted, but also in a way that we may not have thought they would. To
Trevor Connor 33:49
your point, there’s a great study, and I can’t remember the authors of it, but they took the same data from athletes, and we’re looking at how they were distributing their training, and this was cyclists, and when they analyzed their data using heart rate, they were pyramidal. When they analyzed their data using power, they were polarized, because of exactly what you were talking about. Their power was always either in zone three or zone one, but they would spend a lot of time by heart rate in zone two, because it would take a while for heart rate to come up. It would take a while for heart rate to come down. So how you measure it can actually be very important. Yeah,
Dr. Michael Rosenblat 34:26
I think we actually cited the article as as one of the limitations to using this, this method, interestingly, well, actually, I’d say the main reason why we did is because we needed some sort of similar measurement that went across all studies. So that would be probably the main reason why we use that. But a really interesting finding was that when we look at the the intention to treat analysis results were kind of all over the place, but when you look at the distributions of the results, once we reallocated using heart rate, not only did they present somewhat of a similar distribution in terms of the results for the. Different comparisons for vo two Max as well as for time trial. And not only that, the confidence intervals became more narrow. And so it actually tightened things up and made things a little bit more accurate. And the interesting thing about the difference with the VO two Max and time trial was maybe only 70% of those studies were similar. And so the fact that there was different studies, but they still kind of showed a similar distribution overall might speak to I’m always hesitant to say this in science, but the accuracy of the results, given the fact that, yes, there’s some studies for some participants that may have shifted groups. So considering that error, so
Trevor Connor 35:34
I do want to propose one other potential issue. And look, this is not the beat up in your study. I enjoyed reading your study. You guys did a great job on it. You really thought it through. It was an impressive study. Where I’m actually heading, which I’ll get to in a minute, is the question of whether distribution studies can really be conducted in the lab. But one of the other issues that I’d bring up, as you pointed out, was the duration of the study. Some of the studies, at least one of the studies you said was just three weeks in length. I can tell you, as a coach, when I work with an athlete, I always have a bias towards polarized training. If I’m working with an athlete who’s coming off of a threshold model or a high intensity model, I’ll shift into polarized and the first thing I tell them is, you’re going to train really hard for a year and you’re going to see no improvements. We’re not going to see improvements until the second year, when you’re dealing with distribution models and want to see the true benefits. You’re measuring that in years. You’re not measuring that in weeks. And you’ve also brought up the issue of sample size. So I said this is where I was going to go, and it’s the question I want to throw at you, when you’re dealing with distribution models, where it takes a long time to see the benefits, where you probably need a much larger sample size to be able to really see the changes, the effects. And as you know, in a lab, you’re lucky if you get eight people. And to get eight people to come into a lab for six weeks is hard to do. Are we hitting the limitation of the in lab experimentation with this, do we need to find another way to study distribution models? You asked several questions there.
Dr. Michael Rosenblat 37:17
But, but I’ll start by saying, first of all, I think that these are very important questions. And interestingly, I always get more concerned about my work based on the quality of the methods and the statistical approaches, whereas the results are the results. I can’t control those. So I guess what I try to do is, at least with this type of work is based on the literature that’s available, how best can I synthesize this in the cleanest way to provide the best answers from what we know through a scientific lens. And that’s really what I hope to do, because at least you know, what I love is improving the quality of sports science research, regardless of the topic. I think there’s so much better that we can do now, when it comes to am I really showing a lot about which distribution is better. I’ll say that if you look at the title of our manuscript, which is very long, but that’s because that’s how methods go when you do a Prisma guided systematic review. But the one thing that we included in it was training, intensity distribution, intervention. And actually, Dr Seiler, when he stated that at first, so we need to include the term intervention. At first I was like, Yeah, sure, we’ll put it in. But the more that I thought about it, the more I’m happy that we included that term, because we’re just looking at an intervention versus another intervention, and it’s very important to consider that when we’re interpreting these results, we’re not looking at as much as some of these studies are periodized. We weren’t able to analyze how the periodization influenced the results across studies. It just wasn’t possible based on the available data and the differences across the studies. So what I would say is, if we’re just considering these results as just as one, if we’re looking at maybe a mesocycle or a certain type of an intervention, is one beneficial for one type of population versus another. It’s a good place to start, but by no means as a coach or even as a sports science would a scientist, would I say, Well, this is kind of closes the door on pyramidal versus polarized or any type of distribution for recreational or competitive athletes, because we’re only able to capture so much information here. And like you said, yeah, it’s not just over a certain training intervention or mesocycle a year or multiple years. Programs change over time for the right reasons, and hopefully they do right because we should be testing our athletes all the time, and an athlete may respond or adapt to an intervention at a different speed than another athlete, and therefore we’ll have to adapt or shift that intervention accordingly. Basically, I’m saying, does somebody need a more rest period, or should we be shifting how they’re doing their interval training? Do we need to shift how much time they’re spending in the different zones, whether or not they’re even within the same distribution intervention? Or if it means that they need to change their T ID, it’s. Health. So I think it’s very good that you brought that up, because I think it’s important to say, well, there’s, there’s things that we may not be able to capture in sport science studies as it stands, or as as how we’re conducting this research. Now, I know Dr Seiler, as well as myself, have tried doing some crowd source type studies. Mine that was, give me my next question. Yeah, I was gonna say, I want to say mine was a big flop, but it wasn’t in in the sense of, Wow, did I ever learn a lot, and I also learned a lot that communication is extremely important in studies when you’re doing data collection. This whole I did a virtual training study. I didn’t publish the results through a peer review. I actually submitted it to one place. I’m not going to get into that, but it wasn’t accepted because there were no positive results. And so there you have your publication bias. But the methodology was unique. It was the first training study that used an online software platform and answered some interesting questions. Of course, I sent the sample size was large enough, but the dropouts was like seven. I don’t remember. It was like maybe 70% something like that. I And again, it was because people, they come into these studies and they’d say, Oh, yeah, I’m willing to do this. Like, Oh, you want me to only do this? What about my other training? And so some issues on how I would do things differently. But I think there are ways to do this, and I think crowdsource studies is one method. I think we need to be a little more open in terms of how restrictive those studies are. So of course, when you look at a randomized trial, you want to only have one variable that you manipulate. Everything’s the same. We randomize people into different groups if there’s only one variable, because athletes tend to want to change their training a little bit, and there there’s so many differences in terms of location. Even if we do crowdsource studies, you have to have very large sample sizes, which certainly is possible this way, because now you can do multivariate analyzes. The larger the sample, the more differences across these individual training programs and these individuals we can have. So I guess I’m kind of touching on a few things, but certainly you’re right. We need to make some changes in how we’re doing our work, and
Trevor Connor 42:01
for any of our listeners who are wanting crowd source data, is when you basically get an athlete to give all their training data for a month or a year or a couple years, and since they’ve all been recording this, you can actually get a lot of athletes to potentially give their data. So one of my favorite crowdsource studies was from 2011 where they got Strava to basically give them the data of almost every single athlete that was on Strava for the last couple years, and they were able to just go in and look at the trends and do analysis of this huge pool of data. Is
Griffin McMath 42:35
that the privacy policy we wave when we sign into you never read, nothing we’ve never read. It’s like, okay, so Michael, we recently published an episode about the data about women and zone two specifically, and something I’m curious about with the meta analysis that you did, were all the subjects men? No,
Dr. Michael Rosenblat 42:58
I think there’s about 15% of the participants were female. And what would you make of that? I think it’s pretty typical for sports science studies. I think we’re gonna start seeing a little bit of a change as we start collecting data. I think it’s actually very important to see a change, because we need to know how we can help our athletes better. And 15% of the sample size, and so that would be, we had about 350 participants there. And so if we’re looking, you know, fit maybe 50 participants. I can’t do the math off the top of my head right now, even though I am a statistician, so there’s some, but that’s very small sample size. So we can’t really do enough of analysis or a subgroup analysis, specifically on female athletes to determine which distribution model might benefit one versus the other. So
Griffin McMath 43:47
no distinct conclusion that a coach of female athletes or female athletes could make from this.
Dr. Michael Rosenblat 43:55
It’s hard to say that I’m always nervous depending on the question. I’d say I did an analysis on looking at did sex play a role on these results? And the answer was no, but there was no statistically significant difference. So because of saying that, the sample size was too small, so yeah, I guess I’ve answered that question, probably we can’t make a definitive conclusion there.
Griffin McMath 44:15
Thanks also for saying sex and not gender. I appreciate you. All right, where are we at? Now
Trevor Connor 44:22
gonna shift gears a little bit, and first, I’m gonna apologize because you mentioned that you had a very long name for your study, and I realized we’ve been talking about this study and we have not given the title to the listeners. So first, do you remember the title off the top of your head? That’s
Dr. Michael Rosenblat 44:37
a good question, because people ask me such details, but so the title of the manuscript for the study was, which training intensity distribution intervention will produce the greatest improvements in maximal oxygen uptake and Time Trial performance, a systematic review and network analysis time
Trevor Connor 44:54
title performance in endurance. Athlete in endurance,
Dr. Michael Rosenblat 44:59
by the way. Right?
Trevor Connor 45:01
I remembering
Dr. Michael Rosenblat 45:04
that’s the first thing I wasn’t but added the endurance athletes at the end because I was Prisma guide said you have to also include the participant type, and so that I put that in at the end. So I struggling on
Trevor Connor 45:14
that one. And then last part was a systematic review and network
Dr. Michael Rosenblat 45:18
meta analysis of individual participant data. So it is a mouthful, and none of my colleagues like it, but again, I just met the criteria of Prisma guide.
Griffin McMath 45:26
You just made my heart rate jump so much that if any listener was just kind of slowly dozing, there’s no way I’ve just woken up at that.
Trevor Connor 45:34
I was certain you were reading because you were kind of looking at your screen there, but you missed that part. I am impressed you got most of the way through.
Dr. Michael Rosenblat 45:46
But yeah, I was literally in the peer review process that I think it was like, the second round is it was accepted, and I was like, I just, I think we should put that in. And so we put that in. At the end, I think it was in, and we removed it, and then we put it back in.
Trevor Connor 45:59
So I think the last thing that I want to bring up and discuss with you today, and this is talking to you, partially as a researcher, partially just as a coach and an athlete, there has been a big debate lately of polarized versus pyramidal. It has gotten at times very heated. And I really wanted to get your take on this, and we did an episode on this. There was that literally a debate between Dr Seiler and another researcher that was published. And I admit our conclusion was, this is a bit of a nothing burger, but I want to hear your thoughts on this. You have now spent a lot of time researching this. What is common between polarized and pyramidal? So
Dr. Michael Rosenblat 46:39
in terms of what’s common, the amount of time or the volume spent in Zone One. So there’s a huge amount of time, 70, 80% of the time below that first lactate threshold. And then if we were to go to where the differences are, it’s really the time spent in Zone One and zone two. And so pure middle more time spent in Zone Two, whereas polarized, again, there’s that more time spent in Zone One. It’s interesting because I’ve shifted so much away from being more of a cardiovascular exercise physiologist, which is where my my PhD started, and into more of a methodologist and evidence synthesis specialist. So I care more about what does the evidence say, and what can we use from it. And so, yes, it’s been somewhat there has been a heated debate, and it’s been going on for quite some time, but when it comes down to it, I feel like it’s kind of well, how can we best help our athletes? Like, what does the literature tell us, and in what context is it telling us that? Because it varies. I mean, you can see in certain contexts, one distribution might be better than another, but it depends on how you’re looking at it and why, but really it’s like when it comes down to it’s, how can I help my athletes get better? And that’s really all that should matter. And again, I’m also a physiotherapist, so I think of I want to help my patients get better. What’s the best way that I can do this? And that’s all that matters. It doesn’t matter about what I I think, based on what I like better, but it’s what intervention has been shown to be better in the literature, and then also what I see clinically, and those both those things are important. So clinically would be, as a coach in the field, based on what these results show, I would say a good place to start. For competitive athletes would be a polarized distribution. They may need more intensity to see improvement. And then I’d say maybe for recreational athletes, maybe a pyramidal model, as these results are showing, would be beneficial as a good place to start. And maybe it’s because they can’t handle as much intensity. And I don’t necessarily mean that, you know, is a pure middle model necessarily better than a polarized model for recreational athletes. It just might mean that that volume of high intensity exercise might just be a little too much for recreational athletes. Now, again, I’m only speculating, because we don’t know this right, and we’re just kind of looking at these results and saying, Well, what are some possible differences between these two these two groups, and those would be some of them. Another difference could be maybe the where their percentage of their second threshold is relative to their vo two max. So you’d say, well, elite level athletes, or competitive athletes, they have a much higher percentage, so they’re maybe about 90% of their vo two Max is where their critical power is. So maybe they need
Trevor Connor 49:13
to be pushing the top threshold. Can get very close to vo two Max.
Dr. Michael Rosenblat 49:17
Yeah. And so maybe they need more intensity to push the roof of the house instead of the second floor of the house. And so there’s some things to think of
Trevor Connor 49:24
there. I also noticed that doctor stole was one of the authors on your paper, and he wrote a great paper, I guess it’s going back to 2015 now, where he showed that in a lot of competitive endurance athletes, they tend to employ more of a pyramidal approach in the winter or the base season, as they get into the competitive season, switch to a more polarized approach. What’s your feeling on that? You
Dr. Michael Rosenblat 49:50
know, it’s funny, because I’m going to be looking at some data where there’s a 2022, study that came out that actually periodized different train intensity distributions. And so we’re going. To look further into that data at basically on their first and second thresholds of the athletes to see if there’s responders versus non responders. So is it possible? And again, I’m only speculating. I haven’t looked at the data yet. We don’t, I don’t have it yet, and I can only guess here, or give a good educational guess is maybe earlier on in the season, your threshold is going to be a little bit lower. And maybe there’s some specificity to doing more of a pyramidal model initially, and then as someone becomes more fit, maybe they need to start shifting towards a pure, polarized model. I don’t know. I mean, that could be wrong, right? We don’t have enough empirical data from that, but that doesn’t mean again, if you look at an entire approach to somebody trains, maybe if you look at the whole program, it might follow more of a polarized model. Maybe there’s times where it would shift depending on the needs of the athlete. But I can’t, I can’t say I know for sure. I know. I want to look at the data and see if there’s responders versus non responders, and does it matter based on where their thresholds sit? No,
Trevor Connor 50:59
that’s fair, and it’s a good point. I love that you say you’re a statistician. You’re like, I gotta crunch the numbers before I can give you an answer.
Dr. Michael Rosenblat 51:06
Well, even then the sample size is gonna be so small that it might just say, Well, you know, there’s some trends here, and it looks like there might be some responders versus non responders. And then you’d say, well, now we need a study of 500 people to actually figure this one out. So
Trevor Connor 51:19
I will tell you the reason we felt this debate is a bit of a nothing burger is because of how similar the two approaches are and what we were talking about before, which is you take the same training, and if you look at it by power, it could be polarized. You look at it by heart rate, it could be pyramidal, depending how you define the zones, it could be polarized or pyramidal, depending on how much you factor in cardiac drift, it could be polarized or pyramidal. When you’re having just little things like that. Make the difference between whether the training is pyramidal or polarized. Our conclusion was you’re doing pretty similar training here. It’s not like the differences you see between polarized and threshold or pyramidal and high intensity, it’s very minor differences. Yeah, I completely
Dr. Michael Rosenblat 52:04
agree with you. In fact, that’s part of the reason why I was very surprised to see a difference between the recreational and competitive athletes. Difference is so small. And actually, to add to that, that’s why, when you look at the results of the study and the way in which they calculated the different intensity zones with different different methods. Well, there’s going to be some error there. And the fact that there is no statistical heterogeneity means, oh, well, you can be a little bit off. It’s not going to matter as much. And so if you shift a little bit more towards pyramidal or a little bit more towards polarized, I don’t, I don’t think that’s going to play a significant role in how somebody’s going to respond. I do think, though, that how much time you spend in Zone One is a very important factor in training, and those are something that’s very similar with polarized and pyramidal and there’s substantial data retrospective analyzes that show that those who tend to be higher performing athletes tend to do a huge volume of training in Zone One, not even looking at the training intensity distribution literature, just generally looking at how athletes train, they tend to find and not only is it that it’s a huge volume, but those who are doing two hours or longer sessions as well tend to also perform better in aerobic endurance events,
Trevor Connor 53:21
just that value of doing that time at that just low intensity. So Dr Rosenberg, I guess I really have just one last question for you. We’ve been talking a lot about these different approaches. We’ve talked about your meta analysis. This is me really asking you, as a coach, taking a step back and thinking about distribution models, what would be your recommendations to our listeners, anything in general that you could recommend to them? So
Dr. Michael Rosenblat 53:48
it’s funny, talking about training intensity distributions, my recommendations may not even have much to do with the training intensity distributions, and it’s it’s really one is you need to be consistent with your training and don’t push it when you can’t go and so I think that that has a lot to do indirectly, with the training intensity distributions, which has a lot to do with a polarized model. Interestingly, you know, there’s a good place to start. Maybe, if you’re more of a recreational athlete, that’d be beneficial to maybe start with a pyramidal model. If you’re a competitive athlete, it might be beneficial to start with more of a polarized model, and test yourself and make sure you’re monitoring, seeing how your improvement is and how your recovery is, and adapt and change your program as you need
Trevor Connor 54:30
Okay, well, we’re going to finish out with our take homes, which I’ll explain in a second. But before we get there, we do have a question for our forum. Griffin. You want to throw the question,
Griffin McMath 54:42
I would love nothing more. Have you tried a polarized threshold or pyramidal approach to your training? What has been your experience? Give us your experience, your stories, your comments over in the forum, at forum, dot fast, talklabs.com.
Trevor Connor 54:57
All right. Well, Dr Rosenbaum, you’re new to the show. What. We finish out most of our episodes with our one minute take homes. So give you a second to think about it. But what is the most important lesson or message that you hope the listeners take from this episode?
Dr. Michael Rosenblat 55:15
Yeah, I’d say I think it just brings us back to the type of athlete that you are, and where to start, if you’re a recreational athlete, maybe be beneficial to maybe start with a pyramidal model. And if you’re competitive athlete, maybe start more with a polarized model. You know, if anything, you’re allowed to shift a little bit into one model versus another, yeah, I think just kind of see how your trainings going test yourself and making sure that you’re seeing those improvements.
Trevor Connor 55:44
I actually thought you’d go to the statistical significance thing, but that
Dr. Michael Rosenblat 55:48
was, yeah, well, I would, actually, I’d like to do that. I guess I thought it was more on the practical side.
Trevor Connor 55:54
So I’ll tell you what. I will actually make that mine when we get there, and then you can respond to me and finish it out. So Griffin, you want to go next?
Griffin McMath 56:01
I think one of my favorite takeaways, or take homes for today, was just that I didn’t know that I could laugh when talking about millimoles or hearing someone say post hoc very casually in conversation, it’s been a minute. So that was refreshing. But my meaningful take home, I think going back to the beginning, and I know that I typically in podcast episodes, I tend to refer back to conversation points at the beginning of an episode, and that’s because I am a beginner athlete in this so setting the table is so important. And it’s been a while on this podcast since we’ve taken a dive into methodology, so the concept of a network meta analysis and how that relates to or talking about today, is something that I think I’m going to start diving a little bit deeper in as we go through studies on this podcast. So that was my take home, and just what impact that could mean on the results. So
Trevor Connor 56:53
my take home is really the challenge you faced. I think exercise physiology research has really tended towards let’s look at the impact of this intervals or those intervals, because intervals are something where you only have to get somebody into the lab a couple times a week, and you can see the effects of intervals within two to six weeks. They’re relatively easy to study in a lab. When you are talking about something as big as How are you distributing your training? I think with the limits on time and most studies, you might be lucky if you get eight subjects more often than not, you’re not going to achieve statistical significance. But it’s really important for everybody to understand. And I get the feeling you’re going to jump in here. Dr Rosenblatt, but it’s really important for people to understand that saying we didn’t achieve statistical significance isn’t the same thing as saying there was no difference.
Dr. Michael Rosenblat 57:50
Yeah, and I think that’s an important take home, not just for this talk and this study, but whenever we’re looking or anybody, all any of the listeners here are looking at any of the literature, make sure that you see, you know what’s the effect size and what’s the P values, and actually look at these results, because you don’t want to misinterpret what it is that you’re reading, and say, Wow, doesn’t really matter. Well, we might not actually have the answer yet, and I think that’s an important take home when we’re looking at the scientific literature.
Trevor Connor 58:19
Well, Dr Rosenblatt, pleasure to have you on the show. Thanks for joining us.
Dr. Michael Rosenblat 58:23
Thanks so much for having me. It’s definitely been
Trevor Connor 58:25
great. That was another episode of fast talk. Thoughts and opinions expressed in fast talk are those of the individual subscribe to fast talk wherever you prefer to find your favorite podcast, be sure to leave us a rating and a review. As always, we love your feedback. Tweet us at at fast talk labs, join the conversation@forums.fast.labs.com or learn from our experts at fast talk labs.com for Dr Michael Rosenblatt, Dr Steven Seiler and Dr Griffin McMath, I’m Trevor, not a doctor. Connor, thanks for listening. You.
