Chapters
00:00 Start
00:38 Fat-Free Mass Study
01:46 The Researchers
02:23 My Key Disagreement
12:56 Basing Protein Intake On Fat-Free Mass
16:04 Protein Intake Effect While in Higher Deficit
22:59 What is the Highest Protein Intake to Maximize Gains?
Transcript
It looks like protein powder is back on the meal plan, boys. A new meta analysis recommends protein intakes as high as 1.5g per pound of total body weight per day. That’s 3.2g/kg, again of total body weight per day. In this video, I’m going to look at this new meta analysis to see how much sense these recommendations make if you should consume them. And at the end, I’m going to go into two additional findings that the new meta analysis found, which have flew under the radar of most people discussing this and I think are extremely important for most of you to know.
This new study was a meta analysis of 29 other studies, which means they basically looked at kind of the average effect that we see of protein intake on fat free mass changes. So basically our gains fat free mass is used as a proxy for muscle growth in this context. And importantly they looked at studies of non obese individuals in energy deficit. So they were cutting and they were strengthening already before the study. although the requirements only mandated that they had three months of resistance training experience. So many of the individuals were absolute novices. And the researchers looked at protein intakes over a range of 0.8g/kg, up to 3.2g/kg, and they found that there was a positive association between protein intake and gains in fat free mass in this range. the best model was a linear fit, which means that higher protein intakes result in more gains with no seeming limit. and therefore the researchers concluded that very serious trainees who care a lot about their muscle mass should consume the higher end of this range. So bodybuilders, for example, should consume as much as 50% more to even the pro bodybuilding wisdom of one gram per pound. But you might be wondering what researchers came to these preposterous recommendations, but they are actually some of the best researchers in the field. Martin, their follow the lead offer is an up and comer, if not personally, not to him, but he seems like a great guy. And I’ve read multiple of his studies already. Eric Drexler is definitely a great guy. I’ve talked to him many times. Great researcher Eric Holmes. I’ve known for a very long time. I can say our friends, I’ve met him personally many times. Great guy. I have absolutely nothing bad to say about these researchers, but I do not agree with their analysis and in particular their recommendations that flow from their analysis.
In this particular paper. my key beef in this case, is that a positive association in a certain range does not mean that you should be consuming the upper end of that range for maximum benefit. The range went as low as 0.8g/kg per day. We know that there are benefits to higher protein intake standards, so of course there’s going to be some positive association between that low of an intake and higher intakes. And that’s what we are really interested in is what the break point is. in our earlier meta analysis of this research, we found that the break points, although in slightly different subjects, was 1.6g/kg per day. The new meta analysis did not perform a break point analysis. So technically all this analysis shows is that there are benefits above 0.8g/kg, which is just the intake for sedentary individuals. And in that sense, the analysis teaches us nothing like truly novel. the reason that the researchers still recommend that we go up to the top of this range is that the best fit in the data was a linear model. Instead of a break point analysis, the researchers compared a linear model to a quadratic or a cubic model.
Now, if I have you think of the relationship between protein intake and our gains, what type of relationship are we thinking about? Do you think the relationship is linear, whereby it’s just the more protein you consume, the better your gains and there’s no cutoff point. Do you think that’s reasonable? Probably not. Right. Do you think it’s reasonable that the relationship would be quadratic. So maybe it goes up at first and then it kind of plateaus and then it goes down. If you start consuming super high protein intakes, you’re getting start to suffer until they’re just as bad as an a low protein intake. Probably also not right. So that would be something like a quadratic relationship. Or do you think the relationship would be cubic, which would entail that there are maybe inflection points and minimum and maximum along the curve. So maybe your gains start increasing at first and they taper off and then they go up again, something like that. Probably not. Right.
I’m guessing that the vast majority of you think of a diminishing returns curve where you consume protein intake up to a certain point. There are benefits for muscle growth, but at a certain point, enough is enough. You just can’t eat your way to the Mr. Olympia. You have to also do strength training. And there are a lot of other factors that influence how much muscle we build. Protein alone is just going to have some maximum at some point. And we also see this in research is called the muscle fool effect, which sounds like total science, but that’s the actual term in nutritional sciences. It means that there is a limit to how much protein synthesis we can stimulate, because yeah, you just can’t grow infinitely in a single meal or in a single day. So you would probably expect a curve like a ligase, which is basically what we found with our breakpoint analysis. Although a breakpoint analysis can also be argued to be a little bit crude. If you do it in a linear fashion, at least it stimulates the idea that we are looking for some point at which additional protein intake no longer benefits muscle growth. That was not done in the current study, and the researchers, because they only looked at three models, none of which makes any theoretical sense.
Honestly, they chose the one that’s just best best fit the data instead of best fit theory. And as a result of that, they picked the linear model, which is all defensible in statistical terms, but I think just doesn’t make a lot of sense. So assuming it would be a linear relationship, indeed it would be that more is better. But if you are going to recommend more is better on a linear relationship than the cut off point of your recommendation is simply going to be influenced by the maximum protein intake in the data set. Imagine there had been a one more study where for protein intake as high as 5g/kg or 2g per pound. in that case the relationship between protein intake and organs would probably have been basically the same. One additional study is not going to change the data set or the analysis outcomes by a meaningful degree. So we would have the same outcome still linear positive relationship. But now the range goes up to 5 or 2g per pound. And then they would recommend that range. Right.
So you see where I’m going. And if there was another study which had three grams per pound of protein intake, again the relationship and the regression outcomes would probably have been the same. And then the recommendation would have been, oh, so we have to go all the way up to 3g/kg without a break point analysis or fitting the data to some type of logarithmic or asymptotic curve that can capture the diminishing returns effects. We don’t know what the maximum point is at which protein intake starts no longer being beneficial for muscle growth. That’s what we’re really interested in. And unfortunately that was not adequately tested in this analysis. So without any type of break point analysis or looking for some asymptotic curve, essentially all what the analysis shows is that higher protein intakes above the minimum are beneficial. So that’s not very surprising. Another major limitation to the analysis is that only five studies were originally found based on the stringent selection criteria, but because that was too few to conduct a proper meta analysis, the researchers expanded their criteria and then found 29 studies. Unfortunately, and this is a limitation simply of the data that are available when you have these 29 studies, you’re looking at studies that don’t even investigate the effects of higher versus lower protein intakes. You’re looking at completely different studies. Some of these studies were on protein quality. Some of them were on nutrient timing. Some of them were on energy intake. Some of them were on ketogenic dieting or intermittent fasting. so.
Many of these studies are very hard to compare and have many confounding factors in many of the comparisons between higher and lower protein intakes in the studies that were actually even designed for this in the first place. There were also differences in meal frequency, nutrient timing, energy intake, protein quality, or any other confounding factor. So we’re not just looking at the effect of protein intake here. We’re looking at the effect of multiple other factors that we know influence muscle growth. One huge one is energy deficit. In this study, the researchers did look at the effect of energy deficit, but they did not do it in the best possible manner, in my view, because they looked at the effect of protein, and then they looked at the effect of protein, plus a bunch of moderators, which included biological sex of the participants, energy deficits and a bunch of other factors. they found that protein intake alone was actually a better fit. Again, this is an example of trusting the statistics over what makes sense. We have previous meta analytic data and multiple studies showing that greater energy deficits influence fat free mouse change. Not very surprising, right? If you go into a massive energy deficit, the risk of muscle loss is greater than if you are in a small energy deficit, regardless of what the statistics show, when we already know this as fact, pretty much. And theory heavily supports that.
You should keep this in the model. You cannot just blindly trust in anesthetics when the statistics tell you something that doesn’t make any sense, and is also not individually supported by well-controlled randomized trials. So that’s, I think, a huge limitation, especially given how confounded the data set was. And to the researchers, 100% credit. The researchers did note that The meta regression models that are exploratory, as included studies, may not have compared higher versus lower protein intakes in a controlled setting and involve intervention heterogeneity. The last part refers to the fact that many of the studies have habitual strength training, so the groups were not just different in protein intake, but also in the type of strength training like their volume, how hard they trained, etc..
Another very big confounder. An exploratory analysis essentially means that it’s more hypothesis generating, and that you’re just kind of exploring the data to come up with ideas for a more rigorous, controlled trial. So that’s not something the researchers are at fault for. But I think it massively limits how literally we have to take the results. Even if we ignore all of these limitations and we do take the results literally at face value, then we should look at the question how much additional benefit are we getting when we increase our protein intake beyond, say, 1.8g/kg? Is it really worth it to go higher in protein intake? The answer to this from the researchers is increasing protein intake from the lowest 0.8g/kg to the highest 3.2g/kg. Analyzed intakes results in only a small effect size difference. So even if we go from the paltry lowest intake, which is basically for sedentary individuals all the way up to the super high intake, 1.5g per pound, which is like, you know, protein powder at lunch, dinner and breakfast. Then we are still looking at the small effect size difference for your gains. So when we’re looking at just going from, say, one gram per pound or 1.8g/kg to a bit higher than that, you’re looking at a very small effect size difference. And these effect sizes might actually be a bit of an overestimate, because the researchers made a number of statistical choices. The bias, the results in favor of a stronger relationship between protein intake and factory mask gains.
For example, they looked at energy deficits and some of them were implausibly large. so they replaced those with the median. And they found that that resulted in improved model fit. Similarly, there were a couple of outlier studies or outliers because they were outliers only to the extent that they significantly influenced the effect sizes, and removing them improved the relationship between protein intake and fat free mass change, removing outliers because they improved the model in the sense that you want it to be so they basically support your hypothesis is not, in my view, how you should remove outliers. You should remove outliers based on predefined criteria such as a value being over two standard deviations away from the median. That would be a reason not to include the specific study, but just not including a study, because it does not confirm your hypothesis is basically data mining. That said, the analysis was overall very well conducted. I highly respect these researchers. I don’t think there was any intentional type of cherry picking taking place, but it is a fact that these choices a bias the results in favor of stronger relationships. So probably there would be even a slightly smaller effect. I don’t think these would have a big effect or anything in the analysis, but it is a consideration when we were trying to extrapolate whether it’s worth increasing your protein intake.
There were two additional very cool findings in this meta analysis. the first is that using fat free mass to bayshore protein intake on instead of body weight was no more accurate at predicting the relationship between protein intake and fostering mass changes. In other words, it does not appear to be worthwhile to go to the trouble of trying to estimate your body fat percentage, calculate your fat free mass, and then basing your protein intake on your fat free mass. Most people cannot accurately calculate their body fat percentage, most people dramatically underreported, and most commercially available methods are not reliable at all, so it’s probably not worth getting that level of reliability and complexity to estimate your protein requirements.
The model fits statistics, which are the most important things here were virtually identical between protein based on body weights and protein based on fat. Free mouse 54 to 55%, so only a 1% difference in R squared, which is the explained variance of the model. And the standard error was exactly the same at 0.09. So there was really no difference between the relationship between protein intake based on body weight and fat free mouse and change in fat from us Interestingly, the researchers still concluded that fat free mass was better, but they did that based on the 2%. Again, a completely trivial difference in the probability that additional protein intake benefited fat free mass change. Again, this is biased in favor of assuming that there is a relationship between protein intake. A fat free milestone at a stronger relationship means a better model it fits the hypothesis better, but it doesn’t actually fit the data better. What matters for whether the model is better is how well the model fits with the data. And that’s things like R-squared and standard error, which were virtually identical between the two considerations.
Now, I agree that theoretically fat free mass makes more sense than total body weight, but the available research does directly look at this, which is very, very little research does not find very significant effects of fat free miles at all. The reason for that is probably that, yes, fat free mass, like a kilo of muscle mass, is more metabolically active and requires more protein than a kilo of adipose tissue like a kilo of body fat, but both of them completely pale in comparison to organize. Organ mass is by far the strongest predictor of protein requirements. A kilo of organ mass has a far higher protein requirement still than a kilo of muscle mass. Since organ mass scales with body weight and height in particular relatively stable, it just doesn’t appear to be worthwhile to look at differences in one kilo of muscle mass. It’s just like the effect of an additional kilo of muscle mass on your arresting energy expenditure is not as large as some people would like to believe. So if you’re not overweight, most people can simply use total body weights to calculate their protein requirements. And it’s probably just as accurate. In fact, if you cannot reliably determine your body fat percentage, which most people can’t, it might be less accurate to use fat free mass because of the inaccuracy of knowing what your fat three mass index is in the first place.
Now, I will say that the effect of protein intake on fat free miles change increased the lower the body fat percentage of the individuals was, but this was the case in both the models based on fat free mass and body weight, so it may not be related to fat free masks. It might be related to inherent risk of muscle offsets, lower body fat percentages. The second really cool finding in this analysis is that they did a subgroup analysis of smaller and lower deficits, in this case below or above 300 kilocalories. And they found, interestingly, that the effect of protein intake, if anything, decreased when people were in higher deficit. So the thinking of most people in evidence based fitness, myself actually not included, is that higher energy deficits warrant greater protein intakes because it has to offset the risk of muscle loss. That doesn’t logically follow, though, because protein intake and energy deficit likely have separate effects. that. There is only a certain amount of protein that your body will use, so you cannot necessarily compensate for excessive energy deficits by just consuming more protein. It has its own independence and negative effect. The limitation in that case is not protein availability, but the limitation is simply energy availability. and if that results in excess catabolism, it doesn’t seem to be the case that the additional protein can compensate for that. Either way, these results don’t support the theory that higher protein intakes are required to energy deficit.
I know that some people thought that this was the case in this meta analysis, because this matter analysis finds seemingly high protein requirements for at least the researchers recommend them based on research on individuals and energy deficits. But they did not compare protein requirements in individuals outside of energy deficits and in energy deficits. So we cannot say that this analysis supports higher protein index and energy deficit. The subgroup analysis that most directly looked at this, if anything, found that the effect of protein intake on fat must change decreased. So all in all, this analysis does not support that any energy deficit or greater energy deficits. We need higher protein requirements. Now, I might be inherently biased. I wrote an article a long time ago arguing that 1.8g/kg, or 0.8g per pound of total body weight, is the highest protein intake that’s likely required to maximize your gains. This article at the time was based on over 20 studies and nitrogen balance data. And I note up Michael Wolf and Greg Knuckles have taken a subtle shot at me by saying that I rely on nitrogen balance data when there are newer, better methods available.
There are newer, better methods available, but not when I wrote the article over a decade ago. This became quite a prominent article, which I became well known for. It ranked number one on Google for a long time, and it had a big impact on the industry. it in part inspired the meta analysis that I was a part of later, which found the exact same thing as the article, that there is no established research finding benefits over 1.6g/kg per day, so I recommend at 1.8 before the meta analysis already. And I stuck with that after the meta analysis, because it already includes a bit of a markup to account for uncertainty in the data. at the time I wrote the article, and even at the time or the meta analysis was published many years later, those were the most evidence based guidelines based on the available data.
I was the founder of Bayesian bodybuilding, the method of which highly relies on Bayesian thinking, which means that you update your beliefs based on new incoming knowledge. You have a prior based on the data that you already have. New evidence comes in. You update that to form your new belief that takes all the data into account. Since then, there has actually still not been a single, well-controlled trial that has found significant benefits of protein intakes. About 1.6g/kg per day. So what we’re looking at here is what are we can extrapolate from individual studies that did not find significant benefits. Maybe there were some benefits that did not reach statistical significance. And with a meta analysis, we have more statistical power to tease out if maybe all of these small differences were actually consistent enough that we can say there is, in fact, a difference to higher intakes than any controlled study has shown so far.
Also, I would note that there have been some trials published that did find benefits of higher protein intakes, But all of them were confounded by nutrient timing and in most cases also energy supplies. So we cannot say whether it was the difference in energy intake, protein timing or protein intake that made the higher protein intake group have greater gains in those studies. And my memory. That’s only three studies out of a total of about 90 studies. If we don’t control for, confounders that have looked at this, many of which found that in terms of statistical significance, at least even far lower protein intakes, still optimized body recomposition and strength development. So individual studies have not influenced my belief much. It really is up to these meta analysis to see if we can, using statistical wizardry, find out if there are trends in these data that individual studies don’t find to be statistically significant. But maybe in a meta analysis we can tease out there are still some benefits. So it depends on how much stock you put into these types of statistical exercises. and this new meta analysis also was one of these cases.
Another new line of research is the indicator amino acid oxidation research. The ideal method. We have free reliable studies that look at protein requirements based on this method. This is basically a method that looks at the protein intake that’s required to maximize whole body protein synthesis. This is not necessarily a protein intake that maximizes muscle growth. That might be lower, because it’s possible that some protein synthesis can still be increased in, you know, collagen or connective tissue or other types of tissues that muscle mass. But still, it’s a very reasonable proxy.
The first study by Bandwagon Athol in 2017 estimated a protein intake of 1.7g/kg, maximized average whole body protein synthesis on long training days in male bodybuilders. So this was right in line with the data that we had at the time. In 2019, Moloney at all estimated a protein intake of 1.5g/kg per day maximized average will bio protein synthesis on training days in female lifters, so that’s even lower than the protein intakes required. Based on this other research. But then in 2020, Mazwai et al. Estimated a protein intake of 2 gram/kg that’s necessary to maximize whole body protein synthesis. But this wasn’t training days in male bodybuilders. On training days, we know that protein synthesis is higher because that’s when you’re at the peak of the anabolic window. And individual studies show that we can have more protein in that period, and it will increase protein synthesis. So when you look at just the anabolic window period, the post workout time, you extrapolate that to your total protein requirements on average across day, you’re going to overestimate your protein requirements in that sense the two on training days. So maybe 1.6 on rest days would fit perfectly with an average of 1.8. In fact, if you look at the range here, it’s 1.5 to 2, based on the free studies that we have on this, and that fits perfectly with the available other data that we have. I’d also note that the new method is better than nitrogen balance, but it’s also absolutely not perfect. And the assumptions and the calculations used significantly influenced the estimates. For that reason, some people look at the confidence interval around the estimate and go with the higher end of the 95% confidence interval. And they will argue that this is because of individual variation in protein requirements.
This is flawed. The confidence interval is influenced not just by individual differences. Imposing requirements, but also by the error of the measurement technique, by whether the subjects slept well, by how hard they train, by how much protein they had in the days before, whether they were cutting or bulking, not just their individual actual physiological protein requirements. So yes, you could argue based on the uncertainty in the data, not just looking at the question, what is it likely the optimum protein intake, but we’re looking at the question, what could conceivably be the highest protein intake that could be used to maximize our gains? That’s a bit of a different question. It’s all about the level of margin of safety that you want in this case to maximize your gains.
So again, you could use this estimate that I mentioned earlier based on the current meta analysis, whether it’s worth it for you to increase your protein intake substantially to see if the effort and money for the protein is worth the potential addition in gains. All in all, regardless of whether we look at randomized control trials, where we look at nitrogen balance data or the indicator amino acid oxidation methods data, most of those research findings are in line with our protein intake about 1.8g/kg per day, or 0.8g per pound. That will maximize gains and fat free mass. This is the minimum protein intake you have to consume every single day to probably maximize your gains. If we are looking at the highest level found in any breakpoint analysis or asymptotic research, then we’re looking at levels by Missoula at all n gram knuckles and at 2g/kg. this is higher than any well-controlled randomized controlled trial actually supports. So we’re really looking at, a difference in the question of what level of protein intake maximizes muscle growth based on direct research, versus what protein intake, based on the limitations in research, could conceivably maximize muscle growth. And then we’re looking at these higher ranges.
For most individual purposes, though, the additional gains you will get from increasing your protein intake above around 1.8g/kg will be very small. I’m going to conduct another meta analysis based on the methods that I think are superior to see if there is, in fact a benefit of higher protein intakes based on the available research. I don’t know if the research will allow us to tease out that difference. But at this point, really the controversy restaurant, the fact of how much margin of error and margin of safety you need to make sure that you’re maximizing your gains if you want a reasonable estimate. We’re probably looking at the range of 1.6 to 2g/kg per day. Or you could go with the old wisdom of one gram about if you want to be pretty safe. However, if you want to be like ultra, ultra safe and on your death bed, you think, oh, I really regret that I did not reach my mathematics one month earlier than I could have, or I could have gained, you know, one kilo extra fat free mass over the next 3 or 4 years. Then you might want to go with the higher estimates of some of the researchers, like the current math analysis. if, however, you want the best reasonable estimate of what the likely optimum protein intake is, you’re looking at a level of 1.6 to 2 gram/kg based on those research. And then yes, there is a small chance of missing out on a small amount of gains. The research simply does not allow us to say with 100% accuracy, this is the optimum protein intake for anyone. This is exactly how much you should consume.
It doesn’t work that way. There will be differences based potentially on body fat level, potentially on biological sex, almost certainly based on things like muscle memory, whether you are vegan, the level of protein quality in your diet, and individual variants. So unfortunately, based on the current data, we cannot give a heart one size fits all number. I know many of you want that, but it’s simply not how science works. If this type of evidence based content does not make you depressed or bored, I’d be honored. If you like, I subscribe.
