Tuesday, August 18, 2015

More Thoughts on the Recent Low-fat vs. Low-carb Metabolic Ward Study

The recent low-carb vs. low-fat study has provoked criticism from parts of the diet-health community.  Let's examine these objections and see how they hold up to scientific scrutiny.

Before we begin, I just want to re-emphasize that any way you slice it, this study definitively falsifies the version of the carbohydrate-insulin hypothesis that states that carbohydrates must be reduced for fat loss to occur.  Here is a quote from Why We Get Fat:
Any diet that succeeds does so because the dieter restricts fattening carbohydrates …Those who lose fat on a diet do so because of what they are not eating—the fattening carbohydrates.
This hypothesis is readily testable, and Hall's study directly tested it [note: Hall asked me to clarify that the study was not specifically designed to test Taubes's hypothesis, it just happens to do so].  In this case, "fattening carbohydrates" did not prevent a full pound of body fat from evaporating in six days when dietary fat was specifically reduced (1).  This is despite the fact that the low-fat diet was high in sugar (170 g/day; 35% of calories).  This hypothesis has previously been falsified by many other studies, but this new study puts a particularly definitive nail in its coffin.

It is true that this study didn't falsify every possible version of the carbohydrate-insulin hypothesis, of which there are many.  For example, if your hypothesis is that eating carbohydrate makes you hungrier and makes you eat more, and the resulting increase in calorie intake causes weight gain, this particular study doesn't undermine it because calorie intake was strictly controlled.  But again, this study was not intended or designed to test that hypothesis.

OK, on to the critiques.

1.  The study was too short.  Six days isn't long enough for fat adaptation.

This is the big one that people keep bringing up.  The claim is that six days isn't nearly enough time for fat adaptation, so the changes in body fat mass they reported are irrelevant.  Let's have a look.

Now, before we dig into this question, let's get clear on what we're talking about.  "Fat adaptation" refers to the process of shifting to using fat as the body's main source of energy*.  This happens when the diet shifts from carbohydrate-heavy to fat-heavy, or when we're fasting.  This process is associated with measurable metabolic changes.

The question is, how long does it take for those metabolic changes to occur?  Keep in mind that what we care about here is not how foggy your brain feels, how hungry or cranky you feel, how much energy you feel like you have, or how hard you can exercise.  Those things are all irrelevant to the question at hand.  For the purposes of evaluating this study, what we care about is how long it takes for the body to maximize its ability to burn fat.

Scientifically speaking, the claim people are making is that six days isn't long enough for fat oxidation to reach its maximal rate.  In other words, six days isn't enough time for the body to adjust to burning fat, so Hall's volunteers weren't yet able to tap into their own fat reserves effectively (this concept is shaky to begin with; see discussion below*).

Fortunately, we have sufficient evidence to evaluate this claim.  Some of the most relevant data I found are from a 1972 study of prolonged fasting in people with obesity, by William Bortz and colleagues, that Kevin Hall sent me (2).  Their study included indirect measurements of the rate of lipolysis, in other words, the rate at which fat exits fat tissue**.  These measurements reveal how long it took their volunteers to reach the maximal rate of lipolysis, which corresponds approximately to the maximal rate of fat oxidation.

I've graphed the data out so you can see the results.  On the horizontal axis, we have the duration of the fast in days.  On the vertical axis, we have the lipolysis rate:

What you can see is that the lipolysis rate ramps up and then plateaus quickly-- in as little as two days-- and then remains stable out to 23 days.

Here is another graph showing the oxidation of fat, carbohydrate, and protein over time during a prolonged fast, from a textbook chapter that Kevin Hall wrote (3):

As you can see, fat oxidation is fully ramped up after 3 days of fasting.

So the consistent picture that emerges is that the body oxidizes fat at the maximum rate within 2-3 days when it is completely deprived of dietary carbohydrate, including in people with obesity.  That is less than half the six-day duration of Hall's study.

Furthermore, in Hall's study the volunteers weren't completely deprived of carbohydrate.  People in the reduced-carbohydrate arm were still eating 140 grams of carbohydrate per day.  Such a modest degree of carbohydrate restriction requires a lot less fat adaptation than a total fast!  We might expect them to achieve maximal lipolysis and maximal fat oxidation even sooner.

But let's stop speculating, because Hall's team actually measured fat oxidation over time!  In figure 2G, they report the fat oxidation rate on each day of the study for both diets.  Have a look for yourself (RC = reduced carbohydrate; RF = reduced fat):

Both according to Hall's model (line) and the observed data (points), fat oxidation in the reduced-carbohydrate group increased rapidly and reached a plateau by day four-- and possibly as soon as day two.

These data allow us to definitively reject the claim that six days isn't enough time to adapt to burning fat.  Six days is more than enough time for the body to adapt to withdrawing fat from fat tissue and burning it at the maximal rate, including in people with obesity.

Now, I agree that we have to be careful about extrapolating these findings to longer periods of time.  There is still room for longer-term studies to provide direct evidence on what would happen over periods of weeks or months.  But the evidence clearly indicates that it is not possible to dismiss the short-term fat loss results of this study on the basis of insufficient time for fat adaptation.

2.  The primary reason the low-carbohydrate group lost less body fat is that they were burning through their glycogen stores.

This is a good point, and I think it's basically correct.  It is exactly what Kevin Hall's model predicts.

Let me walk through the argument.  The average lean human body contains about 1,800 kilocalories (kcals) of carbohydrate, in the form of glycogen stores in liver and muscle tissue (Keith Frayn. Metabolic Regulation. 2010).  Obese bodies contain somewhat more than that.

Normally, this stored carbohydrate is used to fuel brain and muscle metabolism.  When a person begins a fast, glycogen stores are rapidly depleted in the first few days, and as they go away, the body switches to fat as its primary energy source.  A low-carbohydrate diet is basically a milder version of the same process, and when a person goes on such a diet, the body initially taps into its carbohydrate reserves to make up for the carbohydrate shortfall.  The less carbohydrate the diet contains, the more glycogen stores are depleted.

So anyway, this glycogen contains calories, and every glycogen calorie the body burns displaces a calorie of fat that would otherwise have been burned.  In Hall's study, my calculations indicate that the low-carbohydrate diet caused people to burn 1,920 more kcals of carbohydrate than they ate over the 6-day period.  In other words, they burned 1,920 kcals of their glycogen reserves, most of that in the first four days.  This is consistent with the fact that they lost water weight, which is a sign of glycogen depletion on low-carbohydrate diets.

Now, here comes the interesting part.  If we convert the difference in fat loss between groups into calories, we see that the low-fat group lost 1,962 kcals more body fat than the low-carb group over the 6-day study.  That's almost identical to the 1,920 kcal loss of glycogen, suggesting that the glycogen they burned did indeed displace an amount of fat that could roughly explain the difference in fat loss between diets.

Together, this suggests that glycogen depletion in the first few days of the low-carbohydrate diet is the primary reason it caused less fat loss over the 6-day period.  Without glycogen depletion, fat loss would have been more similar between diets, although Hall's model predicts that the low-fat diet would still have maintained an edge.

Since glycogen stores are modest, glycogen depletion can't go on for very long, and its effects on body fat mass become negligible in the long run.  So it is true that the long-term difference between diets is predicted to be smaller than the 6-day difference Hall's team observed-- a fact they discuss in the paper. Yet the model continues to predict somewhat of a long-term advantage for the very-low-fat diet, primarily due to the fact that carbohydrate has a protein-sparing effect that sustains lean mass and energy expenditure.  Longer studies will be necessary to evaluate that prediction.

So yes, glycogen is important, but this in no way undermines the findings or conclusions of the paper.  It just means we have to interpret the results a bit to understand their full implications.

3. The study controlled calorie intake, so it missed the effects of carbohydrate intake on appetite.

This, of course, is true, but it misses the point of the study.  The purpose of the study wasn't to examine the effects of carbohydrate on hunger or food intake, it was to determine whether dietary carbohydrate suppresses fat loss independently of its calorie content.  If calorie intake hadn't been controlled, the study wouldn't have been able to test this hypothesis, and it wouldn't have provided any new evidence.

4. This study is part of a low-fat conspiracy to hide the truth that low-carb is superior in every way.

Give me a break!


There's a lot to chew on with this study-- it just keeps on giving.

I hope it's clear why, despite vociferous objections from certain parts of the diet-health community, this study and its conclusions remain fundamentally sound.  Yet at the same time, they do require some interpretation to fully understand.

I also hope it's clear why this study directly falsifies the carbohydrate-insulin hypothesis-- at least the version that proposes that carbohydrate restriction is required for fat loss.

*As an aside, I don't think I even believe the concept that the body has to go through an adaptation period to be able to primarily burn fat.  It can primarily burn fat at any time, but whether or not it does so depends on what other fuels are available, because it preferentially burns carbohydrate when it's around (likely because the body's storage capacity for carbohydrate is quite limited, whereas it can store almost unlimited fat).  The only reason it doesn't burn primarily fat immediately when dietary carbs run out is that it's burning stored glycogen.  As soon as that runs out, it's on to fat without a hitch.  There is no period during the transition to primarily fat burning where the metabolic rate drops, suggesting that the body is never struggling to get enough energy out of fat tissue.  The body appears to immediately withdraw as much fat as it needs to meet an energy shortfall, whatever the situation.  What I can believe is that this process of transitioning to predominantly fat burning causes symptoms like brain fog and reduced physical performance, as tissues adjust to the new fuel source.  But this doesn't mean the body isn't burning fat effectively yet-- it definitely is.

** They measured glycerol turnover (= Ra), which is a marker of lipolysis.


Puddleg said...

There are two ways in which this experiment needs to be improved; firstly, start with a baseline diet in which carbs and fats are evenly balanced. 40%CHO 40%FAT 20%PRO would do and because it would be a reasonable approximation of the Western diet, pre-diet heart interference, would also have additional relevance to the hypothesis. Then try say 8% CHO vs 8% FAT.

Secondly, run it for long enough for any changes to be noticable. Also make sure all the subjects eat the right diet and that improbable results are not excluded (if it's a metabolic ward study improbable results need explaining).

Some additional improvements would be, not trying to falsify a heuristic statement in a book (which may well be true for every study published at the time the book was written, and still true with regard to any trial with the kind of long-term or significant effects anyone cares about), because there are obvious exceptions; It's possible that even Gary Taubes suspects that someone eating 100% sugar ad lib will lose fat mass eventually, even though this would refute his hypothesis as stated.

The strength of Taube's statement is shown by the extreme that Hall had to go to to refute it. But it seems to me that the statement "Any diet that succeeds does so because the dieter restricts fattening carbohydrates …Those who lose fat on a diet do so because of what they are not eating—the fattening carbohydrates" is not refuted in that form because Taubes used the word "success" which in this context would be defined as the meeting of personal or medical goals. The context is different from a trial where results are measured in grams and invisible to the participants (the statement also does not mention insulin).

Carl M. said...

The quote from Why We Get Fat is laughable indeed. All you have to do is check out some fruitarians. Fat fruitarians are few and far between. Plenty of "fattening carbohydrates" in a fruitarian diet.

Having once lost 30 pounds without trying in a couple of months on a high sugar diet back in grad school days, I find such overly broad statements to be not worth refuting. (It was an approximation of an eating by instinct raw food diet, where I did cook the meat.) The excessive weight loss is the main reason I quit the diet -- it had some other rather nice health benefits, at least in the short run.

Perhaps a high sugar low fat diet which consists of processed carbs lacking fiber is more fattening than a high fat diet. But I suspect is has more to do with lack of satiation than calorie burning rates.

Roland Denzel said...

Thanks for clarifying the study ever further, and a big thanks for writing in a way that non-scientists can easily understand. It's much appreciated.

Brandon Berg said...

Regarding your first footnote, it's with noting, I addition to the points you made, that even people on fairly high-carb diets are burning fat every day. If that weren't the case, then all dietary fat would be stored in adipose tissue, resulting in extremely fast fat gain.

For example, a person on a 2000-calorie diet with 30% of calories from fat is consuming 600 fat calories per day. If the presence of dietary carbohydrates were preventing that person from burning fat, then that person would be gaining a pound of fat every six days, or sixty pounds per year.

Rap said...

Thanks for pointing out, in response to my earlier email, why the comparison ended up being between very low fat and moderately low carbs, which is because the subjects were already eating less fat in the baseline condition. But this then begs the question, as someone earlier commented, why Hall didn't avoid the confound by equating fat and carbs during the baseline phase. As is, it's obvious why the LC crowd is upset. Yes, the study tests whether carb restriction is necessary for fat loss, but carb restriction isn’t needed for that; seeing an effect with fat restriction alone is sufficient. Presumably, the carb restricted condition was included to compare the effect of carb restriction versus fat restriction. But if you’re going to do that, then the confound becomes a major problem, especially when it can be perceived by LCers (quite rightly) as possibly minimizing the effect of carb restriction.

As for the lack of an effect for the “gold standard DXA analysis,” the article states that “While both diets led to significant decreases in DXA-determined fat mass compared to baseline …, DXA was not sufficiently sensitive to detect a significant difference in fat mass change between the RC and RF diets” (p.3). The authors marshal some evidence for that possibility, but the other possibility, which they don’t acknowledge, is that the DXA analysis – and the high variance it found -- is in fact spot on with what happened. It would be interesting to see what the DXA results were for each individual in each condition. Do you know if that data is available and what it looks like?

Despite my critique, I actually think that fat loss can occur by fat restriction (or calorie restriction) alone and I’m surprised that Taubes would make such a narrow prediction – IF that’s what he meant. I’ve seen very good researchers, in discussing controversial issues, carelessly drift into what seems like an extreme statement but is really only meant as something more general – with little awareness of how their opponents are dissecting every word. But I also hope that Taubes did mean something more extreme. It’s a nice clean, testable prediction, and in fairness to Taubes, he seems to couch his LC view, at least in what I’ve read, as a hypothesis that needs to be tested.

Unknown said...

@rq - I also think that is very interesting. In other studies, obese, but not insulin resistant, people did better on LF because they can handle the carbs. On the other hand, IR people did better on HF.

Its pure speculation from my side that somewhat of the same is going on here.

Guido said...

Table 4 shows insulin data:
Insulin (pg/ml)
Baseline 12.6 ± 2
Reduced Carbs: −2.76 ± 0.77
Reduced Fat: −2.04 ± 0.8

So the low-fat group reduced insulin (almost as much as the reduced-carbs group), with the same amount of carbs as in the baseline diet.

It is possible to burn fat and reduce insulin without cutting carbs. That seems to be an interesting finding even though the low-fat diet is not a real-life diet.

1) How does low-fat diet reduce insulin if carbs are constant in relation to baseline diet?
2) Is reducing insulin still the only mechanism or are any other mechanisms at work? If yes which?

If the researchers wanted to validate whether carb reduction was required to burn fat loss, why introduce a low-carb group. Why not all 19 on low-fat. That would have generated more robust data.

raphi said...

Why not go for an isocaloric reduction in carbohydrates or fat, keeping protein % constant, from a baseline diet of 40% carbs, 40% fat and 20% protein?

Anonymous said...

If you copy a text from an author, please don't remove relevant parts.

Taubes is writing about Weight Watchers, Dean Ornish and the A to Z trial and he writes:

Doesn’t this mean that some of us get fat because we eat carbohydrates and get lean again when we don’t, but for others, avoiding fat is the answer? The simple answer is probably not. The more likely explanation is that any diet that succeeds does so because the dieter restricts fattening carbohydrates, whether by explicit instruction or not. To put it simply, those who lose fat on a diet do so because of what they are not eating—the fattening carbohydrates—not because of what they are eating.

The authors of the Hall et al. article removed the first part, the one that says "The simple answer is probably not. The more likely explanation is that" and they said instead "One influential author concluded that". You even capitalize the first word "Any" to make us think the setence starts with that word. You are creating a dogmatic sentence valid for all diets from one that starts with "The simple answer is probably not. The more likely explanation is that" when talking about commercial diets.

Taubes wrote what he thought was a likely explanation for the eventual success losing weight with commercial diets. He didn't say it was impossible to lose weight following a ultra-low-fat diet for 6 days. Who denies you can lose weight in the short term with ANY hypocaloric diet?

Why did you (and Hall et al.) remove the first part of Taubes text?

Tucker Goodrich said...

"Yet the model continues to predict somewhat of a long-term advantage for the very-low-fat diet, primarily due to the fact that carbohydrate has a protein-sparing effect that sustains lean mass and energy expenditure."

Really? The studies I've seen in suggest the reverse...


Stephan Guyenet said...

Hi rq,

Yes, the DXA results were not significant. However, the difference in fat loss as calculated by fat balance (much more sensitive) was significant. DXA is a great method but it requires larger changes to detect.

You appear to be correct that the subjects were probably surprisingly insulin sensitive for being obese (although it wasn't measured directly). As a general rule, I think results from insulin-sensitive subjects would be less applicable to people who are insulin resistant, and even less applicable to people with diabetes. However, the totality of the evidence suggests that carbs/insulin never trump calories in their effects on fat mass, irrespective of a person's level of insulin sensitivity. It is possible that differences in glucose homeostasis associated with insulin resistance could lead to different appetite responses to carbs though (remains to be tested).

Hi Carl,

I agree it isn't really worth refuting. To be clear, Hall's study wasn't designed with that particular hypothesis in mind, it just happened to test it.

Hi Brandon,

Absolutely-- that's a good point. I suppose a good way to put it is that in the long run, anyone with a stable body composition is oxidizing (or excreting as urea) exactly the amount of carb, fat, and protein they ingest. There is no other possible scenario.

Hi Guido,

Those are fasting insulin measurements. If you look at the C-peptide data, which represents 24hr insulin levels, they suggest that total insulin secretion didn't change in the LF group but declined by 22% in the LC group. That being said, you are absolutely correct that insulin levels can go down on a high-carbohydrate, low fat diet, particularly if a person is in negative energy balance and losing fat. The relationship between diet and insulin levels is a lot more complex than it's made out to be by certain people in the low-carb camp.

Also, they did put all 19 people on the low-fat diet. Since it was a crossover design, each person completed both diet periods.

Hi Raphi,

I think that would be interesting. Hall's model predicts that fat loss would have been more similar between groups in that scenario. The reason he started with a higher-CHO diet for the run-in is presumably that it resembles the typical baseline diet.

Hi novuelvoaengordar.com,

You are accusing me of cherry picking quotes out of context and misrepresenting Taubes's position. Nonsense. First of all, the additional text you added to the quote doesn't change its meaning at all. It still articulates the same hypothesis, all it adds is language like "probably" and "likely", implying less than 100% confidence in the hypothesis.

Taubes has repeatedly, specifically articulated the hypothesis that only carbs matter for fat loss, not calorie intake. I recommend you go to his blog and read his posts. It may be a silly hypothesis but it's definitely what he wrote.

Hi Tucker,

For one thing, it's a mathematical prediction-- it remains to be tested directly. I'm extremely impressed with the predictive power of his model, but it's not a substitute for actual data, particularly in uncharted territory. The second point is that this prediction applies specifically to the macro breakdown they used in the current study-- i.e., super low-fat vs. moderate low-carb. It doesn't necessarily apply to other macro comparisons.

Unknown said...

Cherry picked subjects who were obese and had good insulin sensitivity. Not enough participants. Too short. Low carb diet wasn't, ya know, low carb. It seems those that conducted the study had an axe to grind with the low carbers. The last of the low fat propagandists? Well, these dinosaurs should all be dead soon. Taubes' statement still applies to those with IR. The participants in this study are not the norm. It's common knowledge that those with good IS do fine on low calorie diets. They can easily switch from using glucose for energy and fat for energy. That does not apply to many (most?) obese people.

Unknown said...

Did they release the individual data points for each participant since there were only 19, or just the group averages? Would be curious to see how much individual variation there was.

Anonymous said...

Hi Stephan,

the context of Taubes' text is:

- Weight Watchers, Dean Ornish, A to Z trial diets
- Long term weight loss
- People who do have success on those diets
- He speaks of a likely explanation for that success, in his view

Is he saying that you can't lose fat the first 6 days of a hypocaloric diet? No, he isn't saying that. That idea is nonsense because we all know all hypocaloric diets work in the short term. He is not saying that.

So, how is it possible that Hall et al. say they have proved him wrong? Because they use a straw man. They take out of context Taubes' words and turn them into a dogma they can easily falsify with a 6-day diet (no matter the composition of the diet). That is nonsense. They didn't prove him wrong, they only proved they wanted to attack him.

May be you are right and Taubes doesn't think you can lose body fat the first days with a high-carb hypocaloric diet, but I don't think it is fair to use those words taken out of context. Hall et al. shouldn't have done that.


Stephan Guyenet said...

Hi Brandon,

I re-read my post and I see how my asterisk could imply that people only burn carbohydrate when they're eating a typical diet. I added the work "primarily" in several places to correct that.

Gearóid said...

There's one factor in all of this, which is very relevant to me and that is the effect on hunger. My family tend to be fat and my father and all his 8 siblings were varyingly overweight and his brother and mother died of DM2 (a long time ago).
I've been on low fat diets in the past, even losing 42lb on one and was absolutely starved. I regained all within a year of starting.
I've now been on intially no sugar and then LCHF (that for 1y 8m) and have had my weight down 40lb for 15m. I don't get hungry and I find it easy. I restrict myself to 50g carbs a day and eat all I want of other food. This controls my appetite.
This trial is fundamentally specious and is so because of the caloric restriction.
If you did this trial this way...
Calorie restriction on the LF and real low carb (<50g/day) and no calorie restriction on the LC, you'd get something useful out of it.
You cannot and must not ignore hunger. Hunger will win.

David said...


"You are accusing me of cherry picking quotes out of context and misrepresenting Taubes's position. Nonsense."

Actually what he accused you of is quoting out-of-context, even going so far as to cut off the first part of the sentence you quoted. It is undeniable that you did just this. Whether it was intentional or not, who knows. But don't deny it as if we can't read -- you lose credibility, and look silly.

Unknown said...

Perhaps it's important to note that the "Oxidation Rate" figure (22.4b in the referenced "Quantitative Physiology of Human Starvation: Adaptations of Energy Expenditure, Macronutrient Metabolism and Body Composition") is described as:

"... *simulated* rates of carbohydrate, fat, and protein oxidation in the Benedict experiment..."

So apparently not actual data from Hall's.

The Benedict experiment is, I gather, one done in 1915 on a single (Maltese) gentleman.

Unrelated, but it's disturbing that the indicators are changing drastically from day -1 to day 0. I didn't find an explanation of this in the referenced text on cursory examination.

Zoli said...

Just to clarify, are you using the word, 'falsify,' to mean 'refute'?

Unknown said...

Why are you saying that the tissue is still getting used to fat as the new fuel but then saying that the body has no issue burning fat? The fact that the tissue is fatigued is telling me it is having an issue. Plus the body is a lot less efficient with the fat and ketones witch takes time to improve witch is what I think dr Stephen and Jeff Voleck meant when they said it takes time to adapt plus the body builds more mitochondria. And till then the body will produce an overflow of ketones (with is probably caused by grater rate of released fatty acids and they are secreted in the urine (thanks why someone in the adaptation mode will show a lot more acetoacetate in the urine). So I don't think the body is releasing less fat bec it's not adapted. It's releasing more fat(and producing ketones) but secreting more in the urine bec the body isn't adapted enough yet. Also about what you were saying that the body should need less time to become fat adapted when eating more carbs then in the fasting cases isn't necessarily true bec the body (and brain) can still get lots of its energy from glucose so the body isn't releasing as much fat or producing as much ketones and therefore the body won't need to becom as ketone adapted BECAUSE it's not using as much fat so the adaptation would happen even slower. (I understand your point was that your body in this case has less adaptation to do bec it would need less fat but that is also a reason for it to not need to become as fat adapted ether)
And to regards to Gary taubes he was talking about the a-z trial where they did limit the total carbs and therefor it can be the reason they had a successful weight loss and that they can't say it was bec the calorie restriction.

Stephan Guyenet said...

Hi David,

Either you're unfamiliar with Gary Taubes's writing, or you're trolling me. First of all, those quotes I took verbatim from Kevin Hall's paper, without adding or subtracting anything. Second, on many occasions and in both of his books, he has argued that carbohydrate intake trumps calories. The "context" that was removed from the quotes I used in no way changes the meaning of the text, nor does it misrepresent his position. Leaving out text is acceptable as long as it doesn't change meaning-- which this didn't. Would you prefer I quote the whole chapter in my blog post so I don't leave out any "context"?

I am always dumbfounded by the charitable revisionism people are willing to extend to diet gurus when their hypotheses fail. It's not just that the goalpost gets moved, they act as if the goalpost has always been in its new location. In this case, Taubes has made the argument that carbohydrate, not calories, determines body fatness in both his books, and in his BMJ essay. He shows utter disdain for the prevailing (supported) hypothesis, which is that carbohydrate is fattening by virtue of its calorie content. He states his hypothesis as something to be tested, but then proceeds to make diet recommendations to the public as if the hypothesis were already supported.

Unknown said...

it doesn't mater what Gary said- according to the insulin hypothesis that causing too much insulin secretion is what causes fat storage and limits the use of fat then it will also limit the use of dietary fat and therefore when you limit your dietary fat on a high carb diet with was causing it to get stored now isn't there and you will store less fat witch will cause weight loss

Unknown said...

Thanks for your analysis.

Doesn't this study also disprove (or strongly suggest against) the conventional wisdom that 'a calorie is a calorie is a calorie'?

We had two isocaloric diets with different macros produce statistically significant differences in weight and fat loss.

Under the 'all calories are equal' paradigm, a 30% decrease in calories should yield an equal amount of fat/weight loss no matter how it is achieved.


Subject#1 said...

To sum up:

-Low fat, high carb can work to lose fat if one can maintain a caloric deficit
-low carb, high fat can work to lose fat if one can maintain a caloric deficit

In practise, for me, it's very difficult to control my appetite when I'm on high carb low fat unless I'm eating lots of potatoes and apples.

Perhaps if potatoes and apples are my only carb sources I could revisit high carb while maintaining a calorie deficit and see how I fair on these highly satiating two foods.

Gearóid said...

Stephan,I've read every page of GCBC and my take on what Gary says is not how you take it. It's that carbs are fattening because of hyperinsulinism. If they don't cause it, they're not.I myself think that sugar in excess is the big Cahoona

Stephan Guyenet said...

Hi Christian,

The "calorie is a calorie" thing is more of a rule of thumb than a strict law. At least that's how I view it. There is pretty good evidence that it holds approximately true over most of the macronutrient spectrum in humans. It MAY break down at the extremes, and I think that's part of what we're seeing here (and it is what Hall's model predicts). We may see something similar with the results of the 8-week NuSi metabolic ward study that is currently being conducted on a super-low-carbohydrate diet. Hall's model predicts that at extremely low levels of carb intake, there will be a modest fat loss advantage vs. the typical diet, independent of calorie intake. If this is the case, my prediction is that it will be trumpeted as a huge victory by the low-carb crowd, most of which will ignore the fact that the same thing seems to happen at the other end of the macronutrient spectrum, and therefore cannot be attributed to carbohydrate or insulin per se.

Also, keep in mind that the biggest explanation for the difference in fat loss in Hall's study was glycogen depletion in the LC group, an effect that would have become negligible over longer periods of time.

There is tons of animal work demonstrating that a calorie isn't always a calorie. Rodents will become fat on a refined high-fat diet, even if you restrict their calorie intake to the same as rodents eating regular unrefined chow. I'm pretty convinced that as science advances, the "calorie is a calorie" heuristic will turn out to be approximately true over most of the macronutrient spectrum in humans, but we'll find exceptions.

Hi Subject,

What you're saying is consistent with the evidence.

Peter said...

I've been reading this blog for years, as many of the ideas have been helpful in trying to figure out how to eat. All the trashing of Gary Taubes annoys me, though, because I feel so grateful to him. Not for his speculations about how our bodies work, but for his observation that when relatively healthy cultures start importing our food, principally sugar, flour, and vegetable oil, their rates of obesity and diabetes seem to skyrocket. I feel like the dietary changes I made after I read that have served me well, so I feel defensive when you run him down, not just in this post but in others.

Christopher Kelly said...

Hi Stephan, thought you might like to see my latest podcast.

Why report averages? It's not that much data.

To quote Dr. Feinman from the transcript:

Well, it's worth my attention because I know Kevin Hall and he's a pretty smart guy and this is a pretty distressing paper. One of the reasons is that all of the data is reported as group averages. And the problem is that nobody loses an average amount of weight. The assumption of group averages is that there's a normal distribution or at least essentially limited distribution and the underlying idea behind that is that people are roughly the same and that minor variations account for the spread of the data. Those aren't good assumptions.

But they're okay if the data do not have big variation. Now, that's not true in this study.

Dennis said...


Thanks for your balanced review.

There is a topic which seems not to have been covered though. The determination of the net amount of burnt macro nutrients is based on the measurement of the amount of oxygen used, CO2 produced and nitrogen excreted. These values are input to some formulas in which several constants are used. The values if those constants is based on a number of assumptions. The results are influenced by de novo lipogenesis, but also by the amount of ketone bodies that are being burnt (see for example Fryan 1983, referenced by Hall et al). The latter do seem to have been accounted for in this study.

What I am wondering is whether or not these constants are actually valid for these two very different nutritional patterns. For instance one might expect that the RC diet may indeed result in ketone bodies being formed, potentially requiring different constants to yield the correct results. If the calculations really do require different constants, the results may be drastically different.

Anonymous said...

Excellent insight in your above comment, Stephan. Very nice.

It is only the ignorant Internet diet guru charlatans who abuse the conservation of energy principle and do not understand it who claim that any metabolic advantage contradicts "the first law of thermodynamics." It does no such thing. I am not saying a metabolic advantage does or dos not exist ( I do not know) , but the conservation of energy principle is not at all "contradicted" by its possible discovery etc., nor is it a limiting factor in any way.

There are no "laws" in science, and the "calorie is a calorie" dictate does not even have the same level of support as the actual conservation principles. It is just "unscientific dopey fitness industry speak" often repeated. How our bodies handle this energy is a whole other story.

Even the conservation of energy "law" is a theory. Much remains to be learned about matter and energy. very well supported in the classical world, but energy is ONLY conserved in a time translational invariant system. We tell school kids lies a bit. I suppose it is too complex for them.

As Dr. George Bray himself personally told me , the conservation of energy principle is merely a "state equation", nothing more. He admitted it cannot and does not explain obesity. It is only "relevant" , not sufficient at all. It applies no more to muscular Belgian cattle than it does to overly fat humans, and it "explains" neither scenario at all. And it says absolutely nothing about where the energy came from, how it is stored, how it is used, partitioned, handled, wasted, or absorbed. Dr. Bray totally agreed with me. I bet that would chap the butts of the gurus. LOL !

The "a calorie is a calorie" in the human body rests on a DEEP misunderstanding of what the conservation of energy principle actually means. It does NOT "have to be this way" as fitness gurus claim. Not to any physicists, that is for sure... The conservation of energy would be totally "content" if we ate a large chocolate bar and crapped it out whole again. (Not that the universe operates by OUR models and mandates etc.) It does what it wants. The onus is on us to see how we were wrong.

How the body handles, uses, absorbs, wastes or partitions this energy is definitely not addressed in any way, shape or form by the conservation of energy principle. Those questions are biochemical/genetic/physiological in nature, NOT physics matters. There is lots of money to be made marketing physics to the gullible public who are science illiterate and being hoodwinked by marketing.

And as you noted in the Chris Kresser podcast of May 2010, energy partitioning is a CENTRAL factor in obesity and too much adipose tissue matter and rarely discussed. That was an excellent point. Dr. Garret Fitzgerald stressed to me that energy can be led to many different pathways in the body.

Quantum Chromodynamics explains what is responsible for the mass of our human bodies- at the most fundamental level. It is because of the confined kinetic energy of quarks and gluons. When we gain a kilogram, we are gaining a kilogram of energy. Thanks Albert Einstein. A hot cup of tea has a slightly greater mass than a cold cup of tea.

Take care and I hope you had a great summer, :))

Unknown said...

I'm really bored to know that people is still talking about the disappointing debate carbs vs fat vs protein...hey folks, the greatest thing we learned observing traditional cultures that it's about real food. Food is much more than carbs and fat...we learned that our body can adapt to different macronutrients ratio from the high carb kitava to the high fat inuit, going through hadza who are able to eat more than 4kkalories pet day and stay skinny...fat is much more about the chronic inflammation triggered by grains and other non species appropriate food that set off leptin and insuline resistance.
Really, are human beings so stupid to refuse the evidence for the sake of "science"?

Stephan Guyenet said...

Hi Peter,

Sorry, I don't really want this to be a personal conflict with Taubes, it's just that he has articulated a particularly popular and rigid version of the carb-insulin hypothesis, so the conversation in this area often focuses on his ideas. Add on top of that the fact that Taubes has taken an extremely adversarial/insulting stance toward researchers who disagree with him, including me. I try to keep it civil in my posts, but when I have to defend the most basic facts against some of his acolytes in the comments section, it can sometimes go too far. I acknowledge that.

Hi Christopher,

No doubt it would have been useful to include individual data. Hall has indicated that he will be publishing other papers on this study, so perhaps we'll get to see some of it. But you could make the same critique of 90 percent of other human RCTs, so I think Feinman is taking a legitimate point and applying it a bit too disproportionately to this particular study. To call the paper "distressing" on that basis seems a little much to me.

Hi Dennis,

I have a general understanding of respirometry but I don't know all the details of Hall's methods. I do know they accounted for DNL. Your point about the validity of respirometry for people on different diets is a valid one that has been debated in the field for some time. I don't know what the final verdict is, or if there is one. These are questions that Hall would be able to answer better than me.

Hi Raz,

Thanks. I view energy balance as a link in a longer causal chain. It's a very informative link but it certainly isn't the ultimate explanation for obesity.

Unknown said...

If you have time can you reply to my post

Ashley said...

Let's take a moment to thank Stephan for bothering to (1) write this piece, and (2) to continue to reply to all of these comments. Let's practice some gratitude, and get rid of this snarkiness I'm seeing in some of these posts. We don't need that here. Let's be good consumers of science and thoughtful conversationalists about the science.

Aegirsson said...


I read your post. There is more to it as the body regulates carbs and proteins quite tightly, as opposed to fats. You can stuff yourself with proteins and carbs but it would be hard to gain weight. You would eventually but it is much harder than when you stuff yourself with fats (more so when these fats come with carbs and proteins). Protein regulation is due to the nitrogen balance. Carbs are used by the brain, glycogen store replenishing, and some increased body heat. You;d have to eat an excessive amount of carbs for days and days without basically moving your body before it would trigger an excessive weight gain. DNL is basically insignificant in normal people, and the little created fat is mostly oxidized right then as far as I could read.When it comes to fats, we just suck at regulating excess in terms of increased oxidation and body heat. It is usually stored and burned only when needed (fasting period, inter-prandial phase, etc).

Unknown said...

Where did you get that from?

Unknown said...

Have we seen over feeding studies of carbs in insulin resistant people? The DNL studies I saw were in healthy people - and not for a long period of time.

I'd suspect at some point the body would get wise to what's going on and up regulate carbs -> fat no matter how costly that process is.

Anecdotally, all one has to do is hit up youtube and look at the 30BAD crowd and see how many people abandon the diet because of weight gain.

Aegirsson said...


Nitrogen balance: see e.g. http://www.fastbleep.com/biology-notes/40/116/753

Carb metabolism / DNL : see e.g. http://www.researchgate.net/publication/12932996_De_novo_lipogenesis_in_humans_Metabolic_and_regulatory_aspects

After reading that, being afraid of eating carbs will hopefully seem nonsensical (at least if you are not diabetic).

Mikko Järvinen said...

If there is no adaptation period to using fat as a fuel - why do the ketone measurements with a blood ketone meters or breath ketone meter like Ketonix start producing ketosis level measurements only after several days, even weeks on a nutritional ketosis diet?

Not disputing what you are saying, just would like to understand what is going on then in the body - is ketosis irrelevant, is the body utilizing fat as a fuel equally well in ketosis or out of ketosis? If that is the case, then of course it seems that it is equally easy to lose weight with low fat high carb diet.

That then leaves only the issue of who is able to maintain high carb eating without overeating - very few, it seems when you look around.

Alex said...

Isn't it possible (perhaps even likely given their obesity) that the low fat group was still restricting carbohydrates when compared to their previous diets? I don't think that this is necessarily the proof against Taubes' statement that you propose it to be.

Stephan Guyenet said...

Hi Mikko,

I think full adaptation can take longer, but there are multiple facets to adaptation. The point I made in this article is that total fat oxidation plateaus rapidly, which is what is relevant to fat mass. But that doesn't mean the body has fully adapted to all metabolic aspects of greater fat oxidation. I'm not knowledgeable in this area but I know there is research suggesting that it takes longer than a few days for the body to maximize physical performance after switching to a higher-fat diet. That could be relevant in other contexts like athletic performance, but fat oxidation is what is relevant to fat mass.

Hi Alex,

The run-in period of the study design makes that an unlikely possibility.

Gearóid said...

What is MUCH more relevant to fat mass is hunger or lack of it. Hunger is the elephant in the room. An artificial experiment in which input is controlled by a metabolic ward will not translate into success in the real world if the subject is constantly hungry. Neither, will it succeed if the patient isn't hungry but is not willing to stick to the diet which controls his hunger. In the latter circumstance, one of the big reasons for slipping is cheating. This is analogous to the smoker deciding to have the occasional cigarette.
The low carb diet is much more successful is the subjects get proper psychological support and have it explained to them ab initio, rather than be assigned to limb A or B of a trial by people who are actually trying to show that it won't work longterm.
It's working fine in me. 40lb down and at target weight for 15m. I don't cheat.

Mikko Järvinen said...

Thanks Stephan!

So - fat oxidation on a low carb is slightly lower or about the same as with low fat?

What would you speculate, if they did this experiment with 8% fat and 8% carb version, allowing for an adaptation period of say 2 weeks - would the results change significantly?

I would think that the low carb group here spent less energy just because before the adaptation, you at last feel like you are starving. In the low fat group there was possibly less of a feeling of starvation.

Stephan Guyenet said...

Hi Gearoid,

I agree that hunger is important. However, one of the major points I've tried to make in my writing and speaking is that food intake is the result of many interacting motivations. Hunger is an important motivation, but I'm not convinced it's the primary driver of food intake in the modern world. Most people eat/drink out of habit, because they seek the reward value, because they're self-medicating stress or depression, because it's easy, because they're hungry, and for many other reasons.

One clear example of this is alcohol. One beer contains ~150 calories; one drink of wine or spirits contain ~90, and people rarely drink alcohol because they're hungry. Even moderate drinkers can easily ingest hundreds of excess calories per day purely because of the reward value of alcohol. Most other foods are eaten due to some combination of reward, hunger, habit, and ease of access.

Hi Mikko,

Fat oxidation rate is higher on low-carb than high-carb, given equal calories. However, what is often overlooked is that fat intake is higher on low-carb. What determines changes in body fat mass is fat balance, in other words, fat intake minus fat oxidation. The low-carb group had a less negative fat balance than the low-fat group.

karl said...

My personal experience is that eating carbs induces excess hunger - and that hunger can result in eating not only carbs, but excess fats and protein - not at the time the carbs were consumed, but at the next meal. I would think that you would 'get' that food choice can influence appetite.

Normal people maintain their weight without thinking much about it to within a few percent. The calories-in;calories out are automatically matched by high precision by 'appetite'.

What is not spoken of is the effects on serotonin by carbohydrates - research on this goes back many decades - (see Wurtman ) and led to the development of SSRIs.

Eating carbs increases serotonin - done on a long term basis, the post-synaptic receptors down regulate and you have a rebound effect - an addiction. To re-regulate these receptors has been reported to take 6-weeks when discontinuing SSRIs and one would have to guess a similar time frame from eating carbs - short term studies don't allow time for re-regulation.

I would also note that there are studies that show that the people consuming fructose containing sugars prior to going out to eat ordered and ate more food. (triglycerides blocking leptin?).

In the end, this study does not clarify why we have a pandemic of obesity or what to do about it. Clearly on one level we know it is from eating to much - the why is appetite.

I would also have to say that the editing of Taubs words did change the meaning from a speculation to a claim. (The missing ellipsis (... punctuation) merits a correction by the authors of the paper and puts their objectivity in doubt). While Taubs is a proponent of low-carb diets he has been very clear that the high quality research to understand the question of metabolic differences has yet to be done.

Taubs has also said on many occasions that while the calories-in : calories-out ratio matters - but that the supply side is effected by appetite - and when people are told to simply eat less they fail - almost all the time. Appetite really really does matter. Eating addictive 'comfort foods' (carb heavy food) apparently increases appetite.

We desperately need studies with synthetic diets that test the effects of PUFAs on appetite. (Synthetic diets during the Apollo program taught us that fructose spiked triglycerides (but no one paid attention).) Today we are eating 5x the PUFAs than in 1960's and these are not 'human foods' and have not been studied in careful long term studies on their effect on appetite. (We do know that they have effects at the mitochondrial level that could theoretically change appetite ).

Unknown said...

We discussed this in our lab meeting this week, and here is a couple of points that have been brought up above but we thought were relevant.

* Participants were >30 mg/kg2 BMI, but nondiabetic and had reasonable HOMA-IR scores. This was part of the exclusion criteria, along with lack of medications and being weight stable. The participants might be more reasonably thought of as "healthy obese" than a more typical obese participants.
* For their primary outcome there was no difference between the diets on fat mass or percent fat mass. Fasting that causes loss of both fat and lean mass is not preferable. Based on the data in Table 3, reduced carbs actually increased fat percentage, but the difference between the diets was not significant (p=0.24). They revisit this point looking at the rates of fat loss in Figure 3D/E which is where they see the significant effect, and where again the effect is really only seen looking at total fat loss, not total percent fat decrease. This implies a fairly dramatic (but unreported) decrease in lean body mass. This is consistent with most calorie restricted diets, and the attendant decreases in muscle mass and energy expenditure make it easy to bounce back to elevated body weights.
* Reduced carb reduced energy expenditure even more than reduced fat. This is at odds with the weight loss, then stabilization paradigm in Ebbeling et al. JAMA, 2012 who showed that in their paradigm low carb diets reduced EE less than reduced fat. Ebbeling et al. used a very different paradigm (weight loss, followed by stabilization, followed by diet changes) but its odd that the effects were going in a different direction. Again in Hall, this effect didnt quite reach significance (p=0.099) but its key that it was in the opposite direction. This suggests that this is more of an acute effect of diet, and may not be representative of chronic metabolic adaptation to weight loss (important for their projections in Figure 3.
* There were some really dramatic sex differences presented in the supplement. I don't have access to the supplement from home, but we identified quite a few really dramatic differences in things EE and hormone levels (Ghrelin) between males and females. In a lot of cases, it was probably confusing to combine the genders of the participants (they control for Gender, but probably not the interaction between Gender and Diet). Ill post what we thought of these when I get back to the lab.
* As an aside why are these longs lists of p-values not adjusted for the tons of multiple comparisons? If that was done, many of the marginally significant effects would not be present.

Gearóid said...

I agree with everything Karl said. Full stop!

Unknown said...

Our journal club summary (talking about the gender differences) is here: http://bridgeslab.uthsc.edu/papers/commentary/12

Anonymous said...

This just shows that fat loss on a high carb diet necessitates low fat and calorie restriction. Which is relatively well known.