Sometimes, a meta-analysis (quantitative study of studies) is just what the doctor ordered to inoculate us against the shortcomings of our own cognition. When a topic has been studied extensively and it has produced many studies of varying quality, this lends itself to incorrect conclusions because we can find studies to support almost any belief. This is problematic because we naturally tend to gravitate toward evidence that reinforces pre-existing beliefs, and away from evidence that challenges beliefs. Called confirmation bias, this phenomenon afflicts all of us and has to be actively managed if we want to arrive at reliable knowledge. Systematically examining a body of evidence and integrating it mathematically is a useful tool for combating this bias. A new meta-analysis examines the effect of carbohydrate vs. fat calories on energy expenditure and body fatness, giving us the most objective view of this question to date.
The study
Recently, Drs. Kevin Hall and Juen Guo published what I believe is the first meta-analysis of controlled feeding studies that compared diets of equal calorie content but differing in carbohydrate and fat content (1). They only considered studies in which all food was provided by researchers and protein intake was held constant between diets. Their outcomes were energy expenditure and body fatness.
The results
They identified 28 studies that met their criteria for energy expenditure. Combining the data of these 28 studies, they found that calorie-matched diets predominating in fat vs. carbohydrate have almost identical effects, but higher-carbohydrate diets do lead to a slightly higher energy expenditure. This difference was statistically significant but of little medical or practical relevance, since it only amounted to 26 Calories per day. This slightly higher energy expenditure is consistent with the fact that the metabolism of carbohydrate is slightly less efficient than the metabolism of fat, meaning that a bit more energy is wasted*.
Examining the data, the paper’s result is not hard to believe because only 8 of the 28 studies reported that lower-carbohydrate diets led to a higher energy expenditure than higher-carbohydrate diets, and among those 8, the results were only statistically significant in four. In contrast, 20 studies reported higher energy expenditure with higher-carbohydrate diets, and that was statistically significant in 14. One can choose individual studies that support either belief, but the overall evidence suggests that the relative carbohydrate and fat content of the diet has little impact on energy expenditure.
Onward to body fatness. Hall and Guo identified 20 controlled feeding studies that reported changes in body fatness on equal-calorie diets differing in fat and carbohydrate content. Echoing the energy expenditure finding, they found that diets predominating in carbohydrate or fat have similar effects on body fatness. Yet higher-carbohydrate diets do lead to a slightly greater loss of body fat per calorie, amounting to a 16 gram per day difference. This is actually a larger difference than one would predict from the difference in energy expenditure, which would only be 2.8 g/day.
Discussion
Given this new meta-analysis, I think it’s now fairly safe to say that in a general sense, equal calories from fat and carbohydrate have similar effects on energy expenditure and body fatness, with a possible small “metabolic advantage” for higher-carbohydrate diets. This doesn’t imply very much about the real-world effectiveness of low-fat and low-carbohydrate diets because it doesn’t factor in free-living calorie intake, but it is relevant to certain popular theories about how those diets work. The upshot is that you shouldn’t expect altering the carbohydrate-to-fat ratio of your diet to work magic on your metabolic rate, but rather you should choose a diet that controls your calorie intake effectively and sustainably.
It is no longer tenable to suggest that carbohydrate per se reduces energy expenditure and causes the accumulation of body fat independent of calorie intake. This idea was never very well rooted in evidence, but now that we have a meta-analysis it is clear that it resulted from confirmation bias.
This meta-analysis leaves many questions unanswered. Is the same effect observed during weight gain, weight maintenance, weight loss, and weight loss maintenance, or could the ratio of carbohydrate to fat matter in these scenarios**? Is the effect observed equally at moderate macronutrient ratios and at the extremes? We have individual studies that address these questions, but no meta-analysis yet.
Hall and Guo’s paper is not just a meta-analysis. It’s a review of obesity energetics, body weight regulation, and their relationship to diet composition. It is hands-down the single most informative paper I’ve encountered for explaining the relationship of energy intake and expenditure to body fatness, and the big-picture view of how the body’s energy control systems work. It comes from a research group that is at the leading edge of these questions. It took me years of study to formulate the high-level perspective that you can now get from an hour of reading (1). It also dovetails nicely with chapters 1, 6, and 7 of The Hungry Brain.
* I’m not certain that this explains the higher energy expenditure; I’m just suggesting that it’s an obvious possibility.
** A study by Dr. David Ludwig’s group suggests the possibility that the outcome could be different in the context of weight loss maintenance (2). I doubt there are enough studies on this to support a proper meta-analysis.
What is WMD? I’m assuming it’s not weapons of mass destruction. Sometimes I think I spend more time looking up acronyms than reading the meat of articles.
Hi Gretchen,
That stands for “weighted mean difference”.
I remember once reading that high carbohydrate diets increase the conversion of T4 to T3. Do you think that might be involved in the increased energy expenditure observed?
A major problem with this meta-analysis as it now stands in its e-pub ahead of print status, is that there are no details on how the investigators located the included studies. For example, no search strategy is given. There is no information on which databases were searched, date limits, language limits, etc. This is a huge problem with systematic reviews and meta-analysis in general. If the methodology used to find the relevant literature is questionable then any claimed results must also be viewed with a degree of skepticism simply because of the huge bias introduced by inadequate searching.
Maybe Hall and Guo followed the National Academies’ or Cochrane standards for conducting a systematic review and full details will be included in the final manuscript. At this point we simply don’t know. I made essentially the same comment in PubMed Commons at the citation (PMID: 28193517).
Hi Ben,
The paper has a supplemental methods section in which the meta-analysis methods are described. Have you had a chance to read that? It answers at least some of your questions.
Hi Stephan,
I do see that they mention a methods section in the PDF, but I can’t find it in the PDF itself (other than the figures, legends and forest plots). I am also not seeing a supplementary materials link at the article on the journal webpage (http://www.gastrojournal.org/article/S0016-5085(17)30152-X/fulltext). Could you please email me a screenshot or the file itself or indicate where exactly you’re finding it. I definitely want to remove my comments if I’m missing this.
Thanks,
Ben
The full paper
http://sci-hub.bz/10.1053/j.gastro.2017.01.052
Obesity Energetics: Body Weight Regulation and the Effects of Diet Composition
Kevin D. Hall, Ph.D, Juen Guo, Ph.D.
Charles, you’re a saviour.
As a LCHF fan I know many people will scoff at it, but it is objective data. Being a long time personal trainer I’ve seen ppl get results with both approaches, so I don’t doubt it. I currently enjoy fasting and maybe a low to moderate carb intake. I know this isn’t included in the study, but the difference for me is more than just bodyweight. I feel much better eating less both physically and mentally. I imagine that will be difficult to objectively show, but I think that would be a major factor in picking each strategy. At the end of the day, being in control will get results with either one!
Thanks for the detail for Stephan
Hi Kyle,
I continue to believe that low-carbohydrate diets can be a good choice for certain people, and what you said squares with what I’ve heard from many others. The perspective you have is totally rational.
In real life, I think what diet selection boils down to is comfort. What is the diet/lifestyle that will allow a person to control calorie intake and weight most comfortably? Comfort, or more specifically minimized tension between conscious and nonconscious brain circuits, is the primary thing that determines long-term success. In other words, having appetite and cravings be aligned as well as possible with the conscious, rational goals of eating the right amount of healthy food and being lean and healthy. As opposed to fighting your impulses on a daily basis.
When I lived in Bordeaux I ate highly confected foods. Rewarding food on a world class level, every mouthful, every meal, every day, often in restaurants. I ate well and slept well and I lost about 10 pounds in the 15 months I was there. I walked all the time for transportation and entertainment. And I remained a tourist, always interested and alert to what I experienced daily. Things didn’t work out well when I got back to the US, and I became obese and diabetic over the following 5 years. I’ve tried to go back to those French habits as much as I can, because they were comfortable and maintained lower weight. For me, weight control has been the result of multiple tradeoffs and not just a product of hormonal signaling or macronutrient optimization.
I disagree with your statement, “In real life, I think what diet selection boils down to is comfort. What is the diet/lifestyle that will allow a person to control calorie intake and weight most comfortably”.
How does that answer the millions of skinny-fat people living on a carb addicted diet? Very unhealthy people with nice slender bodies that have heart attacks, strokes and look good naked?
As Denise Minger pointed out. A LCHF OR a HC very LF diet can both work to becoming healthy. But who the heck would choose a high carb diet without almost any fat? I’ll take the coconut, avocado, olives, grass fed and finished butter, cream, meat and cheese any day of the week.
Hi Jim,
When a person is “skinny-fat”, there are two things happening: insufficient muscle mass and excess fat mass. The amount of carbohydrate and fat in the diet has little to do with muscle mass; that is determined primarily by physical activity and protein intake (and genetics of course). And the elevated fat mass is due to excess calorie intake, even in a skinny-fat person. A skinny-fat person who loses fat is just a skinny person.
I don’t think it’s accurate to say that there are “millions of skinny-fat people living on a carb addicted diet”. You are implying that the “skinny-fat” state is caused by a “carb addicted diet”. Yet people don’t become addicted to carbohydrate per se. People don’t eat plain white sugar by the spoonful; they eat foods that are mixtures sugar, starch, fats, salt, and other flavors, and all of those things together are the problem. Yes, fats are part of the problem.
Most people who “look good naked” (low fat mass and adequate muscle mass) are not about to have a heart attack, a stroke, or develop diabetes. Sure, there are always exceptions that can be cited because it’s all a matter of probability, but on average they will be at a much lower risk than the general population.
Stephan,
Thanks for the reply. Fair enough, especially your first and third paragraphs.
Still not sure about your second paragraph completely. I’m using myself and many, many jock friends as an example and guys like Sisson, basketball players and hundreds of others like us that were athletic, ate 4-500 grams of carbs a day or more, looked pretty good, and were sick, tired and beat up a lot. Of course it’s not all the carbs, but a combination of all the other stuff added too! However,is 4-500 grams of carbs a day not an addiction to carbs though? Is it an addiction to the stuff we put on the carbs (salt,fat,etc.)? I think there are millions of skinny-fat people that are addicted to something, and I think it’s the carbs. It’s very difficult to eat way too much protein or fat. I sincerely think that they are much more self-regulating than carbs. Carbs just make you hungry every two hours.
Thanks for your time.
-Jim
Stephen,
What about the idea that eating too much protein and too few carbs leads to too much adrenaline and cortisol in the body? We hear about too much insulin coming from too many carbs but I never hear about the reverse. Why is that?
My understanding is that this is the reason that many people have trouble with low carb diets in the long term. At first they feel great, but over time there are hormonal consequences. The adrenal glands get taxed and lots of health problems can follow. I’d love to hear your thoughts on this. Thanks!
I’m a little confused, were the diets either low fat/high carb or high fat/low carb, or were some moderate fat/moderate carb?
Very interesting, Stephan! Thanks for sharing! 😉
Dr Guyenet
Thank our for sharing this knowledge with us
This is a little off the subject but I really enjoyed your talk on leptin resistance, by far one of the most helpful lectures I have heard in a long time. I appreciate our open minded approach to presenting your data and views, very professional
Continued success to you
Larry
Sorry for the typo, should say: your open minded, not our
L
Do we happen to know the net carb intake/fibre intake of the diets in this grouping? I assume there would be a significant difference between a high carb, high fibre diet and a high carb, high sugar/starch diet
Pretty insignificant study as it does not take into account what sort of carbs or fats are eaten. Sugar is a carb, and so is a cucumber. (mostly)
Or healthy avocado oil vs refined soybean oil.
So it is such a general study, im not even sure it was worth reading.
I am not surprised by the finding that equal calorie intake from different sources results in equal weight gain. In fact I am pleased because it has become ingrained that eating fat makes you fat, when it is really total caloric intake. However my understanding from proponents of a high fat/low carb diet(which includes me) is that you end up eating less total calories then you would with a high carb/low fat diet because you are more satieted with fat. Being on a high carb diet leads to overeating. Eating less calories for both is fine, but one requires less willpower to stop from overeating.
Would you know of studies where caloric intakes are compared with respect to fat/protein/carb consumption?
Gael,
You might look into the potato hack. People eat only potatoes, as much as they want, and lose crazy amounts of weight. They naturally eat fewer calories, because boiled potatoes is the most satiating food ever tested. But potatoes are of course all carbs. Yet, incredibly satiating.
Deep fry them in lard, and we have a different story. But is the problem with french fries the carbs or the fat? Without the fat it was satiating, with the fat not. So is the carbs or the fat the problem?
Hi Dr. Guyenet,
I wonder if age and therefore underlying metabolic health might interact with diet in unique ways in older free living people relative to younger healthier folk? I know this meta-analysis confirms macronutrients don’t matter when calories are held constant but an elderly family member who lost tremendous weight on low carb sent me, a sceptical med student, this egg industry funded RCT link comparing low carb vs low fat w equal protein and the DEXA results are surprising. httpm.fasebj.org/content/31/1_Supplement/lb320.short
Even if it’s just calorie difference that matters why, with equal amounts of satiating protein between groups, are the low carb folks eating so much less? This would have real world implications for the 70 million older adults with prediabetes…thanks for your input, Dr. Guyenet. -Robert