The carbohydrate–insulin model of obesity posits that high-carbohydrate diets lead to excess insulin secretion, thereby promoting fat accumulation and increasing energy intake. Thus, low-carbohydrate diets are predicted to reduce ad libitum energy intake as compared to low-fat, high-carbohydrate diets. To test this hypothesis, 20 adults aged 29.9 ± 1.4 (mean ± s.e.m.) years with body mass index of 27.8 ± 1.3 kg m−2 were admitted as inpatients to the National Institutes of Health Clinical Center and randomized to consume ad libitum either a minimally processed, plant-based, low-fat diet (10.3% fat, 75.2% carbohydrate) with high glycemic load (85 g 1,000 kcal−1) or a minimally processed, animal-based, ketogenic, low-carbohydrate diet (75.8% fat, 10.0% carbohydrate) with low glycemic load (6 g 1,000 kcal−1) for 2 weeks followed immediately by the alternate diet for 2 weeks. One participant withdrew due to hypoglycemia during the low-carbohydrate diet. The primary outcomes compared mean daily ad libitum energy intake between each 2-week diet period as well as between the final week of each diet. We found that the low-fat diet led to 689 ± 73 kcal d−1 less energy intake than the low-carbohydrate diet over 2 weeks (P < 0.0001) and 544 ± 68 kcal d−1 less over the final week (P < 0.0001). Therefore, the predictions of the carbohydrate–insulin model were inconsistent with our observations. This study was registered on ClinicalTrials.gov as NCT03878108.
Paywall - https://doi.org/10.1038/s41591-020-01209-1
Full paper is on the pirate academic sites.
This paper is a EXCELLENT demonstration of why you can't trust abstracts or opinions and you have to look at the data directly.
First kudos to the paper for actually using a ketogenic diet, monitoring ketones! That is rare. Now, on to the bullshit -
From the supplementary material the animal menu was
so this is closer to omnivore keto, with at least 75% plant based energy, AND LOTS OF SEED OILS (fried chicken for goodness sake). Calling this animal based is dishonest. This is plant based with some animal protein thrown in.
Yet the data shows weight going down on low carb, even while energy intake increases.
No run-in period to the diets (no standard baseline).
Diets switched at 2 weeks, no wash out period.... You can't see it in this paper directly, but that means many of the low carb benefits that take time to manifest get demonstrated in the high carb diet due to this rapid switching. This is not a clean demarcation.
Responses to this paper, including data reanalysis
This paper uses the first's paper data and separates out the cross over groups by diet order. When that is done the first paper's conclusion of CIM energy difference not being demonstrated disappear... in a 2 week : 2 week back to back cross over - ORDER MATTERS. LCD reduced energy intake in ad libitum eating, that PERSISTED into high carb diet, and the reverse as well. High carb diet increased energy intake, crossing over into the low carb diet. IT TAKES TIME TO ADAPT TO A NEW METABOLISM.
What I think is demonstrated here is that CICO has not been proven by Hall, and the Carbohydrate Insulin Model still stands viable. Clean studies will be most welcome here.
This was a VERY expensive study. Why Hall, a well funded plant based researcher, is wasting his time on a tiny fringe theory like the carbohydrate insulin model? What made him spend the time and money making these very poorly constructed, yet very expensive, studies? and why compare against plant based and not SAD like he does every other time?
Hall clearly has a bias, and I suspect he engineered his study in such a way he could have the conclusion he wanted. The paper title is more important than the data.
WALTER WILLET, the man himself, called this study “worse then useless because its misleading”… If Walter Willet the king of weak epidemiology is against you… man… you got problems. https://youtu.be/tIPX_TnUAWQ
In the willet video that make the cross over energy difference even more clear. After 2 weeks of low carb energy intake decreased on the low carb eating pattern.
Hall even made a new paper - https://doi.org/10.1016/j.ajcnut.2024.08.013 - Diet order significantly affects energy balance for diets varying in macronutrients but not ultraprocessing in crossover studies without a washout period
But, he doesn't fix or retract his initial paper. So he acknowledges the issue, but not his conclusions.
[Squares are low carb first, circles are high carb first] Look at that! a 2,000 calorie change in eating! that is HUGE.
If you eat low carb for 2 weeks you eat 2,000 calories LESS PER DAY then if you eat low fat for two weeks - FROM HALLS OWN DATA AND GRAPHS - This is exactly inline with the CIM.
Bret Schur did a interview with Taubes and Ludwig on this paper, as well - https://www.youtube.com/watch?v=L39nTrxhBeo
And Hall did a interview about this too: https://youtu.be/zOAapJo9cE0
TLDR - This is a junk paper engineered to push a headline. The data doesn't support the abstract or discussion section findings. Most people will never read the paper, they will just hear from influencer, blogger, etc that keto is bad and there is a study that proves it.
CIM was kinda proven in 197x and 1983 with the in vivo tests demonstrating anti-lipolytic effects of insulin, and in 2000 in vitro. The fact somebody is trying to kill CIM now in 2019 is crazy... you can't mobilize fat with elevated insulin... hence Carbohydrate Insulin Model is just a direct consequence of what we know about insulin and fat cells.
What should be clear from this paper, the responses, the response re-analysis is that very involved people well read on the literature disagree and that CIM isn't disproven or dead by any measure.