Traditional diet advice over the last fifty years, still espoused by most health authorities around the world, holds that if you want to lose weight, you need to cut down your fat intake. In the last few decades, a number of alternative diets have sprung up claiming that you should instead be cutting down your carbohydrate intake. These include the LCHF, Atkins, paleo, and more recently the ketogenic diets. But what do the randomized controlled trials say? Should you cut down on fat or carbs if you want to lose weight?
Before we get in to the meat of the article, we need to clarify one thing. There are basically four different dietary sources of energy available to humans. These are carbohydrates, proteins, fats, and alcohol. If you get less of your energy from one source, that necessarily means you will need to get a greater proportion of your energy from one or more of the others. So a low carb diet generally means you will be getting a greater proportion of your calories from fat, and a low fat diet generally means you will be getting a greater proportion of your calories from carbs.
The idea behind the traditional advice to cut down on fat is largely based around the cholesterol hypothesis, which we have recently debunked. Basically, it was thought (and still is, to a large extent) that a high fat diet made you fat and caused heart disease, which was the basis for recommending that people cut down on their fat intake.
The idea behind the low carb diets is largely based around the fact that fats and proteins are absorbed more slowly than carbohydrates, which leads to a longer lasting sense of fullness after a high fat or high protein meal than after a high carb meal. This should in turn leads to a lower overall intake of calories.
In recent years, there have been two decent quality randomized controlled trials comparing low carb vs low fat diets, one that was published in the Annals of Internal Medicine in 2014, and another that was published in JAMA (the Journal of the American Medical Association) in 2018. We will look at each in turn.
The 2014 trial included 148 participants who were randomized to either a low carb diet or a low fat diet. The trial was funded by the National Institutes of Health (NIH) and was carried out in New Orleans. In order to be eligible for the study, participants had to have a BMI of between 30 and 45 (30 is the lower limit for being diagnosed with obesity) and not have any other chronic health conditions. The average age of the participants was 47 years.
Participants were recruited through mailing lists, flyers, and TV-advertisements. This is important, because it means that the participants were motivated to lose weight and to start leading a healthier lifestyle. Something to be conscious of in light of this is that there was no placebo group – the purpose of this study was not to compare a diet versus placebo, but to compare two diets versus each other. This limits the data somewhat. It would have been interesting to have a third group that was getting some form of placebo treatment, in order to be able to compare the diets with doing nothing. Since the people taking part in the study were in general a least a bit motivated to lose weight, it is quite possible that they would end up losing weight over the coming year even without any specific dietary intervention.
Participants in the low carb group were instructed to keep their daily carbohydrate intake at below 40 grams per day. This isn’t just a low carb diet, it is a very low carb diet (also known as a ketogenic diet). In general, people enter a state of ketosis when their carbohydrate intake is below 50 grams per day. (Ketosis is a state in which the liver starts to produce large amounts of ketones to compensate for the lack of carbohydrates. Ketones are an alternative fuel produced by breaking down fats, which can be used by the heart, brain, and muscles when there is a lack of available carbohydrates).
Participants in the low fat group were instructed to get at most 30% of their calories from fats. Neither group was given a specific calorie target. From my perspective, this is a sensible approach, because people generally aren’t going to be willing to keep counting calories for the rest of their lives, so if a new diet is going to be effective over the long term, then it needs to be relatively effortless to maintain once new dietary habits have been formed.
Both groups were given handbooks explaining what to eat in order to meet the targets, and were provided with one meal replacement per day for the duration of the study, which was either of the low fat or low carb variety depending on which group they were in. Apart from that, the participants had follow-up meetings with a dietician weekly during the first four weeks of the study, and thereafter small group counseling sessions every two weeks for the next five months, and finally small group sessions every four weeks for the final six months.
The participants were followed up at three, six, nine, and twelve months and asked to recall everything they had consumed over the preceding 24 hours. They were also weighed, and had blood tests and urine tests drawn at each of these time points.
So, what were the results?
Overall, 80% of participants followed through with the trial. There was no significant difference between the groups in terms of likelihood of sticking with their respective diets. This suggests that both diets were about equally easy to maintain.
At baseline, the average intake of carbohydrates was 242 grams per day in both groups. The low carb group was able to decrease this to 97 grams at the 3 months mark, but was up to 127 grams at the twelve month mark. The low fat group also decreased its intake of carbohydrates, to 193 grams at the three month mark, which had risen marginally to 197 grams at the twelve month mark.
There are two things about these results which are interesting. Firstly, the low carb group was supposed to cut carbs down to 40 grams per day, but never even came close. To me, this shows the difficulty of maintaining a ketogenic diet. Having said that, they were able to halve their carbohydrate intake, which is still a significant change from baseline. Secondly, the low fat group also decreased their carb intake, which I guess makes sense, since the people taking part in the trial had all signed up to it because they wanted to lose weight and were motivated to do so, so even though there was no formal requirement to cut down on calories, it is reasonable to expect that many participants consciously did so anyway. This could obviously make any difference between a low carb and low fat diet seem smaller than it actually is, since both groups cut down on carbs.
When it comes to fats, the average intake at baseline was 34,7% of total calories in the low fat group, and 32,5% in the low carb group. At the one year mark, this had decreased to 29,8% in the low fat group, but increased to 40,7% in the low carb group.
The average weight reduction at one year was 1,8 kg in the low fat group, compared with 5,3 kg in the low carb group. The difference was highly statistically significant. At the same time, the low carb group increased its muscle mass by 1,3% while the low fat group decreased its muscle mass by 0,4% . This difference was also highly statistically significant.
But that’s not all. One common argument against a low carb high fat diet is that it might be more effective at causing weight loss, but that this is outweighed by the fact that it increases the risk of heart disease. Luckily, the researchers gathered lots of data to help us determine if that is actually the case, and based on this they calculated the ten year risk of cardiovascular disease at baseline and at one year, using a scoring system known as the Framingham risk score. In the low fat group, the ten year risk of heart disease increased by 0,4% . In the low carb group, the risk decreased by 1,0% . Again, this difference was highly statistically significant.
What conclusions can we draw from this study? Basically, the low carb group lost almost three times as much weight as the low fat group, while gaining muscle mass (the low fat group lost muscle mass). Additionally, the low carb group decreased its ten year cardiovascular risk, while the low fat group increased its risk, according to the Framingham risk score. These results strongly suggest that a low carb diet is more effective than a low fat diet, both for weight loss and for protecting against cardiovascular disease.
But before we draw our final conclusions, we’re going to look at the second study, published in 2018. This was a randomized controlled trial that involved 609 participants with a BMI between 28 and 40 (As mentioned before, 30 is the cut-off for obesity. 25 is the cut-off for overweight, so these people were overweight to obese). The average age of the participants was 40 years old. Apart from the excess weight, they had to be fundamentally healthy in order to be included. The trial was carried out in San Francisco and was, just like the previous trial, financed by the NIH. As in the previous trial, participants were followed for one year and were randomized to either a low fat or a low carb diet.
Over the course of the year, the participants took part in 22 small group counseling sessions focusing on ways to achieve the respective diet that they had been assigned to (more frequent earlier in the trial, less frequent towards the end). What participants were told to do in this study varied a bit from the previous one. Participants were told to either limit intake of fats or intake of carbs to 20 grams per day for the first eight weeks, depending on whether they were in the low fat or low carb group. After that, they were supposed to slowly add the fats or carbs back by 15 grams per week until they reached the lowest level that they thought they would be able to maintain over the long term. As in the previous trial, no specific calorie restriction was imposed. And as in the previous study, the actual amount of each substance eaten was determined by interviewing the participants at various time points about what they had eaten during the preceding 24 hours.
On to the results.
As in the other trial, around 80% of participants followed through to the end, with no difference between the groups. At baseline, the low fat group was consuming around 242 grams of carbohydrates per day while the low carb group was consuming 247 grams of carbohydrates per day. At the one year mark, the low fat group had decreased its carbohydrate intake slightly, to 213 grams per day, while the low carb group had decreased its carb intake significantly, to 132 grams per day.
Fat intake at baseline was 35% of total calories per day in the low fat group, and 36% per day in the low carb group. At the one year mark, the low fat group had decreased its relative fat intake to 29% of total calories, while the low carb group had increased it to 45% .
The average weight reduction at twelve months was 5,3 kilograms for the low fat diet group and 6,0 kilograms for the low carb group. The difference was not statistically significant.
That’s odd. The first study found a big difference in weight loss between the groups, while the second study didn’t find any meaningful difference. This is in spite of the fact that the studies were actually constructed quite similarly, and that the changes in carbohydrate and fat intake levels were similar. How do we explain that?
I have no idea. If anyone has any suggestions, please post them in the comment section. In general, the data from second study should be more reliable for the simple reason that it had over four times as many participants, which significantly decreases the risk that random chance will cause a result that isn’t real.
In terms of Framingham ten year risk score, the low fat group decreased its risk by 0,1%, while the low carb group increased its risk by 0,1% . At the end of the trial, participants in the low fat group had a 0,7% ten year risk, while participants in the low carb group had a 0,9% ten year risk. This difference between the groups is too small to be clinically meaningful, but it is quite a difference from the prior study, which found a significant decrease in ten year risk with the low carb diet.
One thing to note before we get to the final conclusion, is that the carb intake in the low carb group was around 100-150 grams in both studies. While this is low carb compared to the standard western diet, it is far above what proponents of a ketogenic diet recommend. It is possible that a more severe carbohydrate restriction would show even greater weight loss benefits. These two studies are not able to answer that question. However, the first study was aiming for a ketogenic diet and didn’t even come close, so it is also quite likely that ketosis is not a realistically achievable long term state for the majority of people, even if it is effective in theory.
Final conclusion: Basically, from these two studies I think we can conclude that a low carb diet is at least as effective as a low fat diet for weight loss. Additionally, the data from these studies does not support the idea that a low carb diet increases risk of heart disease compared with a low fat diet. As an aside, that is another nail in the coffin of the cholesterol hypothesis.
You might also be interested in my article about whether exercise is an effective way to lose weight or my article about whether salt is bad for your health.
Interesting article, thank you. Did the studies break down the results by gender? Low carb / low fat diets have sometimes had different outcomes for males vs females, a difference in the numbers of male vs female participants could help explain why the two studies had different results.
Hi Lynette,
That is a good point. 89% of participants in the first study were women. In the second study, 57% were women. It is possible that a low carb diet is more effective for women than for men.
The “A to Z” study, led by Christoffer Gardener found that the advantage for the Atkins diet compared to other diets was mostly for the individuals with methabolic syndrome. So if you were overwight but without having the metabolic syndrome then low carb made little difference regarding your weight.
Hi Håkan,
Good point. The first study didn’t exclude people with metabolic syndrome while the second study did, so that could be part of the explanation for the difference between the studies.
Hello! What about earlier weightloss episodes among the participants? Often a new method to loose weight gains better effects than people who have tried ones with succes, then goes back to “normal” and gain weight again, will try again the same method but whit not so good result. It seems that the body will learn and tries to keep a balance not to loose energy, even if it is over weighted. Could here be a possible difference?
Hi Theresa,
It is possible. The first study excluded people who had experienced significant weight loss during the preceding six months, while the second study didn’t exclude those people, so that could be part of the explanation for the difference between the studies.
Some years ago I recall reading that one problem with these types of studies is that what people say they were eating often does not match what they actually ate once tests on urine etc were done. ( Sorry, I don’t recall where I read it. ) In many cases, low carb / low fat were not low. In some, the actual diet consumed was effectively the same across groups. Inevitable, outside a metabolic ward I suppose. So I tend to take all such studies with a little scepticism. An interesting article though. Thank you.
Hi Alex,
That is true, and is a weakness in all these types of studies. In the first study, they measured ketones in urine to confirm that the low carb group was indeed consuming less carbs.
Video presentation of A to Z study called The battle of the diets. Check after roughly 40 min for discussion on weight loss with Atkins for metabolically ill (“insulin restistant”) people.
I wondered why you didn’t mention the fact that a high carb diet raises blood glucose, and therefore insulin production which is known as the ‘fat storage hormone’?
That is thought to cause damage to the lining of the blood vessels and contribute to atherosclerosis, plus general inflammation, and type 2 diabetes as well. Fats and proteins, ie ‘low carb’, have far less impact on blood glucose levels as well as keeping us sated for longer.
As someone who grew up in the 60s in the UK, I can remember when we were all slim and anyone who happened to be overweight was said to have a ‘glandular problem’. At that time we had cooked school lunches (“meat and two veg”, not pasta or rice generally, other than maybe macaroni cheese once a month and rice pudding now and then) and I had another cooked meal in the evening when my dad got home from work! I was very slim, looking back at the photos, but I didn’t exercise much more than when dietary advice changed to low fat.
Once that happened, pasta-based meals and rice-based meals took off (they’re cheap, as well, so thought to be a “win/win” situation) – schools didn’t really teach cooking skills as time went on, so people started relying on ready meals and so on and there was hidden sugar in much of it. One of my school friends who was slightly ‘plump’, shall we say, said recently that her mum just fed them on sandwiches every day, as she’d had a school dinner, and processed breakfast cereals in the morning, and considers that was the reason for her slightly excess weight.
Hi Vanessa,
I think what you are writing is very sensible, and foods like rice and pasta that have high amounts of carbohydrates and very low levels of nutrients are probably a big part of the explanation for the global obesity epidemic. The reason I didn’t mention it was because it was outside the scope of the article, which was already on the long side. Since you ask, I will provide the data.
In the first study, fasting glucose levels in the blood stream had dropped 0,10 mmol/l in the low fat group, and increased 0,02 mol/l in the low carb group. Basically there was no statistically significant difference compared to baseline in either group. The same goes for insulin levels. There was a bigger drop in insulin in the low fat group than in the low carb group, but again, the change was not statistically significant.
A similar result was seen in the second study. Both fasting glucose and insulin dropped more in the low fat group than in the low carb group, but the difference between the groups was not statistically significant.
Those results seem strange to me too. I would also have thought you would see a bigger drop in people on a low carb diet than those on a low fat diet, and I don’t know how to explain it. There is clearly a lot going on metabolically that we don’t understand. One thing to note though is that it’s not only carbs that stimulate insulin secretion. Fat also stimulates insulin secretion.
is there an ethnicity aspect to this? I understand some parts of the World thrive better with high fat, low carb diets than other parts. New Orleans is predominately Black American, San Francisco White and Asian. Might it depend on the foods they returned to after the experiment?
Hi Sue,
That is also a possibility. In the first study about 50% of participants were African American, while in the second study only a few percent were. It is possible that people with African heritage respond more to a low carb diet than people with European heritage.
Could it be the meal replacement?
Hi Matt,
I also considered that, and it could be part of the explanation. But overall, the change in carbohydrate and fat intake was similar across the studies, which suggests that it wasn’t the meal replacement that was causing the change, since the people in the second study were accomplishing similar changes in carb and fat intake to the people in the first study even without the meal replacement.
Please, I beg you to read my blog: http://obesidadnoesunaenfermedad.blogspot.com/
WE THINK EXACTLY THE SAME, THAT IS WHY I THINK WE WOULD HAVE TO EXCHANGE OPINIONS (I AM 75 YEARS OLD, I HAVE BEEN A GENERAL DOCTOR FOR 46 AND I GRADUATED AS A GERIATRA FOR 39.
THE BLOG THAT I RECOMMEND YOU IS THE RESULT OF 32 YEARS OF DEEP STUDY OF THE FAT-OBESITY PHENOMENON)
Hi Cesareo,
Thanks for the link. I’m glad to hear you’ve reached the same conclusions as me! Undortunately I can’t read Spanish.
Hi, the second study was Dietfits, where Gardner tried to feed insulin resistant people low carb and insulin sensitive low fat. The liberal add what like -policy regressed to the mean. If you separate men and women, men responded better to low carb. Seems like low fatters reduced both carbs and fats! Insulin sensitive with less calories equals weight loss.
His collegue McLaughling found something more interesting; fat cell size. Low carbers could dimish the size, leaving more availble room for flexible storage. Low fatters did not diminish fat cell size. And logically, insulin fasting and peaks values were much smaller within low carbers. If you consider that LC increased their muscles (heavy) and LF lost a bit, the real difference becomes larger if you use mass comparison.
https://www.cosmosclinic.com.au/blog/low-carb-diet-shrinking-fat-cells/
The latter point is crucial to metabolic health, i.e. once your flexibility is used up, you are diabetic. This tips the scales in favor of LC, and strongly so. Hope that McLaughling publishes her study, so far we have press releases and presentations.
JR
You asked for suggestions that might explain the differences of outcome for the two groups. My experience is in technical marketing, not medicine. My thought is that it might be down to the method used to recruit the two groups. You do not say how the second study recruited their participants, so the groups may be drawing their subjects from different sections of the public possibly with different motivations for taking part.
Thanks Peter, that is definitely a possible explanation.
From the CV risk standpoint every source I have ever seen suggests that the critical factor is reduction of central adiposity by whatever diet is used. Furthermore I don’t think there’s any getting around the laws of thermodynamics so that calories in and out are the critical factor in achieving this goal. The elephant in the room is that strong neurohormonal physiologic mechanisms are in place that resist changes from the status quo.
I agree! I am trying to eat like my grandparents used to.
Interesting article. Thanks for the breakdown. The reason the 2018 study JAMA-published participants all made similar improvements in their health was because they were all given advice on limiting ultra processed foods, sugar, refined white flour, trans fats, and generally junk food.
They were all told to prepare real food from scratch at home. They were advised to focus on nutrient density. So the study didn’t just look at macros. They went out of their way to get people to eat ancestral diets in both groups and to stop eating their Standard American Diet.
I’m not sure why the authors didn’t have that as their main findings. Bit of a head scratcher really.
But this is what they said about their dietary advice (from the supplement on the study protocol):
“The fourth strategy is to promote high nutrient density (Quality). Other Quality concepts included “real food,” “minimally processed,” “seasonal,” “organic,” “grass-fed,” “whole grain,” and “pasture-raised,” depending on diet assignment. Both diet groups will receive similar instructions to drink water and to minimize added sugars, refined white flour products, and sources of trans fats. All participants will be encouraged to take an active role in making food choices; by preparing their own foods at home, reading labels, and asking for appropriate modifications for restaurant menu items.”
https://www.google.com/search?client=firefox-b-d&q=caldwell+esselstyn
https://www.google.com/search?client=firefox-b-d&q=Dr.+T+Colin+Campbell
https://www.google.com/search?client=firefox-b-d&biw=1920&bih=966&sxsrf=ALeKk03_JR38h0mmGmjuriPL9NRoX_AqDQ%3A1609154407896&ei=Z7_pX_SdNs_UkwWwkJHoAQ&q=John+McDougall&oq=John+McDougall&gs_lcp=CgZwc3ktYWIQDFAAWABgrvwBaABwAHgAgAEAiAEAkgEAmAEAqgEHZ3dzLXdpeg&sclient=psy-ab&ved=0ahUKEwi07fumx_DtAhVP6qQKHTBIBB0Q4dUDCAw
https://www.google.com/search?client=firefox-b-d&biw=1920&bih=966&sxsrf=ALeKk0225eJwPjdntTRN-pLV8soz6O0lEQ%3A1609154467117&ei=o7_pX_jUBoL7kwWixLeoDQ&q=Dean+Ornish&oq=Dean+Ornish&gs_lcp=CgZwc3ktYWIQDFAAWABg7BNoAHAAeACAAQCIAQCSAQCYAQCqAQdnd3Mtd2l6&sclient=psy-ab&ved=0ahUKEwi4sprDx_DtAhWC_aQKHSLiDdUQ4dUDCAw
Dead…
https://www.google.com/search?client=firefox-b-d&biw=1920&bih=966&sxsrf=ALeKk0225eJwPjdntTRN-pLV8soz6O0lEQ%3A1609154467117&ei=o7_pX_jUBoL7kwWixLeoDQ&q=dr+atkins&gs_ssp=eJzj4tTP1TcwTClKLzdg9OJMKVJILMnOzCsGAEqHBvg&oq=Dr+&gs_lcp=CgZwc3ktYWIQARgDMgQIIxAnMgcILhDJAxBDMgQILhBDMgQILhBDMgIIADICCC4yAggAMgIILjICCC4yAgguOgoILhDJAxAnEJMCOgcILhAUEIcCOgcIABAUEIcCOgUIABDLAToFCC4QywE6BAgAEEM6CAguEMcBEKMCOgcILhDJAxAnUNldWM-kAWDuwQFoAHAAeACAAZ8BiAGQDpIBBDAuMTOYAQCgAQGqAQdnd3Mtd2l6wAEB&sclient=psy-ab
The study is skewed as 30% of your calories from fat is not a low fat diet, so of course they wouldn’t have lost much weight if they consume that much fat. Low fat is 10-25% of energy from fat. 30% of your calories from fat is actually a lot, at 2000 calories that would be 67 grams of fat a day, which is a rather high amount and by no means can be classified as “lowfat”. This “study” is just trying to push a lowcarb agenda.
Absolutely agree with you. The only unclear question is what is the safe lower limit for fat intake? According to my review, the optimal calorical carbohydrate to fat ratio is 3.5 https://hozar.dreamwidth.org/file/163186.png.
Animal fat seems to be more harmful than vegetable fat https://hozar.dreamwidth.org/file/165469.png https://hozar.dreamwidth.org/715756.html
in my opinion, reducing fat from 34% to 30% apriori could not give an expressive result