Introducing Evidence-based Keto: Your no-hype guide to the ketogenic diet
We've spent the past year analyzing the research on the keto diet, and have just released Evidence-based Keto.
Clocking in at over 200 pages with 500+ references, it's the unbiased guide you need to the ketogenic diet.
If you're a savvy nutrition enthusiast, you probably know that a variety of diets can improve health and lead to weight loss. But in the media, low-carb and low-fat diets have waged an epic battle over the past few decades. The latest comes from a year-long study of obese patients, which claims to shed new light on the topic.
Does this study tell us anything helpful, or does it just confirm what we already know? Headlines in the past 24 hours have largely touted the study as eye-opening, even tying it into the newly low-carb (and much thinner) Lebron James.
While most journalists are chasing the sensationalist story: “low carb proven to be superior,” few actually delved into the study to see what researchers found. When you read the full study, key nuggets jump out that most people have missed. Let's dive right in and break down the details. If you're short on time, read the blue boxes now and then come back and read the details later.
We'll analyze this paper using the PICO method, which researchers often use to frame randomized trials. It stands for:
[P]opulation - The study focused on obese people in Louisiana. Unlike most previous trials, this one included a sizable non-white population. Black people made up a little over half of the group.
[I]ntervention - Participants were split into two groups: the low carb group were told to eat less than 40 grams of carbs per day while the low fat group was to take in less than 30% of their calories from fat. The low fat group also aimed to keep saturated fat under 7% and carbs around 55%, which is in line with the current U.S. dietary guidelines.
[C]omparator - This just refers to the control group. The researchers wanted to know the effects of a low-carb diet (the intervention) and compared it against a low-fat diet (the comparator). The comparator is often the current standard of care. In this case, the low-fat diet was very close to what physicians typically prescribe for people at risk for heart disease.
[O]utcome - The outcomes researchers were most interested in were weight loss and cardiovascular risk factors like LDL, HDL, and waist circumference. They also measured a few other factors, like markers for diabetes.
This study was a bit different than previous low-carb trials. None of the participants had diabetes, heart disease, or kidney disease. Having healthier participants means that the researchers could apply the results to a broader group, not just people who were already sick. One thing that wasn't emphasized in the paper or media: almost 90% of the participants were female! If you're using this study to guide dietary recommendations for obese males, just note that only 9 out of 75 people in the low-carb group were men.
It should be noted that calories were not directly controlled. The participants reported an average intake of 2,000 calories per day before either diet - this seems implausible and emphasizes how inaccurate self-reporting can be.
Furthermore, The low-carb group started at a lower daily caloric intake than the low-fat group. By the end, both were eating slightly more and at roughly the same amount of calories.
Study participants were mostly obese women who didn't have diabetes or heart disease. The low carb group was told to eat less than 40 grams of carbs per day while the low fat group was to take in less than 30% of their calories from fat.
There was more to the intervention and comparator than just "low-carb" and "low-fat." Each group actually had two co-interventions along with their recommended macronutrient targets: Meal replacement: Participant were given a meal replacement bar or shake every day. These meal replacements were either low-carb or low-fat, as was appropriate based on their prescribed diet.
Counseling: The participants met up in small group counseling sessions, along with a dietician, in order to learn about dietary guidelines and receive support.
Both of these co-interventions likely impacted the real-world applicability of the study. If we assume that the best diet is one that a person can stick to, the addition of counseling and meal replacement (to ease the transition) greatly helped to prevent dropout. The completion rate was about 80%, which is very high for a diet study.
Both groups got a daily meal replacement shake or bar, and both groups also had regular dietary counseling. This likely increased compliance but might reduce applicability, especially if sticking to the diet is the hardest part.
Regardless of what the researchers intended, the study ended up comparing a diet that was just barely low-carb (with added protein) and a diet that could barely be classified as low-fat.
The low-carb group, while instructed to consume under 40 grams of carbs, did not manage that low of a number. The lowest average carb intake for the group as a whole was 93 grams per day at six months in, and the highest was 127 grams at 12 months in (down from 242 grams a day). Even the low-fat group ended up decreasing their carb intake considerably, but much less than the low-carb group.
That being said, the similarity of the low-fat diet to the macros of the standard American diet might have played a small role in the study results. The intervention group was prescribed a diet that many of the participants were not familiar with. Shiny new diets often provide benefits just because they are new. On the other hand, the low fat group underwent less drastic dietary changes. Unfortunately it's tough to design the perfect control group, and the perfect control group might even control for too many variables and be too difficult to follow, making the trial results difficult to apply.
As an example, let's say an astute researcher realizes that "low-carb" and "low-fat" are not indeed monolithic diets, and that two fat sources can have drastically different effects (e.g. fish oil versus hydrogenated soybean oil), as can two carb sources (e.g. fruit and a twix bar). If this researcher decides to test low-carb against low-fat but control for food quality and carb/fat sources, the trial could get very messy. Recommending an entirely new diet is hard enough, but getting extremely specific with food recommendations could lead to half the sample dropping out. Point being: everyone's a critic when it comes to analyzing studies, but designing a realistic trial is not that easy.
The low-fat group wasn’t that different in fat content from the standard American diet. The low-carb group was barely low-carb and also ate more protein.
Why did the low-carb group miss their carb targets so badly, not even getting close to 40 grams a day? Well, they are human. Some got close to 40 grams, some probably had life get in the way or just found out they really can't handle low-carb diets. Despite not reaching the target, the group did cut their carb intake by about 50% by the end of the trial, compared to baseline levels.
The low-fat group had a much easier change to make. Their baseline fat intake was around 35%, so to target 30% or less was not a very drastic change. Indeed, they ended up hitting 30% fat almost right on the nose after 12 months. Interestingly, at three months in, 25% of the low-fat group reported headaches, compared to only 8% of the low-carb group. Headache rates were much closer at later time points, and other adverse events were fairly similar between groups.
The low-carb group didn’t get close to the target of 40 grams of carbs - they were roughly 300% higher (127 grams on average). Changing your diet is difficult.
Note that the study did not report much on what exactly the participants ate. But we can surmise that the participants didn't end up eating diets much more rich in whole foods than they did before. Both groups ate around 17-18 grams of fiber at the beginning of the study. The new diets caused the groups to eat less fiber. Although eating less in general can lower fiber intake, the two groups likely did not adhere very strongly to the dietary guidelines encouraged by their counseling sessions, since women were recommended 22-28 grams of fiber a day.
It's also interesting that the low-carb group had a similar fiber intake as the low-fat group, so the argument that avoiding carb-rich grains is detrimental due to lower fiber intake doesn't hold true here.
The study didn’t report on the specifics of what foods were eaten, and neither group likely had large increases in fruit and vegetable intake. Both ate an almost equivalent amounts of fibre.
The most publicized result from this trial was weight loss - that the low-carb group had greater weight loss than the low-fat group. Both groups ate about the same number of calories (low-carb clocking in a bit lower). Both groups also ate more calories as the trial went on, with both experiencing an initial drop of about 500 kcal when their diets were given. Although non-significant, the low-fat group ate around 100 calories more at 12 months than at 3 months and the low-carb group ate around 200 calories more.
When looking at the maintainability of diet, both groups had similar caloric intake at 12 months.
Low-carb eaters also experienced a significant increase in lean mass as well, whereas the low-fat group actually lost a bit of lean mass.
Unfortunately fat mass was measured by bioelectrical impedance, not DEXA or a comparably accurate method. If you own one of those bodyfat-measurement bathroom scales, you know that their large measurement variances makes it hard to draw conclusions. Bioelectrical impedance relies heavily on total body water to calculate fat mass, and low carb diets are known to reduce water weight fairly rapidly which could potentially explain some of the rapid weight loss observed.
Since body fat and lean mass were measured through bioelectrical impedance, one can safely ignore any claims about exact fat loss and muscle gained.
Don’t make the mistake of attributing the group results to individuals. For example, weight loss in the low-carb group varied considerably, with some participants losing around 3 kg and some losing over 12 kg. The trial results don’t mean that you should expect to lose some specific number of kilograms on a low-carb diet. In addition, the first three months is when the weight loss happened in both groups. After that, each group experienced a slight uptick in weight, on average. This is another finding that is sometimes skimmed over by media reports.
Weight loss varied quite a bit between individuals within groups, and mostly happened within the first 3 months of the 12 month study.
A better gauge of fat loss would be the waist circumference. At 3 and 6 months in, the low carb group had a greater reduction in waist circumference. By 12 months in, the low fat group had caught up, and the decrease was the same in both groups.
Waist circumference decreased most rapidly in the low carb group, but was roughly the same between the groups at 12 months in.
Although the paper doesn't get into mechanisms much, we can guess why the low-carb group did better with weight loss and lean mass gain. Protein intake ended up being significantly higher in the low-carb group, which bodes well for weight loss and muscle preservation. The participants were instructed to not change their activity levels, although this was not really measured in the study. Thus, any potential effects of the diets on physical activity were nipped in the bud.
Fat loss in the low-carb group could have been partially due to a higher protein intake, which has been shown to have a positive effect on fat loss.
Considering the similarity in waist circumference reduction and both groups eating similar amounts of calories by the end, coupled with the increase in protein by the low carb group and water weight lost when reducing carbohydrates, it would be disingenuous to state that “low carb is superior to low fat for long term weight loss”.
The low-carb group had a higher HDL to total cholesterol ratio than the low-fat group, which is a strong predictor of heart disease. They also had lower triglycerides, in addition to a lower calculated heart disease risk score. Plus, their LDL dropped a bit more than in the low-fat group. To top it off, the low-carbers had a greater decrease in C-reactive protein, a measure of inflammation in the body.
Despite these improved markers, we can't be quite sure what caused them. Although the low-carb group increased their percentage of saturated fat out of total daily calories, they ended up eating about the same total grams of saturated fat compared to baseline, due to taking in fewer calories per day. The low-fat group on the other hand drastically cut down their daily grams of saturated fat. How much of the improvement in heart disease predictors was due to macronutrient changes rather than weight loss isn't known.
Also, the trial is not really conclusive on the subject of heart health since it didn't measure actual heart disease events nor did it look at LDL particle count and density, which are important predictors of disease. The researchers also measured blood pressure and some indicators of diabetes, but the groups didn't differ much on those parameters.
Many predictors of heart disease were improved by the low-carb diet, but the study was not designed to isolate the specific dietary cause of the improvement.
There were several important questions that the trial didn't answer. For example, was the weight loss and cholesterol improvement due to the low carb level or some other factor? It's possible that the low-carb group ate less simply because there were fewer options for them to eat, since many packaged foods sold at grocery stores are high in carbs.
Would the same result happen if you restricted, let's say, all packaged foods instead of carb-rich foods? Or taken to a ludicrous extreme, what if you limited foods that started with vowels? All of these reduce your food options, which is one way through which low-carb diets could lead to weight loss. Some of the weight loss could also be the result of a bit less glycogen (stored carbohydrate) in the liver and muscles. Glycogen can weigh a couple pounds or more due to its high water content, but less of it is stored in the body during a low-carb diet.
The researchers touted the trial's wider applicability compared to previous trials, since the study included more black people than previous studies. But the trial also happened to include zero Asian people and very few Hispanic people. Not to mention very few males! Combined with the fact that all of the participants were obese but with no history of heart disease or diabetes, this may limit the applicability of the trial.
The study didn’t answer the question of what it is about the low-carb diet that encourages weight loss, as it isn’t necessarily due to carb levels.
This trial is a mixed bag. It included a relatively large number of participants and ensured a high completion rate with counseling sessions and meal replacements. And since there was no calorie goal set, the trial could test just how a low-carb diet affected food intake in obese people.
But rather than proving that a low-carb diet leads to weight loss, the study shows that directing people to consume less carbohydrates might increase dietary protein intake. Was it the increased protein? Was it just fewer carbs (remember, this was not truly low carb)? Was the lower carbs/higher protein/higher fat diet simply more filling? And how does that apply to non-obese individuals? These are four very important questions that come out of this study.
Decreasing carbs and increasing protein leads to superior weight loss. The question is: what caused it, the decrease in carbs or the increase in protein?
At 12 months in, both groups were at roughly the same calories and had lost roughly the same amount on their waist circumference.
The claims of muscle gains and superior fat loss should be ignored as they were measured using bioelectrical impedance, which is essentially useless.
How does this trial inform public policy? Well, the U.S. dietary guidelines have long warned against saturated fat and encouraged carbohydrate intake. Unfortunately weight loss is often oversimplified to carb and fat intake, even though protein could have a big impact. When forming public policy on weight loss, a focus on quality foods rather than micromanaging macronutrients would be a step in the right direction. Factors like sleep and stress might have a bigger impact on weight than any one nutritional factor.
Directing people to consume less than 40g of carbs is not feasible policy. Carb intake may go down, but will likely remain over 100 grams/day.
A low-carb diet improved cardiovascular disease predictors, but the study was not designed to isolate whether the cause was weight loss or macronutrient intake.
Perhaps the biggest takeaway is that media headlines are not always informative, and it takes a thorough reading of the study and some methodological knowledge (much of the nitty gritty, which wasn't mentioned here) to understand what the paper really says.
A more accurate headline would have been: “If you are obese, decreasing carbs and upping protein may lead to greater weight loss, but sticking to any diet that has you eat less will lead to weight loss.”
Examine.com will soon start publishing a digest that analyses the latest nutritional studies and their context in the bigger picture. Be sure to sign up as an Examine.com Insider below to be notified when we publish our next analysis.
ON SALE UNTIL TONIGHT: Evidence-based Keto
Unbiased truth on keto and long-term effects on health, safety, pre-existing conditions, losing weight, and more.Learn the latest on keto