Sunday, 18 September 2016

Post 24: Keto Study Analysis (plus building your toolbox)

Hi guys,

Sorry this one was a little late; I had to work Saturday and I prefer to release posts in the morning, and thusly, this Sunday read was violently thrust upon you…

I also want to admit, that this post is very self-indulgent. Not from a perspective of vanity, but from one of enjoyment. You see, I love bringing you information about the key concepts like intermittent fasting, my experience on the ketogenic diet, and the side effects of excessive fruit, I really do.

Pretty much me every week...

Each time I bring out another post, I want to explain things in ways that people haven’t heard before, and convey mechanisms in a manner which is easily understood. I also want to give you the tools to build back up to the advanced level. I’m hoping in that transferring this knowledge to you, you can improve and use your own critical eye, never taking an article on first impressions again.

The truth is that some people aren’t used to reading academic work, and why would they? Academics have a strong tendency to overcomplicate their vernacular, and use elongated words in order to bamboozle you and compound your confusion.

This leaves the world of research quite opaque to the layperson. What’s more impactful is the knock-on effects. If you can’t go to the source, then you must rely on a middle man. An intermediary who will ferry the information back and forth to you. This middle man is called the media. (Now we’re about to get conspiratorial- C) The problem with the media is that their incentives often are not setup to keep you accurately informed. They are driven to sell their product, and in particular draw eyeballs and clicks to generate advertising revenue.

Remember that today, the primary value you have to give to a company is your views. Your clicks are compiled into big data, given value, and sold.

So that’s why this post is probably quite self-indulgent, because this is one of the things I really want to get across: understanding the mechanisms and building your own toolbox. In doing so, we erode the significance of the middle man, and are a much knowledgeable, happier, and healthier population.

But how are we going to do that?

We're going to build your toolbox, but in a much less sinister way to this one...

If I asked for a drumroll, I’m afraid it would be incredibly anti-climactic, but we’re going to be reading the abstract of a recently published study and understanding what is being said. The title of the paper is “Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men”, which is a true mouthful.

It’s possible that at this point your finger is tentatively hovering over the little X in the corner of your screen but wait a second.

Before you run away, because to be honest, that does sound really, really dull, just read the next paragraph.

What this study is looking at is whether you burn more or less calories when on a ketogenic diet. Does that sound more appealing now? If you really can’t stomach the analysis, skip ahead to the section explaining what the results mean to you.

This study was run by the Nutrition Science Initiative (NuSI) who includes on its board of directors, Gary Taubes, who’s one of the most prominent figures in the low carb/ketogenic world (from now on abbreviated as LC/KD). This is another interesting point: this study is literally run by the people who you might expect to produce studies supporting their ideas! If the sugar industry can do it, then why not the “good guys”!?

The interesting point that emanates from this setup is that the paper rules in favour of ketogenic diet, but only slightly, and is not a complete knock-out. We’ll address this later by looking at the magnitude of the findings.

As well as learning about the results of the study, we’re going to be assembling the beginnings of your own analytical toolbox. Using this toolbox, you can start reading the abstracts and summaries of other articles and truly understanding what they say. Talk about teach a man to fish!

To begin, here’s the link to the study itself: The paper was originally drawn to my attention by the Mark’s Daily Apple blog, which I recommend reading to accompany this post.

First, I want you to familiarise yourself with just the layout of the abstract, which is quite common. People are constantly scared of the unknown, so learn how to get to grips with these things. For the uninitiated, an abstract is the summary at the start of a piece of academic research which lays out the purpose, method, and general results of the paper. It helps you to get the jist of the whole thing, but I’m assuming at this point it just looks like gobbledygook!

Here’s a glossary of the terms used and what they mean:

Energy expenditure (EE): calories burned in a period, typically a day.

Body composition: bodyfat %.

Isocaloric: means “the same calories”; so if two people both eat 2000 calories a day but one only ate butter and the other only ate bread, they would be on isocaloric diets (diets with the same calories).

Carbohydrate-insulin model of obesity: carbohydrate intake leads to insulin release. Insulin is a signal which stops fat burning.

Hyperinsulinemia: abnormally high insulin and hence, insulin resistance (your body is so resistant it has to release loads).

Respiratory Quotient (RQ): measured carbon dioxide production in the breath. We can tell which type of fuel you’re burning by the ratio of CO2 expelled to oxygen taken in. When this ratio is 1, your purely burning glucose. When this ratio is 0.85, you’re mainly using protein. When this ratio is 0.7, you’re purely burning fat. Therefore, we might say as a rule of thumb, that it is better when your RQ is lower, with the lowest being 0.7.

P-values: the research shows that the they have (100-P)% confidence that the true value (the population value i.e. if you averaged everyone in the universe) lies within this range.

I’ve put p-values at the end because they deserve special attention. This is because p-values are a quick way to view whether a result is statistically significant. Results can be statistically significant, and/or significant in magnitude. What this means, depending on which category they fall into, can be seen below:

Is Significant in Magnitude
Is not Significant in Magnitude
Is Statistically Significant
You have enough observations and your data has sufficient strength. The result would also make a material difference to someone’s life. You win.
Your data reveals a strong relationship, but would not make a very big different to someone.
Is Not Statistically Significant
Your data shows a relationship which could make a big difference to someone, but they are probably the lucky ones.
You suck.

So clearly the aim is to be in that top left hand box, where you can reasonably expect a certain result to arise from the process investigated and it will make a sizeable difference.

Taking an example from the paper, energy expenditure measured by doubly labelled water, the p-value was 0.03. That means that the researcher has (100-3)%=97% confidence that his findings are correct.

Commonly, papers will state that their finding are (statistically) significant at the 95, 99, or 99.9% confidence levels. All this means is that their p-values are under 0.05, 0.01, and 0.0001 respectively.

Breaking down the study’s results

So what exactly did the study say? I’ve written a layperson’s translation below, I hope that it reads as simply as possible,


Calories burned and fat loss after a ketogenic diet with the same calories in overweight and obese men.


It is thought by some that high carbohydrate consumption leads to fat gain because this leads to insulin release, which leads to lower calories burned. Therefore, if these people were to eat just as many calories of fat, then their metabolism should increase and they should burn fat. In contrast, when “a calorie is a calorie”, this swap from carbs to fats should have no effect.


We investigated whether a diet of the same calories but low in carbs/ketogenic leads to more calories burned, lower RQ and fat loss.


17 overweight or obese men consumed a high carb diet for 4 weeks followed by a ketogenic diet for 4 weeks which contained the same number of calories. Calories burned, calories burned while asleep and RQ were all measured as was bodyfat %.


The participants experienced a 300 calorie deficit per day causing them to lose water weight and bodyfat. The number of calories burned increased on the ketogenic diet (normal and sleep amounts) and RQ decreased. Body fat loss slowed during the ketogenic diet and loss of muscle increased.

Numerical Results

Energy expenditure on the KD measured by a metabolic chamber increased by 57 kcals per day on average. This result was highly statistically significant (discussion on whether this was significant in magnitude later).

Sleeping energy expenditure on the KD measured by a metabolic chamber increased by 89 kcals per day on average. This result was also highly statistically significant.

RQ (remember we want lower values to burn fat- C), decreased by 0.111 on average and was also highly statistically significant.

Energy expenditure measured by doubly labelled water increased on average by 151 kcals. This result had strong statistical significance, but not as much as earlier, which is shown by the p-value of 0.03.


The ketogenic diet did not increase fat loss but did cause small increases in calories burned

So what does this mean for you?

At this moment, possibly not a lot. This is where looking at the magnitude of the results comes in. Here, we find out whether anyone actually cares about the findings!

It’s so important to make the distinction, and ask whether the findings will make a difference to someone. If the findings were highly statistically significant, but only amounted to an extra one calorie burned, then no one would care. This is a key drawback to relying on a middle man for information because they can manipulate significance; the article might be referring to statistical significance.

So let’s tot up the figures, and we can reveal whether I’ve been wasting your time for 1700 words!

Remember that the results are given “per day”, so we need to account for that. Using the more conservative results provided by the metabolic chamber method:

16 hours awake / 24 hours * 57 = 38
8 hours asleep / 24 hours * 89 = 29.66
Total extra calories burned a day = 38 + 29.66 = 67.66

So that result seems quite piddly. A measly 67 extra calories burned a day is small change! I could eat that many calories in one bite!

But look a little deeper. That was 67 extra calories burned without doing anything but swapping over to a ketogenic diet. Extrapolating those numbers:

67.66 extra calories burned a day
473.62 extra a week

Using the rule of thumb that it takes a 3500 calorie deficit to burn a pound of fat:

0.13532 lbs a week
Over a year, that’s just over 7 lbs, without using calorie restriction at all!

But if only it were that simple!

I would not expect a seven pound weight loss every year from now to the end of the world, and  remember that this study was performed only on 17 obese and overweight people. This number could be different for leaner people and for the population as a whole.

It’s also unreasonable to think that if you go  on a LC/KD that you’re going to lose seven pounds every year without actually experiencing any difficulty. Taking that to it’s logical conclusion, eventually you will disappear!

There’s variation in there as well, which is given by the intervals; some people will burn more calories than others.

Another thing to remember, is that this was the effect after 4 weeks. What happens if you followed it for a year, or five years? Would the results taper off?

There are so many roadblocks that you could erect. You could go blue in the face trying to put this paper in disrepute, but it is the same for every other paper out there. No research is airtight, ever! The only thing you can do is take the information on board and adapt to it. At this point, you now know that it is at least possible to burn more calories while eating a LC/KD. Furthermore, we’ve been given an inkling as to one of the drivers of the insidious weight gain experienced by most westerners in the 21st century. I personally don’t see this paper showing me that I can lose 7 lbs a year by doing nothing; rather I could avoid gaining 7 lbs a year.

Supposedly this picture shows calories burned...

Closing tips and tricks, and takeaways

I hope you found this post useful.

I know I normally go for “enjoyable” or “entertaining” as the hoped for feeling, but this time I really want to start building your toolbox. We have the takeaways to address, but before we do, I want to introduce a list of questions to ask whenever you’re reading an article about health or (God forbid) a research paper:

  • Who funded the research? What are their incentives?
  • What is the aim of the article/paper?
  • Is the information being presented in a “flashy” manner? Why?
  • Does the author go to great lengths to create a balanced argument? Is this to their detriment? Are they clutching at straws?
  • Are you the target market for this piece?

And some LC/KD specific ones:

  • How long were the participants given to adapt?
  • Is the fat content greater than 65% of calories?
  • Were the participants given adequate sodium?
  • Were the participants exercising at the same time?

I’ve included the LC/KD questions specifically because these pitfalls occur all the time. Ideally, you want at least a three week adaptation period, 65% dietary fat, 3g of sodium per day, and some form of resistance training.

And before we go I want to give you your takeaways, and a homework assignment this week!* Your homework assignment is to read an article about health, which definitely doesn’t have to be a research paper. Just pick out an article you find interesting from any web source, and start asking yourself the questions above. Heck, you could even do it for one of my articles!

Here’s this week’s takeaways:
  1. Beware the middle man information - you should be asking questions!
  2. Swapping to a ketogenic diet will probably lead to burning more calories, especially if you are currently overweight or obese
  3. Don’t be scared of research, it can be broken down piece by piece and reassembled as something easy to understand

Until next time,

*To be honest, I’m inspired to write that by the bit in Fight Club where Tyler Durden tells everyone that their homework assignment is to go out and start a fight. So secret homework assignment part 2: start an argument with your colleague, friends, or (preferably and) family regarding carbohydrate consumption. Extra points for irreversibly breaking bonds in such a way you can never see that person again.


  1. That gives off an impression of being fabulous anyway i am still not very beyond any doubt that I like it. At any rate will look much more into it and choose by and by!  تحصیل در انگلستان بدون مدرک زبان