Sunday, February 15, 2009

Paleolithic Diet Clinical Trials Part III

I'm happy to say, it's time for a new installment of the "Paleolithic Diet Clinical Trials" series. The latest study was recently published in the European Journal of Clinical Nutrition by Dr. Anthony Sebastian's group. Dr. Sebastian has collaborated with Drs. Loren Cordain and Boyd Eaton in the past.

This new trial has some major problems, but I believe it nevertheless adds to the weight of the evidence on "paleolithic"-type diets. The first problem is the lack of a control group. Participants were compared to themselves, before eating a paleolithic diet and after having eaten it for 10 days. Ideally, the paleolithic group would be compared to another group eating their typical diet during the same time period. This would control for effects due to getting poked and prodded in the hospital, weather, etc. The second major problem is the small sample size, only 9 participants. I suspect the investigators had a hard time finding enough funding to conduct a larger study, since the paleolithic approach is still on the fringe of nutrition science.

I think this study is best viewed as something intermediate between a clinical trial and 9 individual anecdotes.

Here's the study design: they recruited 9 sedentary, non-obese people with no known health problems. They were 6 males and 3 females, and they represented people of African, European and Asian descent. Participants ate their typical diets for three days while investigators collected baseline data. Then, they were put on a seven-day "ramp-up" diet higher in potassium and fiber, to prepare their digestive systems for the final phase. In the "paleolithic" phase, participants ate a diet of:
Meat, fish, poultry, eggs, fruits, vegetables, tree nuts, canola oil, mayonnaise, and honey... We excluded dairy products, legumes, cereals, grains, potatoes and products containing potassium chloride...
Mmm yes, canola oil and mayo were universally relished by hunter-gatherers. They liked to feed their animal fat and organs to the vultures, and slather mayo onto their lean muscle meats. Anyway, the paleo diet was higher in calories, protein and polyunsaturated fat (I assume with a better n-6 : n-3 ratio) than the participants' normal diet. It contained about the same amount of carbohydrate and less saturated fat.

There are a couple of twists to this study that make it more interesting. One is that the diets were completely controlled. The only food participants ate came from the experimental kitchen, so investigators knew the exact calorie intake and nutrient composition of what everyone was eating.

The other twist is that the investigators wanted to take weight loss out of the picture. They wanted to know if a paleolithic-style diet is capable of improving health independent of weight loss. So they adjusted participants' calorie intake to make sure they didn't lose weight. This is an interesting point. Investigators had to increase the participants' calorie intake by an average of 329 calories a day just to get them to maintain their weight on the paleo diet. Their bodies naturally wanted to shed fat on the new diet, so they had to be overfed to maintain weight.

On to the results. Participants, on average, saw large improvements in nearly every meaningful measure of health in just 10 days on the "paleolithic" diet. Remember, these people were supposedly healthy to begin with. Total cholesterol and LDL dropped. Triglycerides decreased by 35%. Fasting insulin plummeted by 68%. HOMA-IR, a measure of insulin resistance, decreased by 72%. Blood pressure decreased and blood vessel distensibility (a measure of vessel elasticity) increased. It's interesting to note that measures of glucose metabolism improved dramatically despite no change in carbohydrate intake. Some of these results were statistically significant, but not all of them. However, the authors note that:
In all these measured variables, either eight or all nine participants had identical directional responses when switched to paleolithic type diet, that is, near consistently improved status of circulatory, carbohydrate and lipid metabolism/physiology.
Translation: everyone improved. That's a very meaningful point, because even if the average improves, in many studies a certain percentage of people get worse. This study adds to the evidence that no matter what your gender or genetic background, a diet roughly consistent with our evolutionary past can bring major health benefits. Here's another way to say it: ditching certain modern foods can be immensely beneficial to health, even in people who already appear healthy. This is true regardless of whether or not one loses weight.

There's one last critical point I'll make about this study. In figure 2, the investigators graphed baseline insulin resistance vs. the change in insulin resistance during the course of the study for each participant. Participants who started with the most insulin resistance saw the largest improvements, while those with little insulin resistance to begin with changed less. There was a linear relationship between baseline IR and the change in IR, with a correlation of R=0.98, p less than 0.0001. In other words, to a highly significant degree, participants who needed the most improvement, saw the most improvement. Every participant with insulin resistance at the beginning of the study ended up with basically normal insulin sensitivity after 10 days. At the end of the study, all participants had a similar degree of insulin sensitivity. This is best illustrated by the standard deviation of the fasting insulin measurement, which decreased 9-fold over the course of the experiment.

Here's what this suggests: different people have different degrees of susceptibility to the damaging effects of the modern Western diet. This depends on genetic background, age, activity level and many other factors. When you remove damaging foods, peoples' metabolisms normalize, and most of the differences in health that were apparent under adverse conditions disappear. I believe our genetic differences apply more to how we react to adverse conditions than how we function optimally. The fundamental workings of our metabolisms are very similar, having been forged mostly in hunter-gatherer times. We're all the same species after all.

This study adds to the evidence that modern industrial food is behind our poor health, and that a return to time-honored foodways can have immense benefits for nearly anyone. A paleolithic-style diet may be an effective way to claim your genetic birthright to good health. 

Paleolithic Diet Clinical Trials
Paleolithic Diet Clinical Trials Part II
One Last Thought


theoddbod said...

super fascinating. it would have been interesting if they had split the participants up into two groups and done a cross-over type study to see if their conditions worsened again when they returned to their normal diets allowing them to be their own controls.
i'm really diggin' this paleolithic diet info, keep it coming!

Anonymous said...

thanks for the info. I always read stuff on low-carb diets, but this paleo diet trials are awesome.

gunther gatherer said...


Is there a particular reason the researchers specifically singled out potassium chloride as something to take out of the diet? I know if it's an additive, it can't be paleo, but I looked it up and can't find out what foods its in.

I assume they took out table salt and sugar too, though BOTH are present in mayonnaise, so that could confound results. Table salt is a poison used to kill weeds, so imagine what it does to us when we consume it.

I assume also the animal fat wasn't grass-fed or organic, and since that's where the animal stores toxins, the subjects ironically got quite a few more toxins than they might have on a low fat diet!

Still great stuff. Thanks.

Alexandra said...

I'm not a trained scientist, but I do understand the basic premise of a control group, double-blind, etc. However, when you are studying the effect of diet on people, perhaps comparing them to themselves is actually more accurate, due to all the factors you point to that make each person react differently to changes in the diet. Just a thought.

This just confirms what we found in our house doing this ourselves - my husband's weight dropped and his numbers improved when I got him on the Paleo diet. As he slipped off it, he got worse again. Now he really believes it and hopefully is on the wagon for good!

Sonja Pauly said...

I really appreciate this information. I have switched over to a more paleolithic diet and my health has improved dramatically!

My doctor was so impressed after my physical checkup.

Cheers & Good Healthy!

Monica said...

"Mmm yes, canola oil and mayo were universally relished by hunter-gatherers. They liked to feed their animal fat and organs to the vultures, and slather mayo onto their lean muscle meats"

You're cracking me up, Stephan. You sound a bit like Sally Fallon except that you didn't mention the inedible boneless, skinless chicken breast. mmmmmm.. :)

Very interesting results. I'll be sharing this post with various family and friends.

Anna said...

Maybe they made the mayo from scratch, instead of bottled mayo. Sure, that doesn't make it paleo, but it's a heck of a lot better than Hellman's or Miracle Whip.

Robert M. said...


Yeah I felt the insulin drop was the most important result to take home too. What I found especially impressive was how the standard deviation on the insulin levels also went way down, which implies uniform results on the diet, in a few weeks.


The authors are advocates of the acid-base hypothesis, hence the interest in potassium intake.

Stephan Guyenet said...


See Robert's comment below.


It's nice to have subjects be their own controls, but there are a number of things you can't control for in that type of design.


I doubt it was homemade. It was probably one of the designer, high-omega-3 mayonnaises.


Good point, the standard deviation is the place to see that effect. I added that to the post.

Fitnatic said...

So there is hope for me on the not so rigid Paleo diet. good to know. :)

Senta said...

Thanks for another great post, Stephan. It's very impressive that they saw such substantial improvements in measurements so quickly. I imagine there wasn't much funding to allow a larger cohort, control group and longer period of study. Hopefully these kinds of results will encourage larger and better constructed studies.

I too wonder about the quality of the food (grass-fed, organic, or what?) and also, what types of vegetables they ate. Since they eliminated potatoes, did they also eliminate sweet potatoes, yams or other starchy vegetables? Depending on the quality of the animal foods and whether or not they ate organ meats, what quantity of fat-soluble vitamins could they have gotten?

Lots of questions but still encouraging!

Anonymous said...

Eades' blog has a breakdown of what they ate. It appears that fructose intake was pretty high (lots of honey and juices, no starches). Interesting in light of the extensive info on this site and others documenting the potential dangers of fructose. I wonder if Mr. Bruce K will comment on this.

Robert Andrew Brown said...



Do you have a link to Eades blog I would be fascinated to see the diet details.

I am one of those who believes excessive fructose is a significant factor in encouraging weight gain, but always delighted to look again.

Anna said...

Here's Dr. Mike R. Eades' blog post:

daiikkon said...

A little off subject but if you had a choice between consuming regular store bought butter or regular store bought olive oil, which would be the least bad for you?
I ask because sometimes I am at friend's houses and most do not care about what they eat but I do.

Thank You

Stephan Guyenet said...


Good point, it looks like starches were low but sugars were relatively high. Although that doesn't make much sense, it's hard to eat enough sugar to maintain a typical carbohydrate intake when you aren't eating much starch. Could they really have been getting 250g of carbohydrate a day from non-starchy vegetables, fruit and honey? That seems hard to believe.

Anyway, here's my interpretation: getting rid of wheat and processed foods in general, and increasing omega-3, yielded health improvements. The increase in sugar and PUFA didn't impede that in the short term. But I can't imagine high PUFA and high sugar (even from fruit) is healthy in the long run, in any dietary context.


I'd go for the butter. Olive oil is fine too as long as it's extra virgin.

Carl M. said...

If you eliminated cooked starches, you can eat a LOT of fruit -- at least for a while. Google around for "Instinctive Eating." I tried it myself (save that I cooked my meat) and the amount of fruit I ate was way beyond what was ordinary.

But I still lost 30 pounds in a month. And after a while I got sick of fruit. I should have eaten more meat in retrospect, but Severen Shaeffer's book indicated meat as a lower fraction of the typical instinctive eater's diet than for a typical American cooked food eater.

So yes, the 250g figure is believable -- for a 10 day study. Maintaining it for a long time is problematic.

JBG said...

One way to deal with the study as "nine anecdotes" is via the binomial test. (My time of mixing it up with statistics was 40 years ago, but I think this use of the test is kosher.)

The one-tail P value is 0.0195
This is the chance of observing 8 or more successes in 9 trials.

Even if we allowed that the outcome of the experiment might have been that participants would have fared *worse*, the P of a result as extreme as 8 out of 9 would still be less than .04.

The probability of a chance result of all nine being favorable is about .002 (and of all nine being either better or worse, about .004).

Anonymous said...

I think the low carbers need to take a close look at this study. Great insulin improvement in 10 days on 250 grams of carbohydrate - mostly sugar!

I also think this study is interesting in showing how quick and easy it is to achieve a low fasting insulin level. I assume that there is a long road to walk from being the average american to achieving optimal health - possibly years of effort. If insulin levels return to normal after the first week and a half, then what does this say about insulin level as a indicator of optimal health? Perhaps that it is a necessary but nowhere near sufficient condition for optimal health?

Stephan Guyenet said...


Good point.


I agree, insulin is not the end-all health marker. It is important though. It's worth noting that some people begin to feel better in a matter of days when they switch diets.

Robert M. said...

I figure one can in the process of making ghee from commercial butter add powdered fat-soluble vitamin to the mix while boiling the butter. I haven't tried it yet since I still have about a litre of ghee to eat but food for thought (literally).

Anna said...

"one can in the process of making ghee from commercial butter add powdered fat-soluble vitamin to the mix while boiling the butter"

Lately I've been able to buy pastured butter, either raw (Organic Pastures) or pasteurized (Organic Valley Pasture, or KerryGold Irish butter). I stock up and store it in a freezer bag in the freezer.

But when I'm using milk, cream or butter with undetermined access to pasture (like TJ organic butter or half & half, I've been adding a drop or two of the Thorne K2 oil before serving, such a pot of cream soups, a quart of yogurt, etc. I usually add it to something that contains several servings, not just one, as the K2/drop is really concentrated.

As you describe with the ghee making, I think it would be really easy to "re-K2" some dairy foods with the drops. The Carlson D3 2000iU drops would be used the same way.

donny said...

Dr Michael Eades posted about this study the other day. He posted the actual diet, meal by meal. The Paleo phase included four servings of carrot juice a day. I checked fit day, if a serving's a cup, thats around 90 grams of sugar just from the carrot juice. Half of it fructose. You'd get the same from a couple cans of coke. So maybe there's some stuff in carrots that helps you handle the sugar. I think Eades said something about the carrot juice being added as a way to bring the potassium intake up to estimates of paleolithic intake.
Of course it could just be the fact that the diet was strictly controlled, and probably lower sugar than the people's usual diets. But we have too many paradoxes where peoples eat more carbs than we do (to match the ones where people eat more fat than we do, present company excluded) to assume that this is the case.

Robert Andrew Brown said...

Thanks Donny and Anna,

Thought provoking.

If I buy a packet of something sweet I need to finish it, but do not get the same feeling after carrot juice.

I wonder what would have happened if they swopped the carrot juice for a coke or sugar.

What would happened if you spread a regular diet over the same time frame in similar calorie increments.

Jeff Consiglio said...

"It's interesting to note that measures of glucose metabolism improved dramatically despite no change in carbohydrate intake."

This is indeed very interesting. Now of course paleo tends to favor fruits and shun starches. But I'm not so sure that is a real issue. In fact, fructose appears to have more harmful metabolic consequences than even pure glucose. Thus, I am a bit skeptical of Cordain's extreme enthusiasm for fruit, and his fear of starch.

I have to wonder of it's the increased potassium. magnesium, etc, on his diet that allows one to get such good results, even while still consuming ample carbs.

I think they call this the "osmotic theory" of insulin resistance? That certain minerals profoundly affect one's ability to properly process carbohydrate.

Robert Andrew Brown said...

Coach Jeff

Thanks for those thoughts.


DO you have nay links for the impact of minerals on insulin function.

Many thanks.

Stephan Guyenet said...

Hi Jeff,

My suspicion is the diet was effective more for the things that were removed than the things that were added. What do low-carb diets and paleo diets have in common? They're both effective for weight loss and overall health, and they both reduce wheat, processed food in general, and sugar (except in this trial

I agree with you that people in the paleo community may be unnecessarily afraid of starch. I've gradually become convinced that it's not so bad, if you get it in the right form.

Paul Czajka said...

Fantastic series of posts - it's nice to see a level discussion of the real research out there, as sparse as it is. Helps cut through the unsubstantiated, buzzword-ridden hype surrounding most diet-related content online.

So how does Quinoa fit into the paleolithic diet? It's not a grain per-se, so I don't know if it would have the same health impacts that grains cause. Are you aware of any information which suggests Quinoa can be consumed without adversely impacting the benefits of a true paleolithic diet?

fab said...

This is not a good study.... it lacks any scientific rigor. I teach statistics and if any of my students came to me with a biased not randomised trial with nine observations I would have to fail them. You need a minium of 30-40 just for central limit therom to even have a chance.

Please refer to the law of small numbers. The smaller your sample size the higher your variation of results. This is not real research and a credible professor should know better than to pass something like this off as a legitimate clinical trial. I am hoping that someone actually does a robust clinical trial (double blind, randomized control test with a large sample size) on paleo diets because so many people have been selling it hard.

Stephan Guyenet said...

Hi Fab,

I agree about the lack of randomization and a control group. That's a major limitation, and I think the study should be viewed as a hypothesis-generating exercise.

However, I have to disagree with your claim that you need n=30-40 to do proper statistics. If that's the case, you can throw out more than half the published biomedical literature, including my lab's entire body of work. Animal studies rarely use cohorts that large, and the stats work just fine. I've never heard any statistician, including my stat profs or the professional statisticians we consult with, take issue with our cohort sizes. Smaller cohorts simply mean less power to detect differences, it doesn't mean statistical tests no longer apply.

Your expectations for randomized diet trials are unrealistic. It's virtually impossible to do a double-blind trial of any diet, particularly for something like the Paleo diet, because how do you prevent participants from seeing and tasting the food they're eating? If you think we have to do double-blind RCTs to establish knowledge about the healthfulness of dietary patterns, then that implies that we know precisely nothing about the healthfulness of any diet pattern, since no dietary patterns have been tested in double-blind RCTs due to the virtual impossibility of doing so.

Double-blind trials work great for pharmaceutical studies but are unsuitable for studies on diet patterns such as Paleo. The best researchers can do is a randomized, controlled, single-blind study. There are published RCTs of the Paleo diet, although I don't recall whether or not they were single-blinded.