That Paleo Study ...

Random bump because this is making the rounds again due to being published in the IJES this month: 

Unrestricted Paleolithic Diet is Associated with Unfavorable Changes to Blood Lipids in Healthy Subjects

---

Original Publish Date:  5/14/2013

A student, Eric Trexler, at Ohio State did a study on the paleo "lifestyle" for what appears to be an undergraduate honors thesis towards his degree in Exercise Science.  Here is the link to the full text:  Paleolithic Diet is Associated With Unfavorable Changes to Blood Lipids in Healthy Subjects.  So I thought I'd put up a quick blog with some thoughts, because a tweet is just a tweet and I've seen a few comments reading into my supposed motivations for tweeting this.  So ... let's discuss a bit and include some criticisms I've seen elsewhere on the web.

1.  It's not a peer reviewed article or a graduate thesis.  This is true so it hasn't undergone rigorous scrutiny, that shouldn't stop you or I from reviewing it ourselves and seeing if it provides any worthwhile information.

2.  The study is confounded with participation in CrossFit, yet results are attributed to the diet only ... at least this is what the title implies.   Since there is at least a sizeable subset for whom CrossFit and Paleo are synonymous with "being paleo" I might suggest that the "Paleolithic Diet" be replaced with "Popular Paleo Lifestyle".

3.  "So the subjects were told to go forth and eat Paleo? Were guidelines more specific and compliance verified?"  -- this was a response on Twitter, and a valid one.  There is discussion of prior studies of the Paleo diet and the inconsistencies between the various prescribed diets.   Here is the diet description:

A Paleolithic diet, as first described by Eaton and Konner, was implemented for all study participants. Subjects were advised to increase their consumption of lean meat, fish, eggs, nuts, fruit, and vegetables and were instructed to strictly avoid all grains, dairy products, and legumes. All modern, processed foods including any form of processed sugar, soft drinks, and coffees were also excluded from the diets of the subjects.  

Say what??? No coffee!!!  There were no specific compositional recommendations for macro or plant v. animal intake.  As to compliance and evaluation:

Dietary recall logs were distributed to all subjects during the last [tenth] week of the dietary intervention and subjects were asked to record all food and liquids over three days, including a weekend day. 

For better or worse, this protocol is similar to that used in dietary interventions too numerous to count. However:

Our study does have some limitations. A low number of diet logs were returned and deemed sufficient for analysis (n = 8), which calls into question how accurately the returned logs portray the dietary intake of the entire group. Also, our study did not impose a high level of control over our subjects— meals were not prepared for subjects, nor did the subjects stay in a metabolic ward for the duration of the study.

I am concerned that such a small number of the participants returned dietary logs. Only 8 of 43 is less than 20% and I would say this brings into question the entire study in terms of the diet composition. Perhaps logs should have been distributed during week 5 so that those who failed to submit logs could be encouraged to do so during the following week. One wonders what the other 35 ate and/or if there was some reason (like "cheating" and having coffee with cream or butter) they didn't want to submit logs. In this regard this really is a weakness in this study that draws most of the results into question as to their usefulness.  I find that unfortunate as the relatively unstructured ad libitum nature of the diet prescribed, such as it was, already introduced this question of applicability.   However ... the authors stress that they wanted the diet to be "real life" and in this regard, it seems that all subjects at least followed some notion of the paleo diet.  

If any further publication of this study is in the works, I might suggest formally tabulating the dietary logs such as they are and perhaps doing the lipid analysis for reporters v. non-reporters to see if there are any significant differences.  The numbers might not be sufficient for statistical significance but it might be worth a look.  

4.  Unfavorable is In the Eye of the Beholder:  No doubt many will question the title of this article and it might surprise some that I would tend to agree.  The changes in lipids were not necessarily all that "significant" in the practical sense even though they reached the level of statistical significance in many categories.  Here are the results:  

The aerobic capacity and body fat improve while the unfavorable nature of the changes in TC, LDL and n-HDL are not so horrible as to be overly concerning.  Still, in most weight loss trials where participants lose 7 lbs = ~4% body weight, the trend for these lipids tends to be in the direction of lowering not raising.  Further, the unfavorable shifts are similar to the favorable shifts attributed to the paleo diet in other studies -- gotta take the bad with the good!  If you aren't "broken" to begin with, this study presents a cautionary tale as to what miraculous improvements one might expect to achieve by adopting the paleo/CF lifestyle.

5.  Eric Trexler is a Body Builder:  I saw an "oh noes" about this one ... to which I say "so what?"  If this were his n=1 experiment then it might have meaning, but he was only involved in conducting the study, not undergoing the intervention.  His background may have biased his hypothesis, but in the end the results are what they are.

Now ...  Speaking of criticisms like #5, and putting together some of the other points made, one sentiment I've been hearing is that collective brush off that some just want to be negative while this study isn't much if anything to write home about.   In a way I agree.  There simply isn't enough structure and/or documentation here regarding what the diet consumed really was, including variability among the subjects, etc.  That said, compared to other "studies", it can't be ignored.  What other studies?  Well, one issue from which most of the contention with Robb Wolf evolved was basically trying to get a straight answer on just what the diet involved is in his Reno first responder work.  Near as I can tell it's low carb as a general guideline and he's trying to steer it more to Lindeberg-style paleo.  Which brings me to a moment of giddiness for me when I tweeted to Staffan Lindeberg himself and he responded!  Preceding this tweet, someone had wondered if the CrossFit component might have impacted the results.

Eggs cooked in butter - coffee with butter.
 #mmmbutter #lccruise13

As I've discussed on this blog, the paleo diets used by Lindeberg's research group tend to bear little resemblance to the paleo diets we hear about coming from PaleoFX and various gurus in the community.  For starters, they are almost fruit-centric (by mass the largest component) and not fatty meat fests with butter on top!  It's almost laughable that Practical Paleo's Diane Sanfilippo posted the picture at right,  just two days before, of her breakfast on the LC cruise ... Can you say bacon, low-fruit, no-starch "paleo"??  There's more fat pictured at right than consumed in an entire day in the "classic" Lindeberg RCT or perhaps even the Jonsson (his research group) study, and certainly more saturated fat.

I would also mention that eggs are "overconsumed" in the paleo community if Lindeberg's studies are to be any benchmark for a paleo diet conveying health benefits.  In Lindeberg, average egg consumption was 29 g/day, vs 71 g/day for Jonsson ... that's basically one-half to one jumbo egg per day.  (I would note that negligible oils and no butter consumption were reported for each study).

Whatever one's position on saturated fats, if you're going to point to the research out there that is testing paleo, it is disingenuous to promote the diet depicted above as paleo yet point to any of the studies summarized in the table here and say they support your dietary advocacy.  

So in the Trexler study, with the caveat that these were logs for only 8 of 43 participants, the diet came to roughly 50% fat of which 13% was sat fat.  There's no indication of caloric intake but:  

Previous studies have shown positive effects of Paleo interventions on a number of health markers [2, 7-10], and in some cases blood lipids [2, 9]. However, these studies used more restrictive Paleo guidelines, and many used subjects that were initially overweight [2, 8, 9] and were either sedentary [2], had coronary artery disease [8], or had type 2 diabetes [9]. There is a general lack of literature pertaining to Paleo interventions in healthy, highly active populations, and there is reason to question whether previous Paleo studies accurately replicate the actual dietary intake of the free-living Paleo dieter.  

Diet log analysis revealed that subjects in our study reported higher intakes of total fat, saturated fat, and cholesterol than any of the previously mentioned human interventions [2, 8-10]. While our subjects reported daily intakes of 115 g total fat, 29 g saturated fat, and 673 mg cholesterol, the four previously mentioned studies averaged drastically lower daily intakes of total fat (67 g), saturated fat (15g), and cholesterol (426 mg) [2, 8-10].  {10 is the Osterdahl study I haven't included yet, Ryberg in my above table was fairly recent hence not included in Trexler}

So by my calculations, if 115g fat was ~50% of calories, caloric intake averaged 2070 cal/day.  If we presume 20% protein and 30% carbs this works out to roughly 105 and 155g respectively.  Not an over indulgent diet macrowise and the results weren't so bad.  Here's what I think is important:

When stratified by initial blood lipid levels based on the ATP III, there was a significant decrease of HDL among subjects with the highest initial HDL (82.1 ± 3.2 mg/dL to 68.6 ± 4.8 mg/dL; P < 0.05; Figure 2) and significant increases of n-HDL among subjects with the lowest levels of n-HDL (86.6 ± 3.9 mg/dL to 101.4 ± 4.8 mg/dL; P < 0.01; Figure 3).  Furthermore, significant increases of LDL (69.1 ± 3.1 mg/dL to 83.5 ± 4.1 mg/dL; P < 0.01; Figure 4), TC (157.2 ± 0.7 mg/dL to 168.2 ± 0.9 mg/dL; P < 0.05), and TC/HDL (2.5 ± 0.1 to 2.7 ± 0.1; P < 0.05; Figure 5) were observed among subjects with the most optimal initial levels of each respective outcome variable.

If one looks at a really short study, like Frasetto, and the improvements there (on a diet that resembles NO pop Paleo™ version) in 10 days, it is difficult to shrug off the trends seen in this study in a relatively short 10 weeks for those who are already in general good health.  The higher fat consumption at least points to these participants following a more pPaleo™ version of the diet.  This is not surprising to me.  Anyone who reads just a smattering of blog posts or forums like PaleoHacks is likely to encounter the far-more-frequent-than-most-care-to-admit posts about how egads their cholesterol went bananas when they gave up the bananas.  The response to these negative developments are predictable.  Either (1) you're doing it wrong, eat more fat and less carbs or, you may have SIBO or AF or quit the nightshades, or whatever ... just "paleo harder!" and get more sleep ... and if that fails, (2) these biomarkers are meaningless anyway.  Sometimes folks just skip to (2).   In the end, the negative impacts of paleo are rarely addressed seriously, because, after all, what harm can be done eating real food and rebelling against conventional wisdom?  Oh ... and look at all the people fill-in-the-blank-so-and-so is helping!!!

Then there's the knock over the lack of compliance control in this study.  Fair enough.  But many of the critics on this point are all too happy to tout the unverifiable n=1 success stories someone going by the handle IHeartPaleo scatters about the internet or various bloggers feature as success stories (where often we have just pictures and no bloodwork).  Or how about that crowd sourced data over at the Ancestral Weight Loss Registry where you don't even need to provide a real name or email address to be included?  At least here you have 43 people who signed on to a trial and submitted to testing before and after.  In the end, this study is a data point.  One that carries less weight than peer-reviewed RCT's but more than anecdotes on the internet where lipids are not often shared and/or cannot be verified.