Perfect Health Diet Macronutrient Ratios ~ Part II: Omnivores Prefer These Ratios


In the Perfect Health Diet books, the claim is made that omnivores prefer a high fat diet along the lines off the PHD macronutrient ratios.  Two studies on mice are cited as evidence in support of a 65% fat diet.  This post discusses those two studies, as well as how the results are more likely to caution against a high fat diet than advocate for it.  

NOTE:  This is NOT a comprehensive look at the two studies discussed in the post.  In order to keep the post length managable I've only addressed the points relevant to claims made in the PHD books citing these studies.

In the (no longer available) late 2010 print edition of the Perfect Health Diet (also published on Kindle sometime in 2011) by Paul and Shou-Ching Jaminet (who make no mistake lends only her name and credentials to these books),  the justification for a high fat, lowishcarb, "normo" protein diet was laid out. While the recommended macro ratios have subtly changed over time, the reasons for the original ratios have not changed. I blogged about this here: Perfect Health Diet Evolving??.  In  a preceding ebook version and the 2010 printing (hereinafter referred to as PHD-I), Paul laid out his rationale in an early chapter:

Four Reasons to Believe in a 20-65-15 Macronutrient Ratio [Carb-Fat-Protein]
  1. These were the ratios eaten by our hunter-gatherer ancestors
  2. Mother’s milk is a complete and perfect food for infants, and the perfect diet for adults must be similar. 
  3. You should eat what you are.
  4. Omnivorous animals prefer and thrive on low-carb high-fat diets
In Perfect Health Diet Macronutrient Ratios ~ Part I: Breast Milk , I paraphrased these reasons, the four provided above, however, are verbatim from the book.  The late 2012 Scribner version (hereinafter PHD-II) contains these same reasons only in less succinct fashion.  Each is presented in a separate chapter, with #4 incorporated into a chapter on "nutrient transformation".  This concept was in its own chapter in PHD-I.   Paul believes that all mammalian tissues "eat" the same high fat diet regardless of the macronutrients that we consume.  This occurs through the magic of nutrient transformation in digestion and is a concept that Barry Groves came up with ... this absurdity needs its own post to address!

Below, I've screencapped and quoted the relevant portions from PHD-I, and  PHD-II related to omnivore preferences.

This morphed into the excerpt below from PHD-II:
Rodents are omnivores that, in the wild, eat a lot of seeds. Seeds, like nuts, often contain substantial amounts of fat. 
In the laboratory, mice are usually fed a chow that consists mainly of cereal grains, which are very high in starch. So the standard laboratory mouse diets are high-carb diets. 
But what do mice really want to eat? When scientists let mice choose their own food from an unlimited supply of carbs, fats, and protein, most strains choose to get a majority of their calories from fat. In a study of thirteen strains of mice, nine chose to get a majority of calories from fat. Only two chose to consume more carb than fat calories. 
When scientists allowed a wild-type mouse strain that develops obesity and diabetes on a 40 percent carbs, 40 percent fat chow to choose its own diet, it chose a diet of 5.6 percent carbs, 82.5 percent fat, and 12.0 percent protein and “proved highly resistant to the development of obesity and diabetes.” In the same study, transgenic mice genetically engineered to be even more prone to obesity and diabetes self-selected to a diet of 2.2 percent carbs and 85.1 percent fat and “developed obesity [that was] . . . less pronounced than on a high-fat and high-carbohydrate Western diet . . . [and] did not become hyperglycemic; they showed decreased fed and fasted blood glucose levels.” 
We probably can’t infer much from genetically mutated mice that avoided obesity by choosing 85 percent fat diets. But it is telling that most strains of wild-type mice chose to get a majority of calories from fat, a minority from carbs and protein— the same pattern we’ve seen repeatedly in other mammals, in humans, and in breast milk of all mammalian species.
Perfect Health Diet: Regain Health and Lose Weight by Eating the Way You Were Meant to Eat (Kindle Locations 883-897)

Here are links to the full texts of the two cited studies:
Macronutrient diet selection in thirteen mouse strains (I'll refer to this study by lead author Smith)
Self-Selected Macronutrient Diet Affects Energy and Glucose Metabolism in Brown Fat-Ablated Mice.  (I'll refer to this study by lead author Ortmann)

Mice as Poster-Omnivore & Mammalian Diets:

There may be a mistake or two on the list here, but the term omnivore -- an organism that consumes both animal and plant matter -- applies to a long list of diverse species, and several mammals at that.  To make a serious case for a high fat diet for all omnivorous mammals based on the dietary preferences of one species is, frankly, laughable.  

At the very least, for Reason 4 to be included as a BASIS for the PHD, two or three representatives of omnivores would need to enter the discussion.  It is hard to believe that Paul could even think to extrapolate the choices of some of the mice to justify a "perfect diet" for all humans.  But he did.

This is the focus of this post.  

The Healthy Choice Diet??

Paul made a rather outrageous statement PHD-I that didn't make it into PHD-II, and that's this:

Animals instinctively eat a mix of foods that is healthy.

Really?  How so?  Are humans then the exception to this rule, because quite often we don't seem to instinctively choose what's best for us.  Were the "cafeteria rats" who all but ignored their chow in favor of human junk foods acting on healthy instincts?!   

Statements like this may seem inconsequential, but Paul makes a lot of them as regards things like appetite and intake, consuming various components of foods in some "sweet spot", etc.  Anyone who believes every human will do equally well on some narrow range of macro ratios is seriously missing the point.  That even a small part of the reasoning comes from mice in unnatural experiments is troubling to say the least.   Which brings me to ...

Rodent Nutrition:  Lab Chow & The Healthy Choice Diet

I've discussed on a number of occasions some of the benefits and shortcomings of rodent nutrition in laboratory experiments.  Writing this post prompted me to put a few thoughts together in a reference post of sorts rather than adding too much length to this one.   Your typical lab rodent consumes the exact same thing, pressed into a complete homogenous pellet, day in and day out for it's post-weaning existence.  It is usually given free access to this food, but only this food, and fresh water, and there's little else for the animal to do in its confines.  This is quite different in the wild where there would most certainly be variety, and the resourceful rodent will eat seemingly anything they can find, but they have to find it first!  As with all wild animals there's a balance of effort expended to energy obtained that comes into play.

Mice are omnivores, but they consume a lot of plant matter, and eat mostly the seeds of these plants because ... drumroll ... that's where the calories are!!  Your typical rodent chow is often higher in protein and carbohydrate and lower in fat than the human diet:  fat 10-15% , protein seems to vary quite widely from 15-35%, carbohydrate 50%+. The ingredients are usually ground grains, fish/poultry/meat meal, oil or lard, and some vitamin/mineral fortification.  Some feeds, particularly for metabolic studies, contain more refined carbohydrate (starch, sugar and/or maltodextrin) and protein (usually casein).   This will become important here in a minute.

In PHD-II Paul replaced the notion that animals seek healthy foods with this description of rodent diets:
Rodents are omnivores that, in the wild, eat a lot of seeds. Seeds, like nuts, often contain substantial amounts of fat.
In the laboratory, mice are usually fed a chow that consists mainly of cereal grains, which are very high in starch. So the standard laboratory mouse diets are high-carb diets.
This implies that somehow mouse chows are not appropriate diets at all.  Excuse me, but for someone so into his white rice safe starch, Paul seems to like to play awfully fast and loose with definitions and distinctions.  Newsflash:  Grains ARE seeds.  The seeds of grasses. What kinds of seeds do you think a field mouse might encounter to eat?   Many "pseudo grains" are starchy seeds too.  In fact, most seeds are high starch and relatively low fat.

But here's where it gets laughable!  Instead of the usual chow such as this, we should look at mice that are given a choice between macronutrients and see what they prefer!?  Did they give the mice a choice between some real foods that are very high in one of the macros to the near exclusion of the others?  Oh, like, say, white rice, chicken breast and coconut?   Nope!  

In Smith -- the 13 strain study -- the three choice (3CD) diet compositions are shown on the table.  I'm dubbing this the "Healthy Choice" diet so I can chuckle to keep from crying.  These are roughly 75-87 grams of refined macros with vitamins, minerals and a cellulose binder.    

Now, after implying that your basic rodent chow is unnatural, Paul thinks there's something meaningful to be extrapolated to humans from this?  He's kidding, right??  Remember, according to Paul, these "thrive on low-carb high-fat diets".    What diet might that be?  Well, if he's favoring the strains that preferred a high fat diet, then he has inadvertently endorsed vegetable shortening as healthy.    After all, 9 of the 13 strains tested mice "preferred" was this omega-6 dominated, toxic PUFA-laden vegetable shortening.  

The protein is casein, again far from a likely source for a wild weaned rodent. Apparently the mice don't like it too much either.  It is noted that a few strains required some coaxing to get them to eat the protein at all.  So much for "free choice", and one strain failed to adapt the diet at all.  In the end, only one of the remaining 12 strains consumed over 5% of their diets as protein.   This may be of interest in interpreting test diet intake when compared to intake of a high casein protein chow.  Between the veggie oil and casein, this diet would get the PeterD Premium CIAB seal!!   [CIAB = Crap In A Bag]

Lastly the carbs.  Again, Paul makes ground cereals out to be somewhat foreign to the mouse in his lead up to the discussion in PHD-II.  But let's give them cornstarch and sugar in a fixed 2:1 ratio, one-third sugar and 17% fructose?  ...

One final point on the diet composition in this study.  The diets appear to have been formulated to standardize for calories, therefore vitamin & mineral fortification ends up being twice as high per gram of each "diet".  This may have influenced things, though results in the next study suggest it wasn't a factor.

Here are the diets for the second study, Ortmann:

The carb and protein are the same but the fat is mostly half and half lard and coconut oil making it roughly 25% medium chain triglyceride.  While folks might see this and think "saturated fat", the fat is likely relatively low in long chain saturated fats (feel free to do the math, no time).  Here, the micronutrient and cellulose content is (properly IMO) consistent between the "diets" by weight.

Fat Intake and Body Fatness ~ The Smith 13 Strains Study

As with virtually all diet books, PHD-II necessarily included "lose weight" among the outcomes mentioned in the subtitle that you can expect should you adopt the plan.  PHD is really not a weight loss diet, but some guy named Jay lost a lot of weight on a 1200 calorie, one meal a day diet of PHD-friendly foods and Paul got a publisher, so there you go.  But forget weight loss in a dieting population, what we're really interested in is what to eat on a regular basis that should "naturally" lead to a normal weight (and good health).

Ultimately, the Smith study offers little support for the PHD macros.

  • None of the mice came close to 15% protein (if you look at the plots in the paper, all but one strain stays under 5%, some closer to 2%!) 
  • The percentage of fat ranged from 25% to 83%.   
That 9 of the 12 that could even be maintained on the diet prefer more fat sounds convincing, but how does a range from 25 to 83% fat support the 65% fat in PHD for humans??  It doesn't.

But when we look more closely at the results for these mice, things get even more disconcerting!  They are, if anything, an argument against the high fat argued for in PHD-I or II.

Some results from the paper.  The weights were determined before and after roughly 4 weeks on the diets beginning at 7 weeks old.  Note only 12 of 13 were successful on the diet and included in the main results.  The species are listed in the table in descending order of fat percentage intake.  

In the table one can see trends, but there's limited data on body composition.  Epididymal fat is a major depot in mice, but not the only one, and how good a surrogate it is for total fat is unknown.  As you can see, the weights of the mice are quite varied (the smallest CAST/Ei mouse weighing less than half the largest CD-1).  To standardize the epi-fat mass, it is presented as a ratio of initial body weight in the right hand column.  The plot below is of this ratio vs. the percentage of fat in the diet.

The correlation doesn't quite reach statistical significance, but there is a trend.  Here's an interesting plot I created just "playing around".  I calculated the weight gain as a percentage of initial weight and plotted this versus percent fat calories and ...

Whoa.  That's a pretty tight correlation.  The two or three highest fat consumers are likely to be in a ketogenic state even!   Lastly, Ortmann looked at the epi/body ratio for the individual mice in each strain and their total fat intake.  There was a statistically significant positive relationship for two of the top three fat consumers. 

Looking to those lipophilic mice for diet "advice", the more fat they consumed, the more belly fat (as ratio to initial weight) they accumulated.   If you choose a PHD, and add more egg yolks to your diet, this study does not predict a desirable outcome!!  Instead we see that omnivores that prefer high fat diets, and eat more fat, get fatter.  But:
If mice are smart enough to avoid diabetes and weight gain by eating a high-fat low-carb diet, shouldn’t humans be as wise?
~Paul Jaminet, PHD-I
This study does not support this statement.  The "smart" mice that ate higher fat diets didn't avoid weight gain and fat accumulation.  We don't know about diabetes as it wasn't assessed or reported.  Somehow I don't think getting fatter without getting diabetic is NOT the selling point for PHD that Paul is after.  

But ultimately this study is only referred to in passing to imply that high fat preference is common in omnivores and doesn't make us fat.   I'd say this is a "fail" in interpretation two ways:
  • The diversity of chosen diets counters the idea that mice, let alone all omnivores, favor high fat diets.
  • The mice that did prefer high fat intake did get fatter, and even fatter the more fat they ate.

Macros, Intake, Obesity & Diabetes ~ The Ortmann Transgenic Mouse Study

Ultimately, Smith is included to imply safety in numbers, so to speak, as it does not address health at all.   In so much as it addressed weight, these were multipe uncontrolled studies.  Are there any studies for any of these mice comparing "Healthy Choice" to a standard chow?  I'm asking because I actually don't care so I'm not going to waste time looking.  I'm not in the business of trying to convince people that 65% fat is the perfectly healthy way to go, or that following the lead of a few strains of mice on a ridiculously artificial feeding regiment is a good idea.  But this is the sort of thing Paul should have done were he going to mention these experiments at all.

The crux of Paul's mouse-rationale for PHD, however, lies in the Ortmann study.  Were this just a blog post, or an "oh by the way this seems to favor our diet" type mention, I could perhaps forgive the gross misinterpretation of Ortmann and Smith.  But it is an enduring part of the basis for believing in PHD macro magic.  So this is the study I blogged on in Interpreting Study Results.

In PHD-II Paul asks,
What do mice really want to eat?
Who cares?  What does anyone REALLY want to eat?  I have learned that mice apparently don't like casein much.   Go figure.  When Ortmann and colleagues took another strain of mice and exposed them to a similar "Healthy Choice" diet, they sided with the strains that preferred the most fat.  Big deal.

A  Quick Summary of Ortmann:

  • 4 groups of mice weaned at 3 weeks onto their study diet.  Weight and intake recorded weekly from Week 4 through 17.  
  • Two Diets:  Each strain consumed either a standard diet (ingredients unknown, 28% protein, 14% fat, 58% carb) or a "three choice" diet as outlined above.  This diet ended up 12% protein, 82.5% fat, 5.5% carb for WT and 12.7% protein, 85.1% fat, 2.2% carb for the TG mice.
  • Background strain FVB/N mice = WT or wild type, Transgenic TG mice = brown fat uncoupling protein knockout, designated UCP-DTA.
  • Food intake was relatively steady in the WT on a standard chow diet.  Intake appears dysregulated by both the BAT knockout and both species on "Healthy Choice".  The WT mice did eat and digest almost 40% more calories but only gained slightly more weight - note their initial weight is lower as well.   (The only possibly relevant results to humans are the WT mice, and yes, the high-fat mice ate more, digested more but gained only slightly more weight than the low-fat mice.  IF anything similar had ever been shown in humans in a metabolic ward, it might be worth addressing.  Perhaps NuSI will have all the answers soon.  [/snarkasm] ) 
Paul's main conclusion drawn from the study is that:
[WT mice] “proved highly resistant to the development of obesity and diabetes.” [on the high fat 3CD]
This quotation from the paper was highlighted in both versions of the book.  (Corrected the misworded "remarkably" in I to the accurate "highly" in II.)   In PHD-I he followed it with:
If mice are smart enough to avoid diabetes and weight gain by eating a high-fat low-carb diet, shouldn’t humans be as wise?
Again, leaving aside the "wisdom" of mice offered a bunch of purified nutrients, it turns out that the "highly resistant" quote is rather a bit cherry picked.  While not inaccurate it doesn't tell the whole story.  From Ortmann, in context:
Interestingly, whereas other fat-preferring mouse strains, such as AKR/J and C57BL/6J, are also susceptible to the development of obesity (8), FVB/N proved highly resistant to the development of obesity and diabetes when fed a CD, even though they had an overall 38% increase in digested energy intake when fed the CD compared with SD (Figure 3).
Translation:  Two of the fat preferring strains from the 13 strain paper (Smith, ref 8, also cited by Paul) were susceptible to the development of obesity on the uber high fat version of "Healthy Choice" diet.  

NOTE:  Virtually ALL normal mice are highly resistant to the development of obesity and diabetes on their usual very low fat diet.  Even most (but not all) knockout mice require high fat diets to become obese and perhaps also diabetic.  For example Ortmann note in their discussion that the TG mice they used usually get heavier on a standard chow while their's did not.  They then noted that their chow was 12% fat vs. the 17% fat used in other studies.

Take It Away, Jaminet!

So ....  
  • If three strains of mice don't get fat/diabetic on their standard ~15% fat grain-based chow diet, but
  • they do get overweight/obese (and perhaps diabetic) on a ~40% fat "Western" diet, and 
  • two of the three are susceptible to obesity on an ~80% fat "choice diet" of highly refined components
what should you make of such a situation?  Well, apparently if you're Paul Jaminet and you learn of this study in 2007 on the Hyperlipid blog, you arrive at what he wrote in PHD-I&II.  Writing about his preferred state of "physiological insulin resistance", Peter D offers up the following pithy commentary:
NB These German mice should each be given Professorships of Nutrition at medical schools in the most obese nations of the world. Quite what we should do with the current professors I'm not sure, but I bet the mice could think of something.
Anyway, these mice are cool. The only thing that bugged me when I first read the paper was that they had a higher fasting blood glucose than those poor mice fed the normal junk which passes for laboratory mouse "chow".
Can this drip with #IRONY any more???  This choice diet is the ultimate CIAB!!  What?  You want more irony?  From PHD-II:

Nutrients Are Not Food 
Even though we’ll be talking for a number of chapters about nutrients, one thing should be clear: nutrients are no substitute for food— edible plants and animals. 
Michael Pollan has made this point eloquently: “Food rules,” he says, and it is disastrous to pursue “nutritionism”— the idea that collections of nutrients can replace natural foods. 
Plentiful evidence supports this position. For example, laboratory mice and rats are fed either “chow,” which consists of seeds, grains, beans, and alfalfa— foods similar to what rodents eat in the wild— or a “purified-nutrient diet,” whose ingredients are:
Casein, L-Cystine; Corn Starch, Maltodextrin 10, Sucrose; Cellulose; Soybean Oil, Lard; Mineral Mix; DiCalcium Phosphate, Calcium Carbonate, Potassium Citrate; Vitamin Mix, Choline Bitartrate; FD& C Red Dye #40.
Purified-nutrient diets are protein, starch and sugar, fiber, fat, vitamins, and minerals. Nothing more. They are missing a host of biological compounds found in plants and animals. 
Rodents that eat purified-nutrient diets are usually in worse health than rodents that eat chow. Often, purified-nutrient diets make rodents fat. (When you see a “high-fat diet” reported in the news, it means a purified-nutrient diet that is 30 to 60 percent fat by calories.) 
Most likely, purified-nutrient diets are unhealthy because they are malnourishing [...] food is full of nutrients that we don’t know we need. We share a common biology with plants and animals, and their tissues contain premade biological compounds that are valuable to us and that we may not be able to construct in adequate quantities from purified nutrients. 
It’s curious: when humans want to lose weight, they often start eating meal replacement bars and weight loss shakes composed of purified nutrients— the very kind of diet that makes rodents fat! 
~Paul Jaminet, PHD-II (KL ~1300) 
Sorry about that ... coughing fit.  

PeterD's "cool" mice are the ones eating malnourishing purified nutrients, ending up with messed up insulin signaling, but praise Grok are not (yet?) diabetic.  Furthermore, the transgenic mice -- which should NEVER be brought into this discussion except cautiously if it models a real human disease/mutation (which this does NOT) -- did get fat.  But one minute Paul is trying to convince his readers that seeds are high fat and evil researchers are feeding their mice inappropriate grains in their chow.  The next minute he's rightly highlighting the shortcomings of purified nutrients, but back in the first minute he was using mice eating the most purified of all purified "diets" to support his 65% fat diet?   

Calgon, please take me away!

In Conclusion:

In PHD-I, Paul mentions the transgenic mice without caveat.  He highlights how they became less obese than mice on a "Western" chow and managed not to develop diabetes.  In PHD-II, Paul follows the mention of the transgenic mice with this qualifier:
We probably can’t infer much from genetically mutated mice that avoided obesity by choosing 85 percent fat diets. But it is telling that most strains of wild-type mice chose to get a majority of calories from fat, a minority from carbs and protein— the same pattern we’ve seen repeatedly in other mammals, in humans, and in breast milk of all mammalian species.
Allow me to fix that.  Red are my changes, strike-throughs indicate edits, bolded "can't" is my emphasis.
We probably can’t infer much anything from genetically mutated mice that avoided developed mild delayed-onset obesity by choosing 85 percent fat diets, but since PHD is now defined as 50-65% fat, 80+% diets are irrelevant anyway.  But it is telling that most strains of wild-type mice chose to get a majority of calories from fat, a minority from carbs and protein,  In wild type mouse strains that could be maintained on a self-selected choice of isolated macronutrients, the percentage of fat calories selected ranged from 25% to 83%.  An increase in percentage of fat calories was strongly correlated to percentage gain in body weight across these strains.  In two strains that preferred a very high fat diet, an increase in the total amount of fat consumed was significantly correlated with greater "belly-fat" accumulation.  This is the same pattern we’ve seen repeatedly in other mammals, [and] in humans, and in breast milk of all mammalian species. 

Bottom Line:  NONE of Paul's "Four Reasons" plus "nutrient transformation" support the adoption of the Perfect Health Diet's macronutrient ratios.  In many ways, when all facts are considered, they ALL seem to lead towards the opposite conclusion.

Coming Soon to a Theater Near You:  The Perfect Health Storm!