Category Archives: Low-Carb Diets

More Evidence for Low-Carb Diets

In our book we point out a number of dietary tactics that appear to substantially decrease risk of cardiovascular disease. They include:

  • Optimizing tissue omega-6 to omega-3 balance by minimizing intake of omega-6 fats and eating an oily marine fish like salmon or sardines once a week.
  • Optimizing various micronutrients including vitamins D and K2, choline, magnesium, iodine, and selenium.
  • Reducing carbohydrate intake to the body’s natural level of glucose utilization, about 30% of total calories.

We cited two main sources for the claim that reducing carbohydrate intake reduces risk of cardiovascular disease:

–          The Nurses Health Study found that risk of coronary heart disease went down steadily as dietary carbohydrates were reduced and replaced by fat. Those eating a 59% carb diet were 42% more likely to have heart attacks than those eating a 37% carb diet. [1]

–          Replacing dietary carbohydrate with saturated or monounsaturated fat raises HDL and lowers triglycerides, changes that are associated with low rates of cardiovascular disease. Blood lipids are optimized when carb intake drops to 30% of energy or less. [2]

I think this is pretty strong evidence. It is not completely bulletproof, because associations don’t prove causation and improving risk factors doesn’t necessarily improve disease risk; but, combined with supportive evidence from cellular biology and clear evidence that evolutionary selection favors a carbohydrate intake around 30%, I consider it convincing.

However, it’s always good to have more evidence; and two new studies provide some. One directly relates utilization of carbohydrates for energy to atherosclerosis, and the other conducted a 12-month clinical trial of a carbohydrate restricted diet.

Carbohydrate Utilization is Associated With Atherosclerosis

Via Stephan Guyenet comes a study that directly links carbohydrate metabolism to atherosclerosis: “Metabolic fuel utilization and subclinical atherosclerosis in overweight/obese subjects.” [3]

The study used intima-media thickness in the carotid artery, which serves the head and neck, as a measure of atherosclerosis. As Wikipedia notes,

Since the 1990s, both small clinical and several larger scale pharmaceutical trials have used carotid artery IMT as a surrogate endpoint for evaluating the regression and/or progression of atherosclerotic cardiovascular disease. Many studies have documented the relation between the carotid IMT and the presence and severity of atherosclerosis.

To assess metabolism it measured the “respiratory quotient” or RQ. RQ is the ratio of carbon dioxide (CO2) generated in the body to oxygen (O2) consumed in the body.

RQ indicates which fuels are being burned for energy in the body. When carbohydrates are burned, the reaction involves carbon exclusively, so for every O2 molecule consumed there is a CO2 molecule created. This makes the RQ 1.0 when carbohydrates are burned.

Fats, however, donate both carbon and hydrogen, and the hydrogens react with oxygen to make water (H2O). So some of the oxygen consumed when fats are burned goes into water, not carbon dioxide, and the RQ when fats are burned is about 0.7. Ketones also have an RQ around 0.7.

Amino acids from protein have variable amounts of hydrogen and carbon, some amino acids are ketogenic and some are glucogenic, and so the RQ of protein depends on its amino acid mix. Typically RQ from different types of food protein is between 0.8 and 0.9.

However, most people eat a fairly consistent amount of protein, around 15% of energy, so the variable that generally determines RQ in practice is the ratio of carbs to fat in the diet. Higher RQ indicates a higher-carb diet.

Another study had previously shown that calorie restriction, which also reduces RQ by replacing dietary carbohydrate with fat released from adipose tissue, reduces the thickness of the carotid intima-media. [4] This study was the first testing whether the RQ-CIMT relationship holds also in subjects not known to be restricting calories.

The study found that indeed it does: the lower RQ, the less atherosclerosis the subjects had. Unfortunately they don’t present data in a visually useful way (a scatter plot of RQ vs CIMT would have been helpful); here is what they do show:

RQ was better than waist circumference or BMI at predicting degree of atherosclerosis. Only age was a stronger predictor of atherosclerosis than RQ.

RQ predicted atherosclerosis equally well in subjects with and without obesity. This tells us two things:

  1. It supports the idea that it was habitual diet rather than recent calorie restriction (which decreases RQ by replacing food-sourced calories with fat from adipose tissue) that generated low RQ and low CIMT.
  2. As the authors say, it indicates “the main role of metabolic factors rather than BMI” in generating atherosclerosis – metabolic factors meaning burning glucose for energy rather than fat.

It is also supporting evidence for one of the more controversial lines of our book, that “mitochondria prefer fat.”

One caution: Most of the subjects in this study were eating diets that were around 50% to 55% carbohydrate, so the study was testing whether it’s better to eat a little above or below this carb intake. It tells us, I think, that a 45% carb diet is healthier than a diet with more than 50% carbs. It doesn’t tell us what carb intake is optimal.

The Clinical Trial

In a trial lasting 12 months, restricting carbohydrates to 600 to 850 calories per day – that is, about the 30% of energy that we recommend – in the context of a slightly hypocaloric diet improved cardiovascular risk factors. [5]

Overweight and obese subjects in the trial lost 2.8 kg (6 pounds) over the year-long trial, so it couldn’t have been severely calorie restricted. Changes in other risk factors:

–          Blood pressure dropped from 121/79 to 112/72;

–          Fasting blood glucose dropped from prediabetic 106 mg/dl to normal 96 mg/dl;

–          Lipids improved, with triglycerides decreasing from 217 to 155 mg/dl and HDL increasing from 39 to 45 mg/dl.

They conclude:

The results of this study indicate that a moderately restricted calorie and carbohydrate diet has a positive effect on body weight loss and improves the elements of metabolic syndrome in patients with overweight or obesity and prediabetes. These results underscore the need to provide dietary recommendations focusing on calorie and carbohydrate restrictions … Our results are in agreement with reports produced by other authors who also assessed a carbohydrate-reduced diet …


A number of simple dietary and nutritional changes appear to reduce the risk of atherosclerosis and cardiovascular disease generally. One of them is reducing carbohydrate intake.

I believe the optimum carbohydrate intake is around 30% of energy. Many studies generate clear evidence of benefits as carbs are brought down into the range of 20% to 30% of energy, especially in metabolic disorders like metabolic syndrome, diabetes, and obesity. It’s good to see that evidence from other diseases, such as CVD, also supports the same carb intake.

Because most people’s diets are flawed in so many different ways, and fixing an individual factor is often associated with a reduction in CVD risk of 40% to 70%, it’s possible that we could reduce CVD risk by 90% or more by implementing all of the dietary optimizations described in our book.

It’s well worth pursuing all these little optimizations!


[1] Halton TL et al. Low-carbohydrate-diet score and the risk of coronary heart disease in women.  N Engl J Med. 2006 Nov 9;355(19):1991-2002.

[2] Krauss RM. Atherogenic lipoprotein phenotype and diet-gene interactions. J Nutr. 2001 Feb;131(2):340S-3S.

[3] Montalcini T et al. Metabolic fuel utilization and subclinical atherosclerosis in overweight/obese subjects. Endocrine. 2012 Nov 28. [Epub ahead of print]

[4] Iannuzzi A et al. Comparison of two diets of varying glycemic index on carotid subclinical atherosclerosis in obese children. Heart Vessels. 2009 Nov;24(6):419-24.

[5] Velázquez-López L et al. Low calorie and carbohydrate diet: to improve the cardiovascular risk indicators in overweight or obese adults with prediabetes. Endocrine. 2012 Sep 1. [Epub ahead of print]

Very Low-Carb Dieting: Are the Hormonal Changes Risk-free?

I was in Chicago earlier this week to record a video discussion with Dr Ron Rosedale hosted by Dr Mercola. Ron and I have taken opposite sides in several “safe starch debates” (First installment here; reply to Ron here; Ancestral Health Symposium panel discussed here.) This new discussion was intended to be more cordial and uncover common ground as well as differences.

I was intrigued to see that Ron’s lunch consisted mostly of plant foods which he ate avidly; he said he believes that most people on his diet eat a significant amount of plant foods. I came away with the impression that the Rosedale Diet resembles the ketogenic version of PHD, only with less starch and MCT oil.

One of my objections to Ron’s recommendations has been that very low carb and protein consumption can be stressful to the body. Scarcity of carbs and protein invokes certain starvation-associated pathways – for instance, lower T3 thyroid hormone. We discussed this in “Carbohydrates and the Thyroid,” August 24, 2011.

Ron believes that low T3 on low-carb diets is healthy, and other low-carb advocates, such as Sam Knox, have made similar arguments.

I believe that intermittent fasting, which invokes starvation-associated pathways transiently, is usually health-improving – but that you can overdo it. What happens if you invoke these pathways chronically and continuously?

Prof Dr Andro on the “Athlete Triad”

Some light was shed on this question recently by Adel Moussa, aka Prof Dr Andro, who discussed the “athlete triad” in three posts (Part I, Part II, Part III) at his blog Suppversity.

The athlete triad appears most commonly in athletes who undereat and overtrain. Symptoms include low energy, amenorrhea in women and low testosterone in men, osteoporosis, reduced cognitive ability, and impaired immune function. The syndrome is surprisingly common, especially in female athletes:

Although the exact prevalence of the female athlete triad is unknown, studies have reported disordered eating behavior in 15 to 62 percent of female college athletes. Amenorrhea occurs in 3.4 to 66 percent of female athletes, compared with only 2 to 5 percent of women in the general population. [1]

As Adel discusses in Part II, the athlete triad is characterized by the following hormonal pattern:

  • low estrogen and testosterone levels
  • low T4 and low T3 thyroid hormone levels, often with low TSH and high reverse T3
  • a disturbed circadian cortisol rhythm lacking an appropriate cortisol spike in the morning and a normal decline in cortisol levels in the course of the day
  • low leptin, low insulin, and low IGF-1

Precisely the same hormonal patterns, including lower thyroid hormone levels, higher cortisol, and a suppressed circadian cortisol rhythm, are observed in total fasting and starvation. [2] [3]

These hormonal changes conserve glucose and protein, an appropriate step during starvation. The energy-intensive tasks of immune function and reproduction are temporarily suppressed until energy is more readily available.

Similar patterns of reduced T3 and elevated cortisol excretion were recently seen in a clinical trial of a 10% carb weight maintainance diet. [4] This trial shows that even in the absence of calorie restriction, carb restriction is sufficient to reproduce much of the “athlete triad”/starvation hormonal pattern.

This pattern reaches its most extreme form in anorexia:

[H]ypocaloric diets causes changes in thyroid function that resemble sick euthyroid syndrome. Changes consist of a decrease in total T4 and total and free T3 with a corresponding increase in rT3….

States of chronic starvation such as seen in anorexia nervosa are also associated with changes in thyroid hormone, GH, and cortisol secretion. There is a decrease in total and free T4 and T3, and an increase in rT3 similar to findings in sick euthyroid syndrome…. [T]here is an increase in GH secretion with a decrease in IGF-1 levels…. The changes in cortisol secretion in patients with anorexia nervosa resemble depression. They present with increased urinary free cortisol and serum cortisol levels. [5]

In chronic starvation, hunger is replaced by anxiety and a desire to move. In evolutionary context this urge to be active may have stimulated food-seeking, but in modern life it can exacerbate conditions like the athlete’s triad.

In Part II of his series, Adel made an interesting observation. Chris Kresser often mentions a patient who cured his health problems with pizza and beer. Here’s Chris recounting the story to Kurt Harris:

Chris Kresser: Back around 2000, I was interning for a holistic doctor down in San Diego, and this was before I got into Paleo or anything, and I was, I think, a vegan macrobiotic, for crying out loud, at that point!  So, we had a patient who was just really, really sick, and he was just getting sicker and sicker.  He weighed about 90 pounds.  I think he was about 6 feet tall.  And the doctor had him on a restricted diet, you know, one of those food allergy type of diets where all you’re eating is, like, broccoli, venison, and quinoa.

Kurt Harris:  The Specific Carbohydrate Diet?

Chris Kresser:  No, no, just like a really, you know, they do the IgG food testing, which is kinda bunk anyways.

Kurt Harris:  Yeah, that’s pretty bunk.

Chris Kresser:  And then they find out you can only eat strawberries, broccoli, quinoa, and ostrich!  You know?  And so, he was doing that, and he kept removing foods until he was literally down to, like, broccoli and steamed whitefish or something.  That was all he was eating.  And he just kept getting sicker and sicker.  So, he disappears for about six months, comes back a completely different person.  He’s back up to 160 or 170, which was his normal weight, you know, completely normal complexion.  Literally, we didn’t even recognize him, and the doctor was saying:  What happened?  Was it diet?  And the guy was like:  Yep, it was diet.  And he said:  Was it the candida diet?  Was it the Specific Carbohydrate?  What was it?  And he said:  It was the beer and pizza diet!  [laughter]  And this guy literally, I mean, the guy got to this point where he was like:  OK, if this is my life, I’m fine with just flaring out.  You know, this isn’t worth it.  And if I’m gonna go out, I’m gonna have fun.  And so, he started going out.  You know, he wasn’t ever hanging out with his friends anymore because he was on such a restricted diet, he had no social life, so he just said: Forget it.  I’m gonna drink beer and eat pizza at least three times a week, and then the other times I’m gonna do whatever I want.  And that completely restored his health.

Adel speculates (very plausibly in light of the man’s weight of 90 pounds!) that the patient was suffering from the starvation pattern which is replicated in very low-carb “euthyroid sick syndrome” and the athlete triad. What he needed was more calories, especially carb and protein calories. Pizza and beer are great sources!


It was a pleasure to chat with Ron and Dr Mercola in Chicago. We recorded a four hour discussion, which is going to be edited down to an hour or hour and a half.

We found plenty of common ground. We agreed that there are very real health benefits to low-carbohydrate diets. Low-carb diets are helpful against diabetes and metabolic syndrome, and quickly improve cardiovascular risk markers such as blood pressure, triglycerides, and HDL.

But in biology, good things can always be taken too far. One can restrict carbohydrates (and protein) too much. Extremism in carb restriction may, indeed, be a vice.


[1] Hobart J, Smucker D. The female athlete triad. Am Fam Physician. 2000 Jun 1;61(11):3357-64, 3367.

[2] Shimizu H et al. Altered hormonal status in a female deprived of food for 18 days. J Med. 1991;22(3):201-10.

[3] Palmblad J et al. Effects of total energy withdrawal (fasting) on the levels of growth hormone, thyrotropin, cortisol, adrenaline, noradrenaline, T4, T3, and rT3 in healthy males. Acta Med Scand. 1977 Jan;201(1-2):15-22.

[4] Ebbeling CB et al. Effects of dietary composition on energy expenditure during weight-loss maintenance. JAMA. 2012 Jun 27;307(24):2627-34.

[5] Douyon L, Schteingart DE. Effect of obesity and starvation on thyroid hormone, growth hormone, and cortisol secretion. Endocrinol Metab Clin North Am. 2002 Mar;31(1):173-89.

AHS 2012: The Safe Starches Panel

Note: The book has come back to me for copy-editing, that’s why blogging is slow.

The “safe starches” panel turned out to be not about starches, but about carbs. Nobody wanted to contest my assertion that some starchy plants are free of toxins after cooking, so the criticism of starchy foods was solely based on perceived risks from their carb content.

The expectation going in was that Drs. Ron Rosedale and Cate Shanahan would be taking anti-starch/anti-carb positions and that Chris Kresser and myself would take pro-starch/pro-carb positions. But as it turned out, we arrayed ourselves on a spectrum. Ron was resolutely anti-carb, repeating his assertion that we’re all diabetics and intolerant of carbs (see Ron’s summary of the panel here); Cate was more moderate. I supported eating ~30% of calories as carbs, with lower-carb ketogenic dieting as a therapy for certain conditions. Chris took the position that there is little evidence favoring any carb intake over another, and that some cultures have been healthy on carb intakes as high as 85% or more. (See Chris’s summary of his remarks.)

A fair part of the discussion was about longevity and aging, and whether carbs contribute to it. This is a topic that has not been explored much in the Paleo blogosphere, and was the most interesting part of the panel for me.

[Photo from Diana Carr on Facebook.]

My Position: About 30% Carbs is Best

I took an evolutionary perspective. Evolution selected for a carb intake around 30% to 35% of calories. At lower carb intakes, protein is converted to glucose by gluconeogenesis; at higher carb intakes, significant amounts of the excess carbohydrate are converted to fat (not in the liver, but in skeletal muscle and adipose tissue; this is why studies examining lipids exported from the liver show minimal glucose to fat conversion).

If it were equally healthy for the body to have some other glucose supply than the one provided by a carb intake of ~30% of calories, then evolution would not have selected for mechanisms to restore this favored glucose supply by gluconeogenesis or lipogenesis. The body would have accommodated other levels of glucose utilization without trying to alter its glucose supply.

Further, we know that when carb intake is below this natural level, gluconeogenesis does not fully make up the glucose deficit; and when carb intake is above this natural level, lipogenesis does not fully eliminate the glucose surplus. As a result:

  • On high-carb diets, cells/tissues utilize more glucose than in the evolutionarily favored state.
  • On low-carb diets, cells/tissues utilize less glucose than in the evolutionarily favored state.

My thesis is that there are undoubtedly negative effects from over- or under-utilization of glucose by tissues; else evolution wouldn’t be trying to mitigate the over- or under-supply by lipogenesis and gluconeogenesis. And we know at the extremes that negative effects do occur:

  • On very high-carb diets, eg macrobiotic diets, lipid deficiencies appear, reflected in reduced serum cholesterol, impaired immunity, and often mood disorders. We’ve blogged about the effects of lipid-deficient diets in infants.
  • On very low-carb diets, we often see deficient production of mucus and tears due to downregulation of mucin production. We’ve blogged about this.

On less extreme divergences of carb intake from the evolutionary norm, there are no obvious acute effects, but the possibility exists of long-term negative effects.

Ron’s Misunderstanding of My View

Ron Rosedale seems to have misunderstood my argument. In his “A Conclusion to the Safe Starch Debate,” Ron asserts that I am concerned only with blood glucose levels. No, not at all: I am concerned specifically NOT with blood glucose levels but with tissue glucose utilization.

Perhaps a metaphor may help. Imagine an oil well facility connected by pipeline to an oil-burning power-plant. Suppose that it is essential to always maintain a certain pressure of oil in the pipeline, or the pipeline will suffer damage. When the oil wells produce more oil – say, because a new well has become a gusher – the power-plant burns more oil in order to maintain the proper pipeline pressure. When the oil wells produce less oil – say, they’re down for maintenance – the power-plant uses less oil. Always the pipeline has the same amount of oil.

Even though the pipeline always has the same amount of oil and the same oil pressure, that doesn’t mean that it doesn’t matter how much oil is entering the pipeline. The whole complex of wells-pipeline-powerplant may work best and be most robust to trouble if a normal amount of oil is being produced at the well end and a normal amount is being consumed at the powerplant end. Extreme levels of oil production may strain the powerplant’s ability to operate – insufficient oil may shut it down, and excess oil may burn it up or strain its facilities.

In the same way, eating too few or too many carbs will not affect the levels of glucose in the blood – these must be maintained above 60 mg/dl if glucose is to enter the brain which is essential for life even under ketosis – but may strain or stress our tissues which must downregulate or upregulate their utilization of glucose to match the flux of carbs into the body.

Cate Shanahan’s Experience

Cate Shanahan mostly discussed her experiences recommending a low-carb diet to her patients. She says that none of her patients have reported glucose deficiency symptoms, such as dry eyes. For me this mystery was cleared up when she said that she recommends her patients get 70 g of carbs daily. I and others have found that 50 g of starch is a sort of magic level that usually eliminates acute symptoms like dry eyes.  So Cate appears to be recommending a level of carb intake that minimizes the risk of acute symptoms.

It wouldn’t surprise me if most of her patients are eating much more than 70 g. We know the food reward system strongly rewards carb consumption, so that nearly every culture on earth eats at least 45% carbs by calories. Quite possibly most of her patients are eating PHD levels, 100 g to 200 g.

Is there a counter-argument to my evolution based view?

To date, I’ve seen two counter-arguments:

(1)   From the low-carb side, chiefly Ron Rosedale: Evolution didn’t optimize for longevity but for fertility, and we want longevity, therefore we should resist adopting the evolutionarily favored diet.

(2)   From the high-carb side, voiced to some degree by Chris Kresser in the panel: Why isn’t a negative effect of high or low carb intakes apparent in epidemiological data? There seem to be healthy long-lived societies with high carb intakes.

These seem to me to be the interesting issues coming out of the safe starch debate. I will only give brief answers here.

Contra Ron: Very low-carb for longevity

I have two replies to Ron’s argument that very low-carb diets will maximize longevity.

First, evolution DID select for human longevity. The maximum human lifespan is double that of chimps and gorillas. Humans have mitochondrial membranes selected for extreme longevity. Both of these points are discussed in our new Scribner edition. This should give us confidence that the 30% carb intake selected for by evolution may well be the carb intake that maximizes longevity. I believe it is.

Second, Ron’s arguments are based on Cynthia Kenyon’s experiments in the nematode worm C elegans, in which she found that mutations to Daf-2, which is an insulin-like receptor in C elegans, extended maximum lifespan. Cynthia Kenyon famously switched to a low-carb diet after these experiments. However:

  • Humans have multiple genes which are analogs to Daf-2, including insulin-like growth factor 1 as well as insulin, and IGF-1 appears to be more strongly related to longevity than insulin in humans. More significantly, insulin actually antagonizes Daf-2 in worms, inhibiting its signaling; calling into question whether Daf-2 biology can tell us anything about the effects of insulin signaling in humans.
  • More importantly, the mutations to Daf-2 extend maximum lifespan in the laboratory, but they shorten expected lifespan under natural environmental conditions. When Daf-2 mutant worms are placed in the soil, they live shorter lives.

I made the second point during the panel. Here is the reference:

C. elegans mutants that live twice as long as wild-type worms in laboratory conditions typically die sooner than wild-type worms in a natural soil. These results indicate that conclusions regarding extended longevity drawn from standard laboratory assays may not extend to animals in their native environment. [1]

Even if Daf-2 mutant worms were an adequate model of human longevity, we would have to confront the issue of the robustness of health and longevity to stressors such as infections.

The death of Roy Walford, calorie restriction practioner and longevity guru, at age 79 from ALS – a disease that is promoted by fasting and calorie restriction – should warn us about the risks of extreme diets. There are pathogens that can exploit every human environment. “Feed a cold, starve a fever”; for different germs you may need different diets.

Evolution selected for a moderate carb intake because it makes us robust against a wide range of threats. Carb restriction, like calorie restriction, might protect us from some threats but expose us to others. In our natural, free-living situation, we’ll come up against every threat sooner or later. Just because an extreme diet may have the potential to extend our lifespan does not mean that it will. It may reduce our expected lifespan by making us vulnerable to new threats. The safest course, I think, is to follow evolution’s guiding hand.

What Does Experience on High-Carb Diets Tell Us?

Chris Kresser cited a number of healthy high-carb cultures, and noted that Okinawans became the world’s longest-lived culture after being forced to an 85% carb diet during World War II and its aftermath. Stephan Guyenet cites this as a telling point of the debate:

One of the most surreal moments happened right after Kresser brought up the Okinawans, the longest-lived culture and one of the healthiest in the world, and cited a paper showing that their traditional diet was ~85 percent carbohydrate, mostly from sweet potatoes.  Shanahan and Rosedale decided, based on thin air, that the Okinawans actually didn’t eat much carbohydrate, and Shanahan even went so far as to say “I don’t believe you”, even though Kresser was staring right at the citation on his laptop!  This is the kind of head-in-the-sand approach to science that we need to move beyond in the ancestral community.

The source for an Okinawan diet of 85% carbs was published in 1949, and the data was gathered during the post-World War II US occupation. This was a time of desperate poverty, and indeed followed years of poverty and famine during the Great Depression and World War II. Let’s not forget that the Chinese civil war was taking place during this period, and the Korean War was about to begin. Shou-Ching’s parents lived through this period, in China and Korea, and it was a time of starvation for them and for many others in east Asia.

All deeply impoverished people around the world eat high-carb diets, because carb-rich plants are the most readily available “fallback foods” in the natural environment and the cheapest calories available on the market. As soon as animal foods become available, cultures around the world migrate to 50% carb diets.

We have testimony from Okinawans who lived at that time telling us how difficult it was to obtain food. Their diet was severely calorie-restricted. I recall one Okinawan centenarian on television stating that they ate many kilograms of vegetables each day, simply because there was nothing else. (It rather resembled the Terry Wahls diet – very micronutrient rich.)

Then, decades later, the Okinawans become noted for their longevity. What produced the longevity – the carbs or the calorie restriction? Most likely the calorie restriction and high levels of nutrition were more important than the carb-to-fat ratio.

Food quality is also a factor. As readers of our book know, we think traditional Pacific islander diets are the healthiest in the world – composed of safe starches, coconut, and fish, very low in sugar and omega-6 fats. Whatever the carb fraction, such diets are healthier than the American diet. If there is a place in the world to survive a starvation diet of foraged and locally grown foods, it is a Pacific island.

My view: It was silly of the anti-carb panelists to refuse to credit the Okinawan data, but it is also misleading to say that the “traditional diet” of Okinawans is 85% carb based on data from a period of starvation and food scarcity.

As I noted in the panel, when people are able to eat as many calories as they wish, carb intake is generally anti-correlated with longevity at a population level: higher carb intake is associated with shorter expected lifespan. This is mainly due to the correlation of higher carb intake with poverty, but it occurs even within smaller samples of countries at similar income levels. For instance, among the European countries, higher carb intake is associated with shorter lifespan.

As I stated in the panel, this is extremely weak evidence for an effect of carbs; there are many confounding factors and population-level data cannot sort these out (the “ecological fallacy”).

Personally, I give more credence to data on centenarian diets. Few supercentenarians eat high-carb diets, so carbs may indeed reduce maximum lifespan, though eating a high-carb diet certainly doesn’t prevent people from becoming centenarians. (Supercentenarians live to 110.)

How carbs affect appetite may be more important than specific biological effects of glucose (or insulin). As the Simpson & Raubenheimer “protein leverage hypothesis” data show, in rodents higher carb diets are associated with higher calorie intake. Perhaps something similar occurs in humans. We know that energy excess and calorie restriction are major factors in longevity.

In short: I personally think that the relationship of carbs to longevity is U-shaped, with longest expected lifespans at a 30% carb intake; and I think available data is consistent with this. But I think the influence of carbs on longevity is small compared to other factors.


I was concerned going in that the panel would merely re-hash old arguments. I think the anti-carb arguments were, for the most part, familiar and weak; but I think the discussion took us to interesting places. I think the issue of the healthfulness of very high-carb diets, and the data from cultures like Okinawans, is a very interesting topic; and I think the issues of aging and longevity are quite interesting. I enjoyed the panel.

There are some health conditions which benefit from low-carb eating. I am grateful to our moderator, Jimmy Moore, for allowing me to enumerate some of the health conditions that have benefited from the ketogenic version of the Perfect Health Diet.

All in all, I think it was a good discussion but if it is to continue either the very low carb advocates will need to come up with better arguments and better evidence, or the topic will have to shift to exploring the merits of high-carb diets – a topic which the ancestral community hasn’t spent much time discussing.


[1] Van Voorhies WA, Fuchs J, Thomas S. The longevity of Caenorhabditis elegans in soil. Biol Lett. 2005 Jun 22;1(2):247-9.

Perspectives on Low-Carb, I: Dr. Kurt Harris

Last week in An Anti-Cancer Diet (Sep 28, 2011), I recommended that cancer patients eat 400 to 600 carb calories per day, but combine it with a program of daily intermittent fasting plus longer “ketogenic fasts” and periods of ketogenic dieting or low-protein dieting to promote autophagy.

The recommendation to eat some carbohydrates, plus my statement that it was possible for cancer patients to develop a “glucose deficiency” which might promote metastasis and the cancer phenotype, seems to have stirred a bit of a fuss.

In addition to making @zooko sad, it led Jimmy Moore to reach out to a number of gurus to ask their opinion. On Twitter, Jimmy says:

Working on an epic blog post today about @pauljaminet and his “safe starches” concept. Input from numerous #Paleo and #lowcarb peeps.

I’m excited to have this discussion. As Jimmy later tweeted:

Should be fun to hash all this out publicly for ALL of us to understand better about your concepts. Here’s to education.

So far, I have seen responses from Dr. Kurt Harris and Dr. Ron Rosedale. On PaleoHacks, there is an extensive discussion on a thread started by Meredith.

UPDATE: Jimmy’s post is up: Is There Any Such Thing as “Safe Starches” on a Low-Carb Diet?.

I think this discussion is wonderful. With so many people putting effort into this, I have an obligation to respond. I’ll start with Kurt’s perspective today, then Ron Rosedale’s early next week, then whoever else participates in Jimmy’s epic post.

PHD and Archevore: Similar Diets

Kurt and I have essentially identical dietary prescriptions. However, our reasoning sometimes works from different premises. Kurt observes:

My arguments are based more on ethnography and anthropology than some of Paul’s theorizing, but I arrive at pretty much the same place that he does.

An example of a point of agreement is Kurt’s endorsement of glucose-based carbs:

[I] see the human metabolism as a multi-fuel stove, equally capable of burning either glucose or fatty acids at the cellular level depending on the organ, the task and the diet, and equally capable of depending on either animal fats or starches from plants as our dietary fuel source …

We are a highly adaptable species. It is not plausible that carbohydrates as a class of macronutrient are toxic.

I think that if there is no urgency about generating ATP then fatty acid oxidation is slightly preferable to glucose burning. But essentially, I share Kurt’s point of view. Our ancestors must have been well adapted to consuming high-carb diets, and necessity surely thrust such diets upon some of our ancestors. Certainly there’s no reason why consuming starch per se should be toxic.

Kurt and I also agree on which starches are safe:

These starchy plant organs or vegetables are like night and day compared to most cereal grains, particularly wheat. One can eat more than half of calories from these safe starches without the risk of disease from phytates and mineral deficiencies one would have from relying on grains.

White rice is kind of a special case. It lacks the nutrients of root vegetables and starchy fruits like plantain and banana, but is good in reasonable quantities as it is a very benign grain that is easy to digest and gluten free.

We agree that safe starches are a more useful part of the diet than fruits and vegetables:

[E]ating starchy plants is more important for nutrition than eating colorful leafy greens …

I view most non-starchy fruit with indifference. In reasonable quantities it is fine but it won’t save your life either. I like citrus now and then myself, especially grapefruit. But better to rely on starchy vegetables for carbohydrate intake than fruit.

We agree on the optimal amount of carbs to eat:

I personally eat around 30% carbohydrate now and have not gained an ounce from when I ate 10-15% (and I have eaten as high as 40% for over a year also with zero fat gain) If anything I think even wider ranges of carbohydrate intake are healthy.

One can probably eat well over 50% of calories from starchy plant organs as long as the animal foods you eat are of high quality and micronutrient content.

I think being slightly low-carb, in the sense of eating slightly below the glucose share of energy utilization which I estimate at about 30% of energy, is optimal. However, I think we are metabolically flexible enough that a very broad range of carb intake may be nearly as good. I would consider 10% a minimal but healthy intake of carbs, and 50% a higher-than-optimal, but still healthy, intake so long as the carbs are “safe” and the diet is nourishing.

Differing Origins of Our Ideas

Kurt mentions that his ideas are more derived from ethnography and anthropology than mine.

I give great weight to evolutionary selection as an indicator of the optimal diet, and am friendly to ethnographic and anthropological arguments. If I don’t give tremendous weight to such arguments, it’s because I think some other lines of argument give us finer evidence about the optimal diet.

Here, from a paper by Loren Cordain et al [1], are representations of hunter-gatherer diets:

The top graph shows plant food consumption by calories, the bottom graph animal+fish consumption by calories. The numbers are how many of 229 hunter-gatherer societies ate in that range. Typically, hunter-gatherers got 30% of calories from plant foods and 70% of calories from animal foods.

I think the Cordain et al data supports my argument that obtaining 20% to 30% of calories from carbs is probably optimal. However, it’s hardly decisive. There is considerable variability, mainly in response to food availability in the local environment. Inuits, who had few edible plants available, ate hardly any plant foods; tropical tribes with ready access to starchy plants, fruits, and fatty nuts sometimes obtained a majority of calories from plants.

Hunter-gatherer diets, therefore, are a compromise between the diet that is healthy and the diet that is easy to obtain. A skeptic could argue that hunter-gatherers routinely ate a flawed diet because some type of food was routinely easier to obtain than others, and thus systematically biased the diet.

I believe evidence from breast milk is both more precise about what diet is optimal, and much harder for skeptics to refute. Breast milk composition is nearly the same in all humans worldwide, and it has been definitely selected to provide optimal nutrition to infants.

So breast milk, I think, gives us a much clearer indication of the optimal human diet than hunter-gatherer diets. It is an evolutionary indicator of the optimal diet, but it is not ethnographic or anthropological.

There are other evolutionary indicators of the optimal diet — mammalian diets, for instance, and the evolutionary imperative to function well during a famine — which, as readers of our book, we also use to determine the Perfect Health Diet. So, while I think ethnographic and anthropological findings give us important clues to the optimal diet, I think there are plenty of other sources of evidence to which we should give weight. Fortunately, all of these sources of insight seem to be consistent in supporting low-carb animal-food-rich diets — a result which is gratifying and should give us confidence.

Food Reward and Obesity

Kurt seems to have been more persuaded than I am by Stephan Guyenet’s food reward hypothesis (which is, of course, not of Stephan’s creation – it is the dominant perspective in the community of academic obesity researchers). Kurt writes:

Low carb plans have helped people lose fat by reducing food reward from white flour and excess sugar and maybe linoleic acid. This is by accident as it happens that most of the “carbs” in our diet are coming in the form of manufactured and processed items that are simply not real food. Low carb does not work for most people via effects on blood sugar or insulin “locking away” fat. Insulin is necessary to store fat, but is not the main hormone regulating fat storage. That would be leptin.

I agree with Kurt in rejecting what he calls the carbohydrate-insulin hypothesis of obesity, but I am uneasy at the confident assertion that “reducing food reward” is the mechanism by which excluding flour, sugar, and omega-6 fats helps people lose weight.

Let me say first that there is no doubt that the brain has a food reward system that regulates food intake, and also an energy homeostasis system that regulates activity and thermogenesis, and that these systems are coupled. The brain is the coordinating organ of metabolic activity. And the brain’s food reward and energy homeostasis systems are altered in obesity.

But the direction of causality is unclear. Is “reducing food reward” the best strategy against obesity, or is “maximizing food reward with nourishing food” the best strategy?

Some data may illustrate what I mean. Here’s an investigation of how the food reward system in rats controls appetite to regulate protein and carbohydrate consumption. The data is from multiple studies and was collected by Simpson and Raubenheimer [2].

Rats were given a chow consisting of protein and carbohydrate in varying proportions. The figure below shows how much of the protein-carb chow they ate.

I’ve drawn a kinked blue line to show what a “Perfect Health Diet” analysis would consider optimal. Protein needs consist of a fixed amount of protein, around 70 kJ, to meet structural needs, plus enough protein to make up any dietary glucose deficiency via gluconeogenesis. Glucose is preferable to protein as a fuel. Glucose needs in rats are in the vicinity of 180 kJ. When dietary glucose intake falls short of 180 kJ, rats eat extra protein; they seek to make carb+protein intake equal to 250 kJ so they can meet both their protein and carb needs, with gluconeogenesis translating the dietary protein supply into the body’s glucose utilization as necessary.

As the data shows, the food reward system in rats seems to organize food intake to precisely match this:

  • When the chow is low-carb, the food reward system directs rats to eat until carb+protein intake is precisely 250 kJ – then they stop eating.
  • When the chow is high-carb, the food reward system directs rats to eat until protein intake is precisely 70 kJ – then they stop eating.

I interpret this to show that the food reward system evolved to optimize our health, and in healthy animals does an excellent job of getting us to eat in a way that achieves optimal health.

Note that if the chow is high-carb, rats eat more total calories. Is this because their diet has “high food reward”? No, it is because it is malnourishing. It is protein deficient.

Now, a diet of wheat, sugar, and omega-6 fats is malnourishing. There are any number of nutrients it is deficient in. So the food reward system ought to persuade people to eat more until they have obtained a sufficiency of all important nutrients, and rely on the energy homestasis system to dispose of the excess calories in one way or another. But if the energy homeostasis system fails to achieve this, then obesity may be the result.

If this picture is correct, then what is the solution to obesity? Is it to eat a diet that is bland and low in food reward? I don’t think so; the food reward system evolved to optimize our health. Rather the diet that defeats obesity will be one that is efficiently nourishing and maximally satisfies the food reward system at the minimum possible caloric intake.

A good test of these two strategies is the severely calorie (and nutrient) restricted diet. It would be hard to conceive of a diet lower in food reward than one with no food at all. Yet severe calorie restriction produces temporary weight loss followed by regain – often to even higher weights. This “yo-yo dieting” cycle may be repeated many times. I think this proves that at least some methods of “reducing food reward” – the malnourishing ones – are obesity-inducing.

So I would phrase the goal of an anti-obesity diet as achieving satisfaction of the food reward system, rather than as reducing food reward; and would say that wheat, sugar, and seed oils are obesogenic because they fail to provide genuine food reward, and thus compel the acquisition of additional calories.


Jimmy Moore is friends with the smartest people in the low-carb movement, so this discussion is sure to be interesting. I’m grateful that he’s persuaded people to comment on Shou-Ching’s and my ideas, and I’m eager to hear what Jimmy’s experts have to say.

One thing I’m sure of, the discussion will help us understand the many open issues in low-carb science. It should be a lot of fun!


[1] Cordain L et al. Plant-animal subsistence ratios and macronutrient energy estimations in worldwide hunter-gatherer diets. Am J Clin Nutr 2000 Mar;71(3):682-92.

[2] Simpson SJ, Raubenheimer D. Obesity: the protein leverage hypothesis. Obes Rev. 2005 May;6(2):133-42.