***secondaryrss.shtml***

 

Response to T. Colin Campbell

March 4, 2007.

by Chris Masterjohn

In the spring of 2005, I wrote a review of Dr. T. Colin Campbell's book, The China Study. It appeared in the quarterly journal of the Weston A. Price Foundation, Wise Traditions, and is also available here. Since my review was highly critical of his conclusion that a vegan diet paves the road to vibrant health, the Weston A. Price Foundation and I offered Dr. Campbell the opportunity to publish his response on the Foundation's web site.

Although he initially expressed interest in responding, Campbell later informed me that he lost interest after learning that I was not a professional researcher. Nevertheless, we engaged each other respectfully in a lengthy email discussion over a period of several weeks. At the time, I asked Campbell for permission to publish our correspondence on my web site, but he declined.

I was therefore rather surprised to recently see Campbell's rebuttal to my review published on VegSource.Com. The rebuttal unfortunately targets my age and credentials and questions my intellectual independence and the motivations of the publisher of my review rather than making a serious, science-based refutation of my arguments. Nevertheless, Campbell does make several substantive points to which I would like to respond. Below, I will address these arguments after briefly clarifying my credentials and motivations.

In This Article:

Posing as a Qualified Scientist?

Although Dr. Campbell numbers me among those who are "mischievously posing as qualified scientists," I have been entirely open about my credentials. Every page on my web site links to my disclaimer page, on which it is stated:

The information contained within this site has been presented with careful research, and some of the articles contain many references to peer-reviewed journals.

Please understand that I am not, however, qualified to dispense medical or health advice of any kind. Likewise, although the articles reference peer-reviewed publications, these articles are not peer-reviewed themselves.

Although I am trained in lab science techniques and journal article writing at the undergraduate level and plan to pursue a PhD in Molecular and Cellular Biology, I do not currently possess a graduate degree or a medical degree, and I have not published peer-reviewed research.

The disclaimer is actually outdated. I now have two items accepted for publication in peer-reviewed journals: the Journal of the American College of Cardiology accepted a letter of mine on August 28, 2006, and Medical Hypotheses more recently published a ten-page hypothesis paper that I have written on the molecular mechanism of vitamin D toxicity.

More importantly, the strength of my argument and that of Campbell's depends not on our authority nor on our credentials, but on the evidence we use and the logic with which we use it.

The Alternative Agenda

Dr. Campbell argues that my writing reflects an "alternative agenda or bias" and even suggests that the Weston A. Price Foundation (WAPF) is influenced by the heavy hand of industrial agriculture and by factory farm conglomerates.

I find it puzzling that anyone could peruse even a small sample of the materials on the WAPF web site, -- all of which consistently and passionately oppose the practice of factory farming -- and then suggest that factory farms would support the Foundation. Instead, most confinement operations that sell grain-fed animal products through corporate retailers would probably keep their money as far away as possible from an organization that promotes free-range pasture farming and direct farmer-to-consumer sales.

That said, it was arguably a cheap shot for me to have noted in my review that Newsweek -- not my own investigation, as Campbell wrongly reports -- has linked the Physician's Committee for Responsible Medicine (PCRM) to People for the Ethical Treatment of Animals (PETA) and the terrorist group Stop Huntingdon Animal Cruelty (judged a terror threat by the Department of Justice, not by me). Although this connection actually has documentation to support it, unlike Campbell's incorrect speculation, any connection between Campbell and Stop Huntingdon Animal Cruelty is indirect and obviously does not bear on the quality of his arguments. I have therefore deleted the reference to this connection from both the version of the review on my web site and from the version on the WAPF web site.

My personal involvement in WAPF began several years ago when a friend introduced me to some of its information on raw milk. At the time, I had recently reintroduced animal products into my diet after two years of vegetarianism led to anxiety, energy problems and a mouthful of tooth decay. Because the additional changes I introduced after familiarizing myself with the WAPF guidelines made dramatic improvements to my health, I involved myself more deeply in the organization.

I was not, as Campbell writes, on the verge of death, nor do I believe that every vegetarian will "eventually learn of the error of their ways." I believe, instead, that at least some high-quality animal products - not necessarily a diet "high in cholesterol, fat and animal protein" in every case - are required for optimal health. Many of us who have particularly high needs for certain animal-based nutrients -- whether because of heredity, circumstance, or some combination thereof - may indeed need a diet very rich in animal products.

Naturally, my experience influences the way I read and interpret research. Because there is a great degree of ambiguity in science, two reasonable people can sometimes look at the same data and form opposite conclusions. Despite having my own experiences, I still struggle to keep an open mind and to consider all possible interpretations of the data I encounter, not just those I would like to believe. I don't consider having this experience to be an "alternative agenda," and I think most reasonable and honest people would admit that their own past experiences exert considerable influence on how they interpret their present experiences. If that's an agenda, we all have one.

China Study Redux — The Case of the Uncorrected Correlation

According to Dr. Campbell, I have demonstrated "a serious lack of understanding not only of the fundamentals of scientific research but also of the principles of statistics, epidemiology and nutrition," by "superficially citing uncorrected crude correlations" that I have selected "to reflect an alternative agenda or bias that has nothing to do with objective science."

Had he presented corrected correlations in his book, in the original monograph, or elsewhere, I would have used them to evaluate his conclusions. Nowhere in these writings or in Campbell's rebuttal, however, does he provide a reference to such correlations; his response is therefore vague and evasive. [See note 1 on corrected correlations.]

Let us briefly return to the evidence provided by the China Study,1 and that which Dr. Campbell selects to form his argument that animal protein contributes to cancer.2

The China Study looked at cancer in two ways: official cancer mortality statistics and a questionnaire that asked each household whether or not there were cancer patients in the family.

According to the "uncorrected" data, households from villages that had higher average animal protein intakes during the three-day, in-house observation were more likely to have cancer patients within their family according to the questionnaire (Figure 1), but villages with higher average animal protein intakes did not have higher incidences of cancer mortality according to the government statistics (Figure 2).

Yearly meat intake was negatively correlated with cancer whether measured by the official statistics or by the questionnaire, but the association was not statistically significant. [See note 2 on statistical significance.] A much better predictor of cancer by both measures was latitude, which reflects vitamin D levels, but that's another story for another day.

Despite Campbell's criticism of my use of "uncorrected, crude correlations," his own argument that animal protein is linked to cancer in the China Study rests squarely on those very "uncorrected" correlations. He disregards the official cancer mortality statistics and the information about meat intake from the questionnaire, presumably because they do not support his argument. Instead, he relies on the association between animal protein intake during the three-day, in-house observation and the percentages of households claiming cancer patients in their families on the questionnaire; he then supports this flimsy figure with supposed surrogate "biomarkers" for animal protein intake.

On page 89, he writes:

Animal protein intake was convincingly associated in the China Study with the prevalence of cancer in families. This association is an impressive and significant observation, considering the unusually low intake of animal protein. Diet and disease factors such as animal protein consumption or breast cancer incidence lead to changes in the concentrations of certain chemicals in our blood. These chemicals are called biomarkers. As an example, blood cholesterol is a biomarker for heart disease. We measured six blood biomarkers that are associated with animal protein intake.39 Do they confirm the finding that animal protein intake is associated with cancer in families? Absolutely. Every single animal protein-related blood biomarker is significantly associated with the amount of cancer in a family.

In order to find out what these biomarkers are, we have to follow reference 39. This takes us to page 376, where he states:

39. These biomarkers include: plasma copper, urea nitrogen, estradiol, prolactin, testosterone, and, inversely, sex hormone binding globulin, each of which has been known to be associated with animal protein intake from previous studies.

The first observation I made when I read this footnote was that Campbell does not cite any references demonstrating that these biomarkers have "been known" to be associated with animal protein intake. But let us put this question aside for a moment.

The data from the China Study do not quite support Campbell's contention that every single one of these biomarkers is significantly associated with cancer prevalence in families. The associations are shown in Figure 3. Plasma testosterone has no significant association with cancer prevalence in either males or females at any age. Plasma prolactin is positively associated in older subjects, but negatively (though not significantly) associated in younger ones. The inverse association with sex hormone binding globulin (SHBG) holds up, but it is only significant in older subjects. By contrast, the association with estradiol only holds up in younger subjects, and is reversed in older ones. The association with urea nitrogen is not significant. Only plasma copper has a convincing, highly significant association with cancer prevalence.

Although Campbell disregards the official cancer mortality statistics, I have nevertheless shown their associations with these biomarkers in Figure 4. None of the hormones are significantly associated with cancer mortality. However, plasma copper again rears its head as a consistent marker for cancer.

Now, are these measurements truly "biomarkers" for animal protein intake? Since Campbell does not cite any studies supporting this contention, I did a few keyword searches on PubMed to see whether there was any data to convincingly support or refute it.

I did not find any evidence substantiating a link between animal protein intake and plasma urea nitrogen.

I did find several studies showing that SHBG is higher in vegetarians and even higher in vegans,3, 4 but it is increased by lean fish protein compared to fatty animal foods,5 and is decreased by dietary cholesterol.6 This hardly supports its use as an inverse biomarker for animal protein.

Contrary to Campbell's contention, one study found that vegetarians have higher testosterone levels than meat-eaters and that vegans have even higher levels, though because more of it was bound by proteins, the free testosterone available for use by the body was the same in all three groups.3 Lean fish compared to fatty animal products does not affect testosterone levels, despite being higher in animal protein.5 One study found a mere two percent difference in plasma estradiol between vegetarians and non-vegetarians, which the authors attributed to the role of body fat in estrogen production,4 and another found no difference in testosterone levels between vegetarian and non-vegetarian girls.7

Most of this is beside the point. Since plasma copper is the only one of these biomarkers convincingly associated with cancer, what we are most interested in is whether plasma copper can be taken as an indicator of animal protein intake.

According to vegan advocate John Robbins, copper status should not be such a good judge of dietary animal protein:

Vegetarian diets tend to be higher in copper, which overrides any reduced rate of absorption from phytates. Vegans, in particular, consume considerably more copper than meat-eaters.

This is supported by one study of Canadian preschool children showing that they obtained forty percent of their copper from grains but only twelve percent of it from animal products.8 Of course, even if most vegan diets are higher in copper than most omnivorous diets, it would not necessarily mean that the blood levels of vegans are higher in copper than the blood levels of meat eaters. So let us take a look at what the research says.

In very young children, there is no relationship between meat intake and serum copper.9 In Kenyan school children consuming only one percent of their diet as milk and only one percent as meat, the daily administration of a meat-rich and milk-rich meal over the course of a school year had no effect on serum copper levels.10 In Austrian school children of ages four through nineteen, protein itself could only explain two percent of plasma copper status, but meat was able to explain fifteen percent. The contribution of cereals, vegetables, cereal-based desserts, coffee, tea and other hot beverages to plasma copper status was about equal to that of meat.11

At best, then, meat intake may be able to explain a mere fifteen percent of the variation in plasma copper levels. But if plasma copper is associated with cancer in the China Study, does this reflect an association between cancer and meat, or an association between cancer and cereals, vegetables, desserts, coffee and tea?

Thus, having no reliable biomarkers for animal protein intake, we are back full-circle to the original, inconsistent data directly ascertaining animal protein and meat intakes, all the while having considered only uncorrected correlations, because these are the very correlations that Campbell uses to make his argument. As seen here, that argument does not stand up to critical analysis.

The Scientific Value of the Work of Weston Price

Dr. Campbell dismisses the work of Weston Price, whom he refers to as a "dental surgeon," as a set of "meager observations" that can be reduced to the conclusion that processed and sweetened food products contribute to tooth decay. Price did not, according to Campbell, support his "inferences about health in general" with any "reliable empirical data."

Price was not simply a dental surgeon, but a prolific researcher who followed the scientific method from beginning to end. He made observations using physical examinations, case histories, X-ray examinations, collections of dietary information, archaeological excavations, statistics, and chemical analyses of thousands of samples of saliva and tens of thousands of samples of food. From these observations, he developed a specific hypothesis - namely, that tooth decay correlates with other forms of physical degeneration with which it shares a common nutritional cause.

From this hypothesis, he formed predictions and tested them. He generated epidemiological and experimental data supporting the hypothesis, raising it to the status of a well-supported theory. Finally, he applied the theory to his clinical practice, achieving a level of success that made his approach internationally known.

According to the Greater Cleveland Dental Society's History Page, Price developed radiography techniques in Paris in 1900 before radium had even been named. He also studied the physical characteristics of wax and gold, all before assuming his position as the Director of the Research Institute of the American Dental Association in 1916, which was known at that time as the National Dental Association.

Price's twenty-five-year research program at the American Dental Association involved a team of sixty researchers and resulted in the publication of a 1,174-page report and twenty-five scientific articles on the potential of root canalled teeth to act as sources of focal infection.12 Including these writings, Price's bibliography boasts 263 publications.

Over the course of his studies, Price observed that the vulnerability of a patient to systemic health problems originating from root canalled teeth seemed to depend as much on the health of the individual as it did on the infection itself. Observation, of course, is the first step in the scientific method. Price realized the value of having control groups, however, and set out to design the equivalent of the modern case-control study.

This is a study wherein a group of people in which all individuals have a certain disease is compared to a group of people in which none of the individuals have the disease. The disease is called the "dependent variable" because it depends on something else - that is, it has a cause. Researchers look for differences between the two groups in diet, lifestyle, environment, or some other factor, and call these "independent variables."

If such a difference exists, the researchers may offer the hypothesis that the variation in the independent variable is causing the variation in the dependent variable. For example, if patients with tooth decay have lower vitamin A intakes than patients without tooth decay, researchers might hypothesize that vitamin A protects against tooth decay. They would then develop logical predictions based on this hypothesis and test those predictions to support or refute it.

Price did not investigate isolated indigenous populations because he regarded them as "the nearest link to our own past." The reason he sought out indigenous populations was because the prevalence of tooth decay in modern society was so high - approaching one hundred percent - that he could not find suitable controls. One cannot make a case-control study with cases alone.

In his epic work, Nutrition and Physical Degeneration,13 he writes:

It became very clear that [the individual's immunity] was quite as important a determining factor as was the type of infection. This finding led me to broaden the scope of my investigations to include a search for control cases that were free from the degenerative processes. I was not able to find these controls in the clinical material afforded by our modern civilization, and therefore extended the search to isolated primitive racial stocks.

Because Price was driven to look for controls in other populations, his study design came to resemble the modern ecological study in most cases more than it did the modern case-control study. Ecological studies look at groups with varying prevalence of a disease rather than groups made entirely of individuals with or without the disease. However, since many of the groups had tooth decay rates lower than five percent and some approached zero percent, while many other groups had tooth decay rates that exceeded ninety percent, this distinction is rather equivocal.

Price consistently noted that the diets of groups in which tooth decay was prevalent shared certain characteristics - namely, the prominence of refined foods as staples. The diets of groups in which tooth decay was absent or nearly absent varied widely but were all characterized by the absence of refined foods. Not all native diets were equal, however: those in which animal foods were virtually absent, for example, produced much lower rates of tooth decay than found on modern, refined diets, but much higher rates of tooth decay than found on diets containing a significant amount of animal foods.

Price did not hypothesize that the foods determined the vulnerability to tooth decay. Instead, he hypothesized that the nutrients within the foods, regardless of their specific source, determined the vulnerability to tooth decay. To test this hypothesis, he performed chemical analyses of the foods that the groups he studied were consuming. He found that the unrefined foods eaten by groups with immunity to tooth decay were often several times higher in nutrients than the same unrefined foods that were consumed in modernized groups who were not immune to tooth decay.

Price further tested the hypothesis using controlled experiments in laboratory animals and using intervention trials with his patients. Since his patients during the Great Depression often suffered from many other complications of physical degeneration beyond tooth decay and were in great need of immediate help, Price did not use control groups during the human trials. He nevertheless observed dramatic changes in the chemical behavior of the saliva and observed the complete reversal of tooth decay - often accompanied by the reversal of other health problems - using his experimental diet of nutrient-rich whole foods.

Other groups instituted bits and pieces of Price's protocol in controlled studies: for example, a New Zealand study found a forty-two percent increase in resistance to dental caries in a group receiving a daily dose of cod liver oil and malt relative to the control group.

While Price's proof would have been more rigorous if he had used control groups in his human intervention trials, his data are nevertheless astounding. He provides X-ray photographs within his book documenting the ability of his experimental diet to completely inactivate active cavities without the need for oral surgery by inducing the remineralization of dentin and the sealing of the carious tooth with a glassy finish.

Price provided a considerable amount of empirical data supporting the second part of his hypothesis - that tooth decay is but one manifestation of a general process of physical degeneration with which it shares a common nutritional cause.

The connection he demonstrated most convincingly was that between tooth decay and tuberculosis. Price consistently found that tuberculosis was ravaging groups who were susceptible to tooth decay while completely passing over groups who were immune to tooth decay, even when these groups lived close by and were cared for by the same doctors. His consistent observation of this phenomenon in many regions of many continents constitutes a large-scale, ecological study that compellingly demonstrates the association between tuberculosis and tooth decay.

Since many individuals with tuberculosis were cared for in infirmaries, he also had the opportunity to examine the facial structure and tooth decay rates of groups entirely composed of tuberculosis patients. Because he was examining groups of individuals with and without tuberculosis, these observations are equivalent to modern case-control studies. Whereas the prevalence of deformities of the dental arch approached or reached zero percent among individuals immune to tuberculosis, it approached or reached one hundred percent among individuals within tuberculosis infirmaries. Figures for tooth decay were similar. Price observed this phenomenon consistently, whether in Alaska, Africa, or New Zealand.

I read case-control studies frequently, and I never find connections anywhere near as compelling as those that Price provides between tuberculosis, facial structure and tooth decay.

Price also believed that the same nutritional causes that underlie dental disease and tuberculosis underlie heart disease and pneumonia as well. He tested more than 20,000 samples of dairy products sent to him at intervals of two to four weeks over the course of several years from the United States, Canada, Australia, Brazil and New Zealand. Price divided the total area into several dozen districts, each composed of several thousand square miles. Since the vitamin content of the butterfat varied widely according to season, rainfall, and the nutrition of the animal, Price was able to develop separate charts of this fluctuation for each district.

He then obtained government mortality statistics in each of these districts for pneumonia and heart disease. In every case, the seasonal fluctuation in pneumonia and heart disease mortality within the district was inversely related to the seasonal fluctuation in the vitamin content of butterfat. Again, this is a remarkable set of empirical data supporting Price's hypothesis.

Although Price was only able to stay a short time with the indigenous groups he examined, he was sometimes able to obtain important statistics from physicians who practiced among these groups. In this way, he obtained limited data suggesting that nutrition influenced cancer and the ease of childbirth in the same way it influenced tooth decay, tuberculosis, pneumonia and heart disease.

For example, Price visited a group of over four thousand natives eating their native diet and about three hundred whites eating a modern diet, residing together on one of the Torres Strait islands north of Australia. The government physician in charge of supervising this group reported that in thirteen years he had suspected only one case of cancer among the natives and operated on none, whereas he had operated on several dozen cases of cancer in the much smaller white population. Granted, he could not rule out a difference in the genetic propensity of these groups to develop cancer.

In Alaska, however, Price obtained data concerning the effect of diet on the ease of childbirth that was not subject to the confounding effect of genetics. The superintendent of the government hospital in Anchorage reported that in his thirty-six years of experience, the native women gave birth with such ease that he was rarely able to reach them in time to witness a birth. In the most recent generation of young women who had grown up during the shift to a modernized diet, however, the women frequently required the assistance of the hospital after spending days in labor.

Price's research clearly has both strengths and weaknesses. He did not have access in the 1930s and 1940s to the type of laboratory methods we have today. He could not measure plasma urea nitrogen or plasma testosterone or most of the other dozens of measurements recorded in the China Study.

On the other hand, the questionnaire of the China Study did not ask questions about the consumption of organ meats, bone and bone marrow, or insects, nor did it differentiate shellfish from scaled fish. It assumed nutrient intakes based on centralized food composition databases and could not account for the potential wide variation in nutrient contents of foods grown in different soils or by different methods. The fact that Price took care to note the many idiosyncrasies of indigenous diets and measured the nutrient contents of foods directly constitutes a major strength of his work.

Clearly, Price's observations are not "meager" and he did, in fact, publish reliable empirical data supporting his views.

Putting It All in Context — Laboratory Evidence and Clinical Experience

Dr. Campbell rightly points out that only a portion of his book concerns the findings of the China Study. He lists two other lines of evidence that influenced his conclusions: his own laboratory evidence and that of others, and the clinical experiences of physicians using vegetarian or nearly vegetarian diets as treatments or prophylaxis in their practices.

In my review, I chose to approach these forms of evidence simply by pointing out the great selectivity with which Campbell presents the possible interpretations of his laboratory evidence and the range of scientific evidence gathered by others. He concludes that "animal-based nutrients" cause cancer because isolated casein causes cancer, where he should simply have concluded that isolated casein causes cancer. He does not discuss George Mann's post-mortem analyses of hearts and aortas taken from deceased Masai -- cattle herders who subsist primarily on milk, blood and meat -- showing that, with a limit of detection of two percent, the Masai were free of heart disease. Nor does he discuss the clinical experience of many medical practitioners who have worked with indigenous groups consuming diets rich in animal products yet enjoying freedom from cancer, as conveyed by such writers as Weston Price and Vilhjalmur Stefansson.

Campbell carries this selectivity so far as to make the absurd claim that folate is found exclusively in green leafy vegetables when the best source is actually chicken liver. Even in his rebuttal, Campbell ignores most of these points and still has not admitted that folate is present in animal foods.

Nevertheless, since Campbell cites the clinical experiences of several physicians - John McDougall, Caldwell Esselstyn, Jr., Terry Shintani, Joel Fuhrman and Alan Goldhamer - as the "most important evidence of all," these experiences are worthy of comment.

Some of these physicians claim a very high standard of clinical success. In his recent book, Eat to Live,14 for example, Dr. Fuhrman claims that he has treated thousands of patients successfully with a diet that limits animal products to ten percent of calories and requires the consumption of at least two pounds of green leafy vegetables per day. Of several hundred heart disease patients, Fuhrman claims that only one has had to undergo repeat surgery and none have died of a heart attack.

The only two of these practitioners who to my knowledge have published experimental results with vegan diets are Dr. McDougall and Dr. Esselstyn.

McDougall published the results of a small, uncontrolled intervention trial in which twenty-four rheumatoid arthritis patients consumed a low-fat, vegan diet for four weeks.15 Although there was no control group, measures of arthritis improved in the subjects over the course of the study. The experimental diet was twenty-two percent lower in calories than the basal diet and the subjects lost an average of seven pounds. McDougall therefore had no way to distinguish between the effect of weight loss and the effect of the exclusion of animal products.

Dr. Esselstyn published the results of a five-year intervention with a low-fat vegetarian diet combined with the individualized use of cholesterol-lowering drugs to bring cholesterol down to 150 mg/dL.16 Since Esselstyn considered it unethical to allow patients under his supervision to eat a standard diet, there was no control group. Twenty-two percent of those who began the intervention dropped out of the study within the first two years; thirty-five percent of those who completed it did not submit to the follow-up analysis of their cardiovascular health; of the twenty-two patients who began the trial, only eleven remained in the final analysis. Of these eleven, occlusion of the blood vessels became better in five, stayed the same in one, and became worse in four.

Despite the inconsistent results, the average change in the width of the blood vessels was an increase in 0.08 millimeters. This represents a reversal of atherosclerosis - on average. Likewise, on average, the degree to which blood vessels were constricted decreased by seven percentage points. Six of the eleven dropped out of the study after the first five years; in the following five years, there were ten heart attacks among the six that dropped out while there were none among the five who remained on the program.

Since there was no control group and there was such a high drop-out rate, it is difficult to make much sense of the study. Did the people drop out because their health was not important to them? Or did they drop out because the vegetarian diet made them feel fatigued, unsatisfied, and even less healthy than their original diet full of meat and junk food? Were the people who completed the study but did not submit to the final measurements of their blood vessels reluctant for no reason, or were they reluctant because they were afraid of the results they would obtain based on how the diet made them feel?

Despite the lack of high-quality evidence, I have little doubt that many people would improve their health on Esselstyn's plan, and especially on Fuhrman's plan, which emphasizes nutrient density to a greater degree than does Esselstyn's. To the extent that the oxidation of lipoproteins such as LDL within the blood would accelerate the accumulation of atherosclerotic plaque, I would expect these plans to be beneficial in two respects: they are so low in added fats and oils that the subjects eat very little polyunsaturated fat, and the small amount of polyunsaturated fat that the patients do obtain from whole foods are accompanied by a rich array of antioxidants that protect them from oxidation.

Since the plans are so low in fat, the body will produce its own fat from carbohydrates. The primary product of this biochemical pathway is palmitate, which is a saturated fatty acid. Because it is saturated, it is not vulnerable to oxidation. Ironically, one of the benefits of eating a diet so low in fat is that a much greater portion of the total fat obtained is saturated.

The question is whether we can eat a diet that protects our blood vessels from the ravages of oxidized lipoproteins while also eating enough fat and protein to maintain robust physical and mental efficiency and ensuring adequate intakes of nutrients like zinc, vitamin B12, vitamin B6, vitamin D, DHA, taurine, and others that are primarily found in animal products.

Evidence indicates that there is indeed a way to accomplish this.

In 2004, researchers from Tufts University, Harvard School of Public Health and several other institutions published a report in the American Journal of Clinical Nutrition that the editors of the journal called "The American Paradox." 17 The study measured the change in atherosclerosis over the course of three years among postmenopausal women who participated in the Estrogen Replacement and Atherosclerosis trial. Like Esselstyn, the authors measured atherosclerosis directly by coronary angiography.

The results are shown in Figure 5. The higher the saturated fat intake, the slower the progression of atherosclerosis. In the highest quartile of saturated fat intake, atherosclerosis was reversed.

The association is most convincing because it is continuous and monotonic. In other words, the benefit increases consistently from the lowest to the highest quartile.

In this study, the group with the highest intake of saturated fat only achieved a 0.01-millimeter increase in mean coronary artery diameter. Esselstyn's results were eight times more impressive.

Nevertheless, when saturated fat intake was analyzed as a continuous variable, each five percent of calories from saturated fat produced a 0.16-millimeter lessening of atherosclerotic progression. The effect changed from a slowing of progression to a reversal at about thirteen percent of calories from saturated fat. If we extrapolate from these figures, a further increase to eighteen percent of calories from saturated fat would have produced a reversal of atherosclerosis of twice the magnitude produced in Esselstyn's study.

Additionally, every five percent increase in the percentage of calories from polyunsaturated fat was associated with a 0.17-millimeter worsening of atherosclerotic progression. Although extrapolation is by its nature somewhat speculative and inherently inconclusive, the same can be said of intervention trials with no control groups. These results strongly suggest that the restriction of polyunsaturated fat may be the active component in the programs of Esselstyn and Fuhrman, and that by eliminating vegetable oils or replacing them with saturated fats like butter and coconut oil we may be able to achieve similar protection from atherosclerosis without the drawbacks of restricting nutrient-dense animal foods.

One of the most potent inhibitors of the accumulation of cholesterol in atherosclerotic plaque may turn out to be carnosine. Carnosine is a dipeptide - a combination of two amino acids, namely histidine and beta-alanine - found exclusively in animal products.

In the blood, chemicals called aldehydes form when glucose, amino acids or polyunsaturated fatty acids oxidize. These aldehydes damage low-density lipoproteins (LDL) by a process called glycation, which makes them more than twice as likely to accumulate in white blood cells and form the "foam cells" that populate atherosclerotic plaques. Carnosine is able to completely reverse the ability of aldehydes to stimulate LDL accumulation in foam cells.18

Unfortunately, high heat destroys carnosine and converts it into a potent carcinogen and neurotoxin called acrylamide. We have been led to believe that acrylamide is mostly found in potatoes subjected to high heat, but new research suggests that this is only because acrylamide reacts with the creatine in meat, which converts it to compounds such as N-methyl-acrylamide that do not show up on acrylamide tests but are nevertheless just as toxic.19

The exclusion of overcooked and charred meat in a vegetarian program would no doubt improve the health of most people on it. On the other hand, the presence of meat that is not overcooked in an omnivorous diet may constitute a potent weapon against atherosclerosis.

The "most important evidence of all" that Campbell cites, then, hardly constitutes a comprehensive argument for a vegan diet. Rather, the balance of the evidence suggests that there are ways to protect ourselves from atherosclerosis - like restricting polyunsaturated fats and eating minimally cooked meat - that allows us at the same time to reap the many nutritional benefits of animal products.

Animal Products — It's Not All or Nothing

The most egregious error in Dr. Campbell's argument is his false equation of diets that are low in animal products with diets that are completely devoid of them. On the dubious grounds that people in rural China who consume only two percent of calories from animal products are purportedly healthier than those who consume any more than this, Campbell argues that we ought not bother eating any animal products at all.

In reality, the difference between a diet that is one hundred percent animal products and one that is two percent animal products is merely one of quantity, while the difference between a diet that is two percent animal products and one that is zero percent animal products is one of quality. A diet low in animal products and a diet devoid of animal products are simply two fundamentally different things.

Not all animal products are equal. Putting aside all differences in quality such as soil composition, pasture feeding and so on, there are certain animal products that are by their nature vastly richer than most others in important animal-based nutrients.

This is particularly true of shellfish. It would take just over a quarter pound of beef per day to fulfill the minimum requirement for zinc, yet a single serving of oysters per week fulfills the same requirement. One would have to eat two servings of salmon per week to meet the minimum requirement for vitamin B12, but would only have to eat clams once per month to meet the same requirement.

When Weston Price traveled to the Pacific island of Viti Levu, he encountered groups in the interior mountains of the island who subsisted largely on plant foods. He was disappointed to find that they considered it necessary to obtain animal foods from the sea at least once every three months:

This was a matter of keen interest, and at the same time disappointment since one of the purposes of the expedition to the South Seas was to find, if possible, plants or fruits which together, without the use of animal products, were capable of providing all of the requirements of the body for growth and for maintenance of good health and a high state of physical efficiency.

The natives of this island had such a deep respect for the health-promoting value of shellfish that even during times of bitter warfare between those dwelling on the coast and those dwelling on the interior, the latter would bring plant foods down in the middle of the night and store them in caches bordering the zone of warfare, and would return the following day to collect shellfish that the coast dwellers had deposited for them.

The China Study's questionnaire had no questions specific to the consumption of shellfish (Figure 6). How, then, could anyone possibly draw a conclusion from it about what the optimal amount of animal products are, if the amount needed is so different when the nutrition is supplied by shellfish than when it is supplied by meat?

I do not believe that everyone has the same nutritional needs. I personally do best when I eat a diet rich in animal products. The body, however, has no respect for government guidelines or any person's theory of nutritional superiority; instead, its appetite for nutrition is determined by its molecular and physiological needs at any given moment, on any given day, according to the many factors of the environment, circumstance, and the many biochemical idiosyncrasies of the individual that it must face.

Yet I do not see any evidence that any group has maintained superb health - freedom from tooth decay and from the specific degenerative diseases associated with nutritional causes, perfectly developed dental arches that accommodate all teeth, broad nostrils that preclude the need to breathe through the mouth, and broad hips in women allowing smooth passage of the young through the birth canal - through complete abstinence from all animal products. Those groups that have eaten fewer animal products have sought out those animal products with the highest nutrient profiles, like shellfish, fish heads, or whole insects and frogs.

The best theory of nutrition is the one that is able to accommodate all observations, even those that appear on the surface to conflict with one another. The challenge before us is to reconcile these seemingly disparate pieces of evidence, not to ignore the ones we do not like. The theory that optimal health requires abstinence from animal products cannot even approach the ability to make this reconciliation. Any good theory of nutrition must accommodate the observation that every civilization maintaining superb health has used at least some quantity of animal products, and that many have used animal products in copious amounts. Only with a good theory of nutrition can we provide the type of health to our future generations that is their birthright.

Notes

Note 1: Corrected Correlations

Correlations are associations between one variable and another. For example, if people who do not smoke or who smoke fewer cigarettes are less likely to get lung cancer than those who smoke heavily, we would say that smoking is correlated with lung cancer.

Since lung cancer risk goes up as more cigarettes are smoked (in our hypothetical example), one variable increases as the other increases, and we say that the two variables are positively correlated. Now let's say that people who eat more green vegetables are less likely to get lung cancer. Since one variable goes up as the other goes down, we would say that lung cancer is negatively or inversely correlated with green vegetable consumption.

Correlation never proves causation. However, when we accumulate other forms of evidence, we can begin to determine which correlations are causative and which are not.

Sometimes we have a better idea about whether one correlation reflects causation than we do about another. Let's suppose that we have a very good understanding of how smoking causes lung cancer but we are not yet sure whether green vegetables prevent lung cancer. In this case, we might correct or adjust the correlation between green vegetables and lung cancer for smoking in case smoking is a confounding variable.

In other words, if people who eat more green vegetables are less likely to smoke, they might be less likely to get lung cancer for that reason alone. So if we correct the correlation for smoking, we are able to see whether green vegetable consumption is still correlated with lung cancer after adjustment.

If it is not, we might conclude that green vegetables probably do not prevent lung cancer. If it is, we might investigate further the possibility that green vegetables protect against lung cancer.

Adjusting correlations is tricky business. Sometimes we do not know what is causing what, and adjusting the correlation might make it less accurate rather than more accurate. For example, if eating pizza, being obese, and dying of heart disease were all correlated with one another, should we adjust the correlation betwen obesity and heart disease for pizza consumption, or should we adjust the correlation between pizza consumption and heart disease for obesity? Adjusting the pizza correlation might make it more accurate if obesity is the primary, direct contributor to heart disease, but might make it less accurate if saturated fat or refined carbohydrates are making a more important direct contribution to heart disease than obesity.

Since it is not always clear whether adjusting a correlation is making it more or less accurate, researchers often present both adjusted and unadjusted figures in their reports.

[Back]

Note 2: Statistical Significance

Statistical significance is a way of quantifying the likelihood that a correlation is due to chance. This likelihood is usually expressed as the p value.

A p value of 0.01 means that if the study were repeated 100 times, we would expect to find a correlation of equal or greater magnitude once, even if the correlation is due entirely to chance. A p value of 0.05 means that if the study were repeated 100 times, we would expect to see a correlation of equal or greater magnitude five times, even if the correlation is due entirely to chance.

Most often, a correlation is called statistically significant when p < 0.05. Statistical significance should not be taken as an absolute indicator of theoretical signifcance, however. If the p value is 0.06, it is hardly more likely to be due to chance than a p value of 0.04, so the distinction is somewhat arbitrary.

Even the significance level of p < 0.05 is not very rigorous. If the China Study generated 100,000 correlations, we would expect it to generate about 5,000 statistically significant correlations that are nothing more than false correlations due to chance. Using the significance level of p < 0.001, however, we would only expect about 100 false correlations to be generated.

On the other hand, some of the correlations that are not statistically significant may reflect real relationships if the statistical power of the study, due to its design or size, is not great enough to detect the correlation.

As a general rule, then, we should place much more confidence in correlations that achieve a significance level of p < 0.001 than in other findings, a substantial amount of confidence in correlations that achieve a significance level of p < 0.01, and should place our confidence rather reluctantly in correlations that achieve a significance level of p < 0.05.

We should not, however, completely disregard findings that are not statistically significant or that border on statistical significance. Instead, we should consider the possiblity that these correlations are real but that we will need to evaluate them further with larger or differently designed, more statistically powerful studies.

We must always keep in mind that even if a correlation between two variables is real -- that is, it is not due to chance -- it does not necessarily mean that one variable causes the other. For example, countries with higher rates of heart disease may have higher rates of television ownership, but we would not necessarily conclude that owning a television causes heart disease, or that having heart disease causes one to buy television sets.

[Back]

Figures

Figure 1: Associations Between Selected Variables and the Percentages of Families with Cancer Patients in the China Study

 

Values are presented from both the three-day in-house observation and the questionnaire about yearly habits. Data are called significant if p < 0.05, highly significant if p < 0.01, and very highly significant if p < 0.001.

 

Variable

Correlation

Statistical Significance

Total Protein (observed)

+44%

Highly Significant

Animal Protein (observed)

+42%

Highly Significant

Fish Protein (observed)

-15%

Not Significant

Plant Protein (observed)

+7%

Not Significant

Fat (observed)

+29%

Significant

Carbohydrate (observed)

+12%

Not Significant

Fat (questionnaire)

-14%

Not Significant

Green Vegetables (questionnaire)

-30%

Significant

Rice (questionnaire)

-31%

Significant

Meat (questionnaire)

-15%

Not Significant

Increasingly Hot Climate

-42%

Very Highly Significant

Increasing Latitude

42%

Very Highly Significant

 

[Back]

Figure 2. Associations Between Selected Variables and Cancer Mortality in the China Study.

Figures are presented from both the three-day in-house observation and the questionairre about yearly habits. Data are called significant if p < 0.05, highly significant if p < 0.01, and very highly significant if p < 0.001.

Variable

Correlation

Statistical Significance

Total Protein (observed)

+12%

Not Significant

Animal Protein (observed)

+3%

Not Significant

Fish Protein (observed)

+7%

Not Significant

Plant Protein (observed)

+12%

Not Significant

Fat (observed)

-17%

Not Significant

Carbohydrate (observed)

+23%

Not Significant

Fat (questionnaire)

-29%

Significant

Green Vegetables (questionnaire)

-28%

Significant

Rice (questionnaire)

-26%

Significant

Meat (questionnaire)

-20%

Not Significant

Increasingly Hot Climate

-36%

Highly Significant

Increasing Latitude

+30%

Significant

 

 

[Back]

Figure 3. Associations Between Plasma Biomarkers and Percentages of Families with Cancer Patients in the China Study.

 

Data are called significant if p < 0.05, highly significant if , p < 0.01, and very highly significant if p < 0.001.

 

Biomarker

Correlation

Statistical Significance

Plasma Copper

+48%

Highly Significant

Plasma Urea Nitrogen

+18%

Not Significant

Plasma Urea Nitrogen Adjusted for Creatinine

+20%

Not Significant

Plasma Estradiol, Ages 35-44

+27%

Significant

Plasma Estradiol, Ages 45-54

+17%

Not Significant

Plasma Estradiol, Ages 55-64

-10%

Not Significant

Plasma Sex Hormone Binding Globulin, All Ages

-17%

Not Significant

Plasma Sex Hormone Binding Globulin, Ages 55-64

-26%

Significant

Plasma Prolactin, Ages 35-44

-16%

Not Significant

Plasma Prolactin, Ages 45-54

+2%

Not Significant

Plasma Prolactin, Ages 55-64

+33%

Highly Significant

Plasma Testosterone in Males

-3%

Not Significant

Plasma Testosterone in Females

+6%

Not Significant

 

[Back]

Figure 4. Associations Between Plasma Biomarkers and Cancer Mortality Rates in the China Study.

 

Data are called significant if p < 0.05, highly significant if p < 0.01, and very highly significant if p < 0.001.

 

Biomarker

Correlation

Statistical Significance

Plasma Copper

+36%

Highly Significant

Plasma Urea Nitrogen

+17%

Not Significant

Plasma Urea Nitrogen Adjusted for Creatinine

+28%

Highly Significant

Plasma Estradiol, Ages 35-54

-7%

Not Significant.

Plasma Estradiol, Ages 55-64

+6%

Not Significant

Plasma Sex Hormone Binding Globulin

-17%

Not Significant

Plasma Prolactin

-1%

Not Significant

Plasma Testosterone in Males

+9%

Not Significant

Plasma Testosterone in Females

+2%

Not Significant

 

[Back]

Figure 5. Change in Mean Coronary Artery Diameter According to Saturated Fat Intake in the Estrogen Replacement and Atherosclerosis Trial

 

A negative change represents a worsening of atherosclerosis, whereas a positive change represents a reversal of atherosclerosis. Data are called significant if p < 0.05, highly significant if p < 0.01 and very highly significant if p < 0.001.

 

Percentage of Calories as Saturated Fat

Change in Mean Coronary Artery Diameter

Statistical Significance of Difference from Bottom Quartile

3.5-7.0%

-0.22 mm

Not Applicable

7.1-8.6%

-0.10 mm

Highly Significant

8.7-10.5%

-0.07 mm

Very Highly Significant

10.6-16.0%

+0.01 mm

Very Highly Significant

 

[Back]

Figure 6.  The Entire Animal Food Section of the China Study Questionnaire

The questionnaire does not differentiate between fish and shellfish, nor does it ask about the consumption of insects, bones, skin, organ meats, or other foods common in pre-modern diets. 

Animal Food Consumption

How often do you usually have fish or sea food in one month?                         ____ times.

How often do you usually have meat in one month?                                          ____ times.

How often do you usually have eggs in one month?                                           ____ times.

How often do you usually have milk in one month?                                           ____ times.

 

[Back]

References

1. Junshi C, Campbell TC, Junyao L, Peto R. Diet, Life-style and Mortality in China: A Study of the Characteristics of 65 Chinese Counties. Oxford, UK; Ithaca, NY; Beijing, PRC: Oxford University Pres; Cornell University Press; People's Medical Publishing House, 1990. [Back]

2. Campbell TC, Campbell TM. The China Study: Startling Implications for Diet, Weight Loss and Long-term Health. Dallas, TX: Benbella Books, 2006. [Back]

3. Allen NE, Appleby PN, Davey GK, Key TJ. Hormones and diet: low insulin-like growth factor-I but normal bioavailable androgens in vegan men. British Journal of Cancer 2000; 83(1): 95-97. [Back]

4. Thomas HV, Davey GK, Key TJ. Oestradiol and sex hormone-binding globulin in premenopausal and post-menopausal meat-eaters, vegetarians and vegans. British Journal of Cancer 1999; 80: 1470-1475. [Back]

5. Lacaille B, Julien P, Deshaies Y, Lavigne C, Brun LD, Jacques H. Responses of Plasma Lipoproteins and Sex Hormones to the Consumption of Lean Fish Incorporated in a Prudent-Type Diet in Normolipidemic Men. Journal of the American College of Nutrition 2000; 19(6): 745-753. [Back]

6. Bennett FC, Ingram DM. Diet and female sex hormone concentrations: an intervention study for the type of fat consumed. Am J Clin Nutr 1990; 52: 808-12. [Back]

7. Persky VW, Chatterton RT, Van Horn LV, Grant MD, Langenberg P, Marvin J. Hormone levels in vegetarian and nonvegetarian teenage girls: potential implications for breast cancer risk. Cancer Res 1992; 52(3): 578-83. [Back]

8. Vanderkooy, PDS, Gibson, RS. Food consumption patterns of Candaian preschool children in relation to zinc and growth status. Am J Clin Nutr 1987; 45: 609-16. [Back]

9. Taylor A, Redworth EW, Morgan JB. Influence of diet on iron, copper, and zinc status in children under 24 months of age. Biol Trace Elem Res. 2004; 97(3): 197-214. [Back]

10. Siekmann JH, Lindsay AH, Bwibo NO, Denment MW, Murphy SP, Neumann, CG. Kenyan School Children Have Multiple Micronutrient Deficiencies, but Increased Plasma Vitamin B-12 Is the Only Detectable Micronutrient Response to Meat or Milk Supplementation. J. Nutr. 2003; 133: 3972S-3980S. [Back]

11. Konig JS, Elmadfa I. Plasma Copper Concentration as Marker of Copper Intake from Food. Ann Nutr Metab 2000; 44:129-134. [Back]

12. Meinig GE. Root Canal Cover-Up: A Founder of the Association of Root Canal Specialists Discovers Evidence That Root Canals Damage Your Health. Ojai, CA: Bion Publishing, 1998. [Back]

13. Price WA. Nutrition and Physical Degeneration: 6th Edition. La Mesa, CA: Price-Pottenger Nutrition Foundation, 2004. [Back]

14. Fuhrman J. Eat to Live: The Revolutionary Formula for Fast and Sustained Weight Loss. New York, NY: Little, Brown and Company, 2003. [Back]

15. McDougall J, Bruce B, Spiller G, Westerdahl J, McDougall M. Effects of a Very Low-Fat, Vegan Diet in Subjects with Rheumatoid Arthritis. J Altern Complement Med. 2002; 8(1): 71-75. [Back]

16. Esselstyn CB Jr., Ellis SG, Medendorp SV, Crowe TD. A strategy to arrest and reverse coronary artery disease: a 5-year longitudinal study of a single physician's practice. J Fam Pract. 1995; 41(6): 560-8. [Back]

17. Mozaffarian D, Rimm EB, Herrington DM. Dietary fats, carbohydrates, and progression of coronary atherosclerosis in postmenopausal women," Am J Clin Nutr. 2004; 80: 1175-84. [Back]

18. Rashid I, van Reyk DM, Davies MJ. Carnosine and its constituents inhibit glycation of low-density lipoproteins that promotes foam cell formation in vitro. FEBS Letters. 2007; 581: 1067-1070. [Back]

19. Yaylayan VA, Locas CP, Wnorowski A, O'Brien J. The Role of Creatine in the Generation of N-Methylacrylamide: A New Toxicant in Cooked Meat. J Agric Food Chem. 2004; 52: 5559-5565. [Back]


This information is not to be construed as advice.
Please consult a qualified health professional.
Click here for more information.