The China Study: The Most Comprehensive Study of Nutrition Ever Conducted and the Startling Implications for Diet, Weight Loss, and Long-Term Health

Overall score

Scientific accuracy

Reference accuracy

Healthfulness

How hard would it be to apply the book's advice? Very difficult

Book published in 2016

Published by BenBella Books

Revised and expanded Edition, Paperback

Review posted June 29, 2019

Primary reviewer: Stephan Guyenet

Peer reviewer: Travis Masterson

Table of contents

1. Introduction
2. Scientific accuracy
3. Reference accuracy
4. Healthfulness
5. Most unusual claim
6. Other
7. Conclusion
8. Updates

Claim 1

Animal foods such as dairy and meat are a major cause of cancer, particularly due to the type of protein they contain

Supporting quote(s) and page number(s)

Page 57: “…nutrients from animal-based foods increased tumor development while nutrients from plant-based foods decreased tumor development.”

Page 94: “Animal-based foods are linked to higher breast cancer rates.”

P 165: “What is clear is that diets naturally high in fiber and low in animal-based foods can prevent colorectal cancer.”

Criterion 1.1. How well is the claim supported by current evidence?

1 out of 4

This claim received a score of 1, indicating that it is not supported by current evidence. We examined this claim from two angles. First, animal experiments, most of which were conducted by Campbell Sr. himself. Second, observational studies in humans, one of which (“the China Study”, which the book is named after) Campbell Sr. contributed to.

The China Study rests much of its claim on a series of studies in rats. Early studies by Campbell Sr.’s research group compared the effects of a low-protein diet containing, by weight, 5 percent of the milk protein casein vs. a higher-protein diet containing 20 percent casein when rats were exposed to the liver carcinogen aflatoxin. It is important to understand that the diets differed in total protein content as well as casein content. Rats eating the 20 percent casein diet experienced many more pre-cancerous lesions in the liver than rats eating the 5 percent casein diet. In other words, a higher-protein diet in which the protein came from casein increased the potency of a cancer-causing substance.

However, the 5 percent casein diet left the rats stunted. The book does not cite a specific reference for this, but in endnote 39 of chapter 3 it states that “about 12% dietary protein is required to maximize growth rate, according to the National Research Council of the National Academy of Sciences”. On page 49, the book displays a graph from a study that examined the more fine-grained relationship between dietary casein and the promotion of pre-cancerous lesions in the liver after aflatoxin exposure. This graph shows that the lesions become far more numerous as soon as the rats received enough protein to avoid stunting. This suggests a simple explanation for the effect of different levels of protein/casein on cancer in rats: dietary protein deficiency restricts cancer cell growth. This seems to make sense since rapidly-dividing cancerous cells probably require a lot of protein to grow.

On page 51, The China Study goes on to suggest that the type of protein is just as important as the amount, with animal protein (casein) being far more cancer-promoting than plant protein (wheat gluten or soy protein). This is the crux of the book’s claim about food and cancer in rats. The book cites one study to support this, conducted by Campbell Sr. and colleagues. In this study, they found that upon exposure to aflatoxin, wheat gluten as the dietary protein source promoted pre-cancerous lesions in the liver far less than casein.

However, as stated in the paper, gluten is a “low-quality” protein, meaning it doesn’t supply adequate amounts of all amino acids required by the body. This is illustrated by the fact that rats fed the gluten diet were dramatically stunted, just like the animals in previous experiments fed a 5 percent casein diet. In contrast, casein is a “high-quality” protein that supplies adequate amounts of all essential amino acids. When the researchers added the amino acid lysine to the gluten diet– making it a high-quality protein– it was just as cancer-promoting as the casein (and it also did not stunt the rats). Although the book claims that the same protective effect was observed for soy protein (a high-quality plant protein), it does not provide a reference to support that statement and we were unable to find supporting evidence in a quick scientific literature search.

Together, the rat studies suggest that there is probably nothing special about animal vs. plant protein. Evidently, cancer grows poorly when the body is deficient in protein, whether because total protein intake is low or because the protein is low-quality. This is probably not particularly relevant to human cancer, because modern omnivores, vegetarians, and vegans eating reasonably well-planned diets all obtain sufficient high-quality protein. Although many plant proteins are low-quality individually, in a varied diet plant proteins such as grains and beans complement one another to provide adequate amounts of all essential amino acids. If the findings of these rat studies apply to humans, they suggest that the only way to reap the anti-cancer benefits is to eat a diet markedly deficient in protein. The China Study does not communicate this to the reader.

Let’s turn to observational studies. We don’t regard them as a strong form of evidence but we’ll review them anyway since we don’t have more direct evidence in humans. Since they are numerous, we focused on meta-analyses— studies that quantitatively pool data from multiple studies on a topic. We used the following three search terms to identify meta-analyses in the PubMed scientific database: 1) cancer AND meta-analysis AND “protein intake”; 2) cancer AND meta-analysis AND meat; 3) cancer AND meta-analysis AND vegetarian. Due to the large number of results, we used the “best match” function and considered the top few results for each. When possible, we focused on prospective observational studies, which is a stronger study design.

Meta-analyses indicate that total protein intake (predominantly animal) is not correlated with colorectal, prostate, or ovarian cancer risk, and this does not differ when protein intake is divided into animal vs. plant protein. It is worth noting that one observational study reported that a higher-protein diet was correlated with poor health outcomes and a greater risk of death in people between 50 and 65 years, but better health outcomes and a lower risk of death in people over 65 years. In mouse experiments, they were able to partially confirm the observational findings in humans, although as with Campbell Sr.’s experiments the “protective” level of protein was in the markedly deficient range (4%). Importantly, the harmful associations in humans were not observed with plant proteins.

Meta-analyses on meat consumption are somewhat mixed, reporting no association for stomach and bladder cancer, but reporting associations between red and/or processed meat consumption and colorectal, breast, and lung cancer. It’s worth noting that these latter three are among the most common types of cancer.  One meta-analysis included in our review suggested that overall cancer mortality is correlated with processed meat intake but not unprocessed red meat intake. Intake of other animal proteins such as poultry, fish, eggs and dairy are generally not associated with cancer risk.

Meta-analyses on vegetarian and vegan diets suggest that both are associated with a modestly lower risk of total cancer relative to omnivorous diets. However, it is unclear whether this is due to avoidance of animal foods per se or other aspects of an overall healthier lifestyle chosen by a more health-conscious group of people (or other confounding factors). A specific example of this disregard for confounding is when The China Study acknowledges that higher intakes of fiber– typical of plant-based diets– are associated with a lower risk of colorectal cancer. Although this, rather than the lower meat intake, could partially explain the fact that vegetarians and vegans have a lower risk of cancer, the book attempts to sidestep this issue by arguing that there is no need to understand the independent effect of fiber if individuals were to just accept his prescribed diet. However, it is important to note that it is possible to have a diet containing animal foods and high in fiber which may lead to lower colorectal cancer risk.

Finally, let’s examine evidence from the China Study itself, the book’s namesake. To evaluate The China Study’s claim that animal protein consumption is linked to cancer, we headed to the University of Washington medical library and examined data from the original (1990) China Study. It reports no significant association between cancer mortality in people under 65 years and: fish protein intake (page 574), non-fish animal protein intake (page 572), dairy intake (page 630), or egg intake (page 632). Data for older ages were not reported.

However, as Campbell has pointed out, these figures are “unadjusted”, meaning they are simple analyses that don’t control for potential confounding factors. To address these concerns, we digitized data from the China Study on the total cancer mortality rate in people younger than 65, plant protein intake, animal protein intake (total protein minus plant protein), smoking rate, latitude, agricultural and industrial output (a marker of wealth), literacy (a marker of education), and age. We gave the data to a professional statistician, Karl Kaiyala, PhD. He analyzed the data in multiple ways (multivariate regression), none of which supported the book’s claim that people who ate more animal protein died of cancer more often. On this point, Kaiyala’s findings reach the same conclusion as those of Denise Minger, who extensively analyzed China Study data beginning in 2010. Academic researchers have come to similar conclusions regarding the China Study data.

How can The China Study make these claims when the data clearly do not support them? The book accomplishes this by making an indirect argument: animal food intake is correlated with higher cholesterol levels, and higher cholesterol levels are correlated with cancer risk (pages 69-71). These statements individually are at least partially supported by China Study data, but the overall argument is not persuasive. If we want to know whether animal protein intake is correlated with cancer risk, we should examine the direct correlation between animal protein intake and cancer risk, not at an indirect correlation via blood cholesterol that can easily be misleading. In any case, it’s worth keeping in mind that the China Study used an observational design that forces us to take the results with a large grain of salt, regardless of what its findings are.

Overall, the evidence does not offer clear support to The China Study’s claim that animal foods, in general, are a major cause of cancer. Because of the design of the China Study we cannot conclude anything from it with confidence. From other observational studies we were able to find some supporting evidence that processed meat, and perhaps unprocessed red meat, may contribute to the risk of certain cancers, but we did not identify evidence that other types of meat, dairy, or eggs do so.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

1 out of 4

The book’s references for this claim received a score of 1, indicating that they do not support the claim. The book’s claims rest primarily on Campbell Sr.’s extensive research in rodents and the China Study, a large observational study Campbell Sr. played a role in.

As described above, The China Study’s characterization of the rodent research on the connection between animal vs. plant protein and cancer is not well supported.

Also as described above, The China Study’s characterization of the China Study itself is not well supported. As far as we can tell, this study did not identify clear connections between animal food intake and cancer risk, and the arguments made to the contrary in The China Study are unconvincing.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

1 out of 4

This claim received a score of 1, indicating that it is substantially overstated. As discussed above, there is some observational evidence linking the consumption of processed and red meat to the risk of certain cancers, but not much evidence that other animal foods contribute to risk.

Overall (average) score for claim 1

1.0 out of 4

Claim 2

Animal foods are a major cause of cardiovascular disease

Supporting quote(s) and page number(s)

Page 106: “In simple terms, animal foods were linked to higher blood cholesterol; plant foods were linked to lower blood cholesterol. This research clearly implicated diet as one possible cause of heart disease.”

Page 109: “This study suggests that the more animal protein you eat, the more heart disease you have.”

Page XXV: “Heart disease can be reversed by diet alone– and in doing so, reducing animal protein is more significant than reducing saturated fat.”

Criterion 1.1. How well is the claim supported by current evidence?

2 out of 4

This claim received a score of 2, indicating that it is weakly supported by current evidence. We examined this claim from three angles. First, are there plausible mechanisms linking animal food intake with cardiovascular disease? Second, what do observational studies have to say, including the China Study itself? Third, what do intervention studies have to say?

We believe the current scientific literature supports at least two plausible mechanisms linking animal food intake with cardiovascular disease risk. Both mechanisms converge on impacting blood cholesterol. Blood cholesterol is a well-established marker for heart attack risk, particularly the cholesterol carried by the “bad” LDL particle. This particle plays a major role in causing the arterial plaques that drive common heart disease as it carries and deposits cholesterol throughout the blood vessels.

The first plausible mechanism is that animal protein has been shown to increase circulating cholesterol by increasing transit time and reducing cholesterol turnover. This leads to enlarged arterial plaques in animal models with high animal protein intake relative to high plant protein intake. It’s worth noting that this doesn’t hold true for all types of animal and plant proteins. In humans, eating plant protein leads to lower total and LDL cholesterol levels compared to eating animal protein. The second likely mechanism is that animal foods are the primary source of saturated fat in most diets, and saturated fat intake has been shown to increase both total and LDL cholesterol in controlled trials lasting up to a few months. Together, these two mechanisms partially explain why avoiding animal foods reduces total and LDL cholesterol.

Let’s turn to observational studies. As with the first claim, we focused on meta-analyses (studies of studies), particularly of prospective observational studies as this is a stronger study design. We used the following three search terms to identify meta-analyses in the PubMed scientific database: 1) (cardiovascular OR “heart disease” OR infarction) AND meta-analysis AND “protein intake”; 2) (cardiovascular OR “heart disease” OR infarction) AND meta-analysis AND meat; 3) (cardiovascular OR “heart disease” OR infarction) AND meta-analysis AND vegetarian. Due to the large number of results, we used the “best match” function and considered the top few results for each.

The first search on total protein intake did not return any relevant results. The second search suggested that total meat intake and white meat intake are generally not associated with heart attack risk. However, processed meat intake generally is associated with heart attack risk and unprocessed red meat intake is sometimes associated with heart attack risk.  Additionally, stroke risk, another cardiovascular outcome, is associated with total, unprocessed red, and processed meat intake, but not white meat intake.

Our third search produced evidence that suggested fairly consistently that people eating vegetarian diets have a lower risk of heart attacks. However, the effects are larger and more consistent among vegetarian Seventh-Day Adventists than among vegetarian non-Seventh-Day Adventists, suggesting that the vegetarian diet itself may not account for the entire difference in risk. In general, it is unclear to what degree these protective associations are due to avoiding meat vs. other diet and lifestyle habits of health-conscious people.

As with the cancer evidence, on heart disease The China Study places great emphasis on evidence from its namesake China Study. To evaluate The China Study’s claim that animal protein consumption is linked to cardiovascular disease, we headed to the University of Washington medical library and examined data from the original (1990) China Study. It reports no significant association between cardiovascular mortality (myocardial infarction and coronary heart disease) in people under 65 years and: fish protein intake (page 574), non-fish animal protein intake (page 572), dairy intake (page 630), or egg intake (page 632). Data for older ages were not reported.

However, as Campbell has pointed out, these figures are “unadjusted”, meaning they are simple analyses that don’t control for potential confounding factors. To address these concerns, we digitized data from the China Study on cardiovascular mortality (MI and CHD) rate in people younger than 65, plant protein intake, animal protein intake (total protein minus plant protein), smoking rate, latitude, agricultural and industrial output (a marker of wealth), literacy (a marker of education), age, and Apolipoprotein B (a predictor of cardiovascular risk in the blood). We gave the data to a professional statistician, Karl Kaiyala, PhD. He analyzed the data in multiple ways (multivariate regression), none of which supported the book’s claim that people who ate more animal protein died of heart disease more often. On this point, Kaiyala’s findings reach the same conclusion as those of Denise Minger, who extensively analyzed China Study data beginning in 2010. Academic researchers have come to similar conclusions regarding the China Study data. Kaiyala did find in his multivariate models that higher cardiovascular mortality was independently associated with higher Apolipoprotein B, latitude, and plant protein intake. The first two of these were expected, while the third was not.

As in the cancer claim, The China Study argues for a connection between animal food intake and cardiovascular disease using an indirect correlation via blood cholesterol. This argument is no more persuasive for cardiovascular disease than it is for cancer.

Although the China Study itself does not appear to support the claims in The China Study, we examined another study with broadly similar methods: the Seven Countries Study. This massive observational study measured diet and health outcomes in populations of seven countries around the world. Unlike the China Study, its findings seem broadly consistent with the thesis of The China Study: researchers found a correlation between animal food intake and the risk of having a heart attack over a 25-year period. However, “ecological studies” like the China Study and the Seven Countries study are not a particularly strong form of evidence in our opinion. In the scientific community, results from ecological studies tend to be viewed as suggestive rather than conclusive, regardless of how large they are.

Another piece of evidence worth considering is the rate of cardiovascular disease in India, a country that gets almost all of its protein from plant sources due in part to its high prevalence of vegetarianism. According to The China Study, a low intake of animal protein and a high intake of plant protein should protect against cardiovascular disease. However, India has a cardiovascular death rate that is higher than both the global average and the United States. It’s worth noting that this is likely explained, at least in part, by differences in the quality of health care. Nevertheless, this shows that India is not substantially protected against cardiovascular disease.

Let’s now turn to intervention studies. We were unable to find any studies with hard cardiovascular outcomes (e.g. heart attacks) that isolated animal foods as a variable (i.e., that didn’t change other things at the same time). There are randomized controlled trials by Dean Ornish, MD, and others suggesting that a low-fat, vegetarian-focused diet and lifestyle program can partially reverse build up in the heart’s arteries in people with heart disease. At least one of these studies suggests that the program can reduce heart attack risk, but these studies didn’t isolate the effects of animal foods and their contribution to the overall effect remains unclear.

Overall, it seems plausible that diets low in animal foods lead to lower cardiovascular risk than diets high in animal foods. However, the strength of the evidence supporting this position is limited. If we believe the observational studies, the risk is likely to be concentrated in processed and red meat, but not so much in poultry, seafood, and other animal foods like milk and eggs.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

2 out of 4

The book’s references for this claim received a score of 2, indicating that they are weakly convincing. The book’s claims rest heavily on the China Study, a large observational study Campbell Sr. played a role in. As far as we can tell, this study did not identify clear connections between animal food intake and cardiovascular risk (see above). That said, the book accurately cites references suggesting that animal protein tends to increase unhealthy blood lipids in animal models and humans, and increases arterial plaque in animal models, and this lends some degree of support to its claim.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

1 out of 4

The claim received a score of 1, indicating that its strength is substantially overstated. Although there is some evidence supporting the book’s claim, it is not strong or consistent.

Overall (average) score for claim 2

1.7 out of 4

Claim 3

A whole food, plant-based diet prevents and reverses cardiovascular disease

Supporting quote(s) and page number(s)

Page XXV: “Heart disease can be reversed by diet alone– and in doing so, reducing animal protein is more significant than reducing saturated fat.”

Page 124: We now know what is true: a [whole food plant-based diet] can prevent and treat heart disease, saving hundreds of thousands of Americans each year.”

Criterion 1.1. How well is the claim supported by current evidence?

3 out of 4

This claim received a score of 3, indicating that it is moderately well supported by current evidence. Unfortunately, a randomized controlled trial investigating the effects of a whole food plant-based diet alone on the risk of having a heart attack and/or stroke does not exist as it does for other diets such as the Mediterranean diet. However, we did identify randomized controlled trials that included a whole food plant-based diet as part of a broader diet and lifestyle intervention, and other trials that were not randomized or controlled but used a whole food plant-based diet in conjunction with cholesterol-lowering drugs.

The China Study itself reviews much of the relevant research. For example, Dean Ornish, MD, and colleagues have conducted trials suggesting that a low-fat, vegetarian-focused diet and lifestyle program can partially reduce plaque build up in the heart’s arteries in people with heart disease and perhaps even reduce heart attack risk. While these studies generally used a rigorous design and reported encouraging findings, they also included other lifestyle modifications such as regular exercise and smoking cessation making it difficult to isolate the specific effects of the prescribed diet.

More examples can be found in studies from Caldwell Esselstyn Jr., MD, and colleagues. They published a case series describing a group of patients with severe coronary artery disease who adopted a low-fat whole food plant-based diet, and in some cases were prescribed cholesterol-lowering drugs, and had a notably low rate of heart attacks. The goal of the prescribed diet and medication in these case studies was to reduce total blood cholesterol below 150 mg/dL (for reference, the average blood cholesterol in the US is about 189 mg/dL). Initially, Esselstyn’s team published a small study in 1995 that demonstrated partial reversal of coronary artery blockages in some people. Subsequently, they published a larger study in 2014 suggesting that patients who adhered to this diet advice had a much lower rate of heart attacks and strokes than patients who did not adhere. However, this study triggered a lively academic back-and-forth in which researchers and physicians raised concerns about the study’s methods and questioned its conclusions.

These types of studies (case series) are one of the least compelling forms of scientific evidence, however the effect size is large enough that the findings are difficult to dismiss entirely.

Aside from these studies, there is a wealth of research suggesting that substances found in whole plant foods tend to reduce cardiovascular risk. A low-fat, whole food plant-based diet will also tend to cause weight loss, which will improve cardiovascular health. In addition, randomized controlled trials have found that replacing processed food intake with whole plant foods reduces plaque build up in the heart’s arteries and substantially reduces the risk of heart attacks even when people continue to eat animal foods. However, as mentioned earlier, replacing animal protein with plant protein tends to improve blood cholesterol levels. Together, this supports the claim that a whole food plant-based diet probably does substantially reduce the risk of cardiovascular disease.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

3 out of 4

The book’s references for this claim received a score of 3, indicating that they are moderately convincing. They are reviewed in the section above. We believe some of these studies, such as the case series published by Esselstyn and colleagues, should have come with important caveats about the limitations of that study design.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

3 out of 4

This claim received a score of 3, indicating that it is slightly overstated. Although the overall evidence suggests that the claim is probably correct, in our opinion the evidence is not strong enough to justify the supreme confidence of the book. To achieve this level of confidence, we would need at least one well-conducted randomized controlled trial that isolates the unique effect of the diet and directly measures the likelihood of having a heart attack.

Overall (average) score for claim 3

3.0 out of 4

Overall (average) score for scientific accuracy

1.9 out of 4

Reference 1

Reference

Chapter 6, reference 28. Levin et al. J Clin Gastroenterol 8:458. 1986.

Associated quote(s) and page number(s)

Page 131: “These studies document the fact that vegetarians consume the same amount or even significantly more calories than their meat-eating counterparts, and yet are still slimmer.”

Criterion 2.1. Does the reference support the claim?

1 out of 4

This reference received a score of 1, indicating that it does not convincingly support the book’s claim. Although the study does report that vegetarians in Israel were leaner and consumed more calories than omnivores, the data were collected via questionnaire (self-report), which is a notoriously unreliable way of measuring calorie intake. Furthermore, the study did not measure the physical activity level of the two groups, which could have differed. The author’s claim that vegetarians can eat more calories and still remain leaner than omnivores seems unlikely in light of many tightly controlled trials suggesting that calorie intake is the primary dietary determinant of body fatness. We are not aware of high-quality evidence supporting the book’s claim.

Reference 2

Reference

Chapter 9, reference 44. Borchers et al. Autoimmun Rev 9:A355. 2010.

Associated quote(s) and page number(s)

Page 185: “The incidence of Type 1 diabetes is rising rapidly in many parts of the world at an annual rate of 3-5%.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. The paper cites a range of estimates “between 1.3%/year (in Norway) and 9.3%/year (in Poland) for the period 1989–2003.” Although the paper does not specifically include a 5% estimate, it is within the range reported by the paper. However, it is worth noting that the average estimates from a large international World Health Organization study called DiaMond were “2.4% for the period 1990–1994 and of 3.4% for the period 1995–1999.” These are probably the most accurate estimates.

Reference 3

Reference

Chapter 10, reference 74. Jama et al. Am J Epidemiol 144:275. 1996.

Associated quote(s) and page number(s)

Page 209: “Other studies have also found that a low level of vitamin C in the blood is linked to poorer cognitive performance in old age, and some have found that B vitamins, including beta-carotene, are linked to better cognitive function.” The reference follows the phrase “including beta-carotene.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. However, we do note that beta-carotene is not a B vitamin, as suggested by the book’s quote; instead it is a precursor of vitamin A. Since the book’s primary author has a PhD in nutrition, we are going to give the benefit of the doubt and assume this is simply a typo. The study reports that “a lower intake of beta-carotene was associated with impaired cognitive function”, which supports the book’s claim.

Reference 4

Reference

Chapter 18, reference 4. Center for Media and Democracy. “SourceWatch: Physicians Committee for Responsible Medicine: Court Rules Against USDA Secrecy & Conflicts of Interest.”

Associated quote(s) and page number(s)

Page 349: Describes a lawsuit by the Physicians Committee for Responsible Medicine against the USDA Dietary Guidelines Committee exposing undisclosed conflicts of interest with the dairy industry. The book suggests that the chair of the committee, Cutberto Garza, MD, “had failed to declare personal compensation in excess of the amount that required declaration.”

Criterion 2.1. Does the reference support the claim?

3 out of 4

This reference received a score of 3, indicating that it offers moderate support for the claim. It is a link to a web page that describes the lawsuit in question. Some of the links to references on the web page are broken (originally they linked to web pages of the Physicians Committee for Responsible Medicine, a vegan advocacy group) but we were able to access them using the Wayback Machine internet archive. The web page cited in The China Study and its citations support the claim that 6 of 11 committee members had conflicts of interest with the food industry. This included Dr. Garza who received financial compensation from Nestle in excess of $10,000 USD. However, neither the cited webpage nor its links confirmed the claim that Garza “failed to declare personal compensation in excess of the amount that required declaration.” In a quick Google search, we were unable to confirm this allegation.

Reference 5

Reference

Chapter 15, reference 2. National Dairy Council, “Our Story”.

Associated quote(s) and page number(s)

Page 289: “Founded in 1915, the well-organized, well-funded National Dairy Council has been promoting milk for over a hundred years.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. The reference is a web page on the National Dairy Council website confirming the central claim that the organization was founded in 1915.

Reference 6

Reference

Chapter 7, reference 8. West et al. Diabetes 15:9. 1966.

Associated quote(s) and page number(s)

Page 139: This paragraph describes an ecological study suggesting that countries with diets high in calories, fat, animal fat, and animal protein have higher rates of diabetes than countries with diets lower in these qualities and higher in carbohydrate.

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. The book cites this study accurately. Although it is an ecological study of diabetes prevalence across only four countries, which is a relatively weak form of evidence, The China Study freely acknowledges this limitation on the same page: “These old, cross-cultural studies can be crude, resulting in conclusions that are not entirely reliable.”

Reference 7

Reference

Chapter 12, reference 1.  Davey et al. Publ Health Nutr 6:259. 2003. Sobiecki et al. Nutr Res 36:464. 2016.

Associated quote(s) and page number(s)

Page 241. These references are associated with a table with data on nutrient profiles of meat eaters, pescetarians, vegetarians, vegans, and the diet recommended by The China Study, a whole food plant-based diet. The text prior to the reference states “These data are for England.”

Criterion 2.1. Does the reference support the claim?

3 out of 4

This reference received a score of 3, indicating that it offers moderate support for the claim. The China Study cites two papers as the sources of the data in its table. The data on protein, fat, and carbohydrate come from table 2 of Davey et al. 2003. However, neither of the cited papers contains data on dairy or vegetable intake, which represents half the data in the table. It’s unclear where those data come from.

Reference 8

Reference

Chapter 19, reference 7. American Heart Association. “High blood cholesterol and other lipids– statistics.” March 2004.

Associated quote(s) and page number(s)

Page 361: “About 105 million American adults have dangerously high cholesterol levels (defined as 200 mg/dL or higher– heart-safe cholesterol level is under 150 mg/dL.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. We were able to access the web page cited in the book using Wayback Machine, but the page itself did not contain the relevant information. When we tried to click through a link labeled “High Blood Cholesterol and Other Lipids” on the page, we were not able to. However, we found the following statement on a US Centers for Disease Control web page: “More than 102 million American Adults (20 years or older) have total cholesterol levels at or above 200 mg/dL, which is above healthy levels.” This is close enough that it seems likely that the statement in The China Study was accurate, and accurately referenced, at the time it was written, although the reference probably should have been updated with the most recent release of the book.

Reference 9

Reference

Chapter 3, reference 37. Dunaif and Campbell. J Nutr 117:1298. 1987.

Associated quote(s) and page number(s)

Page 48: “But how much protein is too much or too little? Using rats, we investigated a range of 4-24% dietary protein (Chart 3.6). Foci did not develop with up to about 10% dietary protein. Beyond 10%, foci development increased dramatically with increases in dietary protein.”

Criterion 2.1. Does the reference support the claim?

3 out of 4

This reference received a score of 3, indicating that it offers moderate support for the claim. The reference is to a study conducted by Campbell Sr. and a colleague in which they exposed rats to a carcinogen called Aflatoxin B and then measured the development of cancerous lesions in the liver in rats fed different amounts of protein (from the milk protein casein). Although the paper contains a graph (figure 1) that superficially resembles chart 3.6 in The China Study, closer inspection reveals that they are not the same graph and the data in chart 3.6 do not appear in the cited study. Furthermore, chart 3.6 in the book uses the vague term “foci development” for the vertical axis without providing units, while the cited paper uses specific terms like “percent of liver volume occupied by foci”, “foci number”, and “foci diameter” with units. Figure 3.6 is either based on data that do not appear in the cited paper, or it’s an impressionistic version of the data in the cited paper. However, since it conveys the same meaning as the paper, we gave it a 3.

Reference 10

Reference

Chapter 6, reference 26. Duncan et al. AJCN 37:763. 1983.

Associated quote(s) and page number(s)

Page 131: “In some studies, those who follow a whole foods, low-fat, plant-based diet consume fewer calories. It’s not because they’re starving themselves. In fact, they will likely spend more time eating and eat a larger volume of food than their meat-eating counterparts.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. It’s an interesting randomized controlled trial that fed subjects a high-calorie-density diet or a low-calorie-density diet for five days each. The research team provided all foods so they had good control over what people were eating. The paper describes the diets as follows: “The [low-calorie-density] meals contained a large amount of bulk as fresh fruits, vegetables, whole grains, and dried beans with minimum fat. The [high-calorie-density] meals contained minimum bulk and larger amounts of fat and simple sugars as meats and desserts. (See sample menus in the Appendix.) For the [low-calorie-density] diet, energy density averaged 0.7 kcal/g and fiber content 7 g/1000 kcal (8); for [high-calorie-density] meals, energy density averaged 1.5 kcal/g and fiber content 1 g/1000.” The low-calorie-density diet is actually fairly similar to what The China Study recommends. They report that people on this diet did consume a somewhat larger volume of food and eat for a longer period of time than people on the high-calorie-density diet. Their total calorie intake was also barely more than half that of the high-calorie-density period, although they did feel a bit more hungry and less full.

Overall (average) score for reference accuracy

3.4 out of 4

We believe the diet is likely to mostly achieve its goal of reducing the risk of obesity and coronary heart disease. However, the evidence that it reduces cancer risk is not compelling. Its best application may be in improving cardiovascular health in people who are at high risk of heart attack and stroke. Yet the diet may harm health in other ways due to an increased risk of nutrient deficiencies including iron, long-chain omega-3 fatty acids, zinc, and iodine. It may be possible to achieve nutrient adequacy on this diet, but it requires careful planning that is beyond the scope of the information provided to the reader in The China Study, requiring the reader to purchase additional books from the authors.

Summary of the health-related intervention promoted in the book

The China Study recommends a diet that is based entirely on unrefined plant foods: fruits, vegetables, tubers, beans, whole grains, and limited nuts, but no meat, dairy, eggs, added fats or sugars, or processed foods. The book initially suggests that reducing intake of animal products and processed foods will be sufficient to provide general health benefits. We note that such advice is common in the general nutritional science and dietetics community. The China Study goes far beyond this by prescribing a much stricter diet regimen that is low in protein (10 percent; all from plant sources), very low in fat (10 percent), and very high in carbohydrate (80 percent). The book refers to this as a “whole food, plant-based diet.” It is important to note that this specific diet differs substantially from the distribution of protein, fat, and carbohydrate that is typically observed in people who follow a vegan diet, as highlighted within the book itself. The China Study also recommends taking a vitamin B12 supplement, a nutrient found only in animal foods, and vitamin D for people who don’t get much sun (pages 235-238).

Condition targeted by the book, if applicable

The China Study aims to improve general health, but particularly common chronic diseases/disorders like coronary heart disease, diabetes, obesity, and cancer.

Apparent target audience of the book

The China Study appears to be written for a general audience.

Criterion 3.1. Is the intervention likely to improve the target condition?

3 out of 4

The intervention received a score of 3, indicating that it is likely to moderately improve its targeted conditions. We previously reviewed evidence that it will probably reduce cardiovascular risk. It will also probably cause weight loss in people with excess weight, as demonstrated by randomized controlled trials of whole food plant-based diets. Weight loss will improve metabolic health more generally, again as demonstrated by randomized controlled trials of whole food plant-based diets. However, weight loss can be achieved through a variety of dietary and lifestyle changes. It’s not clear how much weight loss this diet causes relative to others, or how much of its health benefits are attributable to the weight loss it causes. Additionally, we aren’t aware of compelling evidence that this diet reduces cancer risk, which is a central claim of the book.

Criterion 3.2. Is the intervention likely to improve general health in the target audience?

3 out of 4

The intervention received a score of 3, indicating that it is likely to moderately improve general health. We have little doubt that the diet will reduce chronic disease risk overall, so in that sense it should improve general health. However, good health is a broader concept than simply avoiding disease. It includes maximizing cognitive, emotional, physical, immune, and reproductive health, and supporting development and resilience. Also, how hard it is to stick with a diet in the long run should be considered. This is where we have greater concerns about the diet recommended by The China Study. Humans have an evolutionary history eating animal foods that stretches back at least 2.6 million years, which may be why our bodies require a vitamin (B12) found only in animal foods. Several studies have indicated that vegan children do not grow quite as tall as their omnivorous peers on average, and people eating a vegan diet are at a high risk of specific nutrient deficiencies including iron. Additionally, due to the diet’s extreme nature it is likely to be difficult to adhere to for long periods of time. We are skeptical that the diet recommended by The China Study optimizes this broader conception of health, although it could still be a good choice for people at high risk of cardiovascular disease.

Criterion 3.3. Does the diet portion of the intervention promote an adequate nutrient intake for general health in the target audience?

1 out of 4

The diet received a score of 1, indicating that it is likely somewhat nutritionally inadequate. Fortunately The China Study recommends vitamin B12 supplementation, without which the diet would be very inadequate. With the addition of B12, the diet is mostly adequate but there is still an elevated risk of suboptimal nutrient intake, particularly of iron, long-chain omega-3 fatty acids, zinc, and iodine. Part of the problem is that certain nutrients, such as iron, are less efficiently absorbed from plant foods, so suboptimal absorption can occur even when dietary intake appears adequate. It may be possible to achieve nutrient adequacy on this diet, but it requires careful planning that is beyond the scope of the information in The China Study.

Overall (average) score for healthfulness

2.3 out of 4