This is something of an ongoing review, chapter by chapter, of Gary Taubes’s extraordinarily dense book Good Calories, Bad Calories, which I usually shorten to GCBC. You might even consider this more of a fact-checking than a review, but whatever. I’m not going to get into a semantic argument. I wrote my first review of this book back in 2012, but after writing it I felt very unsatisfied. GCBC is such a dense book filled with so many unsubstantiated claims that I felt the book demanded a more thorough review. Other bloggers, like James Krieger at Weightology, seem to feel the same way and have tried to provide such a review only to eventually give up once they realize the gravity of the task. I may also give up at some point. I actually have given up a number of times only to feel compelled to hit at least one more chapter.
If you would like to read other parts of this ongoing review go to the table of contents on my Book Reviews page. FYI: All page numbers in this review refer to the hardback version of the book.
Not the Introduction
On pages 60-62 Taubes discusses a publication by the National Academy of Sciences titled Diet and Health: Implications for Reducing Chronic Disease Risk.
I’m not sure that Taubes actually reads it because he gets a few things wrong about it. First (and this is a minor point) he says the publication is thirteen hundred pages long, when in fact it is not even eight hundred pages long. At least that’s the version I have.
Second — and this is probably a more important point — he says about the publication:
It was no longer about the validity of the underlying science, which was no less ambiguous than ever, but about whether Americans should be eating low-fat diets or very low-fat diets.
Actually, Diet and Health recommends a diet of 30% of total calories from fat.1 I don’t see how this could be seen as a low-fat diet, much less a very low-fat diet. Clearly Taubes wants the committee to have recommended something like a very low-fat diet, but they did no such thing so he just lies and says they did anyway. (See: straw man)
Taubes then says:
One striking fact about this evolution is that the low-fat diets now being recommended for the entire nation had only been tested twice… The results of those trials had been contradictory.
Clearly Taubes didn’t bother to actually read Diet and Health. If he did he would have noticed the literally TWENTY-SEVEN PAGES of references regarding dietary fat alone! In fact, the bulk of the publication was devoted to explaining the evidence behind the recommendations. I don’t really blame Taubes for not reading it, though; it was quite long and it didn’t really fit nicely into the narrative he was attempting to craft. As a matter of fact one of the cited trials, titled Low-fat Diet in Myocardial Infarction in The Lancet, had a group consuming about half the percentage of calories from fat than what the National Academy of Sciences was recommending in Diet and Health so the results aren’t really comparable. Nevertheless, the low-fat group actually had beneficial effects to weight and serum lipids compared to the control group.2
* * *
Taubes does some really excellent minimizing on pages 63 and 64 when he discusses some data from the MRFIT trial:
For every one thousand middle-aged men who had high cholesterol—between, say, 240 and 250 mg/dl—eight could expect to die of heart disease over any six-year period. For every thousand men with cholesterol between 210 and 220, roughly six could expect to die of heart disease. These numbers suggest that reducing cholesterol from, say, 250 to 220 would reduce the risk of dying from a heart attack in any six-year period from .8 percent (eight in a thousand) to .6 percent (six in a thousand). If we were to stick rigorously to a cholesterol-lowering diet for thirty years—say, from age forty to seventy, at which point high cholesterol is no longer associated with an increased risk of heart disease—we would reduce our risk of dying of a heart attack by 1 percent.
By the way, I find it humorous that in the previous chapter Taubes was talking about what a disaster the MRFIT trial was, but here he’s discussing it as if it were the capital-T truth.
I think his data is actually wrong, too. He publishes a graph on the same page taken from a MRFIT article, and I think he’s simply guesstimating. The actual data from the trial says that a serum cholesterol value of 220 mg/dl would correspond to a CHD death rate of 5.81/1,000 and a cholesterol value of 250 mg/dl would have a death rate of 9.1/1,000.3,4 I suppose it’s a minor point, but I am going to use the real data in a second.
His choice of how to present the results is a great study of how to deceive with statistics. It also reminds of a scene in Pulp Fiction. You’ve probably seen it, but if you haven’t I’m not going to explain the context because I don’t really need to to make my point.
VINCENT What do you think about what happened to Antwan? MIA He fell out of a window. VINCENT That's one way to say it. Another way is, he was thrown out. Another was is, he was thrown out by Marsellus. And even another way is, he was thrown out of a window by Marsellus because of you.
So Taubes says “These numbers suggest that reducing cholesterol from, say, 250 to 220 would reduce the risk of dying from a heart attack in any six-year period from .8 percent (eight in a thousand) to .6 percent (six in a thousand).” That’s certainly one way to say it. Another way is that it reduces your risk of dying from a heart attack by 36%. Another way is it reduces your relative risk of CHD death from 2.88 to 1.84 (with 1.0 being optimal). And even another way to say it is that this data suggests that 205,954 lives in the US could be spared by lowering serum cholesterol from 250 to 220, assuming 20% of the population has cholesterol in excess of 250 mg/dl.
* * *
Between 1987 and 1994, independent research groups from Harvard Medical School, the University of California, San Francisco, and McGill University in Montreal addressed the question of how much longer we might expect to live if no more than 30 percent of our calories came from fat, and no more than 10 percent from saturated fat, as recommended by the various government agencies. All three assumed that cholesterol levels would drop accordingly, and that this low-fat diet would have no adverse effects, which was still speculation rather than fact.
The Harvard study, led by William Taylor, concluded that men with a high risk of heart disease—such as smokers with high blood pressure—might gain one extra year of life by shunning saturated fat. Healthy nonsmokers, however, might expect to gain only three days to three months. “Although there are undoubtedly persons who would choose to participate in a lifelong regimen of dietary change to achieve results of this magnitude, we suspect that some might not,” the Harvard investigators noted.
The UCSF study, led by Warren Browner, was initiated and funded by the Surgeon General’s Office. This study concluded that cutting fat consumption in America would delay forty-two thousand deaths each year, but the average life expectancy would increase by only three to four months. To be precise, a man who might otherwise die at sixty-five could expect to live an extra month if he avoided saturated fat for his entire adult life. If he lived to be ninety, he could expect an extra four months.* The McGill study, published in 1994, concluded that reducing saturated fat in the diet to 8 percent of all calories would result in an average increase in life expectancy of four days to two months.
Browner reported his results to the Surgeon General’s Office, and only then submitted his article to JAMA. J. Michael McGinnis, the deputy assistant secretary for health, then wrote to JAMA trying to prevent publication of Browner’s article, or at least to convince the editors to run an accompanying editorial that would explain why Browner’s analysis should not be considered relevant to the benefits of eating less fat. “They would have liked it to come out the other way,” explained Marion Nestle, who had edited the Surgeon General’s Report on Diet and Health and had recruited Browner to do the analysis. This put Browner in the awkward position of protecting his work from his own funding agents. As he wrote McGinnis at the time, “I am sensitive to the needs of your office to put forward a consistent statement about what Americans should do, and to your dismay when a project that you have sponsored raises some questions about current policy. I am also concerned that the impacts of recommendations that apply to 240 million Americans are clearly understood. This manuscript estimates the effects of one such recommendation—altering dietary fat intake to 30 percent of calories—based on the assumptions that underlie that recommendation. Shooting the messenger—or creating a smoke screen—does not change those estimates.” JAMA published Browner’s article—“What If Americans Ate Less Fat?”—without an accompanying editorial.
*Browner’s analysis also assumed that restricting dietary fat would reduce cancer deaths, which was speculative then and is even more speculative now.
Although the analysis is perhaps more pessimistic about the benefits of reducing saturated fat, it is also a bit more nuanced than Taubes makes it seem.5 The figures of 42,000 saved lives are in there as is the extra four months of life, but there’s also the following:
The analysis indicates that if those assumptions are valid, restricting dietary fat to a maximum of 30% of total energy intake could reduce CHD mortality rates by between 5% and 20%. Mortality due to cancers of the breast, colon, and prostate might be reduced by an even greater proportion. Similar reductions would also occur in the incidence of these diseases.
From the public health point of view, an increase of 3 months in life expectancy multiplied by 240 million Americans results in about 60 million years of additional life […]
Browner also points out that since the 3-4 month estimate is a statistical average, the real effects of lowering saturated fat would probably translate to several extra years in some individuals and possibly no increased lifespan in others.
But that is not the interesting part… The interesting part is where Taubes states that J. Michael McGinnis tried to shut down the publication of the article. This is based on a letter (presumably) that Browner wrote to McGinnis back then. I have contacted both Dr. Browner and Dr. McGinnis to see if they can confirm this. As of this writing I have received no response from Dr. Browner, but I did receive one from Dr. McGinnis that was very interesting. McGinnis claims he was never contacted by Taubes for his side of the story and that what Taubes wrote “certainly doesn’t ring either familiar or true.” In McGinnis’s own words:
Neither Marion nor I worked in the Surgeon General’s office—rather in a parallel office in HHS—but if the paper was done when she worked with me, it would have been nearly 30 years ago. I really have no recollections on any of what is reported in the passage—including the commissioning of the paper. I do recall the general reported finding on the limited impact on heart disease deaths of the lipid intake reduction indicated, although not the individuals or center (or centers) making the reports. Given the fact that the finding of small effects of isolated dimensions was not unusual (recall MRFIT), it would not have been particularly surprising. It is certainly possible that I would have been of a mind that a narrowcast analysis could be misleading if interpreted as reflecting on the impact of broader associated dietary changes. But unless I was a reviewer commenting on methodologic limits, it would be very unlike me—I cannot recall a single instance—to separately suggest that an article not be published. Indeed, to the contrary, I am squarely in the camp that there is great merit in publicly “calling the question” on these issues to prompt further analytic advances.
Dr. McGinnis also said that he will happily recant if someone were to produce the letter in question. I also contacted Dr. Marion Nestle and she claims that, while she can’t remember the specifics, what Taubes wrote about her remarks are generally true.
* * *
On page 72 Taubes discusses the famous Nurses Health Study and states
In 1992, Willett published the results from eight years of observation of the Nurses cohort. Fifteen hundred nurses had developed breast cancer, and, once again, those who ate less fat seemed to have more breast cancer. In 1999, the Harvard researchers published fourteen years of observations. By then almost three thousand nurses had contracted breast cancer, and the data still suggested that eating fatty foods (even those with copious saturated fat) might protect against cancer.
Walter Willett and Meir Stampfer are two of several authors of these obervational studies.6,7 (Interestingly, Taubes says in a blog post that Willett and Stampfer are not real scientists, they don’t do “real” science, and epidemiology is not to be trusted.) The cited studies do not provide evidence for his claim that copious amounts of saturated fats might protect against cancer. The best thing he can say is that consumption of SATURATED fat doesn’t appear to increase one’s risk of BREAST cancer. Of course the same could also be said of MONOUNSATURATED, POLYUNSATURATED, and even TRANS fats, at least according to these studies.
* * *
On page 73 Taubes gets on the polyunsaturated-fats-cause-cancer bandwagon again stating:
This laboratory evidence that dietary fat caused breast cancer began to evaporate as soon as Diet, Nutrition, and Cancer was published, and researchers could get funding to study it. By 1984, David Kritchevsky, one of the authors of Diet, Nutrition, and Cancer, had published an article in Cancer Research reporting on experiments that had been explicitly designed to separate out the effects of fat and calories on cancer, at least in rats. As Kritchevsky reported, low-fat, high-calorie diets led to more tumors than high-fat, low-calorie diets, and tumor production was shut down entirely in underfed rats, regardless of how fatty their diet was.
However, the results from the study itself are different:8
The data from 3 separate studies altering caloric and fat intake support the conclusions that a high-fat diet influences initiation as well as promotion of DMBA-induced mammary tumorigenesis, that high caloric intake increases tumor yield, and that restriction of caloric intake (even concomitant with the provision of twice as much dietary fat) inhibits tumor formation during the promotion period.
At least he got part of the study correct. By the way, I wonder how Taubes squares this circle in his mind when he later advocates eating all the calories you want — as long as it’s all meat, of course.
* * *
Nonetheless, the pervasive belief that eating fat causes breast cancer has persisted, partly because it once seemed undeniable. Purveyors of health advice just can’t seem to let go of the notion. […] By 2006, with the next release of cancer-prevention guidelines by the American Cancer Society, the ACS was acknowledging that “there is little evidence that the total amount of fat consumed increases cancer risk.” But we were still advised to eat less fat and particularly meats (“major contributors of total fat, saturated fat and cholesterol in the American diet”), because “diets high in fat tend to be high in calories and may contribute to obesity, which in turn is associated with increased risks of cancers.” (Saturated fats, in particular, the ACS added, “may have an effect on increasing cancer risk,” a statement that seemed to be based solely on the belief that if saturated fat causes heart disease it probably causes cancer as well.)
Of course, Taubes ignores the reference in the text that immediately follows “may have an effect on increasing cancer risk” and instead tells his readers that the statement is based on some unfounded belief. In case you’re wondering the reference is a review article that pours over several relevant studies on fat and cancer.9 What Taubes also omits is that the main reasons the ACS recommends avoiding red and processed meats have little to do with fat content.10 I’ll let the ACS explain:
Many epidemiologic studies have examined the association between cancer and the consumption of red meats (defined as beef, pork, or lamb) and processed meats (cold cuts, bacon, hot dogs, etc.). Current evidence supports an increased risk of cancers of the colon and/or rectum and prostate. More limited evidence exists for other sites. Studies that have examined red meat and processed meat separately suggest that risks associated with processed meat may be slightly greater than red meat, but the consumption of both should be limited.
Meat contains several constituents that could increase the risk of cancer. Mutagens and carcinogens (heterocyclic amines and polycyclic aromatic hydrocarbons) are produced by cooking meat at high temperatures and/or by charcoal grilling. The iron content (heme) in red meat may generate free radicals in the colon that damage DNA. Substances used to process meat (nitrates/nitrites and salt) contribute to the formation of nitrosamines that can damage DNA. It is also possible that the fat content in meat contributes to risk. For example, foods that are high in fat increase the concentration of secondary bile acids and other compounds in the stool that could be carcinogens or promoters of carcinogenesis.
These cancer properties of red and processed meats have been well-documented over the years. There is a mountain of evidence for them. But of course Taubes has never been one to let evidence ruin a good story about how people that disagree with him are pinheaded schmucks.
* * *
Pg 75, Taubes lies about the results when discussing a study by the Women’s Health Initiative claiming that
The women on the diet also consumed fewer calories—averaging 120 calories a day less than the controls over the eight years of the study.*
*They did not, however, lose any weight because of this, which is paradoxical, and an issue we will discuss later.
Actually, everyone on the diet lost a significant amount of weight.11
* * *
On page 84 Taubes discusses the results of a Cochrane review*:
The review concluded that the diets, whether low-fat or cholesterol-lowering, had no effect on longevity and not even a “significant effect on cardiovascular events.”
The first part of that claim is actually true. The second lies about the results of the systematic review. It actually reduced cardiovascular events by 16%. The same authors published the Cochrane results in BMJ and they concluded then12:
The findings on cardiovascular events are broadly in keeping with benefits that might be expected from modest lowering of cholesterol concentration and certainly provide support, at an individual level, for the central role of dietary fat intake in the causation of cardiovascular disease.
The results also indicated a reduction in mortality by 9%, but that figure was within the 95% CI. In any case, if you’re curious about the updated results they are basically unchanged.13 From the 2012 Cochrane review:
Dietary change to reduce saturated fat and partly replace it with unsaturated fats appears to reduce the incidence of cardiovascular events, but replacing the saturated fat with carbohydrate (creating a low fat diet) was not clearly protective of cardiovascular events (despite small improvements in weight, body mass index, total and LDL cholesterol).The protective effect was seen almost exclusively in those who continue to modify their diet over at least two years, and in studies of men (not those of women). Dietary advice to those at high risk of cardiovascular disease (particularly where lipid lowering medication may not be available), and probably also to lower risk population groups, should continue to include dietary fat modification, possibly as part of a Mediterranean dietary pattern, and it should be stressed that this is a permanent pattern of eating.
*Something Taubes holds in high regard, unless of course a result might contradict his thesis (as is the case with a Cochrane review of salt and hypertension) in which case they are meaningless. But more on that later.
1. National Academy Press. Diet and Health: Implications for Reducing Chronic Disease Risk. (1989). at <http://www.nap.edu/openbook.php?record_id=1222>
2. A Research Committee. Low-fat Diet in Myocardial Infarction: A Controlled Trial. The Lancet 286, 501–504 (1965).
3. Martin, M. J., Browner, W. S., Hulley, S. B., Kuller, L. H. & Wentworth, D. Serum cholesterol, blood pressure, and mortality: implications from a cohort of 361,662 men. The Lancet 328, 933–936 (1986).
4. Stamler J, Wentworth D & Neaton JD. Is relationship between serum cholesterol and risk of premature death from coronary heart disease continuous and graded?: Findings in 356 222 primary screenees of the multiple risk factor intervention trial (MRFIT). JAMA 256, 2823–2828 (1986).
5. Holmes, M. D. et al. Association of dietary intake of fat and fatty acids with risk of breast cancer. JAMA 281, 914–920 (1999).
6. Browner WS, Westenhouse J & Tice JA. What if americans ate less fat?: A quantitative estimate of the effect on mortality. JAMA 265, 3285–3291 (1991).
7. Willett, W. C. et al. Dietary fat and fiber in relation to risk of breast cancer. An 8-year follow-up. JAMA 268, 2037–2044 (1992).
8. Kritchevsky, D., Weber, M. M. & Klurfeld, D. M. Dietary Fat versus Caloric Content in Initiation and Promotion of 7,12-Dimethylbenz(a)anthracene-induced Mammary Tumorigenesis in Rats. Cancer Res. 44, 3174–3177 (1984).
9. Kolonel, L. N. Fat, meat, and prostate cancer. Epidemiol. Rev. 23, 72–81 (2001).
10. Kushi, L. H. et al. American Cancer Society Guidelines on Nutrition and Physical Activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA. Cancer J. Clin. 56, 254–281; quiz 313–314 (2006).
11. Howard, B. V. et al. Low-fat dietary pattern and weight change over 7 years: the Women’s Health Initiative Dietary Modification Trial. JAMA 295, 39–49 (2006).
12. Hooper, L. et al. Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane Database Syst. Rev. Online CD002137 (2001). doi:10.1002/14651858.CD002137
13. Hooper, L. et al. in Cochrane Database Syst. Rev. (The Cochrane Collaboration) (John Wiley & Sons, Ltd, 2012). at <http://doi.wiley.com/10.1002/14651858.CD002137.pub3>
Awesome work. Keep it up. I started reading GCBC and checked the references as I was going along and I had to stop because it was clear that Taubes was misrepresenting the references he quoted.
I find it a pretty sad state of affairs that this kind of work falls to bloggers and podcasters. Some highly-visible skeptics and science journalists should have been on this case when the book was first published. It might have mitigated the damage Taubes has done and continues to so to public health.
Hi Seth. Love this chapter-by-chapter consideration of the “errors” in GCBC. How many chapters in the book are there?
More than I’d like, unfortunately. There are 24, I think, and I’m only four chapters in… ugh.
Amazing what effort went into this tome of smoke and mirrors in the first place.…almost as if the book’s length itself was devised to dissuade close scrutiny.