Search This Site!
Nutrition Past and Future

4 The Journalist Gary Taubes 4: Ancel Keys Was Very Bad 2


We are now ready to examine the legendary paper which faulted that terrible man Ancel Keys for manipulatively comparing only six countries instead of all the 22 countries with available data. The two heroic skeptics who took on the dark lord of the lipid hypothesis were Jacob Yerushalmy and Herman Hilleboe.

Here is the scatter of the 22 countries that is such an iconic image in the low carb folklore. You will see it a lot online if you look for it. Even with all these dots you should be able to see that at least in a general sense, as fat consumption increased in these countries, so did heart disease mortality. The cholesterol deniers say that they could pick six other countries from this graph and show no increase in deaths with fat intake among those, but to do that, wouldn’t they have to be the ones doing the cherry-picking? When someone like Chris Masterjohn says that he could cherry pick six other countries, as you just heard him say in the previous video, isn't he by definition making a fallacious argument? Listen to him again...

MASTERJOHN: As most of your listeners probably already know, there was data available for 22 countries and when all those countries are included the line doesn’t look very clean at all and in fact you could draw the opposite line if you wanted to cherry pick six different countries.

He says himself that one would have to cherry pick to make this data contradict Ancel Keys and diet-heart. How can he fault Keys if he doesn't understand the problem with this?

Here is the mortality data our authors used. These skeptics, Yerushalmy and Hilleboe, showed a total lack of skepticism about some of these numbers, unlike Ancel Keys.

They said Mexico’s rate of death was “startlingly low”, and you can see how out of place Mexico's dot seems on their graph. They might have been even more startled to learn that this was because of Mexico's badly inaccurate data which was used for calculating their death rates.

As I have noted in a past video, Mexico had not even developed a death certificate system when that data was collected.

Here are other researchers from around that same time who knew that Mexico’s mortality data could not be trusted. Ancel Keys didn’t use this data, yet Yerushalmy and Hilleboe did. That’s because Keys exercised some degree of quality control over his numbers. Yerushalmy and Hilleboe applied no discretion at all.

I have also pointed out France on this graph, another country which had an unusually low number for the classification “B-26”, which represents arteriosclerotic and degenerative heart disease. They seem to have eaten a lot of fat yet they don’t seem to have paid a high price for that in heart disease mortality rates. This apparent phenomenon has famously been called “The French Paradox." I’ll come back to that idea in a moment. Let’s finish with this paper first.

Here is what these guys had to say about Keys’ inclusion of only six countries. They said, “Since no information is given by Keys on how or why the six countries were selected…” Now, having seen my last video, you have seen for yourself that Keys explained his choices and you know that this statement is false. One may criticize his choices. One may say he did not adequately explain himself. But one may not honestly say that “no information” was given by him. The dogmas low-carbers believe the most very often seem to be based on either bad information or outright lies. This statement is a lie.

Yerushalmy and Hilleboe said that practices in each country may have varied with respect to the way deaths were classified. You saw in the last video that Keys did his best to account for this, unlike them.

Here you can see that when later researchers attempted to correct for differences in death classification methods, the Keys graph still mostly held up nicely. This criticism of Keys by Yerushalmy and Hilliboe has not held up with the passage of time.

The two critics also took issue with the inaccuracy of his data for food consumption. They said “it is highly probable that far more edible dietary fat is thrown into waste cans in the United States than in less fortunate countries.” They didn’t know about the specific problem with saturated fat back then.

If they did, they would have realized that saturated fats sourced from animals would have been much less likely to be thrown out than vegetable fats because they are as a rule more expensive and consequently in lesser supply. This fat waste argument makes no sense given our present knowledge of fats so this complaint of theirs doesn’t hold up to scrutiny, either.

There really isn’t any question that the US and Canada were consuming more animal fats than the Italians and the Japanese back then. This was not a reasonable criticism. It's just skeptical-sounding noise which only pandering lightweight bloggers would find persuasive today. They have trouble finding good material to write about so they recycle dumb arguments like this.

So Taubes tells us that when all the 22 countries are put together, the association between fat and heart disease vanished. This is absolutely not true. Taubes is trying to pull a fast one here.
Look at the second and third lines across for calories from fat and calories from animal fat and you will see these correlations remained strongly positive. The association did not vanish. If it did, those numbers would be tiny or negative, like they are for vegetable fat and vegetable protein and carbohydrate. That's a pretty clear lie but this point leaves out something even more important.

I have highlighted animal protein in red. Calories from animal protein had a stronger relationship with coronary deaths than any other element of the diet in their paper. This is what the journalist Gary Taubes decided you shouldn’t know. He's such a thorough researchers, isn't he?

Here is where the two critics discuss this. “In each of the heart disease groupings the strongest association is with total number of calories from animal protein, and in each case the rank correlation coefficient is statistically significant.” Now they do go on to downplay this finding, but the numbers don’t lie, unlike some journalists.

I find it curious that it is so easy to find their scatter plot for fat online, but no one seems to want to show the one they made for animal protein. It looks a lot cleaner than the one for fat, don’t you think? Good luck cherry-picking that one, low carbers! Sorry I let this secret information out!

In 1972, Connor and Connor used data from 29 countries and looked for hazard ratios just as Yerushalmy and Hilleboe did. They, too, found a strong correlation for animal protein and coronary deaths. What do you think of this study, low carbers? They used even more data than Keys or Yerushalmy and Hilleboe did. 29 countries! You wouldn’t want to ignore this important study and appear to be lazy and dishonest researchers yourselves, would you?

Yerushalmy and Hilleboe stated the obvious, saying that animal foods are an index of the wealth of a nation. This is true, but as we know today, this is exactly why developed nations have historically had more heart disease than poor ones. They can afford more of these unhealthy foods.

Taubes leaves the impression that nothing interesting was later put forth on the subject of diet-heart by Yerushalmy or Hilleboe. You might find it interesting that Hilleboe published a paper some years later that declared that high cholesterol was indeed an important risk factor for heart disease. Taubes apparently couldn't find the space to include this information in his big book.

You also might like to know that Yerushalmy actually argued that cigarette smoking did not cause babies to be born underweight. If he could be so wrong about smoking, would you really want to trust him about animal fats?

Here is a news story that makes this very clear for you.

Usually the animal food apologists will confine their criticisms of Keys to that 1953 paper we saw comparing six countries. They usually don’t go after his Seven Countries Study. Gary Taubes is not the usual animal food apologist, though. Lacking judgment, here he does take a clumsy, ill-considered swipe at it. First, realize the Seven Countries Study was a large scale effort with multiple cohorts of patients being examined and interviewed by specially trained medical staff. This was not a simple armchair number crunch of garbage data, like the Yerushalmy and Hilleboe paper was. Look what Taubes has to say about it. “Keys chose seven countries he knew in advance would support his hypothesis.” Let’s take a moment to think this very curious objection through. Had someone done a study like this before him so he could know this in advance? The answer is no. So how could he have known that these countries would be supportive of his hypothesis? More importantly, why would they have turned out to be supportive of his hypothesis if his hypothesis were wrong? Were these seven countries filled with special people who really do get heart disease from saturated fat, unlike the people everywhere else? If so, isn’t Taubes saying that the United States, one of the countries Keys studied in this project, has special people he admits do suffer from saturated fat-induced heart disease? Mr. Taubes, would you care to explain why the Americans they studied seemed to develop heart disease on fatty diets? This objection makes sense to Taubes but it doesn’t make any sense to me. Taubes says the story would have been different had they studied the French or the Swiss. I guess Taubes thinks they are special people of a different kind. I will talk about the Swiss in my Cholesterol Confusion video about poverty, but now is the time to talk about the French.

There has been much written about this topic that you can easily find online. I don’t need to get deep into the usual explanations for the French Paradox here. The various conjectures put forth over the years have included their moderate alcohol consumption, the protective natural chemicals found in their wines, chocolates and plant foods, their lower dairy consumption, and the time lag between their dietary changes toward more fat consumption and their heart disease statistics actually turning upward to reflect that. Of these I think that the various effects of wine have the most bearing here.

But I don’t make much of the French Paradox. Once you look at it, there really isn’t much of a paradox to it. The French do experience more heart disease with higher blood cholesterol just like everyone else. It turns out they aren't so special after all, Mr Taubes. This famous study from 1980, which you seem to have somehow missed, found that their risk factors were not unusual. “Age, blood-pressure, and cholesterol level are ‘responsible’ for comparable risk gradients in France just as in other countries,” the authors wrote.

This paper should have let the air out the French Paradox for good, especially for the time period in question in this video. Referenced in it is a prominent cardiologist living in France in 1958, who stated that their data was either fanciful or wrong. It was simply the convention in France in the 1950s to avoid ascribing cause of death to coronary atherosclerosis.

This author, Pierre Ducimetière, concluded that coronary death rates in France “should be considered as negatively biased estimates” which “cannot be used validly in ecological studies.”

Ducimetière found that their death rates from heart disease were not so exceptional. He noted, however, that there did seem to be a pattern of lower rates of coronary death with lower latitudes. This is an interesting observation.

And it is an observation not so different from a similar one made by Ancel Keys about Italy. More animal foods were consumed in the North, and as a result they had more heart disease risk.

Ducimetière also did not believe the French consumed an unusually high amount of animal fat, which is the supposition at the heart of the myth of the French Paradox.

If Taubes were sincerely interested in learning why those seven countries were selected instead of imagining nonsensical sinister motives, he could simply click over to the website of the university where Keys worked.

The University of Minnesota School of Public Health will spell it all out for him.

He can also pause this video to understand and accept this slide, a photo taken from the actual Seven Countries Study book published by Harvard University Press. Here is an excerpt of Keys’ own account of the selection process. Taubes will read here and in that book that his choices were governed by very practical considerations of cost, contacts, and logistics, as well as a desire to have communities included that had meaningfully contrasting diets.

The Seven Countries Study results were also published in a large supplemental issue of the journal Circulation. In this version, Keys gave us more information about the selection process. I have only reproduced a portion of it to give you a sense of it. Keys was trying to produce the best research he could within the constraints of his limited finances, and this affected his selection of cohorts.

In this passage he preemptively dealt directly with Mr. Taubes’ allegation that he knew what his team would find. The truth is quite different. They did have a rough idea of the situation in each place, but they had no good data for some of the locations they included.

So you see, Ancel Keys was forthright and clear in his methods. The Seven Countries Study, like C. Colin Campbell’s China Study, was produced by many dedicated professionals. It was not an attempt to confuse and distort the science. That’s what Good Calories, Bad Calories is.

Mr. Taubes needs to propagate fictions about Ancel Keys and The Seven Countries Study because for a saturated fat defender like him, the results it produced are quite damaging to his beliefs and/or marketing strategy. Here you see quite starkly how cholesterol and saturated fat both related to deaths from coronary disease. Maybe Mr. Taubes could create an equally clean looking graph comparing book sales with lies told to fat people.

Just as Ancel Keys has been reimagined by the low carbers as a villain, John Yudkin is now someone they think should be regarded as a hero. I’ll offer you a rare skeptical take on him because I’ve been a bit more thorough in my research than the journalist Gary Taubes.