Ancel Keys in the 1950s published the famous Seven Countries Study, which tied saturated fat intake to heart disease (among other findings). Critics claim that he hand-picked data from countries to fit his hypothesis. From wikipedia:

Critics of the Seven Countries Study (like Gary Taubes) say, "Keys chose seven countries he knew in advance would support his hypothesis." If Gary Taubes is correct, then Ancel Keys provides researchers with little to no valuable clinical information.

And from the Big Fat Lies documentary:

Keys had reliable data from 22 countries, and the results were all over the place. from 1:08

Is there any evidence that Ancel Keys cherry-picked the data for his study?

NOTE: I'm not asking whether his conclusion(s) are accurate, nor am I asking about the other claims made in the video. I'm not interested in discussing the merits (or lackthereof) of low-fat diets, etc.

  • 2
    You show praiseworthy restraint by focussing just on the missing data question. Worth an upvote. Hope you don't mind my edit of your title to make it more noticeable.
    – matt_black
    Oct 12 '13 at 12:16
  • 3
    If someone wants to do the effort to look this up, I know the information is out there in reliable sources. Just from memory, the story is that originally the seven countries study was all the data. Later he did a different study that included a lot more countries and found that it contradicted the hypothesis from the first one. It wasn't intentional cherry picking, but also the newer one was not publicized and the policies had already been put in motion based on the study with less data.
    – Matt
    Oct 14 '13 at 19:18

The question above is a little confused mixing claims about Ancel Keys' 7 countries study which he organised, and his 6 (22) countries study. These studies are commonly confused.

The 6 (22) countries study was one he authored himself and presented in 1953 at a symposium at Mt. Sinai Hospital, New York City. Some of the work came from field work done in Europe 1951-1952 when he was a Senior Fullbright Fellow headquartered at Magdalen College, Oxford, and other data came from research grants from the National Heart Institute, and the National Dairy Council, Chicago.

In the paper he describes his methods and he discusses the validity of death certification. He excludes

Western Germany, and Finland because of major population shifts and other effects from the war, and Iceland, Luxembourg and some colonies, and semi-independent states with very small populations.

In the paper, he clearly states that

So far it has been possible to get fully comparable dietary and vital statistics data from 6 countries.

Now if you look at his graph together with Uffe Ravnskov's right hand graph where the claim of cherry picking is made. Ravnskov includes data for all 22 countries for mortality data and consumption of total fat ( not saturated fat as is stated in the question ) which was collected by Yurushalamy & Hilleboe ( using different sources from Keys ).

6 countries graph

You now have a chance to see the data for the countries he dropped. Mexico's (14) data was notoriously unreliable, and appears to go against the trend. But note that Finland (7) is also dropped where the relationship appears to be even stronger (the numbering seems wrong for the countries). So, this is very odd cherry picking that removes data that both supports and weakens the hypothesis. And it is clear that the relationship still is present even if you include the data which was considered by Keys to be unreliable or not comparable, or where the diets were not considered different enough to include.

The 7 countries study was a totally different beast. This was a multicenter prospective long term study which started in 1957:

Beginning in 1957 Keys and his colleagues began what would eventually be known as the Seven Countries Study by surveying 12 000 men aged 40 to 59 from 18 areas of seven countries (Italy, the Greek Islands, Yugoslavia, the Netherlands, Finland, Japan, and the United States). Study communities were chosen for their contrasting dietary patterns and the relative uniformity of their rural laboring populations. Through central chemical analysis of the foods consumed by randomly selected families as well as diet-recall measures, Keys and his colleagues were able to determine that in societies where fat was a major component of every meal (i.e., the US and Finland), both the blood cholesterol levels and the heart-attack death rates were highest. Conversely, in cultures where diets were based on fresh fruit and vegetables, bread, pasta, and plenty of olive oil (i.e., the Mediterranean region) blood cholesterol was low and heart attacks were rare. The report published in 1970 had a decisive impact on CVD prevention, as it described one of the first studies to clearly show that dietary saturated fat leads to CVD, and that the relationship is mediated by serum cholesterol. [3]

Regarding Taubes' purported claim, that Keys picked the 7 countries for the 7 countries study knowing that their data would support his suspicions, then this is already addressed in his paper. The entry criteria included:

Study communities were chosen for their contrasting dietary patterns and the relative uniformity of their rural laboring populations.

It would be logical to assume that further entry criteria were that they shared similar diagnostic criteria, and the countries were stable enough to support such a long term study.

It is absurd for Taubes to write that Keys 7 countries study was invalid as he was somehow able to predict the results 29 years in advance, and that each population would show the same relationship of diet with heart disease both between countries and also inside each study population. (Keys published in 1986 on this study [4])


[1] Keys, A. Atherosclerosis: a problem in newer public health. J Mt Sinai Hosp N Y. 1953 Jul-Aug;20(2):118-39.

[2] Ravnskov, U. The Cholesterol Myths: Exposing the Fallacy that Saturated Fat and Cholesterol Cause Heart Disease

[3] Andrade J et al. Ancel Keys and the lipid hypothesis: From early breakthroughs to current management of dyslipidemia. BCMJ, Vol. 51, No. 2, March 2009, page(s) 66-72

[4] Keys A, Menotti A, Karvonen MJ, Aravanis C, Blackburn H, Buzina R, Djordjevic BS, Dontas AS, Fidanza F, Keys MH, et al. The diet and 15-year death rate in the seven countries study. Am J Epidemiol. 1986 Dec;124(6):903-15. PubMed PMID: 3776973

  • @Flimzy what exactly is the second part of the claim? I only see the cherry picking claim. I'll add something about Taubes though.
    – HappySpoon
    Jun 3 '14 at 20:04
  • I did .. it was in the final large quotation. But I've broken that part out again and quoted it again.
    – HappySpoon
    Jun 3 '14 at 20:35

It occurred to me also that this was a crucial question, yet it was very difficult for me to find any discussion outside of the enormous number of web sites parroting the "low carber" position. It's discussed in great detail here and in subsequent articles. Articles are listed, in numerical order, in the sidebar to the right, this being the first to mention Ancel Keys.

The broader set of articles is an extremely comprehensive attack on the science and promotion of the popular "low carb" position, beginning here.

  • note that the question, along with many "low carbers", confuses the 1953 study with the 1970 "Seven Countries" Study"
    – eeenok
    Mar 8 '14 at 2:49
  • 1
    Please quote the relevant passages from the source and put forth the final conclusion.
    – SIMEL
    Mar 8 '14 at 20:21
  • Your references are not reputable. Can you find better ones?
    – Sklivvz
    Mar 9 '14 at 0:38
  • @eeenok: That's an over-simplification... The study wasn't concluded until the 1980s, but results were being published as early as the 50s.
    – Flimzy
    May 28 '14 at 18:38
  • @Flimzy the study didn't start until 1957. eeenok is correct.
    – HappySpoon
    Jun 3 '14 at 8:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .