{"id":3514,"date":"2017-08-14T10:36:39","date_gmt":"2017-08-14T10:36:39","guid":{"rendered":"https:\/\/www.dietandhealthtoday.com\/?p=3514"},"modified":"2021-02-05T10:15:24","modified_gmt":"2021-02-05T10:15:24","slug":"the-seven-countries-study-part-2","status":"publish","type":"post","link":"https:\/\/www.dietandhealthtoday.com\/2017\/08\/the-seven-countries-study-part-2\/","title":{"rendered":"The Seven Countries Study – Part 2"},"content":{"rendered":"
<\/p>\n
On 1st August 2017, a white paper was published called \u201cAncel Keys and the Seven Countries Study: An Evidence-based Response to Revisionist Histories<\/a>.\u201d It was commissioned by an organisation called The True Health Initiative and it was written by Katherine Pett and three members of the council of The True Health Initiative: Joel Kahn; Walter Willett; and David Katz.<\/p>\n The white paper was focused on \u201cfour primary allegations\u201d, which it claimed had been proposed to discredit Keys and which it claimed were wrong. Part 1 of the review of The True Health Initiative paper<\/a> found that only one part of one of four allegations \u2013 the confusion between the Mount Sinai presentation six countries graph and the seven countries study \u2013 is commonly made in error. The rest was much ado about nothing, as Shakespeare would say.<\/p>\n In this second part, I will present the criticisms of the Seven Countries Study (SCS) that the white paper should have made. There are many limitations of the Seven Countries Study. I will cover the six most damning. The first issue alone renders the SCS redundant. (If you would rather watch a video to tell this story, here\u2019s one from February 2017<\/a> on this very topic).<\/p>\n 1) The SCS is an epidemiological study. <\/strong><\/p>\n The SCS might be of interest if there were not Randomised Controlled Trials (RCTs) and, better still, systematic reviews and meta-analyses of RCTs, examining dietary fat and heart disease. But there are: many of each. At the time the dietary guidelines were introduced (1977 US\/1983 UK) there were five RCTs available for the US authorities to consider (Refs 1-5) and a sixth was available to the UK committee (Ref 6).<\/p>\n Individually, none of these RCTs recommended that dietary guidelines should be introduced. On the contrary, only one study made a positive claim for its intervention after five years (Ref 7) and subsequently, this was moderated (Ref 4). Rose et al<\/em> (Ref 1) warned of possible harm by administering corn oil. The Research Committee concluded \u201cA low-fat diet has no place in the treatment of myocardial infarction<\/em>\u201d (p504) (Ref 2). The MRC Soya-bean oil intervention found no evidence that MI relapse would be materially affected by unsaturated fat in the diet (Ref 3). The LA Veterans study reported that total longevity was not affected and expressed concern about unknown toxicity of their intervention (Ref 5). Woodhill et al<\/em> noted that survival was significantly better in the control than the diet group (Ref 6).<\/p>\n Collectively, in meta-analysis, these RCTs found that dietary interventions made no difference to deaths from Coronary Heart Disease (CHD) or deaths from any cause (Ref 8).<\/p>\n If we generously ignore the fact that the dietary fat guidelines had no RCT evidence base at the time they were introduced, but see if they would be justified now with evidence currently available, I found that they would not be (Ref 9). Others have found similarly. A number of meta-analyses of RCTs, examining dietary fat and mortality, have been undertaken by other authors (Refs 10-13). A meta-analysis of prospective cohort studies has been undertaken by other authors, Siri-Tarino et al<\/em> (2010) (Ref 14). Two additional meta-analyses reviewed both RCTs and prospective cohort studies (Refs 15, 16).<\/p>\n Table 7.1 in this paper (on open view) (Ref 17) summarised the findings from these other meta-analyses of RCTs and\/or prospective cohort studies. There were 39 reports of risk ratios from meta-analysis with 95% confidence intervals. Of these, 4 reported significant findings; 35 reported no significant findings. That is important in itself. 35 out of 39 findings from 8 meta-analyses (Refs 8 and 10-16) found nothing statistically significant in an examination of dietary fat RCTs and\/or cohort studies.<\/p>\n The four significant findings were:<\/p>\n i) Chowdhury et al<\/em> found that trans fat intake was positively associated with coronary disease (Ref 16). I\u2019m not surprised. Having examined the evidence between coronary disease and fatty acids \u2013 including saturated, monounsaturated, polyunsaturated and trans fats, while also reviewing individual chain length fatty acids, palmitic (C16:0) and margaric (C17:0) \u2013 Chowdhury et al\u2019s<\/em> conclusion was \u201cCurrent evidence does not clearly support cardiovascular guidelines that encourage high consumption of polyunsaturated fatty acids and low consumption of total saturated fats<\/em>\u201d (p398) (Ref 16).<\/p>\n ii) Mozaffarian et al<\/em> reviewed the impact of replacing saturated fat with polyunsaturated fat (Ref 12). This study was criticised by Ravnskov et al<\/em> (including me) (Ref 18) for excluding two studies that would have moderated this conclusion \u2013 the Rose Corn oil trial (Ref 1) and the Sydney Diet Heart Study (Ref 6) \u2013 and for including a favourable, but non-randomised, non-controlled, cross-over, trial excluded by robust meta-analyses \u2013 the Finnish Mental Hospital Study (Ref 19). Interestingly, the recent American Heart Association \u2018Presidential Advisory\u2019 paper (Ref 20), also unfairly excluded the Rose Corn oil trial (Ref 1) and the Sydney Diet Heart Study (Ref 6) and also unfairly included the Finnish Mental Hospital Study (Ref 19) and funnily enough reached the same biased conclusion as Mozaffarian et al.<\/em><\/p>\n iii) & iv) The other two significant findings were related to cardiovascular disease (CVD) events and not mortality (Refs 11, 13). In 2011, including RCTs with a minimum of 6 months duration, Hooper et al<\/em> (Cochrane review) found 1 significant result and 11 non-significant results. The one significant result was that, when all RCTs were examined together, the risk ratio (RR) for CVD events from meta-analysis was 0.86 (95% CI 0.77 to 0.96). In 2015, including RCTs with a minimum of 24 months duration, Hooper et al<\/em> found one significant result and seven non-significant results. The one significant result was that, when a reduction in saturated fat was examined, the risk ratio (RR) for CVD events from meta-analysis was 0.83 (95% CI 0.72 to 0.96).<\/p>\n As a result of this 2015 review (an update of the 2011 review), Hooper et al<\/em> suggested that there may be a small reduction in cardiovascular risk with reduction of saturated fat intake (Ref 13). It was further suggested that replacing the energy from saturated fat with polyunsaturated fat \u201cappears to be a useful strategy, and replacement with carbohydrate appears less useful<\/em>\u201d (p2) (Ref 13) and replacement with monounsaturated fat unclear. However, of the 11 interventions contributing to this conclusion, only 1 documented both saturated fat reduction and reported that this was mainly replaced with polyunsaturated fat (Ref 5).<\/p>\n This prompts the question \u2013 why would Hooper et al<\/em> have found even one significant result (among many non-significant results) that other researchers didn\u2019t? The Hooper et al<\/em> meta-analyses of 2011 and 2015 included four small studies (646 people in total), not included in other meta-analyses, which were primarily studies of: diabetes (Ref 21); skin cancer (Ref 22); hypercholesterolemia (Ref 23); and glucose intolerance (Ref 24), but for which Hooper et al<\/em> obtained CVD event information from correspondence with authors. This non-published data was also, therefore, not peer reviewed. Some of it required only a cursory examination to establish that it was nutritionally unsound: Houtsmuller described the intervention diet as \u201ccarbohydrates 50 cal%, saturated fats 35 cal% and proteins 15 cal%<\/em>\u201d (Ref 21). As there is no food containing saturated fat without unsaturated fat, this diet cannot exist.<\/p>\n Notwithstanding that Hooper et al<\/em> reached a conclusion using non peer reviewed, non robust data… Notwithstanding that Hooper et al <\/em>reported a finding about replacing saturated fat with polyunsaturated fat, when only one study had actually done this… (and did not recommend dietary change)… when Hooper et al<\/em> subjected their one significant finding (for CVD events) to the required sensitivity analysis, even this one finding ceased to be statistically significant (Table 8, p121) (Ref 13), leaving no genuine significant findings. (I always credit Dr Trudi Deakin for alerting me to this point).<\/p>\n For the final nail in the Hooper et al<\/em> coffin, the most recent review (2015) (Ref 13) included no study of healthy people of both genders. The one primary, both-sex RCT available was excluded by Hooper et al<\/em>, for not meeting the 24 month duration criteria (Ref 25). Even if the one claim of a small benefit for CVD events had been robust, this had no \u201cgeneralisability\u201d and thus could not be applied to people from the general population. i.e. you and me.<\/p>\n It is worth remembering that dietary fat guidelines were introduced with the ambition of reducing deaths from CHD. No meta-analysis of RCTs and\/or prospective cohort studies has found any significant difference for dietary fat interventions and all-cause mortality or deaths from CHD, or associations with dietary fat and CHD mortality (Refs 8 and 10-16).<\/p>\n 2) The SCS contradicted contemporary epidemiological and RCT evidence. <\/strong><\/p>\n In Part 1 of the review of The True Heath Initiative white paper, in the \u201cIn Addition\u201d section, I quoted an extract from p5 of the white paper: \u201cUltimately, SCS suggested a link between dietary intake, specifically saturated fat, and heart disease. This conclusion, which corroborated other clinical and epidemiological evidence at the time, generated numerous hypotheses and has since inspired countless clinical trials.\u201d<\/em><\/p>\n I said that this was not correct: \u201cThe finding between saturated fat intake and heart disease was not<\/em> replicated by other clinical or<\/em> epidemiological evidence at the time\u201d (Refs 8, 26). The RCT evidence has been presented above. The six RCTs at the time neither individually, nor collectively, corroborated the SCS finding; quite the opposite.<\/p>\n Turning to epidemiological evidence, the six epidemiological studies of the time were: The Western Electric Study (Ref 27); The Seven Countries Study (Ref 28); a study conducted in London and the South East (Ref 29); The Framingham Heart Study (Ref 30); The Honolulu Heart Program (Ref 31); and The Puerto Rico Heart Health Program (Ref 32). The systematic review of these six epidemiological studies was an original publication from my Ph.D. (Ref 26). The summary table of evidence from the six studies is below:<\/p>\n Table 1 Outcome data from included prospective cohort studies for: study name; participant number and age range; years of follow-up; total fat and saturated fat for CHD-free vs. CHD deaths (Refs 30-32) or CHD-free vs. development of CHD (Refs 27-29); and other significant associations found.<\/p>\n Men\/Age (*)<\/p>\n<\/td>\n Follow-up yrs<\/p>\n<\/td>\n Total fat & CHD assoc?<\/p>\n<\/td>\n Sat fat & CHD assoc?<\/p>\n<\/td>\n Other Significant associations with CHD<\/p>\n<\/td>\n<\/tr>\n 1,989 4<\/p>\n<\/td>\n No<\/p>\n<\/td>\n No<\/p>\n<\/td>\n Smoking\/age of death of father\/coffee<\/p>\n<\/td>\n<\/tr>\n 12,770 5<\/p>\n<\/td>\n No<\/p>\n<\/td>\n Yes Previous MI. NO<\/strong> association found with smoking\/ activity or weight<\/p>\n<\/td>\n<\/tr>\n 337 20<\/p>\n<\/td>\n No<\/p>\n<\/td>\n N\/A<\/p>\n<\/td>\n Smoking\/Age of participant\/Inverse with higher calorie intake<\/p>\n<\/td>\n<\/tr>\n 859 4<\/p>\n<\/td>\n No<\/p>\n<\/td>\n No<\/p>\n<\/td>\n Inverse with higher calorie & higher alcohol intake (Note 2)<\/p>\n<\/td>\n<\/tr>\n 7,272 6<\/p>\n<\/td>\n No<\/p>\n<\/td>\n No<\/p>\n<\/td>\n Inverse with higher calorie & higher alcohol intake<\/p>\n<\/td>\n<\/tr>\n 8,218 6<\/p>\n<\/td>\n No<\/p>\n<\/td>\n No<\/p>\n<\/td>\n Inverse with higher calorie & rural living<\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n Table notes: N\/A = data not available; MI = Myocardial Infarction; CHD = Coronary Heart Disease. All six studies, including the SCS, concluded that total fat was not<\/em> associated with heart disease (Ref 26). The white paper acknowledged this: \u201cT<\/em>hough SCS itself never concluded that total fat intake should be restricted<\/em>\u201d (p5). (It does make you wonder how the total fat restriction of 30% ever happened). The SCS was alone in finding that saturated fat and coronary heart disease (CHD) were associated. The SCS also asserted that smoking, activity levels\/exercise and weight played no part in CHD. Much importance was assigned to a study for its saturated fat findings, which contradicted contemporary evidence about smoking, activity and weight.<\/p>\n 3) An inter-country study is fundamentally confounded.<\/strong><\/p>\n P14 of the white paper tried to position an inter-country study as preferable to a study of the same populations: \u201cResearchers carefully selected populations across a wide variation of dietary patterns, but with low variance of other characteristics to limit confounding<\/em>.\u201d Good luck with that one! This is a fundamental flaw of the Seven Countries Study; it is not a positive attribute.<\/p>\n The five non-SCS contemporary epidemiological studies listed above all studied the same populations. They studied men with location, climate, political environment, health care access and much more in common. They were looking to assess which dietary and lifestyle differences, within the same population, were associated with CHD.<\/p>\n The SCS was an inter-country comparison rather than a CHD\/CHD-free comparison. Men who developed CHD in Japan were compared with men who developed CHD in the US, as opposed to men who developed CHD in Japan being compared with men who did not<\/em> develop CHD in Japan. The SCS would have us believe that CHD in Japan vs. CHD in the US was down to the percentage of calories derived from saturated fat in the diet. Not geography, not climate, not lifestyle, not community, not connectivity to the land, not being a fisherman vs. being a railroad worker, not national Gross Domestic Product, not politics, not the amount of fish consumed vs. the number of hot dogs consumed…<\/p>\n 4) The SCS wasn\u2019t a robust dietary study.<\/strong><\/p>\n The dietary information available from The Seven Countries Study was not comprehensive. Dietary information was documented in Volume XVII of The Seven Countries Study (Ref 34) and in a non-mainstream Dutch booklet (Den Hartog) (Ref 35). Most of the volumes in Circulation<\/em> made no mention of dietary fat: total, saturated or any other kind (Refs 36-46). My Ph.D. thesis fully documented the few references to diet\/dietary fat in the original Circulation <\/em>20 volumes and noted the striking omissions. There was no dietary information in either Circulation <\/em>or Den Hartog for the USA railroad cohort or the Rome railroad cohort and Japan was not covered by the Den Hartog publication. The three Yugoslavian cohorts of Velika Krsna, Zrenjanin and Belgrade were reported in Volume XII of The Seven Countries Study (Ref 40), but there were no references to diet or dietary fat. The Den Hartog booklet reported percentage of calories from each macronutrient for these three cohorts, but with no food intake detail underpinning this.<\/p>\n To give you an idea of how poor dietary references were when they were made, here is one from Volume VIII in Circulation <\/em>about the Dutch cohort of Zutphen (which tells us nothing about the Dutch diet): \u201c…the data on the U.S. railroad men are not very precise, but the Americans seemed to be more like the Zutphen men than any other cohort in respect to total fats, kinds of fatty acids, etc<\/em>\u201d (pI-81) (Ref 47).<\/p>\n The opening volume of the study reported that 7-day dietary surveys were carried out on random samples of the cohorts, except in the United States where dietary data were obtained by 24-hour recall interview of all men at the time of entry examination and among the Rome railroad men, where a sample of employees were queried with the sole purpose of detecting whose diets differed from the norm for the region. For Japan, it was noted that a more carefully controlled survey was carried out at Tanushimaru, but not at the fishing village of Ushibuka (pI-6) (Ref 28). Aside from the US survey, the highest sample taken was 8.1% in Yugoslavia \u2013 Zrenjanin \u2013 and the lowest sample taken was 2.8% in Italy \u2013 Crevalcore. From data available, I calculated the average number of men sampled to be 3.4%. That cannot constitute a robust dietary study.<\/p>\n 5) Keys was aware of the impact of war, but the SCS ignored it.<\/strong><\/p>\n The context of The Seven Countries Study was important, as the men were war survivors. Men aged 40-59 years in 1956 (when the SCS started in preparation) were aged between 23-42 years when World War II started and aged between 29-48 years when World War II ended.<\/p>\n The True Health Initiative white paper made 13 references to war\/World War II and Keys also made a number of references to World War II in his publications. Keys was aware of the impact of war on countries studied, Finland especially. Keys excluded Finland from the 1953 Mount Sinai presentation\/six countries graph with the explanation: \u201cThe omissions are Western Germany and Finland, because of major population shifts and other effects from the war<\/em>” (p120) (Ref 48). In the 1954 symposium on atherosclerosis, Keys described the impact of war on the four Scandinavian countries (p190) (Ref 49). In a 1952 paper, Keys narrated: \u201cBesides the dubious point of attributing war-time changes in vital statistics to actual changes in atherosclerosis<\/em>…\u201d (p115-116) (Ref 50).<\/p>\n Despite this knowledge, Keys included Finland (east and west) in the Seven Countries Study. The greatest suffering directly involving any of the 16 cohorts was experienced in east Finland. The Finns fought the Winter War alone against the Soviet Union, the Continuation War with Germany against the Soviet Union and the Lapland War against Germany. Most of Finnish Karelia was lost to the Soviet Union with The Moscow Peace Treaty of 1940. Approximately 400,000 people, virtually the entire population of east Finland, were displaced as a result of World War II. I joked in my 2010 book The Obesity Epidemic<\/em>, I think I would have had a heart attack after all that! For the years following the war (the years of the Seven Countries Study), east Finland experienced the highest rate of CHD deaths among all 16 cohorts \u2013 3 to 4 times higher than the rates in west Finland, despite similar intakes of total fat and saturated fat. Keys was aware of the impact of war, but didn\u2019t acknowledge it in the Seven Countries Study.<\/p>\n The devastation that World War II inflicted upon east Finland\/north Karelia is important to remember because this region is often heralded as an example of how changing dietary fat intake can impact heart disease (Ref 51). The SCS confirmed that CHD mortality was greatest in this cohort. Which explanation do you consider the most plausible (especially given the diet similarity, but immense mortality difference, between east and west Finland)? i) CHD mortality was profoundly impacted by years of war and displacement. CHD mortality reduced thereafter because a) men can\u2019t die twice and b) the next generation did not endure the war. Or ii) north Karelia swapped butter for rapeseed oil based margarine.<\/p>\n\n\n
\n Study<\/td>\n \n \n \n \n \n \n Western Electric Study (Ref 27)<\/td>\n \n
\n(40-55)<\/p>\n<\/td>\n\n \n \n \n \n Seven Countries Study (Ref 28) Note 1<\/td>\n \n
\n(40-59)<\/p>\n<\/td>\n\n \n \n
\n(r=0.84)<\/p>\n<\/td>\n\n \n London bank and bus study (Ref 29)<\/td>\n \n
\n(30-67)<\/p>\n<\/td>\n\n \n \n \n \n Framingham (Refs 30, 33)<\/td>\n \n
\n(45-64)<\/p>\n<\/td>\n\n \n \n \n \n Honolulu (Refs 31, 33)<\/td>\n \n
\n(45-64)<\/p>\n<\/td>\n\n \n \n \n \n Puerto Rico (Refs 32, 33)<\/td>\n \n
\n(45-64)<\/p>\n<\/td>\n\n \n \n \n \n\n
\n <\/p>\n
\n(*) All subjects were free from CHD at baseline except for a number in the SCS.
\nNote 1: The Pearson correlation coefficient reported for the SCS represented 13 cohorts (both Japanese and the Rome railroad cohorts were missing). The coefficient represented the relationship between saturated fat intake for the 13 cohorts and CHD deaths and infarctions. Data for CHD deaths alone were not presented. Data for men without heart disease on entry were not available. The data did not compare fat intake of those with CHD vs. those without. The correlation applied to fat intake for cohorts relative to each other.
\nNote 2: The inverse associations with calories and alcohol are likely to be confounded by affluence \u2013 those who are better off tend to live longer\/be healthier and are more able to afford food and drink.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n