Sceptics undermine effective dietary and heart health advice

Recent reports questioning the link between saturated fats and coronary heart disease fail to convince

“Eat less saturated fat and more polyunsaturated fat” has been the central dietary advice for reducing early death from coronary heart disease (CHD) for more than four decades. This advice was based on evidence accumulated over many years, and the decline in developed countries in the numbers of premature deaths caused by CHD is attributable in some measure to its widespread acceptance.

In 2010 and 2014, two studies based on meta-analyses contradicted this longstanding advice,1,2 and their findings have been widely broadcast in the United States and the United Kingdom by the popular media, including the New York Times, Time magazine, New Scientist and the Independent.

The resulting discussion on the place of saturated fat in the diet and the management of cholesterol in the population has the potential to drastically impede further progress in reducing CHD. Why do the conclusions in these two articles depart so markedly from the international consensus that has been trusted until now?3

The first of these articles was the 2010 meta-analysis by Siri-Tarino and colleagues of prospective cohort studies.1 The authors found no significant association between saturated fat consumption and the risk of CHD. It is remarkable that this unconvincing review stimulated extensive interest in the media, whereas a large pooled analysis by experts from 10 universities, published in the same journal a year earlier, had not attracted the same attention. The 2009 article, by Jakobsen and colleagues, had reported that reducing saturated fat in the diet and replacing it with polyunsaturated fatty acids (PUFAs) was associated with a significantly reduced risk of CHD.4

The importance of experimental design

Experts in the US, Australia, the Netherlands, the UK, New Zealand and Norway have strongly criticised the conclusions reached by Siri-Tarino et al. Six major concerns were voiced.

1. With what did study participants replace dietary saturated fats?

Siri-Tarino et al did not consider most of the prospective studies in Jakobsen et al that reported dietary substitution of saturated fats by PUFAs.4

2. The validity of the dietary assessment methods varied between reports

In several studies, for example, participants were asked to recall what they had eaten during a single 24-hour period, an approach that is inadequate for assessing long-term dietary history.

3. A broader variety of dietary cultures should have been included

The optimum approach for evaluating an association between dietary saturated fats and CHD is to study groups from different dietary cultures. An example of this procedure was the Seven Countries Study (7CS), in which research dietitians watched meals being prepared, and collected foods from the local area and organised their transport to the University of Minnesota for chemical analysis of their fatty acid content.5Saturated fat intake in this study ranged between 2.5% and 22% of total energy intake, and its correlation with the number of CHD deaths was highly significant (r = 0.84).5

The media downplayed this finding when reporting the results, suggesting that the director of the 7CS had “cherry-picked” the seven countries. The cohorts described in the 7CS report had, however, been selected on the advice and support of Paul White, President Eisenhower's cardiologist, and because enthusiastic and reliable local medical researchers, as well as local funding that supplemented US support, were available during the 1950s. Further, a total of 16 cohorts were studied across the seven countries, which also enabled within-country comparisons. Two quite different Finnish cohorts were included, for example, while rural farming and coastal farming villages were compared in Croatia and Japan, as were three dissimilar communities in Serbia.5,6

4. Evidence from metabolic ward studies

The major evidence on the relationship between dietary saturated fat and CHD has been delivered by metabolic ward studies. This type of study has consistently shown since 19567that dietary saturated fats increase plasma levels of total and low-density-lipoprotein (LDL) cholesterol, which are strongly implicated in the pathogenesis of CHD.

5. The evidence from long-term dietary studies

Meta-analysis of the limited number of long-term controlled dietary trials in humans indicates that diets including reduced amounts of saturated fat and an increased PUFA content lead to significantly fewer CHD events.8

6. Historical trends in the epidemiology of CHD

In the early 1960s, experts in Western countries recommended a diet that included less saturated fat and more PUFA for people at risk of CHD, and that advice was soon extended to the entire community. CHD mortality (age-standardised) reached a peak in North America, Australia, Finland and other Western countries around 1965, and has since declined dramatically; in Australian men the rate dropped from 610 per 100 000 in 1965 to 135 per 100 000 in 2000 (the fall was similar in women).3,9 The reduced mortality associated with heart attacks in Australia, the US and in Nordic countries consisted primarily of a reduction in the number of deaths in patients on their way to hospital, indicating that improvements in the management of risk factors, rather than of hospital treatment, explained the change. A cholesterol-lowering diet was a major influence, together with reduced smoking and efficient antihypertensive medication. Cholesterol-lowering statin medications were not introduced into therapy until the 1990s.

Omega-6 fatty acids and heart disease

In the second article that attracted recent media attention, Chowdhury and colleagues analysed four types of evidence from prospective studies or randomised controlled trials concerning the relationship between dietary fat type and CHD events: dietary fatty acid consumption, circulating fatty acid levels, adipose tissue fatty acid levels, and dietary and supplements trials. They concluded that the evidence did not support cardiovascular guidelines that encourage high consumption of PUFAs and low consumption of saturated fats.2 It quickly made headlines around the world, but the findings have been criticised by other experts.10

In my detailed examination of this publication and the associated supplementary material (57 pages in total), in which I focused on the data for omega-6 PUFAs, I also identified a number of problems. In particular, studies were omitted, and the findings of those included were often incorrectly reported. References for these articles are listed in the Appendix.

In their review of prospective studies, Chowdhury et al incorrectly reported the results of two reports (omega-6 PUFAs were protective in the original articles). Seven further published studies were overlooked. In five of these, including two based on large participant numbers,11,12 a negative correlation between omega-6 PUFA intake and the risk of CHD was found. Had all these studies been included in the forest plot in Chowdhury et al, 10 of 15 would have been located on the left-hand side (indicating that omega-6 PUFAs were protective).

In their review of circulating fatty acids and CHD, Chowdhury et al included 10 studies in which levels of linoleic acid, the major PUFA, were measured; two of these were incorrectly placed in the forest plot. Seven studies in the literature were overlooked, including two based on large participant numbers;13,14 linoleic acid tended to be protective in all seven. Overall, this means that 13 of 17 studies should have been placed on the left-hand side of the forest plot (indicating that omega-6 PUFAs were protective).

In their review of adipose tissue fatty acids and CHD, only one study measured linoleic acid, but a further eight reports have been published, of which six found linoleic acid to be protective. A 2007 review included seven articles about adipose tissue fatty acids and CHD,15 none of which were cited by Chowdhury et al.

In their review of dietary and supplements trials, Chowdhury et al made no distinction between these two very different types of investigation. Most of the trials analysed in this section reported simple supplementation with fish oil or capsules containing eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) capsules. It seems inappropriate, however, for supplements trials to be pooled with dietary trials in which participants consumed both less saturated fats and more PUFAs. In the randomised dietary trials, participants were assisted to make large changes to their diets and to maintain them for several years. In the omega-6 forest plot (figure 14 of Chowdhury et al), the Sydney Diet Heart Study (SDHS) is clearly an outlier; it was not included in other meta-analyses of dietary trials,8and its authors wrote that “comparison of the mean diets of those who died and the survivors revealed only trivial differences”.16 If the SDHS trial is removed, the relative risk for CHD in omega-6 dietary trials is 0.81.

In all, Chowdhury et al omitted or incorrectly reported 25 studies of omega-6 PUFAs and CHD. The protective effect of PUFAs would have been clear if all published studies had been included in their meta-analysis. Changes to established public health guidelines should not be advocated unless all the relevant evidence has been reviewed.

A warning about meta-analyses

Professor Walter Willett (Harvard School of Public Health) told Science Insider that “The controversy should serve as a warning about meta-analyses.” These analyses compile the data of individual studies to reach a clear-cut conclusion. “It looks like a sweeping summary of all the data, so it gets a lot of attention. … But these days meta-analyses are often done by people who are not familiar with a field, who don't have the primary data or don't make the effort to get it.” The results of drug trials can be more easily combined because they are similar in design, he explained, but nutritional studies are more diverse. “Often the strengths and weaknesses of individual studies get lost … It's dangerous.”10

In conclusion, the findings of the two articles discussed here do not justify a change of longstanding cardiovascular dietary guidelines that have contributed to massive reductions of CHD in developed countries.