Recently, the world was abuzz talking about the American Heart Association’s (AHA) Presidential Advisory on dietary fats and cardiovascular disease. The report was quickly dismissed by various health figures online (Food Babe, Dave Asprey, Chris Kresser, Sustainable Dish, Mark Sisson, Dr. Axe, Dr. Mercola, JJ Virgin, Medium) but the original research cited throughout the report was never adequately addressed. Although most people likely aren’t interested in the research, there are some news sites out there that do see the unaddressed research as a problem, like Snopes.
I am well aware that there are apparently 17 systematic reviews and meta-analyses that point to no detrimental effect of saturated fat intake and cardiovascular mortality, but I was curious what studies the AHA cited to support their conviction that saturated fats do increase mortality, and reducing their consumption in the diet decreases it. Were those studies well-designed, controlled, and did they adequately address the research question? No one seems to have answered that yet, so I decided to do so myself.
In this post, we will discuss the central argument in the AHA’s report, which was that replacement of saturated fat in the diet with polyunsaturated fat reduces heart attacks and atherosclerosis, by examining the studies they cited. The AHA cited four “core” trials that answered this question, one of which we will discuss in depth, as it was the only one which attempted to accurately describe the diets between two groups. The other studies included a high polyunsaturated fat diet, and a control diet which was not delineated. They also cited three meta-analyses that found decreased cardiovascular mortality with a reduction in saturated fat intake, depending on what numbers were added up. Those studies will be discussed in part II.
This post is on the longer side, so feel free to navigate through it using the table of contents. You are about to discover how the AHA attempte
d to back up their claims, and what kinds of foods people ate in one of the studies I found comprising the core group of literature supporting their viewpoint. We will begin with some background, discussing what has already been covered on this topic, before moving into the research.
Table of contents
- 1 Background
- 2 Replacement of Saturated Fats With Polyunsaturated Fats Reduces Atherosclerosis?
- 3 Interpretation
- 4 Conclusion
After the AHA released their report, the internet was focused on their statements regarding coconut oil. I personally don’t enjoy consuming it internally; it doesn’t taste great, but I do love fresh coconut meat, and occasionally add coconut milk to my cooking, so I’d be really curious if a moderate consumption of these foods could increase my risk of cardiovascular complications. The AHA devoted a few paragraphs explaining why they do not recommend the consumption of coconut oil. An excerpt from that section reads:
“A recent survey reported that 72% of the American public rated coconut oil as a “healthy food” compared with 37% of nutritionists. This disconnect between lay and expert opinion can be attributed to the marketing of coconut oil in the popular press…because coconut oil increases LDL cholesterol, a cause of CVD, and has no known offsetting favorable effects, we advise against the use of coconut oil.”
Their distinction between “lay” and “expert” opinion is certainly a fascinating one, as it’s important to note that nutrition professionals with profoundly “expert” opinions have recommend sugary cereals as a healthy breakfast option. As Food Babe explained, the American Heart Association’s sponsors include companies that make these kinds of cereals, as well as other products which are slowly being phased out of the health-conscious individuals’ awareness.
Perhaps due to surprising conflicts of interest, “laymen” have lost trust in “experts.” Instead, we travel to our favorite blogs. We see how they live their lives, their motivation for pursuing a healthy lifestyle, as well as their conflicts of interest. The degree of transparency in the health blogosphere is greater than that in the medical industrial and big agricultural complexes. Us laymen would rather listen to someone selling snake-oil who is active on social media than the “expert” recommending Froot Loops for breakfast.
Among the list of blogs that dismissed the AHA’s claims, my favorite articles were by Food Babe, Dr. Mercola, and Medium. Dr. Mercola, in addition to Food Babe, exposed the AHA’s conflicts of interest, but also explained the biochemistry of saturated fats and how they may be mechanistically better than polyunsaturated fats for overall health, due to them being less prone to oxidative damage, a cause of inflammation (which is arguably the root cause of most chronic illnesses). What I liked about Medium’s post were the provocative questions, rather than the default approach of making bold assertions and backing them up with evidence. Question-asking is a great educational strategy, one that I will employ here, to stimulate you to make your own decision.
Unfortunately, some people in the influencer list I cited, Dr. Axe specifically, failed to debunk the AHA’s claims by citing research already addressed in the report. Perhaps this is why Snopes sided with the AHA, and disagreed with the notion that the they “cherry-picked” studies to support their view (claims made by Chris Kresser and Sustainable Dish). They maintained that the AHA does indeed have solid research showing that replacing saturated fats in the diet with polyunsaturated fats reduces atherosclerosis. To my delight, they proposed a challenge:
Any response must make a rational counterargument for why the studies they point to are superior to the ones the AHA used. They also must accurately describe the studies they cite.
Thank you, Snopes, for being a little open-minded. Open-mindedness is not listed as one of the steps to following the scientific method unfortunately, but it’s an important skill to cultivate in order to understand the truth, assuming that there is one single truth (a claim Terrence Mckenna, holographic universe proponents, and a Buddhist poem someone read to me recently would counter).
But alas, as long as there is someone out there who agrees with Snopes, I must address the research the AHA cited in order to let readers like you make an informed decision. In the influencer list I cited above, only Dr. Mercola, and Chris Kresser seemed to meet the above criteria, by explaining that the AHA’s studies are over five decades old, and simply aren’t up to par with the newer studies. They still did not discuss them, however. Just because a study is very old does not mean that its conclusions are less accurate than newer studies; in fact, this is a logical fallacy called appeal to novelty.
Let’s now briefly summarize the way the AHA seems to back up its claims, before getting to the studies.
The AHA’s Arguments Can Be Summed Up Quite Simply
The AHA makes three central claims throughout their report to support their viewpoints:
- Saturated fats (SFAs) increase LDL cholesterol, which is associated with an increased risk of atherosclerosis and cardiovascular mortality.
- Replacement of saturated fats with polyunsaturated fats (PUFAs) reduces atherosclerosis and cardiovascular mortality in primates and humans.
- Some studies found increased heart disease when polyunsaturated fats replaced saturated fats in the diet (Sydney Heart Study, and Minnesota Coronary Experiment), but these studies were flawed because the polyunsaturated fats included trans-fats from margarine.
Now in the above list, argument #2 is the most important, as the outcome (atherosclerosis and heart attacks) is always more important than risk factors (LDL cholesterol). If proven correct, it would be the strongest argument favoring reduction of saturated fat intake for prevention of the outcome. Let’s now discuss some nuances of risk factors versus outcomes, so we’re on the same page.
Risk Factors Versus End Outcomes
Risk factors for a disease can only give us a cursory glance at someone’s prospect of developing that disease. Risk factors can include biomarkers present in the bloodstream like LDL cholesterol, gene variants such as variations in the D2 dopamine receptor and risk of addiction (although that is not an accepted risk factor, but there are many studies on it), physical features (which would likely be a product of the gene variant, such as the phenotypic traits in Marfan’s syndrome), family history of having a disease, lifestyle factors such as smoking, and environmental factors such as exposure to pesticides.
A risk factor is not a guarantee you will have a disease or condition in the future. Elevated LDL cholesterol levels are not a guarantee you will have a heart attack. There are multiple other ways to predict risk of having a heart attack, beyond blood levels of LDL cholesterol, such as the Framingham Coronary Heart Disease Risk Score.
What really matters is the end outcome: diagnosed atherosclerosis and cardiovascular events. So if studies can show that reducing saturated fat intake and replacing it with another type of fat, or even carbohydrate, while keeping the total amount of fat, protein, and carbohydrate in the diet the same reduces the number of heart attacks and prevalence of atherosclerosis, there is a strong case for saturated fat consumption being a causative agent. Let’s look at the studies.
Replacement of Saturated Fats With Polyunsaturated Fats Reduces Atherosclerosis?
The four core trials supporting the AHA’s argument are:
- Dayton et al. Male Veterans Study (the only blinded controlled trial). 1969.
- Leren, P. The Oslo Diet-Heart Study. 1970.
- Medical Research Council. Controlled trial with soybean oil. 1968.
- Turpeinen, O. Finnish Mental Hospital Study. 1968, 1972, 1979, 1983. (This is the study we will thoroughly discuss. There are many publications by this research group, I will cite them as we discuss them below).
The three meta-analyses they cited to support their position were (as referenced in their report).
- Mozaffarian et al., Systematic Review & Meta-Analysis. 2010.
- Hooper et al., Cochrane Systematic Review and Meta-Analysis. 2015.
- Chowdhury et al., Systematic Review and Meta-Analysis. 2014.
We will talk about systematic reviews and meta-analyses in part II. Now, let’s discuss the Finnish Mental Hospital Study.
Methods of the Finnish Mental Hospital Study
This study took place over twelve years, divided into two six-year periods (1959 – 1971). Two hospitals, designated N and K, participated. This study used a cross-over design, where participants were first fed one diet, and then switched to the other diet. One diet was a control diet with over 50-60% of fats coming from the saturated type, and the other was a “serum cholesterol-lowering” diet, with 35-45% of fats coming from the polyunsaturated variety. During the first six-year period, participants in hospital N went on the cholesterol-lowering diet (SCL diet), and participants in hospital K stayed on a control diet (NORM diet). The results appeared in three separate publications (references 30, 36, and 37 in the AHA’s report). The first report in 1972 included cumulative data among all participants.
We will discuss below a report from 1979, which included the results for men only. Another report in 1983 shared the results in women; the results of that report will also be found below. There were a lot of people in the study, which gave it the most weight in the AHA’s analysis; almost 30,000 person-years of data in fact.
Below is what the diets looked like (units are in grams per person a day). They attempted to keep the diets the same except for the percentage of calories coming from fat. They created a “filled-milk” product that was rich in soybean oil to increase the concentration of polyunsaturated fats in the SCL diet. The consumption of saturated fat was increased by margarine, whole fat milk, butter, and other animal products. About margarine: in the NORM diet group, participants consumed “common” margarine, and in the SCL group, “soft” margarine.
Margarine is undoubtedly a curse word nowadays. It’s not as popular as it used to be, and unless you’ve been living near the center of the Earth, or in outer space, you know that margarine contains trans-fatty acids, which are universally linked to a higher cardiovascular and all-cause mortality in the systematic reviews that have been done on saturated fat consumption (including one of the reviews I cited already–next post I’ll discuss it more). Since I was born in the early 90s, after margarine seemed to have begun to lose its appeal, I had NO idea what the difference was between common and soft margarine.
According to a New York Times article from 1984, the “common” variety may contain more trans-fats (emphasis mine):
Overall, tub margarines have the best ratios, often 2 to 1 and even 2.5 to 1 of polyunsaturated to saturated fat. The reason is that they contain less hydrogenated oil than many stick margarines; hydrogenation, the process of hardening oil, turns some of the unsaturated fat to saturated.
The hydrogenation process is what creates partially-hydrogenated oils, which essentially are trans-fats. Even though fewer people consume margarine now than in 1984, trans-fats make their way into a variety of processed foods one can find in the aisles of a typical grocery store.
As you can see above, the NORM diet group consumed more of the common/stick/hard margarine, whereas the SCL group consumed soft/tub margarine. The difference was the most important in hospital K, where the SCL group consumed no soft margarine, and consumed 18 g a day of the common margarine.
Below, you will see the percentage of calories from fat in the diet by source. The SCL diet group obtained its fats from “filled” milk (skim milk with soybean oil added to it), and soft margarine mostly, whereas the NORM diet group obtained its fats from whole milk, butter, and meat mostly.
They also looked at the ratio of polyunsaturated fat to saturated fat in each diet. In the SCL diet, the ratio was about six times higher than in the NORM diet. This they hypothesized would reduce cardiovascular events.
Serum total cholesterol levels were lower in the SCL diet group in both hospitals. The picture here shows the total cholesterol levels in each hospital throughout the study. In hospital N, which started on the SCL diet, cholesterol was lower than in hospital K, which started on the NORM diet. After six years, this was switched, and the serum levels of cholesterol in the participants flip-flopped.
They also measured the adipose tissue concentration of linoleic acid, an omega-6 polyunsaturated fatty acid that is commonly found in vegetable oils like corn and soybean oil. They found that during the first six years, the concentration of this fatty acid decreased in participants on the NORM diet (hospital N), while it increased in participants from hospital K on the SCL diet.
In more detail below is their table of fatty acid concentrations in adipose tissue per hospital and diet. As you can see, linoleic acid and palmitic acid (a saturated fatty acid) followed predictably opposite patterns as participants switched diets. On the NORM diet, palmitic acid concentration is higher than when on the SCL diet, and vice versa for linoleic acid.
Lastly, this was a hospital with patients suffering from mental illness who were on psychotropic medications. For whatever reason, during the NORM diet period in hospital N, patients took over twice the dose of the medications they were on, including thioridazine and other phenothiazines (used for schizophrenia).
Now on to the most important part of the results…the researchers here assessed their outcomes in two ways. They looked for coronary deaths and electrocardiographic (ECG) changes indicative of coronary heart disease (CHD–same thing as CVD). The ECG changes they looked for included enlarged Q waves, atrioventricular block, left-bundle branch block, ST depressions, and other changes to Q and T waves. They broke them up into major, intermediate, and minor patterns. Major patters were more suggestive of abnormal changes to the heart, and were based on a set of criteria known as the Minnesota Code (from 1960, outdated now).
Coronary death was assessed by three of the researchers and was based on the following criteria: 1) sudden or unexpected death for which no cause could be determined; 2) slower death with objective diagnostic evidence of coronary heart disease (it’s the same thing as cardiovascular disease); or 3) death associated with autopsy findings indicative of CHD.
Finally, they conducted three analyses: a hard, moderate, and intermediate one. In the “hard” analysis, they only factored in coronary deaths, and excluded the presence of ECG changes. In the moderate analysis, the criteria was coronary deaths or major ECG changes. In the intermediate analysis, coronary deaths or major and intermediate ECG changes were tallied.
Below are the numbers.
In hospital K, 7 out of 213 participants died in the NORM diet group, and 2 out of 196 died in the SCL group. In hospital N, 5/265 died in the NORM diet group, and 4/248 died in the SCL diet. These findings failed to reach statistical significance.
In the intermediate analysis, findings did reach statistical significance in hospital N, where participants took twice as many doses of their psychotropic medications. The findings were not as significant in the intermediate analysis.
In women, there was no trend, although major and intermediate ECG changes increased in the NORM diet group.
The numbers from the above two reports are quite small, but in the report in 1972 showed a higher rate of coronary deaths in the NORM diet group, especially in men, for both hospitals (below).
The total number of coronary deaths is confusing to me personally. In their 1979 and 1983 reports, the numbers of deaths are quite low, and do not reach statistical significance. However in their original report, there were about half as many deaths in the SCL diet group. In addition, the presence of significantly more ECG abnormalities in the NORM diet group led the researchers to believe that replacement of saturated fat with polyunsaturated fat led to favorable cardiovascular consequences. This was further evidenced by a reduction in serum cholesterol levels, which at the time was regarded as the strongest risk factor for cardiovascular disease. It became the strongest piece of supporting evidence for the saturated fat-heart disease hypothesis.
However, the participants were on drugs, like phenothiazines which can affect many aspects of the ECG, including elements that they factored into their analysis. Thioridazine has been known to cause ECG changes since the 60s, such as a prolongation of the QTc interval. It’s also associated with sudden unexplained death. The other drugs they were on were tricyclic antidepressants, like imipramine and amitriptyline, which are also associated with sudden unexplained death.
In hospital N, participants took two and three doses of thiorizadine and other phenothiazines respectively, and also took twice as much TCAs while in the NORM diet. Perhaps this could explain why the highest degree of statistical significant was seen in this hospital. The researchers acknowledged this, and suggested that the drugs the participants were on were not known to cause any changes to the Q waves, and thus would not show up as a “major” ECG change. However, this seems to be false based on the citations above (click on “phenothiazines,” “thioridazine,” and “QTc interval” above for references).
This trend seemed to reverse in hospital K, and a similar pattern of ECG changes was observed in the NORM group, thus there are likely other factors at play. Nevertheless, we cannot conclude from this data that the diets alone resulted in the ECG changes in my opinion or the increased coronary deaths in hospital , as the drugs could have accounted for the worse outcomes in hospital N especially.
Let’s compare what the AHA said about this study in particular.
In each hospital, CHD events were lower during the times when the polyunsaturated fat diet was given. Results were similar in men and women.
Over both hospitals, it is true that there were less deaths in the SCL diet group, half as many in fact. But remember what they said about trans-fats? The AHA excluded two studies from their analysis that found higher cardiovascular mortality in the polyunsaturated fat groups because those diets included trans-fats. However in this case, they did not seem to account for increased trans-fat intake the NORM diet group consuming common margarine, or the increased dose of psychotropic medications in hospital N during the NORM diet period.
Thus, the AHA seems to have fell vulnerable to confirmation bias, and does indeed seem to have cherry-picked the results to support their hypothesis.
How Do the Diets Compare to a Balanced Diet?
I think the most important question that is left out of meta-analyses and systematic reviews on the topic are the foods people consumed. Science is concerned with reducing things to its parts to understand the details. Whole foods aren’t relevant in nutrition science most of the time. Instead, foods are seen as summations of macro- and micronutrients.
In theory, if the amount of saturated fat in the diet decreases, foods containing them decrease as well in the diet, so the whole food may be irrelevant. However, in the Finnish Mental Hospital Study, the percent of energy from fat coming directly from saturated fat ranged from 40-60%, which is on the higher side.
Beef tallow doesn’t even contain this much saturated fat (in fact oleic acid, a monounsaturated fatty acid in olive oil, is the most predominant fat in beef fat, from both the adipose tissue and intramuscular adipocytes, most of the time). People were forced to consume a similar diet on a daily basis for six years at a time. If this study showed anything at all, it was that drinking lots of milk and butter on a daily basis may increase mortality.
It’s not intuitive to eat like that. It’s not balanced either. Patients who live in a hospital, who likely don’t move around a lot, perhaps should not be consuming energy-dense foods like that on a regular basis. According to Ayurveda, this diet is very heavy and rich, and would predispose someone to cardiovascular problems. There is an aspect of common sense to this. (There are also studies on dairy consumption and cardiovascular mortality…this post will be updated with a link to that future post).
Thus when interpreting this study, you must ask yourself how relevant this is to your life. The high saturated fat condition was achieved by forcing participants to consume unbalanced amounts of saturated fat in the diet. I know that I would have to put in considerable effort for my diet to look like that.
In addition, trans-fat rich margarine constituted a significant portion of the saturated fats in the NORM diet group. So the group that consumed more milk, butter, margarine, and drugs that can cause sudden unexplained deaths and arrythmias experienced twice as many deaths than patients drinking soybean oil on a daily basis, not consuming trans-fats, and taking fewer drugs. Hm. Not sure if this applies to my own life and the foods I eat on a regular basis.
Some studies have found increased risk of cardiovascular mortality from replacement of saturated fat with polyunsaturated fat in the diet (Minnesota Coronary Experiment), despite the reduction in cholesterol seen with the diet.
Furthermore, meta-analyses on this topic do not agree with the AHA, including ones the AHA even cited, such as Chowdhury et al. which states in the abstract that the evidence does not support guidelines favoring replacement of saturated fats with polyunsaturated fats.
Most curious to me is that the studies that seem to favor polyunsaturated fat consumption are all coming from Circulation, which is the AHA’s journal. One is simply left to wonder why they so strongly favor polyunsaturated fats when the research does not agree with them, as discussed by others, and as will be addressed in the next post. Maybe it’s due to their ties with manufacturers of the oils, like Mazola. I’ll stick to eating real food, not over-consuming any one category of it, maintain variety, and exercise, rather than worry about saturated fat or consider replacing it it with soybean oil. Gross.