“The
results show that the omega-6 linoleic acid group had a higher risk of death
from all causes, as well as from cardiovascular disease and coronary heart
disease, compared with the control group.”
We should be most skeptical of the research that best confirms our own beliefs. It may not contain an inaccurate result – in the case of the Sidney Diet Heart Study, the result is consistent with the totality of the evidence that is appearing from other sources – but it may not be as convincing to others as it appears to us. We should always be alert to the desirability of designing experiments to produce results that, one way or the other, will convince any reasonable person. In diet studies this may well be unattainable, and usually the “expert evidence” of clinical studies has to be considered in the context of circumstantial and eyewitness evidence from other sources, but it is the scientific ideal.
![]()
We should be most skeptical of the research that best confirms our own beliefs. It may not contain an inaccurate result – in the case of the Sidney Diet Heart Study, the result is consistent with the totality of the evidence that is appearing from other sources – but it may not be as convincing to others as it appears to us. We should always be alert to the desirability of designing experiments to produce results that, one way or the other, will convince any reasonable person. In diet studies this may well be unattainable, and usually the “expert evidence” of clinical studies has to be considered in the context of circumstantial and eyewitness evidence from other sources, but it is the scientific ideal.
On the BMJ website where the study appears, doctors can
send in their responses, and some of these
responses give cogent reasons why a reasonable person, especially one not
engaged with the Paleo discussion of seed oils over the past few years, might
remain unconvinced by the latest revelations.
Perhaps the best response comes from Professor Jean
Gutierrez, assistant professor of exercise science at Washington University.
She points out that one of the products used in the trial probably contained high
levels of trans fats:
Participants in the intervention group consumed “Miracle” Margarine, a product based on safflower oil. Hydrogenation of safflower oil itself creates a grainy product low in linoleic acid, so high-linoleic safflower oil margarine products were created by blending liquid safflower oil with another hydrogenated oil stock (3). Miracle Margarine used in the original study was either low in linoleic acid (due to hydrogenation of the safflower oil itself) or the oil was blended with another commercially hydrogenated fat to create a plastic margarine product. An investigation by Bernfeld, Homburger, & Kelley, published in 1962, indicated that the fatty acid composition of most margarines of the time were about 50-60% 18:1 monounsaturated fats (including oleic and trans isomers) and about 20-30% 18:2 linoleic acid, even in those products having high-PUFA claims on the label (4). None of the 22 margarines studied had a majority of fatty acids coming from PUFA. Another report from the same time period indicates that commercially produced hydrogenated fats, like those added to safflower oil to make margarine, were generally composed of about 25-40% trans fats (5). Fatty acid composition of margarines in the 1960s investigation were not comparable to liquid vegetable oil, despite package claims. The only reference supporting the healthful content of Miracle Margarine is a very general press release from the company who made the product (6). It is probable that Miracle Margarine had significant trans fatty acid content.
This is an obvious confounder, hard to dispute on science grounds. Morally it might not make much difference; if advice to reduce saturated fat and increase PUFA consumption lead to the use of such dodgy products such as Miracle Margarine, as it did, then it still stands condemned historically. Have there been studies where oil but not margarine was the intervention? Was the Rose Corn Oil Study really just a study of corn oil? Comments section please.
In the section "other dietary considerations" it appears that the "prudent diet" intervention group avoided the ordinary margarines allowed in the diet of the controls. We will probably never know which group ate most trans fats. The fact that the intervention group had low cholesterol and trans fats are supposed to elevate LDL cholesterol does seem to indicate that any harm trans fats may have done was not due to "lipid hypothesis" mechanisms.
Participants in the intervention group consumed “Miracle” Margarine, a product based on safflower oil. Hydrogenation of safflower oil itself creates a grainy product low in linoleic acid, so high-linoleic safflower oil margarine products were created by blending liquid safflower oil with another hydrogenated oil stock (3). Miracle Margarine used in the original study was either low in linoleic acid (due to hydrogenation of the safflower oil itself) or the oil was blended with another commercially hydrogenated fat to create a plastic margarine product. An investigation by Bernfeld, Homburger, & Kelley, published in 1962, indicated that the fatty acid composition of most margarines of the time were about 50-60% 18:1 monounsaturated fats (including oleic and trans isomers) and about 20-30% 18:2 linoleic acid, even in those products having high-PUFA claims on the label (4). None of the 22 margarines studied had a majority of fatty acids coming from PUFA. Another report from the same time period indicates that commercially produced hydrogenated fats, like those added to safflower oil to make margarine, were generally composed of about 25-40% trans fats (5). Fatty acid composition of margarines in the 1960s investigation were not comparable to liquid vegetable oil, despite package claims. The only reference supporting the healthful content of Miracle Margarine is a very general press release from the company who made the product (6). It is probable that Miracle Margarine had significant trans fatty acid content.
This is an obvious confounder, hard to dispute on science grounds. Morally it might not make much difference; if advice to reduce saturated fat and increase PUFA consumption lead to the use of such dodgy products such as Miracle Margarine, as it did, then it still stands condemned historically. Have there been studies where oil but not margarine was the intervention? Was the Rose Corn Oil Study really just a study of corn oil? Comments section please.
In the section "other dietary considerations" it appears that the "prudent diet" intervention group avoided the ordinary margarines allowed in the diet of the controls. We will probably never know which group ate most trans fats. The fact that the intervention group had low cholesterol and trans fats are supposed to elevate LDL cholesterol does seem to indicate that any harm trans fats may have done was not due to "lipid hypothesis" mechanisms.
Having
established her scientific objectivity by drawing attention to the study’s biggest
flaw, Professor Gutierrez strikes a home run with her last paragraph:
In addition to increasing PUFA intake, participants in the intervention group reported reduced dietary saturated fat, cholesterol, and calorie intake from baseline. A negative energy balance was verified with a slight mean drop in BMI. As expected, circulating total cholesterol and triglycerides were reduced in the intervention group, but mortality outcomes were not improved consequent to these circulating lipid and anthropometric changes, which is unexpected and interesting. The more important question arising from this study may be why a dietary intervention that improved all of these commonly used surrogate end points did not reduce all-cause mortality?
In addition to increasing PUFA intake, participants in the intervention group reported reduced dietary saturated fat, cholesterol, and calorie intake from baseline. A negative energy balance was verified with a slight mean drop in BMI. As expected, circulating total cholesterol and triglycerides were reduced in the intervention group, but mortality outcomes were not improved consequent to these circulating lipid and anthropometric changes, which is unexpected and interesting. The more important question arising from this study may be why a dietary intervention that improved all of these commonly used surrogate end points did not reduce all-cause mortality?
Why indeed, the world wonders.
Surgeon Basil D Fadipe from Dominica has an explanation for the result:
In a nutshell, the study shows LA is a 'substrate' for oxidative stress convertible to toxic derivatives, mechanistically adverse to the cardiovascular-coronary integrity; An otherwise good dietary item, LA's bad company (smoke/alcohol) wrecks its good potential.
This view
has some merit – LA plus alcohol is definitely destructive to the liver via
these mechanisms, and there are clearly links between high alcohol consumption and the negative effect of LA in the study (though it doesn't seem that abstinence removes the risk), but it does seem like special pleading. People will be exposed to
oxidative stress in various ways, not all easily predicted or avoided. Why
consume a dietary item that magnifies this insult in quantities greater than
those required for essential functioning?
UK GP William
K. Neville sums up the Paleo view;
Linoleic acid is an omega 6 fatty acid. It causes weight gain and inflammation. Omega 3 fats have the opposite effect. The ideal ratio of omega 6 to omega 3 in the diet is about 2:1 [9]. Modern diets have a ratio of 20:1 due to the false belief that linoleic acid is good for us. In pre-historic times weight gain from eating omega 6 fats in nuts and seeds, which were plentiful in the autumn, was an advantage to prepare humans for winter [10]. Meat from grass fed animals has a good amount of omega 3. But 99% of farmed animals are fattened on grain products such as brewery waste prior to slaughter which lowers their omega 3 levels to zero and the meat contains only omega 6. Similarly, farmed fish has reduced omega 3 levels due to being fed on grains.
Linoleic acid is an omega 6 fatty acid. It causes weight gain and inflammation. Omega 3 fats have the opposite effect. The ideal ratio of omega 6 to omega 3 in the diet is about 2:1 [9]. Modern diets have a ratio of 20:1 due to the false belief that linoleic acid is good for us. In pre-historic times weight gain from eating omega 6 fats in nuts and seeds, which were plentiful in the autumn, was an advantage to prepare humans for winter [10]. Meat from grass fed animals has a good amount of omega 3. But 99% of farmed animals are fattened on grain products such as brewery waste prior to slaughter which lowers their omega 3 levels to zero and the meat contains only omega 6. Similarly, farmed fish has reduced omega 3 levels due to being fed on grains.
Heart disease was rare before the introduction of industrial
seed oils at the start of the 20th century [11]. Four meta-analyses recently
prove saturated fat to be harmless [12]. The lesson of this article is that it
is time we stopped demonising the really healthy fats such as butter, lard,
beef dripping, coconut oil and palm oil [13, 14].
He makes
more of the case against dairy than is perhaps justified by the evidence to
date when he says:
The most important oxidising agent of LDL particles is the peptide BCM-7 produced by the digestion of A1 milk. The level of A1 milk consumption in countries is directly proportional to the rate of heart disease. Milk in the UK is mostly A1. Milk in Africa is mostly A2. France is in between. A few supermarkets have recently begun to sell A2 milk which has the potential to prevent many diseases such as type 1 diabetes and heart disease [6, 7, 8].
The most important oxidising agent of LDL particles is the peptide BCM-7 produced by the digestion of A1 milk. The level of A1 milk consumption in countries is directly proportional to the rate of heart disease. Milk in the UK is mostly A1. Milk in Africa is mostly A2. France is in between. A few supermarkets have recently begun to sell A2 milk which has the potential to prevent many diseases such as type 1 diabetes and heart disease [6, 7, 8].
All I can
say is that A2 milk tastes better, but is just as allergenic for my purposes. Milk
is a complex food and there is no one factor that applies to everyone, but the
BCM7 case outlined in “The Devil in the Milk” deserves study. I expect there
are many other things that correlate with heart disease just as well as A1 milk
does. Consider though, that when you eat meat, you eat from one or maybe two
animals (unless it’s processed sausage meat). When you drink milk, the milk
from large numbers of animals has been combined, each with its distinctive
immunological factors. Our milk-drinking ancestors, if we had them, probably
drank from one or two cows at a time. My granddad used to say that one secret
of health was not to mix one’s drinks.
American
Physician Megan I. Maurer makes a valid point that will also appeal to Paleo dieters:
I would like to point out that the source of the omega-6 fatty acids in this case was from safflower oil and safflower oil margarines which, despite being high in omega-6 fatty acids are still very calorically dense and in my opinion are not representative of what you might see had they used a whole food course of omega-6 fatty acids, such as sunflower kernels or walnuts. It has been consistently shown that a whole-food plant-based diet offers great cardiovascular protection. I think a study showing increased intake of omega-6 fatty acids in their whole form vs. saturated fats would have been more telling than just replacing it with oil.
Of course a diet of walnuts (an omega-3 nut, in fact) and sunflower seeds will introduce other confounders. But a worthwhile experiment would be to replace fatty meats with nuts and seeds, and butter and cooking fats with spreads and oils. The difference between intrinsic and extrinsic fats may be as important a confounder as the difference between intrinsic and extrinsic sugars.
![]()
The question is, more studies are needed, but are they ethically justified? It’s easy to test a dietary change when the evidence one has seen and the theories that seem to make most sense suggest it’ll be beneficial. Is it right to continue to experiment in this way as the evidence of harm accrues?
You might suspect it was the trans fats in the margarine that caused those extra deaths, but would you be prepared to re-run the study without them to find out?
I would like to point out that the source of the omega-6 fatty acids in this case was from safflower oil and safflower oil margarines which, despite being high in omega-6 fatty acids are still very calorically dense and in my opinion are not representative of what you might see had they used a whole food course of omega-6 fatty acids, such as sunflower kernels or walnuts. It has been consistently shown that a whole-food plant-based diet offers great cardiovascular protection. I think a study showing increased intake of omega-6 fatty acids in their whole form vs. saturated fats would have been more telling than just replacing it with oil.
Of course a diet of walnuts (an omega-3 nut, in fact) and sunflower seeds will introduce other confounders. But a worthwhile experiment would be to replace fatty meats with nuts and seeds, and butter and cooking fats with spreads and oils. The difference between intrinsic and extrinsic fats may be as important a confounder as the difference between intrinsic and extrinsic sugars.
The question is, more studies are needed, but are they ethically justified? It’s easy to test a dietary change when the evidence one has seen and the theories that seem to make most sense suggest it’ll be beneficial. Is it right to continue to experiment in this way as the evidence of harm accrues?
You might suspect it was the trans fats in the margarine that caused those extra deaths, but would you be prepared to re-run the study without them to find out?
However, the
fact is that the experiment has overtaken our society and most of us have been
enrolled in it, despite the actual results from the Sidney Diet Heart Study
proper having been “lost” since 1973.
If PUFAS do
have some potential for good instead of evil in certain cases, can we avoid
discarding the baby with the bathwater? In 2009 Stephan Guyenet analysed a 1994
study that compared the so-called “prudent diet” used in Sydney with a diet
that supplied limited omega-6 in the context of adequate omega-3; the Lyon
Diet Heart Trial. He writes:
Here's where it gets interesting. The intervention group ate three times as much omega-3 alpha-linolenic acid as the control group, and 32% less omega-6 linoleic acid. The ratio was 20 : 1 linoleic acid : alpha-linoleic acid in the control group, and 4.4 : 1 in the intervention group. This was due to the combination of a low-fat diet and the canola oil goop they were provided free of charge.
Here's where it gets interesting. The intervention group ate three times as much omega-3 alpha-linolenic acid as the control group, and 32% less omega-6 linoleic acid. The ratio was 20 : 1 linoleic acid : alpha-linoleic acid in the control group, and 4.4 : 1 in the intervention group. This was due to the combination of a low-fat diet and the canola oil goop they were provided free of charge.
But it gets even better. The
intervention group reduced their omega-6 linoleic acid intake to 3.6% of
calories, below the critical threshold of 4%. As I described in my recent post on eicosanoid signaling, reducing linoleic
acid to below 4% of calories inhibits inflammation, while increasing it more
after it has already exceeded 4% has very little effect if omega-3 is kept
low*. This is a very important point: the intervention group didn't
just increase omega-3. They decreased omega-6 to below 4% of calories. That's what sets the Lyon
Diet-Heart trial apart from all the other failed diet trials.
After five years on their respective diets, 3.4% of the control (prudent diet) group and 1.3% of the intervention ("Mediterranean") group had died, a 70% reduction in deaths. Cardiovascular deaths were reduced by 76%. Stroke, angina, pulmonary embolism and heart failure were also much lower in the intervention group. A stunning victory for this Mediterranean-inspired diet, and a crushing defeat for the prudent diet!
There's a little gem buried in this study that I believe is the other reason it didn't get accepted to the New England Journal of Medicine: there was no difference in total cholesterol or LDL values between the control and experimental groups. The American scientific consensus was so cholesterol-centric that it couldn't accept the possibility that an intervention had reduced heart attack mortality without reducing LDL. The paper was accepted to the British journal The Lancet, another well-respected medical journal.
After five years on their respective diets, 3.4% of the control (prudent diet) group and 1.3% of the intervention ("Mediterranean") group had died, a 70% reduction in deaths. Cardiovascular deaths were reduced by 76%. Stroke, angina, pulmonary embolism and heart failure were also much lower in the intervention group. A stunning victory for this Mediterranean-inspired diet, and a crushing defeat for the prudent diet!
There's a little gem buried in this study that I believe is the other reason it didn't get accepted to the New England Journal of Medicine: there was no difference in total cholesterol or LDL values between the control and experimental groups. The American scientific consensus was so cholesterol-centric that it couldn't accept the possibility that an intervention had reduced heart attack mortality without reducing LDL. The paper was accepted to the British journal The Lancet, another well-respected medical journal.
[implications of the study here]
So
in both the Sydney study and the Lyon trial, total cholesterol and LDL cholesterol
had nothing to do with whether participants lived, or died of heart disease. Of
course the Lyons “Mediterranean” diet only outperformed the “prudent” diet; the
controls in the Sidney Diet Heart Study already seem to have done that by eating
as people normally did in Sidney between 1966 and 1973.
The last word goes to Professor Dhastagir S. Sherrif of Benghazi University, Lybia:
Polyunsaturated fatty acids (PUFA) are essential fatty acids to be supplied in the diet. These are important for membrane function, producing eicosanoids; the endocannabinoids, the lipoxins and resolvins form lipid rafts for cellular signaling, act on DNA, activating or inhibiting transcription factors such as NF-κB - playing a vital role in physiological functions of the body. Yet there needs to be a balance between its intake and the form of PUFA taken in the diet.
The last word goes to Professor Dhastagir S. Sherrif of Benghazi University, Lybia:
Polyunsaturated fatty acids (PUFA) are essential fatty acids to be supplied in the diet. These are important for membrane function, producing eicosanoids; the endocannabinoids, the lipoxins and resolvins form lipid rafts for cellular signaling, act on DNA, activating or inhibiting transcription factors such as NF-κB - playing a vital role in physiological functions of the body. Yet there needs to be a balance between its intake and the form of PUFA taken in the diet.
Being polyunsaturated, they generate free radicals, cause lipid
peroxidation and damage mitochondrial function. It is suggested that the intake
of PUFA must be accompanied by adequate amounts of intake of anti-oxidants such
as vitamin E. Human body and its metabolism cannot be viewed as parts but as
whole body metabolism. Whole body metabolism needs to be viewed with the
interplay of internal and external factors that regulate and maintain
homeostasis. Extracellular factors including the dietary constituents need to
be balanced with the internal physiological milieu unique to every individual.
We need to remember the Daedalus effect: for every remedy there is an adverse
side effect. How we balance them is the duty of true science that will help
promote health.
(photos by Hayley Theyers (c) 2013)