Recently, it has become a craze to demonize hydrogenated vegetable oils, also known as trans fats.
Indeed, the U.S. Food and Drug Administration (F.D.A.) is seeking to outright ban this class of substances, since they are, after all, “artery-clogging” and “a threat to public health.”
Trans fats are found in most pre-packaged foods, especially anything that contains oils. They are also the main constituent (shortening) of cake frosting (after sugar). And of course, there is margarine, which has become a popular beating stick (no pun intended) of the group of people who seek to vindicate butter and saturated fats (who feel – likely rightfully – that these latter items have been unduly vilified).
The F.D.A. decreed that trans fats should “no longer be ‘generally recognized as safe.'” Why is this so? As explained in a New York Times article:
Partially hydrogenated oils are cheaper than saturated animal fats like butter, and for years were thought to be healthier. They are formed when liquid oil is treated with hydrogen gas and made solid. They became popular in fried and baked goods and in margarine. Crisco, originally marketed in the beginning of the 20th century, was the archetype, although it now contains no trans fat.
But over the years, scientific evidence has shown they are dangerous because they raise the levels of so-called bad cholesterol and can lower the levels of good cholesterol.
Sounds like some awful stuff, right? Their solution:
Banning them completely could prevent 20,000 heart attacks and 7,000 deaths from heart disease each year, the F.D.A. said.
“This is the final slam dunk on the trans fat issue,” said Barry Popkin, a nutrition epidemiologist at the University of North Carolina, Chapel Hill.
So case closed, right?
Well, if you’ve been following along with this series, you’d know I’m going to say not so fast.
How do we know that trans fats are dangerous? As per usual, most of the evidence for the harm posed by trans fats comes from correlations seen in observational studies. As Wikipedia put it:
The major evidence for the effect of trans fat on CHD comes from the Nurses’ Health Study — a cohort study that has been following 120,000 female nurses since its inception in 1976. In this study, Hu and colleagues analyzed data from 900 coronary events from the study’s population during 14 years of followup. He determined that a nurse’s CHD risk roughly doubled (relative risk of 1.94, CI: 1.43 to 2.61) for each 2% increase in trans fat calories consumed (instead of carbohydrate calories). By contrast, it takes more than a 15% increase in saturated fat calories (instead of carbohydrate calories) to produce a similar increase in risk. “The replacement of saturated fat or trans unsaturated fat by cis (unhydrogenated) unsaturated fats was associated with larger reductions in risk than an isocaloric replacement by carbohydrates.” Hu also reports on the benefits of reducing trans fat consumption. Replacing 2% of food energy from trans fat with non-trans unsaturated fats more than halves the risk of CHD (53%). By comparison, replacing a larger 5% of food energy from saturated fat with non-trans unsaturated fats reduces the risk of CHD by 43%.
Another study considered deaths due to CHD, with consumption of trans fats being linked to an increase in mortality, and consumption of polyunsaturated fats being linked to a decrease in mortality.
Medical science, especially that part that gives rise to health prescriptions, unfortunately suffers from an over-reliance on observational studies. These studies only give you correlations. They are incapable of establishing the causal relationship between the phenomena studied. See, again, the words of Gary Taubes:
The shortcomings of observational studies are obvious and should not be controversial. These studies, regardless of their size or number, only indicate associations—providing hypothesis generating data—not causal relations. These hypotheses then have to be rigorously tested. This is the core of the scientific process. Without rigorous experimental tests, we know nothing meaningful about the cause of the disease states we’re studying or about the therapies that might work to ameliorate them. All we have are speculations.
Even more basic, the concern about the effects that trans fats have on cholesterol levels, may be misplaced. The relationship between cholesterol and illness (again, itself gleaned from correlational studies) is hardly as clear cut as it seems. This isn’t even mentioning the fact that statins, drugs used to lower high cholesterol levels, appear to be completely ineffective in reducing mortality in “healthy” subjects. (As well, they may have limited effectiveness for those who do have heart disease).
OK, at this point, you might be thinking that all I have done so far is call into question the evidence on which the assumed dangers of trans fats rests. It’s one thing to point out that the case is not quite as solid as we’d like. That still doesn’t demonstrate that the claim that trans fats are dangerous is wrong. Indeed, you’d be correct on this point. However, now allow me to do you one better:
These graphs show the mortality rates from cardiovascular disease (CVD) in the U.S. over time. As we can see, the present incidence of CVD mortality is considerably lower than it was in the middle of the 20th century.
Now here are some graphs on trans fat consumption (from here and here):
The total trans fat consumption, captured by the use of things margarine and shortening, doesn’t quite appear to follow CVD mortality. I’ve put in a separate margarine graph to show that it may appear to follow CVD mortality if looked at in isolation. But when shortening (which is largely trans fat) and oils (which are often hydrogenated) are considered, the pattern becomes highly questionable.
I did find one paper that tried to estimate the actual trans fat consumption over time: The estimates show great variability, including a marked disparity in the early 1990s – which is related to a shift in the measuring system used. The various methods used are described in the paper. Each has only limited accuracy in assessing population-wide average consumption for reasons described in the paper. As well, unfortunately, this doesn’t go far back enough to see if trans fat consumption tracks the mid-century peak in CVD mortality.
In short, despite the professed dangers of trans fats, one of the most basic signs of this danger is missing. If trans fats were primarily behind the incidence of heart disease, then we’d expect rates of CVD incidents to track (likely in a time-lagged fashion) the level of trans fat consumption. However, it’s not at all clear that we see this. However, our understanding of the true level of trans fat consumption is limited by the inherently poor reliability of the methods used to measure it.
At this point, I should point out the work of Fred Kummerow, who has studied the role of trans fat consumption in heart disease and claims that there is a causal link. However, most of Kummerow’s evidence comes from comparison of diseased patients with “healthy” patients (in a surgical setting), autopsies, animal studies, and limited lab cell/tissue analysis. In short, he has little that truly demonstrates a causal link between dietary trans fat consumption and disease. Blood and tissue analysis, of both living and dead human patients, while interesting, is limited in its ability to shed insight into causation. This is because detecting the presence of a substance in the blood or tissues doesn’t necessarily mean it got there through ingestion. That logic is how we got into the trap with cholesterol’s purported role with heart disease. The human body can manufacture a variety of substances. The extent that it does so is heavily under genetic influence. The presence of the chemicals seen in diseased patients could be symptom, not a cause, of the underlying disease.
So if dietary trans fat can’t explain heart disease, what does? The short answer is we don’t know. Contrary to what health experts might lead you to believe, fundamentally, we have no idea what causes heart disease. There are all sorts of hypotheses, but, as we see, all face significant trouble when subjected to scrutiny. Indeed, see here and here for some additional hypotheses and scrutiny of some popular ones.
The former source touches on a very key point: the failure of randomized controlled trials to affect the incidence of heart disease. We’ve already seen the failure of the Look AHEAD trial. Indeed, a slew of clinical trials testing various prescriptions of have netted negative results:
The MR-FIT trial in the USA was the most determined effort to prove the case. This was a massive study in which over 350,000 men at high risk of heart disease were recruited. In one set of participants, cholesterol consumption was cut by 42 percent, saturated fat consumption by 28 percent and total calories by 21 percent. This should have made a noticeable dent in heart disease rates.
But nothing happened. The originators of the MR-FIT trials refer to the results as “disappointing,” and say in their conclusions: “The overall results do not show a beneficial effect on Coronary Heart Disease or total mortality from this multifactor intervention.”
It is not as if this was one negative to set against a whole series of positive trials. In 1998, the Danish doctor Uffe Ravnskov looked at a broader selection of trials: “The crucial test is the controlled, randomised trial. Eight such trials using diet as the only treatment has been performed but neither the number of fatal or non-fatal heart attacks was reduced.” As Ravnskov makes clear, no trial has ever demonstrated benefits from reducing dietary saturated fat. At this point most people might think it was time to pull the plug.
Indeed, the one “successful” intervention, the Spanish “Mediterranean diet” trial, was rebuked by Stephan Guyenet for being improperly controlled.
That is even going beyond the issue of the generalizability of results from the Spanish population to non-Spaniards.
This brings us to one thing we do know about heart disease: it is (like all things) heavily influenced by genetics. Indeed, a recent GCTA (Genome-wide Complex Trait Analysis) study that looked at 38,167 individuals from Iceland found that the broad-sense heritability of coronary artery disease is at least 55% (at least in the Icelandic population, that is, of course). That is, heredity is the single largest known factor involved your chances of developing heart disease. (As an aside, the study authors claim that their data show that the shared environment is important source of the variance in these traits, based on the differing heritability measures of more distant relatives. However, if you look closely at their confidence intervals, you’ll see that this conclusion is not warranted.)
In fact, a quick return to the age-adjusted incidence of CVD mortality in the U.S. over time shows something interesting:
Assuming that these numbers are trustworthy, (especially for the early 20th century when reliability of cause of death declarations come into question), it seems the total CVD mortality has in fact remained fairly flat throughout for much of the 20th century, only to decline a few decades ago. If we consider “diseases of the heart” alone, we see an interesting rise and fall. What’s up with that?
Allow me to present something else:
These are the top causes of death in the United States since 1900 (from Slate). As we see, prior to the 1930s, the biggest killers were infectious diseases. The increase in modern human lifespan is heavily due to three simple things: improved sanitation, vaccines, and antibiotics. Notice how the incidence of heart disease significantly rises as deaths from infectious diseases fall. This means that the mid-century rise in deaths from heart disease could simply be a result of the fact that more people were living long enough to die from heart disease. Even the age-adjusted rates don’t full account for this factor, because people who lived longer could have systematically more vulnerable to heart disease (heart disease is heritable). The increasing rates of deaths from cancer likely have a similar explanation, and illustrates this process.
It is possible that there was never much we could do about heart disease, aside from getting at better at keeping people alive who’ve already had heart attacks. Unfortunately, it is possible that some people’s bodies are designed in such a way that they develop arterial blockages. The superficial correlations with “lifestyle factors” gleamed from observational studies could be a due to individuals with such a genetic propensity also happening to live certain ways. These factors may have nothing to do with one another causally. Even depression (and by extension, stress), which has been linked to heart disease, may be so linked because of common genetic factors.
With all this talk of genetics, there might be one thing in the environment we might be able to link to heart disease. That thing is smoking (from here):
Smoking rates over time track CVD death rates in the U.S. As well, see this (from Wikipedia):
There appears to be a strong spatial and temporal association between smoking and heart disease. However, that association appears to be primarily due to rates of these things in the Slavic world and the Middle East.
Of course, it’s worth noting that I’ve previously shown that CVD mortality rates can be linked to climate, at least in Europe.
Over the years we’ve been fed all sorts of health advice, different people telling us – preaching to us – what we should do to stay healthy and live as long as possible. It’s becoming increasingly clear that much of it – if not all of it – is bullshit. Indeed, as a result, the health beliefs that most people have today are basically religion. Everyone has their own ideas on how to stay healthy, and little of it is based on any good science (as commenter Sisyphean pointed out). It’s not like we can completely blame folks for this; the advice they’ve been receiving has been confusing and often contradictory, as exemplified by the love-hate-love relationship with butter and saturated fats, and now with trans fats. However, at this point, while I’d agree that, if you want to live as long as you can, it’s good advice to not eat rat poison or go for a stroll into the middle of a busy highway, I can’t vouch for the rest of it. The best advice might just be eat, drink, and be merry – at best, doing what seems to work for you. Least of all, don’t buy what the health experts are pushing until they show you the randomized controlled trials that back them up.
Once again, I’m using Tangerine Dream’s “Cloudburst Flight”, which to me has become a general theme for this series:
On the question of animal fat my attitude — which annoys my wife — is that anything that tastes that good can’t be bad for you.
Climate sure looks like a strong candidate. Especially interesting is the fact that northern countries (in Europe) that have an oceanic climate have relatively few deaths from heart disease. If you look at deaths per 100K inhabitants according to Eurostat we have, Norway 66, Sweden 84 and Finland 120. Those are big difference on genetically and culturally similar populations but they follow the degree of oceanic climate. Maybe it has something to do with really low temperatures?
“If you look at deaths per 100K inhabitants according to Eurostat we have, Norway 66, Sweden 84 and Finland 120. Those are big difference on genetically and culturally similar populations but they follow the degree of oceanic climate.”
Also follows ‘Pure alcohol consumption among adults (age 15+) in litres per capita per year’:-
Norway 7.81; Sweden 10.30; Finland 12.52;
(World Health Organisation recent data).
But as Jayman would point out it’s all correlational. Clearly there are many factors at play.
I think that the people who consume cheap low quality foods which contain ‘trans fats’ would probably be more likely to be from lower socio-economic classes [at least in UK where I live]; those types of people are also far more likely to be smokers too. So Jayman makes a key point about the worthlessness of observational/correlational studies [one that I’ve only recently started to take note of].
The problem with alcohol consumption.
Of course, different peoples are differently adapted to alcohol, so that may have something to do with it. To be honest, I don’t suspect that it does.
Precisely. Actually, I was going to include some studies on trans fat consumption by socioeconomic class. As you’d expect, the lower classes consume more…. :
Maybe it does have something to do with low temperatures. Indeed, the cardiovascular death rate is higher in Quebec than it is in France. Maybe cold isn’t so good for you? But it just as easily be evolution – differing climatic adaptations – even across Scandinavia (the Finns are actually quite distinct from other Europeans genetically) may have affected likelihood of developing CVD. That’s the guess I’m currently running with.
That said, on the direct environmental impact of climate, CVD death rates are much lower in Australia than in any other Anglo country…
True a lot of factors to consider. Eastern Europeans drink and smoke and have lower IQs than the rest of Europe. It’s hard to separate the variables. And the world is changing; Mexico is now the most obese country. Poor people around the world can now access junk food and cigarettes. This trend should increase the importance of IQ and traits like impulsivity and conscientiousness.
I find it hard to get excited about trans fats. Some people are claiming that they are bad for you. I distinctly remember when my mother gave me butter because it was better for me than margarine. Then my wife made us eat margarine because it was better for us. Now I’m back to eating butter – since the divorce. My relationship with the women in my life seems to correlate with my fat consumption.
I read Linus Pauling’s little book on Vitamin C and was blown away. What an argument! No wonder that guy had two more Nobel prizes than I had. He was brilliant but not as it happened correct. But then I placed my faith in Vitamin E. It cured both cancer and heart disease – everyone said so. That is until someone actually ran a real study and not just observe correlations,
Now of course we have regressed in the alphabet. My doctor and all the wise men of earth now recommend Vitamin D. OK , I take Vitamin D and I have to admit I like it better than Vitamin E – the pills are smaller.
I try to eat lamb chops and asparagus only. That way I can ignore trans fats discussions.
This 2009 paper http://www.nature.com/ejcn/journal/v63/n2s/abs/1602973a.html
claims: “Five retrospective case–control studies and four prospective cohort studies demonstrated positive associations between TFA consumption and CHD events.”
and concludes: “Controlled trials and observational studies provide concordant evidence that consumption of TFA from partially hydrogenated oils adversely affects multiple cardiovascular risk factors and contributes significantly to increased risk of CHD events.”
So much debate has been going on about Trans fat and its effect on people’s health. I would say that there are other more risk behaviors that might contribute more to the upsurge of heart condition as well such as smoking. intergratedmedicine.com
The randomized controlled trials they mentioned were few, small, and of short duration. None looked at actual adverse health events. The best they looked at was LDL vs. HDL, finding trans fat consumption seems to raise the former and not the latter (and the short duration didn’t allow us to establish if this result wasn’t only temporary). However, since the role that cholesterol plays in cardiovascular health is far from from fully understood, even if this LDL effect is real, it’s unclear if it’s of concern.
Thanks for pointing me in that direction!
At the start of the article you present 2 graphs with butter consumption in them. One shows a large, almost monotonic increase over the reported time scale. The second shows a shallow decline over the same period. The absolute numbers of each graph for butter are also wildly different. Why is this?
“that guy had two more Nobel prizes than I had”: true, but one was the pishy, politicised Peace Prize. If, like any rational man, you count that as minus one, and add it to his well-earned prize in Chemistry, you can see that effectively he had the same number of Nobel Prizes as you.