Objections, 3

Nullis in Verba

–Motto of the Royal Society of London

In Objections 2 we started talking about fat and explored just a little bit of the history behind the rise and dominance of the lipid hypothesis–that fat is the cause of obesity and health problems.

Now let’s talk a bit about the science.

If you are a literate person, you’ve probably been bombarded your whole life by articles and reports that saturated fat is going to kill you. In fact it’s hard now to read an article without seeing the term “artery-clogging saturated fat,” as if that were the full scientific name for butter.

I’m going to start with a little logic here and then move on to scientific studies.

In what sense is any fat “artery-clogging”?

My favorite illustration of the deadly dangers of delicious fat is a British public health video that shows a hapless person taking a large jug of liquid fat from the refrigerator and pouring it into the room-temperature sink. It then shows how the fat solidifies and and clogs the pipe.

Imagine what it will do to your arteries, we are told grimly.

I don’t know where to begin.

Anyone who has ever owned a bit of lard knows exactly what happens to it in a refrigerator. Anyone?

Now don’t all raise your hands at once.

That’s right–it turns solid. And even if our butter was liquid, what happens to it when it goes from a cooler place-like the fridge–to a warmer place–like the kitchen sink pipe?

That’s right…as anyone who has kept the butter in the trunk of the car a little long knows, the application of heat causes fat to melt. And at the 98.6 degrees you’re probably currently running, you’re warm enough to keep any fat liquid.

If, of course, your stomach worked like an IV and what you ate just went directly into your bloodstream. Which it doesn’t. Which is such an overly simplistic view of human physiology as to be unworthy of any further comment.

So let us move on to the supposed mountain of evidence that proves saturated fat will kill you.

(I’ll give you a hint. There is none.)

That’s right. There’s not one study out there that proves the lipid hypothesis.

Here’s a quote that ought to give you some pause. Before I give it, a little background on the author. His name is George Mann, and he was a medical doctor and doctor of science who co-chaired the Framingham Heart Study. This study was done specifically with the hope of proving the lipid hypothesis. It observed most of the population of Framingham, MA–chosen because it was “reasonably typical.” Every person was examined initially, then reexamined every two years. In fact, the Framingham study is still following the initial participants (from 1948), and two subsequent generations to this day.

The co-founder of this study said this:

The…notion that saturated fats and cholesterol cause heart disease – is the greatest scientific deception of our times…The public is being deceived by the greatest health scam of the century.

Fun, huh? Mann left the study in early 60s because the National Institute for Health refused to allow publication of all the data once they realized it wasn’t proving what they wanted it to prove.

So what kinds of things did this study find out?

Well, at first their data seemed to support the lipid hypothesis and everyone got very excited. But as they looked closer and longer, it didn’t. That’s what happens when your hypothesis is wrong: the more you study it, the less the data supports your theory.

That information was compiled but left unpublished. You can find it only if you really dig, which is what several scientists and doctors have done since. They found, among other things, that ss the participants aged, those who had heart disease were more likely to have low cholesterol than high. They could find no connection whatsoever between cholesterol and heart disease in women over 50, and only a very sketchy one for younger women.

They also ran into trouble when they started examining diets. Turns out that when they compared the diets of men with very high and very low cholesterol, there was no difference in their fat consumption.

Oh dear.

So what else did they find?

Well, they found that the more calories a man ate, the lower his cholesterol was.

That can’t be right. It must be that those men exercised a lot, right? Burnt up all those calories? Except they checked that and found that when examining men with the same activity levels, it still held true that the more they ate the lower their cholesterol.

They found the exact same thing with dietary fat: the more a man ate, the lower his cholesterol. For women it made no noticeable difference. They also, despite their best efforts, couldn’t find any relationship between dietary cholesterol intake and actual cholesterol levels.

In other words, eating eggs and butter wasn’t giving people high cholesterol. You can find a good analysis of the data here.

In fact as Gary Taubes points out in his book, Good Calories, Bad Calories, there is not one study within a single population that can even find a correlation (let alone any proof of connection) between fat and cholesterol or fat and heart disease. You can only find a correlation if you start comparing populations that are different in many respects–and that’s a problem because that suggests fat isn’t the factor causing the heart problems.

That’s all that correlations and associations can really do for you. They can point out where a valid hypothesis might lie, but they can’t prove it. They can, however, show you that your hypothesis has to be dead wrong. Allow me to illustrate.

Say you hypothesize that all dogs are white because you live in Beijing and there are a lot of little Pekes running about. Whenever you see a white dog, you have reason to keep investigating the possibility that your hypothesis is true. The presences of those Pekes can’t prove your hypothesis true, but at least you have a reason to keep researching.

But the moment you see a brown dog, your whole hypothesis dies a very sudden death.

And that’s where the Framingham study can help us. It cannot prove the lipid hypothesis. However, if we found that the data from the study was supporting the hypothesis, we would have reason to keep researching it. But the Framingham study information doesn’t support the hypothesis. It contradicts it. We’re awash in brown and black and tan and roan and spotted dogs, and even some poodles with pink dyed ears.

When you’ve got hundreds of people, and some are eating tons of fat and some are eating low-fat, but all of them have high cholesterol, there’s only one thing you can be sure of: fat is not what’s doing it.

When you have some people with high cholesterol getting heart disease and some people with low cholesterol getting heart disease and you can’t find a link: it means cholesterol isn’t causing their heart disease.

The problem with those articles about fat you read on Yahoo “news” is that those are reporting nothing more than associations and correlations, based on notoriously unreliable food surveys.

Why are food surveys unreliable? Because they are a hundred pages long and are full of questions like this:

Between October and December of 2001, please estimate the number of 3oz portions of 75/25 lean to fat ground beef you consumed. ________

Can you answer that question?

People who fill them out lie, get confused, can’t remember, and sometimes just start making things up because they’re getting dizzy from all the questions.

And then there’s the problem of compliance. People don’t act normally when you watch them too closely, especially over short periods of time. There’s the placebo effect. And finally there’s the most serious problem of association and correlation being mistaken for proof. And then your bias starts showing.

Not long ago a surprising news headline from the British Medical Journal reported, and I quote:

New research shows that men who had had a heart attack were more likely to die from coronary heart disease when they replaced saturated fats with polyunsaturated fat in margarine.

The end of the article said this:

However, experts have been quick to say that we shouldn’t use these findings to abandon vegetable oils found in margarine in favour of butter and lard.

And here’s what I mean about bias. If that study had found the opposite correlation (and that’s all they found; a correlation) we would have seen this headline:


But the correlation did not support what they already hold to be true. It supported the opposite; so suddenly they’re very cautious about advice.

A final problem is that when you dissect these kinds of studies, like one recently that was touted as proving that red meat kills, you find that usually you have two groups on either extreme end. On the one hand you have a group with slightly higher rates of heart disease mortality, and they have the following characteristics:

Older and fatter and don’t exercise; tend to smoke and drink heavily, particularly beer; eat meat, cheese and also lots of fast food, refined flour, and refined sugar. Ignore most of what their doctor tells them.

The ones with, again, only slightly lower rates of heart disease mortality have the following characteristics:

Younger and thinner and exercise faithfully. Don’t tend to smoke and don’t drink as much–and it’s wine and spirits rather than bee. Don’t eat meat or cheese at all or only in very small quantities, but also eat no fast food, no refined flour, and no refined sugar. Religiously follow whatever their doctor tells them to do.

Obviously it’s the meat that’s making the difference, right? The point is there’s no way in the world to tell without further study, but no one wants to do that. They just want to be content with their associations.


Plato says he’s hungry

Help us keep paying for this site and feeding the dogs.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s