Synchronicity strikes again. The seeds of this post were sown when Gary Taubes emailed me about a study published in early 2009 in the New England Journal of Medicine (NEJM) that I had seen at the time, briefly skimmed and tossed aside as worthless. Gary agreed that the study was of little value, but notice that it contained a peculiar statement by the authors, an interesting admission about HDL, the lipophobe’s favorite lipoprotein. And not only had the authors made this strange admission, but so had another prominent lipophobe who wrote the accompanying editorial.
I pulled the study, read it more thoroughly and still found it mediocre at best. But I did come across the strange HDL statements that Gary had mentioned. (More about which later.)
As I was shaking my head over the amount of money spent on what was a truly abominable study, the synchronicity occurred. I got a ding that I had a new email. It was a notice from the American Heart Association telling me that this august body had deemed the very study I was holding in my hands as one of the ten most important papers published in 2009. The sheer stupidity of it nearly took my breath away.
Before we get into the study – which we won’t get into very deeply because, believe me, there’s not much depth – I want to use a parable to show just how silly this study is.
Let’s set our story in the wonderful country of Stupidland where a debate has been raging about the feeding of dogs. A vociferous old woman who kept dogs had been insisting that different breeds of dogs eat different amounts of food The majority of the populace were of the opinion, however, that all breeds eat the same amount (it is Stupidland, after all) and looked down their noses at those who believe a chihuahua may eat less than a collie. To put an end to the bickering, scientists at Stupidland U ( who were believers in the all-dogs-eat-the-same doctrine) decided to do a definitive study. They went to the Stupidland pound and procured a German Shepherd, a Labrador Retriever, an Irish Setter and an Alaskan Malamute.
They provided the four dogs with pleasant accommodations and all the food they wanted to eat. The scientists carefully measured every gram of food eaten by each dog and recorded it. At the end of the two year study, they reviewed the data and confirmed what they already suspected to be the case: the different breeds of dogs ate just about the same amount. They did notice one little disparity, however: the larger dogs ate a little more than the smaller dogs, but they were able to correct for that by controlling for size. Their paper proving that different breeds of dogs ate the same amount of food was accepted for publication in one of Stupidland’s most prestigious scientific journals, The Stupidland Journal of Veterinary Medicine. Buried deep within the paper was a sentence few noticed stating that size was a biomarker for food consumption by dogs.
The Stupidland press picked up on the study and headlines proclaimed that all breeds of dogs eat the same amount. The mainstream Stupidlanders nodded their heads sagely; they, after all, had been right all along. But the old woman, who didn’t actually live within the borders of Stupidland, but who lived close enough to cause trouble, kept insisting that different breeds of dogs didn’t eat the same amounts. She had a beagle and she had a Great Dane, and she had kept careful records of the food consumption of both. She insisted that the Great Dane not only ate more than the beagle, but that it ate a huge amount more. She would bend the ear of anyone who took the time to talk to her, and her data was so persuasive that she was beginning to make converts. Just as the population of Stupidland was once again starting to wonder about the dog breed verses food enigma, the Stupidland Heart Association came out with its annual bulletin announcing that the paper by the brilliant scientists from Stupidland U showing that all breeds of dogs ate the same was the most important paper of the year. The old woman’s first impulse was to attack the Stupidland Heart Association for its sheer stupidity, when suddenly a sense of calmness and clarity settled over her. She experienced a spiritual awakening (just as did the Grinch in another tale) and finally realized the real meaning of Stupidland. She took her dogs and moved far away, leaving the denizens of Stupidland alone to marinate in their stupidity.
The paper that inspired this parable was published in Feb 2009 in the New England Journal of Medicine and titled Comparisons of Weight-Loss Diets with Different Compositions of Fat, Protein, and Carbohydrates. (This is another one of those studies the editors feel is so important that they provide the full text free of charge as a public service.) The authors include Frank Sacks, George Bray, Steven Smith and an entire rogue’s gallery of lipophobes. All the usual suspects, as they say.
What the NEJM study sets out to demonstrate is that different breeds of dogs different weight-loss diets of varying macronutrient compositions all bring about the same loss of weight. According to these authors, it doesn’t matter if you go on a low-carb, high-fat diet or a low-fat, high-carb diet, you’ll lose the same amount of weight. Doesn’t matter how the protein, fat and carbohydrate stack up in your weight loss diet, you’re going to lose the same amount of weight. So, you can go to the bookstore, stand by the diet-book shelf, close your eyes and pick. Whatever diet book you end up with won’t matter because you’ll lose the same amount of weight regardless of which one you choose. And, even more importantly – again, according to the authors of this study – whichever diet book you select will help reduce your heart disease risk factors.
As Dave Barry says: “I AM NOT MAKING THIS UP.” It’s right there in black and white in a study done at Harvard and published in the New England Journal of Medicine.
What’s more, the American Heart Association (AHA) deemed this study to be one of the top ten most important studies published in 2009. And they put it #1 on their list. Now they said that they listed these ten studies in no particular order – and you can call my cynical – but I’m just betting that they put this one right at the top for a reason.
Said the president of the AHA, Dr. Clyde W. Yancy
We all thought the statement made in that study was pretty profound. It really dismissed the notion that there’s something clever about weight loss, [showing] that it really is about calorie consumption or, to make it even more straightforward, portion control. You can spend a lot of time wringing your hands about which diet and the composition of which diet, but it really is a simple equation of calories in and calories out.
Give me strength.
My disgust aside, you may be thinking: Why isn’t the study valid? If they did analyze all those diets and found them to bring about the same results, what’s the problem?
The problem is that the diets they used in the studies were similar. They didn’t vary all that much in carbohydrate. The diet with the highest carb intake contained 65 percent of calories as carbohydrate while the lowest carb diet was made up of 35 percent. To put this into the gram figures we’re all used to, the highest-carb diet contained 325 gram of carb while the lowest-carb version contained 175 gram of carbohydrate. Now, as those of us who have ever followed a low-carb diet know, 175 gram of carbohydrate does not a low-carb diet make. Granted, it’s lower in carb than the diet with the 65 percent of calories as carb, but it doesn’t even approximate a low-carb diet. As I’ve written before, you’ve got to get the carbs substantially below 100 g per day before good things start happening metabolically.
What this study has done is to study roughly similar diets for two years and pronounce that all produce about the same results. What the authors (and, apparently the AHA) want you to take away from this study is that real, honest-to-God low-carb diets don’t perform any better than low-fat, high-carb diets. Which, as most of us know from bitter experience, is not the case.
There are major problems in doing studies such as this one that make their outcomes suspect. And these problems aren’t necessarily the fault of the researchers – they are simply a fact of life.
When you try to do a dietary study by recruiting people who want to lose weight then randomizing them to a particular diet, you are asking for trouble. If you run the study out over a long period of time – two years, for example, as this study did – you are asking for even more trouble. People go into diets with a lot of enthusiasm and pretty rigorously stick to them at first. But as time goes on, people tend to cheat a little, then cheat a little more and pretty soon find themselves pretty much trending back toward and finally squarely back on whatever their regular diet was before they started the study diet. (Sadly, it’s not just subjects in studies who follow this pattern, but is the fate typical of most dieters.) For this reason, after time, all the people in all the different arms of the study are eating about the same thing. This is why you always see the charts showing weight loss and macronutrient composition start out wildly diverging then converge as the end of the study draws near. In other words, they all end up consuming the same diet, so they all end up with about the same result.
How can researchers overcome this dismal outcome. Well, you can put out the call for people who really believe in low-carb diets to fill one arm of the study. And recruit people who love the Ornish diet for another, and the Zone for another. These subjects are more likely to stay enthused and stick with their respective regimens for the duration of the study. But then you haven’t randomized your sample and you will be accused of generating worthless data because your sample groups self selected.
The other way, of course, is to randomize subjects into various diet groups, then put them under lock and key for a year or two and feed them like you would lab animals. Another impractical solution from a cost perspective if in no other reason.
It’s extremely difficult – virtually impossible, I would say – to conduct accurate studies on diet over a long period of time with a large number of subjects. Consequently, it is nonsensical to rely on the data from such studies to make the case for anything other than how difficult these studies are to carry out. I certainly don’t think for all the reasons above that the study in question merits being listed as one of the top ten studies of 2009 by anyone, much less the AHA.
In their discussion of this mishmash of questionable data, however, the authors did make a most interesting statement. Almost an admission, if you will, of the superiority of a lower carb diet. This statement is what Gary emailed me about.
(Before we go on with this, I have to make this aside. HDL and LDL and IDL (intermediate density lipoprotein) and VLDL (very low density lipoprotein) aren’t really cholesterols. Even though we often refer to them as LDL cholesterol and HDL cholesterol, they really aren’t. These different groups of letters refer to transport proteins that carry cholesterol through the blood, not to cholesterol itself. Cholesterol is cholesterol. It is a specific molecule that doesn’t change. Cholesterol is a waxy lipid (fat) that virtually every cell in the body synthesizes (because is it so important). Cholesterol, like all fats, is not soluble in water and therefore can’t dissolve in blood (which is a watery substance), which means that the body has to package cholesterol in a form in which it can be transported from place to place in the blood. The body attaches a specific protein (a lipoprotein) to cholesterol to make it dissolve in the blood. The names LDL, HDL and the rest refer to the specific type of lipoprotein being discussed.)
Here’s what the authors wrote:
There was a larger increase from baseline in the HDL cholesterol level, a biomarker for dietary carbohydrate [my italics], in the lowest-carbohydrate group than in the highest-carbohydrate group (a difference in the change of 2 mg per deciliter at 2 years)…
Even Martijn Katan, a lipophobe if there ever was one, and the author of a number of anti low-carb diatribes that I’ve taken to calling the Katanic Verses echoes the same fact – carbohydrates drive HDL down – in an editorial he wrote about the above paper.
…compliance was assessed with objective biomarkers.
The authors used the difference in the change in HDL cholesterol levels between the lowest- and highest-carbohydrate groups to calculate the difference in carbohydrate content between those diets.
Now the differences weren’t all that spectacular, but the drop in HDL in those on the higher carb diet was there and noticed by the researchers.
I find this extremely revelatory because if there is one lipid parameter a lipophobe loves, it’s HDL. And here you have an entire cluster of lipophobes admitting that HDL varies as the inverse of carbohydrate intake. Take any of these folks individually – or, heck, take ‘em together – and they’ll tell you that low-carb diets are bad because they give you too much fat. Yet they admit that their beloved HDL goes up when carbs go down. Doesn’t make a lot of sense, does it?
When these folks compared these fairly similar diets they found that all of them reduced the risk for heart disease. They used the fact that HDL went up on the lower-carb diets to deem them heart healthful; and they pronounced the higher-carb diets as heart healthful, too, because the LDL declined on those.
As Yogi Berra said: “You can observe a lot by just watching.” And they watched LDL go down on the higher-carb diets and HDL go up on lower-carb diets. But the reverse of the Yogi-ism is also true: you can also fail to observe if you don’t watch.
This refusal to watch is what really gets my dander up.
The researchers whose names are listed at the top of this paper are all affiliated with prestigious institutions. I am quite sure that there is not a single one of them who is unfamiliar with the work over the last 15 years or so of Ronald Krauss, the researcher who made the discovery of the differences between LDL particle sizes. (The same Krauss, by the way, who published the paper about the meta-analysis of saturated fat and heart disease much in the blogosphere currently.) Krauss and his team showed that large, fluffy LDL particles aren’t particularly harmful whereas the small, dense LDL particles are the ones that cause the problems. He also discovered that increasing carbohydrate in the diet caused LDL to shift to a smaller, denser pattern while decreasing carb and adding fat made LDL change to the larger, fluffier non-problematic kind. (You can read a nice review of LDL particle size in this article published in the popular press.)
If you reduce carbs and add fat to the diet, not only does your HDL go up, but your LDL makes a particle size change for the better. However, when you increase carbs and reduce fat, your HDL goes down and your LDL goes down too, but it changes for the worse. So even though the high-carb, low-fat diet decreases LDL, it doesn’t decrease risk – it increases it because even though LDL is lower, it is made up of a dangerous particle size,which negates any possible value of the fall in LDL. All of these researchers know this.
Why didn’t they check LDL particle size on these subjects? Had they done that, they would have found that those subjects on the higher carb diets would have lowered their HDLs and althought they lower levels, would have shifted to more of the dangerous, smaller, denser LDL particles. They couldn’t have then made the case that not only did all diets work the same where weight loss was concerned but they all decreased heart disease risk. They would have had to say that although all diets brought about the same degree of weight loss, the lower-carb diets clearly reduced the risk factors for heart disease the most. And that’s an admission I suspect they didn’t want to make. Therefore they refused to observe.
I don’t know what the deal is with these folks. Why don’t they simply tell it as it is? Do the long-term lipophobes who have ridiculed low-carb diets for years and built their careers on the rickety edifice of the low-fat diet not want to admit they were wrong? That’s understandable, I suppose, but what about the young ones? Why are they stampeding over the low-fat cliff like Gadarene swine? Do the younger lipophobes not want to offend the older ones? Why do they fail to reconcile their theories with what amounts to basic biochemistry and physiology? Whatever the reason, they are fighting a losing battle. Ultimately the truth will out and when it does, all these people who have tenaciously clung to the low-fat, high-carb fantasy will be – like the phrenologists and other failed theorists of the past – so much detritus in the history of medicine. And their books and papers will be displayed as curiosities of the boneheaded thinking of an earlier day. A sad but fitting fate.
Photo: Set of phrenological heads, England circa 1831
via The Pollo Web
Monthly Book Reviews
I have been writing a series of book reviews each month that I email to subscribers. If you're interested and want to get on the list, sign up here (or above where it says Get free email alerts in the upper right). I'll send you an email notice of all new blog posts plus all my monthly book reviews. Also, you will get a link to all the previous month's book reviews I've sent. Hope to see you aboard.