Today I’ve been inundated with comments, emails and even a phone call or two about the ‘study’ that hit the news this morning allegedly showing that the Atkins diet causes blood vessel damage, and increase in ‘bad’ cholesterol and increased levels of inflammation. I figured I would take this opportunity to describe how this kind of information gets out there and discuss this ‘study’ in particular.
To begin with, this isn’t really a scientific study published in a peer-reviewed journal. It was a brief presentation (about 15 minutes including questions) made at the annual scientific meeting of the American Heart Association in Orlando, Florida a couple of days ago. To better understand where presentations like this one fit in the hierarchy of the scientific world, let’s take a look at how these huge meetings are organized.
The annual Scientific Sessions of the American Heart Association is an enormous meeting with thousands and thousands of attendees. This year’s meeting, which is still going on, is being held at the giant convention center in Orlando, Florida. When the organizers of these kinds of meetings start working on putting them together – which they do years in advance – they begin to contact all the big guns for the major lectures. These lectures are presented during the prime times of the conference when nothing else is going on and they can be attended by all attendees. These lectures held in the huge auditorium are usually by well-known, established researchers who present the data from many years of their work on specific inquiries.
Scheduled in around these giant lectures are concurrent meetings, which are held in small conference rooms holding 50-60 or so people. (The photo at the top of this post shows the typical size of one of these sessions) These smaller lectures compete with other lectures taking place at the same time. Usually there are four or five sessions going on at once, which is a major pain if you’re an attendee because invariable two lectures you want to see are going at the same time. (Fortunately, in our case, MD can go to one and I go to the other, so it’s not so bad.) These lesser presentations are often, but not always, made by graduate students or newly minted Ph.Ds or physicians doing postdoctorate work.
At the bottom of the food chain in terms of importance are the poster presentations. These are usually held in a large room and involve a bunch of people with posters describing work that they have done or research in progress.
In between the poster presentations and the concurrent lectures are oral poster presentations. These are lectures about work in progress or small or preliminary studies that have been done but haven’t been published. The ‘study’ in question is an oral poster presentation.
When the organizers of these meetings are in the preparation stage they send out a call for abstracts. They get inundated with abstracts and arrange them sort of in terms of significance with the most interesting or scientifically significant ending up being presented as a concurrent lecture. The next down the list are the oral poster presentations, then finally the posters.
The important thing to understand about these presentations is that they are not peer reviewed as they are if and when they show up in a scientific journal. They are sort of peer reviewed on the spot in the sense that other researchers familiar with the specific field ask questions of the presenters or of the people with the posters. But that’s it. There is no scientific review of the data as there is when it is published in a journal. So people can present all kinds of data with no one looking it over other than the folks who ask questions. And sometimes the exchanges at these meetings can become spirited to say the least, which means that those lecturing are often present data much at odds with what other scientists have found doing the same kind of studies. That’s why it’s always good to take anything that comes from a meeting like this one with a huge grain of salt.
If you take a look at the monster program for the 2007 Scientific Sessions of the American Heart Association (click here) you can find the presentation in question on page 407. It is session #3610. You should roam through this program a little just to get a feel for the scope of a meeting of this size and to see just how much is going on at once, which puts perspective on any given oral poster presentation.
These giant meetings are open to the press, members of which get these huge programs just as do the scientific attendees. The folks with press passes go through these programs looking for lectures that they think will give them a headline. They don’t care how insignificant the talk is, who is giving it, whether it’s a poster or an oral poster presentation – they are looking for headlines. And what better headline than that there is a problem with low-carb diets or even better, the Atkins diet itself.
So the stage is set. I would imagine that of the 30 or 40 people (if that many) who attended this talk a large number had press passes. These media types stay after and talk to the speaker to get a few quotes, then head off to the press room, dash off a few hundred words, send it to their service, and head back off for another headline-grabbing talk (or the bar).
Now that you know the genesis of the news report about this oral poster that many of you have read and maybe worried about, let’s take a look at the study itself.
I emailed the publicist for the University of Maryland Medical Center for an abstract of this talk, and he kindly sent it to me within minutes. (I’ve converted it to PDF for you: miller-2007-aha-abstract-diet.pdf.) Remember, this is all there is. Just an abstract, not a real paper. We can tell only a little about what really went on in this study. But let’s take a look.
The people doing this study were obviously tired of hearing about the supremacy of the low-carb diet as a weight-loss tool, and they probably didn’t want to do yet another study comparing the low-carb diet to the low-fat diet for weight loss or lipid improvement or blood sugar normalization or blood pressure lowering because they knew what that outcome would be. Those studies have been done again and again and the low-carb diet always comes out on top. So, these researchers decided to take a different tack.
They wanted to see what happened when subjects stayed on the diet after reaching maintenance, so they took 18 subjects (they started with 26, but only 18 made it through the entire study) and put them on one of three diets: the Atkins diet, the Southbeach diet or the Ornish diet. The subjects stayed on one of these diets for 4 weeks, then followed their normal diets for 4 weeks, then went on another of the three diets for 4 more weeks, then off for 4 weeks, then on to the last of the three diets for 4 more weeks. But the researchers didn’t put the subjects on the weight-loss versions of these diets, they put them on maintenance versions. They did this by weighing the subjects at the end of each week and adjusting their caloric intake so that they neither gained nor lost weight.
By keeping the weight of the subjects stable, any changes wrought by the various diets could be attributed to the diet and not to the lost weight. Many people believe the changes brought about by low-carb diets occur not because of the diet but because of the weight loss the diet induces. And since in most studies subjects on the low-carb arm lose more weight than those following low-cal diets, it seems reasonable to suppose that it is the lower weight and not the diet composition that results in the better lipids, normalized blood sugars and lowered blood pressure commonly seen in the low-carb group.
In this case the weights stayed the same for all the subjects during the various diets, and as you can see from the abstract, the LDL levels went up (although not significantly) in the Atkins group while LDL levels dropped in the other two groups. The abstract says that the researchers drew blood at the end of each 4 week phase on the different diets, and I assume that they also drew blood at the start of each 4 week dietary trial, but the abstract doesn’t say. For all I know, they could have drawn blood for baseline values at the very start of the study and compared the lab values at the end of each 4 weeks to this baseline number, which would make the later numbers suspect. This is one of the problems with ‘studies’ like this one. These questions would all have to be answered before such a study could be published in a peer-reviewed journal. But in a conference, pretty much anything goes. It would be left up to an attendee to ask such a question.
Along with the LDL levels, the researchers also looked at a number of fairly arcane measures of inflammation: ICAM2, SELL and SOD1. As far as I know, these are not lab studies that the average doctor can order, but are ones that are done in research facilities for research purposes. In any case these indicators went up on the Atkins diet and stayed the same on the Southbeach and Ornish diets.
Finally, the researchers performed a brachial artery reactivity test (BART), which is a somewhat controversial but nevertheless commonly used test to measure endothelial function. (About halfway this old post is a description of how BART works. If you want a more detailed explanation with a discussion the controversial nature of the test, read this full text article.) As to the outcome of BART, the abstract simply reports the following:
BART testing revealed a significant inverse correlation between flow-mediated vasodilation and intake of total fat, saturated fat and monounsaturated fat.
We’re not told if this correlation holds irrespective of which diet the subjects were following or if these were the numbers while they were following the Atkins diet. From the wording of the abstract one would have to think that these values were reported for all the diets. If so, then the researchers have extrapolated from these numbers that the Atkins diet causes more endothelial dysfunction simply because it contains more fat and certainly more saturated fat than the other two diets.
Based on the abstract we find that
In the absence of weight loss, the high fat Atkins diet is associated with increased LDL-C, reduced endothelial vasoreactivity and increased expression of biomarkers of atherothrombosis. As such, these data suggest that isocaloric conversion to the Atkins diet may negatively impact cardiovascular health as compared to the South Beach or Ornish Diet.
Now, let’s probe a little deeper. And we don’t have to get very deep to see a major flaw in this study.
Before we get to the major flaw, though, let’s look at the LDL changes. As far as I’m concerned, they are a big Ho Hum. I mean, who really cares? How many studies do we have to look at to know that when you cut fat from the diet LDL levels decline? We already have dozens showing us this same finding. Dozens of other studies show us that when fat, particularly saturated fat, is increased in the diet that not only do LDL levels rise a little but HDL levels rise a lot. (I posted about this in the past.) So what little risk we might seem to accrue because of increased LDL levels is more than offset by the greater increase in HDL levels. It even gets better. The majority of studies have shown that when LDL levels increase due to high-saturated-fat, low-carb diets, the LDL particle size increases, making the LDL non-atherogenic. And while LDL levels fall with low-fat, high-carbohydrate diets, the LDL particle size decreases and becomes the small, dense type, which is highly associated with cardiovascular disease.
So, this study tells us that when the subjects went on the Atkins diet their lipid profiles actually improved as compared to the other two diets. Why didn’t the researchers just say that? One wonders. One also wonders why – if they went to the trouble and expense to check ICAM2 and the other markers of inflammation – they didn’t bother to check HDL levels or LDL particle size, especially since LDL particle size is lipid parameter that has the strongest correlation to the development of cardiovascular disease.
Back to the major flaw. The abstract doesn’t say how much these subjects weighed. So we can only assume. Let’s assume that they were average weight and required 24oo kcal per day to maintain their average weight. While these subjects were on the Atkins diet part of the study the abstract tells us that they were consuming 50 percent of their calories as fat. Let’s think about this for a minute and do a few back of the envelope calculations. Fifty percent of 2400 calories is 1200 calories. So 1200 calories were fat and 1200 calories were other than fat. Since other than fat means carbs and protein, that means that the other 1200 calories were divided between these two macronutrients. If we assume that 20 percent of calories were protein that calculates out to about 120 grams of protein per day, which is a fair amount and probably more than they actually got. But let’s assume 20 percent. That leaves 30 percent of calories as carbohydrate. Thirty percent of 2400 kcal comes out to 720 calories as carb. And since a gram of carb is 4 calories, dividing it out gives us 720/4 or 180 grams of carb per day. Does that sound like the Atkins diet to you?
Let’s give the researchers the benefit of the doubt and say that they had the subjects on 30 percent protein (180 grams), which I doubt because they would have been scared to death to give someone 180 grams of protein. But let’s assume they bucked up and did it. Running the same calculations on 20 percent of carbs gives us 120 grams of carbs per day – that’s a full 24 teaspoons of sugar’s worth of carbohydrates per day. Or two potatoes and a serving of pasta. Is that the Atkins diet?
See what I mean. It was a flawed study. I already linked to this post above, but it discusses the same issues, i.e., blaming a lab finding on saturated fat when the diet is full of carbs.
What do we care that a low-carb diet showed an increase in inflammation and worsening endothelial function when what was tested wasn’t really a low-carb diet?
In summary: no complete study, no peer review and no Atkins diet. Yet it made the news big time.
Now that you know the story, go back and read the press report of this oral report of a poster. See if your take home message is any different now.
Oh, and one other thing. According to the abstract the study started out with 26 subjects, but only 18 completed it. That’s a dropout rate of over 30 percent. One wonders why these subjects dropped out? And from which dietary protocol? It would be interesting to know, but I can make a pretty good guess.