Thursday, February 26, 2009

Media Distortions and the Lastest Big Diet Study

You may have seen the buzz around the latest diet study, just published today in the New England Journal of Medicine. The New York Times' article about it is quite representative of the coverage this study is getting.


For people who are trying to lose weight, it does not matter if they are counting carbohydrates, protein or fat. All that matters is that they are counting something.


Reading this with my morning coffee threw me into a bit of a panic, given it flies so completely in the face of my personal opinions and experiences. What if my weight loss has all been a lie? Did I really take the blue pill after all?

Then I read the actual NEJM study article. Existential crisis averted. It turns out the study itself has some major flaws, serious enough that the conclusion (which the media picked up on) doesn't even really follow. Let's examine it in detail (below here, quoted sections are directly from the study).

First, it wasn't even designed to address the question of whether caloric restriction works as a general dietary strategy or not. Why not?


Each participant's caloric prescription represented a deficit of 750 kcal per day from baseline, as calculated from the person's resting energy expenditure and activity level.


This one sentence is rather telling. It seems that every group, regardless of macronutrient breakdown, was given a "caloric deficit" up front. So already we are abandoning the idea of testing whether caloric restriction, in general, works. In fact they admit there is no control group whatsoever.


No diet was considered to be a control diet, and the dietary counseling and the attention that we provided were the same for all diet groups throughout the study period.


We're not off to a very good start, but let's continue by looking at the metric they did actually measure, weight loss. Ordinarily here I would raise objections because what most people really want is to lose body fat, not weight, and losing the latter does not imply losing the former. But in this case (the individuals were overweight or obese middle aged men and women) they are probably close to the same, so we'll overlook it. Fortunately, they did measure waist size, a far better indicator of body fat levels. Here is what they found.

They claim the results are equivalent among groups but I'm not so sure. In both the "high protein" and "low carbohydrate" groups, waist reduction was 20% greater than in the "low protein" and "high carbohydrate" groups respectively.

OK, well even with these relatively minor problems, surely the central premise can still be salvaged. I mean, they are actually testing whether it's just the number calories that matter, not what they come from, right? Surprisingly, no! The bombshells (credit goes to Garret Smith and Steven Low for pointing this out to me).


Other goals for all groups were that the diets should include 8% or less of saturated fat, at least 20 g of dietary fiber per day, and 150 mg or less of cholesterol per 1000 kcal. Carbohydrate-rich foods with a low glycemic index were recommended in each diet.



Behavioral counseling was integrated into the group and individual sessions to promote adherence to the assigned diets.



The goal for physical activity was 90 minutes of moderate exercise per week. Participation in exercise was monitored by questionnaire and by the online self-monitoring tool.


First they are mandating a dietary fiber requirement, which clearly falls into the "where they come from" category. But more damningly, by even acknowledging glycemic index, they are freely tacitly admitting that not all calories are equal. That is, after all, precisely what GI is concerned with (favorable versus unfavorable carbohydrates with regard to weight management).

And not only that, but now they are throwing in the undeniably confounding variables of exercise and counseling into the mix. I thought this was supposed to be a dietary study.

So, in the final analysis, this study has very little to do with diet in general, forget the significance of the number of calories consumed. Now, as a fun experiment, let's visit some bizzaro alternate universe where they actually designed this study to test whether it was strictly the number of calories that matters in fat loss.


  • Create real control groups, which are allowed to consume as much food as they want but in the proportions as each of the test groups. This will let us test whether caloric restriction actually works.

  • Get rid of the exercise requirements and counseling. Let's stick to testing diet.

  • Make a real low carb group (35% isn't low enough), and high protein group (25% isn't high enough).

  • Impose no requirements on saturated fat, fiber, or glycemic index. Only calories matter, remember?



Yay - I could be a medical researcher!