April 5, 2008

You Ask, I Answer: Jules Hirsch/Rudolph Leibel Obesity Research

I can't find [their studies].

I do have access to medical and nutrition journals - they are all online, and anyone can access them for about $8 per article - and I still can't find this report.

Nothing from Hirsch and Leibel in 1950s or 1960s.


I have just basic questions about the study - how big was the study (how many people tested) etc - but I can't find that out without finding the original report.

Have you read it?

-- "RicoVado"
Via the blog

I first came across the study two years ago, and was able to find it online for you to peruse -- be sure to download the PDF, rather than simply read the abstract.

In summary, Hirsch and Leibel published a paper in 1992 in the American Journal of Clinical Nutrition, in which they reviewed the records of patients at Rockefeller University's Lipid Laboratory between 1955 and 1965.

They specifically targeted patients -- a total of 16 -- fed different liquid formulas (for at least two weeks) containing an equal amount of calories, but varying fat:carbohydrate ratios.

Although 16 patients might not seem like a very large sample size, the confidence intervals, odds ratios, and power -- a statistical term -- figures of this study demonstrate statistically significant results.

It is worth nothing that none of the patients underwent significant weight changes when their formula was replaced by one of the same caloric amount but a much different (either higher or lower) carbohydrate content.

Gary Taubes has casted off this study as useless in rebutting his argument since none of the patients were obese (he claims to use carbs as the explanation for obesity in individuals already predisposed to it, although that was certainly not made clear in his talk at New York University last month).

Well, Hirsch and Leibel reference several other studies that came to similar conclusions.

Of particular importance is one from 1990 -- also published in the American Journal of Clinical Nutrition -- in which Gilbert Forbes concludes that "although obese individuals appear to deposit a larger proportion of excess intake energy as fat, the energy cost in either lean or obese subjects was not significantly influenced by the composition" of their diet.

Have a slice of whole wheat toast on me tomorrow morning!

6 comments:

AnotherSuggestion said...

Thanks for posting this.

I see where my confusion came from - Kolata refers to "definitive studies done in the 1950s and ’60s by Jules Hirsch of Rockefeller University and Rudolph Leibel of Columbia." She is referring to something that does not exist. Instead what we have is this paper which is based on an analysis done in 1992 on data generated by Hirsch in 1950s and 1960s. As you can imagine, looking for work done by Hirsch & Leibel with publication dates in the 1950s or 1960s was a futile enterprise.

Funny that a science reporter for the NY Times would make such an obvious error.

One other note: when Taubes says the results of the study "might not have been obtained in a group of obese individuals or lean individuals susceptible to obesity" he is quoting Hirsch and Leibel! From this very study! (Page 355 of this study)

What are we to make of a study on diet and weight-gain that the authors agree might not apply to people who are overweight or likely to get that way?

After reading the study for myself, I would emphasize Hirsch and Leibel's repeated calls for restraint in interpreting the results of their study. Their restraint stands in stark contrast to Kolata's claim that this study "definitively proves" anything.

Again thanks for the interesting discussion and for posting the primary document.

Andy Bellatti said...

You will be hard pressed to find any study that does not mention possible limitations in its "results" section.

The mention of a possible limitation should not be seen as invalidation (again, especially when looking at P values and confidence intervals).

This study references others that HAVE studied obese individuals and energy intake and found that diet composition was not a significant influence on any weight changes.

Anonymous said...

In addition to the caveats pointed out in the paper by the authors themselves, there is another reason why this study isn’t terribly relevant. Under the definition used by the Leibel et al, the lowest-carb diet contained 15% of its calories from cerelose, which is a form of glucose. If we do a back-of-the-envelope calculation, we come up with a daily intake of roughly 86 grams of glucose -- an amount that is quite high by Taubsian / Atkinsonian standards. (86 grams x 4 calories per gram = 344 calories, which is about 15% of 2,300 calories). It would be more interesting to see what would have happened had the low-carb diet consisted of, say, 10 or 20 grams of glucose per day.

AnotherSuggestion said...

Andy said: "You will be hard pressed to find any study that does not mention possible limitations in its "results" section."

I agree. And with that as the norm, Hirsch & Leibel are especially modest in their claims and especially open about it's limitations. That's what makes Kolata's "definitive studies" comment so inappropriate.

I am starting to feel like we're peeling back the layers of an onion. Kolata points to Hirsch & Leibel as the "definitive" study but we take a look at the study and see that it is anything but "definitive."

So we go to the next layer - Hirsch & Leibel refer to other studies that perhaps don't have the same limitations. But what are those studies? If we looked at those would we be more convinced of the Kolata hypothesis (the macro-nutrient mix of the diet is irrelevant to weight gain), or would we be less convinced?

Anonymous said...

Small Bites: “Of particular importance is one from 1990 -- also published in the American Journal of Clinical Nutrition -- in which Gilbert Forbes concludes that "although obese individuals appear to deposit a larger proportion of excess intake energy as fat, the energy cost in either lean or obese subjects was not significantly influenced by the composition" of their diet.”

I found the study by Forbes that you mention and that Hirsch & Leibel reference.

Forbes’ study is a review of other people’s published data – including one of his own studies. I believe this is called a meta-analysis.

Forbes only looked at studies in which subjects were fed either high carbohydrate diets or very-high carbohydrate diets. Both made overweight subjects even more overweight. That’s what he means when he says that diet composition didn’t affect weight gain.

(Forbes does mention one small group (n=4) given excess fat on top of their mixed (high carbohydrate)
diet, but concludes "that the number of subjects given excess fat is too small to warrant a regression analysis.")

Forbes does not have anything to say about low-carbohydrate diets because he didn’t review any studies that tested low-carbohydrate diets.

So this doesn’t exactly sever the Gordian Knot.

Andy Bellatti said...

Just a sampling of the research literature arguing against low-carbohydrate diets as the best tool for weight loss:

http://www.atkinsexposed.org/atkins/22/Opinions.htm