Read on Twitter
David Ludwig and co-authors just published an interesting study in JAMA Network Open, and an accompanying NYT op-ed. The JAMA study examines how frequently diet trials stick with their preregistered primary outcomes, compared with drug trials. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2755303
This is a narrow but useful measure of trial quality. Results show that 86% of diet trials had “substantive discrepancies”, while only 22% of drug trials did. This is consistent with my experience reading diet trials & highlights a common shortcoming of diet research methods.
It is indeed very frustrating to read trials and not know how to interpret them because the researchers didn’t stick to their research plan. I think the culture and incentives need to change. Hopefully this study will accelerate that.
The accompanying NYT editorial, by Ludwig and Steven Heymsfield, uses the JAMA study as a launch pad but quickly moves on to issues beyond the scope of that study. https://www.nytimes.com/2019/11/13/opinion/diet-research-nutrition.html
The title is “Why Diet Research Is So Spectacularly Thin”. I don’t know who chose this title (could have been the editors), but it plays into the trope that we don’t know anything about nutrition (often paired with “and so you might as well believe this other hypothesis”).
I don’t like this trope at all because nutrition research has saved many millions of lives, and continues to save and improve lives every day. Here’s one easy example.
https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD008524.pub3/full
https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD008524.pub3/full
Ludwig and Heymsfield critique the “illogical assumption that a diet didn’t work” when a trial returns disappointing results and the subjects didn’t stick to the diet very well. In fact, the conclusion that the diet didn’t work is quite logical.
The reason is simple: life isn’t a metabolic ward. If people can’t stick to a diet in the course of their daily lives, that says something about the effectiveness of the diet (and/or the effectiveness of diets in general).
We already have lots of diet trials comparing many different diets under real-world conditions. Nearly all weight loss diets cause weight loss, some more than others, but none of them lives up to the lofty claims of their public advocates.
In most cases, that isn’t a shortcoming of the diet trials, it’s a shortcoming of the diets themselves! If the diets did what we wanted them to, the education/support offered in diet trials should be more than sufficient to support effectiveness.
This gets back to a key limitation of nutrition science, in my view: it can’t stop us from eating cake. We already know cake, pizza, ice cream, etc. are fattening. So why do we eat those foods? Lack of information isn’t the main problem. #HungryBrain
Ludwig and colleagues want to pit 2-3 diets against one another under long-term metabolic ward conditions. Go for it! It will certainly be interesting. But the information gained will probably have limited public health relevance.
Ludwig and Heymsfield also critique is the short duration of many diet trials. I certainly agree that is a key limitation of most diet trials. I do want to comment on the context of the critique though.
Ludwig believes that elevated insulin secretion is the primary cause of obesity (or at least a major cause? I’m not clear on this), and has advocated for this idea both in the scientific literature and to the public. @davidludwigmd
Recent evidence (mostly from @KevinH_PhD) has shown very convincingly that diet-induced insulin secretion is not a major determinant of changes in body fatness over periods of 3-28 days. Ludwig responded that the trials were too short. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4603544/
We certainly need to be cautious about extrapolating these findings out to months and years. But that’s different than continuing to promote the insulin hypothesis under the assumption that longer studies would have supported it.
IMO, the most favorable rational take for the insulin hypothesis would be: “Yes, these trials up to a month in length are unsupportive, but it’s possible that longer trials would be supportive. That possibility is speculation though.”
Since long-term metabolic ward trials with body fatness as an endpoint haven’t been done, all we have to go on is the shorter trials. And those suggest that insulin per se is not a major determinant of adiposity when kcal intake is controlled.
I agree with the authors’ point that nutrition research is neglected relative to its importance. I would love to see a nutrition “Manhattan project”, as they suggest. From a public health standpoint, I think adherence is the main nut we need to crack though.
In other words, how do we get people to make positive diet and lifestyle changes, and sustain them? Incremental advances in our understanding of carbs vs. fat in metabolic wards are great brain food but they aren't going to answer the most important question.
