Skip to content

Science and Pseudoscience in Adult Nutrition Research and Practice

Stuart Buck writes:

You posted about this once on your blog, i.e., how many times observational studies have been refuted by clinical trials. Check out the following, especially Table 3.

He’s linking to an interesting article by Reynold Spector, who claims that lots of the advice we’ve been hearing from the nutrition experts (eat less fat, eat less carbs, eat less quickly-digested sugar and starch, etc.) is unsupported by evidence. Most of the purported evidence, Spector says, comes from epidemiology/observation studies, which he says have been repeatedly shot down by more careful experiments (see examples in his Table 3).

Spector’s advice is to keep your weight in a reasonable range, practice moderation in nutritional maters, eat what works for you, eat the recommended daily allowances of vitamins, minerals, essential amino acids, essential fatty acids, and high-quality protein–and, except in special situations, forget about everything else.

He also recommends that medical journals stop publishing endless observational studies. I don’t know if this should be such a concern, given that there are medical review journals that summarize the published literature. I guess the problem is that the New England Journal, JAMA, etc., actively promote their publications every week, and the press eagerly reports the results of their latest studies. Spector points out the incentives involved in all of this.


  1. Ian Fellows says:

    I'm glad to see this issue getting some attention. The whole problem though is that we really have no idea how to accomplish even the modest "common sense" advice you have given.

    1. There is no proven way to "keep your weight at a reasonable level." Repeatedly, long term clinical trials fail to show weight loss 3 years after intervention. The body has a natural (though possibly unnaturally created) fat level homeostasis that is VERY hard (impossible?) to change. We need to stop blaming the obese for something that there is no proven way to change.

    2. "practice moderation." I hear this one all the time, though there is no consensus on just what moderation is. Is moderate no meat, some meat or all meat? You could make a case for any of these. As for portion size, people will eat until satiated. Trying to fight hunger is like trying to be celibate… possible but not something I would be prepared to do.

  2. Russell says:

    I strongly recommend reading Gary Taubes – he's the first citation in the linked article. It is interesting that Dr. Spector cites Taubes to show the unrealiability of existing research, but then falls back on the old "everything in moderation." He attributes that idea to the "ancient" Aristotle.

    Aristotle lived less than 2500 years ago, but humans and our "pre-human" ancestors (and therefore our genes) have existed for how many hundreds of thousands of years? What did they eat? What did they not eat? Something to think about.

    So I agree with Dr. Spector's criticism of nutritional studies, but find his advice for what to do to be the status quo. If you read Taubes, then I think you will find this status quo advice sorely lacking.

  3. garth says:

    Russell your comments are right on.

    Spector serves up the same advice that he just
    a few sentences earlier warned was unproven.

    Classic error but I am afraid hardly unusual.

  4. garth says:

    Related to the Expert bias I came across this link from the Predictably Irrational site.

    Here an experiment was done where an cocky expert was viewed as more reliable even when
    past history showed his predictions were no better than anyone else.

    Does the name Jim Cramer come to mind?

    Here's the link:

  5. Keith O'Rourke says:

    Does anyone know how to reliably process that kind of information? – poorly done and selectively published (mis) reports/summaries of observational studies!

    Re: "He also recommends that medical journals stop publishing endless observational studies. I don't know if this should be such a concern, given that there are medical review journals that summarize the published literature."

    I'd second "the don't publish" until studies can be "managed" and published/archived in a way it can be safely interpreted.



  6. Joseph Delaney says:

    Re: "medical journals stop publishing endless observational studies"

    I am (obviously) a big skeptic of a lot of observational research. However, there are some points to keep in mind:

    1) Some problems can only be tackled by passive observation due to the ethics of experimentation. Clever data collection, monitoring and analysis can help but there are limits

    2) Careful observational research (with the first word in the sentence being crucial) is likely better than the generalization of anecdote that we'd be forced into otherwise

    3) Long term experiments run into issues (adherence, loss to follow-up, treatment cross-over) that may begin to blur the distinction between trials and observational studies; this is even more true when trials are of things that can't be done blinded (i.e. exercise)

    4) A solid body of observational work may make a previously inconceivable experiment possible as it can introduce equipoise into a situation where there was previously a consensus.

    That being said, in my own area (pharmacoepidemiology) I see a LOT of studies on beneficial drug effects that fail to consider the likely impact of confounding by indication.

    It's a hard area . . .

  7. WL says:

    Stephan Guyenet has a really good blog post (Part I and Part II) on these issues.

  8. Michael Bishop says:

    Medicine isn't unique. See social science.