Advocates of low-carbohydrate diet are voicing concern about a recent paper that suggested the diet could cause weight gain, contrary to previous research. One expert has even called for its retraction.
The study, published in Nutrition & Diabetes in February, also found that the low-carb diet did little to prevent the progression of type 2 diabetes. Researchers have since criticized the study for drawing these conclusions based on data from a handful of mice, using a poor proxy for the human version of the diet.
One expert took to social media to warn against the study’s message regarding the low-carb, high-fat diet (LCHF); for instance, obesity & nutrition researcher and author, Zoë Harcombe from Newport, Wales, tweeted:
9 mice – yes 9 mice! were fed sucrose & oil/fat & these dudes think they've unpacked LCHF! https://t.co/pw6cdQo6mi Try meat & veg on people!
— Dr Zoe Harcombe, PhD (@zoeharcombe) February 19, 2016
Harcombe told us:
The paper was incredible (as in — lacking in credibility). It took 17 mice, genetically bred to be fat and sick, and put 8 on standard chow and 9 on a LCHF diet. You only study mice when it would be unethical to study humans. There are many human LCHF studies already…so why study mice? (Maybe because you didn’t like the human conclusions?!)
In addition, the diet given to the mice was far from what people on a low-carb diet eat, Harcombe said.
But based on its conclusions, the study was reported sensationally in the media, Harcombe noted, with headlines such as the Daily Mail‘s “The Paleo diet is ‘dangerous and increases weight gain’, diabetes expert claims:”
The paper was one thing. The way it was reported (sensationally and completely irresponsibly by one of the researchers) was another and the way the media reported it (with no critique/review/analysis/effort) was another still.
Another researcher, Richard David Feinman, a biochemist based at the SUNY Medical Center in New York, posted a letter to the journal on Facebook outlining his concerns with the paper, including:
The recommendation [against the low-carb, high-fat diet] follows from presumed risk based on experiments with 9 mice from a strain bred for susceptibility to metabolic abnormalities. The study ignored the dozens of studies, listed in their own references, comprising hundreds of human subjects, that contradicted the mice data.
In the letter, Feinman calls for the paper to be retracted, and subject to new peer review from experts with experience in the LCHF diet in humans. He explains in the letter:
There are numerous papers demonstrating the value of dietary carbohydrate restriction. Our review ([1], reference (5) in Lamont, et al.) includes 77 references and documents the evidence that nothing is better than LCHFD for obesity, diabetes and metabolic syndrome. We do not think that this is the final word and we are open to criticism. However, a study of 9 mice for eight weeks does not raise questions about this extensive body of knowledge.
Calling the results “ambiguous,” he adds in the letter:
1. The LCHFD had better improvement than control on [triacylglycerol] TAG and HDL (“good cholesterol”) and there were no significant differences in total cholesterol or LDL. The ratio of HDL/TAG is considered one of the best indicators of CVD risk: the LCHFD conferred decreased risk.
2. The result that a reduction in dietary carbohydrate worsened glycemic control, if reproducible, would be a landmark phenomenon demonstrating that conditions could be found where an NZO mouse showed metabolic behavior completely the opposite of what is seen in humans. That increasing dietary carbohydrate will decrease blood glucose in humans is absurd and has been contradicted by thousands of recorded measurements.
3. Similarly, the result (Figure 1) that the low-carbohydrate mice gained more weight even though they consumed fewer calories makes no sense at all. In fact, in the numerous diet studies in humans, the low-carb arm almost always loses more weight and critics claim that that is only because they took in fewer calories.
“A low-carbohydrate high-fat diet increases weight gain and does not improve glucose tolerance, insulin secretion or β-cell mass in NZO mice,” has not yet been cited, according to Thomson Reuters Web of Science.
Co-author of the study, Sof Andrikopoulos, who is based at the University of Melbourne in Australia and is the current president of the Australian Diabetes Society, declined to comment. He told the Daily Mail:
Low-carbohydrate, high-fat diets are becoming more popular, but there is no scientific evidence that these diets work.
The news story also includes this quote from Andrikopoulos:
In fact, if you put an inactive individual on this type of diet, the chances are that person will gain weight…This level of weight gain will increase blood pressure and increase your risk of anxiety and depression and may cause bone issues and arthritis.
For someone who is already overweight, this diet would only further increase blood sugar and insulin levels and could actually pre-dispose them to diabetes.
Feinman, however, called Andrikopoulos’ reported comments “completely out of line,” warning in his letter:
In the end, it is people struggling with obesity and diabetes who are the real victims of this breakdown in scientific practice. There is a serious ethical issue in that Andrikopoulos’s paper and his reported public statements are intended to deny a treatment to people with obesity, diabetes and metabolic syndrome, a treatment that, in fact, is established as effective and safe and is even included in the “individualization” recommended by health agencies. It is impossible to tell how much harm has been done but retraction of the paper will be a start to correction.
But Feinman told us he doesn’t think the journal will retract the paper:
The Journal will not retract the paper since [there] is no evidence of fraud or falsified data. I will not pursue it since it is [biased] and poor judgement that is involved. It’s real importance is an example of the poor standards in the literature.
A spokesperson from Nutrition and Diabetes added:
We can confirm that Nutrition & Diabetes received the letter from Dr Feinman. We take issues related to research published in the journal seriously and we have offered Dr Feinman the option to submit a formal “Letter to the Editor” for publication in the journal.
Harcombe, on the other hand, said she doesn’t think the paper should be retracted:
I’m not a fan of censorship — although that is what the conventional side tries to do with our “unconventional” articles all too often.
Hat tip: Ben Gallarda
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
My very first paper in my next life will be entitled, “Nutrition advice upregulates bovine credulity in humans.” In the real lives of patients who are being harmed by their obesity, differences in dietary composition make so little difference that this whole argument should be viewed as a distraction. Emphasis on tweaking dietary composition is the mother of hundreds of bogus, promotional books.
Tom, may I say:
AMEN, AMEN.
This entire story highlights what I deeply believe about nutritional science: Most of it is voodoo. And even the parts thtat are not are obviously ineffective in real life.
Anyway, a call for retraction here is downright ridiculous. Unless you have reason to believe that the results were generated erroneously, you have to live with the results even if they seem to “make(s) no sense at all”, as Dr. Feinman said.
I am not saying that this particular paper represents great science. I did not look into it and I simply do not know. But you can alway challenge the wording of conclusions and discussions, whether or not the data really supports the conclusions and so on. But if every paper with disagreement is retracted, there will be not much literature left…..
As crappy as this science may be, I think I agree with the editors that there are no grounds for retraction. However, this is the exact kind of thing that plays to the strength of post-publication peer review. The detractors should write their concerns in full on pub-peer and let science take its course.
Ancel Keys and Walter C. Willet and their “food issues think tank” all over again. The effects on public health damaged by the lunacy of “high carb, low saturated fat” diet on current generation should be enough. Healthcare will be paying $$$ long time for the past mistakes in public nutrition policies.
THANK YOU! So much ink and bandwidth wasted on silly diet “information.”
Both sides are sounding hyperbolic, at least as far as I can tell from the quotes above. For example, the statement “You only study mice when it would be unethical to study humans” is misleading. A mouse study may be a prelude to mechanistic studies; only the latter might be unethical to do on humans but you wouldn’t start there. But meanwhile, Andrikopoulos’s comments certainly seem to be going far beyond the totality of the data. I wish scientists would resist the urge to puff up the importance of their work when the newspaper starts asking questions.
I did not suggest retraction as censorship. I specifically said the paper should be re-submitted and the experimental work evaluated. The hyperbole in interpretation is harmful to patients. Minimally, the statement that “there is no scientific evidence that these diets work” should be removed. Our review on diabetes (Feinman, et al. Nutrition, 2015), cited by Lamont, et al. is not sensibly called hyperbole. We laid out the 12 points of evidence and provided examples and references. We do not think that we know all the answers and specifically sought feedback. Lamont, et al. listed the reference but did not address any of the 12 points.
Steven St. John: In private email to the editor I pointed out that we do animal studies that practically or ethically cannot be done on humans or where we are interested in species differences and similarities. Lamont, et al., like work in my own lab (Borghjid & Feinman, Nutrition&Metabolism, 2008) showed that mice are a very poor model for response to diet. We think that the origins rest with the very minuscule rodent brain and the limited demand on insulin. So, it is almost impossible to study a low-carbohydrate diet as analog for humans.
In the NZO mouse model of diabetes, only the male mouse is susceptible to obesity-induced diabetes. Females are immune to this effect. All of the mice in this experiment were male mice – yet the conclusions and press release did not specify that the results only applied to males.
Of course it would have been ridiculous to do so, but no more ridiculous than stating the conclusions that were somehow reached.
It seems to have escaped the attention of some commenters on this post that this was an experiment supposed to shed light on the causes of diabetes. It was not intended as a “weight loss diet” test in any way. As a diabetes-and-diet experiment it does of course run counter to the results of human experiments going back for a couple of centuries.
The question is whether, under any circumstances, an experiment in (male but not female) mice (with a genetic defect never seen in humans) can be used to refute the consistently established results of human experiments.
For example, if it were discovered that cigarette smoking renders the (male) NZO mouse immune to cancers, what would we think of a scientist using this information to criticise current human recommendations?
If it were a weightloss experiment it would also run counter to human experiments, but that is another matter, and a very easy one to test at home. No-one is writing a diet book for mice at the present point in time.
Our letter in response to this study was published by Nature Nutrition and Diabetes.
http://www.nature.com/nutd/journal/v6/n8/full/nutd201631a.html
I think that I was wrong in calling for retraction in the sense that, as I said above, there was “no evidence of fraud or falsified data” but I didn’t see how one could deal with such an egregious breakdown in scientific judgement. On reflection, however, it is that the paper was not subject to real peer review which would sensibly have included comments from people who had actually studied low-carb diets in humans, (or even in mice but who had come to different results or interpretation). Comments from me and others noted above, are what a reasonable reviewer would have said. That is different than saying that I am right but the editor would then have had the task of evaluating conflicting reviews.
Whether Richard Atkinson, the editor, did not recognize that this was a controversial subject (unlikely), or himself maintains bias against low-carb diets (possible — refusal to face the benefits of such diets has a long tradition in medicine) or whether he just did not adequately attend to the problem (most likely) is unknown, but that is where the buck stops. He is the gatekeeper. It seems reasonable however that mIistakes should be recognized and corrected. Post-publication peer review can put things in perspective, as suggested by SB above, although I think PubPeer, while a step in the right direction, does not provide the critic with equal footing. I have suggested to Dr. Atkinson that he invite such a post-publication review from me or one of the authors in our diabetes review (reference (5) in Lamont, et al.
There remain other big issues along the lines of citing conflicting data in the literature, or even data that show consistency but are more reasonably interpreted. In general, what level of breach of research integrity applies to failure to cite clearly relevant literature. The study his question brings out s many of the excesses that have made the medical literature the target of serious scientific criticism and have encouraged the loss of faith of the public in medicine, especially in nutrition.
New rule: Editors have the obligation to recognize controversial subjects and to seek reviews from both sides of the controversy. Failure to do so constitutes de facto bias.