Fact Check: Eat Fat, Cut The Carbs and Avoid Snacking To Reverse Obesity and Type 2 Diabetes (Section 1 & 9)

The joint report Eat Fat, Cut The Carbs and Avoid Snacking To Reverse Obesity and Type 2 Diabetes published by selected members of the National Obesity Forum (NOF) and Public Health Collaboration (PHC) has caused quite a stir. A number of the NOF board have now resigned with some distancing themselves entirely from the report and the The Royal Society for Public Health went as far as labelling it a “muddled manifesto of sweeping statements, generalisations and speculation”. Ouch.

I thought it would be worth doing a quick fact check of some of the claims made in this report. To avoid the inevitable strawman criticism which will come my way, I am not asserting that the overall message of the report is right or wrong, nor am I advocating for a particular diet in this post.

What I am saying however is that if you wish to make a coherent argument it should be i) internally consistent and ii) supported by citations which say what you claim they say.

Comments are enabled so let me know if I have made any errors myself, or if you spot anything I have missed. I’ll be happy to correct.

Section 1. Eating Fat Does Not Make You Fat

“Evidence from multiple randomised controlled trials have revealed that a higher fat, lower carbohydrate diet is superior to a low-fat diet for weight loss and cardiovascular disease risk reduction [8, 9]”

The first reference [8] is a meta-analysis by Sackner-Bernstein et al. The cited paper does indeed conclude that Low-Carb diets are superior for weight loss and reducing markers of CVD risk. However the nuance that their superiority is “numerically modest” is excluded from the NOF/PHC report:

The second reference [9] also supports that claim. A minor gripe here though, if we are making the case about multiple trials showing benefit why is this particular individual Randomised Controlled Trial (RCT) Bazzanno et al  worthy of note when there are meta-analyses available?

“An exhaustive analysis of 53 randomised controlled trials involving 68,128 participants conducted by the Harvard School of Public Health concluded“ when compared with dietary interventions of similar intensity, evidence from randomised controlled trials does not support low fat diets over other dietary interventions for long term weight loss. In weight loss trials, higher fat weight loss interventions led to significantly greater weight loss than low fat interventions.”

No citation is provided for this claim. Sloppy, but not a big problem as it’s a recent and relatively easy to find paper with the information provided: “Effect of low-fat diet interventions versus other diet interventions on long-term weight change in adults: a systematic review and meta-analysis” by Tobias et al. The quote cited does indeed appear in this meta-analysis.

Again, a minor gripe but the Bazzano et el paper cited by the NOF/PHC separately as reference [9] is one of the papers considered within this meta-analysis, so there is a minor element of double counting of evidence here. Its hardly significant, but it is indicative that a systematic approach has not been taken by NOF/PHC in term of identifying and presenting the body of evidence.

One other point which we will come back to is that this meta-analysis considers not just low-carb, high fat and low-fat dietary comparisons – where data is available – but also separates the trials into studies aimed at weight loss (with and without weight loss goals) and weight maintenance.

Things now begin to go downhill as they focus on one particular trial:

“Furthermore the Women’s Health Initiative was the largest randomised controlled diet trial ever performed”.

Its not clear why the NOF/PHC feel that this RCT trial in particular needs to be discussed further, especially when it was another RCT which was included within the previously cited meta-analysis by Tobias et al. 

48,835 post-menopausal women were randomised to either their usual diet or a low-fat, calorie reduced diet with increased exercise with the hypothesis that this would reduce cardiovascular disease.

Inaccurate in one important aspect. According to the citation it was a dietary modification trial in which participants were counselled to change the composition of their diet but not to reduce energy intake (i.e. there were no weight loss goals):

“The intervention achieved an 8.2% energy decrease in total fat intake and a 2.9% energy decrease in saturated fat intake, but did not reduce risk of CHD or stroke.[10]”

Correct.

While not specifically a weight loss trial, nevertheless, the reduction in dietary fat and total daily calories (361 calories/day reduction) also failed to produce any significant weight loss over the duration of the study.

It was not a weight loss trial, it was a weight maintenance trial.

The reduction in daily calories does not appear in the cited paper anywhere. There is however a separate paper from the Women’s Health Initiative trial titled “Low-Fat Dietary Pattern and Weight Change Over 7 Years” so it looks like a citation error. The reduction in calories is taken from Table 2 of this paper and it represents the final nutrient intake (baseline minus follow-up) estimated by Food Frequency Questionnaire (FFQ):

It is extremely misleading to take a figure from the follow-up FFQ  and present this as if this was a calorie deficit which was achieved and maintained for the duration of the trial (“361 calories/day reduction”). It gives a false impression that large and sustained energy deficit was achieved, yet significant weight loss was not. We also don’t know, of course if participants were at energy balance at baseline.

It is even more misleading when you consider that data obtained from FFQs are well known to have weaknesses and that the authors of the paper explicitly highlights this issue to prevent misinterpretation:

This isn’t the first time the PHCUK have misused self reported data without acknowledging the uncertainty in the measurement – even when the dangers of this are explicitly stated by their source.

As for the claim that Women Health Initiative trial failed to “produce any significant weight loss over the duration of the study” I’ll leave you to decide if that claim is true:

(But the answer is: no, it’s entirely untrue).

As an aside, the dietary pattern and weight change paper is well worth reading in full because it also includes statements like the one below which are, ahem…let’s say inconvenient to the pro-fat argument that the NOF/PHC are trying to craft:

“This rejected the notion that the low-fat diet is either beneficial for cardiovascular disease or weight loss.”

I think the NOF/PHC have a problem here with the consistency of their argument. It’s rather strange to put forward the argument that low fat diets are “not beneficial for cardiovascular disease or weight loss” based solely on one (albeit large) RCT when they have already cited other evidence which tends to refutes that claim.

Remember the Sackner-Bernstein et al meta-analysis cited at the start of this section of the NOF/PHC  report which they used to argue for the superiority of low-carb diets over low-fat diets? It concludes:

What about the Tobias et al meta-analysis – which includes the WHI study – and looks at weight loss and weight maintenance? It shows that the low fat diet is superior to the standard diet for weight loss (with and without goals) and for weight maintenance, and that they were broadly not significantly different from higher fat diets unless they also restrict carbohydrate.

“We recommend that guidelines for weight loss for the UK should include an ad libitum low refined carbohydrate and a high healthy high fat diet (i.e non-processed foods or “real” foods) as an acceptable, effective and safe approach for preventing weight gain and aiding weight loss.”

It is unclear which data is being used to support the use of low refined carbohydrate diets for long term weight maintenance. According to the Tobias et al meta-analysis cited by NOF/PHC “no long-term non-weight loss or weight maintenance trials compared low-fat with low-carbohydrate dietary interventions”:

Section 9. Snacking will make you fat (Grandma was right!)

There have been two major changes in our dietary habits since the 1970s, prior to the onset of the obesity epidemic. The change to a high carbohydrate, low fat diet has been well documented and has played an important role in causing obesity.

Assertion and questionable language. The change to a high carbohydrate diet is causing, or is associated with increasing obesity? (Answer: the latter, if you care about not committing the post ergo propter hoc fallacy).

The other change, the increase in meal frequency plays an equal if not larger role and has been largely ignored. In the 1970s, the average number of eating opportunities was three – breakfast, lunch and dinner. Fast-forward to 2005 and that number has almost doubled. [40]

If I’ve understood this properly the position taken by the NOF/PHCuk is that obesity in the UK has been driven not by increasing calorie intake, but by an increase in insulin secretion caused by frequent high carbohydrate meals. The citation leads to this cross sectional study. The cited study does indeed confirm that frequency of eating opportunities has increased since the 1970’s, but in the USA.

If we accept that the UK and USA are broadly similar we are still left with a couple of issues before we accept the above argument: i) cross-sectional studies cannot show causation ii) the citation makes the case that the increase in eating opportunity in the USA also correlates with an increase in overall energy intake (i.e. not one of the two major changes to diet identified by the NOF/PHC):

To further developing the idea that meal frequency is one of the major factors in increasing obesity the following claim is made:

“Eating continuously from the moment we arise to the moment we go to sleep does not allow our body to digest and use some of the foods that we eat. The entire day becomes an opportunity to store food energy without a chance to burn it. Eating six times a day does not result in weight loss (41), but tends to increase overall consumption of food.”

The third sentence is interesting and the citation for this is “Increased meal frequency does not promote greater weight loss in subjects who were prescribed an 8-week equi-energetic energy-restricted diet” by Cameron et al. This is a small RCT (n=16) looking at the hypothesis that a high meal frequency (MF) might lead to a greater weight loss than that obtained with a low MF under conditions of similar energy restriction.

The absolute claim that “eating six meals a day does not result in weight loss” is directly contradicted by the citation. Over eight weeks participants in the high-MF group (3 meals/day and 3 snacks/day) did in fact lose weight:Weight loss is entirely unsurprising, since despite having six meals per day the protocols were designed to achieve an energy deficit in both low-MF and high-MF groups:

The claim that six meals per day “tends to increase overall consumption of food” might be true, but due to the design of this study, which involved the participants actively trying to lose weight by adhering to a prescribed calorie deficit, it cannot support that claim.

Authorship and Peer Review

While initially the document was issued without any acknowledgement of authorship we now know via a statement on the NOF website that a large number of individuals were co-authors. Some of these have competing interests (e.g. David Haslam sits on the scientific advisory board for Atkins Nutritional, Sam Feltham has a diet book for sale based on the principles in this report) and for transparency these really should have been identified in a statement of competing interests.

It seems unusual that there was also an “expert panel” who “supported” the paper as well as “peer reviewing” it before publication? Where is the line drawn between support and co-authorship? – this isn’t at all clear.

Some of the people are listed as “co-author” and “peer reviewer” (is that even possible?) while others are clearly not experts – “Film maker and health activist”, “Best-selling author and health activist”.

I think it would have been better to simply say the paper has not been externally peer reviewed and leave it at that.

Conclusion

Having reviewed a couple of sections of the report in detail and skimmed the remainder the bias within this document is plain to see and there are serious issues about the way this report was authored and checked.

There is little evidence that a systematic approach has been taken to identify and appraise relevant evidence to inform their claims and it relies quite heavily upon simple unreferenced assertion to make the case for higher fat diets.

The benefits of high fat and low-carb diets are presented in a way which would lead the casual reader to expect major benefits and superiority over alternatives, when in fact this is often shown to be marginal, or there is simply insufficient data to justify the conclusions drawn by NOF/PHC  (e.g. low-carb diets for weight maintenance).

It’s apparent that there is a sloppy approach to citations, and some of the claims made are just plain wrong. It’s baffling that this large cohort of co-authors and an equally large cohort of “peer reviewers” didn’t check the claims they were making sufficiently, or possess sufficient understanding/expertise or impartiality to interpret the data in a more even handed manner.

The NOF/PHC need to acknowledge and correct the obvious errors within this report.

The Public Health Collaboration UK: Wrong about energy intake

Yesterday saw the launch of a funding campaign on indiegogo to establish a new charity called the Public Health Collaboration UK.

Without wishing to pre-judge, this group of doctors and low carbohydrate diet proponents who aspire to:

……collaborate our efforts into one singular organisation to inform the public and empower the medical community on the science and solutions of health. ¹

……seems to look suspiciously like a thinly disguised lobby group who’s activities will produce opinion papers which will – inevitably – be sexed up to the level of fact on the BBC Breakfast sofa and in press releases, by the media savvy members to promote their diet de jour.

So what are they claiming?

…….in the UK 25% of adults are obese, the highest prevalence in Europe, and type 2 diabetes has risen by 65% in the past 10 years with no sign of slowing down. Together they cost the NHS £16 billion a year and the UK economy at large £47 billion a year. ¹

Strong start.

These perilous percentages and shocking statistics have presented themselves despite the fact that as a population we are closely following the dietary advice that is being recommended to us. ¹

Forgive me just one argument from incredulity, but really? We’ve followed the guidelines closely? 

I’m going to need some evidence for that.

Based on the latest National Diet and Nutrition Survey published in 2014 by Public Health England, our total food consumption is on average 383 calories below the recommended……¹

Jesus. They appear to have taken the data literally.

A quick skim through the National Diet and Nutrition Survey report itself reveals the following:

Dietary surveys are reliant on self-reported measures of food intake. Misreporting of food consumption, generally underreporting, in self-reported dietary methods is a well-documented issue. The under-reporting of energy intake (EI) is known to be an issue in past and current NDNS, as for all dietary surveys and studies. This is an important consideration when interpreting the findings from this survey.²

The energy and nutrient intakes presented in this report have not been adjusted to take account of underreporting ³ 

It is not possible to extrapolate this estimate of underreporting to individual foods and nutrients because they may be affected differentially. ³

The report also discusses how the doubly labelled water technique was used to validate the figures for a sub set of the surveys and it highlighted potentially large discrepancies between reported energy intake (EI) and total energy expenditure (TEE):

In the NDNS RP, estimates of EI from the four-day diary were compared with measurements of total energy expenditure (TEE) using the DLW technique in a sub-sample of survey participants. The results of this analysis indicated that reported EI in adults aged 16 to 64 years was on average 34% lower than TEE measured by the DLW technique, 12% lower in children aged four to ten years, 26% lower in children aged 11 to 15 years, and 29% lower in adults aged 65 years and over.  ³

In short, these figures might show trends or where intakes of certains aspects of diet are below recommended levels but they are not robust enough to conclude that the UK is getting more obese and (T2) diabetic despite eating less. And if you must use the data state the level of uncertainty in the data. I raised the issue on twitter with the collaboration of experts but none were willing to acknowledge or correct the erroneous claim on energy intake which does not bode well for a wannabe charity.

It is also quite surprising to see the collaboration citing this as evidence when Exit Door favourite Aseem Malhotra has previously stated on national television that this type of survey is:

….heavily flawed because it relies on personal reporting which we know classically under reports calories consumed….

I’m not detecting much of a collaborative effort here – did Aseem even read this before release, and if so – did he recognise the mammoth level of hypocrisy involved in putting his name to this?

The call for funding goes on make further broad, unsubstantiated ideologically driven claims without even attempting to evidence them:

…..the Eatwell plate and simple calorie restriction, that have been used for the past 20 years with no improvements in public health ¹

All of which is irrelevant conjecture until you have completed the monumental first step of demonstrating that the population have been following the guidelines throughout this period (good luck with that one).

I sense entertainment lies ahead – it will be interesting to see who of the experts are willing to be a trustee of a charity which is already playing fast and loose with the evidence, and more interestingly one whose begging bowl approach to funding seems to places their dubious claims within the remit of the Advertising Standards Authority.

---

References

¹ https://www.indiegogo.com/projects/public-health-collaboration#/

2 National Diet and Nutrition Survey Results from Years 1, 2, 3 and 4 (combined) of the Rolling Programme (2008/2009 – 2011/2012) p74 

3 National Diet and Nutrition Survey Results from Years 1, 2, 3 and 4 (combined) of the Rolling Programme (2008/2009 – 2011/2012) p75 

 

Nina Teicholz and the BMJ mislead about the US Dietary Guidelines and low carbohydrate diets

 I’ve not blogged for a long while but Nina Teicholz seems to be making quite a splash with her latest BMJ piece ¹ . The bit that stands out for me – in a piece which is supposed to be showing the weakness of the science behind the 2015 Dietary Guidelines Advisory Committee (DGAC) Scientific Report – is how this piece is used (and the BMJ) to attack recommended diets whilst surreptitiously promoting the idea that low carbohydrate diets are the hard done-by, ignored, always-the-bridesmaid-never-the-bride solution to our problems.

The sophist begins

Another important topic that was insufficiently reviewed is the efficacy of low carbohydrate diets.

Low carbohydrates are certainly an important topic, the DGAC states they have been of “public interest” ² .

Again, the 2015 committee did not request a NEL systematic review of the literature from the past five years.

“Again” seems to imply a continued refusal to perform a review of the literature. This is of course misleading – a review specifically of low carbohydrate diets was unlikely to be performed in 2015 – because the DGAC is moving away from percentages of fat/carb/protein, and moving towards recommending dietary patterns¹  (defined as “the quantities, proportions, variety or combinations of different foods and beverages in diets” ²).

In fact the NEL Systematic Review which Teicholz implies should have been carried for 2015 would have been rather difficult as the DGAC report clearly states that the low carbohydrate studies “generally did not meet the DGAC’s definition of a dietary pattern study unless a full description of the dietary pattern consumed was provided and appropriate methods were used to adjust for the confounding of foods and nutrients”.¹

Teicholz in a roll here though, and continues with the DGAC are-ignoring-low-carbohydrate-diets spin

The report says that this was because, after conducting “exploratory searches” of the literature since 2000, the committee could find “only limited evidence [on] low-carbohydrate diets and health, particularly evidence derived from US based populations yet many studies of carbohydrate restriction have been published in peer review journals since 2000, nearly all of which were in US populations.

This is again misleading, as it makes it sound like the DGAC were simply dismissive of the benefits of low carbohydrate diets. However the report states that “The most evidence available focuses on low-carbohydrate diets and body weight. The 2010 DGAC examined the relationship between macronutrient proportion and various body weight outcomes“. The DGAC did in fact carry out a previous systematic review on weight loss, which they lean on in the 2015 report and state there are no further studies which change the conclusions.

This is probably why the 2015 DGAC report specifically identifies “Low-carbohydrate (initially less than 20 g/day carbohydrate) diet without formal prescribed energy restriction but realized energy deficit” as one possible method of weight loss.

So after her attempt to portray the DGAC as ignoring low carbohydrate diets, Teicholz offered some evidence of her own and a counter argument: there are lots of studies since 2000 showing benefit of low carbohydrate diets, you just need to look!

These include nine pilot studies, 11 case studies, 19 observational studies, and at least 74 randomised controlled trials, 32 of which lasted six months or longer (see table C on thebmj.com)”

Table C: Published research on low carbohydrate diets ⁴ is full of bloat, and many of the studies cited are totally irrelevant to the questions the DGAC were considering. For example the first study on her list: did an inconclusive case study of 5 people with reflux ⁵ who self administered a low carbohydrate diet really merit formal review for the guidelines?

Do any of the cited studies in Table C provide evidence about the relationship between low carbohydrate dietary patterns and:

• Risk of cardiovascular disease? (question 1)
• Measures of body weight or obesity ? (question 2)
• Risk of type 2 diabetes? (question 3)
• Cancer? (question 4)
• Risk of congenital anomalies? (question 5)
• Risk of neurological and psychological illnesses? (question 6)
• Bone health? (question 7)

Because they were the specific questions which the evidence needs to address, everything else is simply window dressing.

What about some of her other citations, the effect of low carbohydrate diets on epilepsy (numerous), migraines, physical activity, hunger, IBS, and rare conditions like Sturge Weber Syndrome? Are these generally applicable to the population and relevant to answering Questions 1-7?

The task of the DGAC was not to review what Teicholz calls the “many studies of carbohydrate restriction” and dredge for benefits. Does she believe an NEL should be carried out on every medical condition, just to satisfy her that some benefit of her favoured diet has not been cruelly overlooked?

In addition for someone preaching about high quality evidence why is she even listing case studies in Table C anyway? These would fail to meet the standards required for being considered in the NEL systematic review she implies was required. And why does someone who is adamant that conflicts of interest should be avoided produce a table of studies so heavily infected with Atkins Foundation funding?

If only Teicholz was intellectually honest enough to apply the standards to her own work which she demands of others.

Thats not to say she’s doesnt make some valid criticisms

The report provides no documentation of these “exploratory searches,” 

With hindsight this was a mistake.

They should have listed the papers found since 2000 on low carbohydrate diets and health in questions 1, 3, 4, 5, 6, 7 and perhaps those on body weight (question 2) since the previous guidelines.

By not doing so, they have allowed Teicholz and the BMJ to create an impression that the guidelines are ad hoc and not based in science and confuse the public further.

Mission accomplished for the headline grabbers and book sellers.

---

References

¹  The Scientific report guiding the US dietary guidelines: is it scientific? BMJ 2015;351:h4962 http://www.bmj.com/content/351/bmj.h4962

² Dietary Guidelines Advisory Committee. Scientific report part D: chapter 2. Dietary patterns, foods and nutrients, and health outcomes—continued. http://health.gov/dietaryguidelines/2015-scientific-report/07-chapter-2/d2-2.asp.

³  Dietary Guidelines Advisory Committee. Scientific report part D: chapter 2. Dietary patterns, foods and nutrients, and health outcomes — introduction. http://health.gov/dietaryguidelines/2015-scientific-report/07-chapter-2/

⁴  The Scientific report guiding the US dietary guidelines: is it scientific? BMJ 2015;351:h4962 http://www.bmj.com/content/bmj/suppl/2015/09/23/bmj.h4962.DC1/teicholzmaster2609.wt3_default.pdf

5 Improvement of gastroesophageal reflux disease after initiation of a low-carbohydrate diet: five brief case reports (PMID:11712463) http://europepmc.org/abstract/med/11712463

Eat Like an Ancient Egyptian

Nina Teicholz  is again busy rewriting history to make it fit better with her hypothesis that a diet heavy in meat is the best for health:

Ancient Egyptian tomb find: cattle barn. How we used to eat. We forget history at our peril. pic.twitter.com/02IRMwXi2b

— Nina Teicholz, PhD (@bigfatsurprise) March 27, 2015

Ancient Egyptian tomb find: cattle slaughtered and blood collected. How we used to eat. Forget history at our peril pic.twitter.com/69s1fClNpy

— Nina Teicholz, PhD (@bigfatsurprise) March 27, 2015

Lets deal first with the premise that these two images are representations of how we used to eat. The models in question are funerary objects which were found in the tomb of an Egyptian noble called Meketre¹, who was a chancellor and chief steward during the reign of Mentuhotepo II and III during the Middle Kingdom (between about 2000 BC and 1700 BC).² 

What is funny here is that Teicholz blatantly omits other parts of the find from the very same tomb, which are housed in the very same Museum, which run counter to her thesis. For example the tomb also contained a model of a bakery and brewery ³ :

There is also a model of a granary ⁴ which according to the Museum description includes an “…accounting area…[…]…Keeping track of grain supplies was crucial…”:

This should come as no surprise, since the Egyptians were an agricultural society who pretty much invented , and their success was attributable to large scale agriculture and irrigation techniques which they used to exploit the nutrient rich flood areas surrounding the Nile to grow crops.

It also seems that Teicholz fails to understand the purpose and context of these objects. These are not necessarily a record of what everyday people were eating at the time, they were specifically placed in the tombs of the elite – nobles and priests – along with actual food to sustain them in the afterlife, in the way in which they had become accustomed to during life.

Rather than showing how we used to eat, the “we” being the general population, these objects (including those awfully inconvenient depictions of grains) should more accurately be labelled: how the Ancient Egyptian elite ate.

Finally, lets looks consider the idea that when it comes to diet we forget history at our peril. While this appears to be a simple fallacious appeal to tradition (there is little reason to conclude this way of eating was the best for health back then, or that it is now) I’m going to agree with Teicholz on this one.

With this in mind, I look forward to hearing her thoughts on research ⁵ which shows that the mummified remains of Ancient Egyptian Priests (who ate the fairly well documented food offerings which were made to the gods – a meat heavy 50%+ fat diet high in saturated fat) showed the signs of what looks suspiciously like vascular calcification.

---

References

¹ http://en.wikipedia.org/wiki/TT280

² http://en.wikipedia.org/wiki/Middle_Kingdom_of_Egypt

³ http://www.metmuseum.org/collection/the-collection-online/search/544258

⁴ http://www.metmuseum.org/collection/the-collection-online/search/545281

⁵ http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(10)60294-2/fulltext

A letter to the BMJ: Lets talk about Hansen!

Over the past few weeks I have increasingly found myself – of my own volition I might add – whack-a-mole-in’ fallacious claims about the Swedish Government recommending a low carbohydrate diet.

I have traced the source of many of these claims as a particularly bad BMJ news article by Anders Hansen, so I thought it might be more productive to cut from the root and write to the BMJ calling for retraction, or substantial correction.

Now I know writing whiny complaint letters is more Clark Kent than Superman, but sometimes a few key points from a moaning pedant is a practical way of tackling the great menace of inaccuracy. So, lets see how we get on with this:

Dear BMJ

I am writing to express my concern regarding the news article “Swedish health advisory body says too much carbohydrate, not fat, leads to obesity”. ¹

While I appreciate this is a now a historical news article from 2013, it contains numerous factual inaccuracies which continue – based on the reputation of the BMJ as a trusted source  – to be widely propagated and cited, wrongly, as conclusive evidence that low fat diets are ineffective for the treatment of obesity.

Indeed, despite the attempts of the chair of the HTA committee Nina Rehnquist to clarify matters in her rapid response to the article, it has now influenced a much wider audience, as it was cited in the best selling book ‘The Big Fat Surprise’ as the source of the following erroneous claim:

“…in 2013 in Sweden, an expert health advisory group, after spending two years reviewing 16,000 studies, concluded that a diet low in fat was an ineffective strategy for tackling either obesity or diabetes.”

Having read the english summary of the Health and Technology Assessment, I believe that the level of inaccuracy in this news piece combined with its now wide audience requires the BMJ to either retract the article, or substantially correct both the headline and content. Firstly, the title of the piece is inaccurate:

Swedish health advisory body says too much carbohydrate, not fat, leads to obesity

The Health and Technology Assessment cited did not consider the cause of obesity, only the treatment of obesity a point made clear by both the title of the report ² and Nina Rehnquist’s rapid response.

As the article metrics show, the vast majority of readers have seen only the abstract of this article and I feel that this inaccurate title leaves the piece open to being mis-cited. With regards to the content of the article, there are a significant number of inaccuracies. The first line claims:

An influential Swedish health organisation has recommended a diet that is low in carbohydrates but not low in fat for people who are overweight or obese or have diabetes.

Unfortunately, the report did not consider the treatment of diabetes because this was the subject of an earlier separate Health and Technology Assessment in 2010.³ It is worth noting however that the earlier report on diabetes did not recommend a diet “low in carbohydrates but not low in fat”, in fact is states something quite different:

In type 2 diabetes, low-fat and moderate low-carbohydrate diets (30–40% of the energy from carbohydrates) have similar, favorable effects on HbA1c (long-term blood glucose) and bodyweight. The absence of sufficient-quality studies in people with diabetes prevents evaluation of the long-term effects of more extreme diets involving low-carbohydrate and high-fat intake, eg, so-called “low-carb, high-fat” (LCHF) diets. Hence, safety aspects become particularly important in clinical follow-up of individuals who choose extreme low-carbohydrate diets (10–20% energy from carbohydrates).

With regards to the claim that a diet “low in carbohydrates but not low in fat” was recommended for the treatment of obesity, this is correct only in the short term, and the report has more nuanced conclusions which suggest a range of equally effective options over the longer term:

Weight loss in adults. A range of advice on alteration of eating and drinking habits can result in obese individuals losing weight or reducing their waist size. In the short term (six months), advice on strict or moderate low carbohydrate diets is a more effective means of achieving weight loss than advice on low fat diets. In the long term, there are no differences in the effect on weight loss between advice on strict and moderate low carbohydrate diets, low fat diets, high protein diets, Mediterranean diets, diets aimed at achieving a low glycaemic load or diets containing a high percentage of monounsaturated fats. Advice on increasing the intake of dairy products (primarily milk) or reducing the intake of sweet drinks may also lead to weight loss.

Hanson further states:

The guideline advises that meat and fish rich in fat, along with nuts and olive oils, should form a large part of a healthy diet, while the consumption of pasta, potatoes, and white bread should be reduced.

No such recommendations are made within the report. In fact there were virtually no findings on individual foods, as the HTA report ² clearly states:

However, all in all these studies provide no clear evidence for advice on individual foods for obese individuals in order to prevent morbidity or achieve weight loss.

In addition I can find no support in the HTA for the following claim made by Hansen:

The recommendation contradicts the generally held belief that people should avoid foods that are rich in fat, especially those high in saturated fat.

The HTA report ² itself stating:

The studies relating to strict low carbohydrate diets which were included in the report give no indication of whether low carbohydrate diets should provide small portions or not include saturated fat…[…]…is not possible to draw any conclusions on the link between low carbohydrate diets – irrespective of fat content type – and cardiovascular morbidity. The precautionary principle could be applied here. This may result in restraint on the intake of saturated fat when advice is given on low carbohydrate diets, as long as the documentation on the long-term effects is so inadequate.

I would hope that you agree that there are serious inaccuracies in this article which continue to compromise the reputation of the BMJ. I would ask that you consider a formal retraction, or substantial correction of the article to address the above points.

Regards,

Slipp Digby

References

¹ Swedish health advisory body says too much carbohydrate, not fat, leads to obesity. BMJ 2013;347:f6873

² Swedish Council on Health Technology Assessment. Dietary Treatment of Obesity: A Systematic Review (No 218/2013), September 2013, ISBN: 978-91-85413-59-1.

³ Swedish Council on Health Technology Assessment. Dietary Treatment of Diabetes: A Systematic Review (No 201), August 2010, ISBN: 978-91-85413-37-9.

Update 30/5/2015

In response to my letter the BMJ have issued a very substantial correction which can be found here:

In this News story, “Swedish health advisory body says too much carbohydrate, not fat, leads to obesity” (BMJ2013;347:f6873, doi:10.1136/bmj.f6873), the headline and some of the text were incorrect. The report did not say that too much carbohydrate leads to obesity, as stated in the headline. It said that low carbohydrate diets were more beneficial for reducing obesity in the first six months of treatment, when compared with low fat diets, but made no difference at 12 months.

The report said that, in the longer term, “there are no differences in the effect on weight loss between advice on strict and moderate low carbohydrate diets, low fat diets, high protein diets, Mediterranean diets, diets aimed at achieving a low glycaemic load, or diets containing a high percentage of monounsaturated fats.” The report did not conclude that “the scientific evidence did not support a low fat diet.”

In addition, the report made few recommendations with regard to specific foods and did not say that “the consumption of pasta, potatoes, and white bread should be reduced.” We apologise for these errors.

Kudos to the BMJ for making a very full and formal retraction of these erroneous claims.

Finally an errata of my own! – In my original letter when referring to Nina Renquist’s rapid response I state “her” when it should in fact be “his” response. Apologies.