Low carbohydrate-high protein diet and incidence of cardiovascular diseases in Swedish women: prospective cohort study

BMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e4026 (Published 26 June 2012)
Cite this as: BMJ 2012;344:e4026

Recent rapid responses

Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on bmj.com. Although a selection of rapid responses will be included as edited readers' letters in the weekly print issue of the BMJ, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window.

Displaying 1-10 out of 14 published

Congratulations for publishing such a pain staking study! This article has generated lot of interest. I feel that this observation study with an average follow-up of 15.7 years is one of a kind. Most available studies are short term and often ignore a lot of clinically important issues and leave readers as confused as before. It would be nice to have similar studies in other developed countries. If you refer to WHO Country profile 2011, in most countries where obesity is prevalent physical inactivity is one of the major concerns. Let us not confuse the clear mandate that to lose weight, eat a balanced diet and increase physical activities. This study is comparable to DCCT and UKPDS, which were pioneer studies that cleared up so many doubts among practising physicians about microvascular and macrovascular complications in Diabetes.

Competing interests: None declared

R.K Saxena , Internist and Geriatrician

nil , Post Box 11137 , Wellington, 6142 , New Zealand

Click to like:

Reply to letters by Drs. Campillo-Soto and Freedhoff

In his criticism of our paper Dr. Campillo-Soto repeatedly invokes “biases”, but uses a technical epidemiologic term rather loosely and unconventionally, to say the least. Selection bias is not really a concern in a cohort study. Nor is measurement bias - assessment of exposures cannot be differentially affected by the outcome when the latter is not known at the time the exposure is reported and non-differential misclassification is expected to attenuate a true association. In his reference to “compliance bias” and “expectation bias”, again interesting terminology, Dr. Campillo- Soto confuses interventional with observational epidemiological study designs. As for follow-up bias, it is relevant when losses to follow up are concerned, which is not the case in our study - linkage to registries allowed us practically complete follow up of all our study subjects.

Criticizing our paper, Dr. Freedhoff indicates what is already stated in our manuscript: that the diets of our study population are less extreme in their low carbohydrate content compared to advertised dietary regimes. The associations we detected, however, are monotonic, indicating that the risk is expected to be higher in more extreme low carbohydrate-high protein regimes. We disagree that the study is misinformative – all aspects are clearly presented in the manuscript, but more importantly, we disagree that the increase in risk is miniscule.

Both Dr. Campillo-Soto and Dr. Freedhoff criticize our use of a food frequency questionnaire, but they disregard the fact that such questionnaires, notwithstanding their limitations, are a standard tool used in large nutritional epidemiologic cohort studies. Also, both Dr. Campillo-Soto and Dr. Freedhoff point out that the assessment of the diet at recruitment only in our study generates misclassification. We agree, but in a cohort study this misclassification is non-differential and, thus, much more likely to attenuate an existing association rather than generating it.

Pagona Lagiou1, 2
Sven Sandin3
Marie Lof3, 4
Dimitrios Trichopoulos2, 5
Hans-Olov Adami2, 3
Elisabete Weiderpass3, 6, 7

1 Department of Hygiene and Epidemiology, University of Athens Medical School, 75 M. Asias Street, Goudi, GR-115 27, Athens, Greece.
2 Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA.
3 Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, SE-171 77, Stockholm, Sweden.
4 Department of Clinical and Experimental Medicine, University of Linkoping, SE-58185 Linköping, Sweden
5 Bureau of Epidemiologic Research, Academy of Athens, 28 Panepistimiou Street, Athens, GR-106 79, Greece
6 The Cancer Registry of Norway, Montebello, N-0310, Oslo, and Department of Community Medicine, University of Tromso, Tromso, Norway.
7 Folkhälsan Research Center, Samfundet Folkhälsan, Helsinki, Finland.

Competing interests: None declared

Pagona Lagiou, Professor of Hygiene and Epidemiology

Sven Sandin, Marie Lof, Dimitrios Trichopoulos, Hans-Olov Adami, Elisabete Weiderpass

University of Athens Medical School, 75 M.Asias Street, Goudi, GR-115 27, Athens, Greece

Click to like:

Although the topic which diet is best and safest is hot in a progressivly obese world. I feel this article does not take into account the effect of a few very strong predictors of cardiovascular disease. There is no information of a family history of cardiovascular disease, nor is there any information with regards to renal disease or function. Furthermore it is hard to believe that with such a long interval between exposure and oucome that a significant number of subjects would be on their original diet. This would make conclusions with regard to the outcome and diet invalid.

Competing interests: None declared

Peter JH Smak Gregoor, Internist-Nephrologist, vascular specialist

Albert Schweitzer Hospital, PO Box 444

Click to like:

As someone who studies diet from an evolutionary perspective - and its impact on the human microbiome - I always find diet studies that draw a correlation between increased protein or fat intake (or some combination of the two) with some unfavorable outcome. The implication always ending in some cautionary tale of the woes of eating too many animals.

In the current study, the authors suggest a low-carb high protein diet is significantly associated with increasing incidence of cardiovascular disease. In conclusion, they suggest attention be given in future studies to the nature of the protein - that is, plant versus animal. The implication is that one is better than the other and the reader can only assume the latter given the current wind direction in dietary advise on meat consumption. Interestingly, even though the conclude with this cautionary tale, their own study found no significant difference in cardiovascular health based on the difference of protein intake from either animal or plant origin.

That aside, a recent paper published in JAMA found that an extreme low-carb Atkins-like diet increased biomarkers of inflammation (e.g., CRP), even though the diet resulted in greater weight loss compared to other diets. The current study and its increase cardiovascular risk from low-carb intake and the JAMA study may be pointing us to the role of the microbiome.

In a series of elegant studies, Cani and colleagues ( 2-4) have shown that holding calories constant and varying macro levels of fat can induce low-grade metabolic endotoxemia which can lead to complications associated with cardiovascular health. As fat intake, so do serum levels of LPS and associated biomarkers. However, in high-fat diets with prebiotic oligosaccharides added (derived from chicory roots), serum levels of LPS drop, as do the metabolic markers of inflammation.

In the current study, the implication that increased protein – implied to be a bad thing if derived from animal sources – will increase risk of cardiovascular disease. By default, as either protein or fat intake go up and carbohydrates go down in ANY diet, we can assume that the physical and chemical diversity of what plants that are being consumed will decrease. With a drop in diversity comes a drop in the diversity of available fermentable oligo- and polysaccharides available to the microbiota.
In the Cani studies (and there are numerous others), a prebiotic oligosaccharide was shown to stimulate the growth of the LAB Bifidobacterium, which in turn improved gut permeability through several well-known mechanisms (5).

Rather than implying that higher protein and fat consumption may have a less than desirable outcome in cardiovascular health, maybe we should be talking more about the diversity and quantity of fermentable substrates that is lacking in the diet.(6) In other words, let us not always conclude that an association between high protein (and fat) and cardiovascular disease is the result of meat consumption.

(1) JAMA. 2012;307(24):2627-2634. doi:10.1001/jama.2012.6607

(2) Cani, Patrice D., Rodrigo Bibiloni, Claude Knauf, Aurélie Waget, Audrey M. Neyrinck, Nathalie M. Delzenne, and Rémy Burcelin. 2008. "Changes in Gut Microbiota Control Metabolic Endotoxemia-Induced Inflammation in High-Fat Diet–Induced Obesity and Diabetes in Mice." Diabetes no. 57 (6):1470-1481. doi: 10.2337/db07-1403.

(3) Cani, Patrice D., Jacques Amar, Miguel Angel Iglesias, Marjorie Poggi, Claude Knauf, Delphine Bastelica, Audrey M. Neyrinck, Francesca Fava, Kieran M. Tuohy, Chantal Chabo, Aurélie Waget, Evelyne Delmée, Béatrice Cousin, Thierry Sulpice, Bernard Chamontin, Jean Ferrières, Jean-François Tanti, Glenn R. Gibson, Louis Casteilla, Nathalie M. Delzenne, Marie Christine Alessi, and Rémy Burcelin. 2007. "Metabolic Endotoxemia Initiates Obesity and Insulin Resistance." Diabetes no. 56 (7):1761-1772. doi: 10.2337/db06-1491.

(4) Cani, P. D. 2012. "Crosstalk between the gut microbiota and the endocannabinoid system: impact on the gut barrier function and the adipose tissue." Clinical Microbiology and Infection no. 18:50-53. doi: 10.1111/j.1469-0691.2012.03866.x.

(5) Tuohy KM, Rouzaud GC, Bruck WM, Gibson GR (2005) Modulation of the human gut microflora towards improved health using prebiotics—assessment of efficacy. Curr Pharm Des 11:75– 90

(6) Leach, J.D., K. D. Sobolik (2010). High dietary intake of prebiotic inulin-type fructans in the prehistoric Chihuahuan Desert. British Journal of Nutrition, 103 , pp 1558-1561 doi:10.1017/S0007114510000966

Competing interests: None declared

Jeff Leach, Archaeologist

Human Food Project, 48 Swallow, New Orleans LA 70124

Click to like:

I have two main concerns regarding the article entitled “Low carbohydrate-high protein diet and incidence of cardiovascular diseases in Swedish women: prospective cohort study published in BMJ 2012;344:e4026 doi: 10.1136/bmj.e4026 (Published 26 June 2012).

1. The conclusions said that “low carbohydrate-high protein diets, used on a regular basis and without consideration of the nature of carbohydrates or the source of proteins, are associated with increased risk of cardiovascular disease”. The authors did not say that such observations are observed in women.

2. The study results revealed an association between low carbohydrate-high protein diet consumption and overall incidence of cardiovascular disease with incidence rate ratio estimates 1.04 (95% confidence interval 1.00 to 1.08), 1.04 (1.02 to 1.06), and 1.05 (1.02 to 1.08). Although such incidence rate ratio estimates are statistically significant since 95% confidence intervals (CI) do not overlap 1, the magnitudes of the strength of association are very minimal – the point estimates are closed to 1, the lower limits of 95% CI are very closed to 1 and the highest upper limit of 95% CI is only 1.08. These incidence rate ratio estimates from this cohort study should have no clinical importance and these associations should be due to a very large sample size. The authors should consider the appropriate number of the subjects to test the magnitude of the strength of association that has clinical importance (such as incidence rate ratio of 2) before conducting the study. The authors should include the aforementioned comments in terms of clinical importance of their findings in the article in order to avoid misunderstanding and misinterpretation of the readers and the media who do not have sufficient capacity to interpret the point estimate and 95% CI. Finally, the authors should say in the conclusions that the magnitude of the strength of the association they observed is minimal.

Competing interests: None declared

Visanu Thamlikitkul, Physician

Siriraj Hospital, 2 Prannok Road, Bangkok 10700, Thailand

Click to like:

Let me start this response with a question. Is your style of eating - your dietary repertoire if you will - the same today as it was 15 years ago? Mine's certainly not. 15 years ago I was a bachelor doing my residency. My diet consisted of burgers, steaks, sausages, pizza, chicken wings, french fries, restaurants, take-out, pots and pots of coffee, and horrible, horrible hospital cafeterias. The fact is, whether it's consequent to life changes, medical conditions, dietary fads, new relationships, new jobs - eating style for many varies wildly with time and circumstances.

Yet the researchers of this paper, one that looked at the impact dietary style would have on cardiovascular risk among 43,396 Swedish women, which concluded that women who followed low-carb diets had an increased risk of heart disease, used only one single baseline dietary data set upon which to base their 15 years long study and its conclusions.

And about that single, solitary, data set. It consisted of a food frequency questionnaire in which subjects were asked to identify how frequently they consumed 80 different food and beverage items over the course of the past 6 months.

Think that'd be accurate? Can you remember how many of anything, let alone 80 items, you've had over the past 6 months? And what if you ate and drank more than 80 items?

But let's not guess about accuracy, let's actually look at it. Taking the study subjects' reported average caloric intake, it's immediately evident that their dietary recall data is inaccurate as the authors report an average daily calorie consumption of only 1,561 calories. That's an incredibly small number, and one which flies in the face of the Food and Agriculture Organization of the United Nations reporting of a Swedish per capita average calorie consumption of 2,990 calories during the same time period this study's data was collected.

But let's for a moment pretend that the data was in fact accurate and that you could fairly extrapolate that your style of eating today will remain the same for the next 15 years. My next question would be do you think the quality of your diet's various proteins, carbohydrates and fats might have a bearing on your risk of developing cardiovascular disease over the next 15 years? Would eating a diet higher in trans-fats confer a different degree of risk than a diet higher in unsaturated fats? Would eating a diet rich in whole grains confer a different risk than a diet rich in ultra processed pulverized flours? How about if you ate out regularly versus cooked from scratch? Deep fried versus baked? Sausages versus salmon? Quinoa versus white rice? Kale versus potatoes? Of course it would matter, and thats a fact that I'd bet even a straw poll of 10 year olds would agree with.

Not this paper's researchers. You see they simply reported and analyzed the data from that single time point questionnaire by means of 10 different centiles of carbohydrate consumption. They didn't pay any attention to the quality of the macronutrients being consumed, just their total quantities. The only consideration they gave as to the quality of macronutrients was a very broad "animal protein" vs. plant sources. There was no consideration given to quality of fats (this despite the well established impact of trans-fat on increasing cardiovascular risks and on unsaturated fats for decreasing them), or carbohydrates (again despite data suggesting cardio-protective benefits from whole grain and risk from refined).

But ok, let's even pretend for a moment that the data was in fact well controlled for dietary factors known to impact on cardiovascular disease, is the risk reported here worthy of concern and press release? Here the relative risk increase of 5% per increase in low-carb score means an extra 4-5 cases of cardiovascular disease per 10,000 women in the lowest carb group per year. Should we stop the presses? Given how incredibly small that is absolutely, and given how incredibly complicated it is to control for confounding variables (and here I'm not just talking about the unbelievably glaring confounders that were explicitly ignored by the researchers but non-dietary and lifestyle variables of which this study looked at very few), can and should cautionary conclusions truly be drawn let alone publicized?

Lastly let's pretend for a moment that data is in fact conclusive and that the risk is real and scary enough to shout from the rooftops. Do you think then it would matter if the diets studied and were being reported as risky weren't in fact "low-carb" given that the caution being shouted is that low-carb dieting increases the risk of heart disease? Looking at the data, the 1st quartile of lower carb consumption is in fact a diet where the 154.7g of carbohydrates account for 40% of calories. Simply put, that's not a low-carb diet! Atkins for instance starts at just 20g and then over time people tend to work themselves up to between 50-100g. 154.7g? That's. Not. Low. Carb. Nor is even their reported 10th percentile at 123.7g or 32% carbohyrdate.

So to review. The paper's basing all of its 15 years worth of conclusions off of a single, solitary, and clearly inaccurate, baseline food frequency questionnaire, it didn't control for clearly known smack-you-in-the-face dietary confounders, it found just a miniscule absolute increase in risk, and the diet it's reporting on can't even be fairly referred to as a low-carbohydrate diet.

Useful? Conclusive? Press worthy?

It gets worse.

Here the BMJ didn't just publish a completely useless paper, they gave this very clear, yet completely non-evidence based advice to clincians in their accompanying editorial,

"Despite the popularity of these diets, clinicians should probably advise against their use for long term control of body weight"

Worse still highly reputable socially networked curators of medical information tweeted the resultant media stories as relevant and even Journal Watch, a New England Journal of Medicine publication reported it as valuable to scores of physician subscribers who trust Journal Watch to keep them abreast of the latest important journal studies.

As far as the true state of the evidence on the long term risks or benefits of low-carbohydrate diets goes this paper, and the BMJ's editorialized conclusions, add absolutely nothing. Given the paper's scientifically horrifying flaws and the editorial's completely clueless conclusion they're useless contributions to the medical literature. In fact I'd describe them as worse than useless in that they're unforgivably, irresponsibly and shamefully misinformative - something our already incredibly nutritionally confused world really didn't need.

(Originally published on my blog at http://www.weightymatters.ca/2012/06/what-reading-that-low-carb-gives-yo...)

Competing interests: Author of Simon & Schuster published book on fallacies of modern day dieting slated for publication Spring 2013

Yoni Freedhoff, Physician

University of Ottawa, 575 West Hunt Club, Suite 100

Click to like:

I've read the article with high expectations but, after reading it, I'm very sad, because I don´t understand why the BMJ, one of the best medical journals in the world, has published an article so biased. If we look over it we can see the following biases:

1º.- Methodological bias: The authors only send a questionaire in 1992 and 15 years later without: 1) New questionaires; 2) Patients´ follow-up; 3) Information about the patients' habits. They jump to conclusions based on biased and misleading arguments.

2º.- Selection bias: At least 1 out of 2 patients rejected being included in this study.

3º.- Measurement bias: Authors know nothing about the life habits of patients.

4º.- Follow-up bias: Only one questionnaire is NOT appropiate follow-up.

5º.- Expectation bias: The absence of masking or blinding may err in measuring data toward the expected outcome.

6º.- Insensitive measure bias: The measurement tool used in this study is not sensitive enough to detect important differences in the variable of interest.

7º.- Compliance bias: Nobody knows if there is a adherence to the reported diet habits.

8º.- Misclassification bias: It's impossible to know if patients are classified correctly (Low carb vs High carb).

9º.- Confounding bias: In this study no more than 51% of potential patients were included.

10º.- Non-response bias (49% of potential patients!): Limits generalizability, not validity.

So far as the scientific methodology is concerned, the conclusion of this article should be the following:

“This medical paper has both a very low quality and a lot of major biases, therefore it shouldn´t be published because its flaws not only have a confounding effect in nutrition science but can generate emotional shock in people under these kind of diets"

According to Stephen Lock (ex-editor of BMJ): "Medical journals will soon be wrapping up next week´s fish and chips".

Competing interests: None declared

Álvaro campillo-Soto, General Surgeon & Scientific Metodologist

University General Hospital "JM Morales Meseguer", Marqués de los Vélez s/n Murcia. Spain. CP: 30008

Click to like:

This paper has attracted much interest, but there are several reasons to be cautious in considering its conclusions.

The "opt-in" enrolment is a methodological weakness, and, although the study had more than 40 000 subjects, it also had a less than 50% recruitment rate, which creates the potential for bias. The authors make no comment on this matter.

Dietary habits and the presence of cardiovascular risk-factors were self-reported via a questionnaire at enrolment but there are no follow up data to look at changing trends for any of these indices, despite the mean follow-up being 15.7 years. This represents a significant shortcoming and, again, the authors do not address this issue.

The study does seem to confirm the well-established risks of hypertension, obesity, and smoking in terms of subsequent cardiovascular disease, with the relative risks between 2.48 and 2.73 for the presence of these conditions at enrolment.

However, it is less clear that the 5% increase in risk (RR 1.02-1.08) found in those on the lowest carbohydrate and highest protein diets can be considered a real finding, given the lack of longitudinal data on the more significant risk factors, and especially as the cases were defined by the occurrence of a cardiovascular event, and then mapped to the dietary habits. It would be interesting to know the current/exit blood pressure, BMI, and smoking status of these cases.

These women (the cases) developed premature cardiovascular disease, and it is suggested that dietary habits conferred a small excess risk. That is possible, but it is certainly true that the cases must have had significant predisposition to cardiovascular disease - one wonders what would have happened to them if they had had different dietary habits, and if they subsequently became obese and developed the metabolic syndrome?

The study design makes it impossible to answer this key question. Randomised controlled trials, of adequate power, with appropriate analysis of risk factors are necessary, but are lacking at this stage.

Competing interests: None declared

Sean D P Mackay, Senior Lecturer, Surgeon

Monash University, Eastern Health Clinical School, Arnold St, Box Hill, Melbourne, Australia

Click to like:

The following points need to be made in response to this article:

1) The claimed association, not causation, is extremely small

There were 1,270 recorded incidents of all cardiovascular disease (ischaemic heart disease, ischaemic stroke, haemorrhagic stroke, subarachnoid haemorrhage and peripheral arterial disease) among 43,396 women over 680,745 study years. An incident rate of 0.19%. This means that the women in the study had a 99.81% chance of NOT suffering any cardiovascular events during the long study.

The 8,395 women with a LCHP score of ≥ 16 on their scoring system had a 0.23% incident rate during their 131,262 woman years. This 'highest risk' group, therefore, had a 99.77% chance of NOT suffering any cardiovascular events whatsoever during the 15-16 year study.

The 8,343 women with a LCHP score of ≤ 6 on their scoring system had a 0.14% incident rate during their 130,965 woman years. This 'lowest risk' group, therefore, had a 99.86% chance of NOT suffering any cardiovascular events whatsoever during the long study.

So the 'highest risk' group might expect a 0.04% lower chance of remaining cardiovascular disease free than the average woman in the study. This does not justify the BMJ headline of "Experts warn of significant cardiovascular risk with Atkins-style diets." (My emphasis) (Ref 1)

2) The scoring system is absurd

Bjorn Hammarskjold captures it best: “Low protein is 1, high protein is 10, low carbohydrates are 10, high carbohydrates is 1. So to have sum score of 11 it can be low protein (1) and low carbs (10) or high protein (10) and low carbs (1) or any sum between: 1+10, 2+9, 3+8, 4+7, 5+6, 6+5,7+4, 8+3, 9+2 or 10+1, all give 11 as score.”

What conclusions would this lead to?

3) Atkins is not low carb high protein
Atkins is low carb, whatever protein naturally occurs in food and high fat. This brings us on to:

4) The 'researchers’ do not seem to know the nutritional composition of food

There are three macro nutrients: carbohydrates, fat and protein. All food has at least one macro nutrient, most have two and some have three.

The only foods on this planet with no protein are sucrose and oils. Every other food contains protein. Sucrose is 100% carbohydrate – no fat or protein and oils are 100% fat – no carbohydrate or protein.

Most foods are predominantly carbohydrate proteins (with no fat or little fat) or fat proteins (with no carbohydrate or little carbohydrate). The carb proteins come from the ground and trees and will fall under plant proteins in this article. Fruits, vegetables, grains, pulses, starchy foods are in this category. The fat proteins come from ‘things with faces’ and will fall under animal proteins in this article. Meat, fish, eggs and dairy products are in this category.

This article assumes that the higher the carbohydrate intake and the lower the protein intake, the lower risk of CVD the women shall have. Hence our Swedish women should have eaten nothing but sucrose for 15.7 years and they would have had 100% carb intake and 0% protein intake (and 0% fat intake – presumably also ideal). They would have had no vitamins or minerals, no protein, let alone complete protein and no fat, let alone essential fats, but they would have met the recommendations emanating from this article.

5) Table 1 of the study should have provided the headlines

Table 1 of the study looks at “non-nutritional variables”. Table 1 tells us the following:

- AGE: Women in the 35-39 age group had twice the incidence of cardiovascular events as women in the lowest age group. Women in the 40-44 age group had nearly four times the incidence of cardiovascular events as women in the lowest age group. Women in the 45-49 age group had nearly eight times the incidence of cardiovascular events as women in the lowest age group.

- EDUCATION: Women with fewer than 10 years education had nearly three times the incidence of cardiovascular events as women with more than 13 years education. Women with 11-13 years education had less than half the incidence of cardiovascular events as women in the lowest education group.

- HEIGHT: The shortest group of women in
the study had nearly three times the incidence of cardiovascular events as women in the tallest group.

- BMI: Those with normal BMI have the lowest incidence of cardiovascular events. Underweight woman had almost twice the incidence of cardiovascular events as normal weight women. Overweight women experienced less incidence of cardiovascular events than underweight women, but still 1.6 times that of normal weight women. Obese women experienced nearly three times the cardiovascular events as normal weight women.

- SMOKING: Not surprisingly, smokers have nearly three times the incidence of cardiovascular events as non smokers. Giving up smoking helps but you’re better off not ever having smoked.

- EXERCISE: The lower the activity group, the greater the incidence of cardiovascular events.

The headline should have been "The older women get, the more incidence of cardiovascular events they experience”, but then that wouldn't have been headline news.

Ref 1: http://group.bmj.com/group/media/latest-news/experts-warn-of-significant...

Competing interests: None declared

Zoe Harcombe, Author & obesity researcher

None - self employed, Llanvaches, Gwent , NP26 3AY

Click to like:

The conclusions of the paper are not supported by the data. The evidence does not demonstrate that low carbohydrate diets increase the risk of cardiovascular disease.

The fundamental problem with this research, as with most diet studies, is that the food intakes of subjects were not adequately measured (1) . Indeed, it suffers a double problem. First, it used an inaccurate measuring instrument at baseline. Second, it conducted no measure of intake at all over the following 15+ years.

The initial instrument was a food frequency questionnaire (FFQ), popular in large studies because results are easily processed. But FFQs are just another form of “self-report”. They rely on subjects honestly telling researchers what they eat.

But they do not. Subjects consistently report a healthier diet than they actually eat, smaller in volume and a more nutritious mix. Technically, this is known as “under-reporting”. In plain English, it is lying.

By whatever name, the problem is well-known, long established and large. In Britain, the accuracy of national diet surveys is checked by using “double-labelled water”, a biochemical measure independent of subjects’ self-reports. Among adults, average under-reporting of calorie intakes is 25% (2). No such independent measure was employed in the study reported here.

Instead, the authors claim they used a “validated” FFQ. But such “validations” only compare FFQs with other forms of self-report, vulnerable to the same problem of misrepresentation. The only thing “validated” is the errors of both.

Subjects’ intakes were never measured again, by any instrument. Nonetheless, the authors imply that their initial diets were maintained for a decade and a half, concluding that low carbohydrate diets are risky if used “on a regular basis”. This is crucial to their argument because they dismiss supportive evidence for such diets as only coming from “short-term” studies.

The authors acknowledge the difficulty, but try to defend themselves by claiming that lack of repeat intake measures is common in large cohort studies. They are certainly right about that.

Despite their limited intake data, the authors perform numerous statistical analyses to establish relationships with various forms of cardiovascular disease. None of these relationships are credible.

Like many before them, they conduct sophisticated secondary analyses of bad primary data. This is a statistical smokescreen that conceals the fundamental problem.

And the problem truly is fundamental. Like many others, they seek to show that bad diets cause diseases, without measuring diets accurately. This a a fundamental flaw in scientific method: trying to establish a causal relationship without measuring the independent variable properly.

In sum, this study does not allow us to conclude anything about low carbohydrate diets, for or against. It is just another partisan shot in the long-running ideological war between the proponents of low fat and low carbohydrate diets.

(1) Winkler J, The Fundamental Flaw in Obesity Research, Obesity Reviews, 2005, 6, 199-202. jtw@blueyonder.co.uk

(2) Rennie K, Coward A, Jebb S, Estimating under-reporting of energy intakes in dietary surveys using an individualised method, BJN, 2007, 97, 1169-1176.

Competing interests: None declared

Jack T Winkler, Professor of Nutrition Policy

London Metropolitan University (Retired), 28 St Paul Street, London N1 7AB

Click to like:

THIS WEEK'S POLL