Low carbohydrate-high protein diets are associated with increased risk of cardiovascular disease
Guest Article by Martin MacDonald
Are you scared after reading that blog title? Are you worried that I've suddenly realised that the EatWell Plate is in fact great? Fear not, nothing has changed its just another crappy study getting infinitely more media attention than it deserves. If you've not heard about it yet, there was a big study published this week in the British Medical Journal titled: Low carbohydrate-high protein diet and incidence of cardiovascular diseases in Swedish women: prospective cohort study. Many of you got in contact through various avenues to make sure that I was aware of it ….and more importantly could I perhaps give a rundown of why these results were found because "my colleague at work sent me a link to this article on the Express.co.uk website saying 'I think you should read this.'" This wasn't the only place it was reported, the Daily Mail and Telegraph both reported it. What is it with the papers, when is a paper going to realise that if they actually got someone qualified to write their articles they could have unique/original news stories when everyone else is writing the same old unsubstantiated crap! I'd be MORE than happy to do this for them so if you're reading this Editors…. ๐
What is interesting is that the Atkins diet has been receiving the brunt of the negative attention… despite the fact that the Atkins diet is high in fat… not protein?! The Dukan diet should really have been the baddy here! Similarly, more than one person has mentioned that this study seems to refuse the Mac-Nutrition message despite the fact that the only real message of Mac-Nutrition is that people should eat real food that is as minimally processed as possible. Of course this tends to mean that protein and fat are higher and carbohydrates are lower than is currently recommended by the UK government but by no means do I advocate a one size fits all or extreme diet for all… I'm boring and uncontroversial like that… ๐ (As a side note, the article in the Express actually gives the final word to an Atkins spokesperson who clears it up nicely!)
So, onto this study! This study is open access and anyone, even journalists, who feel they're qualified to write on health news like Jo Wiley from the Express, can read all of it! This is not going to be my usual research update, I'm simply going to tear this utterly terrible study apart as quickly as possible because it's already had too much attention. What makes the situation slightly worse is that I don't think it is just terrible media reporting! Even the authors of the study (so called scientists) seem to be jumping head first with recommendations based on their massively flawed study.
The Study
A quick recap on some terminology: simply reading the blog title means alarm bells should ring with 'association does not prove causation'. 'Association Does Not Prove Causation', 'ASSOCIATION DOES NOT PROVE CAUSATION'… thinking about it, that would be quite a catchy ring tone, no? Randomized intervention trials on high protein diets are the ones that can show a degree of causation. Second term is 'Prospective' which is often seen as the best type of observational study because you're able to get baseline readings and then 'follow' the subjects through the longitudinal process. At this point critics might be wondering if I'm simply throwing stones at this article because it says low carb is bad and because it is an observational study… oh no my friends… there's more.
This study took ~43,000 Swedish women, aged 30-49 years at baseline and followed them for ~15 years. After 15years they measured the association between incidence of cardiovascular diseases with intakes of carbohydrate and protein. They attempted to control/adjust for intakes of energy and fat intake as well as other variables such as smoking, physical activity etc.
Problem 1 – Over 15 years, diets change!
In this study, I kid you not, they only asked the women what they ate once!!! At the beginning of the 15.7 years! How has your diet changed over the last 15 years? If you're a 47.7 year old woman reading this post, is your diet the same now as it was when you were 30 years old? Or even better, if you're 62.7 years old, is your diet the same as when you were 45?? OK, maybe I'm wrong and a single, weighed food diary is good enough.. But oh no, wait!
Problem 2 – They didn't use a weighed food diary… they used a questionnaire.
Yes. Instead of people measuring how much food they ate, they filled in a questionnaire that asked them how many times on average they ate different foods! How many you ask? 80! There were 80 different foods and the women were asked with what frequency they ate them. Anyone see a problem?
Problem 3 – What about foods that I eat that aren't in the 80 on the questionnaire?
OK, so maybe not a big problem but still not the best. So, how far back do you want me to record for?
Problem 4 – Subjects had to remember what they had eaten in the last 6 months?
Yup, the women were expected to record their food consumption of 80 different items over the last 6 months! Come on, you're starting to question this study now too, aren't you? By the way, 583 of the women couldn't face filling in the questionnaire so were not included in the study. I'm surprised it wasn't more!
So after all of this, the researchers gathered up all the data and then 'converted' this to Energy, Protein, Carbs and Fats in Kcals and Grams per day. Et Voila! We have our data set.
Problem 5 – Doesn't diet quality have anything to do with disease risk?
The results of this study are based on comparing intakes of protein vs carbohydrate; that's pretty much it. Now, some muppets out there are still saying that a calorie is a calorie but let's assume for a minute that those people are too stupid for comprehension and that getting your carbohydrates from coca cola is different to getting it from vegetables however crazy that may seem…. this study did not distinguish between getting carbohydrates from sweet potato vs a 'heart (not) healthy' Nutrigrain bar or protein from billy bear ham vs grass-fed beef! Oh, lastly, there is no mention of controlling for trans fat intake either.. but never mind… they're only the most harmful 'food stuff' available but lets just ignore that… What's more, the 'conversion' seem to bring about some dubious data…
Problem 6 – Are you sure those numbers are right?
If you have a look at Table 2 from the study we can see that the average energy intake of women was ~1500kcals. Something a bit fishy there, considering average population intakes are over 1000kcals higher than that! Can we speculate that perhaps some of the women 'forgot' how many times they had ice cream or cupcakes in the last 6 months?
Reporting Error!
So, with the headlines telling people that this study has proven that eating a low carbohydrate diet is bad and will give you cardiovascular disease and with the researchers telling us that those with the lowest carbohydrate intakes were most at risk… Do you care to guess what the 'low' carbohydrate intake in this study was? To me, very low carbohydrate is less than 20g per day… I've only had a handful of clients in this range. Low carbohydrate is 20-100g or somewhere in the region of 0.5 – 1g/kg bodyweight. Well, in this study the lowest quartile of carbohydrate intake in this study was 154.7g of carbohydrate!?
I'm going to leave it there. Hopefully you've got some idea of just how bad this study was. It's a pity that Dr Clare Walton of the Stroke Association and Victoria Taylor of the British Heart Foundation seemingly thought the study was sound enough to spout off recommendations based on it and that the BMJ released an editorial condemning lowering carbohydrate intake.
Edit: Just a quick update, someone mentioned an area to me that I hadn't thought to include. Bear in mind the huge impact this study had in the media… you'd assume the figures they produced would save hundreds if not thousands of lives? Well, not so much… in real terms, if 10,000 extra people starting eating the way the people in the highest protein and lowest carbohydrate group were, there would only be an additional 4-5 cases of CV disease each year. Ridiculous.
Update: Can't believe I didn't include this the first time! I have a list of studies that I keep in tabs to write about if I get the inclination/time. I completely forgot about this one; Low-carbohydrate, high-protein score and mortality in a northern Swedish population-based cohort. Remember, the study above was done in Swedish people (women) too! Well, it seems maybe being from the northern part of Sweden is protective against death… or perhaps as I said above, the previous study was a load of rubbish. This study also used "low-carbohydrate, high-protein (LCHP) scores" and found the following:
High LCHP score (14–20 points) did not predict all-cause mortality compared with low LCHP score (2–8 points),..
For cancer and cardiovascular disease, no clear associations were found.
I can't get access to the full paper here to give you a critique as above but… what's the need, I'm not saying it proves anything other than the fact that these studies are no good for drawing conclusions and they most certainly don't deserve big media attention… although notice how this study got absolutely no attention… hmmm.