"GOOD and Whole Foods Market have teamed up to bring you a series of infographics and videos on what we eat: where it comes from, what’s in it, and what choices you can make to ensure you’re eating as well as possible."
On November the 4th they published the second infographic: Unfortunately it is a little large to embedd here, but essentially it takes 9 countries: The USA, Australia, Sweden, France, Brazil, Bangladesh, Japan, China and Eritrea and gives you 8 main statistics about each country:
1) The proportion of the population that is overweight (as defined by a BMI of 25 or higher)
2) The average number of calories eaten
3) The percentage of the diet attributable to 'cereals'
4) The percentage of the diet attributable to 'fruits, vegetables, pulses and nuts'
5) The percentage of the diet attributable 'meat , milk, fish, eggs'
6) The percentage of the diet attributable to 'oils, fats, sugars'
7) The percentage of the diet attributable to 'roots and tubers'
8) The percentage of the diet attributable to 'other'.
Rough analysis:
Firstly, we the relationship between calories ingested, and % of the population overweight is not entirely linear; although there is a trend for more calories to = more overweight, the pattern does not fit that well, in fact 3 (France, China and Japan) fall way off the regression line, perhaps Brazil and Sweden do too. So it really is not as simple (according to this infographic) as energy in = weight out.
So, cereals:
We see that the lower the percentage of cereals in your diet, the higher the weight. In general. Although the USA eats a higher % cereal than than Australia, but it more overweight and there is the same problem with Sweden and France and then Bangladesh and Eritrea. Plus it does not seem neatly proportional.
Fruit, vegetables, pulses and nuts:
This is essentially uninformative - the % of your diet attributable to these foodstuffs does not affect the proportion of people who are overweight.
Meat, Milk, Fish and Eggs
Meat, milk fish and eggs seem to increase weight as, as they increase as a proportion of your diet, weight goes up. The same is the case for fats, oils and sugars:
Fats, oils and sugars:
This shows a much clearer linear trend with more fat oil and sugar = higher BMI. The question is, do any of the % dietary variables tell us anything independent - i.e. given all the other information what is most associated with weight? Well, a very quick 'n' dirty analysis, at an alpha level of .5; oils, fats and sugars are the only significant predictor. If we relax this alpha level to .7 (for non statisticians this basically says that a smaller effect is still important. I am doing this given how small the amount of data is, but I admit would not pass muster in a peer-reviewed science journal of any repute and is a heinous crime against the discipline of statistical inference), the categories 'cereals and grains' and 'meat, milk, fish and eggs' are also predictive. So, according to WF's infographic we should:
-Watch our calories a bit as they have a general impact, but not worry too much about them
-Drop fats, oils and sugars
-Eat more cereals and grains, and less meat, fish and eggs
Here is my probelm with this infographic and it's proclivity to engender such conclusions:
1) Population statistics as a whole
I am not going to trash population statistics. I use them. I made a PhD out of them. Heck I even make a living out of them now, to a point. But, people need to understand population statistics and a few things about them. The most important here being that the results are a general trend over the whole group, they may not apply to everyone in the group.
Let me give an example. Let's say I have a pill 'X' that makes women's breasts grow by 4cm a month. I don't know that it is sex-specific and I give it to a general population sample that is 50% male. I measure the breasts at monthly intervals and after 3 months say 'the average breast growth was 2 cm a month (the women's breasts grew 4cm, the men's not at all)'. This information is not useful to men - taking the pill will not affect their breast growth. So, trends over the whole popualtion may only apply to some people; they may even work in the opposite direction for others.
What does this mean for our interpretation? Well, these conclusions may not apply to us all. Especially, I would say, conclusions regarding fat. There is good evidence to think that fat, for example is not dangerous, with a zero / very low / low carbohydrate diet. However, some fats, when consumed at high levels with carbohydrates (particularly refined ones) can be very detrimental to health. So the problem here is 'population heterogeneity' - too many unmeasured differences among people in your sample, making it difficult to generalise any trends to anyone else. Of course, there will always be these differences and problems with generalising from one sample to another, but here, I think there are far too many sub categories to make a useful conclusion.
2) Comparing the countries
Linked to the above points there is too much information missing to say 'X' is thinner than 'Y' and 'X' eats less fat than 'Y' so fat makes you fat. We need to very much think about composition of the diet, daily life, exercise, biological sensitivities to certain food ratios and so on.
3) The categories
So far, I think I have only highlighted that the data, as presented, are not that informative. And heck, if they get people thinking about nutrition and making generally sensible changes to their diet, OK. I don't agree with the above conclusions as a mantra for general health (I will be honest, am a proponent of a very low / low carbohydrate diet when used alongside other restrictions). But, there is a darker side here subtly contained within the categories chosen.
What kind of a category is 'fats, oil and sugars', when we are talking about health, and weight gain? Why on Earth would you lump the mono- and poly- unsaturated fats which have been shown to contribute to a reduction in BMI, a reduction in waist circumference and an independent improvement in blood lipid profiles with trans fats (I will leave saturated fats out of this for a moment)? More to the point, why the heck is sugar in here? Fat and sugar in the diet are not synonymous, nor interchangeable in their effects. Fat + sugar = bad. In my opinion, and many others; fat in the absence of sugar (and starch) = OK / good. For some people (e.g. the American Heart Association) sugar in the absence of fat, is OK. I do not think this is telling us anything about health.
What kind of category is 'meat, milk, fish and eggs'? The advice may not all concur with suggested intake, but I cannot think of a dietary plan that treats fat-free milk the same as steak; one is high-ish sugar and low fat; the other the opposite. Chicken breast vs. liver - same effect on BMI? Well, this likely depends on the composition of the rest of your diet, but for the majority of the USA - no. Perhaps a better example is 'fruits, vegetables, pulses, nuts'. This runs the gamut from high sugar low fat (fruits) to low sugar low fat (many vegetables) to low sugar / high fat (nuts). Really, these things are going to have very different effects on BMI, and differential relative effects within different diets. And why are some 'fruits' not in the 'sugar' category, aside of vitamin content, I have yet to be convinced that watermelon should not be treated as fat free candy, biologically speaking in terms of weight-per-weight effects on BMI.
The problem is that this categorising encourages bad thinking about diets. We need to be more informed and more careful about our selection. Telling people to eat more 'cereals and grains', less 'fats, oils and sugars' and less 'meat milk and eggs' could lead to a very harmful diet if the healthy fats, and protein sources in someone's diet e.g. some avocado and some chicken were replaced with masses of refined flour (e.g. pretzels and white bread). There is no mention of processed food, nor fibre. These categories are too crude to be helpful, and possibly encourage simplistic thinking that could be detrimental to health.
Conclusions
Well, I will always be critical of anything that does not encourage very low / low carbohydrate diet, a cutting out of processed foods and grains, ingestion of natural fat (within this frame), more movement and so on. But, I hope I have given everyone some food for thought in interpreting the arguments we are presented with for following the prevailing dietary advice (which I think stands at 'high carb, low fat, low calorie' first and foremost, and increasingly 'medium sugar' as an addition).
