The Obesity Blame Game

I want to tell you something you’ve known all along.

I’ve written a fair bit about food and health over the years. I have tried, occasionally with success, to counter some of the common misunderstandings and bad science in the world of food. But there is one thing that I dearly wish I could get through to people, because it is the source of so much of that misinformation. It concerns obesity and why Blondie releasing ‘Parallel Lines’ didn’t cause a global health crisis. And frustratingly, it takes a bit of explaining.

It all comes down to a strange quirk of how we talk about obesity and body weight. We generally use a system of BMI measurements to get an impression of how much excess weight people in the population have. BMI is a flawed system, especially when used as a diagnostic tool for individuals, but as a measure of how body shapes are changing in a human population, it’s not that bad. We take someone’s weight in kilograms and divide it by the square of their height in metres to give us a number, which we can then use to compare against other people. In short, it is a rough measure of how fat people are.

The quirk comes not from that measurement, but from how it is interpreted at a population level. When anyone talks about changes in population body weight over time, they will not talk about changes to the average weight, or the average BMI. Instead, they will almost always refer to the percentage of a population who are overweight or obese. This is the percentage of people who have a BMI over 25 for overweight, or 30 for obese.

Why 25 and 30? Perhaps surprisingly, it is just because somebody at the WHO decided that they are nice round numbers that should be easy to remember. In fact, up until 1997, the cut off for overweight was 27, but this was changed to make things simpler, although that suddenly gave 30 million previously healthy people a medical condition, moving them into a different category overnight. Occasionally in research and public health communication, an ‘overweight’ BMI of 25-30 is termed as ‘pre-obese’, suggesting that as soon as you pass a BMI of 25 a slide to obesity is inevitable (this author sincerely hopes that is not the case).

The number of people in a population that have a BMI over 30 tells us something about that population, but is a slightly odd way to talk about things. It is a bit like measuring changes in population height, then instead of working out the average, reporting how many people are over 6 foot 2. Or perhaps it is similar to trying to understand improvements to our education system by investigating how many geniuses there are. It gives us some information, but it is slightly odd information that is difficult to interpret.

If changes to the average weight or average BMI of a population were regularly reported alongside the percentage of obese and overweight people, I could perhaps understand it. I suppose the main argument is that people with a BMI over 25 have higher risk of health issues, but even that is contentious, varying massively depending on age and other factors. And if it is health rather than BMI that we are worried about, why don’t we just report how many people have weight related health issues? Why are we so concerned about the percentage of people who have crossed an arbitrary BMI threshold?

This has always seemed odd to me, and I could leave it alone if it was not for one thing. This way of measurement produces a statistical quirk that, one way or another, has proven one of the most consistent sources of dietary misinformation over the past 30 years. And it is one that however much I shout about it, is not going away.

Look everywhere in the diet, food and health space and you will hear that at some point around the late 1970s and early 1980s, obesity in the populations of the US and UK started to rise very quickly, doubling within a decade (graph below is one example, there are hundreds of others). Although I wish this narrative of an obesity epidemic would just go away, there is no denying that the rate of people crossing the threshold for obesity did rise rapidly at that time.

Blondie released parallel lines at exactly the point obesity started to rise - coincidence?

This then leads to the inevitable question, what happened? Was it the release of dietary guidelines that promoted low-fat, low-calorie diets, leading to an increase in carbohydrate consumption? Was it the sudden rise of sugary cereals as a breakfast of choice, instead of bacon and eggs? Was it the increased use of chemicals in food packaging? The increased use of pesticides on crops? The rise of Ultra Processed Foods? The decline of home cookery? Increases in women entering the workplace? The runaway commercial success of Blondie’s third studio album?

I have seen all of the above blamed for obesity (although Parallel Lines has only been the culprit in previous Angry Chef blog posts). The reasons why this sort of blame game is popular is for two reasons, both of which will be familiar to Angry Chef readers. Firstly, humans love to look for corelations, to find patterns that explain the world. But as we know, correlation is not causation, and you really need better evidence than ‘these two things happened at the same time and it’s a nice story that agrees with some of my existing prejudices’

Secondly, the ‘nice story’ people tell about this correlation is almost always one that appeals to antiquity. Most of us, for one reason or another, are inclined to believe that things were better at some point in the past, often one when we ourselves were younger and more vibrant (and maybe a bit thinner). The idea that humanity has ruined a perfect world is at the heart of most major religions, probably because believing things were better ‘back in the day’ is a near universal part of the human condition. So, a story about how great things were pre-1978, despite almost every available metric of health, happiness and longevity telling us otherwise, is usually a winner. Blame UPFs, women’s liberation, carbs, dietary guidelines, chemicals, or female fronted New Wave pop bands. Stories that combine spurious correlations and the evils of modernity are an easy root to a dietary best-seller.

There’s just one problem. The entire premise that something dramatic happened in the late 1970s is a false one. It is simply an artefact of us looking at the crossing of BMI thresholds, rather than changes in average BMI. Don’t believe me. Below is a graph showing changes to BMI over time in different US populations groups. There is no dramatic change in the late 1970s. So, what is different about this graph when compared to the standard one we see everywhere else?

This graph shows changes to BMI over the 20th Century using something called a birth cohort analysis

The BMI of a population at any one time tends to follow a pattern that is quite similar to a normal distribution bell-curve, but skewed a little to the right side, largely because people with very low BMIs are not very good at staying alive (see 2 figures below). From the graph above, we can see that the average population BMI in the US has been increasing steadily since the turn of the century, which is quite understandable, given that food became way cheaper and more plentiful, meaning that far fewer people were experiencing hunger. In richer and better fed groups of the population such as college graduates, there is evidence that the average BMI was increasing faster and earlier than the overall picture. There is also evidence that some of the most rapid gains in US BMI happen from the late 19th to mid 20th century, a time which saw dramatic reductions in poverty and hunger. An example below illustrates this in the changing BMI of US army recruits. If you can picture this, it seems clear and obvious that when people have enough to eat, they gain weight. So, when populations have enough food, average BMI increases.

This chart shows BMI and weights of 18 year old American men. Data going back this far are quite rare.

For this reason, the average BMI of the population increased throughout the 20th century, and the bell curve of BMI distribution moved slowly and steadily to the right, as would be expected. There’s an example of this below from late in the century, to illustrate how the distribution moves and changes shape over time.

This chart illustrates how BMI changes over time. Note how a small movement in the average BMI leads to a large change in the number of people crossing the obesity threshold.

 Generally the average BMI increased by about 2-2.5 BMI points per decade, with a slight uptick around 1960. For quite a while, during this steady advance, the number of people with obesity kept quite low. But when we reach the late 1970s, the rate of obesity suddenly starts to rise, despite the rise in average BMI remaining the same. Why? Simply because, at this time, the steep side of that bell-like distribution curve starts to cut through that arbitrary BMI line at 30. If the cut off was at 32, that sudden rise would have started around 1988, and we’d been blaming Yuppies, Mobile Phones and Green by REM. If it was 28, we’d be wanting to ban Astral Weeks and Rosemary’s Baby.

So, why did obesity suddenly start to rise in the late seventies? It’s because someone at the WHO thought 30 was a nice round number. Any other explanation is missing the point.

And why do we use increases in obesity prevalence as opposed to average BMI when discussing weight related health issues? I suggest that one reason is that it makes things sound a lot more dramatic and sudden, which is good for pedlars of false correlations. In the period from 1980 to 1990, we often hear that US obesity prevalence doubled. But the reality was that average body weight increased by around 3kg. That’s a significant amount, but perhaps not the sudden apocalypse often reported. And it is about the same rate of increase that occurred over the preceding few decades. In short, there was no obesity epidemic in the 1980s.

Don’t believe me on this? Here’s a quote from the epidemiologist Katherine Flegal, one of the most respected and cited obesity researchers in the world.

‘Obesity might more properly be regarded as endemic, not epidemic. Data dating back to the Civil War, though limited, suggests that weight has been increasing fairly steadily ever since and increasing at a slower rate now than it did in the last half of the 19th century. A broader perspective then would suggest that the current trends in obesity are a further manifestation of this longer-term trend and not a sudden outbreak of a disease.’

Will anything happen to counter this misinformation? Let’s be clear, I am not saying that obesity is good, or that it should be ignored. I am certainly not saying there are no issues within our food system. I am just pointing out that BMI rose steadily for decades throughout the 20th century, and that rise was almost certainly because the food supply improved dramatically over that time, meaning that fewer people were going hungry. Although there is a strong argument that there has been a disproportionate increase in extremely high BMIs in more recent decades (a different issue entirely in my opinion), nothing special happened in 1978. The more we try to build narratives around an arbitrary cut off point, the further we will move away from solutions that might actually help.

Posted at 11.59 on Friday 10th January