Sunday, 6 February 2011

Global Average Weight

Did you know that over half of American adults are overweight (55%)? (1)

Is obesity rising worldwide? Or is it that an affluent portion of the world is skewing results? And how would we measure it anyway? What is the "average weight" of, say, a 20 year old, worldwide? Or a thirty year old? How does it compare with previous records? And how can we interpret the results?

I've chosen this as an example of global averages because, in theory at least, it appears reasonably easy to get a statistical handle on. After all, we are talking about individuals, and while we can't count anyone, we can take samples across different parts of the globe. Weight may differ from day to day, but there are not going to be that many massive fluctuations, when weight surges up or down - unless, of course, there are natural disasters.

What is "overweight" is a matter of definition, and it is not just what you show up when you get on the scales. Your height also needs to be taken into account, as does gender. For the World Health Organisation,

for adults, a body mass index of 25 or more is considered "overweight" and a BMI of 30 or more is considered "obese"

The body mass index (BMI), or Quetelet index, is a heuristic proxy for human body fat based on an individual's weight and height. BMI does not actually measure the percentage of body fat. Body mass index is defined as the individual's body weight divided by the square of his or her height.

Now for starters, the formula used has built in measurement biases and it doesn't result in perfect measurements. Some of the exceptions are described below (1):

- Body Mass Index changes with Age, obviously in Children but also in Adults.
- Men and Women are different, so why should adult men and women have exactly the same BMI?
- Short adult women have higher BMI than taller women.
- Race/ethnicity and nationality affect body composition and BMI.
- Muscular people, athletes and bodybuilders particularly, have high BMI values, but are not fat.

Given the exceptions, it is perhaps surprising that the index is still used. And what on earth can it mean to say that over half of American adults are overweight (55%)?

A recent study suggests South Africans are putting on weight faster than the global average. The study was published in The Lancet Magazine and is based on body mass index, the ration between weight and height. It shows that three in four South African women are too heavy, while two in three men have the same problem. (2)

That is journalism, though, but a document on "Heart Disease in South Africa" supplied to the media by Department of Medicine, University of Cape Town & Chronic Diseases of Lifestyle Unit, at the Medical Research Council, has this to say:

In 1998, 56% of women and 29% of men, aged 15 years or older in South Africa, were overweight or obese. These high rates had not changed by 2003. (3)

Hidden away in a footnote is the comment that this data comes from a random sample of about 14 000 South Africans. Given that South Africa's population is around 49 million, we are talking about a sample size of 0.03%.. The survey has more detailed breakdown, and notes that:

Urban people had higher weights than people who were living outside the cities

For men the obesity rate was the highest in whites (18%), followed by Indians and coloureds (8%) and then by Africans (6%) (3)

But where is it easier to get samples that are roughly equivalent? In the cities. Outside the cities, weights may vary wildly depending on the terrain and availability of food, and it is unlikely that such a small survey could provide much in the way of accurate quantification of any diversity there. Note also the different ethnic groups were all treated alike in measurement of BMI, even though we know this is a conceptual flaw in the measurements.

But what did the original article mean by "the global average"? To make a global estimate, different statistical results have to be taken by all the different reporting agencies and combined. This means the total BMI of those sampled, and the total count, so that the results can be combined. If you just take averages from one area and obtain an average of averages, that would result in a different figure unless the samples sizes were both the same, so a lot of reverse calculation has to take place to ensure that the correct figures are obtained.

Yet even so, as we have seen from the South African figures, the sample size is very small, and is probably biased towards urban populations. The global average BMI, which seemed such a simple ideal, is in fact, fraught with problems, both with the instrument of measurement (the BMI) and the sampling of data over a planet. How easy is it to sample famine ridden troublespots of the world? How easy is it to sample people living in  the more remote and inaccessible areas of the planet? Is there a measurement bias towards locations that can be easily measured?

But one thing is surely clear - it is a matter of how human beings feed themselves. They are the causal agents for their own obesity. Which makes a paper in "The Scientist" entitled "Animals are getting fatter too" somewhat startling in its findings:

Obesity levels have risen dramatically in research animals and others living close to humans, suggesting environmental factors are encouraging everyone to gain weight, according to new findings in the Proceedings of the Royal Society. It's no secret that obesity has become an epidemic in humans -- among American adults, nearly one in three is obese, defined as having a Body Mass Index (BMI) greater than 30. Researchers have pointed their fingers at everything from a lack of physical activity to the highly processed foods that so many of us eat.

But what if something in the environment was at least partly to blame, as well? To investigate, David Allison, a statistical geneticist at the University of Alabama, Birmingham, and his colleagues gathered data on body weights of more than 20,000 adult animals from 24 populations of 8 different species from around North America. The authors included only those mammals for which there were two weight measurements in the past 50 years, and whose weight was not deliberately manipulated as part of research or a livestock feeding program.

All 24 populations of animals, which ranged from primates housed in research facilities to feral rats living in the greater Baltimore area, showed significant increases in body weight...Not only did body weight increase significantly, but so did the chances than an animal would be obese. In 23 out of the 24 populations, animals were more likely to be obese -- defined as weight above the 85th percentile at the initial time-point -- at the second time-point than at the first.  What's more, the increased body weights and increased likelihood of obesity were found even in animals whose diets and physical activity levels were known to be the same throughout the study period. So if dietary changes and energy imbalance weren't responsible for the rise in obesity, said Allison, it may be some environmental factor. (5)

Now the scientists in question don't think that obesity cannot be improved by better diet and exercise, but they do show that there may be other factors at work, which may also contribute to trends. While this is obviously only a preliminary study (published last November), it has not been criticised on any political ground by health groups or taken up by food manufacturers keen to show that their food does not contribute to rising obesity! But it does show that a single causal explanation may not be the only one, and weight rise across the Western world might be more complex than first thought.

Returning to our starting point:

Obesity is a growing global health problem (4)

This sentence illustrates another problem with talking about average global trends. By simplifying the figure to one unit (an average), the spread, the statistical measures of diversity, are smoothed out, and this can be dangerous because it hides the complete picture of what is going on in the world. While parts of the world may have a major problem with obesity, starvation, famine, and high rates of mortality are the lot of other parts of the globe, where any food is in short supply. Extremes are flattened out with averages, and sometimes it is more important to look at the extremes, because they can give a more complete picture than the average, as can measures of spread, like standard deviations from the mean.

There can be little doubt that obesity, in the wealthier countries of the world, is increasing, and is in part due to diet and exercise, although the interplay of other factors may be more subtle but more important than we care to realise. For example, the perceived risk of being on roads cycling with higher volumes of traffic may lead to less children cycling; the perceived risk of safety from predatory individuals, or just plain health and safety may lead to children being less likely to be encouraged to go out walking on their own. Increased pressure on parents working may lead to children left in sedentary pastimes more often. Health officials often come out with overly simplistic mantras of how to solve the problem of obesity, but the structural nature of our society also has a contributory effect that gets overlooked, and should also be tackled.

Measures of obesity can be useful in highlighting a trend, but the more they simplify the data, the less they are likely to provide a sharp focus for action. Figures can be helpful, but people don't react well to figures. The change in advertisements, as detailed in Vance Packard's seminal book "The Hidden Persuaders", shows that early advertisements, packed with facts, facts, and yet more facts, did not sell the products well, because people are not computers, hard-wired to respond to "facts". When asked "What do you think about..?", people often express feelings, not thoughts.

The "fact" is, after all, an abstraction, taking a measurement from a whole, and while we can do a lot with statistics, getting people to respond to bare statistics (or just hitting them over the head with a barrage of statistics), just doesn't work. As Karen Armstrong has noted:

When we're really, really frightened or a child or animal is frightened, it's no use telling them to pull themselves together, be more rational, get on with life and face facts.  It doesn't work like that.  People become paralyzed. (6)

We shouldn't dispense with facts, but as writers like Charles Dickens or Victor Hugo showed so memorably, facts alone are not enough. In Dickens, the teacher Mr Gradgrind is obsessed with facts, and pushes them at the children all the harder when the children don't respond like robots. He is a monster, whose obsession has caused him to become totally unbalanced as a human being. But that didn't mean Dickens was against facts, simply that he could see the dangers of a purely fact based monomania.
 
Dickens wanted to reach out and engage with the widest population that he could on his social concerns about poverty and injustice. It is well known that he first considered writing a polemical essay or pamphlet on the subject - no doubt emotive, but primarily factual; instead he wrote "A Christmas Carol". If he had presented the bare facts, would his writing still capture the minds of our day? By using both imagination and fact, he engaged not only with the public of his day, but future generations, in the call for generosity to the poor, as he takes us through the journey of Scrooge, and at the end, not only is Scrooge changed, so is the reader. We need to recover something of this approach, what Karen Armstrong termed "mythos" to reach people today; bare "logos" is not enough.

Links

(1) http://www.halls.md/body-mass-index/bmirefs.htm
(2) http://www.eyewitnessnews.co.za/articleprog.aspx?id=58714
(3) http://www.heartfoundation.co.za/docs/heartmonth/HeartDiseaseinSA.pdf
(4) http://bigthink.com/ideas/25517
(5) http://www.the-scientist.com/news/display/57821/
(6) http://faithljustice.wordpress.com/2010/05/08/karen-armstrong-part-ii/

3 comments:

st-ouennais said...

The obesity increase in animals is interesting. In farm animals, there is the possibility this is selective breeding. In pets it may be environmental. One possibility is the impact of endochrine diruptors in common platics.

plastics

st-ouennais said...

The obesity increase in animals is interesting. In farm animals, there is the possibility this is selective breeding. In pets it may be environmental. One possibility is the impact of endochrine diruptors in common platics.

plastics

Nick Palmer said...

I've always thought that BMI was a rubbish idea to establish "obesity". The obvious falsifier is the athletic heavyweights like Muhammad Ali (showing my age here).

I reckon a relatively simple addition to the BMI score would correlate it with a "can you pinch more than an inch" measurement of surface body fat. The two together should give more realistic results.

Like St Ouennais, my gut feeing was that the hormone mimicking plasticisers, such as BPA, might be responsible not only for fatter animals but also a whole slew of other 20th/21st C anomalies in people - ADHD anyone?