Saturday 27 August 2011

Obesity babble

Last night, the BBC News reported as fact that the obesity rate will nearly double to 40% by 2030. It was the considered view of "doctors and scientists" and therefore immune from doubt and criticism. Yesterday, I mentioned the last time "experts" tried to predict the future of obesity in Britain—and managed to get both the magnitude and the direction of the trend wrong.

In fact, I have yet to see an obesity forecast that has not been wildly off base. In 2009, for example, a study in the Journal of Epidemiology and Community Health tried to predict obesity rates in 2012. Not a huge challenge, you might think, to see three years in advance. The researchers concluded that:

If recent trends in adult obesity continue, about a third of all adults (almost 13 million individuals) would be obese by 2012.

To be fair, there's still four months left before we can call this a fail. All that needs to happen in the next four months is for ten percent of the population to suddenly become obese and we will have a winner.

"Off the mark" is probably the kindest way of describing the predictions made about the obesity "pandemic". Take this, from 2004, for example:

Rising obesity will condemn one in ten Britons to diabetes by 2010

At current rates, doctors predict six million British cases of the condition by 2010

One in ten Britons can expect to suffer from diabetes by 2010 as the full impact of the country’s soaring levels of obesity takes its toll on public health.

Doctors’ leaders gave warning yesterday that a diabetes epidemic was now inevitable as new figures revealed that the number of cases has gone up by more than one third in the past eight years.

The report, the first indicator of the repercussions of Britain’s obesity problem, shows that there are 1.8 million diabetics in Britain — up 400,000 cases since 1996. A further million people are estimated to be living with diabetes, but with the condition as yet undiagnosed.

While the Government has predicted a rise of up to three million cases by 2010, doctors said last night that it would be closer to double that level on current rates.

Both estimates turned out to be too high, but the government got much closer than the doctors. By 2010, there were 2.6 million cases.

The phrase "at current rates" is the key to all this. Although the BBC might treat "doctors and scientists" as if they were infallible, the forecast in the Lancet—which is "based on historical trends"—is rooted in the assumption that existing trends will continue at the same rate indefinitely. Actually, it is rather worse than that, because the forecasters have chosen to ignore the flattening out of obesity levels in the last decade and have instead based their figures on the increase between 1985 and 2002, as Nigel Hawkes has noticed:

The actual figures show that the trend from 2002 has actually been flat. For 2009 (not included on the graph, I notice) the proportion of all men who were obese was 22.1 per cent, the same as in 2002. That is far from the impression given.

Although the obesity rate has not risen at all in the last 9 years, the Lancet predicts that it will rise by more than 100% in the next 19 years. This strikes me as being a questionable assumption at best. But leaving the dubious methodology aside, obesity predictions rest on the observation that rates have increased over time and therefore obesity is effectively caused by time itself. That being so, obesity will continue to rise at a consistent and predictable rate each year. It sounds ridiculous when put in these terms—and it is—but that is fundamentally what the "science" of these projections rests on.

It is the same logic that led the banks to believe that property prices would to go up forever. It is the same logic that makes gamblers put large sums of money on "home bankers" (the best football teams playing lesser football teams at home). It is the logic that is superbly documented in Dan Gardner's recent book Future Babble, which shows that tossing a coin is a more reliable way of predicting the future than asking an expert.

One of the reasons expert predictions are so unsuccessful is that they see the future as an exaggerated version of the present. All too often, they do not really understand why the current trend is happening in the first place, but having observed it, they expect it to continue.

This is not a good way of seeing the future at the best of times, but it is particularly foolish when current trends are exceptional. After 9/11, many predicted that massive acts of terrorism would become regular events in the years to come. They didn't. When the economy had an exceptional boom, the experts called it a new paradigm. It wasn't. When the economy came out of recession, some predicted a double-dip recession. It didn't happen (or hasn't yet). The experts were reacting to what was freshest in their minds. They reacted to exceptional events and assumed that they were now the norm.

There is no doubt that the rise in obesity in the last half-century has been exceptional. It is possible that the rise might increase at the same rate and become still more exceptional, or it might rise a little bit, or it might flatten or fall (as the most recent figures from the UK and USA suggest).

The Lancet obesity experts do not take their logic to its natural conclusion and tell us when 100% of the population will be obese, but based on their trend line it will be shortly before the end of this century. If they told us that everybody will be obese by 2100, we would (I hope) find their prediction risible. But it would be an entirely fair prediction to make, based on the assumptions that underpin their current report.

This is not just an reductio ad absurdum argument. Researchers have done exactly that. Take this 2008 study published in the journal Obesity, for example:

If these trends continue, by 2030, 86.3% adults will be overweight or obese; and 51.1%, obese. Black women (96.9%) and Mexican-American men (91.1%) would be the most affected. By 2048, all American adults would become overweight or obese, while black women will reach that state by 2034.

Let us imagine how that work. No American athletes. No American anorexics. Everybody, including anti-obesity campaigners, public health researchers and the author of this study—assuming he lives that long—will be overweight or obese by 2048.

It sounds absurd. It is absurd. Only by the logic of "expert" obesity predictions does it make any sense, because those experts do not acknowledge that people are autonomous beings capable of changing, adapting and taking control of their own lives

And so, while we can't entirely rule out the possibility that the new obesity prediction will prove to be correct, all the previous forecasts have been wrong and so, "based on historical trends", if we were given to predictions, we might predict that this one will be wrong too.


Curmudgeon said...

Forecasts based on the relentless extrapolation of statistics with no thought for real-world counfounding factors inevitably prove to be wrong.

I'm sure I've read that, on current trends, by 2040 the Yemen will have a higher population than Russia. Umm, might some of them emigrate to Russia?

Mike Spilligan said...

I've had to go on a "crash" diet and I lost a stone in a month. If I continue at that rate I'll have disappeared before Christmas 2012.

Ann W. said...

Chris, my understanding is that countries have been lowering the BMI over the last decade, thus creating the epidemic.

In 2008 India lowered the cut off from a BMI of 25 to a BMI of 23.
"A person with a body mass index of 23 kg/m2 will now be considered overweight and below that as one with normal BMI—unlike the cut-off limit of 25 kg/m2 earlier.
Those with BMI of 25 kg/m2 will be clinically termed obese (as opposed to 30 kg/m2 at the international level) and those with BMI of 32.5 kg/m2 will require bariatric surgery to eliminate excess flab."

Malenfant said...

Was this research done by the same team that claims that, at the present rate of growth, 2/3rds of the world's population will be Elvis impersonators by 2040?

Anonymous said...

Over at Frank Davis's place, we just a lttle conversation about the non-doctors who inhabit the DoH.

We have a perfectly good word in our language to describe these non-doctors. The word is QUACKS.

I will be using that word in future to describe all quacks - including scientific ones!

Tomrat said...

Still maintain we are likely to see a contraction in the economy, though do still owe you a crisp fiver for the last time (just).