Professor Sir Michael Marmot warns of a cancer epidemic brought on by increased levels of obesity in the UK and across much of the world,splashed across
page 9 of the Observer.
On page 27, the Observer's food critic Jay Rayner rails against fast food and sings the praises of the Government's
Change4Life (a.k.a. "Official: it's OK to bully fat kids and sneer at their parents") programme.
Is this true? Will cancer, as a cause of death, double in the next 40 years? Is it true, as Professor Marmot suggests, that
one third of all cancers are caused by obesity and lack of exercise?
If so, this is scary stuff.
But there are reasons to be cautious.
First, there is a very good reason for cancer deaths to be increasing. Cancer is a disease of the elderly. Younger people do get cancer and do tragically die from it sometimes, but the vast majority of people who get cancer are old - you're most at risk if you're over 80. More of us are dying of cancer because
we're living longer. It's a result of our astonishing success in improving our diet and conquering a host of killer diseases like TB, dyptheria and smallpox that did for so many of our ancestors.
Second, figuring out what causes something like cancer is a tricky business and there simply hasn't been enough research to give us clear and definitive answers.
What do we even mean by saying "
obesity causes cancer"? Do we mean that the obese person who gets cancer wouldn't have got it had they been slimmer? Or that they would have been slightly
less likely to get cancer? How much slimmer do you have to get? Is there some magical chemical change in your body when your BMI passes through the overweight or obese thresholds? Sir Michael suggests we should be
as lean as possible so even having a BMI of under 25 may not be enough to save us.
Take a look at the US National Cancer Institute
factsheet on obesity and cancer. It gives rather different figures.
In 2002, about 41,000 new cases of cancer in the United States were estimated to be due to obesity. This means that about 3.2 percent of all new cancers are linked to obesity.
A recent report estimated that, in the United States, 14 percent of deaths from cancer in men and 20 percent of deaths in women were due to overweight and obesity.
The US page also suggests that, whatever the link between obesity and cancer, there's little evidence that losing weight is helpful anyway.
There is insufficient evidence that intentional weight loss will affect cancer risk for any cancer. A very limited number of observational studies have examined the effect of weight loss, and a few found some decreased risk for breast cancer among women who have lost weight. However, most of these studies have not been able to evaluate whether the weight loss was intentional or related to other health problems.
Over at
JunkFoodScience there's a detailed examination of the 2007 report that also casts doubt on the whole hypothesis of a causal link between obesity and cancer. Sandy Szwarc is of the opinion that the report's authors started off with the assumption that there
was a link and, from the selection of studies for the meta-analysis to the interpretation of data, incorrectly drew conclusions that supported the initial assumption. She goes through the evidence in some detail - take a look.
There's disagreement over both how much evidence we have and what that evidence shows. In short, a classic case for doing more research,
not for declaring the question solved and launching a war against obesity.
A misleading graph
I probably wouldn't have blogged about this at all, were it not for the graph that accompanies the article in the print version of the Observer.
The graph annoyed me sufficiently to take action (yes, I really am sad enough to get annoyed by a graph).
I can't find it online, so I've scanned it in for your pleasure. It's a clear example of trying to twist the data to prove something that probably isn't there. The graph wants to make us believe there's a strong correlation between obesity and cancer. You
are going to believe, whether you want to or not.
What the data actually shows is two vaguely upward lines. From the
data you might as well correlate cancer to the sales of Nintendo Wiis, the amount of recycling we all do or visits to the cinema.
How does the graph try to fool us?
- It carefully matches up the two scales so the lines are close together. The cancer scale is "rates per 100,000 of the population" and the overweight/obese scale is a percentage. The two have nothing in common, but they're helpfully matched up as if 55% overweight and obese is obviously equivalent to 360 people per 100,000 getting cancer.
- Both scales start nowhere near zero. Again, the starting points (350 and 50%) are carefully chosen to give the impression of a strong correlation.
- Both scales only run a short distance (350-380 and 50-65%) so we have nice big, steep lines. If the cancer scale had been, say 0-400, the cancer line running from 356 to 373 per 100,000 would look more or less flat. By zooming in it makes the lines look steep and scary.
- Finally, the start and end years are carefully chosen to scare us more. The cancer data doesn't go beyond 2005, so you'd think it would make sense to stop the graph there. But look! If they'd done that, the last year would have shown obesity falling and cancer rising : definitely not on-message. By tacking on an extra year of obesity data, the graph looks better. Similarly, choosing 1996 or 1997 as the first year would have made the graph much less convincing.
The way I look at it, if Cancer Research UK is putting out such blatant propoganda, such utter misuse of statistics, to scare us into seeing a link, we should treat other pronouncements with a
great deal of caution.
[
Update:
Here's another example of a misleading graph, this time from a Sun opinion poll. ]
Final wordI don't want to give the impression that I'm rejecting a link between cancer and obesity. Sir Michael is an expert who, with his colleagues, has spent a lot of time looking at the evidence. I'm a layman who's done a bit of digging on Google. But I don't think it's nearly as clear-cut as the Observer article would have us believe; nor do I think that this evidence, even if true, supports Government action to force us to eat more healthily and get fit.