Could 45 per cent of all cancers be prevented by improvements in diet, lifestyle and environmental risk factors? That was the message from Cancer Research UK in the wake of a series of papers in British Journal of Cancer by Professor Max Parkin of Queen Mary College, University of London, and colleagues.

CRUK’s press release says: “Overall the review shows that 45 per cent of all cancers in men could be prevented – compared to 40 per cent of all cancers in women.”

But if we read the conclusions of the paper that summarises all the data, we find a more nuanced message. The population-attributable fractions used to make the calculations “should not be used to indicate the percentage of cancers that can currently be prevented by practical means without reference to the individual sections that discuss some of the uncertainties involved.”

But isn’t that exactly what CRUK’s press release has done? It also suggests that the majority of these preventable cancers, more than 100,000 of them, are being caused by smoking, unhealthy diets, alcohol and excess weight, while for Professor Parkin tends to use “caused” in inverted commas.

The truth is that in calculations like these, the uncertainties are often greater than the outcomes. It entirely depends on what attributable fractions you choose to use, and there is a very wide range to choose from. For dietary fibre, for example, some studies have shown no link at all with the risks of colorectal cancer, others a significant one. Men eat more fibre than women, so ought to suffer less colorectal cancer: in fact, they suffer more. The summary paper cites estimates that range from 5 per cent of cases being attributable to low fibre intake to 18 per cent, and chooses 12 per cent.  

Smoking is, of course, the exception – the one example where an exposure can be unambiguously linked to a cancer without any room for argument. Take cigarettes and lung cancer out of the equation, and the number of cancers attributable to diet, drink and excess weight is 40,000 a year. So for non-smokers, which is almost 80 per cent of us, adopting a perfect diet, drinking no alcohol, eating no red or processed meat, and maintaining a body mass index of less than 25 would actually reduce total cancer incidence by 13.3 per cent. Many people might think the sacrifice scarcely worth the reward.

Besides, what really matters to most of us is not dying of cancer – we have to die of something – but dying prematurely of cancer.  About 30 per cent of all deaths are caused by cancer and if nothing else gets us first, cancer eventually will.

The death registration tables published by ONS show that of 141,446 deaths caused by cancer in England and Wales in 2010, more than half (74,450) were in people over 75. Nobody wants to die at any age, but these are not premature deaths. These people have exceeded the life expectancy they were born with.

There’s no harm in pointing out that lifestyle and dietary changes can reduce the risk of cancer. But the actual effects – smoking excluded – are not really as dramatic as the headlines implied. Some of the changes, such as reducing occupational exposures, infections, and radiation, are not really a practical proposition for individuals. Others, such as doubling fibre intake, are implausible.

Professor Parkin himself concludes that although 50 per cent of colorectal cancer cases are attributable to diet, alcohol, physical inactivity and being overweight, only about half is preventable in a 20-year timescale.

That wasn’t really the impression conveyed by CRUK or the media that reported the studies.