Should we eat less red meat?
Eating less red meat could save 17,000 lives, reports The Daily Telegraph in response to a press release from the World Cancer Research Fund, a charity whose aim in life is to demonstrate that cancers are caused by diet.
Some certainly are. Prominent among the suspects are red meat and processed meat, both of which have been linked to an increased risk of colorectal cancers. WCRF urges people to eat less red meat and avoid processed meats altogether.
WCRF issued a press release about its new report embargoed for Monday morning, a good choice as daily newspapers are always anxious to find a “Sunday for Monday” story that doesn’t require too much work. This one could be written on Friday and left for those in charge on Sunday to slot into the paper.
Lots of papers took the hook: apart from The Daily Telegraph, they included the Daily Mail, The Guardian, the Daily Express, and a number of papers who published a different angle on the story, provided by the Press Association’s Jane Kirby – that a diet high in fibre protects against bowel cancer. Those publishing the PA story (or one along similar lines) included The Press and Journal, The Independent, The Western Mail, The Herald, The Daily Mirror, The Evening Times, and The Sun.
The WCRF’s website includes the press release and a link to the report, which is an update of a document first published in 2007. The research team has sought out papers that have been published since then to add weight to its findings.
The effect is rather different from the impression given by the newspaper reports. The extra research may have strengthened the evidence for a link between both red and processed meat and bowel cancer, but it also had the effect of reducing the size of that risk. See Table, below. .
So the actual risk estimates are lower than they were at the time of the 2007 report, which included data up to 2005. The updated report adds ten new studies to the 14 that featured in the 2007 report. It produces more convincing estimates of the risk, with smaller confidence intervals, but a lower overall risk for red and – to a lesser extent - for processed meat.
The data on dietary fibre shows little change, with a relative risk of 0.90 for colorectal cancer for every 10 grams a day consumed, (0.89 for colon cancer and 0.91 for rectal cancer). There’s no reason to question any of this evidence.
I do question, however, the headline figures quoted by WCRF and its partner, the American Institute for Cancer Research, for the number of cases that could be prevented by dietary and behavioural change. WCRF quotes 43 per cent of cases, AICR 45 per cent, but without specifying what people would need to do to achieve this objective, or how large the change in risk of death would be for any individual.
Teresa Nightingale of WCRF says in the press release: “This latest report shows there is enough evidence to recommend that people can reduce their bowel cancel risk by consuming less red and processed meat and alcohol, having more food containing fibre, and by maintaining a healthy weight and being regularly physically active.
“This report confirms that bowel cancer is one of the most preventable types of cancer and we estimate that about 43 per cent of bowel cancer cases in the UK could be prevented through these sorts of changes. That is about 17,000 cases a year.”
I can’t find any mention in the report about this conclusion, how it was reached, and what it means. (The report is 855 pages long, so I may have missed it.) How much less red meat? How much less alcohol, and how much more fibre? Does it really mean eliminating all processed meat, such as bacon or ham, from the diet?
If, as one might suspect, implausible changes in personal behaviour have been assumed in order to achieve this figure (moving from the top quartile to the bottom quartile in red meat consumption, for example) then it is a misleading claim – a pity, since the report itself gives every evidence of having been properly carried out.
Still, the WCRF can’t be blamed for the Daily Telegraph, which says in the headline that 17,000 lives might be saved (it meant cases), and in the text that there are 15,000 deaths a year in the UK from bowel cancer. You can’t save more lives than there are deaths
Daily Telegraph, May 23 2011
David Colquhoun (not verified) wrote,
Tue, 24/05/2011 - 18:14
The claim for lives that could be saved depends entirely on the observed association representing a causal relationship. If it doesn't, then you can stop eating meat but there will be no lives saved.
It is characteristic of chance effects that they tend to get smaller and smaller as more data accumulates (e.g. http://www.dcscience.net/Ioannidis-amer-j-epidemiol-2008.pdf ), so it is even possible that the association is more apparent than real. But even if the association is genuine, it it isn't useful to know it without evidence for causality.
Almost the only form of evidence that one can get in cases like this is a good dose response curve. The relation between number of cigarettes smoked and risk of lung cancer is clear and unambiguous. In contrast, the dose response curves for red meat were all (with one possible exception) horizontal, when I first wrote about this question in 2009 http://www.dcscience.net/?p=1435
The new dose-response curves confirm this: they are more or less horizontal too.
The lack of any reasonable relationship between dose and response makes the evidence for causality very thin indeed, and getting thinner with each new report.
That being the case, one might ask whether the annual shock warnings issued by WCRF are taking the precautionary principle too far?
Richard Evans, World Cancer Research Fund (not verified) wrote,
Wed, 25/05/2011 - 09:26
It is incorrect to say we are a charity “whose aim in life is to demonstrate that cancers are caused by diet”. We are committed to finding out how diet, physical activity and weight affect cancer risk and then communicating the findings to the public, whatever those findings are.
The new report’s judgements, like those of our 2007 Expert Report, are not those of WCRF but of an independent panel of some of the world’s leading experts in the field.
To answer the question about the preventability estimates, they come from Policy and Action for Cancer Prevention, a report we released in 2009 and which can be downloaded at www.dietandcancerreport.org. The report contains a full explanation of the methodology for these estimates, or there is an explanation aimed at a non-scientific audience at www.wcrf-uk.org/research/cancer_stats_preventability_estimates.php
But briefly, for each cancer risk factor dietary information for the UK was used to divide the population into three groups: low consumption, medium consumption, and high consumption. The preventability estimates are based on what would happen if, for each risk factor, everyone did the same as the healthiest group do at the moment. For processed meat, for example, this would mean everyone eating under 10g per day.
The reason for this methodology is so that the figure is something that is realistically achievable.
Nigel Hawkes (not verified) wrote,
Wed, 25/05/2011 - 12:55
Thanks, Richard, for clarifying that. It would have been good to have a link to this methodology in the press release. It enables people to judge if you estimates for preventability are credible.
David Colquhoun (not verified) wrote,
Sat, 28/05/2011 - 15:10
@Richard Evans
I'm not suspecting your motives, which I'm sure are entirely good. but I do wish you'd address my major concern which is the lack of any detectable dose response relationship. That has been confirmed in the new data. I simply don't understand how you can sound so confident about causality when the one bit of evidence that you could adduce for causality is not there.
Each time I raise the matter, it is evaded. The same happened when Wiseman replied to my original blog. I got a lot of stuff about the qualifications of the people who'd written the report, but no comment at all on the real question. I'd really like to hear your comments.
Anonymous (not verified) wrote,
Sat, 28/05/2011 - 21:10
I don't think that WCRF are ever likely to talk about methodology in their press releases because press releases are designed for impact and therefore rarely emphasise the limitations of the methods used or the minute absolute risks the "findings" reveal.
The expert panel Richard talks about consists mainly of nutritionists who unsurprisingly believe that diet is important in cancer cause /prevention.
I am sure that well meaning people are involved with WCRF but I am concerned by their methods and also their ethics. The dubious nature of the latter were brought into focus by a fund raising letter syndicated through local newspapers in 2010 that encouraged readers to make out a will in favour of WCRF using WCRFs Personal Will Organiser. A free organiser was offered to the first 100 readers to call WCRF. Readers were encouraged to leave a legacy in WCRFs favour in order to ensure that:
“even more people know about our life-saving message that a third of the most common cancers could be prevented by following a healthy diet, being physically active and maintaining a healthy weight.”
I find it difficult to relate this somewhat evangelical claim with Richard’s comment that WCRF is about “communicating the findings to the public, whatever those findings are”. Are 33% of cancers really lifestyle related and is it ethical to use that statistic to encourage people to make a will in your favour? I am not convinced but am open minded.
Richard Evans, World Cancer Research Fund (not verified) wrote,
Tue, 31/05/2011 - 11:21
David, I’m happy to respond to your suggestion that the dose response relationship is “more or less horizontal”. Just to make it clear, though, that the judgment that the evidence is convincing is not mine but that of an independent panel of some of the world’s leading experts in the field.
On the issue of dose response, you are right that the graph for smoking looks much more dramatic than that for meat, but this is always going to be the case for a risk factor (such as meat and bowel cancer) where the increase to risk is much smaller amount than the large increase that smoking has for lung cancer. But although the graph might not look dramatic, the dose response is there and the statistics suggest it is very unlikely to be there by chance.
The forest plots that are based on a “per 100g” basis are precisely a representation of a dose response; and statistical analysis is a better way of assessing that than looking at the pictures, which are designed to show the shape of the dose response, rather than its presence.
It was this dose response that was one of the main reasons that the Panel judged that there is convincing evidence that both red and processed meat increase bowel cancer risk.
Richard Evans, World Cancer Research Fund (not verified) wrote,
Mon, 06/06/2011 - 15:34
We are able to give independent health advice because we reply on funding from ordinary members of the public and one of the ways people can support us is by leaving a gift in their will.
We also promote the fact that a third of most common cancers in the UK could be prevented through diet, physical activity and weight maintenance. We do this to highlight how the importance of cancer prevention, and also to help raise awareness of these risk factors because it is too low at the moment.
But while this is an estimate, it is an evidence-based one and comes from our 2009 report, Policy and Action for Cancer Prevention.