Vegetables and you

Today's big health news (or, at least, headline)? A vegetarian diet prevents cancer! A study of 60,000 people in the UK, published in the British Journal of Cancer, reports that vegetarians get less cancer of the blood, bladder and stomach. Of 100 meat eaters, 33 will eventually get cancer, while of 100 vegetarians, 29 will do so. If we feel a need to point out that this difference is rather unimpressive, the study does seem to have been independent--at least, it was not paid for by the carrot and broccoli industry! But, vegetarians are more than half as likely to get cancers of the blood and lymph, although the actual number of most cancers, in this sample, was quite small.

The protective effect of the vegetarian diet isn't always true, however--cervical cancer is higher among vegetarians than in meat eaters, though the number of cases was very small, and bowel cancer was slightly higher among those who don't eat meat (contrary to decades of reports that meat-eating, for various reasons having to do with bacterial metabolism of animal fat, increased colorectal cancer).

What mechanism do the authors suggest to explain their findings? Perhaps there are viruses or mutation-causing compounds in meat, or protective compounds in vegetables. Indeed, at least stomach and cervical cancers are known -- and this knowledge does seem to be real! -- to be caused by viruses.

However, something that at least the BBC write-up of the story doesn't point out, a notorious problem with these kinds of studies that should always be pointed out right at the top, is the problem of environmental confounding, in which one measure is correlated with an unmeasured factor. In that case, it is wrong to attribute causation to the former.

It should be clear even to the most obtuse that vegetarians and meat-eaters probably have different life styles in all sorts of ways, which may increase or decrease their risk of exposure to causative environments, having nothing at all to do with diet itself. In this case, diet is merely a marker of life style and risk, not a causative factor, and indeed the study does nothing to control for any such differences. Maybe the dedication of vegetarians to things like Zen meditation affects cancer risk!

Most interestingly, one of the authors of the study was interviewed on the BBC radio program, Newshour, this morning with a fascinating lead-in. Owen Bennett-Jones, the interviewer, pointed out that dietary findings come and go, often being contradicted by subsequent findings--red wine protects against cancer, or it doesn't, dietary fat causes breast cancer, or it doesn't--so why should we believe the results of the vegetarian diet study?

The author himself acknowledged that the findings are not earth-shattering, and may eventually be contradicted, and may only apply to vegetarians in the UK, and these were small numbers of each cancer anyway. He quite burst his own balloon, albeit with the help of his interviewer. His rationale is that after smoking, everything else is a minor risk factor. But, apparently this shouldn't stop researchers from spending large amounts of taxpayer money to look for these minor effects anyway. And hyping them to the media. Hmm. Maybe the vegetable industry should pay for this research!

Bennett-Jones played this story well. He had a science journalist on along with the scientist, and he asked her why so many unreliable stories appear in the media. She said the explanation is very easy--health news sells, especially when it's scary and about cancer. She blames the hype on scientists for wanting to publicize their iffy findings, the industry for wanting to promote the latest food that will prevent cancer, and the media for having to fill the papers and airwaves on deadline. Literally every day, she said, she gets calls from the food industry, or scientists, wanting to tell her about their latest findings.

So, follow a vegetarian diet if you choose to. But don't do it because of the promise that it will prevent cancer.

Fittingly enough, there's a story in the New York Times today, by Gina Kolata, updating us on yesterday's big health news, which suggested that c-reactive proteins, a marker of inflammation, cause heart disease. People were already developing tests for CRPs, counting on the promise that CRPs, rather than cholesterol, were causative and thus were going to need to be tested for in everyone, multiple times. But a new study in the Journal of the American Medical Association, suggests that, yes, CRPs are associated with heart disease, but are not causative.

Interestingly, this study uses a relatively new epidemiological method called Mendelian randomization to show that some people are genetically predisposed to make more CRPs than others, but that CRPs levels themselves aren't associated with heart disease. This is an appropriate use of genetic data.

Well, we have to go now. It's time for lunch. We wonder what's on the menu today.....