Vegetarian Cancer Statistics

Popped up on the news, but some more detail from the Guardian about how vegetarians are less likely to contract cancer. Let's look at the figures.

61,000 people were 'followed' by scientists. 3, 350 contracted cancer during the study. So let's stop there and assume that the people in the study were representative of the population of the UK in terms of age, ethicity, etc. That means the chances of getting cancer are 3350/61000*100 or 5.5%.

Of that number 2, 204 were meat-eaters, 800 were vegetarians and 300 ate fish but not meat (shall I ask why fish isn't meat? No let's just say they didn't eat red meat). They helpfully provide percentages too; their figures 68%, 24% and 9.5% totalling 101.5% um.

Okay first off 2204+800+300=3304 and not 3,350 so are they basing the percentages on a total of 3,350 or 3,304 (obviously not 61,000)?

If it's 3,350 then the results are 65.8%, 23.9% and 9.0% with rounding (1.4% not counted);
If it's 3,304 then the results are 66.7%, 24.2% and 9.1% with rounding; again
If it's Guardian the results are 68%, 24% and 9.5%... yeah.

Okay ignoring the bit about the specific blood cancer and focussing on the overall cancer rate the article led off with the point that vegetarians "are 12% less likely to develop cancer overall." So add the Guardian's 68% and subtract the 24% and 9.5% and you get 34.5%; hold up that's a bigger number so why wasn't it published and where did 12% come from? Even if you take it as a percentage of 68% you get 35.3%

Let's go back to the total study figure of 61,000 and break it down from there. That gives 3.6%, 1.3% and 0.5%; to clarify if that's representative then you have a 3.6% chance of contracting cancer, but only a 1.3% (or 1.8%) if you're a vegetarian.

Even if I take the overall 5.5% and use that as the base i..e 3.6/5.5*100 etc. I get 65.5% and 23.6 (ignoring the 0.5%)

So still no sign of the 12%, but this morning the figures I heard were that the chances of getting cancer were held to be 33% and 29% respectively and that is indeed a 12% difference (or 4% if you just subtract them).

Yet from the figures here I can't pull out that 33% or 29% from anywhere. It seems I have to turn to the full study which makes me wonder what the point of a newspaper is?

Oh look those figure of 2204, 800 and 300 have become 2204, 829 and 317 which adds up and are shown in the very first paragraph of the report the Guardian links to, did anyone actually read it?

Yes it's a bit dense, but the 12% is there in the data. It's not easy as they've combined some figures as well as split them up for instance there's a combined figure for C18-20 and then they've been separated out into C18 and C19-20, or in some case they overlap without a split so you can't simply add up the figures and get the same totals, which is bad; bad data smack on the nose. That means all we have is their relative risk totals and for meat eaters it's set to 1.00 (for obvious reasons as it has to be relative to something) for fish-eaters it's 0.82 and vegetarians 0.88. The difference between the first and last being 0.12 which is the 12%.

So the headline figure is accurately based on relative risk from data that can't be totalled as presented in the report. To put all this back in perspective however lets go back to the overall figures once more and turn them into layman language.

If you know 1000 people odds are that around 55 of them will develop cancer sometime within twelve years. Of those fifty-five 36 of them will be meat-eaters and 18 won't be. If you consider that a good reason to give up red meat then by all means do so.

Still not sure where the 33% and 29% came from though.