Women, Drinking and Cancer.

[rhythmaning]

There have been lots of reports linking an increase in a range of cancers in women with drinking even small amounts of alcohol - a glass or two of wine a day.

This must have been worrying.

The blog Pyjamas in Bananas has taken the data and analysed it (via @BenGoldacre on Twitter - he runs the blog Bad Science).

To quote from their analysis:

I think it is clear that the difference between a drink a day and less than 2 drinks a week (this was the reference group because non-drinkers often includes ex-alcoholics or those who have given up because they are already sick) is less than convincing.

Looking at their analysis, it seems to make sense. It is so long since I have done that kind of stats that I couldn't challenge it, but based on the graphs they present, any increase in risk at low level of alcohol is not statistically significant.

I'll drink to that!

Comments

[kittenexploring.livejournal.com]

The smoking link is interesting, although the increase in risk still seems so small I would hesitate to call it meaningful. That may just be that after running my own experiments I hesitate to believe any graph based on three points. Three is what you use when you don't have good data. Two means you repeat the work (aka give up and find funding for a new project). Four means either you have limited funding and luckily nothing went wrong or you have good funding but the experiment went badly wrong. Five and up means you either have good luck in funding, good luck in experiments or you're just really good (and combinations thereof).

Apart from that, putting lines to data when it's a biological system is iffy when you approach zero. Small amounts can be protective and large dangerous - like seems to be the case with radiation. Small amounts may be dealt with with no issue and only at large amounts do you get a nice graph - like alcohol poisoning. If you look at the large amounts part for either you get the small amounts wrong.

Then there's all the other influences you get at the same time which confuse this sort of data.

... so, erm, I think I'm trying to say I agree with you. :)

[rhythmaning.livejournal.com]

It's OK - I thought you were agreeing with me, too!