Why not subscribe?

Thursday, March 05, 2009

Alcohol-Cancer link in women questioned

http://www.spiked-online.com/index.php?/site/article/6316/

 

Patrick Basham and John Luik aren’t convinced the data show there’s a link:

 

“[Oxford’s Naomi] Allen came across with scary news for Americans, telling the Washington Post that the ‘take-home message’ was this: ‘If you are regularly drinking even one drink per day, that’s increasing your risk for cancer [since] there doesn’t seem to be a threshold at which alcohol consumption is safe.’ “

They offer several criticisms of the study, the most serious of which is this:

“the study is full of significant puzzles that suggest that its results are unreliable. For example, it reports that the incidence of all types of cancer studied in its non-drinking subjects was 5.7 per cent compared with 5.3 per cent for those subjects who had at least a drink a day, and up to 14 drinks a week.

“In other words, not only was there no dose-response in terms of cancer risk, but teetotallers had a higher population incidence of cancer than those consuming up to 14 drinks a week!

“Even those women in the study who drank the most (15 or more drinks a week) had a cancer incidence of 5.8 per cent, which is virtually identical to those who drank nothing.”

Allen’s results aren’t impossible – by the time you control for factors such as age, income, urban/rural, etc. it’s possible to come up with results that are the reverse of the ones reported at first. The statistician Donald Rubin provides some good examples in his articles on propensity scores.  But reversals are relatively rare, and deserve scrutiny.  If I see a rejoinder by Allen, I’ll link to it here.