I'm A Doctor, Not A Statistician

It would appear that alchoholic beverages are linked to several kinds of cancers (link). Doctor Paolo Boffetta of the International Agency for Research on Cancer announced the results of studies showing the link, and said "Given the linear dose-response relation between alcohol intake and risk of cancer, control of heavy drinking remains the main target for cancer control." That's pretty damning if you're a whiskey. "Linear dose-response relation" is science talk for "the bigger the dose, the bigger the response," and since the response in this case is cancer, a zero-dose would be safest. But Boffetta and his team aren't advocating teetotaling, because a LITTLE bit of alchohol offers some protection against cardiovascular disease. And here's where I think they blow it with the numbers game: From the linked article... The World Health Organization estimates year 2000 alchohol-related deaths at 185,000 men and 142,000 women. It estimates alchohol prevented 71,000 male deaths and 277,000 female deaths the same year. These statistics do not show that alchohol in moderation should still be okay. They show that the net effect of alchohol was 114,000 dead men and 135,000 living women. Those are the "Developed Countries" numbers. In "Developing Countries" where cardiovascular disease is less of a problem, alchohol was linked to 1.52 million dead men and 301,000 dead women, and wasn't linked to ANY prevention of heart disease. I don't care how you crunch those numbers (all of which came from the same Reuters/CNN article, and, I presume, the press release from the IARC and WHO): the statistics do not show that moderation in alchohol is better. They show the following: 1) Further research is needed into why developing nations have a lower incidence of heart disease. 2) If you want to live longer, drink less (where "less" equals "approaching zero") and then find OTHER ways to take care of your heart. See #1 above. (Obligatory disclaimer: I haven't been given access to the raw data. I'm not your doctor. I'm not your priest. I'm your cartoonist, and then only once each day.) They say you can prove anything with statistics. The trick lies in providing numbers that support the point you want to prove. Dr. Boffetta has blown it -- He needs to take a line from Doctor McCoy, and admit that he's out of his depth, because the numbers he uses undermine his point rather than supporting it.
Next Post » « Previous Post
blog comments powered by Disqus