Big studies: Size isn't everything

My newsfeeder this morning had a news release and a couple of news articles about a big study in Japan that showed that consumption of green tea can reduce deaths from all causes, in particular lowering death from cardiovascular disease. As one of the trendiest, not to mention lucrative, dietary ingredients to hit Western culture in a long time, green tea consumption has been begging for a good study. Personally, since I'm now in the habit of laying out $3.50 a pop for a Starbucks green tea frappuccino, I'm curious to know whether I'm getting anything for my money besides sugar, milk, and a mild caffeine buzz. I hope I'm not alone in my curiosity - as I've said before, consumers have a right to know whether the supplements they shell out money for actually do what they're supposed to do.

The study is big, really big, not just in terms of all participants, but also in the numbers of those who died; both are important for robustness - it means the sample will hold up to a lot of subgroup or multivariable analyses with which one controls for confounders. The researchers followed 40,530 healthy Japanese adults, observed what they ate (among 36 foods) and drank (black, green, and oolong tea, and coffee), obtained information at baseline on a few other lifestyle factors, and measured their causes of death. The cohort was followed for 11 years for deaths due to all causes and up to 7 years for deaths due to specific causes.

People who drank 5 or more cups a day experienced the most apparent benefit; death from all causes was reduced by 12% for men and 23% for women, and women who drank 3-4 cups also experienced statistically significant benefit. When the sex groups were combined, there was an all-cause mortality benefit only in the 5+ cups/day group. Similar benefits to the overall mortality results were seen for cardiovascular disease deaths (after 11 years of follow-up), but no apparent benefit (or risk) was seen for cancer deaths.

So what does it all mean? Well, first of all, this is a pretty darn good study, but it can't prove that green tea actually prevents death from cardiovascular disease and all causes. That's because there may be other factors that the study was unable to anticipate or measure that were associated with both green tea drinking and reduced mortality (though they did account for some good ones, as noted above). Also, dietary measurement is notoriously lousy, and that could have resulted in some undercounting as well, possibly even in a systematic way between heavy and lighter tea drinkers, for example. A study that could prove benefit, and also prove safety, would also have to be fairly large to look at deaths as an outcome, but would also have to be randomized, and since ethical issues associated with randomizing people to drink green tea are obscure to this writer, you might be able to do it. A large randomized study would be quite expensive though, and who would pay for it? Green tea manufacturers? I doubt it. Until consumers really come to understand the hierarchy of study designs - hint! - there just may not be a demand for it. Until then I guess I'll just have to keep jumping up and down.

Something else to think about - the very large number of people in the study who drank 5 or more cups of green tea a day. Could we do it in the U.S.? That is, sign up several thousand people and get them to drink all that tea, or some placebo, and what would that placebo be?

I'll be interested to see whether actual public health recommendations come out of this study.
green tea
complementary medicine
epidemiology
evidence
prevention