Wednesday, November 2, 2011

Why the new study on alcohol and breast cancer got it wrong - again

Big news! The latest study on the association between alcohol and breast cancer found what all the numerous prior studies using the same methods found: even small amounts of consumption increase the risk, regardless of the type of alcoholic beverage, even red wine. But as I point out in my book Age Gets Better with Wine, they are simply repeating the same mistakes and failing to see the big picture. Here’s why:

Self-reporting bias. Studies such as this, which seem to derive power from their large numbers, only magnify the errors if the data isn’t reliable. The nurses in this study were asked to fill out questionnaires on their drinking habits and other lifestyle factors every 6 months. It is widely acknowledged that this retrospective self-reporting is highly unreliable. So having a hundred thousand or even a million participants doesn’t yield stronger data, it just magnifies the error. Statisticians are of course aware of this and attempt to make adjustments according to known behaviors, but in a sense it would be better to use a smaller number of subjects and observe them more closely.

No distinction between different drinking patterns. If we were to design a study that could accurately measure the effects of say red wine vs beer or spirits, it would look like this: one group drinks only red wine, in the same amounts, every day, while the others do the same for their assigned beverage. They would be closely followed for many years. This is clearly not the case with the nurses study, which simply asked people what sort of drinks they prefer. By far the vast majority have mixed drinking patterns, both in amounts, types of drinks, and daily patterns. There is simply no realistic way to infer anything about the different drinks from this. On the other hand, studies from areas where drinking patterns are consistent for wine show a substantial decrease in breast cancer incidence.

Ignoring the big picture: Let’s put the numbers in perspective: the overall lifetime risk of breast cancer is around 9 or 10 percent, so a 10% increase in risk from a couple of drinks a week raises it to around 11%, and a 50% increase from heavy drinking brings it up to 15%. But far and away the leading cause of death in women is heart disease (1 in 3), and regular wine consumption clearly reduces that risk. Breast cancer, at 1 in 36, is a ways down the list. Add in also Alzheimer’s, hip fractures from osteoporosis, and diabetes, all of which are reduced among wine drinkers, and you get a very different picture.

That’s why I tried to portray the bigger picture in my book. There is little question that the net effect of regular wine consumption, especially wine with meals, is positive both in terms of disease incidence, lifespan, and quality of life.