Monday, February 16, 2004


The media never get tired of telling us - especially the female half of us - that something is going to make us sick. Today's example is an article in the Washington Post: "Study Links Breast Cancer to Antibiotics Use."
Antibiotic use is associated with an increased risk for breast cancer, a new study has found, raising the possibility that women who take the widely used medicines are prone to one of the most feared malignancies.
There it is, right in the first sentence: the risk is almost impossible to avoid (who hasn't taken antibiotics?), and the disease is one of the most feared (it's not the biggest killer of women, or the most lethal cancer in women, or anything like that, so they're pretty much relying on subjective dread to get our hearts pumping here).

When a report of a scientific study uses the term "association," usually they mean a correlation. Correlation simply means that two factors tend to co-occur. It doesn't tell you a thing in the world about why the two factors co-occur, which is kind of a bitch for people who want to interpret correlations in exciting ways. There are three main causal possibilities when you're dealing with a correlation: (1) Factor A could cause factor B. (2) Factor B could cause factor A. (3) A third factor (C) could cause both A and B.

Despite the shock headline of the Post article and its moral that doctors should be even yet still more careful about prescribing antibiotics to women, this story seems like a prime candidate for explanation (3). Women who take a lot of antibiotics are probably sicker than women who don't take a lot of antibiotics. If they're coming down with a lot of infections, they probably have weaker immune systems. Guess what else the immune system fights? Developing tumors.

Why didn't they control for the initial health of the women in the study, or collect measures of immune function? Because they didn't have the opportunity. This was what's called retrospective research. They had a lot of data on the health care used by women belonging to a particular health plan, and they sifted through it looking for relationships between medications and cancer. So they didn't have much information on the functioning of these women's immune systems, or on other breast cancer risk factors. The authors of the study admit as much in their abstract in the Journal of the American Medical Association, but the same information is buried in the popular press article.

We have to read through to the last paragraph to get some words of sense from the American Cancer Society:
One finding that cast doubt on the possibility that it will turn out that antibiotics increase the risk for breast cancer was that the study found the risk for all types of antibiotics, said Debbie Saslow, director of breast and gynecologic cancer for the American Cancer Society. That makes it unlikely it's the antibiotics because they different classes work in very different ways, she said.
They'll do some more research on this one, and probably they'll find out that the pills you innocently took for strep throat when you were twenty aren't going to kill you when you're seventy. But in the meantime, lots of women are being made to feel very, very afraid.