A well-done analysis in the BMJ this week calls into question previous research that has been used to tout mammography as an effective tool for lowering breast cancer mortality in Denmark. That previous study compared breast cancer death rates in Copenhagen, where women were offered screening mammography in 1991, to areas in Denmark where mammograms were not offered until 17 years later, and concluded that the introduction of mammogram screening resulted in a 25 % reduction in breast cancer mortality in screened areas.
The new study adds an additional county where screening was offered (with a little implication that perhaps the previous researchers should have included this other area, but I’ll stay out of the academic finger pointing) and then reanalyzes the data.
The researchers found that breast cancer deaths declined nationwide during the time period studied, in all areas, regardless if that area was one offering the screening program. Much of this decline occurred in women ages 40-49, who were too young to have been offered screening. This suggests that it is breast cancer treatment rather than screening that should take the credit for most of the mortality declines in Denmark over the time period studied.
The researchers then go on to make this statement:
We believe it is time to question whether screening has delivered the promised effect on breast cancer mortality.
This practically guaranteed that I’d have to read their paper and comment on it. So I am.
I’m not convinced that this paper makes the point that mammograms are ineffective. The authors themselves argue that the effect size of mammography, estimated at about 15-16% in randomized trials, is too small to be measured in epidimiologic studies. I agree. It is just impossible to control all the confounding factors inherent in an entire population of individuals to tease out the effect of a single intervention over time, particularly when breast cancer treatment was evolving so rapidly over the time period being studied.
I have to admit that I have a hard time believing that, for 17 years, women living outside of Copenhagen never entered that fair city to have a mammogram on their own dollar once they found out that their city-dwelling friends were being offered the test and they weren’t. (The paper used to support the claim doesn’t make the case in my opinion.) I know that Europeans have not bought the whole mammogram thing hook, line and sinker the way we here in the States have, but I don’t think it is as black and white as the Danes would like us to believe. (If you are a Danish woman reading this, feel free to enlighten us…) But that’s just an aside.
This is an important paper in that it effectively refutes previous conclusions about the Danish mammogram screening program. Unfortunately, I don’t think this study stands on any stronger ground in arguing that mammograms are ineffective. Nothing in this paper rescinds the results of the randomized trials, which the authors themselves state find a mortality reduction of about 15-16% for mammography. In my opinion, their results primarily show us that population based data is nearly impossible to use to make any valid conclusions about mammogram screening — either for or against it.
Hat tip to Gary Schwitzer for pointing me to this study
*This blog post was originally published at The Blog that Ate Manhattan*