The BMJ study isn’t the final word on mammograms

by Allan Joseph

By now, nearly everyone who’s interested in health policy has heard about the huge study published in the BMJ this week — a 25-year randomized, controlled trial of mammography that shows no effect on mortality. (If you haven’t, Aaron Carroll’s excellent summary at the Incidental Economist is a good place to start.) The study comes at a time when breast-cancer screening is hotly debated, as is other preventive screening for cancers. Much has been made over the difference between survival rates and mortality rates, and whether the study is actually an appropriate randomized controlled trial (RCT), but that’s not what I want to talk about.

Let’s just, for the moment, assume that we all agree that the BMJ is methodologically perfect and externally valid. (If you’re relying on survival rates, let’s set that aside for now and go with mortality rates as the best measure of patient death due to illness). If we assume all of that, we come to a pretty sobering conclusion: mammograms don’t save lives.

So is that it? Should women not be getting mammograms? No — that would be far too strong a conclusion to draw, for two reasons.

1. Don’t forget about high-risk women. The BMJ study examined results among the population at large. Women with a history of breast cancer (who have a higher risk) weren’t in the study, nor did they analyze the data for women with family histories of breast cancer, BRCA genes, or biological risk factors such as lobular carcinoma in situ (LCIS). These are the women at highest risk for developing breast cancer, and the research doesn’t really say anything about them — only about the population at large. Until we have better data on screening high-risk women, we should be very cautious about changing the way high-risk women are screened. (As an aside, it’ll be hard to get a good piece of data on high-risk women — how could you possibly run an experiment?)

2. Mortality isn’t everything. Mortality rates are really important for measuring whether an intervention actually saves lives. But we’re not only interested in whether women with breast cancer live, and for how long. We also want to know about the quality of life. Does early detection mean less chemotherapy? Does it mean fewer full mastectomies and invasive surgeries? The BMJ study reported that tumors in the non-screening group were bigger than those in the mammography group (2.1 cm v. 1.4 cm, p<0.001), suggesting that they were detected later. It’s plausible that the two groups could get the same results in terms of mortality, but that the control group had to undergo more aggressive treatment to get the same results. Mammograms might not save lives, but they might save breast tissue and reduce chemotherapy regimens. If that’s the case, they very well might still be worth the price of over-diagnosis.

This study isn’t the final word on mammography. It’s certainly impressive and eye-catching research, and it should be — it’s extraordinary on scale alone. But even with some strong assumptions (external validity being the biggest), we’re not yet able to make blanket statements on mammography. We’re getting closer to knowing more, but we don’t have a final answer yet.


Allan Joseph is a first year medical student at the Warren Alpert Medical School of Brown University, where he is pursuing an MD/MPP. You can follow him on Twitter @allanmjoseph.


One thought on “The BMJ study isn’t the final word on mammograms

  1. Jim Whiting says:

    No one wants to believe that Mammograms are only marginally effective!

    Jim Whiting, Never Stop Exploring

    Sent from my iPad


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: