The U.S. preventive services task force (USPSTF) is updating its breast cancer screening recommendations. You might think it’s a slam-dunk this time around, that the data against mammography are overwhelming, so persuasive that we should just give it up. In fact, the Swiss Medical Board has considered doing just that. But a close read reveals flaws in the evidence.
For today, I’ll focus on dates. As it turns out, recent analyses cite studies carried out during the 1980s, 1970s and 1960s. In other words, researchers are analyzing our grandmothers' mammograms to inform women’s health and screening decisions today.
2013 Cochrane review on mammography, with magnifying lens
This wouldn’t matter so much if mammography were a static tool, a method that hasn’t improved much since 1985. But it’s not a simple lab test, like a hemoglobin level or a PSA (what might be used to screen for prostate cancer.) Mammograms are x-ray images of the breast that might detect a small cancer before you or a doctor can feel it.
Consider how much televisions and cameras have improved since 1985. Would you choose a cell phone or TV set based on how reliably and clearly those worked in 1990? Unlikely.
A recent New York Times “Upshot” article emphasized the concepts of the Number Needed to Treat (NNT) and Number Needed to Harm (NNH). For screening mammography, the authors presented dismal statistics. But the single reference listed (and to which the authors link) is a June 2013 Cochrane review on Screening for breast cancer with mammography.
In that 2013 two-author Cochrane review on mammography, all but one of eight studies the authors evaluated started in 1982 or earlier, as follows: 1963 (the New York trial); 1976 and 1978 (the Malmö trial, parts I and II); 1977 (the “Two-County trial,” Kopparberg and Östergötland – aka the Swedish trial); 1978 (the Edinburgh trial); 1980 (the Canadian trial); 1981 (the Stockholm trial); 1982 (the Göteborg trial); and 1991 (the UK age trial). In their final analysis, the two Cochrane authors discounted the Edinburgh trial; only seven papers were analyzed, and six of those seven began between 1963 and 1982. The most recent and only “modern” trial began in 1991.
TheNNT.com website, a potentially helpful resource to which some health journalists refer, offers this page on mammography which lists an even older Cochrane study as its source (tab: “Details for this Review”).
A recent paper found that after 2009, when the USPSTF last changed its recommendations for breast cancer screening, fewer U.S. women with health insurance went for mammograms. The biggest decline, of 10 percent, occurred among women between the ages of 40 and 49.
Kind of scary – that so much rests on old studies of a technology that’s gone through several “generations” of improvement since the 1980s. Many women and their primary care physicians (who may not read these papers so closely), think it wise to follow evidence-based medicine. I do too. But the Cochrane review and its archaic data should have no role in how we consider breast cancer screening in 2015.
