Prescriptions

My Own Private Screening

What watching ESPN could teach us about mammograms.

A woman receives a mammogram

It’s not often that a football game can teach us something useful about mammography. But look what happened on Sunday after New England Patriots coach Bill Belichick decided to go for a short fourth-down conversion from his own 28 yard line, with a late six-point lead. The Colts stopped the Pats cold to take possession, and star quarterback Peyton Manning quickly fired off a touchdown pass to win the game for Indianapolis.

Like a woman considering whether to have a mammogram, Belichick—widely considered one of the wiliest and most data-driven coaches in the game—had to process complex probabilities to make a decisive, if controversial, call. What’s amazing is how mathematically sophisticated fans and sports commentators are in their analysis of Belichick’s decision not to punt. Sports Illustrated labeled Belichick’s call “I’m-smarter-than-they-are hubris,” but only after it reviewed the preceding plays in numerical detail. On the other side, using a complex mathematic model, two fans gave Belichick a “thumbs up” for a “courageous and correct call”: They calculated the Patriots’ chances of winning at roughly 77 percent by going for it and 76 percent if they had punted instead. The New York Times football blog and the New Republic also dissected probability equations on the matter.

Journalists and physicians explaining the U.S. Public Health Service’s widely publicized  change in mammogram recommendations—it now suggests most women start receiving them every other year at 50, instead of 40—might take a page from the sports commentators’ playbook. If the sports media have no problem filling newspapers and the airwaves with complex statistics—and often discussing them clearly—why do the health media treat the same consumers like innumerate dolts, especially when women’s lives are at stake? Hopefully it’s not because they think only testosterone-drenched sports fans can handle the math.

Even though they use the same data, the American Cancer Society, U.S. Public Health Service, and National Cancer Institute—not to mention Canadian and British health authorities—now fail to agree on when women should start getting mammograms. Some say at 40, others at 50. That’s why it’s critical to separate each organization’s opinions from the facts. Encouragingly, there’s really no disagreement about the latter.

Here’s the bottom line about mammography: Getting screened or not screened for breast cancer is a gamble. There is no right answer, but there is helpful statistical guidance. There are two broad ways to handle mammography for women under 50 years old: to treat them as homogenous and statistically naive (the cookie-cutter approach, usually favored by policymakers) or to assume greater patient savvy (the personalized approach, increasingly favored by clinicians and patients).

Let’s first review the “cookie cutter” approach the way a sports analyst might. For the average woman turning 40, there is a 1.4 percent—about one in 100—chance of getting invasive breast cancer before the age of 50. In baseball terms, she bats .986 against breast cancer. (For comparison, a woman turning 50 has a 2.4 percent chance of breast cancer over a decade; she bats .976.) Further, most cancers are treated successfully, no matter how they’re found, so survival rates are even more favorable.

How does mammography improve these stats? Researchers generally agree that mammograms save lives, but—this is critical—catching breast cancer early changes the outcome in only 15 percent of cases. So consider the actual numbers: For the average 40-year-old woman, annual mammography for a decade increases one’s overall chance of breast cancer survival from roughly 99.7 percent to 99.8 percent. That is, it increases the final batting average by only 0.001. According to the National Cancer Institute, there’s also a downside. During this time, half of all screened women will have at least one suspicious mammogram, and one-quarter of them will end up getting a biopsy. Mammograms in women from 40 to 50 years old cause a huge number of false positives, resulting in about 100 biopsies for every life saved. Even more worrisome: It’s possible the radiation from those mammograms may end up causing more cancers than they prevent.

But what if you’re the one whose life is saved—doesn’t that outweigh the dozens of fruitless biopsies and thousands of scans done on other women to save you? Now that’s something to talk about. There: You now know pretty much everything the experts do in the debate over whether women under 50 should get mammograms.

Still, there’s another way to discuss mammography, which is potentially more useful. When the people who know the most about the subject—like the folks at the National Cancer Institute, Centers for Disease Control, and American Cancer Society—disagree so strongly about mammograms, the lesson is that people handle risks in idiosyncratic ways. As a result, there is a growing but small movement to indentify precisely a given woman’s risk of breast cancer and help her make the decision that’s right for her alone.

Developed by a statistician at the National Cancer Institute, the so-called “Gail model” allows women to estimate their risk of breast cancer based on family history, age, menstrual history, and other personal factors. (You can calculate your risk of getting breast cancer here. It’s helpfully delivered as a percent over five years and over a lifetime.) This allows higher-risk women to make more personalized decision. For example, a woman under 50 who has a sister or mother who’s had breast cancer has the same overall risk as someone over 50 and thus would likely benefit from routine mammograms. Based on her risk profile, a woman might make a choice that adheres to her comfort level, without being at the mercy of arguing authorities. To better guide women, national organizations could band together and propose mammogram screening predicated on personal risk percentages, instead of only the blunt instrument of age.

Earlier this year, the American Society for Clinical Oncology proposed another personalized strategy. For women with a Gail score predicting a five-year breast cancer risk of more than 1.66 percent, the organization recommends considering the drug tamoxifen, which reduces breast cancer risk by about one-third to one-half—yet the guideline is widely ignored.

A third personalized strategy involves targeted genetic testing for the “BRCA” genes, which are responsible for one-tenth of all invasive breast cancers. A woman carrying the gene has an almost 60 percent chance of developing breast cancer during her life—and yet, no organization has yet developed guidelines on who should get tested. Though data are still sketchy, it is reasonable to assume that carriers would benefit more than the average woman from early mammograms or tamoxifen therapy.

Over time, medicine is bound to become more individualized. Unfortunately, national health organizations have failed to take the lead in such cancer screening, though the tools exist. But if people given the right statistical tools can manage a fantasy sports teams over a season, it stands to reason they can probably take charge of their personal health the same way.