Confusion, confounding and cancer screening

A recent proposed change in breast cancer screening recommendations is causing a furor in the US, even prompting pop-singer (and breast cancer survivor) Olivia Newton John to issue her own recommendations.

Why the fervor? A panel that makes prevention recommendations to the government reviewed the evidence and concluded that low risk women in their 40’s don’t need mammograms, and women 50 and older can screen less often. Advocates for more stringent screening practices are crying foul: This will cost women’s lives, is motivated by government penny pinching, or worse, patriarchal devaluing of women’s lives. The airwaves are echoing with testimonials from breast cancer survivors in their 40’s convinced that early detection saved them. In some cases it may have. In others it may hasten death.

Many assume that cancer is one disease, but there is no one thing that can be looked at and proclaimed “cancer.” It is instead a catch-all term for cells behaving badly. When the behavior can’t be labeled anything else, it may end up in the cancer category. In about 25% of breast cancer cases, the body cures itself: growing tumors are killed off by the body’s own immune response, with no treatment. Of course we don’t know which 25% of cancers will die off, and which might threaten the patient. What we do know is that fully 25% of women who have a diagnosis of breast cancer and receive aggressive treatment including mastectomy, radiation and chemotherapy, were never at risk from the disease itself. The assumption is that the treatment saves lives, and if they survive the odds are about 3 to 1 that that is true. In a small minority of cases women may even get cancer and die from the treatment itself. We do not hear testimonials from these women for obvious reasons.

Screening inflates our sense of being able to cure disease. Here’s an example. Consider cancer X, which kills 90% of patients, for which we have no treatment, and no screening. Lets say 50,000 people a year are diagnosed with X once it produces symptoms, 45,000 of whom die. If we start a screening regimen, we will begin to detect more X. That’s what all screenings do. Suddenly we may have 100,000 cases, but still only 45,000 deaths. The death rate has dropped to 45%. Use new technology to increase early detection by picking up miniscule traces of X, and you may increase the number of cases to 500,000. Now the death rate is 9%. We are still not treating anyone, and still have 45,000 deaths. In addition to improving the survival rate, we are also increasing survival time for those patients who eventually die. It’s called lead-time bias: Finding a growing tumor earlier will always increase the time that you live with it, whether you receive treatment or not.

Health decisions are not nearly as absolute as we’d like to believe, and like it or not we must all live with the uncertainty.