From the archives: Don't do stupid sh*t in cancer screening

Originally posted on Common Sense Family Doctor on August 25, 2014.

**

The electronic medical record that my office uses features a clinical protocol button that we are encouraged to click during patient visits to remind us about potentially indicated preventive services, such as obesity and tobacco counseling and cancer screenings. I once tried it out while seeing a 90 year-old with four chronic health problems. The computer suggested breast cancer, colorectal cancer, and cervical cancer screenings - three totally inappropriate tests for the patient.

At the residency program where I precept one afternoon a week, we recently held a "chart rounds" on an elderly patient with advanced dementia: when should you stop cancer screening? The answer boils down to the patient's predicted life expectancy compared to the number of years needed for a patient to benefit from a test. Although forecasting how long someone has left to live is not a precise science, knowing averages is essential to deciding if the inconvenience, expense, and potential adverse effects of screening (and treatment, if an abnormality is discovered) can be justified by the potential benefit. Since advanced dementia is a terminal disease, with more than half of nursing home residents in a National Institutes of Health-sponsored study dying within 18 months, there is virtually zero chance that a patient with this condition would benefit from cancer screening of any type. The same statement applies to a healthy 90 year-old in the U.S., who is expected to live around 4-5 more years.

But as one might expect in our crazy health non-system, cancer screening in patients with limited life expectancies happens all the time. A study in JAMA Internal Medicine found that one-third to one-half of surveyed Americans with a 9-year mortality risk of more than 75 percent reported receiving recent cancer screenings. 55% of men in this group had knowingly been screened for prostate cancer within the previous 2 years - a test that, if it works at all, requires a decade to show a mortality benefit.