Does the United States Over Diagnose Cancer?

Ezra Klein challenges the notion that patients in the United States get better cancer treatment than patients in other developed countries. Klein was writing in response to the Commonwealth Fund’s comparison of health systems in eleven developed countries. As I noted previously, one problem with this survey is that there is no apparent relationship between ranking on the survey and health outcomes. Although the United States does poorly in the survey, it does well in health outcomes, especially cancer outcomes.

Or maybe not, according to Klein:

Most of the studies that highlight America’s skill in treating cancer do so by measuring survival rates—that is to say, they measure how many people survive for a certain number of years after the cancer is diagnosed. So if a certain cancer kills 50 percent of people within five years, then the five-year survival rate is 50 percent.

The problem here is simple: survival rates don’t necessarily measure when people die. They also measure when they’re diagnosed—and sometimes, that’s all they measure.

“Let’s say there’s a new cancer of the thumb killing people,” writes Aaron Carroll, a professor of pediatrics and assistant dean at Indiana University’s School of Medicine. “From the time the first cancer cell appears, you have nine years to live, with chemo. From the time you can feel a lump, you have four years to live, with chemo. Let’s say we have no way to detect the disease until you feel a lump. The five year survival rate for this cancer is about 0, because within five years of detection, everyone dies, even on therapy.”

Carroll goes on to imagine a remarkable machine: “a new scanner that can detect thumb cancer when only one cell is there.” Congress immediately orders that every American be scanned for thumb cancer. “We made no improvements to the treatment,” he writes. “Everyone is still dying four years after they feel the lump. But since we are making the diagnosis five years earlier, our five year survival rate is now approaching 100%!” That’s how survival rates can mislead.

Klein goes on to report that actual death rates (in the U.S. population) have not really changed for many cancers, despite much greater detection. Further, this epidemic of over diagnosis can cause harm, because people will undergo surgery who don’t need it. (This argument is made thoroughly in Gil Welch’s Overdiagnosed.)

Although this argument is not trivial, I find it hard to accept. First, almost all stories of harmful treatment concern breast cancer or prostate cancer. There are lots of other cancers, but their victims do not have the benefit of early screening. For example, lung cancer is often diagnosed after a person has a troubling cough. Only recently did the American Society of Clinical Oncologists suggest annual screening with low-dose computed tomography (LDCT) for smokers and former smokers at high risk for developing lung cancer. Surely, opponents of “over diagnosis” do not want us to “uninvent” mammography, and have women wait until lumps appear before they are diagnosed.

Also, early detection has a social benefit, because it leads to greater learning about the nature of a disease, which leads to better therapies. For example, I cannot imagine how we would have any genomic medicine if organized medicine had decided that it did not want to push back the time of diagnosis before symptoms appear.

Further, it is not clear what “over diagnosis” has to do with “universal health care”. Klein cites a study by Tomason and colleagues, which compares U.S. survival rates with international survival rates. However, that study did not describe which health insurance the U.S. patients had. There are very few comparisons of cancer outcomes that compare U.S. uninsured and insured patients. A 2008 study showed that privately insured breast-cancer patients had earlier diagnosis and better outcomes than Medicaid and uninsured patients. So, score one for private health insurance and early preventive screening.

Finally, as Klein notes, the alternative to measuring survival is measuring death rates. However, that ignores all medical interventions. Knowing how many people die of a certain cancer in a population tells us nothing about the state of the healthcare system. For such a purpose, survivability appears to be the superior of two limited measurements.