In the past few years, we have steadily accumulated research that questions the benefit of mammograms for women. Last week, a new study that examined the breast cancer screening histories of 16 million US women over a 10-year period added significantly to the research, concluding that there is “widespread overdiagnosis” of breast cancer due to mammograms.
Before reviewing the study’s findings we should consider the term “overdiagnosis”, which many commentators rightly criticize.
Overdiagnosis occurs when we wrongly identify someone who doesn’t have a given disease as having that disease. However, that’s not how mammograms are creating controversy: the small tumors that mammograms find are called cancer because pathologists who assess biopsy samples with microscopes and chemical tests identify them to be cancers. No different from biopsies from tumors that we discover without mammograms.
A better term for what mammograms cause is “overtreatment”. That’s because the new study, like earlier ones, finds that the tumors that mammograms find might not need immediate attention.
In fact, a 2008 study from Norway showed that some breast cancers discovered by mammograms seem to disappear without any intervention. The researchers reached that conclusion because after dividing women into two groups followed for six years, they found 22% more biopsy-confirmed tumors in the group checked every two years than in the group checked only once at the end of the six years. It’s difficult to find an explanation for that result other than that some tumors that mammograms identify in two-year interval testing will vanish on their own.
Of course, knowing that some tumors disappear isn’t very helpful to a woman who has a new breast tumor found by mammogram if we don’t have a way to tell apart those cancers that grow from those that shrink away. That’s obviously an important target for further research.
But what if we don’t actually need to tell threatening from non-threatening cancers in order for women to make decisions about whether or not to have mammograms?
There are two arguments that support normal-risk women (those without a family history of breast cancer) who want to say no to mammograms.
The first is based on whether or not mammograms have contributed to the welcome and continuing decline in the proportion of women who die from breast cancer. Because the trend began in the 1980s, the same time that we introduced mammography screening, the reasonable assumption was that mammograms caused the reduction in death rate.
However, a landmark 2011 study on the relationship between mammography and breast cancer mortality cleverly refutes that assumption. Using three pairs of neighbouring countries with similar genetics and lifestyle but with large differences in the year of launching mammography screening, the researchers showed that the drop in mortality was the same in both countries in each pair.
That means that the steady reduction in deaths was already under way when the countries with later starts to mammogram programs first began to introduce them. The study thus confirms other research that suggests that improvements in treatment, not earlier detection with mammograms, are responsible for better breast cancer outcomes.
The second argument against the benefit of mammograms concerns the way that research studies have gathered data and reported results in studies of mammography. Specifically, the focus is overwhelmingly on the rate of breast cancer deaths, which epidemiologists refer to as “disease-specific mortality”.
While that measure is important it’s not what we really want to know. Instead we want to know how many women with breast cancer die from any cause, whether it’s from the breast cancer or from some other event or disease.
Why? Because sometimes the processes of diagnosing and treating a disease damage a person’s health without us recognizing the link between the disease and the cause of death. For example, people who have chemotherapy might be more likely to die of strokes than people who don’t, or have higher rates of second cancers. But if we don’t include deaths from any cause – what’s called “all-cause mortality” – in the research results, we can’t properly calculate the total benefit and risk of a test that leads to diagnosing a disease.
Unfortunately, almost no studies of mammograms measure all-cause mortality. Those that do don’t find that mammograms make a difference. That isn’t a unique failing of mammograms: this study of screening tests for diseases where death is a common outcome concludes that reductions in all-cause mortality due to the tests are “very rare or non-existent”.
Returning to the just-released study, it doesn’t report all-cause mortality. However, it found no significant difference in breast cancer death rates between women who had mammograms and those who didn’t.
More important is a subtle finding that requires careful thought to understand: the study showed that mammograms detected more small cancers without finding fewer large tumors in women who underwent mammography over a 10-year period. If mammograms helped to find tumors earlier, we would expect the greater number of small tumors would be offset by a lesser number of large tumors because mammograms catch the small tumors before they grow to a bigger size.
Instead, the new study shows that mammograms find extra cancers without lowering the number of larger tumors or the chance of a woman dying from breast cancer. Like the results of the Norwegian study from 2008, that finding suggests that the extra cancers found by mammograms aren’t dangerous. And that challenges our conventional thinking that sooner is better when it comes to finding cancer.
It also means that women at average risk for breast cancer don’t lower their risk of dying by participating in current screening. Which is why it’s reasonable for a woman to say “no thanks” to mammograms.