Why mammogram overdiagnosis estimates can be so far apart

One of the hardest questions in cancer screening is not whether mammograms find cancers early—they do—but how often they find a cancer that would never have caused symptoms during a person’s lifetime. That is called overdiagnosis, and published estimates have ranged from very low to nearly 50%.

A new meta-analysis revisited randomized mammography trials with close attention to timing. The researchers accounted for how long people were followed and for screening that happened after trials ended—details that can make a temporary early increase in detected cancers look like lasting overdiagnosis. Using that framework, they concluded the patterns were compatible with overdiagnosis below 5%, rather than estimates approaching 50%.

It is not the final word on screening. This was a reanalysis of older trials, not a new trial, and other researchers and guidelines may reasonably interpret the trade-offs differently. But it usefully shows why a headline percentage can conceal important choices about how a study counts, and when it stops counting.