Overdiagnosing overdiagnosis

Abstract

Solid evidence shows that screening mammography contributes to a substantial reduction in mortality from breast cancer, but it is far from a perfect test. Estimates of overdetection of breast cancer, based on extrapolation of SEER data, may suffer from considerable uncertainty because of major assumptions required in such calculations and the difficulty in correcting for confounding factors such as lead time and demographic variations. Some of the charged language used and the extreme positions taken in the discussion of screening distract from coherent communication of the complex issues surrounding this topic. More balanced discussion would better serve the public.

Figure 1.

Surveillance Epidemiology and End Results 9 incidence and U.S. mortality, 1975–2010, for females of all races. Rates are age adjusted.

Figure 2.

The size of of box (left) represents breast cancer incidence from 1976 to 1978 for women older than age 40. It is assumed that very little screening is occurring. The unshaded region (upper right) represents the expected incidence (by extrapolation of data for younger women using the methods of Bleyer and Welch) that would occur in the absence of screening. Excess of observed incidence compared with expected incidence (shaded region) is attributed to overdetection. Consideration of well-screened and unscreened or poorly screened subcomponents in the Surveillance Epidemiology and End Results data (lower right). Abbreviations: E, early stages; L, later stages.