The overdiagnosis nightmare: a time for caution.

Overdiagnosis (and overtreatment) of cancers not bound to become symptomatic during lifetime is an unavoidable drawback of mammography screening. The magnitude of overdiagnosis has been estimated to be in the range of 5-10%, and thus acceptable in view of screening benefits as to reduced mortality. In a recent research article in BMC Women's Health, Jørgensen, Zahl and Gøtzsche suggest that overdiagnosis may be as high as 33%, based on their analysis of breast cancer incidence in screened and non-screened areas in Denmark. Here we consider how reliable such analyses can be, why it might have been useful to adjust comparisons between screened and non-screened areas for early detection lead time, and what further evidence might be needed to build on or confirm these results.