Overdiagnosis occurs when a tumor is detected by screening but, in the absence of screening, that tumor would never have become symptomatic within the lifetime of the patient. Thus, an overdiagnosed tumor is a true extra diagnosis due solely to the existence of the screening test. Patients who are overdiagnosed cannot, by definition, be helped by the diagnosis, but they can be harmed, particularly if they are treated. Therefore, knowledge of the likelihood that a screen-detected cancer has been overdiagnosed is critical for making treatment decisions and developing screening policy.
The problem of overdiagnosis has been long recognized in the case of prostate cancer and is currently an area of extreme interest in breast cancer. Published estimates of the frequency of overdiagnosis in breast and prostate cancer screening vary greatly. This presentation will investigate why different studies yield such different results. Dr. Etzioni will explain how overdiagnosis arises, catalog the different ways it may be measured in population studies, and discuss different approaches that are used to estimate overdiagnosis.
Many studies use excess incidence under screening relative to incidence without screening as a proxy for overdiagnosis. Others use statistical models to make inferences about lead time or disease natural history and then derive the corresponding fraction of cases that are overdiagnosed. Each approach has its limitations and challenges, but one thing is clear: estimation approach is a major factor behind the variation in overdiagnosis estimates in the literature. The talk will conclude with a list of key questions that consumers of overdiagnosis studies should ask to determine the validity (or lack thereof) of study results.