Objectives: To estimate overdiagnosis (detection of latent carcinomas) as a consequence of screening for prostate cancer. Design: Based on actual screen (first or repeat) detected and interval prostate cancer rates observed in the Florence screening pilot study, a scenario was simulated where males aged 60 years (or 65) had six biennal screens and were followed up for four years. Overdiagnosis was determined as the proportional excess of cancers detected by screening with respect to that expected in its absence. Setting: City of Florence, Italy, from 1992 through 1995. Population: 2,740 resident males, aged 60 to 74 years. Results: Overdiagnosis was estimated to be 51% (95% confidence limits: 44%-55%) or 93% (85%-101%) for age 60 or 65 at entry. Comparison with other screening experiences obtaining higher detection rates suggests that a more aggressive screening approach could be associated with overdiagnosis estimates as big as 200%-250%. Conclusions: Screening for prostate cancer is associated with a relevant risk of overdioagnosis. As latent carcinomas can not be presently identified, this would lead to overtreatment in most overdiagnosed cases. The negative consequences of overdiagnosis (knowledge of having a cancer) and of overtreatment (impotence, incontinence, perioperatory death) may be extremely serious. In absence of any scientific evidence of screening benefits (if any) screening should not be recommended as a current practice, but should be limited to prospective controlled studies designed to assess its cost-effectiveness.
