Introduction: Policy decisions about prostate cancer screening require data on the natural history of histological cancers and the resulting impact of screening. However, the gold standard procedure required to identify true positive histological cancer is a full autopsy of the gland which is not possible in screening studies, leading to verification bias. We aim to estimate the sensitivity of a prostate cancer screening round (PSA result to diagnosis) relative to histological cancer.
Methods: We developed a framework combining data on UK screened and non-screened prostate cancer populations originating from a single round of population-based PSA testing among UK men aged 50-69 years, prostate cancer incidence data, and needle biopsy data from the published literature.
Results: Sensitivity of a screening round was highest at age 65-69 years at 33% (95% CI: 30%-37%) and 24% (95% CI: 21%-28%) for PSA cut-off levels of 3 ng/ml and 4 ng/ml, respectively. Sensitivity was lowest at age 50-54 at 15% (95% CI: 12%-17%) and 9% (95% CI: 8%-11%) for PSA cut-off levels of 3 ng/ml and 4 ng/ml, respectively. In contrast, the clinical detection rate in the absence of mass screening, relative to histological cancer, varied between 0.2%-0.7% at age 50-54 and 1.2%-2.7% at age 65-69 from 1995 to 2012.
Conclusions: The framework enabled the sensitivity of a prostate cancer screening round relative to histological cancer diagnosis to be estimated and provides a basis to determine the impact and cost-effectiveness of prostate cancer screening. The approach could be adapted to inform the sensitivity of other biomarkers, cancers and screening programmes.
Keywords: Evidence synthesis; Overdetection; Overdiagnosis; Prostate cancer; Screening; Sensitivity.
Copyright © 2017. Published by Elsevier Ltd.