Sensitivity in cancer screening

Abstract

Objective: We propose three concepts of sensitivity in cancer screening and apply to data on prostate cancer. Conceptual entities: Sensitivity is the indicator on the ability of screening to find cancer in the detectable preclinical phase (DPCP). The ability is usually specified as to the screening test. We call this entity the test sensitivity. Test positivity with histological confirmation refers to the full diagnostic process and we call the corresponding entity as episode sensitivity. Ultimately, a screening programme identifies a proportion of cancers in the DPCP in the total target population, that we call programme sensitivity. We derive the formulae for these three sensitivities consistent with the incidence method. Example: Our example on estimation of the three sensitivities is from a randomized screening trial for prostate cancer in Finland. The estimates by incidence method were substantially different, 85% for test sensitivity, 48% for episode sensitivity and 36% for programme sensitivity. Conclusion: More than one concept of sensitivity with standard method of estimation is needed to describe the ability of screening to identify the disease in the DPCP.