Modelling the cumulative risk for a false-positive under repeated screening events

Journal Article

Screening examinations are widely utilized in detecting the presence of medical disorders, for instance, screening mammograms and clinical breast examinations for detection of breast cancer. Such procedures are invaluable in enabling early treatment but produce the possibilities of false-positive and false-negative diagnoses. Focusing on false-positive results, with increasing number of screening events, it is clear that the risk of a false- positive increases. The objective of this paper is to quantify the cumulative risk associated with repeated screening. We provide a very general framework within which to investigate this risk, both at the population and the individual level. The latter allows incorporation of evolving patient medical history to permit individualized assessment of risk. We model cumulative risk in terms of the number of screening events until first false-positive. We develop models which are essentially familiar actuarial models for life table data adding a Cox regression to enable individual level modelling. Because it offers several advantages, we employ a Bayesian inference framework and apply our modelling to the analysis of 9773 screening mammograms collected from 2227 women at an HMO serving nearly 300 000 adults in and around Boston, MA. Copyright (C) 2000 John Wiley and Sons, Ltd.

Full Text

Duke Authors

Cited Authors

  • Gelfand, AE; Wang, F

Published Date

  • 2000

Published In

  • Statistics in Medicine

Volume / Issue

  • 19 / 14

Start / End Page

  • 1865 - 1879

PubMed ID

  • 10867676

Digital Object Identifier (DOI)

  • 10.1002/1097-0258(20000730)19:14<1865::AID-SIM512>3.0.CO;2-M