Vitenskapelig artikkel   2018

Lilleborge, Marie; Hofvind, Solveig; Sebuødegård, Sofie; Hauge, Ragnar

Publication details

Journal:

Statistics in Medicine, vol. 37, p. 1531–1549, 2018

Publisher:

John Wiley & Sons

Issue:

9

International Standard Numbers:

Printed: 0277-6715
Electronic: 1097-0258

Links:

DOI: doi.org/10.1002/sim.7601

This study proposes a method to optimize the performance of BreastScreen Norway through a stratified recommendation of tests including independent double or single reading of the screening mammograms and additional imaging with or without core needle biopsy. This is carefully evaluated by a value of information analysis. An estimated graphical probabilistic model describing the relationship between a set of risk factors and the corresponding risk of breast cancer is used for this analysis, together with a Bayesian network modeling screening test results conditional on the true (but unknown) breast cancer status of a woman.

This study contributes towards evaluating a possibility of improving the efficiency of the screening program, where all women aged 50 to 69 are invited every second year, regardless of individual risk factors. Our stratified recommendation of tests is dependent on the probability that an asymptomatic woman has developed breast cancer at the time she is invited to a screening.