Email updates

Keep up to date with the latest news and content from Breast Cancer Research and BioMed Central.

This article is part of the supplement: British Society of Breast Radiology Annual Scientific Meeting 2012

Open Badges Poster presentation

Repeatability of breast density visual assessment

L Walshaw1*, JC Sergeant2, M Wilson3, S Steed1, N Barr3, U Beetles3, C Boggis3, S Bundred3, S Gadde3, Y Lim3, S Whiteside4, DG Evans3, A Howell3 and SM Astley2

  • * Corresponding author: L Walshaw

Author Affiliations

1 Manchester Medical School, University of Manchester, UK

2 Institute of Population Health, University of Manchester, UK

3 Nightingale Centre and Genesis Prevention Centre, University Hospital of South Manchester, Manchester, UK

4 Department of Medical Statistics, University Hospital of South Manchester, Manchester, UK

For all author emails, please log on.

Breast Cancer Research 2012, 14(Suppl 1):P27  doi:10.1186/bcr3282

The electronic version of this article is the complete one and can be found online at:

Published:9 November 2012

© 2012 Walshaw et al.; licensee BioMed Central Ltd.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Breast density, measured as the proportion of the breast occupied by fibroglandular tissue in a mammogram, is a strong and modifiable risk factor for breast cancer. Area-based estimates made by expert observers are a practical approach, but are subjective. Here we investigate repeatability of visual assessment of percentage breast density.


Seven mammographic film readers re-assessed the density of 100 normal full-field digital mammogram cases for which they had made density estimates at least 1 year previously as part of the Predicting Risk of Cancer at Screening (PROCAS) study. The mammograms for a given reader were selected to show a range in density, by randomly sampling 10 cases from each decile of density assessed by that reader. They were reviewed in similar reading conditions on both occasions using a visual analogue scale to record the assessments.


For the majority of readers the difference in mean density between the two sets of readings was less than 6%, but the largest discrepancy between means was 14.7%. Bland-Altman plots were generated for each reader and showed considerable variation between readings on the two occasions. At best, the limits of agreement were -12.46% to +17.02%, and at worst they were -14.50% to +40.98%. The largest difference between first and second readings for each reader ranged from 26 to 65%.


Although density estimates made by a subset of these readers have been strongly related to cancer risk, the variability in reproducibility calls into question the usefulness of subjective assessment without prior evaluation of reader performance.