Monitoring and evaluating the UK National Health Service Breast Screening Programme: evaluating the variation in radiological performance between individual programmes using PPV-referral diagrams.

Abstract

A high quality breast cancer screening programme can be defined as one offering both a high cancer detection rate and a low referral rate of women for further investigation. Such a programme will have as few women as possible undergoing further investigations who do not have a final diagnosis of breast cancer--that is, a high positive predictive value of referral for further investigation. This paper introduces a graphical technique to illustrate individual programme performance. The graph plots positive predictive value of referral against referral rate, with the cancer detection rate expressed as "isobars" on the graph. Confidence limits can be expressed as "boxes" on the diagram. The graph not only illustrates programme performance but also enables suggestions to be made to improve performance. The definition of high quality screening is seen to have a subjective element as well as an objective element, as radiologists have to balance screening sensitivity with specificity. The technique is illustrated using data from the individual screening programmes in the UK National Health Service Breast Screening Programme for the screening year 1 April 1998 to 31 March 1999. The methodology could also be applied to other national screening programmes.