Influence of number of views and mammographic film density on the detection of invasive cancers: results from the NHS Breast Screening Programme.

Abstract

The National Health Service Breast Screening Programme (NHSBSP) has recommended the adoption of two view mammography at the prevalent screen, and the use of a target film density in the range 1.4-1.8. The aim of this study was to review the impact of number of views and optical density on the detection of invasive cancers. The last four annual returns for screening centres in the NHSBSP have been analysed retrospectively for 2827342 women aged 50-64 years attending their first (prevalent) screening examination. The detection of invasive cancers was assessed in relation to the number of views and film density using the age adjusted, Standardized Detection Ratio measure of screening performance. Typical film densities were reported for each screening year by local physicists, and the average value for all mammography sets at each programme calculated, and found to vary from 0.85 to 1.85. The mean film density across the NHSBSP rose progressively from 1.30 (SD = 0.21) in 1991/2 to 1.57 (SD = 0.12) in 1994/5. Programmes using single view mammography (MLO) and an optical density less than 1.4 detected 76% (95% CI 74-79%) of the expected invasive cancers. Programmes using two view mammography (MLO and CC) and an optical density equal to or greater than 1.4 detected 95% (95% CI 92-98%) of the expected invasive cancers. In 1994/95 when more programmes used the recommended screening modes, the NHSBSP detected 96% (95% CI 92-101%) of the expected invasive cancers at prevalent screening. The detection of invasive cancers was highest where programmes used two views with a film density in the range 1.4-1.8. The results provide evidence of the benefit of the recommended protocol for prevalent screening and indicate that from 1995/96 when all programmes will be using the recommended protocol, it is likely that the detection rates and interval cancer rates from prevalent screens in the NHSBSP will be close to the figures in the Swedish-Two County Trial.