Screening test accuracy is a function of both sensitivity and specificity. Sensitivity is the ability to detect true positives (where a tumour exists and is identified on mammogram) and true negatives (where no tumour exists and no tumour is identified on mammogram). Specificity is the ability to exclude false positives (where no tumour exists but one is indicated on mammogram) and false negatives (where a tumour is present but missed on mammogram).
The sensitivity (true positive rate) of screening mammography was estimated in a meta-analysis to be between 83% and 95%, with a range of 39% to more than 90% in an International Agency for Research on Cancer (IARC) review. The specificity of screening mammography exceeds 80%, with many studies showing specificity above 95%.[1,2] However, these estimates are based on data published before 2002, therefore their relevance to current mammography methods is unclear.
Mammography is around 10% less sensitive in women under 50 than in women aged 50-59, the meta-analysis showed. This is because younger women have more dense breast tissue, making their mammograms harder to read. Women in their 40s screened within the UK Age Trial had a 5% risk of a false-positive result at their first screen.
Around 1 in 13 (7.7%) women in the NHSBSP in England were recalled for further tests after their first mammogram, data for financial year (FY) 2010/11 show. Around 1 in 10 (9.5%) of those women were diagnosed with cancer. Recall rates are higher for first screens than subsequent ones, typically because unusual findings on first screens (which may be normal for the woman) are investigated in full. This means the positive predictive value(PPV); the proportion of women with positive screening results who are diagnosed with cancer) is lower at first screens. At subsequent screens the previous investigations can be referred back to without needing to recall the woman, so only new unusual findings are investigated in full. This means the recall rate is lower at subsequent screens, but the PPV is higher. In FY2010/11 in England, around 1 in 34 (2.9%) women were recalled after a subsequent screen, and of those almost 1 in 4 (25.5%) were diagnosed with cancer.
Number of views taken at mammography
Two-view mammography – in which one image is taken from above and one from the side for each breast – is used in all the national breast screening programmes across the UK.
When the NHSBSP began in England, single view mammogram (from the side) was used. In 1995, two-view mammography was introduced for all first (prevalent) screens, after an RCT showed a 24% increase in cancer detection rates and a 15% decrease in recall rates for two-view versus single-view at this first screen. In 2003, two-view mammography was extended to all subsequent (incident) screens in England, after research showed that two-view mammograms were particularly effective in detecting small invasive cancers. In Northern Ireland, two-view mammograms have been used at all screens since the programme started. Two-view mammography at every screen was implemented in Wales in 2001, and in Scotland between 2008-2009.[10,11]
Digital mammography is being introduced into the UK breast screening programmes, with equipment in at least 85% of screening units. Cancer detection rates are slightly higher with digital mammography than conventional film mammography, with no detectable impact on recall rates or PPV, a meta-analysis showed.
Double-reading and computer-aided diagnosis
In the majority of UK breast screening units, mammograms are read by two people (double-reading): either two radiologists, or a radiologist and an experienced radiographer.[14,15] In some units, mammograms are read by two experienced radiographers and a radiologist is consulted if the radiographers disagree. Double-reading increases cancer detection rates by 10% in comparison with reading by one person, a meta-analysis showed. Computer-aided diagnosis (in which a single human reader uses a computer programme designed to bring his/her attention to suspicious features) is the subject of ongoing research, but the evidence is presently inconclusive.
HRT and mammography
Users of hormone replacement therapy (HRT) are more likely than HRT non-users to receive false negative or false positive screening results, a systematic review and large cohort studies have found.[18-20] This may be because HRT increases breast density, making mammograms harder to read.
Sensitivity was 83% in current HRT users versus 92% in never-users; and specificity was 97% in current users versus 98% in never users, the UK Million Women study found. Current users were 46% more likely than never users to receive a false positive recall, and past users were 12% more likely, a meta-analysis showed.