Original PaperOptimal screening mammography reading volumes; evidence from real life in the East Midlands region of the NHS Breast Screening Programme
Introduction
The NHS Breast Screening Programme (NHSBSP) aims to maximize the detection of breast cancer whilst minimizing anxiety to women by avoiding unnecessary recalls. Double-reading, particularly with arbitration, has been shown to increase cancer detection without increasing recall rates 1, 2 and most mammograms in the NHSBSP are double-read.
Performance measures have been in place since the outset of the programme with some additions and modifications over time. Key performance indicators for radiologists and other film readers are cancer-detection rates, small cancer-detection rates, recall rates, positive predictive value of recall (PPV), and interval cancer rates. Further information on missed cancers can be obtained by looking at cancers detected by only one reader and then recalled following arbitration. This information is more readily available than data on interval cancers, with which there is naturally a delay.
Although performance data from individual units are regularly scrutinized and assessed, it has only been since the introduction of the new National Breast Screening Service (NBSS) computer system in 2003 that individual film-reader performance can be accurately assessed. Unit-based performance data can mask variations in individual performance and it can be helpful for film readers to assess where their performance lies in relation to peers.
In the UK, film readers involved in breast cancer screening are encouraged to take part in a voluntary self-assessment programme (PERFORMS — Personal Performance in Mammographic Screening), which involves reporting on a test set of cases. However, results from this programme cannot be a substitute for real-life performance. It is a national requirement that all film readers should read at least 5000 cases per year.3 Although previous studies have assessed the relationship between reader volume and performance outcomes these have largely been based on laboratory based “PERFORMS” type work rather than real life.4
The present study investigated the relationship between real-life reader performance and both years of film reading experience and volume of mammograms read over a 3 year period in a multicentre study.
Section snippets
Materials and methods
Consent was obtained from all film readers working in the East Midlands Breast Screening Programme to analyse individual-based data anonymously for the purposes of this study. No film reader withheld consent. Individual performance data was retrieved from the NHSBSP computer system using a report-generating tool (crystal) for all film readers with continuous service between April 2005 and March 2008. Data from one unit were excluded as they were not comparable with the other units because of
Results
During the 3 year period, the number of sets of mammograms read by individual film readers ranged from 4423 to 38,632 (median 16,195). This was equivalent to an average of 1474 to 12,877 (median 5398) cases per year. Table 1 lists the characteristics of the film readers included in the study. The study group included 20 consultant radiologists, one breast physician and 16 advanced practitioners (radiographer film readers). Radiologists had both a higher median reading volume, and more years of
Discussion
Accurate reading of mammograms is crucial to the quality of a breast-screening programme affecting both the cancer-detection rate and false-positive recall rate. The present study showed that higher-volume readers had slightly lower cancer-detection rates than other readers but at much reduced recall rates. However, their small cancer-detection rates were comparable.
In the East Midlands programme all film readers review screening assessment cases and interval cancers. This is in addition to the
References (13)
- et al.
The pathological and radiological features of screen-detected breast cancers diagnosed following arbitration of discordant double reading opinions
Clin Radiol
(2005) - et al.
Is radiologists’ volume of mammography reading related to accuracy? A critical review of the literature
Clin Radiol
(2005) - et al.
Cancer detection and mammogram volume of radiologists in a population-based screening programme
Breast
(2006) - et al.
Flobbe k. Performance of radiographers in mammogram interpretation: a systematic review
Breast
(2008) - et al.
A comparison of cancer detection rates achieved by breast cancer screening programmes by number of readers, for one and two view mammography: results from the UK National Health Service Breast Screening Programme
J Med Screen
(1998) Quality assurance guidelines for breast cancer screening radiology
(January 2005)