Does access to prior mammograms improve the performance of radiographers in interpreting screening mammograms?

The impact of previous screening mammograms on radiographers' performance in mammography interpretation is unknown. This study assesses the impact that previous screening mammograms has on radiographers’ interpretation of mammograms. Thirteen Australian radiographers working for the national breast screening service independently interpreted a mammography test-set containing mammograms of 28 women based on the Royal Australian and New Zealand College of Radiologists’ classification. Twelve radiographers completed the “No prior test-set” (no previous mammograms available) while one radiographer completed the “Prior test-set” (most current screening mammograms with access to previous mammograms) in the first reading session. In the second reading session, 12 radiographers completed the “Prior test-set” and one radiographer completed the “No prior test-set”. Their performance with and without previous mammograms were calculated and compared. The availability of prior mammograms significantly improved specificity [81(range:58–95) vs. 60(range:37–79); p = 0.002], ROC [91(range:80–99) vs. 82 (range:57–91); p = 0.003], and JAFROC 87(range:73–99) vs. 79 (range:52–91); p = 0.01]. Prior mammograms also significantly reduced false positives (p = 0.002). No differences were observed between readings with and without previous mammograms in terms of sensitivity (p = 0.70) and lesion sensitivity (p = 0.82). Years qualified as a radiographer did not modify the influence of previous mammograms on specificity, ROC, and false positives. Years specialised as breast radiographer slightly modified the influence of previous mammograms in radiographers with ≥25 years of experience but not those with <25 years of experience as breast radiographers. The availability of previous screening mammograms improves radiographers’ ability to discriminate between normal and abnormal mammograms and reduce the false positive rate without affecting the detection of breast cancer. The findings highlight the need for practices to store screening mammograms and for radiographers to actively refer to previous screening mammograms when interpreting mammograms from the current screening round. It also highlights the need for policies to establish a national accessible mammographic database platform for integrated clinics and to account for population mobility across states.