[Assessment of diagnostic accuracy of mammography carried out for secondary prevention. Results of a test with a sample caseload conducted by 75 Italian radiologists]

Abstract

Purpose: To report the results of a test performed by 75 volunteer radiologists involved in screening program in Italy.

Material and methods: The test includes 12 screening detected cancers and 6 cancers (4 screening detected and 2 interval lesions) with minimal signs at review of a previous negative mammogram. Data on previous experience in mammography (years of activity, years of screening activity, total number of mammograms read and number of mammograms read per year) were also collected.

Results: Standards for passing the test were at least 83.3% for sensitivity for screening detected cancers and at most 30% for recall rate among negative cases. Mean sensitivity was 83.9% (median 83.3%, range 50-100%) and mean recall rate was 16.3 (median 14.4%, range 3.8-42.4%). In all, 44 radiologists (58.7%) passed the test. On average 1.4 of 6 cancers with minimal signs were correctly identified. According to readers' experience (at least 5 years in mammographic activity, at least 3 years in screening mammography, at least 10,000 total mammograms read, at least 5,000 per year) radiologists were classified as expert (13) or not (62): no difference in test performance was observed between the two groups. The only significant association occurred for years of mammographic activity and sensitivity.

Discussion and conclusions: Test results suggest the need for proper training of radiologists prior to involvement in a screening program. Seeding proficiency tests with cancers with minimal signs was of no benefit as far as evaluation is concerned.