Quality of mammography screening in the Milan programme: Evidence of improved sensitivity based on interval cancer proportional incidence and radiological review

Abstract

Background Interval cancer (IC) incidence and review-based initial evaluation of Milan service screening (ASLMI1) suggested suboptimal performance. We report results in a subsequent screening round to further determine screening quality. Methods IC was identified in subjects with a negative screening episode in 2005 ( N = 33,258) by linkage with 2005–2007 Hospital Discharge Records. IC proportional incidence-based sensitivity was estimated. Radiological review, with case-mix, was conducted in a blinded manner. Linkage and review modalities were maintained as for initial evaluation. Results 30 identified IC accounted for 6.99% (year 1), 27.84% (year 2), and overall 17.44% (2-year screening interval) proportional incidence. Estimated sensitivity was 93.01% (year 1), 72.16% (year 2) and 82.55% (2-year interval). Review classified 5 (16.6%) as screening error, 2 (6.7%) as minimal signs, and 23 (76.7%) as occult. Conclusion Programme sensitivity is now within recommended European standards. Performance indicators improved relative to initial evaluation. Both increasing experience and formal training of radiologists are likely to have contributed to this improvement