Design: Review of national external quality assessment cases, 2015–2025
Population: 263 breast cases; up to 22 expert coordinators and about 678 participants per case
Primary outcome: Diagnostic concordance and features of low-concordance cases
Effect: 15 cases (5.7%) below 80% expert agreement; mean agreement 57% in those cases

The UK breast screening external quality assessment scheme reviewed 263 cases circulated between 2015 and 2025, each read by up to 22 expert coordinators and an average of 678 participating pathologists.

Overall agreement was high: two-thirds of participants agreed with the consensus on every case. But 15 cases (5.7%) fell below 80% agreement even among the experts, with mean agreement of 57%. These were mainly atypical epithelial proliferations and papillary lesions at the boundaries between categories: lobular neoplasia subtypes, apocrine atypia, flat epithelial atypia, and in situ versus invasive papillary carcinoma.

The finding is less about individual error than about where the classification itself is fragile. For a reporting pathologist, it identifies the lesions where a second opinion and clear wording of uncertainty in the report are most worthwhile. It was published in May 2026.

Seek a second opinion on atypical epithelial and papillary breast lesions near category boundaries.
State uncertainty explicitly in the report when a lesion sits between categories.
Use multidisciplinary discussion to align management with a borderline diagnosis.
Take part in EQA for breast pathology, which is where these weak points show.

Why it matters

It shows where borderline breast diagnoses are least reproducible, even among experts.

The statistics, in plain English

Agreement of 57% means that on average just over half of experts chose the same diagnosis for these cases. The 5.7% figure is the share of circulated cases that fell below the 80% agreement threshold, not an error rate in routine practice.