Two models for radiological reviewing of interval cancers

Abstract

Objectives To compare two different review methods of examining how many of our interval cancers could be regarded as missed cases (overlooked and misinterpreted owing to observer's error). Setting A mass screening programme in Stockholm 1989–91, performed at five independent screening units. 107 846 women attended for screening (70.6% of those invited), and 207 women with interval breast cancers were identified. Interval cancers from two of the units, 104 cases, are reviewed in this study. Methods Screening examinations preceding the interval cancer diagnoses were reviewed both mixed with other screening images in a ratio 1:8 and non-mixed. Both internal reviewers (from the two units responsible for the screening mammograms) and external reviewers (from the other units) took part in the study. Results The proportion regarded as missed cases varied between 7% and 34%, depending on what review method was used, and on the number of reviewers included to identify a case as missed. Mixed reviewing reduced the number identified as missed cases by 50% compared with non-mixed reviewing. Whether the reviewer was internal or external made no difference to the results. Conclusions Comparing the rate of missed cases from different studies may be misleading unless the same review method is used. No difference in detection rate could be shown whether the radiologist reviewed images from his/her own screening unit or not. Most of our interval cancers were not regarded as missed cases by either of the two methods.