PB.16. Breast screening mammograms: recall or not to recall. What is the golden ratio?

Introduction

The accuracy of prospectively categorised screening mammograms prior to assessment was evaluated and the individual recall and cancer detection rates for each category were studied.

Methods

The screening mammograms of women who attended the National Health Service Breast Screening Program over a period of 5 years (April 2008-March 2013) were included in the study. The recalled patients' films were prospectively categorised into one of six groups: patients with normal mammograms requiring clinical recall; probably benign; probably benign/suspicious; suspicious; suspicious/malignant; and malignant. Recalled patients were subsequently assessed with additional mammographic views, ± ultrasound and/or needle biopsy. The recall and cancer detection rates of each category were individually calculated.

Results

A total of 86,405 mammograms (17,881 prevalent and 68,524 incident rounds) were read and 4,125 patients were recalled for assessment (overall recall rate = 4.8%). A total of 721 patients were diagnosed with cancer (0.8%). The prevalent recall rate was 9.1% (recommended programme target <7%) and incident recall rate was 3.6% (target <5%).

Conclusion

Mammograms showing minimal signs of cancer that were categorised as 'probably benign' comprised the largest recall group with a relatively low detection rate (7%). If patient recall was limited to only more suspicious groups (that is, PB/suspicious and above), the overall recall rate would have dropped to 1% from 4.8%; however, 28% of the cancers would have missed. Patient anxiety generated due to false positive recalls and the cost of assessment clinics needs to be carefully balanced against a higher cancer detection rate.