Lehman · Radiology 2017 · multi-center registry-based cohort study · n=792808

National Performance Benchmarks for Modern Screening Digital Mammography: Update from the Breast Cancer Surveillance Consortium.

Cited 679 times in the scientific literature.

Level 3 - non-randomized controlled study

Multi-center observational registry cohort study evaluating screening performance metrics.

PubMed 27918707 · doi:10.1148/radiol.2016161174 · record verified 2026-08-31

What was done

Performance metrics for digital screening mammography were evaluated among 359 radiologists across 95 facilities in six Breast Cancer Surveillance Consortium (BCSC) registries between 2007 and 2013. The study included 1,682,504 digital screening mammograms in 792,808 women. Performance measures were calculated using American College of Radiology BI-RADS 5th edition criteria and compared with published benchmarks and expert recommendations.

What was found

Mean performance measures were: abnormal interpretation rate (AIR) of 11.6% (95% CI: 11.5–11.6); cancer detection rate (CDR) of 5.1 per 1,000 screens (95% CI: 5.0–5.2); sensitivity of 86.9% (95% CI: 86.3–87.6%); specificity of 88.9% (95% CI: 88.8–88.9%); false-negative rate of 0.8 per 1,000 screens (95% CI: 0.7–0.8); PPV1 of 4.4% (95% CI: 4.3–4.5%); PPV2 of 25.6% (95% CI: 25.1–26.1%); and PPV3 of 28.6% (95% CI: 28.0–29.3%). Of detected cancers, 76.9% were stage 0 or 1, 57.7% were minimal, and 79.4% were node-negative invasive cancers. While 92.1% and 97.1% of radiologists achieved recommended CDRs and sensitivity, only 59.0% achieved recommended AIRs and 63.0% achieved recommended specificity.

Why it matters

This study provides updated national benchmarks for digital mammography in community practice. It demonstrates high sensitivity and cancer detection rates nationwide, while identifying high recall rates as an ongoing area for quality improvement.

Limits

The abstract does not provide subgroup analyses by age, race/ethnicity, or breast density, nor does it account for radiologist reading volume or facility characteristics. Long-term clinical endpoints such as mortality reduction or overdiagnosis were not measured.

Cited by