Diagnostic mammography: identifying minimally acceptable interpretive performance criteria

Area of Science:

• Radiology and Medical Imaging
• Quality Improvement in Healthcare
• Breast Cancer Screening

Background:

• Mammography interpretation is crucial for early breast cancer detection.
• Establishing performance standards for physicians is essential for quality assurance.
• Variability in interpretive performance can impact patient outcomes.

Purpose of the Study:

• To define minimally acceptable performance thresholds for physicians interpreting diagnostic mammography.
• To develop criteria for evaluating mammography interpretive performance in clinical practice.

Main Methods:

• The Angoff approach was utilized to establish performance criteria.
• Normative data from the Breast Cancer Surveillance Consortium (BCSC) informed cut-point identification.
• Simulations using BCSC data estimated the clinical impact of performance thresholds.

Main Results:

• Specific performance cut points were defined for both abnormal screening workups and breast lump evaluations, including sensitivity, specificity, and positive predictive values.
• For abnormal screening workups, thresholds include sensitivity <80%, specificity <80% or >95%, abnormal interpretation rate <8% or >25%, PPV2 <15% or >40%, PPV3 <20% or >45%, and cancer diagnosis rate <20/1000.
• For breast lump workups, thresholds include sensitivity <85%, specificity <83% or >95%, abnormal interpretation rate <10% or >25%, PPV2 <25% or >50%, PPV3 <30% or >55%, and cancer diagnosis rate <40/1000.

Conclusions:

• Physicians performing below identified thresholds require review within their practice context.
• Remedial training should be considered for physicians whose performance falls outside acceptable ranges.
• Adherence to these criteria can improve cancer detection rates and reduce false positives.