When Radiologists Have Risk Profiles, Fewer Breast Cancers are Missed

June 20, 2014

When mammograms are read by radiologists who are familiar with the patients’ risk factors, there are fewer missed diagnosis and false positives.

Knowing a patient’s risk profile for breast cancer when reading mammograms reduces missed cancers and false positives, according to a paper presented at the conference of the Institute for Operations Research and the Management Sciences (INFORMS).

Researchers from the University of Texas and Southern Methodist University in Dallas analyzed the decision performance of mammogram-only reading (no risk profile information available), an unbiased reading (radiologists consult the risk profile after examining the mammogram), and biased readings (radiologists consult a woman's risk profile as they examine the mammogram). The researcher then examined the conditions in which profile information could help improve biopsy decisions. Risk factors included family history, reproductive history, age, ethnicity and others.

The concern with providing this information to radiologists is whether this would introduce bias and if the bias would hinder or help the accuracy of the readings.

The simulated results showed that readings performed by radiologists who were aware of the patients’ risk profiles had fewer false positives by 3.23 percent (from 27,158 to 26,282 for every 100,000 cancer-free patients) and a reduction of 3.70 percent in number of false negatives (from 196.78 to 189.49 for every 1,000 cancer patients), when compared to readings that were performed without risk profile knowledge.

“Currently, no guidelines exist as to whether and when to use the risk profile information while interpreting mammograms. We find that using the risk profile information at the right time and assigning it an appropriate weight can spare women of unnecessary procedures and can help us with better cancer detection,” coauthor Mehmet Ayvaci, assistant professor in information systems and operations management, Naveen Jindal School of Management, University of Texas at Dallas, told Diagnostic Imaging. “If supported by future research, guidelines can incorporate the optimal use of risk profile information in managing the breast disease.”