A new US study found that interpretation of diagnostic mammograms varied widely in a way that could not be explained by variations in patients but might be explained by differences in radiologist attributes, such as experience. The researchers expressed concern that such a variation could influence clinical decisions affecting large numbers of women.
The study is published in the December 11th online issue of the Journal of the National Cancer Institute and was carried out by Dr Diana Miglioretti of the Group Health Center for Health Studies in Seattle, Washington, and colleagues.
Other studies have already shown extensive variation in the intepretation of screening mammograms by radiologists, but less is known about whether this is also a feature of diagnostic mammograms, which are used to further examine suspicious findings from screening mammograms or physical examinations.
In this latest study the researchers investigated radiologist attributes in relation to accuracy of diagnostic mammogram interpretation. They focused on experience primarily, and also on one or two other characteristics such as whether the radiologist was working in an academic medical centre.
The investigation pulled data from the Breast Cancer Surveillance Consortium on 36,000 diagnostic mammograms interpreted by 123 radiologists working at 72 facilities.
In terms of radiologist experience, the researchers ascertained that:
- 75 per cent of the radiologists had been interpreting mammograms for at least 10 years.
- 87 per cent of them spent less than 40 per cent of their time interpreting mammograms.
- About 25 per cent of them worked on no more than 100 mammograms a year.
- The median age of the radiologist group was 49 years.
- 75 per cent of them were male.
The results showed considerable variation in the performance of the radiologists, for which there was no consistent link with patient characteristics.
About 80 per cent of breast cancers were diagnosed correctly, but more than 4 per cent of female patients without cancer were tentatively diagnosed with cancer from their mammogram interpretation (false positives).
Radiologists who were in academic settings were more likely to diagnose a cancer correctly and were less likely to give a false positive diagnosis.
However, the study only included 7 radiologists working in academic settings, which is rather a low number from a statistical point of view and the results should be interpreted with caution said the researchers.
The more experienced radiologists were more likely to have a higher threshold for recommending a biopsy, while those radiologists who took longer over interpreting mammograms usually had a lower threshold.
Such a variation in the diagnostic performance of radiologists is worrying warned the researchers, because the rate of breast cancer in 10 times higher from diagnostic mammograms compared to screening mammograms. It is likely to affect many women, with and without breast cancer.
“To realistically improve mammography interpretation for women across the United States, we need to identify ways to improve accuracy and reduce variability among all radiologists who interpret mammography,” wrote the researchers.
“Radiologist Characteristics Associated With Interpretive Performance of Diagnostic Mammography.”
Diana L. Miglioretti , Rebecca Smith-Bindman , Linn Abraham , R. James Brenner , Patricia A. Carney , Erin J. Aiello Bowles , Diana S. M. Buist , and Joann G. Elmore.
Journal of the National Cancer Institute Advance Access published on December 11, 2007.
DOI 10.1093/jnci/djm238.
Written by: Catharine Paddock