Accurate diagnoses are directly related to the number of mammograms interpreted by a physician, according to a UCSF study published in the March 6 issue of the Journal of the National Cancer Institute.
The study was done to understand factors that affect quality in mammography. The participants were each given a set of 60 mammograms to interpret. From the U.S., 60 radiologists were divided into three groups: low volume radiologists who read 100 or fewer mammograms per month, medium volume radiologists who read 101-300 mammograms per month, and high volume radiologists who read more than 300 mammograms per month. Joining them were 194 radiologists from the U.K, all high volume readers. The results of two Swedish radiologists were used to compare high volume results but not included in the data analyses.
Sensitivity, or the ability to detect cancer, was higher among the high volume readers than medium and low volume readers. Furthermore, the U.K. high volume readers were as accurate or more accurate than the high volume U.S. readers were. The high accuracy achieved by the U.K. readers is significant, said the authors, because mammograms are far more likely to be followed by biopsies in the U.S. than in the U.K. “These results suggest that higher biopsy rates are not necessary to find more cancers,” said lead author Laura Esserman, MD, MBA, UCSF assistant professor in residence and director of the Carol Franc Buck Breast Care Center at UCSF.
Overall, high volume readers were shown to have better accuracy – they found more cancers and more often were correct in identifying non-malignant abnormalities. Low volume U.S. readers identified fewer abnormalities than any of the high volume readers. High volume U.S. readers were 76 percent and U.K. readers 79 percent accurate; medium volume U.S. readers were 70 percent and low volume U.S. readers 65 percent accurate.
The conditions under which physicians provide mammography vary widely. In both Sweden and the U.K. mammography is performed at centralized centers. All of the participating U.K. radiologists were part of the National Screening Programme, which specifies a minimum annual reading volume for participation at 5000 mammograms per year. In the U.S. mammography is provided in a variety of settings, and the minimum reader volume, as set by the Mammography Quality Standards Act of 1992, is 480. Most U.S. radiologists, therefore, do not read a high volume of mammograms. “There may be an opportunity to improve quality and efficiency by re-engineering the organization of U.S. mammography screening programs,” said the study authors.
The difference between the studied nations in how mammography results are handled is also significant: in the U.S., physicians are much more likely than physicians in the other studied countries to recommend a mammogram be followed with a biopsy. Non-surgical biopsy (stereotactic or fine-needle aspiration) in the U.K. and Sweden reveals a cancer 30–60 percent of the time. This rate is similar to some of the most experienced breast cancer screening programs, which report finding cancer in 37-40 percent of non-surgical biopsies. The largest U.S. study found just 20-25 percent of non-surgical biopsies revealed cancer, and some studies report rates a low as 11 percent, meaning most were performed in response to mammograms interpreted falsely as positive.
“Finding ways to keep accurate cancer detection high while avoiding unnecessary biopsies would improve women’s experience of mammography,” said Esserman.
The intent to maximize sensitivity – to identify all potential cancers – is one of the reasons sometimes cited for the relative acceptance in the U.S. of false positives. This study maintains that minimizing false positives has not been shown to be associated with a lack of thoroughness in detecting cancer. The bias for more biopsies in the U.S. may actually reflect a cultural bias rather than a quality advantage, according to the authors.
“Many women undergo biopsies for benign findings, which causes great emotional distress. Furthermore, the cost of potentially unnecessary biopsies for the U.S. as a whole is more than $1 billion annually,” said Esserman.
The report states that there are economic, medical and humanitarian reasons to improve quality and efficiency by re-engineering the organization of U.S. mammography screening programs. “Having high volume, experienced readers involved with mammography screening as the primary reader, or as double reader for smaller localities, could increase the quality of care while simultaneously decreasing the cost,” said the study authors.
The study was funded by the California Breast Cancer Research Program. Additional authors are: Carey Eberle, UCSF analyst; Sophia Chang, MD, UCSF assistant clinical professor of medicine; Helen Cowley, University of Derby, U.K. research assistant; Alastair Gale, BSc PhD FErgS CPsychol FBPsS, professor of psychology; Alistair Kirkpatrick, MD, Scottish Breast Screening Programme, South-East Scotland Division, Edinburgh, Scotland.
Journal
JNCI Journal of the National Cancer Institute