Research has already proven that the screening of mammography results is enhanced considerably when two readers analyze the results as opposed to only one. Although this has become a standard practice in Europe, it is not widely used in the U.S. However, according to a group of British researchers, it has now been shown that a single reader can use computer assistance in performing the analysis with the effectiveness of having two readers according to a new study.
Mammograms are performed women to screen for early signs of breast cancer. In the United States, X-rays are read by single radiologists, which allows for cancers to sometimes be missed. Experts maintain that there are not enough radiologists available to be able to give mammograms two readings. In addition, insurers will not pay for a second reading.
Computer-aided detection (CAD) was developed to assist radiologists in spotting more cancers. These computer programs were approved ten years ago and Medicare will pay an additional $15 for the use of CAD. However, CAD is currently only used for about one third of U.S. mammograms due to concerns regarding the accuracy of the technology.
Dr. Fiona J. Gilbert, from the University of Aberdeen in the UK, and colleagues conducted a randomized controlled trial that involved 31,057 women’s mammograms in comparing single reading utilizing CAD with double reading. The study found that the sensitivity, specificity, and positive predictive value for the two reading practices were about the same.
The research found that double readings detected 87.7 percent of cancers whereas single reading with computer assistance detected 87.2 percent. The actual numbers of cancers detected were 198 out of 227 for the single reader with CAD, compared to 199 for the two readers. In addition, 3.4 percent of women were recalled for further assessment where double reading was used, white the rate for single reading plus computer assistance was slightly higher at 3.9 percent.
The study results for the CAD single reading included sensitivity at 87.2 percent, specificity at 96.9 percent and a positive predictive value of 18 percent. For the double reading the percentages were 87.7 for sensitivity, 97.4 for specificity and 21.1 for a positive predictive value.