Study: Radiologists more confident in double reading than CAD
The expansion of CAD in the U.S. since 1998 has been accompanied by a sharp decline in double reading, but this trend has progressed with little attention paid to radiologists’ perceptions toward both, the authors contended.
Tracy Onega, PhD, of Dartmouth Medical School in Lebanon, N.H., and colleagues sought to measure the rates of use and the perceptions of CAD and double reading in a survey of 257 radiologists from the Breast Cancer Surveillance Consortium. The sample included community-based radiologists as well as primary and adjunct professors of radiology. Eighty-six percent of the participants reported reading 1,000 or more mammograms per year.
Seventy-five percent of the radiologists surveyed reported using at least some CAD, compared with 28 percent of radiologists using at least some double reading and 23 percent using some CAD and some double reading. Nearly half of radiologists surveyed reported using CAD for all mammography readings.
Seventy-five percent of participants also responded that they believed double reading improved cancer detection rates, while 55 percent answered that the same was true of CAD. Eighty-one percent of radiologists said that double reading reassures reviewers, compared with 65 percent for CAD.
“These perceptions suggest greater confidence in colleagues’ interpretations (double reading) compared to those of a computer,” the authors explained.
The most common complaints levied against double reading by respondents were that the method takes too much time and is not profitable, whereas about half of respondents said that CAD increases recall rates and makes practices more profitable.
The study also found that 98 percent of primarily academic radiologists looked favorably upon double reading, compared with 69 percent of adjunct professors and 59 percent of entirely private practice radiologists.
Three times as many radiologists also said that double reading better protects physicians against malpractice cases, a major issue among practices screening for breast cancer.
The study did not control for the different CAD programs or methods of double reading that radiologists used, which may have influenced participants' perceptions of the techniques. The authors also acknowledged that some of the subgroups in their sample were small and susceptible to error.
The authors expressed the importance of further studies to understand the higher confidence but low utilization of double reading compared to CAD.