Computer-aided detection, structured reporting templates and various education strategies all can be effective as correctors of perceptual errors by radiologists. Meanwhile, double reporting represents a validated and arguably simpler solution than any of those approaches—and it may be more cost-effective too.
That’s according to a review of the literature at the University of Sydney in Australia led by Richard Pow, MBBS, the report of which was published online March 21 in the Journal of Medical Imaging and Radiation Oncology.
The team systematically reviewed 41 studies meeting inclusion criteria related to diagnostic efficacy.
Of these, 23 related to mammography (22 screening, one diagnostic), the imaging domain in which double reporting—two radiologists interpreting the same studies with one or more arbitrators available to settle discordant findings by reaching consensus—has been most widely deployed and validated.
Also meeting the inclusion criteria were six studies related to thoracic imaging, four related to neuroradiology, five to gastrointestinal, three to barium enema, two to CT colonography, two to general oncology and one to trauma.
Pow et al. point out that independent double reporting, where observers have no knowledge of one another’s reports, is generally considered to be the most effective form of the practice.
Among the 41 studies they reviewed was one looking at the impact of double reporting on plain chest imaging.
In this study, researchers presented 100 chest radiographs, 35 of which contained subtle and initially missed lung cancer nodules.
The sensitivity and specificity of individual observers varied from 20 percent to 60 percent (mean: 33 percent) and from 87 percent to 95 percent (mean: 92 percent), respectively.
Double reporting increased the sensitivity range to 33 to 67 percent, but it reduced the specificity to 82 from 94 percent.
From this and other examples presented in their report, Pow and co-authors conclude that double reporting is likely to be most useful “where emphasis is placed on high sensitivity, for example, in the case of cancer screening, where a significant benefit is achieved through early detection and missing lesions may have serious consequences.”
As for cost comparison with other methods, the authors point out that computer-aided detection, structured templates for reporting and various education strategies “can be costly, involve significant technological development and often have been shown to have minimal impact on the experienced observer. Double reading represents a relatively straightforward solution with proven benefits even among the most expert radiologist.”