2196. Subspecialty-Specific Analysis of Discrepancy Rates of Examinations Outside of Radiologists’ Fellowship Training
Authors * Denotes Presenting Author
  1. Adam Roth *; Indiana University
  2. Tarek Hanna; Emory University
  3. Christine Lamoureux; Virtual Radiologic
  4. Jamlik-Omari Johnson; Emory University
  5. Tianwen Ma; University of Michigan
  6. Timothy Johnson; University of Michigan
  7. Suzanne Chong; Indiana University
Previously published summary results lent support to multispecialty practice for acute studies in the community setting. This follow-up study analyzed the subspecialties separately to determine the relationships between major/minor discrepancy rates, fellowship training, and common/advanced studies.

Materials and Methods:
This institutional review board-approved and HIPAA compliant retrospective study used the databank of a teleradiology company. Between January 1, 2012 and December 31, 2016, 1,776,043 neuroradiology, 3,045,198 abdominal radiology, and 1,062,730 MSK radiology studies were preliminarily reported by 269 teleradiologists fellowship trained in neuroradiology, abdominal radiology, or MSK radiology followed by final reports made by on-site client radiologists. Studies were classified as common or advanced and concordant or discordant with the teleradiologists’ fellowship training. Client radiologists’ submitted discrepancy reports adjudicated by the teleradiology company’s quality assurance committee whether a major or minor discrepancy occurred were used. A 3-way conditional analysis was performed. Statistical significance was set at p = .05.

For studies concordant with fellowship, major discrepancies were lower for common studies for neuroradiology (RR=0.42, p < .001) and abdominal radiology (RR=0.41, p < .001) but higher for MSK radiology (RR=2.88, p = 0.01), while minor discrepancies were lower for neuroradiology (RR=.51, p < .001), but not significantly different for abdominal or MSK radiology. For studies discordant with fellowship, major discrepancies were higher for common studies for neuroradiology (RR=1.28, p = 0.02) and lower for abdominal (RR=0.72, p < .001) and MSK (RR=0.73, p = 0.002) radiology while minor discrepancies were higher for neuroradiology (RR=1.46, p < 0.001), lower for abdominal radiology (RR=0.85, p = 0.005), and not significantly different for MSK radiology. For common studies, major (RR=.52, p < .001) and minor (RR=.43, p < .001) discrepancies were lower for neuroradiology studies concordant with fellowship, but for abdominal radiology no significant difference was observed for major discrepancies but minor discrepancies were higher (RR=1.54, p < .001) while for MSK radiology, major discrepancies were higher (RR=1.84, p = .001) and minor discrepancies were lower (RR=.70, p = .021). For advanced studies, major discrepancies were higher for neuroradiology (RR=1.59, p = .001) and abdominal radiology (RR=1.79, p < .001) when concordant with fellowship and not significantly lower for MSK radiology, while minor discrepancies were higher for abdominal radiology (RR=1.43, p = .002) but not significant for neuroradiology or MSK radiology.

Our results suggest the need for neuroradiology training to reduce discrepancy rates for common neuroradiology studies. For common abdominal and MSK radiology studies, our results support multispecialty practice. For common MSK and advanced neuroradiology as well as abdominal radiology studies, increased major discrepancies despite concordance with fellowship was unexpected and warrants further study.