Second opinions in orthopedic oncology imaging: can fellowship training reduce clinically significant discrepancies?

To determine factors that lead to significant discrepancies in second-opinion consultation of orthopedic oncology patients, and particularly if musculoskeletal fellowship training can decrease clinically significant discrepancies.

A PACS database was queried for secondary reads on outside cross-sectional imaging studies, as requested by orthopedic oncology from 2014 to 2017. Comparison of original and secondary reports was performed using a published seven-point scale that defines clinically significant discrepancies. An online search was performed for each original radiologist to record if a fellowship in musculoskeletal imaging was completed. Additionally, years of post-residency experience, number of Medicare part B patients billed per year (marker of practice volume), and average hierarchical condition category for each radiologist (marker of practice complexity) was recorded.

A total of 571 patients met the inclusion criteria, with 184 cases initially interpreted by an outside fellowship trained musculoskeletal (MSK) radiologist and 387 cases initially interpreted by a non-MSK trained radiologist. The rate of clinically significant discrepancy was 9.2% when initially interpreted by MSK radiologists compared with 27.9% when initially performed by non-MSK radiologists (p < 0.05). After adjustment by both patient characteristics and radiologist characteristics, the likelihood of clinically significant discrepancies was greater for initial interpretations by non-MSK radiologists compared with MSK radiologists (OR = 1.36; 95% CI = 1.23-2.49).

In orthopedic oncology patients, the rate of clinically significant discrepancies was significantly higher when initially interpreted by non-MSK radiologists compared with MSK radiologists. The lower rate of clinically significant discrepancies demonstrates that subspecialty training may direct more appropriate diagnosis and treatment.