A feedback protocol improves the diagnostic performance of MR arthrography by experienced musculoskeletal radiologists in patients with traumatic anterior shoulder instability

• Feedback improves the radiologic diagnostic reproducibility of TASI related MRA.
• Feedback also improves the radiologic diagnostic accuracy of TASI related MRA.
• Feedback enables radiologists and orthopaedic surgeons to discuss discrepancies.
• Feedback enables fine-tuning of TASI related definition interpretation agreement.
• Feedback increases congruent interdisciplinary professional development.

PurposeTo prospectively evaluate the diagnostic performance of magnetic-resonance-arthrography (MRA) by experienced musculoskeletal radiologists in patients with traumatic-anterior-shoulder-instability (TASI), after feedback protocol execution.Materials and methodsForty-five surgically confirmed MRA’s were used to enhance personal feedback, to discuss differences in outcome between MRA assessment and surgical findings and to fine-tune definition interpretation agreement of 7 different TASI-related lesions, between experienced musculoskeletal radiologists and experienced orthopaedic shoulder surgeons. After execution of the feedback protocol 20 new, surgically confirmed, MRA’s were assessed by 2 experienced musculoskeletal radiologists using a seven-lesion standardized scoring form. Kappa coefficients, sensitivity, specificity, and differences in percentage agreement or correct diagnosis (p-value, McNemar’s test) were calculated per lesion and overall per 7 lesion types to assess whether diagnostic reproducibility and accuracy was improved.ResultsPer 7 lesion types, the overall kappa and percentage of agreement, between the 2 radiologists, were dramatically increased in comparison with our former study (k = 0.81 versus k = 0.48 and 90.7% versus 78.2%, respectively). The overall sensitivity of radiologist 1 increased from 45.9% to 87.8%, the overall sensitivity of radiologist 2 increased from 63.5% to 79.6% and the overall specificity of radiologist 2 increased from 80.1% to 85.7%. Furthermore, the overall percentage of correct diagnosis of both radiologist was also exceedingly higher (85.7% and 83.6%) compared to our former study (74.4% and 74.8%).ConclusionThe implementation of our feedback protocol dramatically improved the reproducibility and accuracy of high field MRA by experienced musculoskeletal radiologist in patients with traumatic anterior shoulder instability.