Does Bone Loss Imaging Modality, Measurement Methodology, and Interobserver Reliability Alter Treatment in Glenohumeral Instability?

Purpose: To determine, in the context of measuring bone loss in shoulder instability, whether measurement differences between magnetic resonance imaging (MRI) and computed tomography (CT), linear-based and area-based methods, and observers altered the proposed treatment when a standardized algorithm was applied. Methods: This was a retrospective, comparative imaging study of preoperative patients with anterior shoulder instability with both an MRI and CT scan within 1 year of one another. On parasagittal images reoriented en face to the glenoid, 2 attending orthopaedic surgeons measured glenoid width, glenoid area, glenoid defect width, and glenoid defect area. On axial images maximal Hill-Sachs width was measured. From these, linear percent glenoid bone loss (%GBL) and area %GBL were calculated, and on-versus off-track was determined. With these results, a recommended treatment was determined by applying a standardized algorithm, in which the Latarjet procedure was selected for %GBL >20%, arthroscopic labral repair and remplissage for off-track lesions with %GBL <20%, and arthroscopic labral repair on-track shoulders with %GBL <20%. Results: In total, 53 patients with mean +/- standard deviation 45 +/- 83 days between scans were include with a CT linear %GBL of 23.5 +/- 9.6% (range 0%-47%). CT lead to larger measurements of %GBL than MRI (linear P = .008, area P = .003), and fewer shoulders being considered on-track (33.0% vs 40.5%), which would alter treatment in 25% to 34%. Linear measurements produced larger values for %GBL (CT, P < .001; MRI, P < .001), which would alter treatment in 25%. For %GBL, inter-rater reliability was good, with intraclass correlation coefficients varying from 0.727 to 0.832 and Kappa varying from 0.57 to 0.62, but these inter-rater differences would alter treatment in 31%. Conclusions: The significant differences in bone loss measurement between imaging modality, measurement method, and observers may lead to differences in treatment in up to 34% of cases. Linear CT measurements resulted in the most aggressive treatment recommendations.