Joseph W. Galvin, Stephen A. Parada, Xinning Li, Josef K. Eichinger

Posterior shoulder instability is less common and potentially more difficult to diagnose clinically and radiographically compared with anterior shoulder instability. Radiographic findings including posterior labral tears, increased retroversion, presence of glenoid dysplasia, and increased capsular area are associated with symptomatic recurrent posterior shoulder instability. This study aimed to determine the prevalence and severity of associated radiographic parameters found on magnetic resonance arthrograms (MRAs) in patients with arthroscopically confirmed isolated posterior labral tears and symptomatic recurrent posterior shoulder instability, compared with an age-matched cohort of patients without posterior instability or labral injury confirmed with shoulder arthroscopy. Cross-sectional study, Level of evidence, 3. Patients who received a preoperative standard shoulder MRA at an academic institution over a 5-year period and had symptomatic posterior instability and received a repair of an arthroscopically confirmed posterior labral tear (n = 63) were identified. These patients were compared with an age-matched control group of patients without posterior instability (n = 49) who underwent an isolated arthroscopic distal clavicle resection that included an arthroscopic glenohumeral joint evaluation. Glenoid version, posterior humeral head subluxation, glenoid dysplasia, and linear and capsular area measurements were evaluated between the 2 groups. Interobserver reliability for continuous and categorical variables was assessed for all measurements. Multivariate logistic regression revealed that the presence of increased glenoid retroversion (P = .0018), glenoid dysplasia (P = .03), and increased axial posterior capsular cross-sectional area (P = .05) were significantly associated with posterior labral tears and symptomatic posterior shoulder instability compared with the age-matched control group. Posterior humeral head subluxation was found to be a statistically significant variable with univariate analysis (P = .001) for posterior shoulder instability but not with multivariate logistic regression (P = .53). Interobserver reliability was good to very good for all measurements (intraclass correlation coefficient [ICC] = 0.74-0.85; κ = 0.64) but was moderate for total capsular area and sagittal capsular area measurements (ICC = 0.43-0.56). The presence of increased glenoid retroversion, glenoid dysplasia, and increased posterior capsular area on MRA are significantly associated with posterior labral tears and symptomatic posterior shoulder instability. Identification of these critical radiographic variables on magnetic resonance arthrography assists in the accurate diagnosis and management of clinically significant posterior shoulder instability.