Assessment of painful total shoulder arthroplasty using computed tomography arthrography.

Background: This study assessed the accuracy of computed tomography (CT) arthrography when evaluating glenoid component stability in the setting of postarthroplasty shoulder pain.

Methods: We retrospectively reviewed all patients presenting to the clinic during a 5.5-year period to identify those with a painful shoulder arthroplasty more than 1 year after the index procedure. We excluded reverse and hemiarthroplasty procedures, patients with a clearly identifiable cause for pain, such as rotator cuff insufficiency or gross component loosening as seen on plain radiographs, and those with culture-positive aspiration. There were 14 patients with suspected glenoid component loosening but inconclusive plain radiographs. Each of the 14 patients underwent a CT arthrogram that was evaluated by the senior author (J.J.P.W.) for the presence or absence of contrast material underneath the polyethylene component. Operative reports and surgical videos from subsequent arthroscopy were reviewed to assess glenoid component stability as determined by direct arthroscopic visualization.

Results: CT arthrography suggested glenoid component loosening in 8 of 14 patients (57.1%), and arthroscopic inspection identified loosening in 10 of 14 patients (71.4%). In 3 of 10 patients (30%), CTA suggested a well-fixed glenoid component, but gross loosening was identified during arthroscopy. In this study, CTA yielded a sensitivity of 70%, a specificity of 75%, a positive predictive value of 87.5%, and a negative predictive value of 50.0%.

Conclusion: CTA had a low negative predictive value (50%), and therefore, the prediction of component stability based on the absence of contrast between the glenoid component and the bone-cement interface does not always reflect true stability.