Effect of Axillary Radiograph Rotation and Medial Scapular Visualization on Glenoid Retroversion Measurement

Axillary radiographs enable the measurement of glenoid retroversion, which is associated with worsened clinical outcomes and glenoid loosening following total shoulder arthroplasty. Due to the variability in radiographic technique, this study aims to determine if the accuracy of retroversion measured by axillary radiograph is affected by 1) scapular rotation and/or 2) proper visualization of the medial scapula.

Using five cadaveric scapulae, investigators obtained axillary radiographs in true neutral position as well as in 10° and 20° of anterior and posterior rotation. For each radiograph, two fellowship trained shoulder surgeons measured glenoid retroversion with complete visualization of the scapula (Technique 1) and with visualization limited to the lateral half of scapula (Technique 2). The observers also measured glenoid retroversion by CT scan to use as a gold standard technique. Spearman's Rho was used to assess agreement between measurements.

Average glenoid retroversion of the five scapulae assessed by CT scan was 3.8° (R: 1.5-6.9). Measurements obtained using Technique 1 demonstrated improved levels of interobserver agreement (ICC: 0.412) compared to measurements obtained with Technique 2, which demonstrated no agreement (ICC: 0.103). Scapular rotation was inconsistently associated with agreement using both techniques.

The reliability of glenoid retroversion measurements was limited by incomplete visualization of the medial scapular spine. When measuring retroversion to the base of the scapular spine, improved agreement and accuracy were seen with various degrees of scapular rotation.