Comparison of two- and three-dimensional implant-based scapulohumeral rhythm measurements derived from two radiographic images in shoulders with reverse total shoulder arthroplasty

Background: The accuracy of two-dimensional (2D) scapulohumeral rhythm (SHR) measurements in comparison to the precise three-dimensional (3D) SHR measurements in shoulders with reverse total shoulder arthroplasty (rTSA) has not been fully validated. The primary aim of this basic science study was to assess the validity of the mean SHR, calculated from the arm at the side to maximum abduction using 2D measurements derived from 2 radiographic images, in comparison to 3D measurements obtained through 3D to 2D model-image registration in shoulders with rTSA. Methods: The study included 35 shoulders from 34 patients who underwent rTSA at a single institution. Each shoulder underwent computed tomography and single-plane fluoroscopy at an average of 14 (range, 12-27) months post-operatively. Fluoroscopic images were acquired during scapular plane abduction and 2 fluoroscopic images taken with the arm at the side and at maximum abduction were used for the mean SHR measurement from the arm at the side to maximum abduction. For 3D measurement using model-image registration techniques, the poses of 3D models were iteratively adjusted to match the silhouettes in the fluoroscopic images. For 2D measurement, a line drawn along the lateral side of the humeral stem, a line connecting the top and bottom edges of the backside of the glenosphere, and a vertical thoracic reference line were used. SHR was defined as (ΔH-ΔS)/ΔS, where ΔH is the increment in humeral abduction angle and ΔS is the increment in scapular upward rotation angle. The agreement between 3D and 2D measurements was assessed using Bland-Altman plots. Pearson's correlation coefficient was used to evaluate the correlation between the mean SHR of 2 methods. Results: < .001). Conclusion: This study demonstrated a significantly strong positive correlation between 2 measures of SHR. However, both fixed bias and proportional bias were observed. Consequently, while a 2D measurement of SHR may offer some clinical utility, it is crucial to recognize its tendency to underestimate the mean SHR compared to 3D measurement.