To investigate the influence of bone morphology on the accuracy of estimating glenohumeral external rotation, we examined accuracy of estimating glenohumeral external rotation for different combinations of variables.
Thirteen dominant shoulders of healthy adult males were examined. Glenoid sagittal orientation and humeral retroversion were measured using a three-dimensional bone model, which was generated from computed tomography. Shoulder kinematics were evaluated during shoulder external rotation at 90° abduction using single-plane fluoroscopic images. Kinematics parameters, including shoulder external rotation, glenohumeral external rotation, scapular posterior tilt, were measured using 3D to 2D model-image registration techniques. To examine the accuracy of estimating glenohumeral external rotation angle at maximum shoulder external rotation, multiple regression analyses were performed with glenohumeral external rotation as the dependent variable and shoulder external rotation, scapula posterior tilt, glenoid sagittal orientation, and humeral retroversion as independent variables, and combining all variables including shoulder external rotation as an independent variable. Residuals of measured and predicted values of glenohumeral external rotation were calculated from the obtained regression equations, and the validity between measured and predicted values was examined using intraclass correlation coefficients (ICC(3,1)).
Significant regression equation was obtained only for model 2, in which shoulder external rotation, scapular posterior tilt, and humeral retroversion were independent variable (adjusted R² = 0.582, p = 0.04). Residuals for each model were as follows: Model 1, in which all variables were independent variables, showed 0.0 ± 3.9° (95% Confidence Interval: 95% CI: -7.6 to 7.6°); Model 2 showed 0.0 ± 3.9° (95% CI: -7.7 to 7.7°); Model 3, in which shoulder external rotation, scapular posterior tilt, and glenoid sagittal orientation were independent variables, showed 0.0 ± 5.7° (95% CI: -11.3 to 11.3°); and Model 4, in with only shoulder external rotation were independent variable, showed 0.0 ± 5.9° (95% CI: -11.5 to 11.5°). ICC(3,1) was significant only in model 1 (ICC=0.741, p0.01) and model 2 (ICC=0.736, p0.01).
Models 1 and 2 demonstrated narrow 95% CI and significant ICC, indicating high validity in estimating glenohumeral external rotation. Furthermore, no difference in estimation accuracy was observed between Models 1 and 2.
More accurate estimation of glenohumeral external rotation can be achieved by measuring not only shoulder external rotation and scapular posterior tilt but also bone morphology. In clinical setting, due to need to minimize the number of evaluation variables from the perspectives of time and effort, this study suggests that measuring humeral retroversion is particular crucial for estimating glenohumeral external rotation.
Scapular morphology
Humerus morphology
