A modification to the Walch classification of the glenoid in primary glenohumeral osteoarthritis using three-dimensional imaging.

Background: Since Walch and colleagues originally classified glenoid morphology in the setting of glenohumeral osteoarthritis, several authors have reported varying levels of interobserver and intraobserver reliability. We propose several modifications to the Walch classification that we hypothesize will increase interobserver and intraobserver reliability.

Methods: We propose the addition of the B3 and D glenoids and a more precise definition of the A2 glenoid. The B3 glenoid is monoconcave and worn preferentially in its posterior aspect, leading to pathologic retroversion of at least 15° or subluxation of 70%, or both. The D glenoid is defined by glenoid anteversion or anterior humeral head subluxation. The A2 glenoid has a line connecting the anterior and posterior native glenoid rims that transects the humeral head. Using 3-dimensional computed tomography glenoid reconstructions, 3 evaluators used the original Walch classification and the modified Walch classification to classify 129 nonconsecutive glenoids on 4 separate occasions. Reliabilities were assessed by calculating κ coefficients.

Results: Interobserver reliabilities improved from an average of 0.391 (indicating fair agreement) using the original classification to an average of 0.703 (substantial agreement) using the modified classification. Intraobserver reliabilities improved from an average of 0.605 (moderate agreement) to an average of 0.882 (nearly perfect agreement).

Conclusion: When 3-dimensional glenoid reconstructions and the modified Walch classification described herein are used, improved interobserver and intraobserver reliabilities are obtained.