Area Measurement Percentile of 3-Dimensional Computed Tomography Has the Highest Interobserver Reliability When Measuring Anterior Glenoid Bone Loss

Abstract

To determine the accuracy of glenoid bone loss measurement and the difference between 3 methods of measurement, as well as the measurements application to previously published studies. A list of patients with anterior bony glenoid defects was created through a search of electronic medical records. Three surgeons reviewed each patient's advanced imaging (computed tomography [CT], 3-dimensional [3D] CT, or magnetic resonance imaging), and glenoid bone loss was measured by 3 different methods: (1) linear measurement percentile (LMP), (2) area measurement percentile (AMP), and (3) circle-line method (CLM). The intraclass correlation coefficients between reviewers and mathematical differences between measurement techniques were calculated. The images of 125 patients with anterior glenoid bone loss were measured. For all imaging studies, the intraclass correlation coefficient was greatest with the AMP (0.738) and CT with 3D reconstruction (0.735). Within the entire sample, average bone loss measured 21.3% (range, 5.6%-43.5%) by the LMP method, 15.7% (range, 1.6%-42.2%) by the CLM, and 16.5% (range, 2.3%-40.3%) by the AMP method. On average, the difference between the LMP and AMP methods was 4.8%. When the AMP and LMP methods were compared, the greatest difference in measurement was 5.9%, and this occurred at an LMP of 19.1%, which was an AMP of 13.2%. When measuring anterior glenoid bone loss, CT with 3D reconstruction and the AMP method have the greatest interobserver reliability. Furthermore, the greatest difference between the LMP and AMP methods occurs at an LMP between 18.3% and 20.0% and an AMP between 12.4% and 14.2%, with the difference ranging from 5.7% to 5.9%. When measuring anterior glenoid bone loss, evaluation of CT with 3D reconstruction is more reliable than magnetic resonance imaging evaluation. Furthermore, the AMP method has the greatest interobserver reliability when compared with the LMP method and CLM.