Intra- and Interobserver Reliability of the CT-Based Glenoid Arc-Angle Method Using Injured-Side and Contralateral Reference Circles: A Diagnostic Reliability Study

Background/Objectives: This study aimed to compare the intraobserver and interobserver reliability and absolute agreement of the CT-based glenoid arc-angle method when the reference circle was constructed either directly on the injured glenoid or transferred from the healthy contralateral glenoid. The objective was to assess measurement reproducibility rather than anatomical validity, diagnostic accuracy, or clinical utility. Methods: Preoperative CT scans of 31 patients with anterior shoulder instability who underwent surgery were retrospectively analyzed. Three independent observers with different levels of experience performed measurements at two separate time points using two geometrical approaches. In the contralateral-reference method, the diameter of the best-fit circle was determined on the healthy contralateral glenoid and transferred to the injured side. In the injured-side method, the best-fit circle was determined directly on the injured glenoid. The defect angle and glenoid diameter were measured, and the defect angle was converted into an area-based percentage of glenoid bone loss. Intraobserver reliability was assessed across the two measurement sessions, whereas interobserver reliability was assessed across observer-specific means from the two sessions, using intraclass correlation coefficients with 95% confidence intervals. Absolute agreement, systematic bias, and measurement error between sessions were additionally evaluated using Bland–Altman analysis, the standard error of measurement, and the minimum detectable change. Results: The contralateral-reference method yielded numerically higher ICC point estimates for both measured parameters. For defect-angle measurements, the interobserver ICC based on observer-specific two-session mean estimates was 0.890 with the contralateral-reference method (95% CI: 0.811–0.942) and 0.603 with the injured-side method (95% CI: 0.399–0.767). For glenoid-diameter measurements, the corresponding interobserver ICCs were 0.900 (95% CI: 0.660–0.961) and 0.812 (95% CI: 0.672–0.900), respectively. Several confidence intervals crossed conventional reliability category boundaries and partially overlapped between methods; therefore, these findings represent numerical differences in point estimates rather than statistically established superiority. Conclusions: Within this cohort, the contralateral-reference method yielded numerically higher ICC point estimates, smaller session-related biases, and lower measurement-error estimates than the injured-side method. These descriptive findings do not establish statistically significant superiority, diagnostic accuracy, clinical utility, or treatment thresholds. The results are limited by the relatively small retrospective cohort of 31 surgically treated patients and the absence of anatomical or clinical outcome validation.