Algorithm for Evaluation of Bipolar Defects in Anterior Instability of the Shoulder

Purpose of the study — to justify the algorithm for evaluation of bipolar defects in anterior shoulder instrability using the most accurate, statistically significant and reproducible methods which would make the algorithm applicable in practical surgery.Materials and methods. The authors established 4 groups with 6 patients in each with shoulder instability, group distribution was based on glenoid defect size: small (15%), moderate (15-19%), large (20-25%) and massive (25%). All 24 patients underwent 3D-CT, 3D VIBE MRI and shoulder arthroscopy. Measurements were taken by 7 specialists 5 of whom measured defect during arthroscopy. Glenoid defect was measured by linear and sectional relation methods. Pico method on 3D-CT was taken as the “golden standard”. Accuracy was verified by analysis of variance with post-hoc comparison. Reproducibility was evaluated by intraclass correlation coefficient.Results. All groups excluding the one with massive glenoid defects demonstrated significant differences from the model (p≤0,05) for measurements during arthroscopy and examinations by 3D-CT and 3D VIBE MRI. Restrospective analysis confirmed the least accuracy and the worst reproducibility of visual evaluation of glenoid defects less than 25%. Sectional relation method on 3D-CT had the maximum accuracy and reproducibility in all groups (PE = 1,29%±2,39%, ICC = 0,756-0,856), excluding the group with massive defects, where researched measurement methods had close accuracy when applied on 3D-CT, 3D VIBE MRI and during arthroscopy. Linear relation method on 3D-CT overestimated the defect volume at 2,1-7,9% and demonstrated less reliable reproducibility (PE = 3,22%±5,31%, ICC = 0,612-0,621). The highest error (up to 7,9%) was demonstrated by linear method in case of borderline defects in the III group of 20-25%. Insufficient conformity of results for linear (ICC = 0,42) method and moderate conformity for sectional (ICC = 0,62) method were observed during comparison of 3D VIBE MRI with 3D-CT. MRI underestimated the value of small defects and overestimated large defects. Reproducibility of measurements on 3D-CT by different operators was moderate for visual (ICC=0,594) and linear methods (ICC = 0,621) and good for sectional method (ICC = 0,756). Reproducibility of measurements by each operator also was moderate for visual and linear methods (ICC = 0,553 и ICC = 0,612) and good for sectional method (ICC = 0,856). The authors suggested an algorithm for selection of examination