LAUSR

Background The objective of this study was to investigate inter- and intraobserver reliability of the morphological Mutch classification for greater tuberosity (GT) fragments in consecutive proximal humerus fractures (PHF) regardless of the number of parts according to the Codman classification system for three different imaging modalities (plain radiographs, two-dimensional [2-D] computed tomography [CT], and reformatted, three-dimensional [3-D] CT reconstruction). Materials and methods One hundred thirty-eight consecutive PHF with GT involvement were identified between January 2018 and December 2018 in a supraregional Level 1 trauma center. GT morphology was classified by three blinded observers according to the morphological Mutch classification using the picture archiving and communication software Visage 7.1 (Visage Imaging Inc., San Diego, CA, USA). Fleiss’ and Cohens’ kappa were assessed for inter- and intraobserver reliability. Strength of agreement for kappa (k) values was interpreted according to the Landis and Koch benchmark scale. Results In cases of isolated GT fractures (n = 24), the morphological Mutch classification achieved consistently substantial values for interobserver reliability (radiograph: k = 0.63; 2-D CT: k = 0.75; 3-D CT: k = 0.77). Moreover, use of advanced imaging (2-D and 3-D CT) tends to increase reliability. Consistently substantial mean values were found for intraobserver agreement (radiograph: Ø k = 0.72; 2-D CT: Ø k = 0.8; 3-D CT: Ø k = 0.76). In cases of multi-part PHF with GT involvement (n = 114), interobserver agreement was only slight to fair regardless of imaging modality (radiograph: k = 0.3; 2-D CT: k = 0.17; 3-D CT: k = 0.05). Intraobserver agreement achieved fair to moderate mean values (radiograph: Ø k = 0.56; 2-D CT: Ø k = 0.61; 3-D CT: Ø k = 0.33). Conclusion The morphological Mutch classification remains a reliable classification for isolated GT fractures, even with 2-D or 3-D CT imaging. Usage of these advanced imaging modalities tends to increase interobserver reliability. However, its reliability for multi-part fractures with GT involvement is limited. A simple and reliable classification is missing for this fracture entity.