Classification of proximal humeral fractures based on a pathomorphologic analysis

Background: The purpose of this study was to analyze the pathomorphology of proximal humeral fractures to determine relevant and reliable parameters for fracture classification. Methods: A total of 100 consecutive acute proximal humeral fractures in adult patients were analyzed by 2 non-independent observers from a single shoulder department using a standardized protocol based on biplane radiographs and 3-dimensional computed tomography scans. A fracture classification system based on the most reliable key features of the pathomorphologic analysis was created, and its reliability was tested by 6 independent shoulder experts analyzing another 100 consecutive proximal humeral fractures. Results: The head position in relation to the shaft (varus, valgus, sagittal deformity) and the presence of tuberosity fractures showed a higher interobserver reliability (k > 0.8) than measurements for medial hinge, shaft, and tuberosity displacement, metaphyseal extension, fracture impaction, as well as headsplit component identification (k < 0.7). These findings were used to classify nondisplaced proximal humeral fractures as type 1, fractures with normal coronal head position but sagittal deformity as type 2, valgus fractures as type 3, varus fractures as type 4, and fracture dislocations as type 5. The fracture type was further combined with the fractured main fragments (G for greater tuberosity, L for lesser). Interobserver and intraobserver reliability analysis for the fracture classification revealed a k value (95% confidence interval) of 0.700 (0.631-0.767) and 0.917 (0.879-0.943), respectively. Conclusion: The new classification system with emphasis on the qualitative aspects of proximal humeral fractures showed high reliability when based on a standardized imaging protocol including computed tomography scans. (C) 2016 Journal of Shoulder and Elbow Surgery Board of Trustees.