Classification of proximal humerus fractures according to pattern recognition is associated with high intraobserver and interobserver agreement

BACKGROUND: The Mayo-Fundación Jiménez Díaz (FJD) classification for proximal humerus fractures aims to identify specific fracture patterns and apply displacement criteria to each pattern. The classification includes 7 common fracture patterns: isolated fractures of the greater or lesser tuberosity, fractures of the surgical neck, impacted fractures involving head rotation in a varus and posteromedial direction or in valgus, and fractures where the humeral head is dislocated (head dislocation), split (head splitting), or depressed (head impaction). The purpose of this study was to evaluate the intraobserver and interobserver agreement of the Mayo-FJD classification system using plain radiographs (xR) and computed tomography (CT). METHODS: Three fellowship-trained shoulder surgeons blindly and independently evaluated the xR and CT of 103 consecutive proximal humerus fractures treated at a Level I trauma center. Each surgeon classified all fractures according to the Mayo-FJD classification system on 4 separate sessions at least 6 weeks apart. K values were calculated for intraobserver and interobserver reliability. RESULTS: The average intraobserver agreement was 0.9 (almost perfect) for xR and 0.9 (almost perfect) for CT scans. The average interobserver agreement was 0.69 (substantial) for xR and 0.81 (almost perfect) for CT scans at the first round, and 0.66 (substantial) for xR and 0.75 (substantial) for CT scans at the second round. CONCLUSION: The pattern-based Mayo-FJD classification scheme for proximal humerus fractures was associated with adequate intraobserver and interobserver agreement using both xR and CT scan. Interobserver agreement was best when fractures were classified using CT scans.