Intraobserver and interobserver agreement in the classification and treatment of proximal humeral fractures

Background There is controversy surrounding the reliability of radiographic measurements and existing classifications for proximal humeral fractures. Methods Ten orthopedists, divided into 2 groups by length of experience, evaluated radiographs in 3 views from 40 proximal humeral fractures. We evaluated 11 radiographic criteria (including the Neer and pathomorphologic classifications, head-shaft angle, displacement of the humeral shaft, and lesser and greater tuberosities) and treatment indication. We also analyzed the criteria that most influenced the choice of treatment. Results Interobserver reliability was substantial for the presence of fracture of the greater tuberosity (κ = 0.749) and medial metaphyseal comminution (κ = 0.627) and moderate for the pathomorphologic classification (κ = 0.504), displacement of the greater tuberosity (κ = 0.422), and treatment decision (κ = 0.565). Intraobserver reliability was substantial for treatment indication (κ = 0.620) and presence of displacement of the fracture of the greater tuberosity (κ = 0.627 and 0.611) and moderate for the Neer (κ = 0.490) and pathomorphologic (κ = 0.607) classifications. The results were influenced by the observer's experience. The surgical indication was influenced by the pathomorphologic classification in 50% of the evaluators (odds ratio, 4.85; range, 3.30-8.65). Conclusion The pathomorphologic classification has higher reliability than the Neer classification and was the factor that most influenced the surgical decision. The determination of the presence of fracture and displacement of the greater tuberosity and medial metaphyseal comminution is reliable with the use of simple radiographs, and the results were influenced by the observer's experience.