Abstract

Our purpose was to verify if the Neer and AO-ASIF classifications for fractures of the proximal humerus satisfy the requisites of simplicity and reproducibility and if the parameters that they consider to establish the severity of the fracture are similar. Two of the authors classified the proximal humeral fractures of 227 patients based on plain radiographs, and they repeated the classification five years later. The reliability, reproducibility and coherence of the classifications were investigated. Inter-observer reliability was K = 0.77 (Neer) and K = 0.65 (AO-ASIF) while intra-observer reproducibility was K = 0.68 (examiner I) and K = 0.63 (examiner II). In 1/5 of the cases, disagreement led to a different classification of the same fracture. Furthermore, neither classification establishes a linear scale of gravity able to provide an indication for treatment. The Neer and AO-ASIF classifications have a low reproducibility and reliability when fractures, especially those with 3 or 4 parts, are assessed by means of plain radiographs. Therefore, patients with complex fractures should be submitted to CT to have a correct pre-operative diagnosis.