Interobserver and intraobserver analysis of the Neer and AO/OTA classifications for proximal humerus fractures

Abstract

Objectives: To evaluate the intra- and inter-observer agreement of the Neer and AO/OTA classification systems for proximal humeral fractures based on imaging analysis. Methods: A retrospective imaging-based study was conducted including radiographs and computed tomography scans of adult patients diagnosed with proximal humeral fractures. Three orthopedic surgeons independently classified all fractures according to the Neer and AO/OTA systems. After a minimum interval of four weeks, the same evaluators reassessed the images in a randomized order to determine intra-observer agreement. Inter- and intra-observer reliability were calculated using the Kappa coefficient, and agreement levels were interpreted according to the criteria proposed by Landis and Koch. Results: Intra-observer agreement was high for both classification systems. The Neer classification demonstrated almost perfect to substantial intra-observer reproducibility, while the AO/OTA classification showed consistently substantial agreement among evaluators. In contrast, inter-observer agreement was lower for both systems. The Neer classification presented fair inter-observer agreement, whereas the AO/OTA classification demonstrated similarly limited concordance, particularly in fractures with more complex patterns. Conclusion: Both classification systems showed satisfactory reproducibility when applied repeatedly by the same observer; however, inter-observer agreement remained limited, especially for the AO/OTA classification. These findings indicate that while the Neer and AO/OTA systems are reliable for individual clinical use, caution is warranted when they are employed as standardized tools for comparative research or multicenter studies.