The reliability and reproducibility of the Neer classification system--digital radiography (PACS) improves agreement

We investigated if the introduction of digital radiography, with its software permitting enhancement of plain radiographs, improved inter- and intra-observer agreement in the Neer classification.

Five observers participated in classifying 50 randomly selected radiographs with a confirmed proximal humeral fracture. The observers included a junior registrar, an upper-limb fellow, a lower-limb orthopaedic consultant and two orthopaedic shoulder consultants. Agreement was quantified via kappa values.

In general, good (0.61-0.80) results were obtained using kappa value for inter-observer reliability throughout all grades. On further analysis, however, discrepancies persist in the classification between Neer type 1/2 and Neer type 5 categories. The latter was not restricted to more inexperienced surgeons. Intra-observer agreement (> 0.81) was excellent throughout all grades.

The introduction of digital radiography aids the improvement of Neer classification of proximal humeral fractures across all grades. With superior agreement in displacement and fracture patterns, improved communication and discussion of these injuries and similar treatment plans can be expected. This may help negate one aspect of the variability in outcome of proximal humeral shaft fractures. More complex fracture configurations continue to have difficulty in interpretation and may require further imaging analysis to conclude definitively.