Introduction: Treatment strategies for management of proximal humeral fractures are assisted by an understanding of the fracture morphology, and in particular the viability of the humeral head. Although widely accepted, the AO and Neer classification systems show poor interobserver reproducibility, and generally do not provide a basis to guide treatment regimens. Hertel described a comprehensive binary (Lego) classification system, which defines fracture plane and parts, as well as incorporating calcar length, attachment and angulation that is vital in predicting humeral head ischemia. The sequential numerical form of the classification makes it complex, and prone to categorisation error. Sandow has extended this to a more descriptive system by naming proximal humeral parts (H-head, G-Greater Tuberosity, L-lesser Tuberosity, S-shaft), recording the fracture plane, and optionally incorporating calcar length and head angulation or displacement.: The aim of this study was to compare the inter- and intraobserver reliability of this new classification system with the AO and Neer Classification, and its usefulness as a guide to management.

Patients and Methods: 49 proximal humeral fractures in 49 consecutive patients treated at the department of orthopaedics and trauma, Royal Adelaide Hospital were identified in the period of July 2007 till January 2008. All fractures of the proximal humerus were examined using AP, lateral and axial radiographs. Three independent reviewers, looking specifically at interobserver correlation and the indication of humeral head viability, classified the fractures using the AO, Neer and “HGLS Classification”.

Results: The median age of patients was 72 (range 50 to 85). Based on the interobserver correlation analysis, the AO (κ-value 0.47) and Neer κ-value (0.44) classification systems were graded as poor and were consistent with values published in articles in the past. The HGLS Classification” showed good interobserver agreement for all three examiners (κ-value 0.73). Similar κ-values were also seen for intraobserver agreement.

Conclusion: While the parts system of Neer and AO-system can still provide a general impression of the fracture form, the “HGLS classification” for proximal humeral fractures provided a more precise description of the fracture pattern which has important prognostic and therapeutic implications. It is quick to apply and easy to use as it does not require the memorising of a numerical classification. Our study showed a good reliability for the classification system, however further studies seem necessary to assess validity of the HGLS-system.

Treatment strategies for the management of proximal humeral fractures are assisted by an understanding of the fracture morphology and, in particular, the viability of the humeral head. Although widely accepted, the AO and Neer classification systems show poor interobserver reproducibility and generally do not provide a basis to guide treatment. The aim of this study was to compare the interobserver and intraobserver reliability of a new classification system with the AO and Neer classifications and review its usefulness as a guide to management.

Hertel described a comprehensive binary (Lego) classification system, which defines fracture planes and parts, as well as incorporating calcar length, attachment and angulation. This facilitates predicting humeral head ischemia; however the sequential numerical form of the classification makes it complex and prone to categorisation error. Sandow has extended this to a more descriptive system by naming proximal humeral parts (H-head, G-greater tuberosity, L-lesser tuberosity, S-shaft), recording the fracture plane and optionally incorporating calcar length and head angulation or displacement.

50 proximal humeral fractures in 50 patients treated at the Department of Orthopaedics and Trauma, Royal Adelaide Hospital, were identified from the period of July 2007 to January 2008. All fractures of the proximal humerus were examined using AP, lateral and axial radiographs. Three independent reviewers classified the fractures using the AO, Neer and “HGLS Classification”. The findings were analysed specifically for intra/interobserver correlation and the indications for humeral head viability.

The median age of patients was 72 (range 50 to 85). Based on the interobserver correlation analysis, the AO and Neer Classification systems were graded as poor. The ‘HGLS’ Classification showed good interobserver agreement for all three examiners and more consistently provided guidelines for management based on humeral head viability.

While the parts system of Neer can still provide a general impression of the fracture form, the “HGLS classification” for proximal humeral fractures provided a more precise description of the fracture pattern which has important prognostic and therapeutic implications. It is quick to apply and easy to use as it does not require memorisation of a numerical classification and can help to understand fracture patterns and thus aid planning of a reduction and fixation strategy. Good interobserver correlation makes it a useful tool for communication between surgeons.