Inter- and intra-observer reliability of the new AO/OTA classification of proximal femur fractures.

Introduction: Hip fractures are a global health burden, with an incidence that is projected to increase from 66,000/year currently in the United Kingdom to 100,000/year by 2033. The classification of intertrochanteric fractures is key to the treatment algorithms advising on their surgical management. The AO/OTA classification is the most commonly used system, initially published in 1990 and subsequently shown to have poor inter- and intra-observer reliability, it was revised in 2018 with the main aim of re-classifying and further defining the 31-A2 group.

Methods: 150 plain film anteroposterior and lateral plain film radiographs of intertrochanteric fractures from three hospitals were classified using the 2018 AO/OTA classification of intertrochanteric fractures by six Orthopaedic Surgeons (2 Consultants, 4 Trainees), all were blinded to the definitive surgical treatment for patients. Radiographs were re-classified after a minimum of 3-months, Cohen's Kappa for inter-observer reliability was calculated from first round classifications and intra-observer reliability from first and second classifications.

Results: Mean Kappa for inter-observer reliability for AO group classification (e.g. 31-A1) was 0.479 (0.220 - 0.771, for sub-group classification (e.g. 31-A1.1) reliability reduced to 0.376 (0.276 - 0.613). Intra-observer reliability was comparable for both group and sub-group classifications, 0.661 and 0.587 respectively.

Conclusions: The revised 2018 AO/OTA classification aimed to simply the classification of intertrochanteric fractures, however it remains unreliable with only a "moderate" inter-observer reliability at group level with this falling to "fair" when sub-group classifications are made. Identification of stable and unstable injuries using the new AO/OTA system remains fraught with difficulties and appears difficult to apply with consistent accuracy.