PURPOSE: In periprosthetic femoral fractures, our hypothesis was that when the bone and implant are stable, the fracture does not occur at the biologic or cement fixation regions but occurs at the no fixation region. The aim of this study was to investigate the validity of our new classification for periprosthetic femoral fractures and compare reliability of radiographic evaluation for implant stability between our classification and the Vancouver classification. PATIENTS AND METHODS: Sixty-six patients with periprosthetic femoral fracture were operatively treated by us between 2005 and 2013. We investigated the sensitivity and specificity of our new classification with actual implant stability. Twenty patients were randomly selected from 66 patients. After fully explaining the Vancouver and our new classification to four orthopaedic surgeons, plain radiographs acquired in two directions at the time of injury were presented, and the interobserver reliability based on the two classifications and accuracy rates of stem stability were investigated. RESULTS: The specificity of the new classification was 89%, and sensitivity was 94%. The positive and negative predictive values were 84% and 96%, respectively. Interobserver agreement was separately assessed among all possible pairs of orthopaedic surgeons. The κ values for the Vancouver and the new classification were 0.36 (0.19-0.49) and 0.76 (0.66-1.0), respectively. CONCLUSIONS: Our classification is based on a completely new concept and was prepared to overcome periprosthetic femoral fracture failures by objective evaluation. We believe this new classification is useful to establish a therapeutic strategy for femoral fractures around the stem.
PURPOSE: In periprosthetic femoral fractures, our hypothesis was that when the bone and implant are stable, the fracture does not occur at the biologic or cement fixation regions but occurs at the no fixation region. The aim of this study was to investigate the validity of our new classification for periprosthetic femoral fractures and compare reliability of radiographic evaluation for implant stability between our classification and the Vancouver classification. PATIENTS AND METHODS: Sixty-six patients with periprosthetic femoral fracture were operatively treated by us between 2005 and 2013. We investigated the sensitivity and specificity of our new classification with actual implant stability. Twenty patients were randomly selected from 66 patients. After fully explaining the Vancouver and our new classification to four orthopaedic surgeons, plain radiographs acquired in two directions at the time of injury were presented, and the interobserver reliability based on the two classifications and accuracy rates of stem stability were investigated. RESULTS: The specificity of the new classification was 89%, and sensitivity was 94%. The positive and negative predictive values were 84% and 96%, respectively. Interobserver agreement was separately assessed among all possible pairs of orthopaedic surgeons. The κ values for the Vancouver and the new classification were 0.36 (0.19-0.49) and 0.76 (0.66-1.0), respectively. CONCLUSIONS: Our classification is based on a completely new concept and was prepared to overcome periprosthetic femoral fracture failures by objective evaluation. We believe this new classification is useful to establish a therapeutic strategy for femoral fractures around the stem.