Stuart A Aitken1, Ian Johnston2, Angus C Jennings2, Ivan T H Chua2, Richard E Buckley2. 1. The Department of Orthopaedic Trauma, Foothills Medical Centre, Calgary, Alberta, Canada. Electronic address: saaitken@me.com. 2. The Department of Orthopaedic Trauma, Foothills Medical Centre, Calgary, Alberta, Canada.
Abstract
BACKGROUND: Despite its use in the literature, the application of the Herscovici classification system for medial malleolus fractures has not been evaluated. METHODS: We aimed to determine the reliability and accuracy of the Herscovici classification. The blinded radiographs of 130 patients were independently classified by four orthopaedic trauma surgeons. We held a consensus meeting where observers agreed on a final classification and this served as our reference standard. We used weighted kappa (κ) coefficients of agreement. RESULTS: Twenty-four fractures (18%) were deemed unclassifiable. The classification system demonstrated moderate inter-observer reliability (κ=0.54, 95% CI 0.40-0.68) but substantial reproducibility (κ=0.64, 95% CI 0.51-0.79). Accuracy, when compared with the reference standard, was κ=0.54 (95% CI 0.40-0.66). CONCLUSIONS: The obliquity of the fracture line, and fracture extension, created difficulty in classification in 26% of cases. 18% of our cases could not be classified by majority decision. Our results emphasise the challenges faced in classifying these fractures. Future work should focus on refining the Herscovici classification.
BACKGROUND: Despite its use in the literature, the application of the Herscovici classification system for medial malleolus fractures has not been evaluated. METHODS: We aimed to determine the reliability and accuracy of the Herscovici classification. The blinded radiographs of 130 patients were independently classified by four orthopaedic trauma surgeons. We held a consensus meeting where observers agreed on a final classification and this served as our reference standard. We used weighted kappa (κ) coefficients of agreement. RESULTS: Twenty-four fractures (18%) were deemed unclassifiable. The classification system demonstrated moderate inter-observer reliability (κ=0.54, 95% CI 0.40-0.68) but substantial reproducibility (κ=0.64, 95% CI 0.51-0.79). Accuracy, when compared with the reference standard, was κ=0.54 (95% CI 0.40-0.66). CONCLUSIONS: The obliquity of the fracture line, and fracture extension, created difficulty in classification in 26% of cases. 18% of our cases could not be classified by majority decision. Our results emphasise the challenges faced in classifying these fractures. Future work should focus on refining the Herscovici classification.