Shortcomings of Posterior Malleolus Fractures Classifications: A Independent Inter and Intra-Rater Agreement Study Was Proposed

CATEGORY: Ankle INTRODUCTION/PURPOSE: Posterior malleolus fractures (PMFs) occur up to 50% of ankle fractures. Several classification systems exist for their characterization, specially under computed tomography (CT). However, those classifications do not report or report incompletely the level of agreement and reliability. Also, there are no studies that have validated those classifications further. The aim of this study was to perform an independent agreement study for comparing the level of agreement of Haraguchi, Bartonícek/Rammelt, and Mason's classification; among physicians with different levels of orthopaedic training in the management of PMFs. METHODS: This study was designed according to the 'Guidelines for Reporting Reliability and Agreement Studies' ('GRASS'). Ninety-four CT scans of patients with ankle fractures that had PMF were included. Exclusion criteria were previous history of fracture, surgery, tumors, and infections of the ankle; and the presence of pilon fracture. PMF were assessed by six evaluators (3 foot and ankle surgeons and 3 orthopaedic surgery residents) according to Haraguchi, Bartonícek/Rammelt; and Mason 's classifications. All images were reassessed by the same evaluators in a random sequence three weeks later. The new sequence of CT images was done by one of the authors, who did not participate in the evaluation of the cases. The Kappa (k) oefficient was used to determine the intra and interobserver agreement. Statistical significance was established using p < .05, with a 95% confidence interval (CI). RESULTS: We performed a total of 564 observations for each classification. The overall interobserver agreement was moderate for all classifications, with a (k) value of 0.52 (95% CI: 0.43-0.61) for Haraguchi, 0.53 (95% CI: 0.46-0.61) for Bartonícek/Rammelt; and 0.54 (95% CI: 0.47-0.62) for Mason's classification, without significant differences between them. Meanwhile, the overall interobserver agreement was substantial for all classifications, with a (k) value of 0.70 (95% CI: 0.64-0.75) for Haraguchi, 0.73 (95% CI: 0.69-0.78) for Bartonícek/Rammelt; and 0.73 (95% CI: 0.69-0.78) for Mason's classification, without significant differences between them. No differences were found in the inter and intraobserver agreement among orthopaedic surgeons and residents. CONCLUSION: This is the first agreement study that compares the most utilized classifications for PMFs, according to 'GRRAS'. Our results showed substantial intraobserver agreement, but only moderate interobserver agreement for all classifications; limiting adequate communication in clinical and research field. These findings did not change regardless of the level of experience of the evaluators. Due to moderate level of agreement and that no classification is superior to the others, we estimate the need to generate a new classification for PMFs, that gives a higher level of agreement, as well as, orientation to the surgeon during the decision-making process.