Intraobserver and Interobserver Reliability of the Computed Tomography-Based Classifications Used in Posterior...

Three main computed tomography (CT)-based classification systems have been defined and used for posterior malleolar fractures. The intraobserver and interobserver reliability of two of these systems has never been investigated. The aim of this study was to investigate the intraobserver and interobserver reliability of the Haraguchi (HC) and Bartoníček-Rammelt (BRC) classifications.

Axial and sagittal CT sections and three-dimensional reconstruction images of 60 fractures were evaluated twice by six observers at a 6-week interval. Cohen κ values for intraobserver reliability and Fleiss κ values for interobserver reliability were calculated.

The interobserver reliability of the HC was fair in both assessments (Fleiss κ: 0.263 and 0.249 for the first and second evaluations, respectively). The interobserver reliability of the BRC was moderate in both evaluations (Fleiss κ: 0.535 and 0.447, respectively). The intraobserver reliability values (Cohen κ) of the HC were determined to be between 0.532 and 0.927 for the six observers. These values indicated moderate agreement for one observer, substantial for four, and very good for one. Intraobserver reliability values for the BRC were determined to be between Cohen κ of 0.626 and 0.884. Based on these values, the BRC's intraobserver reliability indicated substantial agreement for two observers and very good for four observers.

Intraobserver and interobserver reliability values of the BRC were higher than those of the HC.