INTRODUCTION: Tibial plateau fracture classification systems have limited interobserver reliability and new systems emerge. The purpose of this study was to compare the reliability of the Luo classification and the Schatzker classification for two-dimensional computed tomography (2DCT) and to study the effect of adding three-dimensional computed tomography (3DCT).
MATERIALS AND METHODS: Eighty-one observers, orthopedic surgeons and residents, were randomized to either 2DCT or 2D- and 3DCT evaluation of a spectrum of 15 complex tibial plateau fractures using web-based platforms in order to classify according to the Schatzker and according to Luo's Three Column classification. Reliability was calculated with the use of Siegel and Castellan's multirater kappa measure. Kappa values were interpreted according to the categorical rating by Landis and Koch.
RESULTS: Overall interobserver reliability of the Schatzker classification was significantly better compared to the Luo classification (kSchatzker=0.32 and kLuo=0.28, P=0.021), however, 'fair' for both fracture classification systems. For the Schatzker classification observers agreed significantly better on 2DCT compared to 2D- and 3DCT (k2DCT=0.37 and k2D+3DCT=0.29, P<0.001). The addition of 3DCT did not improve the overall interobserver reliability for the Luo classification as well, as kappa values were not significantly different on 2DCT and 2D- and 3DCT (k2DCT=0.31 and k2D+3DCT=0.25, P=0.096).
CONCLUSIONS: The agreement between observers was significantly better for the Schatzker classification compared to Luo's Three Column classification, however agreement was fair for both classification systems. Furthermore, the addition of 3DCT reconstructions did not improve the reliability of CT-based evaluation of tibial plateau fractures. Considering that new classification systems and 3DCT do not seem to improve agreement between surgeons, other efforts are needed that lead to more reliable diagnosis of complex tibial plateau fractures.