Since the reproducibility of the Schatzker and AO tibial plateau fracture classification systems has already been assessed, the goal of this study was to evaluate the Duparc classification system and compare it to the other two.

CT scan is better than X-rays for analyzing and classifying tibial plateau fractures. The Duparc classification system is more effective than the other two systems but could be improved by adding elements of each.

Six observers analyzed images from 50 fractures and then classified them. Each fracture was evaluated on X-rays. Two weeks later, these same fractures were evaluated on X-rays and CT scans. The same process was repeated four weeks later. The Kappa coefficient (κ) was used to measure agreement and contingency tables were built.

The interobserver reproducibility for the X-ray analysis was poor for the Duparc and AO classifications (κ_Duparc=0.365; κ_AO=0.357) and average for the Schatzker classification (κ_Schatzker=0.404). The reproducibility was improved overall when CT scans were also analyzed (κ_Duparc=0.474; κ_AO=0.479; κ_Schatzker=0.476). A significantly greater number of fractures could not be classified in the Schatzker system than in the others (14.3% versus 2% for Duparc and 7.33% for AO). Review of the contingency tables revealed that the Schatzker and AO classification systems did not take certain fracture types into account. Seventy-one percent (71%) of the lateral unicondylar split fractures were found to be combined fractures when CT scan analysis was added.

Our results showed CT scan to be better at analyzing and classifying fractures. We also found the Duparc classification to be advantageous because it allowed more fractures to be classified than in other classification systems, while having similar reproducibility. Based on our study findings, the Duparc classification was revised by adding elements of the other two. We propose using the modified Duparc classification system to analyze tibial plateau fractures going forward.

Level IV. Retrospective study.