The Comprehensive AO CMF Classification System for Mandibular Fractures: A Multicenter Validation Study

The AO CMF has recently launched the first comprehensive classification system for craniomaxillofacial (CMF) fractures. The AO CMF classification system uses a hierarchical framework with three levels of growing complexity (levels 1, 2, and 3). Level 1 of the system identifies the presence of fractures in four anatomic areas (mandible, midface, skull base, and cranial vault). Level 2 variables describe the location of the fractures within those defined areas. Level 3 variables describe details of fracture morphology such as fragmentation, displacement, and dislocation. This multiplanar radiographic image-based AO CMF trauma classification system is constantly evolving and beginning to enter worldwide application. A validation of the system is mandatory prior to a reliable communication and data processing in clinical and research environments. This interobserver reliability and accuracy study is aiming to validate the three current modules of the AO CMF classification system for mandible trauma in adults. To assess the performance of the system at the different precision levels, it focuses on the fracture location within the mandibular regions and condylar process subregions as core components giving only secondary attention to morphologic variables. A total of 15 subjects individually assigned the location and features of mandibular fractures in 200 CT scans using the AO CMF classification system. The results of these ratings were then statistically evaluated for interobserver reliability by Fleiss' kappa and accuracy by percentage agreement with an experienced reference assessor. The scores were used to determine if the variables of levels 2 and 3 were appropriate tools for valid classification. Interobserver reliability and accuracy were compared by hierarchy of variables (level 2 vs. level 3), by anatomical region and subregion, and by assessor experience level using Kruskal-Wallis and Wilcoxon's rank-sum tests. The AO CMF classification system was determined to be reliable and accurate for classifying mandibular fractures for most levels 2 and 3 variables. Level 2 variables had significantly higher interobserver reliability than level 3 variables (median kappa: 0.69 vs. 0.59, p  < 0.001) as well as higher accuracy (median agreement: 94 vs. 91%, p  < 0.001). Accuracy was adequate for most variables, but lower reliability was observed for condylar head fractures, fragmentation of condylar neck fractures, displacement types and direction of the condylar process overall, as well as the condylar neck and base fractures. Assessors with more clinical experience demonstrated higher reliability (median kappa high experience 0.66 vs. medium 0.59 vs. low 0.48, p  < 0.001). Assessors with experience using the classification software also had higher reliability than their less experienced counterparts (median kappa: 0.76 vs. 0.57, p  < 0.001). At present, the AO CMF classification system for mandibular fractures is suited for both clinical and research settings for level 2 variables. Accuracy and reliability decrease for level 3 variables specifically concerning fractures and displacement of condylar process fractures. This will require further investigation into why these fractures were characterized unreliably, which would guide modifications of the system and future instructions for its usage.