OF-Pelvis classification of osteoporotic sacral and pelvic ring fractures

Objectives: Osteoporotic fractures of the pelvis (OFP) are an increasing issue in orthopedics. Current classification systems (CS) are mostly CT-based and complex and offer only moderate to substantial inter-rater reliability (interRR) and intra-rater reliability (intraRR). MRI is thus gaining importance as a complement. This study aimed to develop a simple and reliable CT- and MRI-based CS for OFP.

Methods: A structured iterative procedure was conducted to reach a consensus among German-speaking spinal and pelvic trauma experts over 5 years. As a result, the proposed OF-Pelvis CS was developed. To assess its reliability, 28 experienced trauma and orthopedic surgeons categorized 25 anonymized cases using X-ray, CT, and MRI scans twice via online surveys. A period of 4 weeks separated the completion of the first from the second survey, and the cases were presented in an altered order. While 13 of the raters were also involved in developing the CS (developing raters (DR)), 15 user raters (UR) were not deeply involved in the development process. To assess the interRR of the OF-Pelvis categories, Fleiss’ kappa (κ F ) was calculated for each survey. The intraRR for both surveys was calculated for each rater using Kendall’s tau (τ K ). The presence of a modifier was calculated with κ F for interRR and Cohen’s kappa (κ C ) for intraRR.

Results: The OF-Pelvis consists of five subgroups and three modifiers. Instability increases from subgroups 1 (OF1) to 5 (OF5) and by a given modifier. The three modifiers can be assigned alone or in combination. In both surveys, the interRR for subgroups was substantial: κ F  = 0.764 (Survey 1) and κ F  = 0.790 (Survey 2). The interRR of the DR and UR was nearly on par (κ F Survey 1/Survey 2: DR 0.776/0.813; UR 0.748/0.766). The agreement for each of the five subgroups was also strong (κ F min.–max. Survey 1/Survey 2: 0.708–0.827/0.747–0.852). The existence of at least one modifier was rated with substantial agreement (κ F Survey 1/Survey 2: 0.646/0.629). The intraRR for subgroups showed almost perfect agreement (τ K  = 0.894, DR: τ K  = 0.901, UR: τ K  = 0.889). The modifier had an intraRR of κ C  = 0.684 (DR: κ C  = 0.723, UR: κ C  = 0.651), which is also considered substantial.

Conclusion: The OF-Pelvis is a reliable tool to categorize OFP with substantial interRR and almost perfect intraRR. The similar reliabilities between experienced DRs and URs demonstrate that the training status of the user is not important. However, it may be a reliable basis for an indication of the treatment score.