Inter- and intraobserver reliability and critical analysis of the FFP classification of osteoporotic pelvic ring injuries

INTRODUCTION: The classification and management of osteoporotic pelvic ring injuries (OPRI) remain challenging. The fragility fractures of the pelvis (FFP) classification system proposed by Rommens and Hofmann constitutes the first comprehensive classification system of OPRI and may contribute to a more thorough assessment and grading of these injuries. The reliability of the FFP classification system, however, has not been evaluated yet. The purpose of this study therefore was to determine the inter- and intraobserver reliability of the FFP classification system and to critically analyse its strengths and weaknesses.

MATERIALS AND METHODS: One-hundred pelvic CT scans obtained from a consecutive series of patients aged 70 years and older who sustained a low-energy pelvic ring injury were included in this study. Three orthopaedic traumatologists of varying levels of experience (one experienced pelvic surgeon, one consultant, one resident) and one musculoskeletal radiologist independently classified each OPRI according to the FFP classification in two different sessions. Intra- and interobserver reliability were determined using percentage agreement and Cohen's Kappa coefficients.

RESULTS: The observed relative distribution of FFP fracture types was comparable to that reported in the original study by Rommens and Hofmann. Overall interobserver reliability for all 100 cases was moderate with Kappa values from 0.42 to 0.59 (mean percentage agreement: 61% (54%-68%)), while intraobserver reliability was substantial with Kappa values from 0.68 to 0.72 (mean percentage agreement: 77% (76%-78%)). Subgroup analysis revealed lowest reliability for the classification of Type IIc, IIIc and IVb injuries (32 cases). Within this subgroup of combined anterior and posterior OPRI involving a complete nondisplaced or displaced (uni- or bilateral) sacral fracture, Kappa values for interobserver reliability ranged from 0.10 to 0.52, while those for intraobserver reliability ranged from 0.29 to 0.66.

CONCLUSION: Overall interobserver reliability of the FFP classification system was moderate, while intraobserver reliability was substantial. Despite the acceptable overall reliability, classification of FFP subtypes involving a complete nondisplaced or displaced sacral fracture showed relatively poor reliability. The latter limits the usefulness of the FFP classification for both clinical and research purposes and needs to be addressed in future studies.