Reliability of the FFP classification in pelvic fractures

2021 
BACKGROUND Pelvic fractures are typical for frail geriatric patients. They are characterized by increasing pain and loss of mobility. As geriatric pelvic fractures differ from the typical high velocity injuries, Rommens and Hofmann recommended a new classification for fragility fractures of the pelvis (FFP) in 2013. In addition to the location of the fracture they also assessed the degree of dislocation. OBJECTIVE Compared to known fracture classifications of the pelvis, the FFP classification appears complex. Therefore, this study was designed to investigate the interobserver reliability of the FFP classification. MATERIAL AND METHODS The members of the Section of Geriatric Traumatology (DGOU) were presented with 10 DICOM data sets with fractures of the pelvis for classification. As a reference the classification of P.M. Rommens, the author of the FFP, was established. RESULTS In this study 24 consultants (47%) and 27 (53%) residents took part. Also, six radiologists were invited to participate. A total of 493 assessments were made. In 184 (37%) cases there was agreement with the reference, in 183 (37%) computed tomography images the fractures were classified lower, in 26 (26%) higher than the reference. This initially resulted in a Cohens κ coefficient of 0.36. This corresponds to a sufficient agreement (fair) according to Garbuz. With the reduction of the classification to four main groups, the Cohens κ coefficient increased to a satisfactory value. CONCLUSION The reasons for the partly low agreement could be the participants' lack of experience and an inaccurate description of the classification levels. In the end, however, the FFP classification is the only one that meaningfully represents pelvic fractures of geriatric and frail patients. By simplifying to the four main groups, a better interobserver reliability is achieved. For a successful treatment, however, attention to the individual patient and the "fracture personality" is essential.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    29
    References
    1
    Citations
    NaN
    KQI
    []