Accuracy in radiographic assessment of pelvic ring fracture deformity: analysis of current methods

To our knowledge, measuring pelvic ring displacement on plain radiographs has not been scientifically evaluated. This study aims to assess the validity and reliability of existing measurement systems utilizing plain radiographs.

A Sawbones model, validated as part of this study, was fractured in a controlled fashion to simulate known pelvic ring fracture patterns. Computed tomographic scans with 2- and 3-dimensional reconstructions of intact and injury models were obtained to measure the "true" displacement of anterior and posterior disrupted surfaces. The computed tomographic data set was used to create computer-reconstructed radiographs (CRR) simulating standard anteroposterior, inlet and outlet views. Currently described measurement techniques were utilized by 8 senior orthopaedic residents and 3 fellowship-trained orthopaedic trauma surgeons on the computer-reconstructed radiograph of each model. These measurements were than assessed for interobserver reliability and validity.

Average observer measurements of anterior symphysis diastasis and vertical translation had good validity and reliability with each having an average confidence interval of 4.6 mm. The observer measurements of sacroiliac (SI) joint displacement were significantly less reliable with average confidence interval of 5.8 mm. Kappa correlation calculations for whether the SI joint was displaced more than 1 cm showed an overall agreement of 0.702 and kappa coefficient of 0.404.

Computer-reconstructed radiographic, and thus plain radiographic, assessments of deformity at the SI joint have poor reliability and accuracy. Clinicians should recognize these limitations when measuring pelvic displacement of the SI joint and cautiously interpret clinical results of treatment of SI joint injuries where displacement is measured on plain radiographs.