Reproducibility of radiographic measurements for subaxial cervical spine trauma

Radiographic measurement study.

To assess the interobserver reliability of radiographic measurements for subaxial cervical spine trauma; to compare the reliability of measurements made on computed tomography (CT) versus those made using plain radiographs.

Despite the importance of imaging and classification of subaxial cervical injuries, the persistent lack of a uniform measurement protocol remains an obstacle. Recently, the Spine Trauma Study Group published a proposal for a standardized set of measurement techniques for the radiographic evaluation of subaxial cervical spine trauma. While a worthwhile venture, the observer error of these methods was not tested.

Lateral cervical plain radiographs and CT images of 30 patients who sustained a broad spectrum of subaxial cervical spine injuries were distributed to surgeons. Participants were asked to measure kyphosis, translation, vertebral body height loss, and facet joint apposition. Each rater was provided with a pictorial diagram illustrating the prescribed measurement technique. All measurements were made using plain radiographs and CT images with the exception of facet joint apposition, which was assessed using only CT. Reliability was examined by calculating the ICC and Pearson correlation coefficients. RESULTS.: Vertebral body translation was the most reproducible method on both CT images and plain radiographs. Kyphosis measurements were less reproducible, though the endplate method demonstrated superior reliability to the posterior tangent method. Plain radiographic measurement of anterior vertebral body height loss demonstrated moderate reliability while all other height loss measurements were found to show poor reliability. Facet joint apposition measurement demonstrated poor reproducibility.

Despite a consensus regarding their importance in directing treatment, radiographic measurements for subaxial cervical spine trauma demonstrate inconsistent reliability. Even in the idealized setting used in this investigation, there was limited agreement between observers. Although translation and kyphosis showed satisfactory reproducibility, results for vertebral body height loss and facet joint apposition were unreliable. On the basis of these findings, it may be more appropriate to describe facet joint apposition binomially as "present" or "not present" instead of a numerical value; vertebral body height loss may be more appropriately characterized in quaternary terms, such as less than 25%, 25% to 50%, 50% to 75%, and more than 75%. Though simpler, such descriptions would need to be validated in future studies.