This is a well-designed, clinically relevant study. It raises an excellent, previously unaddressed question to which the answer seems intuitively obvious: whether neural axis level of injury or osseous level is more predictive of neurological outcome. As the authors mention, neural axis level of injury in patients with thoracolumbar trauma is typically classified on the basis of osseous vertebral level of injury. This practice may lead to misclassification of neural axis level of injury, depending on the level of conus medullaris termination. This study provides novel information about prognosis, based solely on magnetic resonance imaging findings, by correlating neural axis anatomic variation to osseous level of injury.
The study design was straightforward and well described. Prospectively gathered data were used to retrospectively analyze magnetic resonance imaging findings as a predictor of outcome. The inclusion and exclusion criteria generated a clinically relevant population while eliminating confounding factors. Therefore, clinicians can easily determine how best to apply the findings of this study to their own patients.
Assuming that the authors' methods of reading posttraumatic spine magnetic resonance images are associated with good interobserver reliability, there are several conclusions that readers may draw from this study. First, when considering an imaging-based prognosis for patients with thoracolumbar injury, predictions will be more accurate if the level of conus medullaris termination is considered along with the level of osseous injury. Second, the neuraxial level of maximal canal narrowing is predictive of initial and final motor scores. And third, confounding factors, such as the presence or absence of initial anal sensation and the morphology of the osseous injury, must always be considered when determining prognosis.
Readers should keep in mind the several limitations to the study that were discussed by the authors, including the 41% inclusion rate, the variability in patient follow-up, and the significant difference in fracture types found between groups. A limitation not in study design but in clinical relevance is that the method of determining level of neuraxial injury was solely based on magnetic resonance imaging. Therefore, the clinical applicability is limited because findings from the physical examination supersede those from magnetic resonance imaging for determination of neurological level of injury. Even in the case of a patient who is not able to cooperate with a physical examination (e.g., an obtunded patient), and the only data available for determination of prognosis is the magnetic resonance image, the study is still likely not applicable because such a patient would have been excluded from this study due to the authors' exclusion criteria.
The final caveat to readers is that the authors do not explicitly mention the potential discrepancy between the neural axis level of injury as assessed on magnetic resonance imaging and the true neural axis level of injury as based on the findings of neurological deficits during the physical examination. The neural axis level of injury on magnetic resonance imaging is assessed according to the level that shows maximal canal narrowing, which is still a definition by level of osseous injury. The methods used in the study by Kingwell et al. are different from the methods used in prior studies in that Kingwell et al. adjust their interpretation on the basis of neural axis anatomy. Although it is reasonable to suspect that the actual neural axis level of injury is the level of maximal canal narrowing, this has not been proven. Moreover, because an unstable injury can collapse or realign between the time of injury and the time that magnetic resonance imaging is performed, the mechanics of the neural injury and level of actual neurological damage may be substantially different from the estimation that was based on the results of magnetic resonance imaging. This limitation does not preclude usefulness of the study; it is simply a concept that clinicians should keep in mind.
As the authors state, one of this study's largest contributions to the literature is its likely effect on future research involving thoracolumbar trauma. Because Kingwell et al. have shown that injury level becomes more predictive of outcomes after adjustment for neural axis anatomic variation, researchers who use their method will better avoid misclassification bias.