Methylprednisolone for the Treatment of Patients with Acute Spinal Cord Injuries: Response

Dear Editor,

We are grateful for Dr. Edward D. Hall's comments and are sensitive to his concerns regarding the possible role of methylprednisolone in treating acute traumatic spinal cord injury (SCI); however, we believe that our findings have been objectively reported and support our conclusions. ¹

Both Dr. R. John Hurlbert and Dr. Michael Fehlings have published previously on the subject of methylprednisolone in SCI, both have expertise in basic science and clinical SCI research, both contribute actively to the Rick Hansen Spinal Cord Injury Registry (RHSCIR), and both are in agreement with respect to the methodology, results, interpretations, and conclusions voiced in our article. It may be pertinent to disclose for Dr. Hall that Dr. Fehlings was a principal investigator for the Third National Acute Spinal Cord Injury Study (NASCIS-III) trial whereas Dr. Hurlbert has no ties to the NASCIS studies nor any other vested interests in methylprednisolone.

We agree that our finding of a decrease in the frequency of NASCIS-II methylprednisolone administration over time suggests that the NASCIS-II protocol has generally been abandoned in Canadian trauma centers. Our finding that NASCIS-II methylprednisolone started within 8 h of injury did not improve motor score recovery, and was associated with a higher rate of total complications, validates this trend and supports current guideline recommendations against its routine administration. Our nonsignificant results for individual complication types (e.g., pneumonia) in the two groups may have been related to the small sample size, as expected for this type of study, but the difference in total number of complications between groups was statistically significant.

We also agree that differences in the timing of patients' baseline neurological examinations and differences between the centers at which they were treated could have introduced confounding; therefore, we used propensity-score matching and performed sensitivity analyses with negative binomial regression to adjust for potential imbalances. As we reported in Table 2, there were no significant differences in any baseline characteristics between the two groups after matching, including time from injury to baseline neurological examination. We did not match on the basis of RHSCIR site, because most of the patients who received NASCIS-II methylprednisolone did so at non-RHSCIR community hospitals prior to being transferred to RHSCIR sites; however, we adjusted for RHSCIR site in an analysis that compared the NASCIS-II methylprednisolone group against the full cohort of unmatched potential controls, and our results did not change.

Further, it is well known that many variables can affect neurological improvement after injury, including the neurological level and baseline severity of injury (American Spinal Injury Association Impairment Scale [AIS] A–E). Although some studies in this field match groups based on level or severity of injury, very few consider the effect of their joint distribution (for example, AIS A for C1–4). We previously demonstrated that even distinguishing between high cervical (C1–C4) and low cervical injuries (C5–T1) within each AIS level is crucial for tracking motor score improvement, because spontaneous recovery can vary based on this joint distribution of anatomical region and severity of injury. ² If this heterogeneity is not accounted for when two groups are compared, differences in spontaneous motor recovery may incorrectly be attributed to the effect of an intervention when it may be the result of an imbalance in the amount of spontaneous recovery between the groups. ²

We did not investigate the extent to which those clinicians who administered methylprednisolone in our study did so because they “viewed it as a last resort,” but earlier survey data suggest that other reasons may be more likely. ³ We also did not collect data to explore the extent to which methylprednisolone begun within the first 3 h after injury would have been just as ineffective as if it had been delayed until between 3 and 8 h; however, we did prospectively verify that the patients who received methylprednisolone in our study did so within 8 h of their injuries as per the NASCIS-II protocol. We discussed the results of NASCIS-III on page 1680, paragraph 6, and the 8 h subgroup of NASCIS-II on page 1681, paragraph 1.

Finally, we acknowledged that our small sample size may limit confidence in our findings, and we noted for comparison that the analyses of NASCIS-II showing motor score improvements relied on only 65 patients who received methylprednisolone within 8 h. We therefore share Dr. Hall's concern that decision makers would be misguided if they only applied the results of individual studies to the care of patients with SCI. Rather, we suggest that clinicians, researchers, and other evidence users should consider the totality of the evidence in the context of a methodologically rigorous systematic review and meta-analysis. According to pooled data from four randomized controlled trials and 17 observational studies, methylprednisolone is not associated with a significant long-term benefit in patients with acute traumatic SCI, and may be associated with an increased risk for gastrointestinal bleeding. ⁴