This is an interesting conclusion to draw from an analysis of, essentially, only negative studies. NASCIS 1 (1984) was statistically negative - but was discounted because the dosing was possibly too low. NASCIS 2 (1990) was also statistically negative, except for pinprick and light touch at six months, which disappeared at one year. The supposed positive outcome comes from a post-hoc analysis in which the patients who received their steroids between 3 and 8 hours after injury shook out to have a statistically significant improvements in motor score at six months and one year. However, post-hoc subgroup analysis cannot be considered practice-changing evidence until confirmed in subsequent studies. Otani (1994) was statistically negative for the primary outcome, but post-hoc analysis identified greater sensory improvement in the steroid group - which therefore implies greater motor improvement in the control group, as the overall combined neurologic scores were not different. NASCIS 3 is not placebo-controlled.
There is also no mention in the Cochrane Review of adverse events - the only mention of the safety profile of high-dose steroids in the discussion section references a systematic review of high-dose steroids given to general surgical patients, both elective and trauma. This is rigorously invalid, as the correct assessment of the safety profile of an intervention should be derived from the safety outcomes of the studies included in the analysis - nearly all of which had consistent, non-significant (underpowered) trends towards increased infectious complications.
Would it surprise you to discover that the author of the 2000, 2002 and 2012 Cochrane Review articles is the same first author of NASCIS 1, 2, and 3?
for acute ."