Source: 8 trials including 3 high quality and 5 low quality, listed below
Efficacy Endpoints: Motor function improvement and mortality
Harm Endpoints: Infectious complications, gastrointestinal complications, and mortality
Narrative: The pharmacologic treatment of acute spinal cord injury (ASCI) has been a matter of contention for over 20 years.1 Studies in animal models suggest steroids may improve neurologic recovery2, 3 and beginning in the 1980s studies were performed testing the theory in humans.4, 5, 6, 7 Despite what most considered to be weak evidence of potential benefit and seemingly clearer signals of harm,6, 7, 8 and perhaps in part because of a dearth of treatment options, there was soon widespread adoption of high-dose methylprednisolone (MP) for the treatment of ASCI.9
The Cochrane Collaboration published an updated systematic review of the topic in 2012.10 Dr. Michael Bracken, the sole author, is a long time advocate of MP for ASCI and was principle investigator of the first large scale trial to assert a benefit with MP.4 In the review the effects of four interventions were analyzed, however the emphasis of our summary is the first of these (high-dose MP versus placebo or none), which was evaluated in multiple trials and is the focus of contemporary debate. Outcomes in the MP versus no MP or placebo groups were measured based on the change in neurologic scores between initial presentation and follow-up evaluation. Motor strength was classified from 0-5 at each of 14 muscles (10 in Petitjean) with a maximum score of 70 (100 in Petitjean). Pin prick and light touch were evaluated on a 3 point scale at 29 sensory levels (28 in Petitjean), for a maximum score of 87 (112 in Petitjean).
No difference was seen between groups overall in motor function or functional improvement. When restricted to patients who received MP within 8 hours there was a small but statistically significant improvement in motor function score (16.0 vs 11.2, p=0.033). No significant difference was found in mortality, long term neurologic function, wound infection, or gastrointestinal complications. Pulmonary complications, identified in later trials as a problem in groups receiving steroids, were not reported.
The remaining interventions were assessed in a single trial each, without placebo controls, and are noted below. Two studies discussed in the Cochrane review, one involving whiplash and the other lumbar disc herniation, are not addressed here due to their unclear relevance to ASCI.
Caveats: The author’s conclusion that "methylprednisolone sodium has been shown to enhance sustained neurologic recovery" is not, in our view, supported by these data, and in fact, no trial has demonstrated benefit for this as a primary outcome. This statement is instead derived from a post-hoc analysis using what we believe are arbitrary time cut-offs (both for time-to-administration of MP and for follow-up endpoint evaluation). In addition, the reported difference may not represent a clinically meaningful benefit, as a five point improvement based on a 0-70 scale, and spread over 14 muscles, seems unlikely to impact function.11 Moreover, there appear to be consistent findings of harm across studies, both dose-dependent and otherwise, that lead us to believe that infectious and gastrointestinal complications are very likely increased by steroids in the doses used.
This belief informed our color recommendation of Black—that the harms of high dose MP likely outweigh the benefits. We did, however, have a spirited discussion with the editorial board regarding the ‘Red’ vs ‘Black designation for this intervention based on the validity issues listed below in regard to studies documenting harms. Ultimately, we believe at TheNNT that randomized trials typically focus most on potential benefit in the form of primary outcomes, and are much less diligent and reliable when it comes to harms. This has created a systematic bias in the medical literature, and made harm-benefit balance calculations fraught,12, 13, 14 requiring attempts at systematic correction that continue to be, at best, weak. Therefore we believe it is most likely that steroids have real and consistent harms that, virtually however small, amount to greater harm than benefit for patients with spinal cord injuries.
To further examine the evidence for MP in ASCI, we believe the six prospective, randomized trials for ASCI should be evaluated separately and we have therefore listed them below with trial characteristics. None of these trials showed a treatment effect of MP on their primary outcomes, rather only in post-hoc analyses of multiple subgroups. The finding of a possible benefit in a subgroup from NASCIS II (steroids within 8 hours), was subsequently tested by Otani and Petitjean, neither of whom were able to replicate a finding of clinically meaningful benefit. Moreover, global improvement scores and functional improvement scores were not altered by MP.4, 6, 8 In light of increased infectious complications shown in the NASCIS I7 and Matsumoto et al15 trials, as well as multiple chart review studies,16, 17 it appears unlikely that the benefits of high-dose MP outweigh the risks. While the studies by Chikuda and Suberviola have all the attendant weaknesses of most chart reviews, and the Suberviola study was quite small, the documented harms were clear and consistent with those found in prospective trials. We therefore have assigned this intervention a rating of Black (harms>benefits), disagree with the Cochrane conclusions, and agree with the Congress of Neurological Surgeons, the American Association of Neurological Surgeons1, the Consortium for Spinal Cord Medicine18, the Canadian Association of Emergency Physicians19 and the American Academy of Emergency Medicine20, in recommending that "MP should not be routinely used in the treatment of patients with ASCI." 1
NASCIS, 19847, 21
• Population: Age > 13 years, treatment within 48 hours of injury
• N= 330
• Treatment: high-dose MP (1000mg bolus, then 250mg Q6 hours for 10 days) versus low-dose MP (100mg bolus, then 25mg Q6 hours for 10 days)
• Outcome: no benefits found
• Comment: stopped early due to high fatality rate, wound infections three times more likely in high-dose (9.3% versus 2.6%, NNH 17; 95% CI 11-328).
NASCIS II, 19904, 5
• Population: Age>13 years, within 12 hours of injury
• Treatment: high-dose MP (30mg/kg bolus, then 5.4mg/kg over 23 hours) versus naloxone versus placebo,
• Outcome: small benefit in sensation at 6 months, lost at 1 year; subgroup <8 hrs showed increased neurological score improvement compared with placebo (16.0 in treatment group versus 11.2 in placebo, p=0.033)
• Comment: <8 hrs from injury subgroup excluded 62% of subjects, <8 hr placebo group more severely injured than the >8hr placebo group 22, 23
• Population: Hospitalized patients in Japan age 11-65, within 8 hours
• Treatment: high-dose MP (same as NASCIS II) versus placebo,
• Outcome: no difference in global improvement or motor function, statistically significant improvement in pin-prick and touch sensation
• Comments: Not intention-to-treat, 41/158 patients excluded after randomization, not blinded
NASCIS III, 199724, 25
• Population: Patients age >14 able to receive MP within 8 hours
• Treatment: all received initial bolus of high-dose MP (30mg/kg); high-dose MP for 24 hours (same as NASCIS II) versus 48 hours versus lipid peroxidation inhibitor for 48 hours
• Outcome: no change in functional improvement scores or motor scores. Subgroup <8 hours showed improved motor function score in 48 hour group
• Comment: Severe pneumonia increased in 48 hour MP group (NNH 26)
Petitjean, 19988, 26
• Population: Hospitalized patients <8 hours
• Treatment: high-dose MP versus nimodipine versus both versus placebo
• Outcome: no difference at 1 year
• Comment: trend towards increased infectious complications in ICU patients (66% v 45%) in MP group. Republished in English with author change.
• Population: Hospitalized patients <8 hours of injury
• Treatment: high-dose MP versus placebo
• Outcome: complications, trend towards increased complications in MP group, increase in pulmonary (pneumonia and atelectasis) (34.8% versus 4.3%, NNH 4; 95%CI 3-31) and GI complications (bleeding gastric ulcers and ileus) (17.4% versus 0%, NNH 6; 95%CI 6-infinite)
Author: Joseph W Watkins MD, Christopher R Carpenter MD, MSc
Published/Updated: May 7, 2015
The title bar is color-coded with our overall recommendation.
If you have suggestions, requests, or questions about a particular NNT review, please send us a message and we’ll try to address it as soon as possible.