High-Dose Methylprednisolone in Acute Spinal Cord Injury: NASCIS Against a Real Infection Risk
A single patient, five hours into an incomplete cervical spinal cord injury and already showing early signs of aspiration pneumonitis. The disagreement isn't about the old NASCIS trials' own flaws — it's about whether a contested benefit still clears the bar against a harm that, in him, is no longer hypothetical.
J.P., a twenty-two-year-old man, has spent the last two summers working as a lifeguard at the same lake he was swimming in, off duty, with three friends, when he dove in headfirst the way he has probably done a hundred times before. What he didn't account for is that the lake has dropped nearly four feet since June, after a summer with almost no rain, and the water where he entered was shallower than it has been in years. He struck the bottom, and his friends pulled him out within a minute, already unable to move his legs and reporting he couldn't feel them either.
He struck the water at about one in the afternoon and was in the trauma bay inside forty minutes; it is now five hours since the dive. He is otherwise healthy — no prior spine or neurologic history, no medications — and swims competitively when he isn't on duty at the lake. CT showed a burst fracture at C5-6 with cord signal change on the MRI that followed; on exam he has some preserved pinprick and light-touch sensation below the level of injury but no volitional motor function in his legs and only trace movement in his hands — an incomplete injury, ASIA B, with real if uncertain potential for some recovery. He also swallowed and likely aspirated a meaningful volume of lake water before his friends got him out, and it is only now, five hours on, that the second problem is becoming as concrete as the first: his oxygen saturation has drifted down to 91% on room air, he's started a low-grade fever, and his admission chest X-ray, repeated an hour ago, now shows a new right lower lobe infiltrate — early aspiration pneumonitis, not yet frank pneumonia, but a real and active process, not a theoretical risk sitting in his history.
The spine team is now deciding whether to give him high-dose methylprednisolone under the old NASCIS-derived protocol, still administered at some centers within the first eight hours after acute spinal cord injury on the strength of a result now three decades old. His own clock is what complicates the question. NASCIS III sorted its patients by how long after injury the drug was started, and prescribed different courses accordingly: the 24-hour infusion for those treated inside three hours, the 48-hour infusion for those treated between three and eight. At five hours he falls in the second band. The protocol-faithful regimen for him is therefore not the shorter course most clinicians picture when they say “the NASCIS protocol” — it is the longer one, and the longer arm is the one that carried that trial's excess of severe pneumonia.
In the trauma bay, inside the eight-hour window
I want to at least put the case for it on the table before we rule it out. NASCIS II, Bracken and colleagues' 1990 trial, remains the only randomized evidence for any pharmacologic intervention improving motor recovery after acute spinal cord injury, and its subgroup finding for patients treated within eight hours is the basis nearly every subsequent discussion of this drug still centers on. He's an incomplete injury, ASIA B, five hours out, inside that window, with real preserved function that gives him something to actually protect. I'm not pretending the trial is clean. I am saying that when the alternative is offering him nothing pharmacologic at all, a contested but real signal is still worth more than silence.
I want to be honest that this position gets weaker every year the field has had to sit with NASCIS's own methodology — I'm not arguing this is a strong recommendation, only that it isn't nothing.
You're right that it's the only randomized signal we have, and I'm not going to pretend a positive trial doesn't matter just because it's contested. But the harm side of this drug isn't a subgroup finding the way the benefit is — and I want to state it precisely, because it is usually stated too loosely. In NASCIS II, Bracken's own complication data were a trend, not a result: wound infection roughly twice as common on steroid, confidence interval crossing one. It is NASCIS III, in 1997, that produced the hard number, and it is a duration effect rather than a dose effect — both arms got the identical 30-milligram-per-kilogram bolus and the identical hourly rate, and differed only in how long the infusion ran. Severe pneumonia was 5.8 percent on the 48-hour course against 2.6 percent on the 24-hour, which reached significance; severe sepsis ran 2.8 against 0.6 and did not. That is a narrower claim than "steroids cause infection," and it is the one that applies to him. He aspirated lake water five hours ago, his oxygen is already drifting down, and his chest film already shows a new infiltrate. Giving a steroid whose most reproducible harm is impaired infection control into a chest that is actively seeding an infection right now isn't weighing a contested benefit against a hypothetical risk. It's weighing a contested benefit against a risk I can already see on the film in front of me.
And the eight-hour window you're citing as the basis for treating him was never the trial's own prespecified primary analysis — every preplanned comparison in NASCIS II was negative, and the window emerged afterward from a time-stratified subgroup. That reading is not mine; it is the substance of Coleman and colleagues' critical appraisal of the NASCIS reporting and of Nesathurai's 1998 reassessment, and it is why the AANS/CNS guidelines eventually stepped back from calling this a standard of care.
I don't think either of you is wrong about your own piece of this. What I'd add is that the protocol people mean when they say they'll give it precisely is usually not the protocol this patient is actually owed. The regimen most clinicians have in mind is NASCIS II's — 30 milligrams per kilogram as a bolus, then 5.4 per kilogram per hour for 23 hours. But Bracken's 1997 follow-up, NASCIS III, assigned the course by time-to-treatment: 24 hours for patients dosed inside three hours of injury, 48 hours for those dosed between three and eight. He is at five. Faithful administration in his case means the 48-hour infusion, not the 23-hour one.
Which is where the two of you stop disagreeing, I think. The 48-hour arm is precisely the arm that carried the severe-pneumonia excess my colleague just quoted. So the protocol-fidelity argument and the infection argument are not in tension here — they converge. Giving him an approximated short course would expose him to the harm profile while abandoning the only version of the benefit claim that describes his time window; giving him the faithful long course would double the steroid exposure into a chest that already has an infiltrate. There is no version of this that is both faithful to the evidence and safe for him.
Agreed: high-dose methylprednisolone was not started. The team's judgment turned on two findings that pointed the same way — the infiltrate already visible on his chest film, and the fact that his five-hour presentation assigned him to the 48-hour course rather than the shorter one, doubling the exposure at issue. Not a rejection of the drug in every acute SCI presentation, but a conclusion that a contested benefit doesn't clear the bar when the faithful version of the regimen is also the version carrying the harm, in a patient with a currently active infectious process. Infectious disease was consulted directly for the aspiration pneumonitis, and the spine team documented the reasoning explicitly in case his neurologic exam changes over the coming days and the decision needs revisiting.
Not agreed: whether this would have gone differently without the aspiration event. The neurosurgeon believes a genuinely clean incomplete injury, without a competing infectious risk, would have been a real, defensible case for the steroid course despite its contested evidence base, and would have pushed harder for it under those circumstances — conceding that the 48-hour assignment would still have applied, and that he would have been arguing for the more exposed version of the regimen, not the milder one. The infectious disease physician isn't convinced the calculation would have changed much even then, given how consistently the harm signal has held up across the literature independent of any one patient's other risk factors. Both agreed the aspiration made today's answer easier than it would otherwise have been, without agreeing on what today's answer would have been without it.