The Last Tier: Deciding to Start a Barbiturate Coma
A college student's pressures are still climbing after decompressive surgery and every drug tier below this one. Pentobarbital can flatten the pressure, but the evidence behind it has never promised it will save him.
C.T., a 19-year-old college sophomore, was riding home from a study group when the car he was in was struck by a driver running a red light. He underwent a bifrontal decompressive craniectomy within hours of arrival for pressures that hadn't responded to the standard first tier — sedation, osmotic therapy, CSF drainage — and for a day it worked. Now, forty-eight hours in, his pressures are climbing again despite every one of those same measures optimized to their ceiling, sustained in the high 20s and low 30s on a skull that already has room to expand. He's already on norepinephrine just to keep his cerebral perfusion pressure above 60, which means the next tier of treatment has to work without a hemodynamic margin left to spend carelessly.
Pentobarbital coma is the drug at that next tier, and its evidence base is worth stating plainly rather than assuming: Eisenberg and colleagues' 1988 randomized trial found high-dose barbiturate therapy did lower refractory ICP in patients who responded to it, but a subsequent Cochrane review by Roberts and Sydenham, pooling the available randomized data, found no overall mortality or functional-outcome benefit from barbiturate therapy in severe TBI — the ICP control is real, the survival benefit has never been shown, and hypotension requiring vasopressor support followed roughly one in four patients in the pooled trials, on top of the vasopressor he is already running. Whatever the decision, it has to be made honestly as an ICP-control measure with an uncertain outcome benefit, not as a step assumed to save his life because it lowers a number reliably.
His parents, who drove through the night from three states away, have been told plainly by the team that a craniectomy this extensive combined with pressures still this high carries real uncertainty regardless of what happens next — a conversation held before this specific drug decision, not contingent on it, so that whatever tonight's outcome turns out to be, the family isn't hearing about the severity of his injury for the first time through the lens of a single treatment choice.
The tier nobody wants to reach
We've already made the biggest physical concession available — half his skull is off. Pressures are still climbing through every drug tier below this one. Pentobarbital reliably lowers ICP in patients who respond, and uncontrolled refractory hypertension carries its own mortality that isn't zero either. I don't think withholding the next tier because its outcome data is uncertain is actually the more conservative choice here.
I take the exhausted-tier logic seriously, but I want us honest about what Roberts and Sydenham's review actually found: real ICP control, no demonstrated mortality or functional benefit, and hypotension in roughly a quarter of pooled patients. He's already on norepinephrine to hold his CPP above 60. Adding a drug whose main systemic effect is vasodilation on top of that isn't a small additional cost — it could push him into the exact secondary-ischemia pattern we're trying to prevent.
Then let's not treat this as start-or-don't. Start pentobarbital with an explicit threshold set in advance — if his norepinephrine requirement climbs past a level we agree on now, or his CPP can't be held above 60 despite that support, we stop the trial and reassess rather than pushing forward on momentum. That gives the neurosurgeon's rationale for trying it and the neurointensivist's caution about its real cost both a genuine answer, instead of one overriding the other.
Pentobarbital achieved burst suppression within four hours, and his ICP fell to the mid-teens and held there. His norepinephrine requirement climbed but stayed within the pre-agreed threshold, and CPP was maintained above 60 throughout.
Left explicitly unresolved, and named as such in the chart: whether this outcome reflects a real mortality benefit from the intervention or simply the successful management of its known hypotension risk in a patient who happened to respond — the neurointensivist's reading of the evidence was never proven wrong, only not tested against a worse trajectory. The pre-agreed threshold was credited by the team as the reason the disagreement didn't have to be resolved in the moment.