Palliative Sedation for Refractory Dyspnea and Terminal Agitation
A dying man's breathlessness and agitation have stopped responding to standard comfort measures, and the question is no longer whether to sedate him but which drug gets there reliably in someone whose body has spent a decade building tolerance to the first, most obvious choice.
D.F. has lived alone since his wife died six years ago, in the same apartment above the hardware store he ran for three decades before COPD forced him to sell it. A home health aide had been coming twice a day, then more, in the months before he was admitted to the inpatient hospice unit four days ago, breathless at rest despite maximal bronchodilators, opioids titrated past the point of meaningful relief, and oxygen that no longer touches the sensation of air hunger driving his agitation. He has been on clonazepam for restless-leg symptoms for eleven years, a detail the team almost treated as incidental until it started determining how this decision actually plays out.
Refractory symptoms at the true end of life meet a threshold for sedation that is deliberately narrow, and Cherny and Radbruch's EAPC framework states it in roughly those terms: disease-directed and standard symptom-directed therapy exhausted, a prognosis measured in days, and distress severe enough that the doctrine of double effect genuinely applies — the intent is relief, not death, even though the sedating dose required may hasten it. D.F. meets each of those conditions on today's chart, which is why nobody in the room is arguing about whether to sedate him. Midazolam is the near-universal first-line agent for exactly this threshold, titratable and familiar to every hospice team in the country. But eleven years of continuous clonazepam has left D.F.'s GABA-A receptors adapted to a baseline level of agonism that a naive receptor population never sees — chronic exposure shifts subunit composition and blunts the response to any fresh benzodiazepine at a given dose. That is real, measurable, receptor-level tolerance, and it means a midazolam titration that would settle most patients within an hour may run considerably longer, and to a considerably higher dose, in him before it reaches the same effect.
His aide, who has managed his clonazepam refills for years and knows his home dosing history better than the admission chart did, was the one who first flagged the length on it when the team's initial titration plan assumed a treatment-naive response. What that history does not do, and what the team has not yet said out loud, is make the next drug safer. Every agent that could reach him from here works at or near the same chloride channel his receptors have spent eleven years adapting around, which means the dose that finally settles him will be arrived at by climbing rather than by calculation — and on a unit without continuous monitoring, the margin between the dose that works and the dose that does more than intended is the thing nobody can size in advance.
At the bedside, after the third failed titration
Two midazolam boluses in ninety minutes with no meaningful change in his RASS score is exactly the pattern I'd expect given eleven years of chronic benzodiazepine exposure — his GABA-A receptors have already adapted to a baseline level of agonism that clonazepam has been supplying continuously. I want to escalate the midazolam infusion rather than switch drugs; we haven't actually exhausted the class yet, we've under-dosed for his tolerance.
I'd push back on escalating within the same class before switching. Continuing to raise a midazolam infusion against demonstrated benzodiazepine tolerance risks stacking doses faster than we can titrate safely on an inpatient hospice unit without the monitoring an ICU would have — and the EAPC framework's own guidance is explicit that proportionality is judged against the monitoring actually available, not the monitoring we'd like. Propofol works through a different, additive mechanism — direct GABA-A chloride channel effects alongside glutamate modulation — and its onset and offset are measured in minutes, which matters when we're titrating against a moving target like his tolerance.
The doctrine of double effect doesn't change with the drug chosen — the intent stays relief of refractory suffering either way — but propofol does require continuous infusion and closer monitoring than midazolam, which is a real logistics question for an inpatient hospice unit, not just a pharmacology one.
Both of those points are right, and I don't think they're actually in conflict. Phenobarbital sits between them — it acts through prolonging GABA-A chloride channel opening rather than just potentiating it the way midazolam does, so cross-tolerance with his chronic clonazepam is real but partial, not complete, and it doesn't require the same continuous-infusion monitoring propofol does. Given this unit can manage a phenobarbital loading dose safely, I'd try that step before committing to propofol, and keep propofol in reserve exactly where the anesthesiologist wants it — as the next step if phenobarbital doesn't reach him either.
Agreed: phenobarbital loading dose given, followed by a titrated infusion, with a target RASS of -2 to -3 reassessed hourly. His niece was told directly that a phenobarbital titration in someone with his tolerance history could still take longer than a typical patient, and that this was expected, not a sign anything had gone wrong.
Not raised as a disagreement, but recorded as a standing contingency: if the phenobarbital infusion does not bring his RASS to target within four hours, propofol starts next, with the closer monitoring that requires already arranged with the unit's charge nurse rather than negotiated in the moment it's needed.