Haloperidol for Hospital Delirium: A Reflex the Largest Trial Didn't Support
An agitated, delirious inpatient is pulling at his IV lines. The order for haloperidol is nearly automatic — the disagreement is whether the largest trial ever run on this exact practice changes what "automatic" should mean.
Robert N., 74, retired firefighter, was admitted three days ago for a urinary tract infection that has since been appropriately treated with antibiotics, and he was oriented and pleasant on admission. He still coaches a youth flag-football team on weekends and lives independently with his wife, and aside from well-controlled hypertension and a hip replacement two years ago, he has no significant medical history — certainly no prior cognitive impairment. Overnight he became acutely confused — not recognizing the nurse he'd joked with the evening before, trying repeatedly to climb out of bed convinced he was late for a shift, and pulling at his IV line hard enough that it infiltrated once already tonight. His daughter, reached by phone, says he was completely himself two days ago, sharp enough to needle her about a fantasy football trade.
The CAM screen is positive for delirium, hyperactive subtype, and the working differential covers the usual hospital-delirium contributors — the resolving infection itself, sleep disruption from overnight vitals and lab draws, a new opioid started for a hip injury sustained during a fall risk assessment two days ago, and simple disorientation from an unfamiliar environment. None of those contributors has yet been addressed tonight; the acute problem in front of the team is behavioral, not diagnostic. The night resident's reflex, watching him try to climb over the bed rail for the third time in an hour, is to order haloperidol — the most familiar tool on the floor for exactly this scenario, ordered here almost as often as it's questioned.
At the bedside, on the medical floor
I want to order haloperidol tonight, low dose. He's pulled one line already and he's trying to climb the bed rail. I'm not trying to fix his delirium with this — I'm trying to get through the next four hours without an injury or another infiltrated IV.
I want to name what the evidence actually shows before we default to that. MIND-USA — Girard and colleagues, New England Journal of Medicine, 2018, 566 delirious ICU patients randomized to haloperidol, ziprasidone, or placebo — found no significant difference between any group in days alive without delirium or coma, and no difference in 90-day survival. This is the largest, most rigorous trial ever run on this exact question, and it found the drug doesn't do what we usually assume it does.
I take the safety concern seriously, but I'd push back on reaching for haloperidol as though it's known to help here — the trial evidence says it specifically doesn't shorten the delirium itself, which is different from what's actually being asked for right now.
I think you're both right, about different things. MIND-USA is real and it means we shouldn't tell ourselves haloperidol is treating his delirium — it isn't, by the best evidence we have. But there's a real difference between that claim and tonight's actual question, which is narrower: does he need something for the next few hours to prevent an injury.
The PADIS guideline from the Society of Critical Care Medicine, 2018, is explicit that non-pharmacologic measures — reorientation, minimizing overnight disruption, a sitter if one is available — come first. Given how close we already are to an injury tonight, I'd support a single low dose as a bridge while we get a sitter arranged, with a hard stop on repeat dosing without reassessment.
Agreed: a single low dose of haloperidol given tonight as a bridge, a 1:1 sitter arranged for the remainder of the night, and the recently started opioid reviewed for alternatives. No standing or repeat antipsychotic order is placed; any further dosing requires a fresh bedside reassessment.
Not agreed: whether tonight's single dose, even bridged and time-limited, risks becoming exactly the kind of reflexive practice the pharmacologist wants to move the floor away from. The hospitalist views a clearly-bounded, safety-specific exception as compatible with that goal; the pharmacologist worries that any use, however narrowly framed tonight, makes the next night's order easier to write without the same scrutiny.