- Design
- Post hoc Bayesian analysis of a randomised placebo-controlled trial
- Population
- 987 adult ICU patients with delirium (540 hypoactive, 447 hyperactive)
- Primary outcome
- Days alive and out of hospital at 90 days
- Effect
- Benefit in both subtypes; 76.5% probability of larger effect in hyperactive (difference 2.9 days, CrI −4.8 to 10.6)
AID-ICU randomised 987 adult ICU patients with delirium to haloperidol or placebo. This post hoc Bayesian analysis asked whether the effect differed by motor subtype: 540 had hypoactive and 447 hyperactive delirium at randomisation.
Haloperidol was associated with more days alive and out of hospital at 90 days in both subtypes, with a 76.5% probability that the effect was larger in hyperactive delirium (difference 2.9 days, 95% credible interval −4.8 to 10.6). Results were similar for days alive without delirium or coma and without ventilation. Mortality was lower with haloperidol in both groups, with a 73% probability of greater benefit in hyperactive delirium. Serious adverse reactions and rescue medication use did not differ.
Many clinicians reserve haloperidol for agitated patients. These data suggest it is not harmful, and may help, in quiet delirium too. The analysis was not prespecified, and the credible intervals for the differences between subtypes are wide.
- Haloperidol was not associated with harm in either delirium subtype in this analysis.
- Treat reversible causes — infection, drugs, hypoxia, retention, pain — before or alongside any antipsychotic.
- Consider haloperidol in hypoactive delirium as well as agitated delirium, weighing QT and extrapyramidal risk.
- Check QTc and electrolytes before and during treatment.
Why it matters
It challenges the habit of treating only the delirium that disturbs the ward.
Don't overread it
This is a post hoc subgroup analysis; the difference between subtypes is uncertain and needs confirmation.
The statistics, in plain English
Bayesian results give the probability that an effect exists rather than a p-value. A 76.5% probability of a larger effect in hyperactive delirium is suggestive, not conclusive, and the credible interval crossing zero shows the subtypes could also respond equally.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for emergency & critical care, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free