Design: prospective, double-blind, randomised, placebo-controlled trial at two ICUs in Melbourne
Population: 120 mechanically ventilated adults already receiving an opioid infusion for analgosedation, excluding cardiac surgical patients
Primary outcome: hourly dose of opioids in fentanyl equivalents
Effect: 64 microgram/hr (IQR 36-89) with ketamine vs 77 (47-100) with placebo; median difference -13.0 (95% credible interval -26.6 to 2.4), probability of benefit 95.1%; no difference in delirium

A double-blind trial at two Melbourne ICUs randomised 120 mechanically ventilated adults already receiving an opioid infusion to low-dose ketamine at 0.15 mg/kg/hr or placebo for the duration of ventilation. Median hourly opioid dose in fentanyl equivalents was 64 microgram/hr with ketamine and 77 microgram/hr with placebo.

Delirium — the outcome clinicians most fear when ketamine is raised — showed no difference between groups, nor did significant adverse events, ventilator-free days, ICU-free days or hospital-free days. The trial is small, and the authors are explicit that larger trials are needed for patient-centred outcomes.

What this adds is a reasonably clean safety signal for a practice that is already widespread and largely unevidenced. Low-dose ketamine as an opioid-sparing adjunct is used in many units on the strength of pharmacological reasoning; this is the randomised evidence that it does reduce opioid dose and does not obviously cause the harm people worry about.

Whether a 13 microgram/hr reduction in fentanyl equivalents is worth having is the open question. The trial cannot answer it, because it was not sized for the outcomes that would — time to extubation, opioid withdrawal, long-term opioid exposure after discharge.

Treat low-dose ketamine as a reasonable opioid-sparing adjunct in ventilated adults, not as a sedation strategy in its own right
Do not withhold it on delirium grounds alone; no difference was seen here
Record cumulative opioid dose, which is the outcome this affects and the one units rarely track
Note the exclusions: cardiac surgical patients were not studied
Watch for the larger trials before building a protocol around it

Don't overread it

The endpoint was opioid dose, not extubation, delirium duration or survival — none of which this trial was sized to detect.

The statistics, in plain English

The median difference of -13.0 microgram/hr has a 95% credible interval of -26.6 to 2.4, which crosses zero — in Bayesian terms the authors report a 95.1% probability that ketamine reduces opioid dose, meaning roughly a 1 in 20 chance the true effect is in the other direction. That is suggestive rather than conclusive, and 120 patients from 538 screened is a heavily selected population. The primary outcome is opioid dose, a surrogate: it tells you the drug does what it was given for, not that any patient was better off.