Sedating mechanically ventilated patients with the inhaled anesthetic isoflurane instead of standard intravenous drugs did not reduce their chance of dying in hospital, according to a randomized trial of 780 adults with acute hypoxemic respiratory failure published Oct. 14 in JAMA. Hospital mortality was 42.5% with inhaled sedation and 43.8% with intravenous sedation, and patients in the inhaled group needed more additional treatment for persistent low oxygen and spent fewer days free of delirium and coma.
The SAVE-ICU trial was a pragmatic, open-label, superiority trial run in 14 Canadian intensive care units by the SAVE-ICU Investigators and the Canadian Critical Care Trials Group, with Angela Jerath of Sunnybrook Research Institute and the University of Toronto as first author. Between September 2020 and March 2025 it randomized 786 adults who were on a ventilator for acute hypoxemic respiratory failure, either from COVID-19 or from other causes with a ratio of arterial to inspired oxygen below 300 mm Hg, and who had been sedated for less than 72 hours. The primary analysis included 780 patients, 385 assigned to inhaled sedation and 395 to usual intravenous sedation. Their mean age was 61.1 years, 34.0% were women and 73.7% had non-COVID respiratory failure. The peer-reviewed paper reports the trial registered as NCT04415060.
In the inhaled arm, isoflurane was given through a vaporizing device such as an anesthetic conserving device, with intravenous sedatives added when clinically needed; on any given day in the first two weeks, between 49.9% and 55.4% of inhaled-group patients also received intravenous sedation. Both groups were titrated to a sedation target until extubation or death. All-cause hospital mortality, the primary outcome, was 161 of 379 (42.5%) with inhaled sedation versus 171 of 390 (43.8%) with intravenous sedation, a risk ratio of 0.98 (95% confidence interval 0.84 to 1.14; p=0.85). The result did not differ by age, sex or COVID-19 status.
Days free of the ventilator, the ICU and the hospital were similar between groups, as were quality-of-life and disability scores at three months. Two secondary outcomes favored the intravenous strategy: 63.9% of inhaled-group patients received adjunctive therapies for refractory hypoxemia versus 55.0% of controls, a difference of 8.89 percentage points (1.95 to 15.83), and the inhaled group had fewer delirium- and coma-free days by day 14 (difference −0.95; −1.69 to −0.22). “Among mechanically ventilated patients with acute hypoxemic respiratory failure, an inhaled sedation strategy with predominantly isoflurane did not reduce hospital mortality compared with usual intravenous sedation,” the authors write in the paper’s key points.
The trial was funded by the Canadian Institutes of Health Research (grant 447612), the Government of Ontario, the Sunnybrook Medical Services Alternative Funding Plan Association and the London Health Sciences Foundation, according to the published protocol in BMJ Open, which states the funders had no role in design, conduct or reporting. The protocol planned for 758 patients to detect a 10-percentage-point absolute reduction in mortality from a baseline of 40%, with 80% power.
Why it matters
Inhaled anesthetics have attracted interest as ICU sedatives because they relax airway smooth muscle, may have anti-inflammatory and lung-protective effects seen in animal studies and a small human trial, are eliminated by exhalation rather than through the liver or kidneys, and may allow faster awakening, the protocol explains. Shortages of intravenous sedatives during the COVID-19 pandemic added a practical reason to look for alternatives. Conventional intravenous sedatives carry their own burdens: in a JAMA editorial, Matthieu Jabaudon of CHU Clermont-Ferrand in France notes that benzodiazepines are linked to delirium and coma, propofol can cause hemodynamic instability or infusion syndrome, and dexmedetomidine may cause a slow heart rate or insufficient deep sedation.
SAVE-ICU is the second large trial in two years to find no advantage for inhaled sedation. The SESAR trial, led by Jabaudon and published in JAMA in March 2025, randomized 687 patients with early, more severe acute respiratory distress syndrome in 37 French ICUs to sevoflurane or propofol; sevoflurane produced fewer ventilator-free days and lower 90-day survival (47.1% versus 55.7%; hazard ratio 1.31, 1.05 to 1.62). In an accompanying editor’s note, Natalie McEvoy of the Royal College of Surgeons in Ireland and JAMA associate editor Christopher Seymour frame the findings as new evidence for debate rather than a settled answer.
The caveats
The trial was open label, so clinicians knew which strategy each patient received, which the editors note can introduce bias. Roughly half of the inhaled-group patients also received intravenous sedatives on any given day, so the trial tested a pragmatic strategy rather than isoflurane alone. The study was powered to detect a 10-point absolute reduction in mortality, so smaller benefits or harms cannot be excluded. The mortality denominators, 379 and 390, are slightly smaller than the analysis population, and the abstract does not explain the gap. The paper’s disclosures list grants to co-author Claudio Martin from the Province of Ontario’s COVID-19 Rapid Research Fund and to another co-author from Canadian federal and Quebec research agencies; the protocol declared no competing interests.
What to watch
How critical care societies weigh two negative trials in future sedation guidance, and whether the protocol’s suggestion that inhaled agents could help patients with high or complex sedation needs, such as those with burns or substance use, is tested in a dedicated trial.
This article is for general information and does not replace advice from a qualified clinician. Corrections: email the editor using the address on the Contact page.
