Autonomous oxygen titration increased the proportion of time hospitalized adults receiving supplemental oxygen spent within a targeted normoxemia range and reduced time spent in hypoxemia compared with manual titration by clinicians in the SAVE-O2 AI randomized clinical trial published in JAMA Internal Medicine.
Investigators conducted the multicenter, unblinded, parallel-group trial at 4 US tertiary care hospitals from May 2024 to November 2025. Adults hospitalized with acute respiratory illness, trauma, burns, or acute care surgery were eligible if they were receiving 1 to 10 L/min of supplemental oxygen and could undergo randomization within 36 hours of hospitalization. Patients with imminent plans to escalate to high-flow nasal oxygen, noninvasive ventilation, or invasive mechanical ventilation were excluded. Participants were randomly assigned 1:1 to autonomous oxygen titration or usual care with manual titration by clinicians.
Both groups had a target peripheral oxygen saturation of 93%, with an acceptable range of 90% to 96%, and were connected to 2 pulse oximeters for up to 72 hours following randomization or until hospital discharge. The autonomous group used the O2matic PRO100 closed-loop system, which automatically adjusted supplemental oxygen flow based on pulse oximetry measurements. The system was not cleared by the US Food and Drug Administration for clinical use in the US during the trial. Clinicians manually titrated supplemental oxygen according to local hospital protocols in the usual care group.
The primary outcome was the proportion of observed time spent within the targeted normoxemia range of peripheral oxygen saturation of 90% to 96%. The key secondary outcome assessed time with peripheral oxygen saturation below 88%. Investigators performed the primary analysis according to intention to treat and adjusted for study site and indication for hospitalization.
The trial population included 300 patients, with 152 assigned to autonomous titration and 148 to usual care. Median age was 66 years, 54% were female, 69% were admitted with acute respiratory illness, and 52% were receiving at least 3 L/min of supplemental oxygen at randomization.
Patients assigned to autonomous titration spent a mean 85% of observed time in normoxemia vs 63% with usual care, an adjusted difference of 21 percentage points. Mean time spent in hypoxemia was 2% vs 4%, respectively, and mean time spent in hyperoxemia was 9% vs 29%. In a post hoc analysis, mean time spent with peripheral oxygen saturation below 85% was 1% with autonomous titration vs 2% with usual care.
“The ability of autonomous oxygen titration to correct hypoxemia once it has been detected—in a more quickly and sustained fashion than manual oxygen titration—may improve clinical outcomes when applied at a population level,” wrote lead author David J. Douin, MD, MSc, of the Department of Anesthesiology at the University of Colorado Anschutz School of Medicine in Aurora, and colleagues.
Prespecified subgroup analyses of the primary outcome favored autonomous titration across skin pigmentation, home oxygen use, baseline oxygen flow rate, and indication for hospitalization. Sensitivity analyses using measurements from either the autonomous device pulse oximeter or the hospital pulse oximeter separately also favored autonomous titration.
Mortality, respiratory failure, supplemental oxygen-free days, and hospital-free days were exploratory outcomes. Four patients assigned to autonomous titration and 5 assigned to usual care died in the hospital by day 28. No serious adverse events were reported. Epistaxis occurred in 2 patients and nasal cannula or face-mask discomfort occurred in 2 patients, all in the autonomous group.
The investigators noted that the unblinded design could have influenced participant behavior or clinicians’ treatment decisions. They cautioned that findings from the PRO100 device may not apply to other systems used for autonomous oxygen titration. Because the system titrates oxygen according to pulse oximeter measurements, it does not correct for inaccuracies or imprecision in those measurements.
Disclosures: The study received support from the Defense Health Agency Research and Engineering Directorate through the Medical Technology Enterprise Consortium, with additional funding from the National Center for Advancing Clinical and Translational Sciences. IDTS Medical served as the investigational device exemption sponsor, and PRO100 devices were rented from O2matic for the trial. The funders and sponsors had no role in the study design or conduct, data collection or analysis, manuscript development, or decision to submit the manuscript for publication. Several authors reported grants and/or personal fees from government, academic, or industry sources during or outside the conduct of the study.
Full disclosures can be found in the published study.
Source:JAMA Internal Medicine
