Automated Oxygen Therapy Adjustments in Hospital Improves Oxygenation

Key Takeaways - In a randomized trial, a novel system for autonomous control of supplemental oxygen for inpatients who needed it improved the proportion of time spent with adequate oxygen levels. - The study showed safety of the approach and lower oxygen use overall compared with usual care. - The trial could not assess clinical outcome effects like reduced respiratory decompensation, shorter hospital stays, or improved survival. A novel system for autonomous control of supplemental oxygen for adult inpatients who needed it for acute illness or injury improved the proportion of time spent with adequate oxygen levels, the SAVE-O2 AI trial showed. The proportion of time spent in the targeted normal oxygen saturation range of 90-96% was an adjusted 21 percentage points higher compared with manual oxygen titration by clinicians over the first 72 hours (85% vs 63%, P<0.001), reported David Douin, MD, MSc, of the University of Colorado Anschutz School of Medicine in Aurora, and colleagues in JAMA Internal Medicine. Use of the closed-loop system to automatically adjust the flow of oxygen to maintain the target range also cut mean time spent in the hypoxic range under 88% compared with usual care (2.0% vs 3.6%, P=0.002), they said. The findings also were presented at the Military Health System Research Symposium in Kissimmee, Florida. Autonomous titration systems solve some of the resource intensive aspects of oxygen administration in hospitals, cutting down on the requirements for monitoring and repeated adjustments to flow rate by clinical staff, but also stand to improve clinical outcomes by reducing both the undercorrection and overcorrection that intermittent manual titration are prone to, said Melanie Weingart, MD, and Michael Matthay, MD, both of the University of California San Francisco, in an accompanying invited commentary. The findings showed safe optimization of oxygen delivery while reducing oxygen utilization in hospitalized patients requiring low-flow oxygen, with a modest but statistically significant reduction in patients experiencing hypoxemia, they said, but "whether these improvements in oxygenation metrics translate into meaningful patient outcomes, including reduced respiratory decompensation, shorter hospital stays, or improved survival, remains uncertain." "However, the absence of demonstrated patient-centered benefits in this trial should not diminish the potential health system value of technologies that improve oxygen stewardship and reduce resource utilization," they added. The trial included 300 patients on 1-10 L of oxygen acute hypoxemic respiratory failure in the setting of acute respiratory illness, trauma, burn, or acute surgery who were admitted to four U.S. hospitals. They were randomized to open-label use of autonomous oxygen titration with the investigational O2matic PRO100 device or manual titration by clinicians during the first 72 hours after randomization. Oxygen titrations of at least 1 L/min per day averaged 1.0 with the intervention and 1.3 with usual care, with a median of 6.9 and 7.3 flow rate changes recorded per day overall, respectively. Participants had a median age of 66, 54% were female, and around a quarter of patients were in the ICU. A substantial proportion had darker skin tones as defined by the Monk Skin Tone Scale (56% medium and 22% dark), which the editorialists pointed to as enhancing the generalizability of the findings. The primary endpoint of proportion of time spent in normoxemia (SpO2 90%-96%) was consistent across skin pigmentation and other prespecified subgroups including by pulmonary comorbidities, and oxygen flow rate at randomization. Among secondary findings, autonomous oxygen titration reduced the total volume of oxygen administered per participant per day (2,310 vs 2,992 L, incidence rate ratio 0.83, 95% CI 0.67-1.02) without a difference in supplemental oxygen-free days or in progression to more severe respiratory failure. The intervention had a relatively small impact on the low proportion of time spent in borderline hypoxemia (3.4% vs 4.0% with SpO2 88%-89%) or severe hypoxemia (1.0% vs 1.9% with SpO2 <85%, albeit statistically significant). There was a bigger impact on reduction in mean proportion of time spent in hyperoxemia (9.2% vs 29.1% with SpO2 >96%), which has "long been implicated in the pathophysiology of acute lung injury through the production of reactive oxygen species," Weingart and Matthay noted. However, they pointed to no impact of a conservative oxygen strategy to avoid hyperoxemia in critically ill patients receiving mechanical ventilation in prior trials. "This brings into question whether hyperoxia in hospitalized patients receiving low-flow oxygen is clinically deleterious," they wrote. "Although autonomous oxygen titration clearly reduces hyperoxemia and excessive resource use, the clinical significance of reducing transient hyperoxemia in patients receiving low-flow supplemental oxygen remains uncertain." No serious adverse events occurred in either group, and 28-day in-hospital mortality was similar between groups The trial wasn't blinded, assessed only one autonomous oxygen titration systems with unknown generalizability to others, and relied on pulse oximeters, with their known limitations, the researchers acknowledged. Weingart and Matthay also pointed to a lower than typical hypoxemia rate in the control group, which could mean enhanced usual care or selection bias if clinicians preferentially excluded patients at greatest risk of rapid fluctuation in oxygen or deterioration.