Highlight
• The UK-ROX randomized clinical trial enrolled 16,500 mechanically ventilated adult ICU patients and compared a conservative oxygen strategy (SpO2 target 90%, range 88–92%) with usual clinician-directed oxygen therapy.
• Conservative oxygen reduced measured supplemental oxygen exposure by 29% but did not reduce all‑cause 90‑day mortality (35.4% vs 34.9%; adjusted risk difference 0.7 percentage points; 95% CI, −0.7 to 2.0; P = .28).
• No significant differences were observed in ICU or hospital length of stay, days alive and free from organ support at 30 days, or mortality at other time points.
Study background and disease burden
Oxygen is one of the most commonly administered therapies in the intensive care unit (ICU). It is used to treat hypoxaemia and to support patients receiving invasive mechanical ventilation. While oxygen is life‑saving when needed, observational and physiological data have long suggested potential harms from both hypoxaemia and hyperoxaemia. Excess oxygen can increase oxidative stress, promote vasoconstriction, worsen pulmonary inflammation and impair microcirculatory flow; conversely, insufficient oxygenation increases the risk of tissue hypoxia and organ failure.
Despite biological plausibility and observational signals, randomized trial data addressing optimal oxygen targets for general mechanically ventilated ICU populations have been limited and sometimes conflicting. This uncertainty leaves clinicians with heterogeneous practice: some aim for higher SpO2 targets (≥96%) to avoid hypoxaemia, while others adopt lower targets to minimize oxygen exposure and potential oxygen toxicity. A large, pragmatic randomized trial was needed to define whether a conservative SpO2 target could improve patient‑centered outcomes such as mortality.
Study design
The UK‑ROX randomized clinical trial (Martin et al., JAMA 2025) was a multicenter, pragmatic, parallel‑group randomized trial conducted in 97 UK ICUs between May 2021 and November 2024, with follow‑up completed in February 2025.
Population: Adult patients receiving invasive mechanical ventilation and supplemental oxygen in ICU (n=16,500 randomized). Median age was 60 years (IQR 48–71), and 38.2% were female; baseline characteristics were balanced across groups.
Interventions:
• Conservative oxygen therapy (n=8,258 randomized; primary analysis n=8,211): clinicians titrated to the lowest fraction of inspired oxygen (FiO2) necessary to maintain pulse oximetry saturation (SpO2) at 90% (allowed range 88–92%).
• Usual oxygen therapy (n=8,242 randomized; primary analysis n=8,183): oxygen administered at treating clinician discretion with no trial‑mandated SpO2 target.
Primary endpoint: All‑cause mortality at 90 days.
Secondary endpoints: Duration of ICU and acute hospital stay among survivors, days alive and free from organ support at 30 days, and mortality at other time points.
Pragmatic features: Open‑label design reflecting real‑world practice and clinician decision‑making, central randomization, and intention‑to‑treat primary analysis. The intervention focused on bedside FiO2 titration with continuous pulse oximetry guidance rather than strict ventilator protocols.
Key findings
Primary outcome:
Of 16,500 randomized patients, 16,394 had primary outcome data (8,211 conservative; 8,183 usual care). By 90 days, 2,908 patients (35.4%) in the conservative oxygen group and 2,858 patients (34.9%) in the usual oxygen group had died. After adjustment for prespecified baseline variables, the absolute risk difference was 0.7 percentage points (95% CI, −0.7 to 2.0), P = .28 — a statistically non‑significant result indicating no evidence that the conservative strategy reduced 90‑day mortality.
Secondary outcomes and process measures:
• Oxygen exposure: The conservative strategy achieved a 29% reduction in exposure to supplemental oxygen compared with usual care (a meaningful reduction in delivered oxygen dose across the cohort).
• ICU and hospital length of stay: No statistically significant differences were observed in ICU length of stay or total hospital length of stay among survivors.
• Organ support: Days alive and free from organ support at 30 days were similar between groups.
• Other mortality time points: No clinically or statistically significant differences were reported at other predefined time points.
Safety and important subgroup considerations:
• The trial reported no signal of increased harm in the conservative group across the overall population. However, as with any pragmatic trial, subgroup heterogeneity is possible, and the main report did not show a convincing benefit or harm at the aggregate level.
Interpretation of effect size and confidence intervals:
• Point estimates favored neither strategy. The 95% CI for the adjusted risk difference crossed zero and excluded large beneficial effects of the conservative approach. Clinically important benefits larger than ~1% absolute reduction were unlikely on average in this population.
Expert commentary and contextualization
How UK‑ROX fits into existing evidence
Large randomized trials and meta‑analyses over the past decade have produced mixed results regarding conservative versus liberal oxygen strategies in different acute care settings. Prior small trials suggested possible benefit from conservative oxygen in certain populations, whereas others found no benefit or potential harm. The UK‑ROX trial is notable for its scale, pragmatic design, and focus on invasively ventilated ICU patients — a group that consumes substantial oxygen resources and where small per‑patient risks might translate into large public health impact.
Biological plausibility and mechanistic considerations
From a mechanistic standpoint, reducing hyperoxia exposure plausibly lowers oxidative injury, mitochondrial dysfunction, and pulmonary inflammation; it also avoids oxygen‑induced vasoconstriction. However, mechanical ventilation itself introduces many competing risks (ventilator‑associated lung injury, infection, hemodynamic instability) that may dilute any modest benefit from optimized oxygen dosing.
Limitations to consider
• Open‑label design: While pragmatic, lack of blinding may affect care in subtle ways (though the primary endpoint — mortality — is objective).
• SpO2 as target: Pulse oximetry is readily available and practical, but SpO2 can be affected by factors that introduce measurement error (poor perfusion, motion artifact, skin pigmentation); these issues may have affected target attainment.
• Heterogeneity of illness: The trial enrolled a broad, unselected mechanically ventilated population. If specific subgroups (for example, severe ARDS, COPD with chronic hypercapnia, or ischemic brain injury) benefit from more tailored oxygen targets, a trial of an overall conservative strategy may not detect subgroup effects without prespecified, adequately powered analyses.
• Generalizability: Conducted in UK ICUs, the findings are most directly applicable to similar high‑resource settings; practices and baseline oxygen use may differ elsewhere.
Clinical implications
• Routine adoption of a strict SpO2 90% target (88–92%) across all invasively ventilated ICU patients is not supported by evidence of reduced mortality from this large trial.
• The conservative approach did significantly reduce oxygen exposure without causing an apparent aggregate increase in harm; that may be relevant for oxygen stewardship and resource use, particularly in settings where oxygen supply is constrained.
• Clinical decisions should continue to individualize oxygen targets based on underlying pathology. Patients with conditions where cerebral or myocardial oxygen delivery is critical, or those with severe hypoxaemia, may need higher targets; conversely, avoiding routine hyperoxia remains reasonable.
Conclusions
The UK‑ROX trial provides robust, pragmatic evidence that a strategy targeting SpO2 90% (range 88–92%) to minimize supplemental oxygen exposure does not reduce 90‑day all‑cause mortality in a large population of mechanically ventilated adult ICU patients. Conservative titration cut oxygen exposure by nearly a third, but there was no measurable benefit in survival, ICU or hospital length of stay, or organ‑support–free days. These results argue against a one‑size‑fits‑all mandate for strict conservation of oxygen in this population and support continued individualized targets guided by clinical context.
Future research priorities
• Prespecified, adequately powered subgroup analyses or trials focused on populations with high biological plausibility for benefit (severe ARDS, ischemic brain injury, specific inflammatory phenotypes).
• Studies integrating more direct measures of tissue oxygenation or strategies combining oxygen targets with ventilation and haemodynamic optimization.
• Work examining applicability in low‑resource settings where oxygen scarcity, cost, and supply logistics are major constraints; here, oxygen‑sparing strategies may have different risk‑benefit calculus.
References
1. Martin DS, Gould DW, Shahid T, Doidge JC, Cowden A, Sadique Z, et al.; UK‑ROX Investigators. Conservative Oxygen Therapy in Mechanically Ventilated Critically Ill Adult Patients: The UK‑ROX Randomized Clinical Trial. JAMA. 2025 Aug 5;334(5):398–408. doi:10.1001/jama.2025.9663.
2. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al. Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2021. Intensive Care Med. 2021;47(11):1181–1247.
(For clinicians seeking broader context, consult local and international critical care guidelines and institutional protocols when implementing oxygen targets.)