Table 1. Summary of recent studies that compared low versus high oxygenation targets in ICU patients.

FiO2, fraction of inspired oxygen; ICU, Intensive Care Unit; eICU-CRD, eICU Collaborative Research Database; MIMIC, Medical Information Mart for Intensive Care III database; IQR, interquartile range; LOS, length of stay; mRS, modified Rankin Scale; PaO2, supranormal arterial oxygen; RCT, randomized controlled trial; SaO2, saturation of arterial oxygen; SpO2, saturation of peripheral oxygen.

Study	Design	Sample Size	Reason for admission	Intervention	Main Findings	Conclusion
Mackle et al. [33]	RCT	1,000	Acute brain disease, surgery	In conservative intervention, the limit for SpO2, which was 97% FiO2, decreased to 0.21.	No significant difference in ventilator-free days. A difference of >28 h was reported duration of ICU stay in the conservative group and control group. Conversely, the conservative-oxygen cohort exhibited a diminished duration of time with SpO2 >96% (median time: 27 h [IQR: 11.0–63.5]) compared to 49 h [IQR: 22–112] in the usual-oxygen cohort). At 180 d, mortality rates were insignificant in both groups.	No difference was found between the conservative and normal groups.
Panwar et al. [26]	RCT	103	Trauma, Surgery, and medical	SpO2 levels 88–92% (conservative); SpO2 levels ≥96% (liberal).	SpO2, SaO2, PaO2, and FiO2 significantly differed between the conservative and liberal oxygenation groups (P<0.0001). No significance difference was found in mortality or organ dysfunction. The conservative arm had a higher percentage of time spent with SpO2 <88%.	Conservative oxygenation therapy is feasible compared to liberal.
Young et al. [27]	Post hoc analysis	251	Sepsis	In conservative intervention, the limit for SpO2, which was 97% FiO2, decreased to 0.21.	Patients with sepsis in conservative oxygenation therapy exhibited reduced time in ICU with SpO2 ≥97%. No significant difference was reported in 90 d mortality rates.	Conservative therapy is a better option in patients with sepsis.
Helmerhorst et al. [28]	Single-center pilot prospective before-and-after trial	15,045	N/A	Conservative oxygenation (PaO2: 55–86 mmHg).	PaO2 levels elevated from 47% at baseline to 63% and 68% during phases 1 and 2, respectively (P<0.0001). No significant differences in ICU fatality or ICU-free days were noted.	Conservative oxygenation targets were feasible.
van den Boom et al. [36]	Replicate retrospective analyses	26,723 in eICU-CRD and 8,564 in MIMIC	Atrial fibrillation, sepsis, stroke	–	A significant inverse relationship was observed between the time spent within the optimal SpO2 range and hospital mortality, with an odds ratio of 0.42 for eICU-CRD and 0.53 for MIMIC.	The most suitable range of SpO2 was 94–98%.
Suzuki et al. [29]	Pilot before-and-after trial	105	Cardiovascular, sepsis, respiratory	Conservative = SpO2 of 90–92%.	Time-weighted average SpO2 and PaO2 levels were significant between conservative and conventional oxygen therapy. The median SpO2 was 95.5% and 98.4% during conservative oxygen and conventional therapy, respectively (P<0.001). No significant differences were observed in the PaO2/FiO2 ratio or any other biochemical or clinical outcomes between the two therapy periods.	Conservative oxygen was more suited in terms of clinical outcomes.
Gelissen et al. [34]	RCT	400	Systemic infection, stroke, cardiac arrest, pneumonia	PaO2 = 8–12 kPa (low-normal); PaO2 = 14–18 kPa (high-normal).	No significant difference was reported in both groups regarding the median duration of mechanical or in-hospital mortality. Mild hypoxemia occurrences were more frequent in the low-normal group.	No difference was found in low or high oxygenation targets.
Girardis et al. [30]	RCT	434	Surgical, medical	Conservative therapy = SpO2 94–98%; Conventional therapy = SpO2 97–100%.	The conventional group exhibited higher median PaO2 than the conservative group (P<0.001). The conservative arm showed decreased mortality (11.6%), whereas conventional therapy showed increased mortality (20.2%).	Conservative protocol was better in terms of ICU mortality.
Azoulay et al. [37]	RCT	776	Acute hypoxemic respiratory failure	PaO2 <60 mmHg.	Patients who received high-flow oxygen therapy had a higher PaO2:FiO2 ratio and a lower respiratory rate after 6 h. No significant differences were observed regarding LOS, infections, and dyspnea.	No difference was observed regarding mortality outcomes.
Schjørring et al. [35]	RCT	2,928	Pneumonia, Cardiac arrest, Myocardial infarction, Traumatic brain injury	Lower oxygenation target = 60 mmHg; Higher oxygenation target = 90 mmHg.	At 90 d observation, the mortality rate was 42.9% vs. 42.4% in the low and high oxygenation target groups, respectively. No significant difference was observed between the groups regarding survival without life support or posthospital discharge survival rates.	No difference was found between low and high oxygenation targets.
Asfar et al. [31]	RCT	442	Sepsis	FiO2 at 1.0 (hyperoxia); SpO2 = 88–95% (normoxia)	At 28 d follow-up, 43% of patients in the hyperoxia group had died, while 35% of patients in the normoxia group had died. Adverse events were significantly different in both groups, with almost double the incidence observed in the hyperoxia group compared to the normoxia group.	Arterial hyperoxia increases the risk of mortality.
Taher et al. [32]	RCT	68	Traumatic brain injury	Experimental = 80% oxygen via mechanical ventilator; Control = 50% oxygen via mechanical ventilator.	The median duration of ICU stay was less in the experimental group (P=0.280). After 6 mo. of injury, the moderate outcome score was 16 and 9 in the control and experimental groups, respectively; mRS at discharge was 2.6 and 2.3 in the control and experimental groups, respectively (P=0.320).	Experimental oxygen therapy was better suited for critically ill patients.