How Much O2 Is Right in COVID?

Background: Hypoxemic respiratory failure is a frequent complicating feature of severe COVID-19 infection. Early in the pandemic, extensive efforts were made to identify the best approach to oxygenation in this group of patients. Best practices settled on aggressive use of non-invasive ventilation (NIV) and delaying invasive mechanical ventilation (IMV) where feasible. Advanced interventions include proning (both on NIV and IMV) and extracorporeal mechanical oxygenation (ECMO). As therapies are escalated, it’s critical for clinicians to have high-quality data on target oxygen levels.

Paper: Nielsen FM et al. Lower vs higher oxygenation target and days alive without life support in COVID-19. The HOT-COVID Randomized Controlled Trial. JAMA 2024. PMID: 38501214

Clinical Question: In COVID-19 patients with severe hypoxemia, should clinicians be targeting a higher (90 mm Hg) or lower (60 mm Hg) PaO2?

Population: Adult patients (> 18 years of age) admitted to the ICU with COVID-19 and severe hypoxemia (defined as receiving supplemental O2 with a flow rate of at least 10L/min or receiving mechanical ventilation or non-invasive ventilation.

Intervention: Target PaO2 = 60 mm Hg

Control: Target PaO2 = 90 mm Hg

Outcome (primary): Absolute number of days without life support in 90 days defined as the absence of mechanical ventilation, circulatory support, or kidney replacement. 

Outcomes (secondary): 

  • 90-day all-cause mortality.
  • Adverse events in the ICU within 90 days (new shock, cerebral ischemia, myocardial infarction, or intestinal ischemia).
  • Absolute number of days alive and out of hospital in 90 days.

Design: Multicenter, multinational, non-blinded, parallel-group randomized controlled trial.

Exclusion: Patients who could not undergo randomization within 12 hours of ICU admission, patients in whom consent could not be obtained, and those previously included in HOT-ICU or HOT-COVID trials.


Primary results

  • 726 of the preplanned 780 patients were enrolled
    • Trial stopped early due to slow enrollment in the last year of the study.
    • Low O2 Group: n = 365
    • Standard O2 Group: N = 361
  • Primary outcome data available for 96% (697/726) of patients.

Critical results*Statistically significant difference

High O2 Group Low O2 Group Bootstrapped Mean Difference (95% CI) Statistically Significant?
Primary Outcome Days Alive w/o Support 72 days 80 days 5.8 days (0.2 – 11.5) YES
Risk Ratio (98.6% CI)
Secondary Outcomes 90-day Mortality 34.7% 30.2% 0.83 (0.66 – 1.13) NO
Adverse Events 51.7% 47.5% 0.94 (.79 – 1.13) NO
Days Alive Out of Hospital 48 days 59 days NO


  • Asks a clinically important question and furthers our understanding of COVID-19.
  • Multi-center, international study increases external validity
  • Baseline characteristics were well-balanced between groups.
  • Minimal exclusion criteria increasing generalizability of the data. 


  • Study was non-blinded: treating clinicians, patients and family members were all aware of which arm the patient was randomized to. 
    • This may bias clinicians in terms of interventions rendered to patients.
    • There was not a standard protocol instituted for decisions to intubate or wean from therapy.
    • Unclear how this bias affects outcomes.
  • Investigator initiated enrollment instead of consecutive enrollment.
  • Stopped early due to low enrollment
    • This may lead to overestimation of the benefit of the intervention.
  • Unclear if therapeutic interventions were balanced between the two groups as this data is not provided.
  • Did not protocolize the measurement of PaO2 but 4 measurements per day were expected.
  • Cohort was older (mostly > 65) and predominantly male which may reduce generalizability.
  • Patients lost to follow up were excluded from the analysis.


  • PaO2 of 60 mm Hg vs 90 mm Hg is essentially an O2 sat difference of 90-92% vs 95-96%.
  • Difference in oxygen levels between the control and intervention groups:
    • Figure 2 in the manuscript demonstrates that there was a difference in PaO2 and O2 saturation between the two groups.
    • The difference, however, was minimal:
      • PaO2: 90 mm Hg vs 68 mm Hg
      • O2 sat: 96% vs 93%
  • With this small difference in oxygenation between groups, it’s important to try to understand why there was a difference in the primary outcome.
    • The difference may have been driven by other features of care not discussed here. For instance, additional therapeutics including steroids, JAK inhibitors etc.
    • More aggressive care to maintain a higher PaO2 may have been deleterious to patients:
      • Mechanical ventilation, prone positioning and transfusions were all higher in the high PaO2 group.
      • These interventions may have harmed patients instead of improving outcomes
    • The absence of blinding and the non-consecutive enrollment introduce biases that may also influence the outcome.
  • Accumulating evidence over the last decade has shown that an O2 sat of > 90% should be targeted in the critically ill and that a sat > 95% doesn’t bestow additional benefit in most circumstances. This data supports current practice of targeting a lower O2 sat.

Author Conclusion:In adult ICU patients with COVID-19 and severe hypoxemia, targeting a PaO2 of 60mmHg resulted in more days alive without life support in 90 days than targeting a PaO2 of 90mmHg.”

Bottom Line: Targeting a lower PaO2/O2 sat (60 mm Hg/90-92%) resulted in better outcomes than targeting higher PaO2/O2 sat in patients with COVID-19 and severe hypoxemia. 

Clinical Take Home Point: Titrate O2 therapy to an O2 sat > 90% but don’t strive for normoxia in COVID-19 patients with severe hypoxemia. Once again we see that less is often more. Interventions directed at increasing PaO2 levels above 60 mm Hg (O2 sat > 90%) did not result in better outcomes and appear to harm patients.

Read More

Post Peer-Reviewed By: Marco Propersi DO (Twitter/X: @marco_propersi)

Cite this article as: Anand Swaminathan, "How Much O2 Is Right in COVID?", REBEL EM blog, April 22, 2024. Available at:

Like this article?

Share on Facebook
Share on Twitter
Share on Linkdin
Share via Email

Want to support rebelem?