COVID-19 Hypoxemia: A Better and Still Safe Way

I have been thinking a lot about patients with COVID-19 and the pulmonary pattern that they develop.  This disease process has been categorized like ARDS, but the reality is it is not like “typical” ARDS.  Lung compliance is often normal in these patients, and many patients are not in respiratory distress despite low O2 saturations.  Patients can have a bizarre hypoxemia that does not correlate with their symptoms.  I have even read reports of patients looking comfortable and speaking in full sentences with oxygen saturations in the 40 – 80% range.  There are also more traditional patients in respiratory distress with similar oxygen saturations.  This is a situation where we cannot treat a patient based solely on a number – pulse oximetry may not be a reliable marker of respiratory compromise.

Approaches to oxygen supplementation have stressed minimizing aerosolization of viral particles by avoiding HFNC and NIV.  This appears to be a fear-based statement as opposed to an evidence based one.  If we go straight from nasal cannula to intubation, we will simply run out of ventilators.  Then, more challenges present themselves like rationing mechanical ventilation and trying to figure out how to split ventilators due to the lack of resources.

Finally, I have yet to find a study that shows a mortality rate <50% once a patient is intubated.  Maybe a better way to deal with these patients is an intermediary step using HFNC or CPAP while proning patients while they are awake, before considering intubation.  In this post, I want to review some evidence to support my thoughts on this and, just assume that in every scenario we are discussing full PPE (eye protection, N95/PAPR, gown, gloves, and face shield).

Aerosolization [1][2]

The big fear with using modalities such HFNC and NIV is the aerosolization of viral particles.  The risk of aerosolization has been based on expert opinion from prior outbreaks, but I was unclear of the evidence for this.  First, some definitions:

Droplet transmission: expelled particles that can settle quickly and can travel short distances (within 1m) from the source.

Aerosol transmission: expelled particles with smaller size distribution that can travel farther.

The WHO defines this difference between airborne and droplet transmission based on the size of the particle (≤5um is airborne and >5um is defined as droplet). I was able to find two papers that support the safety of HFNC and wanted to summarize them here.

HFNC relieves respiratory distress by:

  • Matching the patient’s demand
  • Reducing anatomical dead space by decreasing the extent of rebreathing
  • Providing a positive pressure in the upper airway
  • Reducing inspiratory effort
  • Improving oxygenation
  • Improving dynamic compliance

Although the guidelines tell us that patients with COVID-19 should be preferably managed in negative pressure isolation rooms for infection control, the reality is we simply do not have enough of these rooms for the volume of patients that are coming in.  Therefore, there has been an increasing number of people being treated with HFNC or CPAP in the ICUs, wards, and EDs outside of negative pressure rooms.  From an infection control point of view, it is important to understand the exhaled air dispersion distance and direction during application of HFNC and CPAP at different airflow rates.

The first study [1] was a simulated patient examining exhaled air dispersion and directions in a hospital isolation room with application of HFNC and CPAP with two nasal pillows in a negative pressure isolation room with 16 air changes per hour.

There was a negligible lateral dispersion of exhaled air when nasal pillows were tightly connected to the tubing.  It is important to note that exhaled air dispersion extended to 620mm laterally with a loose connection between the cannula and the interface tube when HFNC was at 60LPM when delivered to the simulated patient with a normal lung condition.

With CPAP, using a Quattro air mask there was no significant leakage when applied at 5, 10, 15, or 20 cmH20. With nasal pillows the dispersion was a bit more than HFNC in normal, mild, and severe lung injury.

Summing up the tables above, the mean exhaled air dispersion was up to 172 +/- 33mm along the sagittal plan via HFNC at 60LPM and CPAP via nasal pillows used up to 20cmH20 similar leakage distances could be detected up to 264 and 332mm.  As the severity of lung injury worsened the exhaled air dispersion distances became shorter for both HFNC and CPAP via nasal cannula.  In contrast when CPAP was increased up to 20cmH20 via the Quattro Air mask, there was no significant leakage, irrespective of the severity of lung injury. One important note that may be lost outside of a study is the importance of a tight mask seal.  This is key to minimize aerosolization.

The second study was performed by VapoTherm [2] looking at high velocity nasal insufflation (HVNI) and aerosolization with HFNC in a standard non-negative pressure room with 6 air changes per hour. They performed a simulation with HFNC and a surgical mask on the patient showing that there is a leak around the mask of 11.6% with no therapy, 12.6% with low flow oxygen (1 – 6LPM NC), and 16.5% with HVNI.  The mask itself filtered most of the particles flowing through the mask with particles that did leak having a path length of <1m. The images below show velocity and the table shows leakage.

Another study discusses the scientific evidence supporting the risk of HFNC induced bio-aerosol dispersion [10].  They summarize the risk of aerosolization utilizing a smoke simulation via a manikin model. A summary of their findings is below:

An additional layer of protection can be added with a clear plastic drape over the face of the patient.  With the use of a surgical mask on the patient and a clear plastic drape there should be minimal aerosolization.

Bottom Line: A surgical mask placed over the mouth and nose of a patient significantly reduces the spread of these particles by reducing the flow velocity and trapping particles

Risk of Aerosolization from Different Oxygen Delivering Modalities [7]

  • This was not a head to head comparison, but a compilation of individual studies…

Escalation of O2 Therapy [7]

Starting with patients with O2 sat <90% and mild to moderate work of breathing.  Goal should be to maintain O2 sat >88 – 90%

  • NC 1 – 6LPM + Surgical Mask
  • Venti Mask
  • NRB + NC
  • HFNC + Surgical Mask
  • NIV (i.e. CPAP)
  • Escalate therapy to the next step if O2 sat, RR, and work of breathing not improved.  Also perform awake proning where possible in every step of the escalation.

Infographic Created by Mark Ramzy, DO (Twitter: @MRamzyDO)

HFNC/NIV Proof of Concept [8]

  • Multicenter, retrospective observational trial evaluating the effectiveness of HFNC/NIV in patients with confirmed COVID-19 pneumonia
    • 318 patients screened
    • 27 (8.4%) received HFNC/NIV
      • HFNC 1st Line: 17 patients
      • NIV 1st Line: 9 patients
      • Invasive Ventilation 1st Line: 1 patient
    • HFNC failure defined as need for NIV or intubation as rescue therapy
      • 7/17 (41%) HFNC 1st patients experienced HFNC failure
      • 2/7 (29%) failed NIV and required intubation
      • HFNC failure rate:
        • PaO2/FiO2 > 200mmHg: 0/6 patients (0%)
        • PaO2/FiO2 ≤200mmHg: 7/11 patients (63%)
        • p = 0.04
      • Respiratory rate significantly decreased after 1 – 2hrs of HFNC in successful group:
      • Median 26 vs 23 compared to the unsuccessful group (p = 0.03)

    • Limitations:
      • Retrospective trial design can lead to patient selection bias
      • Unblinded study can lead to physician bias (Transition to NIV or intubation was decided by physician which is a subjective decision that will vary from physician to physician)
    • Bottom Line:
      • HFNC non-inferior to NIV for intubation rate
      • HFNC more comfortable than NIV for patients
      • Manipulation of HFNC is much easier than NIV

ROX Index [3]

An important consequence of using HFNC is the risk of delaying a needed intubation which can lead to worsened outcomes for patients due to respiratory fatigue.  ROX is the first prospectively validated and accepted intubation criteria for acute hypoxemic respiratory failure.

ROX is defined as the ratio of oxygen saturation as measured by pulse oximetry (SpO2) divided by the fraction of inspired oxygen (FiO2) in relation to the respiratory rate ((SpO2/FiO2)/RR).  The variables with a positive association with HFNC success were placed in the numerator and the variables with an inverse relationship with HFNC success were placed in the denominator

There was a 2-year multicenter prospective observational cohort study including 191 patients with pneumonia treated with HFNC.  In this study 68 (35.6%) of patients required intubation.  A ROX index ≥4.88 measured at 2, 6, and 12 hours after HFNC initiation was consistently associated with a lower risk for intubation.

Bottom Line: The ROX index potentially could be used to help predict which patients are failing HFNC and will require expeditious intubation versus those who can be maintained on HFNC. It is important to understand, that this was a fairly small smaple size of patients with none of the patients having COVID-19 where hypoxia can be out of proportion to exam exam findings frequently.  This could potentially lead to more inapproprate intubations but one possible option.

Awake Proning [4][5]

Is awake proning a thing?  The answer is YES.  Although the evidence is not robust for this modality, we have to understand that we are at the bleeding edge of a disease process that came into being only 3 – 4 months ago.

Previous studies  have shown that prone positioning can improve the PaO2/FiO2 ratio by +35mmHg and help reduce mortality in moderate to severe ARDS patients [6].  Prone positioning has only been recommended in severe ARDS with PaO2/FiO2 <100mmHg who are intubated. It has not been recommended in non-intubated patients with mild to moderate disease.

Early use of non-invasive ventilation (NIV) and high flow nasal cannula (HFNC) could potentially reduce the need for intubation of mild to moderate ARDS patients.  NIV applies end-expiratory positive airway pressure (PEEP) and pressure support (PS) to help improve functional residual capacity, open collapsed alveoli, improve ventilation-perfusion mismatching, and reduce intrapulmonary shunting.

Awake Proning (Link to Image is HERE)

Logistically, how do we make this happen? Since the patients are awake, they should be able to prone themselves. The goal target should be a SpO2 >88 – 92% with FiO2 ≤0.6 [4].  If this is not achieved patients can be proned in the following manner:

Potential COVID-19 Hypoxemia Workflow

  1. SpO2 > 88%: reposition to prone position 1hr after the HFNC initiated
  2. SpO2 <88% on HFNC for >10min: place in prone position with same setting of HFNC
  3. If unable to achieve SpO2 and FiO2 goals above, switch to NIV. If stable SpO2 reposition to prone position 1 hr after the NIV initiated
  4. SpO2 <88% on NIV for >10min: place in prone position with the same setting of NIV
  5. If none of the above are working, then consideration for intubation will be needed however comfort, RR, tachycardia, and perfusion have to all be considered in the decision to intubate

Patients can essentially have one of 4 options:

  • HFNC alone
  • HFNC + prone positioning
  • NIV alone
  • NIV + prone positioning

Patients can remain in the prone position for at least 30 minutes and if well tolerated potentially even longer (Up to 6hours).  In one study [4] this was performed at least two times a day for the first 3 days.  No sedation was required, and patients could be monitored at the bedside to ensure their comfort and tolerance was successful.

In a small study of 20 non-intubated ARDS patients [4] something very similar to this was done.  Their primary outcome was the rate of avoidance for intubation and secondary outcomes for efficacy were the increase in PaO2/FiO2.

11/20 (55%) of patients avoided intubation and of these 8 (73%) of patients had moderated ARDS and 3 (27%) had severe ARDS. To be fair 9/20 (45%) patients required intubation of which 2/9 (22%) had moderate ARDS and 7 (78%) had severe ARDS and finally 3/9 intubated patients required ECMO.  As for the PaO2/FiO2 (secondary outcome) prone positioning improved this measure as well in the following way:

  • HFNC + Prone Positioning = 130 +/- 35mmHg
  • HFNC Alone = 95 +/- 22mmHg
    • P = 0.016
  • NIV + Prone Positioning = 166 +/- 12mmHg
  • NIV Alone = 140 +/- 30mmHg
    • P = 0.133

Bottom Line: In a crisis situation, where ventilators are running low, awake proning is a feasible option in non-intubated awake patients with mild to moderate ARDS.  Awake prone positioning appears to add 25 to 35mmHg for the PaO2/FiO2 [4] when compared to the HFNC or NIV without prone positioning.

COVID19 Lung Injury is NOT HAPE [9]

  • HAPE
    • Excessive and uneven hypoxic pulmonary vasoconstriction
    • Increased pulmonary artery pressure in certain regions of the lung
    • Increased pulmonary capillary hydrostatic pressure
    • Leakage of fluid out of vascular space into alveolar space
    • Treatment: Supplemental O2 alone OR descent to lower elevation (Resolves alveolar and interstitial edema with resolution within hours to days)
  • COVID-19 Lung Injury
    • Viral mediated inflammation
    • Alveolar epithelial inflammation/dysfunction
    • Impaired surfactant function/alveolar fluid clearance
    • Alveolar collapse and/or filling (V/Q mismatch)
    • Increased pulmonary artery pressure is a consequence, NOT the cause, of alveolar edema
    • Treatment: Supplemental O2 (Improves hypoxemia but does not resolve underlying inflammation or lung injury)

COVID-19 Hypoxemia: A Better and Still Safe Way

  • PPE should be the initial consideration when managing patients with COVID-19
  • Use of HFNC/NIV are unlikely to represent a disproportionate risk of droplet exposure as compared to other forms of ventilatory support
  • Use of HFNC at a flow rate of 40LPM with a surgical facemask in place over the patients mouth and nose is no different and shows no greater dispersal of particles than low flow oxygen with a facemask in place
  • Awake proning is feasible and may stave off early intubation
  • In my mind patients come in 3 flavors…
    • Silent hypoxemia = do not need intubation/trial of NIV/HFNC + Awake Proning
    • Intermediate Hypoxemia = may need intubation/trial of NIV/HFNC + Awake Proning
    • Respiratory Distress = Need intubation/ too far gone for NIV/CPAP
  • A better and still safe way to manage COVID-19 patients with ARDS:
    • NC 6LPM + Surgical Mask (Goal SpO2 88 – 92%) if this fails…
    • HFNC + Surgical Mask + Awake Prone –> Calculate ROX Score (Pulse Ox/FiO2/RR) at 2hr, 6hr, and 12hrs –> if <4.88 –> Consider Intubation
    • Another option would be CPAP + Awake prone position instead of HFNC before considering intubation


  1. Hui DS et al. Exhaled Air Dispersion During High-Flow Nasal Cannula Therapy Versus CPAP via Different Masks. Eur Respir J 2019. PMID: 30705129
  2. Leonard S et al. COVID-19 Transmission Assessment Report. High Velocity Nasal Insufflation (HVNI) Therapy Application in Management of COVID-19. Vapotherm. [Vapotherm Website]
  3. Roca O et al. An Index Combining Respiratory Rate and Oxygenation to Predict Outcome of Nasal High-Flow Therapy. Am J Respir Crit Care Med 2019. PMID: 30576221
  4. Ding L et al. Efficacy and Safety of Early Prone Positioning Combined with HFNC or NIV in Moderate to Severe ARDS: A Multi-Center Prospective Cohort Study. Crit Care 2020. [Epub Ahead of Print]
  5. Sun Q et al. Lower Mortality of COVID-19 by Early Recognition and Intervention: Experience from Jiangsu Province. Ann Intensive Care 2020. [Epub Ahead of Print]
  6. Munshi L et al. Prone Position for Acute Respiratory Distress Syndrome. A Systematic Review and Meta-Analysis. Ann Am Thorac Soc 2017. PMID: 29068269
  7. Whittle J et al. Respiratory Support for Adult Patients with COVID-19. JACEP Open 2020. [Epub Ahead of Print]
  8. Wang K et al. The Experience of High-Flow Nasal Cannula in Hospitalized Patients with 2019 Novel Coronavirus-Infected Pneumonia in Two Hospitals of Chongqing, China. Ann Intensive Care 2020. PMID: 32232685
  9. Luks AM et al. COVID-19 Lung Injury is Not High Altitude Pulmonary Edema. High Altitude Medicine & Biology 2020. PMID: 32281877
  10. Li J et al. High-Flow Nasal Cannula for COVID-19 Patients: Low Risk of bio-Aerosol Dispersion. European Respiratory Journal 2020. [Epub Ahead of Print]

For More on This Topic Checkout:

Post Peer Reviewed By: Anand Swaminathan, MD (Twitter: @EMSwami)

Cite this article as: Salim Rezaie, "COVID-19 Hypoxemia: A Better and Still Safe Way", REBEL EM blog, March 31, 2020. Available at:
The following two tabs change content below.

Salim Rezaie

Emergency Physician at Greater San Antonio Emergency Physicians (GSEP)
Creator & Founder of REBEL EM

Like this article?

Share on facebook
Share on Facebook
Share on twitter
Share on Twitter
Share on linkedin
Share on Linkdin
Share on email
Share via Email

Want to support rebelem?

24 thoughts on “COVID-19 Hypoxemia: A Better and Still Safe Way”

  1. Thank you for this post. This article is very interesting : Modified N95 Mask Delivers High Inspired Oxygen Concentrations While Effectively Filtering Aerosolized Microparticles. Ann Emerg Med 2006;48:391-399.

  2. Thanks for the post. We in Germany do use cpap via mask while patients are awake for at least 36 hrs if tolerated before we think about intubation. I used this method almost all my fellowship period on ICU to avoid intubation and it helped in almost 90% of cases.

    • Hello Reza,
      TY for sharing…the more I read about COVID-19 the more I think we should try to stave off intubation as long as possible. Its good to know anecdotally that this has worked for you in the past. I can only hope we have similar success with COVID-19.


  3. FWIW … I work at 9200 feet of elevation. We have HAPE patients all the time with very low Sp02 (you name it … I’ve seen 12%) and they are often walking / talking / comfortable. Obviously different pathophysiology, but the “silent hypoxemia” is very reaL. 85-87% is nothing and I bet most of those folks will do just fine on nasal cannula.

    • Hello Jack,
      Appreciate your input. Agree with you 100%. I think we are doing more harm than good intubating patients early. We have an adopted HFNC + Awake Proning with a surgical mask over the patient. Thus far the results have been rather promising. Better than the 50 – 90% mortality we are seeing in intubated patients globally.


  4. I’ve seen some theorizing that the virus is actually interfering with or damaging maturing rbc hgb in the marrow as an explanation for this bizarre hypoxia syndrome. Has anyone tried exchange transfusion as a hail mary or are we not to the point of trying completely unproven ideas?

    • Hey Brad,
      Completely aware that there is a chance that there is also a hemoglobinopathy of some sort at play. Think this is all theoretical at this point and really proven. If confirmed to be true, then it makes sense why exchange transfusion would work. I think before that is done we are using patients like guinea pigs which goes against our mantra of do no harm. Would certainly like to see more data on this before we attempt something like this as this is not an inexpensive or without risk treatment modality. Hope that makes sense.

    • TY for sharing…have read this study and know there is some effect on hemoglobin…however it this was the only reason this was happening then you would expect your ABG and SpO2 to disagree which is not what is seen clinically. This may be a component, but most likely a very small component of the hypoxemia we are seeing.


  5. Do you recommend oxygen levels greater than 6 liters via the nasal cannula when the patient has the non-rebreather on? We have heard about using 10-15 liters nasal cannula with 10-15 liters non-rebreather together and I can’t find literature to support this.

    • Hello Diane,
      The recommendation of 10 – 15L NC with NRB 10 – 15L seems to be expert opinion. I have yet to find a study in my readings that state this exact fact. Theoretically, if you are using NC + NRB the patient is requiring more support/O2 so it makes sense why you have higher flows. As for aerosolization risk, I have no idea…if we can run HFNC at 40 – 60L with a surgical mask safely, I don’t know why we couldn’t do 10 – 15L with NC + NRB. This however is speculation/opinion.

  6. This is actually very scary and not rosy. Even a little dumb to read about it. Hah But the information is quite useful and constructive. Thanks for this. In fact, the options are really suitable and I wonder if they are used so widely. I myself do not belong to the category of doctors, but in principle this is really interesting to me. Therefore, I came to your article. And did not lose. By the way, I read recently about non-invasive ventilation. And I can’t understand if these are similar cases or not?


Leave a Comment

Time limit is exhausted. Please reload CAPTCHA.