Skillful use of BiPAP and high-flow nasal cannula (HFNC) can avoid intubation and improve outcomes.  However, there isn't comprehensive evidence about the nitty-gritty details of these techniques.  In this post I will use my opinions to fill some gaps in the evidence (1).  Noninvasive respiratory support remains more of an art than a science, perhaps a dark art at that.

Fundamental concepts


  1. Inadequate monitoring: Techniques described here are designed for an environment with close monitoring and staff available to intubate 24 hours a day.
  2. Multi-organ failure: Noninvasive respiratory support works best in patients with single-organ respiratory failure.

Pathophysiology of failure:  why do patients require intubation for respiratory failure?

To avoid intubation, we must first understand why patients require intubation:

  1. Hypercapneic encephalopathy (“CO2 narcosis”). These are patients with extremely high CO2 levels (usually pCO2 > 100 mm) causing obtundation.
  2. Refractory hypoxemia: Inability to oxygenate despite HFNC or BiPAP.
  3. Respiratory muscle exhaustion: This is the most common reason for intubation, because it represents a final common pathway of respiratory failure.  Any type of respiratory failure increases the work of breathing.  Eventually, respiratory muscles fatigue and fail.  As the diaphragm fails, the ability to cough and clear secretions is lost.  This may lead to mucus plugging, which causes acute deterioration.

Given the importance of respiratory muscle fatigue, this post will focus on:

Goals of therapy

The goal of HFNC or BiPAP is to stabilize the patient, in order to buy time for the underlying disease process to improve.  Therefore, my goals are as follows:

  1. Maintain adequate oxygenation.
  2. Provide enough ventilatory support so that the patient is comfortable and doesn't develop respiratory muscle fatigue.
  3. Ensure that the patient is protecting their airway.
  4. Serial examination (focused on #1-#3) shows that the patient's trajectory is either stable or improving (2).

That's all.  Please note that these goals don't include an immediate improvement in the pH or pCO2.  For example, consider the following scenario:

Some would call this a “BiPAP failure” because the pCO2 is unchanged, leading them to intubate her.  However, this is actually BiPAP success, because her dyspnea has resolved.  Her pCO2 will eventually improve, after steroids and bronchodilators have had more time to work.  Treat the patient, not the ABG.  This concept was proven by Brochard 1995 in a RCT evaluating the use of BiPAP in COPD:  BiPAP improved mortality despite having no effect on ABG parameters after one hour.  This study shows that BiPAP can be successful, even without any immediate effect on the ABG.


For most patients on noninvasive respiratory support, clinical assessment is all that’s needed.  This involves three or four pieces:

Look at the monitor and the patient, talk to the patient, perhaps try to wake them if they are sleeping.  That's all.  An experienced clinician will perform this assessment in about a minute.

The most common problem with assessment is excessive reliance on ABG values.  This is problematic for many reasons:

One situation where ABG/VBG is necessary is the somnolent patient who has received sedation (e.g. to facilitate tolerance of the BiPAP mask).  Blood gas analysis is needed to sort out whether somnolence is due to hypercapnia or medication effect.  Please note that VBGs are generally completely adequate to monitor a patient with respiratory failure.  If the patient has a peripheral IV line with blood return, this is a humane and rapid approach to obtain VBGs.

The key to assessment is frequent assessment by multiple providers.  Patients will be serially evaluated by a nurse, a respiratory therapist, and a physician.  When in doubt, all providers should discuss their impressions.  A team approach combines dozens of assessments by several experienced clinicians.

serial assessment by an experienced nurse or respiratory therapist is roughly a billion times more accurate than an ABG

— josh farkas (@PulmCrit) February 4, 2018

Device selection

Below is my general rubric for noninvasive respiratory support.  Let's walk through it step by step.

Is immediate intubation needed?

The decision to intubate is beyond the scope of this post.  However, it's worth mentioning a few “fake-out” situations, where patients look horrible but usually don't require intubation:

Are there contraindications to full-face BiPAP?

Contraindications include:

There are nearly no contraindications to HFNC, so any patient with a contraindication to BiPAP can be treated with HFNC (6).  BiPAP with a nasal interface can also be considered here, especially if HFNC isn't available (7).

BiPAP-sensitive conditions

BiPAP has some important advantages compared to HFNC:

For patients with these conditions, I will generally make a real effort to use BiPAP.  If the patient can't tolerate BiPAP due to anxiety, it may be worth using sedation to facilitate BiPAP tolerance.  Sedation is particularly useful for patients with COPD or asthma, who require a slow respiratory rate in order to exhale properly:

Some brief comments on various sedatives to facilitate BiPAP:

General BiPAP titration schemes are shown below (11):

HFNC for patients with parenchymal lung disease (e.g. pneumonia)

Parenchymal lung disease refers to any primary disease of the alveolar tissue itself.  The advantages of HFNC here include the following:

Compared to heart failure and COPD, parenchymal lung disease is a less BiPAP-responsive physiology.  Studies have found questionable benefit from BiPAP in this situation.  The FLORALI trial showed that in a group of patients with hypoxemic respiratory failure (mostly from pneumonia), HFNC was more successful than BiPAP at avoiding intubation and improving mortality.

Treat pleural disease with drainage

If a pneumothorax or pleural effusion is causing acute respiratory failure, the best treatment is drainage.  The concept of immediate drainage of a pneumothorax is universally understood.  Unfortunately, emergent drainage of a pleural effusion remains under-utilized.

Temporize upper airway obstruction with Heliox

Heliox is a fixed-ratio mixture of helium and oxygen which has lower viscosity than air.  This may be used to stabilize patients with upper airway obstruction.  One limitation is that these ratios contain a fixed amount of oxygen (typically 70-30 mix or 60-40 mix with 30% or 40% FiO2), so they cannot be used for patients who require >40% inspired oxygen.  Heliox isn't a cure for upper airway obstruction, but it may be useful:

Avoid BiPAP or HFNC in primary CNS disease

One practice which drives me crazy is the use of BiPAP for treatment of hypercapnia due to drug intoxication.  If the patient is so severely intoxicated that they truly need ventilatory support, then they should be intubated (aspiration risk precludes the use of BiPAP here).  A reasonable approach to these patients may be as follows:

Consider HFNC if the diagnosis is unclear

For a patient with unclear diagnosis, either BiPAP or HFNC could be used.  My preference is HFNC for the following reasons:

Choice of BiPAP vs. HFNC depends on the diagnosis, not the ABG

For example:

Selecting the treatment based on the diagnosis streamlines management.  The diagnosis can generally be determined rapidly on the basis of history, physical exam, and bedside ultrasonography.  This allows for immediate treatment, without waiting for an arterial blood gas measurement.

Parting shot: the value of empiricism

This post is intended to provide a rational framework, but not a rigid one.  Due to individual factors (e.g., mask seal, anxiety), the response of individual patients isn't entirely predictable.  If one device isn't working, try something else.  As long as you are monitoring patients closely to ensure that they are responding adequately, any strategy is acceptable.

  1. I initially gave a grand rounds presentation about this topic two years ago, with the intention of posting a blog about it shortly thereafter. I wrote a blog but wasn't happy with it.  Re-wrote the blog a year later, but still wasn't happy with it.  The problem was that there isn't much evidence, so the blog ended up being more expert-opinion type stuff than I would like.  This time around I committed to just clean it up as best as possible and post it.  I expect that there will be some lively debate about much of this, which is a good thing.
  2. BiPAP cannot be continued without a break for too long (>24-48 hours) without causing nutritional problems and pressure necrosis of the nasal skin. Thus, if the patient fails to improve on BiPAP for 1-2 days, then a transition to HFNC or intubation is needed.
  3. Hypercapnia is generally extremely well tolerated. The way that patients die from hypercapnia is usually hypercapneic encephalopathy causing somnolence, airway loss, apnea, and hypoxemia.  Therefore, a patient who is mentating well is unlikely to have life-threatening hypercapnia.
  4. The amount of minute ventilation per liter that patients will need depends on their size, their metabolic activity, and the amount of dead space in their lungs (pulmonary dysfunction). This is an important parameter to pay attention to in any patient on BiPAP or invasive ventilation.  Over time, practitioners should gain a general sense of how much minute ventilation various patients will need.  An average patient at rest might need about 6-7 liters/minute.  A patient with COPD exacerbation will have poor pulmonary function and might need more (e.g. perhaps 8-10 liters/minute). Thus, if you encounter a COPD patient on BiPAP who is receiving 4 liters/minute, this should be concerning as it is likely rather low.  Like ABG numbers, the minute ventilation is only one number which must be placed into proper clinical context.  Trending the tidal volume may be more useful than focusing on a single number.
  5. References: Umenda 2008,Sasse 1994Thorson 1983Hess 1992.
  6. A patient with bilateral nasal packing couldn't receive HFNC. I suppose that if a patient just had nasal surgery then HFNC might not be a terrific idea, but I've never encountered this situation.
  7. The preferred mask type for BiPAP is a full-face mask, given some evidence that this has a lower failure rate compared to a nasal mask (Girault 2009). This post is written in reference to full-face BiPAP.  However, it is certainly possible that nasal BiPAP could be used, which would avoid problems regarding aspiration.  Some evidence does support the use of BiPAP with a nasal interface (e.g. Bott 1993).  The helmet mask seems extremely promising, but this isn't currently available in the United States.
  8. There is outstanding evidence for the use of BiPAP in COPD, but much less evidence regarding the use of BiPAP in asthma. This likely reflects the fact that status asthmaticus is less common than exacerbated COPD and harder to study.  Both disease processes have similar physiology, so the fact that BiPAP works well in COPD implies that it should also work well in asthma.
  9. Due to the potential for hemodynamic instability with boluses of dexmedetomidine, my usual practice is to start the infusion without a bolus at a relatively high level (e.g. 1-1.4 mcg/kg/hr) and then gradually titrate down over the next 30-90 minutes. This is safer than using a bolus, but doesn't achieve immediate sedation.
  10. The best opioid here is arguably a remifentanil infusion, because it is very titratable. Eventually if remifentanil becomes generic and comes down in cost, this could be an enormously valuable agent for these patients.  Currently remifentanil isn't available in most emergency departments or ICUs.
  11. There are some very passionate arguments about how much PEEP should be used in various conditions. For example, many people argue that asthmatic patients should be placed on zero PEEP, to facilitate exhalation.  However, this isn't possible using my machine (to achieve pressurization, the lowest possible PEEP is ~3 cm).  Other folks argue that higher levels of PEEP are useful in COPD/asthma to stent open the airways during exhalation.  The BiPAP settings recommended here are reasonable in a variety of situations, but not necessarily perfect.  An inspiratory pressure >20-25 cm may cause gastric insufflation and promote vomiting.
  12. Ideally these patients should be monitored with end-expiratory CO2 to further trend their ventilation.


Josh Farkas
Social Me

Share this:

Would love your thoughts, please comment.x