The n e w e ng l a n d j o u r na l of m e dic i n e

Review Article

C. Corey Hardin, M.D., Ph.D., Editor

Noninvasive Respiratory Support for Adults

with Acute Respiratory Failure
Laveena Munshi, M.D., Jordi Mancebo, M.D.,* and Laurent J. Brochard, M.D.​​

A
From the Interdepartmental Division of cute respiratory failure is a common indication for admission
Critical Care, University of Toronto (L.M., to an intensive care unit. Invasive mechanical ventilation, particularly
L.J.B.), the Critical Care Department Sinai
Health System (L.M.), and Keenan Re- positive-pressure ventilation, has been the cornerstone of the management
search Centre for Biomedical Science, Li of severe forms of acute respiratory failure since the 1950s.1 However, despite
Ka Shing Knowledge Institute, Unity Health major advancements in critical care management, the complications and mortal-
Toronto (L.J.B.) — all in Toronto; and the
Intensive Care Department, Hospital ity associated with intubation and positive-pressure ventilation are not insignifi-
Universitari de La Santa Creu I Sant Pau, cant.2 Efforts to circumvent invasive mechanical ventilation through the use of
Barcelona (J.M.). Dr. Brochard can be con- noninvasive devices have therefore garnered much attention. For some conditions,
tacted at ­laurent​.­brochard@​­unityhealth​.­to
or at the Keenan Research Centre for Bio- such as cardiogenic pulmonary edema and chronic obstructive pulmonary disease
medical Research, Li Ka Shing Knowledge (COPD) exacerbations, noninvasive respiratory support is highly beneficial,3 where-
Institute, Unity Health Toronto, 209 Vic- as for hypoxemic respiratory failure, the presence of associated conditions such as
toria St., Toronto, ON, M5B 1T8, Canada.
sepsis and shock4 may make the use of noninvasive respiratory support risky and
*Deceased. its benefits more difficult to delineate.
N Engl J Med 2022;387:1688-98. Three main methods of noninvasive support are used in the acute care setting:
DOI: 10.1056/NEJMra2204556 a high flow of gas delivered through a large-bore nonocclusive nasal cannula (i.e.,
Copyright © 2022 Massachusetts Medical Society.
high-flow nasal cannula), continuous positive airway pressure (CPAP), and nonin-
vasive ventilation (i.e., pressure-support ventilation with positive end-expiratory
pressure [PEEP]). In this review, we provide an overview of the physiological ef-
fects, different configurations, clinical indications, and evidence for the use of
noninvasive respiratory support in adults with acute respiratory failure.

Ph ysiol o gic a l Effec t s

Respiratory failure has two main components: ventilatory dysfunction and hypox-
emia. Ventilatory dysfunction leads to dyspnea, increased work of breathing, use
of accessory muscles, and hypercapnia: this situation is best managed with the use
of a method that offers frank ventilatory support. Hypoxemia reflects inadequate
gas exchange and warrants different forms of oxygen therapy and specific device
settings (mostly positive pressure) to improve gas exchange. Clinical respiratory
distress and severe hypoxemia often go together in various combinations because
injuries that cause abnormal gas exchange often result in abnormal mechanics and
high work of breathing, but they can also be dissociated.5 The different noninva-
sive respiratory support interfaces and methods are shown in Figure S1 in the
Supplementary Appendix,6 available with the full text of this article at NEJM.org,
and Table S1 describes each of the physiological effects.

High-Flow Nasal Cannula

High gas flow rates (≥30 liters per minute and up to 60 to 80 liters per minute)
with a set fraction of inspired oxygen (Fio2) of 0.21 to 1.0 can be administered
through a nasal cannula.7 Heating (to 34° to 37°C) and humidification make gas
delivery comfortable,8 and the high flow, usually higher than the patient’s own

1688 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

inspiratory peak flow, allows for the effective thus improving arterial oxygenation.20-22 When
delivery of the intended Fio2. In addition, high PEEP is applied to improve oxygenation, it is
flows generate a small level of nonadjustable important to clinically estimate its effect on the
PEEP — slightly higher when the mouth is kept work of breathing. Excessive PEEP can induce
closed — and reduce ventilation requirements hyperinflation and reduce the efficiency of dia-
and inspiratory muscle effort through a washout phragmatic contraction.
of the dead space in the upper airway during In cases of left ventricular dysfunction,19,23
expiration.9 This allows fresh gas with a con- PEEP may have beneficial effects by increasing
trolled Fio2 to be available at the beginning of the intrathoracic pressure and thus decreasing
each inspiration, thereby slightly reducing the preload, as well as by reducing the work of
quantity of ventilation that needs to be generat- breathing. By decreasing the negative intratho-
ed by the patient to clear carbon dioxide. The racic pressure swings generated by the activity of
use of a high-flow nasal cannula is often associ- the respiratory muscles, PEEP decreases the af-
ated with a prolonged expiration through a re- terload on the left ventricle. These effects are
sistive effect that reduces the respiratory rate.10,11 negligible in the context of normal cardiac func-
This method may also assist in mucociliary tion, but PEEP in the form of CPAP or noninva-
clearance of secretions through humidified sive ventilation can be very effective in relieving
gas,12-15 is easy to apply, and generally causes respiratory distress in patients with cardiogenic
minimal discomfort.16 pulmonary edema by improving both cardiac
and respiratory function.
CPAP and PEEP
With CPAP, the patient breathes with a constant Noninvasive Ventilation
level of positive pressure that is maintained dur- Noninvasive ventilation is a patient-triggered,
ing both inspiration and expiration.17 CPAP may pressure-targeted mode of ventilation in which
be deployed with intensive care mechanical ven- positive inspiratory pressure is delivered above a
tilators or with continuous-flow open circuits. PEEP level at each patient-triggered breath. In-
The latter have a high-gas-flow generator that spiratory pressure and PEEP are set by the health
allows the Fio2 to be adjusted up to 1.0 and a care team. The inspiratory pressure directly in-
PEEP valve with minimal resistance. CPAP can creases the tidal volume by raising the pressure
also be delivered through open-to-atmosphere gradient between the mouth and the alveoli, al-
valves that have internal microchannels through lowing the patient to reduce the required breath-
which a jet of oxygen is delivered.17 These sys- ing effort. Situations involving hypoventilation
tems can also be humidified.18 and respiratory acidosis are best treated with
Although PEEP does not have a direct effect noninvasive ventilation, which results in a sub-
on ventilation, it can indirectly act as assistance stantial reduction in work of breathing.24 De-
to ventilation through various mechanisms: creasing work of breathing can also decrease
counterbalancing the mechanical load imposed oxygen consumption and further improve gas
by residual end-expiratory alveolar pressure (dy- exchange.25,26 In patients with hypoxemic respi-
namic hyperinflation at the end of expiration) in ratory failure, the effect of positive inspiratory
COPD exacerbations, combating atelectasis (e.g., pressure needs to be monitored to ensure that it
postoperative hypoxemia after abdominal or does not lead to excessive tidal volumes, a situa-
thoracic surgery), providing a mechanical stent tion that is predictive of subsequent failure.27,28
of the upper airways (e.g., in patients with ob- This mode can be applied with the use of in-
structive sleep apnea), or acting as a threshold tensive care ventilators that compensate for leaks
external pressure to overcome the critical open- or by dedicated “bilevel” positive-airway-pres-
ing pressure of the airway (e.g., in patients with sure machines.29-31 Dedicated bilevel machines
obesity).19 In patients with hypoxemia, when have sophisticated algorithms to compensate for
impairment of oxygenation is secondary to a leaks, but they also have fewer monitoring capa-
loss of aerated alveoli (e.g., in the context of bilities than conventional ventilators. Leaks dur-
severe infectious pneumonia or acute respiratory ing noninvasive ventilation are dependent on the
distress syndrome) leading to intrapulmonary preset pressures and can make the delivery of
shunt, PEEP may facilitate alveolar recruitment, ventilation uncomfortable for the patient.17

n engl j med 387;18 nejm.org November 3, 2022 1689

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 The n e w e ng l a n d j o u r na l of m e dic i n e

CPA P a nd Nonin va si v e expired tidal volume is, however, not usually

V en t il at ion In ter face s feasible with standard helmet noninvasive venti-
lation.37
Implementation of noninvasive support involves
selecting the appropriate interface. The most Indic at ions a nd Cl inic a l
commonly used patient interface for CPAP and E v idence
noninvasive ventilation is the oronasal face mask,
which covers the nose and the mouth and is The indications and evidence from clinical stud-
secured firmly with head straps. Gas leaks ies for the use of different methods of noninva-
around the mask limit the efficacy of the device sive respiratory support are discussed below.
and may induce failure of this method, given the Figure 1 also summarizes the various clinical
inability to effectively deliver desired pressures30; conditions and associated evidence.
gas leaks also make monitoring of the tidal vol-
ume less precise. Oronasal face masks that are Acute Hypoxemic Respiratory Failure
tightly fitted to minimize leaks may result in Acute hypoxemic respiratory failure is often
facial ulcerations and cause discomfort, making characterized by a combination of lung inflam-
this method substantially less acceptable to pa- mation or infection, permeability pulmonary
tients — particularly with prolonged use. Total edema, and atelectasis resulting in impaired
face masks, which exert no direct pressure on oxygenation, ventilation, and respiratory me-
the nose, can be used with less skin breakdown, chanics.38 Invasive mechanical ventilation, deliv-
and their efficacy is similar to that of masks ered with targeted pressures and volumes to
with lower internal volumes.32 Despite their prevent ventilator-induced lung injury, is used in
larger internal volume, they rarely increase func- the context of worsening gas exchange39 and
tional dead space.32,33 Nasal mask interfaces are high effort to breathe. This invasive approach,
not commonly used in the acute care setting, however, often involves heavy sedation. Nonin-
given the limited pressures that can be delivered. vasive respiratory support may facilitate gas ex-
Comfort of the interface is critical for the use change while maintaining wakefulness and
of CPAP or noninvasive ventilation. Clinicians spontaneous breathing.40 Moreover, the sponta-
should ideally have a variety of interfaces and neous generation of negative intrathoracic pres-
sizes available so that individual patients’ needs sures can have beneficial effects on gas exchange
can be properly met. and the distribution of ventilation. However, at
The helmet is a larger interface for the deliv- least experimentally, prolonged exposure to vig-
ery of CPAP or noninvasive ventilation. It is a orous spontaneous breathing under conditions
cylinder-shaped hood made of transparent plas- of worsening severity can also be associated
tic that is fitted around the neck with a metal or with harms.41 In addition, the presence of non-
plastic ring and a soft collar. The helmet is fixed pulmonary organ dysfunction (most commonly
in place with two under-arm straps attached to brain or cardiovascular dysfunction) may neces-
the neck ring.34,35 Although some patients may sitate intubation to protect the airway and re-
experience claustrophobia or report excessive duce oxygen consumption. Despite uncertainty
amounts of noise, discomfort from the helmet is surrounding its effectiveness, noninvasive respi-
generally minimal, and it does not exert direct ratory support is used frequently with the hope
regional pressure on the skin. The helmet allows of reducing the need for intubation.4 Indeed,
for more prolonged use of CPAP or noninvasive during the coronavirus disease 2019 (Covid-19)
ventilation than does the oronasal face mask. pandemic, noninvasive ventilatory support has
Recent designs have improved the patient–venti- been delivered in up to 41% of patients with
lator interaction, allowing for higher levels of acute hypoxemic respiratory failure caused by
PEEP (10 to 12 cm of water) than with tradi- severe Covid-19; the types of support included a
tional interfaces.14,36 Higher levels of PEEP are high-flow nasal cannula, CPAP, and noninvasive
necessary to avoid collapse of the hood in pa- ventilation.42-44
tients with large tidal volumes. Furthermore, A large number of clinical trials have been
high gas flows may be necessary to avoid carbon performed, and a series of meta-analyses have
dioxide rebreathing. Accurate measurement of shown that the risk of endotracheal intubation is

1690 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

Before Invasive
Mechanical
Ventilation After Invasive Mechanical Ventilation

Prevention of To facilitate early In patients at risk for As rescue strategy

intubation extubation extubation failure (respiratory distress)

Cardiogenic Pulmonary Edema

COPD

Obesity

Mild-to-Moderate Acute Hypoxemic

Respiratory Failure
Moderate-to-Severe Acute Hypoxemic
Respiratory Failure

Preoxygenation during Intubation

After Surgery

Evidence of benefit Uncertainty of evidence No benefit or potential harm

Figure 1. Summary of Evidence for Noninvasive Ventilation across Acute Care Conditions.
The greatest benefit of preemptive use of noninvasive ventilation in the context of acute hypoxemic respiratory failure
is seen in selected high-risk patients (e.g., those with obesity or cardiac conditions). Helmet noninvasive ventilation
and therapy with a high-flow nasal cannula are under investigation for moderate-to-severe acute hypoxemic respira-
tory failure; the risk of failure is increased in patients who have a ratio of the partial pressure of arterial oxygen to
the fraction of inspired oxygen of less than 150 while receiving face-mask noninvasive ventilation. In the context of
extubation after surgery, no benefit has been found for preemptive use of continuous positive airway pressure after
abdominal surgery; however, a potential benefit of preemptive use of a high-flow nasal cannula has been found in
this context in higher-risk patients and patients with hypoxemia. A potential benefit of rescue noninvasive ventila-
tion after abdominal surgery has also been found. COPD denotes chronic obstructive pulmonary disease.

lower among patients with hypoxemic respira-

NIV use NIV failure
tory failure treated with either a high-flow nasal 50 47
cannula or noninvasive ventilation than among 45 42
those treated with conventional oxygen therapy.
40
However, no effect on mortality has been shown
35
consistently for patients with this indication.3,45
Frequency (%)

30
A landmark trial compared three methods
— high-flow nasal cannula, conventional oxy- 25 22
gen therapy, and face-mask noninvasive ventila- 20 17
tion with a high-flow nasal cannula used in be- 15 14 13
tween sessions — in patients with a ratio of the 10
partial pressure of arterial oxygen to the fraction 5
of inspired oxygen (Pao2:Fio2) of less than 300
0
and showed that the 90-day risk of death was Mild ARDS Moderate ARDS Severe ARDS
higher with conventional oxygen and noninva- (PaO2:FiO2 >200 to 300) (PaO2:FiO2 100 to 200) (PaO2:FiO2 <100)

sive ventilation than it was with the use of high-

Figure 2. Frequency of Use and Failure of Face-Mask Noninvasive Ventilation
flow nasal cannula alone.46 High tidal volumes
(NIV) for ARDS.
and low Pao2:Fio2 ratios (<200) were associated
Failure was defined as endotracheal intubation. The figure is based on data
with an increased risk of intubation (Fig. 2). from Bellani et al.4 ARDS denotes acute respiratory distress syndrome, Fio2
High tidal volumes 1 hour after the initiation of fraction of inspired oxygen, and Pao2 partial pressure of arterial oxygen.
noninvasive ventilation were associated with in-

n engl j med 387;18 nejm.org November 3, 2022 1691

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 The n e w e ng l a n d j o u r na l of m e dic i n e

creased mortality, which aroused concerns about In synthesizing the literature on acute hypox-
whether pressures delivered under noninvasive emic respiratory failure, we found that all non-
ventilation potentially precipitated ventilation- invasive respiratory support devices have been
induced lung injury. A network meta-analysis shown to potentially decrease the risk of endo-
comparing all methods of noninvasive respira- tracheal intubation more effectively than con-
tory support across 25 trials showed that all the ventional oxygen therapy, at least among pa-
methods were associated with a lower risk of tients with mild-to-moderate respiratory failure
intubation than conventional oxygen therapy.47 and in the absence of associated conditions (e.g.,
A small, single-center, randomized, con- severe organ failure or shock). However, the cur-
trolled trial evaluated different interfaces for rent evidence for a benefit of noninvasive respi-
noninvasive ventilation.36 A lower risk of intuba- ratory support for more severe forms of respira-
tion and lower 90-day mortality was found with tory failure (Pao2:Fio2 ratio <150) is less clear,
the use of a helmet than with the use of a face with some data suggesting potential risks asso-
mask for noninvasive ventilation. Although the ciated with face-mask noninvasive ventilation.4
trial was stopped early, the results were intrigu- Important uncertainties surround questions re-
ing, since randomized clinical trials evaluating garding whether to use noninvasive respiratory
these interfaces head-to-head had previously support, which device to use, the risk factors
been lacking. Patients in the helmet group ap- associated with failure, and how to monitor for
peared to have less discomfort from noninvasive failure in patients with higher severities of
ventilation at higher PEEP values (median, 8 cm hypoxemia. Table 1 outlines considerations in
of water) than patients in the face-mask group the selection of a noninvasive device for acute
(median PEEP, 5 cm of water).36 A trial evaluat- hypoxemic respiratory failure and the factors as-
ing a helmet for noninvasive ventilation as com- sociated with failure.
pared with high-flow nasal cannula was con-
ducted during the Covid-19 pandemic.14 The Cardiogenic Pulmonary Edema
patients who had been randomly assigned to In patients with cardiogenic pulmonary edema,
receive noninvasive ventilation with a helmet noninvasive respiratory support strategies are
received therapy with a high-flow nasal cannula used as bridging therapy during hypoxemia and
in between sessions of noninvasive ventilation. respiratory distress while urgent medical thera-
Lower rates of intubation (secondary outcome) pies (e.g., diuretics and vasodilators) are admin-
were found in the group assigned to receive istered; in this clinical context, noninvasive re-
noninvasive ventilation through a helmet than in spiratory support serves to decrease the work of
the group assigned to a high-flow nasal cannu- breathing, increase functional residual capacity,
la. Other trials involving patients with Covid-19 and enhance cardiac function. CPAP and nonin-
have shown CPAP and therapy with a high-flow vasive ventilation with a face mask have been
nasal cannula to be more effective in decreasing evaluated extensively in patients with cardio-
the risk of intubation than conventional oxygen genic pulmonary edema.55,56 A series of system-
therapy.48,49 atic reviews has shown a reduced risk of endo-
Immunocompromised patients have histori- tracheal intubation and reduced in-hospital
cally been deemed to be good candidates for mortality associated with these methods.3 In the
noninvasive respiratory support and in particular absence of shock or an indication for urgent
for noninvasive ventilation. This idea was driven revascularization, clinical practice guidelines
by early trials that were conducted when invasive recommend the use of CPAP or noninvasive ven-
ventilation was associated with extremely high tilation in this context. From a clinical stand-
mortality in this cohort.50,51 However, mortality point, when these patients present with both
among immunocompromised patients with re- hypoxemia and hypercapnia, it seems advisable
spiratory failure has decreased substantially over to use noninvasive ventilation as a first choice.
time.52 Given this change, the contemporary
data currently do not support avoiding invasive COPD Exacerbation and Hypercapnic
ventilation at all costs and do not support adopt- Respiratory Failure
ing a strategy that differs from that used for Noninvasive ventilation with a face mask has
nonimmunocompromised patients.47 been very effective in the context of COPD exacer-

1692 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

Table 1. Practical Considerations in the Management of Acute Respiratory Failure with Noninvasive Respiratory Support to Prevent Intubation.*

Consideration Recommendations and Common Practices Areas of Uncertainty

Where to monitor patient Patients should be treated in a monitored setting (e.g., The Covid-19 pandemic forced many institutions
step-down unit or ICU) by an experienced health to use HFNC outside the traditional setting of
care team with expertise in noninvasive respiratory a step-down unit or ICU. Thresholds for safe
support (e.g., physician, respiratory therapist, and use of HFNC therapy in the inpatient wards
nurse). The team should have knowledge of moni- have not been defined.
toring, adjustments of settings, identification of
failure, and urgent intubation.
Which method to use Use face-mask NIV for COPD, OHS, or congestive heart Criteria for selecting between HFNC and CPAP
failure exacerbation. or NIV are not clearly defined in the literature.
Consider intubation for acute hypoxemic respiratory fail- Treatment should first include the method the
ure (in the absence of an underlying chronic condi- health care team is most comfortable with, fol-
tion) in patients with high severity of illness, shock, lowed by an early evaluation of the presence of
acute kidney injury, decreased level of conscious- risk factors associated with failure.
ness, or severe hypoxemia.
Consider trial of HFNC or trial of CPAP or NIV for acute
hypoxemic respiratory failure (in the absence of an
underlying chronic condition) in patients with mild
or moderate hypoxemia.
Risk factors associated with Pao2:Fio2 <150 with presence of ARDS indicates high Promising evidence has suggested benefit (re-
failure risk of face-mask NIV failure and death.4 duced risk of intubation) with helmet NIV as
Pao2:Fio2 <200 1 hr after initiation of face-mask NIV, compared with face-mask NIV and HFNC in
particularly in the context of large tidal volumes patients with moderate hypoxemic respiratory
(>9 to 9.5 ml/kg),27,28 indicates high risk of face- failure. Ongoing clinical trials are under way.
mask NIV failure. Institutions with familiarity with helmet NIV
If above features are present, consider time-limited may consider its use in patients with moderate
trial of HFNC or helmet NIV (if there is institutional hypoxemia, particularly in the presence of high
familiarity) with frequent reevaluation (e.g., every work of breathing.
1 or 2 hr).
Measures to monitor for Monitor the respiratory rate, Pao2:Fio2 trajectory, level
failure of consciousness, and ROX index53 (HFNC) or
HACOR score54 (NIV).†

* ARDS denotes acute respiratory distress syndrome, COPD chronic obstructive pulmonary disease, CPAP continuous positive airway pressure,
Covid-19 coronavirus disease 2019, HFNC high-flow nasal cannula, ICU intensive care unit, NIV noninvasive ventilation, OHS obesity hypo­
ventilation syndrome, Pao2 partial pressure of arterial oxygen, and PEEP positive end-expiratory pressure.
† The ROX index is the ratio of the oxygen saturation divided by fraction of inspired oxygen (Fio2) to the respiratory rate. The HACOR scale is
based on heart rate, acidosis, consciousness, oxygenation, and respiratory rate.

bations, since it efficiently offloads respiratory dence supports the routine use of noninvasive
muscles and counteracts dynamic hyperinflation. ventilation in the context of asthma exacerba-
This method often prevents intubation as a bridge tions.57 Lastly, patients with obesity hypoventila-
to administering effective therapies (e.g., glucocor­ tion and mixed forms of respiratory failure rep-
ticoids, bronchodilators, and antibiotic agents). resent an increasingly large group of patients
A series of randomized, controlled trials have with hypercapnia and respiratory acidosis.58
evaluated the effectiveness of face-mask nonin- These patients may also benefit from both the
vasive ventilation as compared with conventional PEEP and the driving pressure of noninvasive
oxygen therapy for COPD exacerbations. Face- ventilation.59
mask noninvasive ventilation consistently showed
success in preventing intubation and decreasing After Extubation
hospital mortality among these patients.3 Non- Noninvasive respiratory support strategies have
invasive ventilation is therefore strongly recom- been evaluated to facilitate early liberation
mended as the first-line therapy for this popula- (weaning) from invasive mechanical ventilation,
tion. Currently, there is insufficient evidence to prevent extubation failure in high-risk pa-
surrounding the role of a high-flow nasal can- tients, and as a rescue strategy in acute respira-
nula for COPD exacerbations. Much less evi- tory failure after extubation (Fig. 1). Early libera-

n engl j med 387;18 nejm.org November 3, 2022 1693

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 The n e w e ng l a n d j o u r na l of m e dic i n e

tion from invasive ventilation through transitioning ure, noninvasive respiratory support could have
to face-mask noninvasive ventilation has been a promising role. As compared with usual care,
evaluated as a weaning strategy. This strategy the application of preemptive CPAP was not
has had great success among patients with shown to decrease a composite of pneumonia,
COPD specifically, with a meta-analysis showing endotracheal intubation, or death within 30 days
a reduced length of hospital stay and lower mor- after postoperative extubation in a recent large,
tality.60 randomized trial involving patients undergoing
Despite fulfilling criteria for successful extu- abdominal surgery.65 However, in a meta-analysis
bation, 12 to 20% of patients may be determined of 11 trials, preemptive use of a high-flow nasal
to need reintubation within the week after extu- cannula decreased the risk of intubation more
bation. As compared with conventional oxygen effectively than conventional oxygen therapy.
therapy, the application of noninvasive ventila- This benefit, however, was driven by higher-risk
tion or a high-flow nasal cannula immediately populations (e.g., patients with obesity).66 CPAP
after extubation has been successful in prevent- and noninvasive ventilation have been effective
ing reintubation in certain high-risk popula- in decreasing the incidence of reintubation and
tions, such as patients with COPD,60 coexisting complications in patients who have postextuba-
cardiac conditions, or obesity.61 In a comparative tion hypoxemia after abdominal surgery.67,68
evaluation, application of noninvasive ventilation
in combination with a high-flow nasal cannula Moni t or ing a nd Iden t ific at ion
in patients who were at risk for being reintu- of Fa ilur e a nd Sel ec t ion
bated was associated with a lower risk of rein- of De v ice s
tubation and postextubation respiratory failure
at day 7 than the use of a high-flow nasal can- Failure of noninvasive respiratory support is re-
nula alone.62 In a post hoc analysis, heterogene-
ported in only 15 to 20% of COPD exacerba-
ity in the treatment effect was found, with lower
tions69 but in up to 40 to 60% of cases of acute
risks of reintubation and death in the intensivehypoxemic respiratory failure.4,70 The likelihood
care unit (ICU) by day 7 associated with nonin- of failure increases with the severity of respira-
vasive ventilation than with a high-flow nasal tory failure (Fig. 2) and associated coexisting
cannula among patients with obesity (body-mass conditions. The decision to use noninvasive
index [BMI, the weight in kilograms divided by methods in patients with brain or circulatory
the square of the height in meters], ≥30) or over-
dysfunction should be made very cautiously, ex-
weight (BMI, 25 to 30) but not among patients cept when the dysfunction can be reversed by
of normal or lower-than-normal body weight.63 noninvasive ventilatory support.
The application of noninvasive respiratory Noninvasive ventilation failure has been
support as a rescue maneuver in the context of found to be an independent risk factor for death
postextubation acute respiratory failure has notin the ICU in patients with hypoxemic respira-
shown great success. The use of face-mask non- tory failure.4,71 In a secondary analysis of the
invasive ventilation in this context has been as-
trial evaluating helmet noninvasive ventilation as
sociated with delayed intubation and increased compared with a high-flow nasal cannula,14 pa-
mortality.3,61,64 These findings, however, may not
tients with a low partial pressure of arterial
be generalizable to patients with COPD exacer- carbon dioxide (Paco2) (<35 mm Hg) derived the
bations or cardiogenic pulmonary edema, since greatest benefit from helmet noninvasive ventila-
the trials of rescue noninvasive ventilation have
tion with respect to a decreased risk of intuba-
predominantly involved patients with pneumo- tion. This effect was not seen among patients
nia. Ultimately, close monitoring and frequent with a normal or higher Paco2 (≥35 mm Hg).
reevaluation is essential to monitor for failureThe authors postulated that the low Paco2 may
and avoid delaying reintubation. represent high inspiratory effort and may define
a subgroup of patients who are at greatest risk
Postoperative Respiratory Failure for patient self-inflicted lung injury during spon-
Given the prominent role of atelectasis or pul- taneous breathing.72 Although moderate-to-high
monary edema in postoperative respiratory fail- levels of PEEP can reduce the high inspiratory

1694 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

Table 2. Monitoring for Failure of Noninvasive Respiratory Support in Patients with Acute Hypoxemic Respiratory Failure.*

Variable Device Evaluated Description

Pao2:Fio2 4,27,28
Face-mask NIV Pao2:Fio2 <200 at 1 hr after NIV associated with increased risk of intubation; Pao2:Fio2
<150 associated with increased risk of death (as compared with up-front strategy of
­invasive mechanical ventilation)
Tidal volume4,27,28 Face-mask NIV Tidal volume >9 to 9.5 ml per kilogram of predicted body weight 1 hour after NIV associated
with increased risk of intubation and death
Respiratory rate53,54,75 Face-mask NIV Low or decreasing respiratory rate associated with greater likelihood of NIV success; respi-
ratory rate does not always correlate with inspiratory effort
Simplified Acute Face-mask NIV Higher scores indicate higher severity of illness, which is associated with higher likelihood
Physiology Score II27 of failure and receipt of invasive mechanical ventilation; no definitive threshold defined
in the literature
Composite scores
ROX index53 HFNC Tool for prediction of HFNC therapy failure and receipt of invasive mechanical ventilation,
validated in patients with acute hypoxemia due to pneumonia who were receiving HFNC
therapy; index evaluated at 2 hr, 6 hr, and 12 hr after initiation
HACOR score54 Face-mask NIV Evaluation of heart rate, acidosis, consciousness, oxygenation, and respiratory rate; thresh-
old of >5 at 1 hr after initiation of NIV associated with subsequent receipt of invasive
mechanical ventilation
Measures under evaluation
Paco272 Helmet NIV Possible surrogate for inspiratory effort; Paco2 <35 mm Hg associated with a greater like­
lihood of success with helmet NIV than with HFNC in reducing the risk of invasive me-
chanical ventilation (effect not seen when value is ≥35 mm Hg)
Changes in esophageal Helmet NIV Possible surrogate for inspiratory effort in patients with Pao2:Fio2 <200; lack of reduction
pressure at onset of in the change in esophageal pressure to <10 cm of water with application of helmet
inspiration76 NIV in patients with a baseline value of >10 cm of water associated with a higher risk
of intubation
Point-of-care lung ultra- Face-mask NIV, Lung aeration and morphologic abnormalities on ultrasonography quantified with a sim-
sound score77 HFNC plified protocol in six lung ultrasound areas and assigned score of 0 to 3 for each lung
area; the total lung ultrasound score was significantly higher in patients with Covid-19
who had HFNC or NIV failure leading to invasive mechanical ventilation

* The predominant cause of acute hypoxemic respiratory failure evaluated across these studies was pneumonia. Face masks may have been
an oronasal mask or a full mask. Paco2 denotes partial pressure of arterial carbon dioxide.

effort through reducing atelectasis and diaphrag- in esophageal pressure during inspiration has
matic effort and create a more homogeneous been investigated as a measure of inspiratory
delivery of pressures across the lungs,73 some effort and as an early warning signal.76 Compos-
patients may continue to generate large intratho- ite scores incorporating a combination of respi-
racic pressure swings and large tidal volumes, ratory variables and trends over time have also
which may ultimately lead to excessive work of shown potential for the identification of nonin-
breathing, oxygen consumption, cardiac over- vasive respiratory support failure. The ROX index
load, or patient self-inflicted lung injury.74 De- (the ratio of the oxygen saturation divided by
layed intubation may result in suboptimal intu- Fio2 to the respiratory rate) calculated at multi-
bating conditions, since these patients will have ple points after initiation of the use of a high-
little physiological reserve. flow nasal cannula has been associated with
A series of physiological variables have been failure of this mode of therapy in patients who
identified as being associated with failure of have acute respiratory failure in the absence of
noninvasive respiratory support, as shown in an underlying chronic condition.53 The HACOR
Table 2.4,27,28,76 Precise and reliable methods to scale (heart rate, acidosis, consciousness, oxy-
measure a threshold of inspiratory effort that is genation, and respiratory rate) has been evalu-
harmful remain under investigation; the change ated after 1 hour of face-mask noninvasive ven-

n engl j med 387;18 nejm.org November 3, 2022 1695

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 The n e w e ng l a n d j o u r na l of m e dic i n e

tilation to predict failure.54 Physiological factors A r e a s of Uncer ta in t y

associated with failure of the methods in pa-
tients with acute respiratory failure in the ab- Many questions remain with respect to the use
sence of an underlying chronic condition are of noninvasive respiratory support. The effective-
outlined in Table 2. ness of noninvasive respiratory support in com-
Given the risk of failure and the mortality bination with interventions such as prone posi-
associated with this risk, caution must be exer- tioning, the accurate measurement of inspiratory
cised when deciding to treat a patient with non- effort, the role of sedation in reducing harms
COPD, noncardiogenic, acute hypoxemic respi- associated with spontaneous breathing, and the
ratory failure. The approach to selecting a role of extracorporeal gas-exchange methods
noninvasive device, settings, adjustments, moni- coupled with noninvasive respiratory support to
toring, and decisions to transition to invasive avoid intubation are all areas in need of further
ventilation require a health care team–based ap- investigation.
proach. At many institutions, this team includes
respiratory therapists who have familiarity and C onclusions
expertise in the range of noninvasive devices and
interfaces, bedside nurses with experience in The different methods of noninvasive respiratory
managing respiratory failure, and physicians. support are important tools to support oxygen-
Specific factors regarding the patient, the physi- ation and ventilation across a variety of indica-
ological aspects of acute respiratory failure, the tions. Noninvasive respiratory support has a role
health care team, and institutional factors that in preventing intubation and decreasing mortality
may guide decision making regarding the selec- for patients with specific conditions. The benefit
tion of noninvasive respiratory support selection of averting intubation needs to be balanced
are outlined in Table S1. Table 1 provides an against the harms of delaying intubation, espe-
outline of considerations for the use of the dif- cially in patients with acute hypoxemic respira-
ferent noninvasive devices, which measures to tory failure. Early clinical identification of fail-
monitor, and factors that may be associated with ure is important to circumvent delayed intubation,
failure. It is likely that patients with different and therefore careful monitoring is required.
phenotypes respond differently to the noninva- Disclosure forms provided by the authors are available with
sive respiratory supports available. the full text of this article at NEJM.org.

References
1. Berthelsen PG, Cronqvist M. The first from bench studies to practical issues. Physiol (1985) 2022 May 05 (Epub ahead
intensive care unit in the world: Copenha- Minerva Anestesiol 2012;​78:​1146-53. of print).
gen 1953. Acta Anaesthesiol Scand 2003;​ 7. Drake MG. High-flow nasal cannula 12. Mauri T, Turrini C, Eronia N, et al.
47:​1190-5. oxygen in adults: an evidence-based as- Physiologic effects of high-flow nasal
2. Ranieri VM, Rubenfeld GD, Thomp- sessment. Ann Am Thorac Soc 2018;​15:​ cannula in acute hypoxemic respiratory
son BT, et al. Acute respiratory distress 145-55. failure. Am J Respir Crit Care Med 2017;​
syndrome: the Berlin Definition. JAMA 8. Richards GN, Cistulli PA, Ungar RG, 195:​1207-15.
2012;​307:​2526-33. Berthon-Jones M, Sullivan CE. Mouth leak 13. Möller W, Feng S, Domanski U, et al.
3. Rochwerg B, Brochard L, Elliott MW, with nasal continuous positive airway Nasal high f low reduces dead space.
et al. Official ERS/ATS clinical practice pressure increases nasal airway resistance. J Appl Physiol (1985) 2017;​122:​191-7.
guidelines: noninvasive ventilation for acute Am J Respir Crit Care Med 1996;​154:​182-6. 14. Grieco DL, Menga LS, Cesarano M,
respiratory failure. Eur Respir J 2017;​50:​ 9. Goligher EC, Slutsky AS. Not just oxy- et al. Effect of helmet noninvasive ventila-
1602426. gen? Mechanisms of benefit from high- tion vs high-flow nasal oxygen on days
4. Bellani G, Laffey JG, Pham T, et al. flow nasal cannula in hypoxemic respira- free of respiratory support in patients with
Noninvasive ventilation of patients with tory failure. Am J Respir Crit Care Med COVID-19 and moderate to severe hypox-
acute respiratory distress syndrome: in- 2017;​195:​1128-31. emic respiratory failure: the HENIVOT
sights from the LUNG SAFE study. Am J 10. Adams CF, Geoghegan PH, Spence CJ, randomized clinical trial. JAMA 2021;​
Respir Crit Care Med 2017;​195:​67-77. Jermy MC. Modelling nasal high flow 325:​1731-43.
5. Tobin MJ, Laghi F, Jubran A. Why therapy effects on upper airway resistance 15. Pinkham MI, Domanski U, Franke
COVID-19 silent hypoxemia is baffling to and resistive work of breathing. Respir K-J, et al. Effect of respiratory rate and
physicians. Am J Respir Crit Care Med Physiol Neurobiol 2018;​254:​23-9. size of cannula on pressure and dead-
2020;​202:​356-60. 11. Vieira F, Bezerra FS, Coudroy R, et al. space clearance during nasal high flow in
6. Sferrazza Papa GF, Di Marco F, Akou- High f low nasal cannula compared to patients with COPD and acute respiratory
mianaki E, Brochard L. Recent advances continuous positive airway pressure: failure. J Appl Physiol (1985) 2022;​132:​
in interfaces for non-invasive ventilation: a bench and physiological study. J Appl 553-63.

1696 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

16. Cuquemelle E, Pham T, Papon JF, Noninvasive positive-pressure ventilation. 43. COVID-ICU Group. Benefits and risks
Louis B, Danin PE, Brochard L. Heated N Engl J Med 2015;​372(23):​e30. of noninvasive oxygenation strategy in
and humidified high-flow oxygen therapy 30. Carteaux G, Lyazidi A, Cordoba-Izqui- COVID-19: a multicenter, prospective co-
reduces discomfort during hypoxemic re- erdo A, et al. Patient-ventilator asynchrony hort study (COVID-ICU) in 137 hospitals.
spiratory failure. Respir Care 2012;​ 57:​ during noninvasive ventilation: a bench Crit Care 2021;​25:​421.
1571-7. and clinical study. Chest 2012;​142:​367-76. 44. COVID-ICU Group. Clinical charac-
17. Carteaux G, Pons M, Morin F, et al. 31. Córdoba-Izquierdo A, Drouot X, Thille teristics and day-90 outcomes of 4244
Continuous positive airway pressure for AW, et al. Sleep in hypercapnic critical critically ill adults with COVID-19: a pro-
respiratory support during COVID-19 care patients under noninvasive ventila- spective cohort study. Intensive Care Med
pandemic: a frugal approach from bench tion: conventional versus dedicated venti- 2021;​47:​60-73.
to bedside. Ann Intensive Care 2021;​11:​ lators. Crit Care Med 2013;​41:​60-8. 45. Rochwerg B, Einav S, Chaudhuri D,
38. 32. Fraticelli AT, Lellouche F, L’her E, et al. The role for high flow nasal cannula
18. Thille AW, Bertholon JF, Becquemin Taillé S, Mancebo J, Brochard L. Physio- as a respiratory support strategy in adults:
M-H, et al. Aerosol delivery and humidifi- logical effects of different interfaces dur- a clinical practice guideline. Intensive
cation with the Boussignac continuous ing noninvasive ventilation for acute re- Care Med 2020;​46:​2226-37.
positive airway pressure device. Respir spiratory failure. Crit Care Med 2009;​37:​ 46. Frat J-P, Thille AW, Mercat A, et al.
Care 2011;​56:​1526-32. 939-45. High-flow oxygen through nasal cannula
19. Pinsky MR, Brochard LJ. CPAP to 33. Fodil R, Lellouche F, Mancebo J, et al. in acute hypoxemic respiratory failure.
counterbalance elevated pleural pressure Comparison of patient-ventilator inter- N Engl J Med 2015;​372:​2185-96.
in obese patients: restoring functional re- faces based on their computerized effec- 47. Ferreyro BL, Angriman F, Munshi L,
sidual capacity or simply keeping all air- tive dead space. Intensive Care Med 2011;​ et al. Association of noninvasive oxygen-
ways open? Chest 2021;​159:​2145-6. 37:​257-62. ation strategies with all-cause mortality
20. Nava S, Hill N. Non-invasive ventila- 34. Patroniti N, Foti G, Manfio A, Coppo in adults with acute hypoxemic respiratory
tion in acute respiratory failure. Lancet A, Bellani G, Pesenti A. Head helmet ver- failure: a systematic review and meta-
2009;​374:​250-9. sus face mask for non-invasive continuous analysis. JAMA 2020;​324:​57-67.
21. Florio G, De Santis Santiago RR, Fu- positive airway pressure: a physiological 48. Perkins GD, Ji C, Connolly BA, et al.
magalli J, et al. Pleural pressure targeted study. Intensive Care Med 2003;​29:​1680-7. Effect of noninvasive respiratory strate-
positive airway pressure improves cardio- 35. Chiumello D, Pelosi P, Carlesso E, et al. gies on intubation or mortality among
pulmonary function in spontaneously Noninvasive positive pressure ventilation patients with acute hypoxemic respiratory
breathing patients with obesity. Chest delivered by helmet vs. standard face failure and COVID-19: the RECOVERY-RS
2021;​159:​2373-83. mask. Intensive Care Med 2003;​29:​1671-9. randomized clinical trial. JAMA 2022;​327:​
22. Radermacher P, Maggiore SM, Mercat 36. Patel BK, Wolfe KS, Pohlman AS, Hall 546-58.
A. Fifty years of research in ARDS: gas JB, Kress JP. Effect of noninvasive ventila- 49. Ospina-Tascón GA, Calderón-Tapia LE,
exchange in acute respiratory distress tion delivered by helmet vs face mask on García AF, et al. Effect of high-flow oxygen
syndrome. Am J Respir Crit Care Med the rate of endotracheal intubation in pa- therapy vs conventional oxygen therapy
2017;​196:​964-84. tients with acute respiratory distress syn- on invasive mechanical ventilation and
23. Lenique F, Habis M, Lofaso F, Dubois- drome: a randomized clinical trial. JAMA clinical recovery in patients with severe
Randé JL, Harf A, Brochard L. Ventilatory 2016;​315:​2435-41. COVID-19: a randomized clinical trial.
and hemodynamic effects of continuous 37. Cortegiani A, Navalesi P, Accurso G, JAMA 2021;​326:​2161-71.
positive airway pressure in left heart fail- et al. Tidal volume estimation during hel- 50. Hilbert G, Gruson D, Vargas F, et al.
ure. Am J Respir Crit Care Med 1997;​155:​ met noninvasive ventilation: an experimen- Noninvasive ventilation in immunosup-
500-5. tal feasibility study. Sci Rep 2019;​9:​17324. pressed patients with pulmonary infil-
24. Brochard L, Mancebo J, Wysocki M, 38. Thompson BT, Chambers RC, Liu KD. trates, fever, and acute respiratory failure.
et al. Noninvasive ventilation for acute ex- Acute respiratory distress syndrome. N Engl J Med 2001;​344:​481-7.
acerbations of chronic obstructive pulmo- N Engl J Med 2017;​377:​1904-5. 51. Antonelli M, Conti G, Bufi M, et al.
nary disease. N Engl J Med 1995;​333:​817- 39. The Acute Respiratory Distress Syn- Noninvasive ventilation for treatment of
22. drome Network. Ventilation with lower acute respiratory failure in patients un-
25. Brochard L, Harf A, Lorino H, Le- tidal volumes as compared with tradition- dergoing solid organ transplantation:
maire F. Inspiratory pressure support pre- al tidal volumes for acute lung injury and a randomized trial. JAMA 2000;​283:​235-
vents diaphragmatic fatigue during wean- the acute respiratory distress syndrome. 41.
ing from mechanical ventilation. Am Rev N Engl J Med 2000;​342:​1301-8. 52. Azoulay E, Pickkers P, Soares M, et al.
Respir Dis 1989;​139:​513-21. 40. Antonelli M, Conti G, Rocco M, et al. Acute hypoxemic respiratory failure in
26. Brochard L, Isabey D, Piquet J, et al. A comparison of noninvasive positive- immunocompromised patients: the Efraim
Reversal of acute exacerbations of chronic pressure ventilation and conventional me- multinational prospective cohort study.
obstructive lung disease by inspiratory as- chanical ventilation in patients with acute Intensive Care Med 2017;​43:​1808-19.
sistance with a face mask. N Engl J Med respiratory failure. N Engl J Med 1998;​ 53. Roca O, Caralt B, Messika J, et al. An
1990;​323:​1523-30. 339:​429-35. index combining respiratory rate and oxy-
27. Carteaux G, Millán-Guilarte T, De 41. Yoshida T, Fujino Y, Amato MB, Kava- genation to predict outcome of nasal
Prost N, et al. Failure of noninvasive ven- nagh BP. Fifty years of research in ARDS: high-flow therapy. Am J Respir Crit Care
tilation for de novo acute hypoxemic re- spontaneous breathing during mechani- Med 2019;​199:​1368-76.
spiratory failure: role of tidal volume. Crit cal ventilation. risks, mechanisms, and 54. Duan J, Han X, Bai L, Zhou L, Huang
Care Med 2016;​44:​282-90. management. Am J Respir Crit Care Med S. Assessment of heart rate, acidosis, con-
28. Frat J-P, Ragot S, Coudroy R, et al. Pre- 2017;​195:​985-92. sciousness, oxygenation, and respiratory
dictors of intubation in patients with 42. RECOVERY Collaborative Group. Toci­ rate to predict noninvasive ventilation
acute hypoxemic respiratory failure treat- lizumab in patients admitted to hospital failure in hypoxemic patients. Intensive
ed with a noninvasive oxygenation strat- with COVID-19 (RECOVERY): a random­ Care Med 2017;​43:​192-9.
egy. Crit Care Med 2018;​46:​208-15. ised, controlled, open-label, platform trial. 55. Bersten AD, Holt AW, Vedig AE,
29. Kelly CR, Higgins AR, Chandra S. Lancet 2021;​397:​1637-45. Skowronski GA, Baggoley CJ. Treatment

n engl j med 387;18 nejm.org November 3, 2022 1697

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.
 Review Article

of severe cardiogenic pulmonary edema tubation failure: a randomized clinical 70. Menga LS, Berardi C, Ruggiero E,
with continuous positive airway pressure trial. JAMA 2019;​322:​1465-75. Grieco DL, Antonelli M. Noninvasive re-
delivered by face mask. N Engl J Med 63. Thille AW, Coudroy R, Nay MA, et al. spiratory support for acute respiratory
1991;​325:​1825-30. Beneficial effects of noninvasive ventila- failure due to COVID-19. Curr Opin Crit
56. Masip J, Betbesé AJ, Páez J, et al. Non- tion after extubation in obese or over- Care 2022;​28:​25-50.
invasive pressure support ventilation ver- weight patients: a post hoc analysis of a 71. Demoule A, Girou E, Richard JC,
sus conventional oxygen therapy in acute randomized clinical trial. Am J Respir Taille S, Brochard L. Benefits and risks of
cardiogenic pulmonary oedema: a ran- Crit Care Med 2022;​205:​440-9. success or failure of noninvasive ventila-
domised trial. Lancet 2000;​356:​2126-32. 64. Esteban A, Frutos-Vivar F, Ferguson tion. Intensive Care Med 2006;​32:​1756-
57. Althoff MD, Holguin F, Yang F, et al. ND, et al. Noninvasive positive-pressure 65.
Noninvasive ventilation use in critically ill ventilation for respiratory failure after ex- 72. Grieco DL, Menga LS, Cesarano M, et al.
patients with acute asthma exacerbations. tubation. N Engl J Med 2004;​350:​2452-60. Phenotypes of patients with COVID-19
Am J Respir Crit Care Med 2020;​202:​1520- 65. PRISM trial group. Postoperative con- who have a positive clinical response to
30. tinuous positive airway pressure to pre- helmet noninvasive ventilation. Am J Respir
58. Adler D, Pépin J-L, Dupuis-Lozeron E, vent pneumonia, re-intubation, and death Crit Care Med 2022;​205:​360-4.
et al. Comorbidities and subgroups of pa- after major abdominal surgery (PRISM): 73. Morais CCA, Koyama Y, Yoshida T, et al.
tients surviving severe acute hypercapnic a multicentre, open-label, randomised, High positive end-expiratory pressure ren-
respiratory failure in the intensive care phase 3 trial. Lancet Respir Med 2021;​9:​ ders spontaneous effort noninjurious. Am
unit. Am J Respir Crit Care Med 2017;​196:​ 1221-30. J Respir Crit Care Med 2018;​197:​1285-96.
200-7. 66. Chaudhuri D, Granton D, Wang DX, 74. Brochard L, Slutsky A, Pesenti A. Me-
59. Mokhlesi B, Masa JF, Brozek JL, et al. et al. High-flow nasal cannula in the im- chanical ventilation to minimize progres-
Evaluation and management of obesity mediate postoperative period: a system- sion of lung injury in acute respiratory
hypoventilation syndrome: an official atic review and meta-analysis. Chest 2020;​ failure. Am J Respir Crit Care Med 2017;​
American Thoracic Society clinical prac- 158:​1934-46. 195:​438-42.
tice guideline. Am J Respir Crit Care Med 67. Squadrone V, Coha M, Cerutti E, et al. 75. Yoshida Y, Takeda S, Akada S, Hongo
2019;​200(3):​e6-e24. Continuous positive airway pressure for T, Tanaka K, Sakamoto A. Factors pre-
60. Burns KEA, Stevenson J, Laird M, et al. treatment of postoperative hypoxemia: dicting successful noninvasive ventilation
Non-invasive ventilation versus invasive a randomized controlled trial. JAMA 2005;​ in acute lung injury. J Anesth 2008;​22:​
weaning in critically ill adults: a system- 293:​589-95. 201-6.
atic review and meta-analysis. Thorax 68. Jaber S, Lescot T, Futier E, et al. Effect 76. Grieco DL, Menga LS, Raggi V, et al.
2022;​77:​752-61. of noninvasive ventilation on tracheal re- Physiological comparison of high-flow
61. Fernando SM, Tran A, Sadeghirad B, intubation among patients with hypox- nasal cannula and helmet noninvasive
et al. Noninvasive respiratory support emic respiratory failure following abdom- ventilation in acute hypoxemic respiratory
following extubation in critically ill adults: inal surgery: a randomized clinical trial. failure. Am J Respir Crit Care Med 2020;​
a systematic review and network meta-analy­ JAMA 2016;​315:​1345-53. 201:​303-12.
sis. Intensive Care Med 2022;​48:​137-47. 69. Jolliet P, Ouanes-Besbes L, Abroug F, 77. Biasucci DG, Buonsenso D, Piano A,
62. Thille AW, Muller G, Gacouin A, et al. et al. A multicenter randomized trial as- et al. Lung ultrasound predicts non-inva-
Effect of postextubation high-flow nasal sessing the efficacy of helium/oxygen in sive ventilation outcome in COVID-19 acute
oxygen with noninvasive ventilation vs severe exacerbations of chronic obstruc- respiratory failure: a pilot study. Minerva
high-flow nasal oxygen alone on reintu- tive pulmonary disease. Am J Respir Crit Anestesiol 2021;​87:​1006-16.
bation among patients at high risk of ex- Care Med 2017;​195:​871-80. Copyright © 2022 Massachusetts Medical Society.

1698 n engl j med 387;18 nejm.org November 3, 2022

Downloaded from nejm.org by I can't believe they gave me a medical degree on November 4, 2022. For personal use only. No other uses without permission.