Critical care

Original research

Non-­invasive ventilation versus invasive weaning in

critically ill adults: a systematic review and meta-­
analysis
Karen E A Burns  ‍ ‍,1,2,3,4,5 James Stevenson,6 Matthew Laird,6 Neill K J Adhikari,1,7,8
Yuchong Li,2,5 Cong Lu,2,5 Xiaolin He,5 Wentao Wang,9 Zhenting Liang,8 Lu Chen,2,5
Haibo Zhang,1,2,5,10 Jan O Friedrich1,2,3,5

► Additional supplemental ABSTRACT

material is published online Background  Extubation to non-­invasive ventilation Key messages
only. To view, please visit the
journal online (http://​dx.​doi.​ (NIV) has been investigated as a strategy to wean
org/1​ 0.​1136/​thoraxjnl-​2021-​ critically ill adults from invasive ventilation and reduce What is the key question?
216993). ventilator-r­ elated complications. ⇒ What is the effect of non-­invasive and invasive
Methods  We searched MEDLINE, EMBASE, the weaning in randomised or quasi-­randomised
For numbered affiliations see trials of critically ill adults on mortality (primary
Cochrane Central Register of Controlled Trials,
end of article.
proceedings of four conferences and bibliographies (to outcome), pneumonia, weaning failures and
June 2020) for randomised and quasi-­randomised trials other clinically important outcomes?
Correspondence to
Dr Karen E A Burns, Critical that compared extubation with immediate application
What is the bottom line?
Care, Unity Health Toronto, of NIV to continued invasive weaning in intubated
Toronto, Canada; ⇒ High-­quality evidence suggests that non-­
adults and reported mortality (primary outcome) or
​karen.​burns@​unityhealth.​to invasive (vs invasive) weaning significantly
other outcomes. Two reviewers independently screened
reduced mortality, pneumonia and the
Received 5 February 2021 citations, assessed trial quality and abstracted data.
proportion of weaning failures, and subgroup
Accepted 15 September 2021 Results  We identified 28 trials, of moderate-­to-­good
analyses found that non-­invasive weaning
Published Online First quality, involving 2066 patients, 44.6% with chronic
29 October 2021 reduced mortality, pneumonia, reintubation
obstructive pulmonary disease (COPD). Non-­invasive
and intensive care unit length of stay in trials
weaning significantly reduced mortality (risk ratio (RR)
enrolling chronic obstructive pulmonary disease
0.57, 95% CI 0.44 to 0.74; high quality), weaning
(COPD) (vs mixed) populations.
failures (RR 0.59, 95% CI 0.43 to 0.81; high quality),
pneumonia (RR 0.30, 95% CI 0.22 to 0.41; high Why read on?
quality), intensive care unit (ICU) (mean difference (MD) ⇒ Pooled data support the net clinical benefits
−4.62 days, 95% CI −5.91 to −3.34) and hospital stay associated with the non-­invasive approach
(MD −6.29 days, 95% CI −8.90 to −3.68). Non-­invasive to weaning on important clinical outcomes,
weaning also significantly reduced the total duration of particularly in patients with COPD, while more
ventilation, duration of invasive ventilation and duration careful patient selection is required in non-­
of ventilation related to weaning (MD −0.57, 95% CI COPD patients.
−1.08 to −0.07) and tracheostomy rate. Mortality,
pneumonia, reintubation and ICU stay were significantly
lower in trials enrolling COPD (vs mixed) populations. health-­related quality of life,4 5 post-­traumatic stress
Conclusion  Non-­invasive weaning significantly reduced disorder,6 7 depression,8 delirium9 and cognitive
mortality, pneumonia and the duration of ventilation impairment.10 Consequently, minimising patients’
related to weaning, particularly in patients with COPD. exposure to invasive mechanical ventilation has been
Beneficial effects are less clear (or more careful patient
identified as a key research priority by critical care
selection is required) in non-­COPD patients. societies.11
PROSPERO registration number  CRD42020201402. Non-­invasive ventilation (NIV), administered with
a patient–ventilator interface, can provide partial
ventilator support to patients with ARF without the
need for an artificial airway. Interfaces include nasal
INTRODUCTION mask, oronasal mask, full face masks, mouthpiece,
Mechanical ventilation is often required to support nasal pillows and helmets.12 NIV does not provide
© Author(s) (or their critically ill patients with acute respiratory failure airway protection. Similar to invasive ventilation,
employer(s)) 2022. No (ARF). Although life-­ saving, invasive ventilation is NIV can augment tidal volumes, reduces respiratory
commercial re-­use. See rights associated with ventilator-­ related complications, rates, apply positive end-­ expiratory pressure and
and permissions. Published such as ventilator-­associated pneumonia (VAP). VAP, improve gas exchange.12 13 Unlike invasive mechanical
by BMJ.
in turn, is associated with attributable mortality of ventilation, NIV preserves patient’s ability to cough,
To cite: Burns KEA, approximately 13%.1 Cumulative exposure to inva- swallow and speak.14 Patients who are treated with
Stevenson J, Laird M, et al. sive mechanical ventilation is associated with long-­ NIV may receive less invasive monitoring and seda-
Thorax 2022;77:752–761. term sequelae including muscle weakness,2 3 reduced tion15 and experience less psychological distress.16
752   Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993
 Critical care
For these reasons, investigators have studied NIV as a method to we included the more protracted measure in pooled analyses (eg,
reduce patient’s exposure to invasive mechanical ventilation during mortality).
weaning. With this technique, patients who do not meet conven- We evaluated the impact of statistical heterogeneity among
tional criteria for extubation (ie, fail a spontaneous breathing trial pooled studies for each outcome using the Cochran Q statistic
(SBT) or are not ready to undergo an SBT) are extubated directly (threshold p<0.10)19 20 and the I2 statistic21 22 with threshold
to NIV. Non-­invasive support is then reduced over time, minimising values of 0%–40%, 30%–60%, 50%–90% and >75% representing
patients’ exposure to invasive ventilation and related complications. heterogeneity that might not be important or represent moderate,
To better understand the net clinical benefits associated with substantial or considerable heterogeneity, respectively.22 If a hetero-
non-­invasive weaning, we sought to critically appraise, summarise geneity value overlapped two categories, we assigned it the higher
and update a prior systematic review and meta-­analysis17 of the heterogeneity rating. In sensitivity analyses, we assessed the impact
effect of non-­invasive weaning compared with invasive weaning on of excluding quasi-­randomised trials on estimates of mortality and
important outcomes in light of new evidence. VAP. We planned subgroup analyses to compare the effects of non-­
invasive weaning on mortality and weaning failures in studies of
METHODS patients with chronic obstructive pulmonary disease (COPD) only
Data sources and searches compared with mixed populations (any non-­COPD) and in studies
We updated a previously conducted search of MEDLINE (January that enrolled >50% versus <50% patients with COPD. We assessed
1966 to July 2021), EMBASE (January 1980 to July 2021) via for differences between subgroup summary estimates using the χ2
Ovid SP and the Cochrane Central Register of Controlled Trials test.23 We examined funnel plots visually for evidence of publica-
(CENTRAL) (the Cochrane Library 2012, July 2021) without tion bias. We used the principles of grades of recommendation,
language restrictions. Details of the search strategies used are assessment, development, and evaluation (GRADE)24 to assess the
provided in online supplemental appendix 1. Two reviewers (JS, quality of the body of evidence associated with specific outcomes
ML) independently screened citation titles and abstracts. One (mortality, weaning failure, VAP, duration of ventilation related to
reviewer (JOF) updated manual searches of abstracts from confer- weaning and reintubation).
ence proceedings published in the American Journal of Respiratory
and Critical Care Medicine, Intensive Care Medicine, Critical Care RESULTS
Medicine and Chest from June 2013 to August 2020. We reviewed Trial identification
bibliographies of retrieved articles to identify potentially relevant We previously identified 16 trials25–40 meeting our eligibility criteria.
trials, contacted authors to obtain additional information regarding In updated searches of 971 records (online supplemental appendix
study methods where needed and searched for ongoing trials on 1), we assessed 20 new articles for eligibility (figure 1). Of these, we
www.isrctn.com and www.clinicaltrials.gov. included 12 new trials41–52 that evaluated NIV as a weaning strategy.
One trial28 was previously assessed to be quasi-­randomised. We were
unable to clarify whether another trial42 reported as ‘randomised’
Study selection
was truly randomised or quasi-­ randomised. One trial did not
We included randomised and quasi-­ randomised trials that (1)
include outcomes data by group and only contributed data to the
enrolled adults with respiratory failure who required invasive
quality assessment.41 In summary, we included 28 trials reporting on
mechanical ventilation for at least 24 hours but before they met
2066 patients in our updated review.
conventional criteria for extubation, (2) compared extubation with
Of the 28 included trials, 4 trials were published only in abstract
immediate application of NIV to continue the ventilation weaning
form,27 30 41 43 8 trials were published in Chinese,28 31 33 34 42 48 49 52
process with continued invasive weaning and (3) reported at least
1 trial was a dissertation subsequently published in full36 and
one of mortality (primary outcome), VAP, weaning failure (using
another was a pilot randomised controlled trial (RCT).40 We
authors’ definitions), intensive care unit (ICU) or hospital length of
excluded 19 studies53–71 including 8 newly excluded publications
stay, total duration of ventilation, duration of ventilation related to
(figure 1).72–79 The two reviewers (JS, ML) achieved complete
weaning, duration of invasive ventilation, adverse events or quality
agreement on study selection.
of life. We excluded trials that compared non-­invasive with inva-
Fourteen trials included exclusively COPD (n=922) patients
sive weaning in the immediate postoperative setting (typically with
(table 1).25 28 30–34 36 38 42 43 45 47 52 Of non-­COPD trials, COPD
less than 24 hours of invasive ventilation), compared NIV with
was diagnosed in approximately 75% of patients in three
unassisted oxygen supplementation and investigated NIV after
trials,26 29 37 approximately one-­third of patients in two trials,27 35
unplanned extubation.
over 20% of patients in another trial39 46 and unspecified in a
final trial.41 Three trials excluded patients with COPD40 49 51 and
Data abstraction and risk of bias assessment one trial each focused exclusively on specific patient populations
Two unblinded authors (JS, ML) abstracted data regarding risk including after cardiac surgery,44 elderly patients with severe
of bias (randomisation, allocation concealment, completeness of community-­acquired pneumonia49 and adults with acute respira-
follow-­up, selective outcomes reporting) using a standardised tory distress syndrome48 or hypoxemic respiratory failure.51 The
data abstraction form. We did not formally assess blinding in largest trial stratified on the presence of COPD but included less
these necessarily unblinded trials. Disagreements regarding than 4% of patients with COPD.50
study selection and data abstraction were resolved by consensus Patients were considered difficult to wean in two trials26 37 and
and arbitration with a third author (KEAB). persistent weaning failures in another trial.29 Five trials31–34 43
included COPD patients with respiratory failure due to pulmo-
Data synthesis and statistical analysis nary infection. No included trials evaluated high-­ flow nasal
We pooled data across studies using random effects models. We cannulae in the intervention group.
derived summary estimates of risk ratio (RR) and mean differ-
ence (MD) with 95% CIs for binary and continuous outcomes, Quality assessment
respectively, using Review Manager V.5.4 (Cochrane Collabora- Overall, the quality of the included trials was moderate-­to-­good
tion, Oxford).18 If an outcome was reported at two different times, (figure 2). In most trials, allocation to treatment group was
Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993 753
 Critical care

Figure 1  Trial selection process. This review represents an update of a previously conducted systematic review and meta-­analysis.16

by random assignment with one quasi-­randomised confirmed to 0.74; p≤0.0001; high quality) with moderate heterogeneity
to allocate patients according to hospital admission order.28 (I2=27%) (figure 3).
We judged allocation concealment to be adequate in 13
trials,25 27 29 30 37–40 44 47 50–52 unclear in 14 trials26 31–36 41–43 45 46 48 49 Secondary outcomes
and inadequate in a quasi-­randomised trial.28 In three trials,33 34 41 Eleven trials involving 829 patients, using variable defi-
denominators were not provided in binary outcomes to ensure nitions, reported the proportion of patients successfully
complete outcomes reporting. One trial49 was deemed to be at weaned.25–27 30 37–40 43 46 51 The pooled data demonstrated a
high risk of bias due to incomplete outcomes reporting as >5% significant reduction in the proportion of weaning failures using
of the patients (4 NIV arm, 1 control arm) were excluded post non-­invasive weaning (RR 0.59, 95% CI 0.43 to 0.81; p=0.001;
randomisation. With regard to selective outcomes reporting high quality) with low heterogeneity (I2=22%).
two trials were deemed to be at high risk of bias29 41 and three Pooled data from 23 trials involving 1581 (increase of 628)
trials were deemed to be at unclear risk of bias28 44 45 (online patients25 26 28–39 42 43 45–49 51 52 that reported VAP for which criteria
supplemental appendix 2). On inspection of a funnel plot for for the diagnosis were provided in 10 trials25 28 29 31–36 39 42 45–48 50 51
the primary outcome, we noted the potential absence of small showed that non-­invasive weaning was associated with decreased
negative trials (online supplemental appendix 3). VAP (RR 0.30, 95% CI 0.22 to 0.41; p<0.00001; high quality)
with low heterogeneity (I2=24%) (figure 4).
Primary outcome: mortality Meta-­analysis also supported that non-­ invasive weaning
Twenty-­seven trials involving 2042 patients (increase of significantly reduced ICU (MD −4.62 days, 95% CI −5.91 to
1048) provided mortality data. Mortality was reported at −3.34; p<0.00001, I2=80%) and hospital (MD −6.29 days,
30 days,36 39 45 50 60 days,25 49 90 days,26 29 50 180 days,50 95% CI −8.90 to −3.68; p<0.00001, I2=75%) stay, total dura-
ICU37 40 46 47 50–52 and hospital discharge26 30 32–35 38 40–44 46 50 51 tion of mechanical ventilation (MD −5.26 days, 95% CI −7.86
and at undefined times.27 28 31 48 There was strong evidence that to −2.67; p<0.0001, I2=83%) and duration of invasive venti-
non-­invasive weaning reduced mortality (RR 0.57, 95% CI 0.44 lation (MD −7.75 days, 95% CI −9.86 to −5.64; p<0 00 001,
754 Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993
 Critical care

Table 1  Populations and interventions in studies of non-­invasive ventilation (NIV) in critically ill adults
No. of Inclusion criteria
Study patients Inclusion criteria (patients) (weaning eligibility) Experimental strategy Control strategy
25
Nava et al 50 Exacerbation of COPD. Intubated for at least Simple weaning criteria, 1 hour SBT Non-­invasive PS on conventional Invasive PS
36–48 hours failure ventilator delivered with face
mask
Girault et al26 33 Acute-­on-­chronic respiratory failure (COPD, Simple weaning criteria, 2 hours SBT Flow or pressure mode with Flow or pressure
restrictive or mixed populations). Intubated for failure nasal or face mask mode (PS)
at least 48 hours
Hill et al27 21 Acute respiratory failure (ARF) 30 min SBT failure NIV using VPAP in ST-­A mode Invasive PS
28
Chen et al 24 Exacerbation of COPD. Intubated for at least Day 3+ weaning criteria Bilevel NIV (pressure mode)  Invasive PS
48–60 hours. Saturations>88% on FiO2 40%
Ferrer et al29 43 ARF and persistent weaning failure. Intubated 2 hours SBT failure on 3 consecutive Bilevel NIV in ST mode delivered AC or invasive PS
for at least 72 hours days with face or nasal mask
Rabie et al30 37 Exacerbation of COPD 2 hours SBT failure NIV (proportional assist in timed Invasive PS
mode) delivered by face or nasal
mask
Wang et al31 28 COPD. Bronchopulmonary infection PIC window NIV (pressure mode) delivered SIMV+PS
by mask (unspecified)
Wang et al32 90 COPD with severe hypercapnic respiratory PIC window Bilevel NIV (pressure mode) SIMV+PS
failure. Pneumonia or purulent bronchitis.
Age≤85 years. Capable of self-­care in past year
Zheng et al33 33 COPD. Severe pulmonary infection PIC window Bilevel NIV (pressure mode) Invasive PS
delivered by face or nasal mask
Zou et al34 76 COPD with severe respiratory failure. PIC window Bilevel NIV (pressure, ST mode) SIMV +PS
Pulmonary infection delivered by nasal or oronasal
mask
Trevisan et al35 65 Invasively ventilated >48 hours 30 min SBT failure Bilevel NIV (pressure mode) Invasive MV
delivered by face mask
Prasad et al36 30 COPD. Hypercapnic respiratory failure 2 hours SBT failure Bilevel NIV (pressure mode) Invasive PS
delivered by full face mask
Girault et al37 138 Chronic hypercapneic respiratory failure 2 hours SBT failure Non-­invasive PS±PEEP or bilevel Invasive PS with
invasively ventilated for at least 48 hours NIV with face mask (initial once daily SBT with
choice) T-­piece or PS±PEEP
Rabie Agmy et al38 264 Acute on chronic exacerbation of COPD 2 hours SBT failure NIV (pressure, ST mode) Invasive PS
Tawfeek and Ali-­ 42 Invasively ventilated for >48 hours 2 hours SBT failure Non-­invasive PAV ventilation SIMV
Elnabtity39 delivered by face mask
Vaschetto et al40 20 Hypoxemic respiratory failure invasively PS with PEEP +inspiratory support, Helmet NIV Invasive PS with
ventilated for at least 48 hours <25 cmH2O and PEEP 8–13 cmH2O; SBT when P/F
PaO2/FiO2 200–300 mm Hg with ratio >250 mm Hg
FiO2≤0.6
Charra et al41 24 Invasively ventilated for greater than 48 hours T-­piece SBT failure Face mask bilevel NIV Classic wean with
in medical intensive care unit invasive PS
Rong42 64 COPD with respiratory failure PIC window Bilevel positive airway pressure SIMV+PS
Mohamed and 30 COPD with infective exacerbation Not reported NIV Invasive MV
Ibrahim43
Laiq et al44 60 Cardiac surgery patients invasively ventilated 30 min T-­piece SBT failure NIPPV Invasive MV with
greater than 48 hours daily SBT
El-­Shimy et al45 40 COPD on MV 30 min–2 hours T-­piece SBT failure NIV (BiPAP) SIMV+PS
46
Carron et al 64 Intubated for  ARF≥48 hours 30 min PSV SBT failure Helmet NIV PSV weaning
Mishra et al47 50 COPD exacerbation requiring invasive MV for T-­piece SBT failure Full face mask NIPPV (BiPAP) PSV
at least 48 hours
Wang et al48 53 Surgical patients requiring invasive MV for PaO2/FiO2 200–250 mm Hg with NIV Invasive MV with
acute respiratory distress syndrome PEEP 8 and PS 12 cmH2O, acute daily SBT
infiltrates resolved
Guo et al49 96 (5 Age≥75 years and community-­acquired Failed SBT NIV PSV weaning with
excluded) pneumonia requiring MV SBT
Perkins et al50 364 Invasive MV for >48 hours SBT failure Face mask NIV Invasive MV with
daily SBT

Continued

Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993 755

 Critical care

Table 1  Continued
No. of Inclusion criteria
Study patients Inclusion criteria (patients) (weaning eligibility) Experimental strategy Control strategy
Vaschetto et al51 130 Hypoxemic respiratory failure invasively PS with PEEP +inspiratory support, Oral-­nasal or full face mask NIV Invasive PS with
ventilated for at least 48 hours <25 cmH2O and PEEP 8–13 cmH2O; 30 min PS 5/PEEP
PaO2/FiO2 200–300 mm Hg with 5 SBT when P/F
FiO2≤0.6; PaCO2 ≤50 mm Hg and ratio >250 mm Hg
pH ≥7.35,  risk ratio≤30, Vt≤8 mL/kg
IBW, T<38.5°C, GCS 10T, suction <2/
hour
 Chen et al52 106 COPD exacerbation requiring invasive MV Criteria for moving to NIV in NIV Invasive MV until
intervention group not clear successful SBT
AC, assist control; BiPAP, bilevel positive airway pressure; COPD, chronic obstructive pulmonary disease; GCS, Glasgow Coma Scale; IBW, ideal body weight; MV, mechanical
ventilation; PEEP, postive end-­expiratory pressure; P/F, partial pressure of oxygen/fractional concentration of oxygen; PIC, pulmonary infection control; PS, pressure support; PSV,
pressure support ventilation; SBT, spontaneous breathing trial; SIMV, synchronised intermittent mandatory ventilation; ST, spontaneous timed; ST-­A, spontaneous timed; VPAP,
variable positive airway pressure; Vt, tidal volume.

I2=94%), all with considerable heterogeneity. Non-­ invasive than 50% patients with COPD. Similarly, we found significant
weaning significantly reduced the duration of mechanical venti- differences between subgroups evaluating non-­invasive weaning
lation related to weaning (MD −0.57 days, 95% CI −1.08 to on VAP and ICU length of stay in COPD versus mixed popula-
−0.07; p=0.03; moderate quality due to inconsistency with tions. A post hoc subgroup analysis demonstrated significantly
I2=88%) favouring non-­invasive weaning (figure 5). No study lower reintubation rate (p=0.02) in COPD trials (RR 0.48,
reported quality of life (table 2). 95% CI 0.34 to 0.67) versus mixed patient populations (RR
0.89. 95% CI 0.59 to 1.35) (table 3).
Adverse events
The pooled result showed no difference in arrhythmias (RR DISCUSSION
0.70, 95% CI 0.41, 1.20; four trials, 565 patients),26 36 37 50 In this updated systematic review, we identified 28 trials
non-­significantly lower reintubation (RR 0.69, 95% CI 0.47 to (2066 patients) enrolling mechanically ventilated patients that
1.01; 14 trials, moderate quality due to inconsistency, 1336 compared weaning with extubation to non-­invasive ventilation
patients)26–29 32 34 35 37–40 46 48 50 and tracheostomy rates (RR 0.25, to ongoing weaning on mechanical ventilation. Compared with
95% CI 0.10 to 0.61; 10 trials, 1130 patients)26 29 35 37–40 46 50 51 invasive weaning, non-­invasive weaning reduced mortality (high
with variable heterogeneity (table 2). quality), VAP (high quality), weaning failures (high quality),
length of stay in the ICU and hospital and tracheostomy. More-
Sensitivity and subgroup analyses over, non-­invasive (vs invasive weaning) significantly reduced
Exclusion of two potentially quasi-­randomised trials28 42 main- the duration of invasive ventilation, total duration of ventilation
tained significant reductions in mortality (RR 0.62, 95% CI 0.47 and the duration of ventilation related to weaning. The effects
to 0.80) and VAP (RR 0.30, 95% CI 0.22 to 0.43) both favouring on mortality and VAP remained significant after exclusion of
non-­invasive weaning. quasi-­randomised trials. Subgroup analysis suggested that the
We noted a significant difference in RR between subgroups benefits of non-­invasive weaning were higher in COPD versus
(p=0.0003) evaluating non-­ invasive weaning on mortality in mixed patient populations, with significant between-­ group
COPD (RR 0.36, 95% CI 0.25 to 0.51; 14 trials, 922 patients; differences in mortality, ICU length of stay and reintubation
p<0.00001, I2=0%) versus mixed population (RR 0.81, 95% CI favouring patients with COPD.
0.62 to 1.05; 13 trials, 1120 patients; p=0.11, I2=7%). A Compared with our previous review, this update includes
subgroup analysis that compared trials enrolling at least 50% (RR data from 12 additional trials and 1072 additional patients.17
0.44, 95% CI 0.31 to 0.63; 18 trials, 1200 patients) versus less Patients with COPD accounted for half the trials and 44.6%
than 50% patients with COPD (RR 0.80, 95% CI 0.61 to 1.06; of the patients overall. Overall trials included in the current
nine trials, 842 patients) also showed a larger and statistically review were of moderate-­to-­good quality. Lack of blinding and
significant (p=0.009) mortality reduction in subgroup analysis the inconsistent use of standardised weaning protocols in both
favouring COPD trials. The effect of non-­invasive weaning on arms and absence of a sedation protocol in the invasive weaning
weaning failures did not differ significantly between COPD and arm raise the possibility that control patients in some trials
mixed populations and trials enrolling at least 50% versus less may not have received optimal care, biasing in favour of non-­
invasive weaning. The largest trial included patients with vari-
able reasons for ARF and carefully protocolised weaning in both
groups.50 Similar to our systematic review, Perkins et al found
significant differences in the duration of invasive ventilation and
total duration of ventilation. Unlike our review, this study did
not find differences in rates of mortality or tracheostomy.50 The
low proportion of patients with COPD included in their trial
(4% COPD) versus the larger number of patients with COPD
included in our review (44.6%) may, at least in part, explain the
discordant findings. In general, results of large trials have been
Figure 2  Risk of bias of the included trials. Green, yellow and red found to agree with meta-­analyses to which they contribute,80
circles represent low, unclear and high risk of bias, respectively. but differences between large trials and meta-­analyses of smaller
756 Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993
 Critical care

Figure 3  Effect of non-­invasive weaning on mortality. COPD, chronic obstructive pulmonary disease; M-­H, Mantel-­Haenszel.

Figure 4  Effect of non-­invasive weaning on ventilator-­associated pneumonia. COPD, chronic obstructive pulmonary disease; M-­H, Mantel-­Haenszel.
Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993 757
 Critical care

Figure 5  Effect of non-­invasive weaning on duration of ventilation related to weaning. COPD, chronic obstructive pulmonary disease; IV,
intravenous.

trials addressing the same question have been demonstrated in large-­scale observational studies of weaning practices83 84 should
some fields81 and may depend on the selection of included trials, provide more data on the real-­world use of non-­invasive weaning.
methods used to summarise data and outcomes reported.82 Clinicians weighing the trade-­off between the risks associated with
Non-­invasive weaning may be particularly suitable for patients failed extubation versus prolonged invasive ventilation can be reas-
with COPD, whose failure to wean is characterised by respiratory sured that non-­invasive weaning is superior to invasive weaning, for
muscle weakness and gas trapping leading to intrinsic positive end-­ patients with COPD. However, enthusiasm should be tempered by
expiratory pressure, both of which are assisted by non-­ invasive considerations of experience required among physicians, nurses and
ventilation. In contrast, patients with non-­hypercapneic respiratory respiratory therapists to safely implement non-­invasive weaning,
failure may fail a SBT for other reasons (eg, excess sputum) or may reintubation if required, and the need for standalone non-­invasive
have other reasons why extubation should be deferred (eg, low ventilators or ventilators capable of both forms of ventilation. This
level of consciousness). A survey of self-­reported practices found is particularly important in non-­COPD patients where benefits are
that intensivists commonly reported using non-­invasive weaning in less clear based on our subgroup analyses.
patients with COPD (>50% of respondents in most regions) but
were less likely to do so (<30% in most regions) for other indica-
tions such cardiogenic pulmonary oedema or postoperatively.83 Two Comparison with other studies
A clinical practice guideline that cited our previous review85
gave a conditional recommendation in favour of non-­invasive
Table 2  Summary estimates of effect of non-­invasive ventilation in weaning for patients with hypercapneic respiratory failure but
critically ill adults made no recommendation for patients with hypoxemic respira-
No. of studies Summary estimate tory failure. Compared with the 2018 review by Yeung et al,86
Outcome (no. of patients) (95% CI) I2 (%) we included five additional trials.27 30 38 51 52 Similar to their
Mortality 26 (2042) 0.57 (0.44 to 0.74) 26 study, we documented beneficial effects of NIV on mortality,
VAP, duration of invasive ventilation and ICU stay.86 In a sensi-
VAP 23 (1581) 0.30 (0.22 to 0.41) 24
tivity analysis of nine trials (n=788 patients who failed an initial
Weaning failures 11 (829) 0.59 (0.43 to 0.81) 22 SBT), they reported beneficial effects of non-­invasive weaning
Length of stay on hospital mortality but wide ‘highest posterior density inter-
 Intensive care  22 (1804) −4.6 (−5.9 to −3.3) 80 vals’, from Bayesian estimates, precluded a definitive statement
 Hospital  13 (1061) −6.3 (−8.9 to −3.7) 75
of the effect of NIV on this outcome. Unlike their study, we also
identified beneficial effects of non-­invasive (vs invasive) weaning
Duration of mechanical ventilation
in reducing the proportion of weaning failures and tracheos-
 Total  12 (687) −5.3 (−7.9 to −2.7) 83 tomies, as well as, hospital length of stay and the duration of
 Related to weaning  13 (1163) −0.6 (−1.1 to −0.1) 88 ventilation related to weaning with considerable heterogeneity.
 Invasive ventilation  19 (1214) −7.8 (−9.9 to −5.6) 94 Moreover, we found significant effects of non-­invasive weaning,
compared with invasive weaning on mortality, VAP, ICU stay and
Adverse events
reintubation in COPD versus mixed populations. Similar to our
 Reintubation  14 (1336) 0.69 (0.47 to 1.01) 59 previous meta-­analysis,17 we found that non-­invasive weaning
 Tracheostomy  10 (1130) 0.25 (0.10 to 0.61) 56 significantly reduced mortality, weaning failure, VAP, ICU and
 Arrhythmia  4 (565) 0.70 (0.41 to 1.20) 0 hospital lengths of stay, total duration of ventilation and reintu-
*Risk ratio. bation compared with invasive weaning. In this updated review,
†Mean difference. we also found that non-­invasive (vs invasive) weaning signifi-
‡Invasive ventilation. cantly reduced the duration of ventilation related to weaning but
VAP, ventilator-­associated pneumonia. did not significantly reduce the rate of reintubation. Although
758 Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993
 Critical care

Table 3  Secondary analysis of summary estimates of effect of non-­invasive ventilation by subgroup

Outcome Subgroups No. of trials Effect estimate Test for subgroup differences
Mortality COPD 14 RR 0.36 (0.25, 0.51) P=0.0003
Mixed 12 RR 0.81 (0.62, 1.05)
Weaning failure COPD 4 RR 0.52 (0.37, 0.73) P=0.56
Mixed 7 RR 0.62 (0.37, 1.06)
Ventilator-a­ ssociated pneumonia COPD 14 RR 0.22 (0.15, 0.33) P=0.03
Mixed 9 RR 0.42 (0.28, 0.64)
ICU length of stay COPD 12 MD −6.1 (−8.1, –4.0) P=0.03
Mixed 10 MD −3.1 (−5.0, –1.1)
Hospital length of stay COPD 7 MD −7.4 (−10.9, –4.0) P=0.16
Mixed 6 MD −4.0 (−7.4, –0.6)
Total duration of mechanical ventilation COPD 7 MD −6.4 (−10.4, –2.4) P=0.15
Mixed 5 MD −3.2 (−5.0, –1.4)
Duration of ventilation related to weaning COPD 6 MD −1.3 (−2.2, –0.4) P=0.17
Mixed 7 MD 0.3 (−1.8, 2.3)
Duration of invasive ventilation COPD 11 MD −8.0 (−11.0, –5.1) P=0.81
Mixed 8 MD −7.4 (−11.3, –3.6)
Reintubation COPD 4 RR 0.48 (0.34, 0.67) P=0.02
Mixed 10 RR 0.89 (0.59, 1.35)
Tracheostomy COPD 1 RR 0.04 (0.00, 0.60) P=0.15
Mixed 9 RR 0.31 (0.13, 0.71)
Arrhythmia COPD 1 RR 2.00 (0.20, 19.78) P=0.36
Mixed 3 RR 0.66 (0.38, 1.15)
COPD, chronic obstructive pulmonary disease; ICU, intensive care unit; MD, mean difference; RR, risk ratio.

our prior review noted a significant difference in mortality trials, primarily driven by potential publication bias, specifically
between COPD versus mixed populations overall, it did not find the absence of small negative trials and imprecision. Overall, the
significant differences comparing trials in which at least 50% of quality of the evidence was graded as high for key outcomes
enrolled patients had COPD with trials in which less than 50% in our review including mortality, weaning failure and VAP.
of patients had COPD. By contrast, in our updated review, we Notwithstanding, some outcomes, in particular VAP, were
found significant differences between COPD versus mixed popu- subject to ascertainment bias, since microbiological confirmation
lations in four outcomes (mortality, VAP, ICU stay and reintuba- via sputum culture is easier in intubated patients. Highly vari-
tion) and in mortality in a subgroup analysis that compared trials able control group event rates for mortality and VAP may reflect
enrolling at least 50% versus less than 50% patients with COPD. heterogeneity in the patient populations included and selection
These findings underscore the greater net clinical benefits asso- criteria used between studies or different usual care practices and
ciated with the non-­invasive approach to weaning for patients contribute to indirectness of evidence.
with COPD. The findings of our updated review extend the
findings of our previous review by identifying beneficial effects
CONCLUSION
of non-­invasive (vs invasive) weaning in reducing the duration
Pooled data support the net clinical benefits associated with
of ventilation related to weaning, hospital length of stay and
the non-­ invasive approach to weaning on a wide range of
the proportions of weaning failures and tracheostomies. More-
clinical outcomes. In subgroup analysis, we found significant
over, they highlight the robustness and stability of the evidence
benefit of non-­invasive weaning in trials enrolling COPD versus
favouring non-­invasive weaning for patients with COPD as new
mixed populations on mortality and other important outcomes
trials were added to the prior pool of trials. At present, however,
including pneumonia, reintubation and ICU length of stay.
our analyses do not support the use of non-­invasive weaning in
mixed populations. Aligned with our findings, a recent system- Author affiliations
atic review and individual patient meta-­analysis in six trials high- 1
Interdepartmental Division of Critical Care Medicine, Temerty Faculty of Medicine,
lighted the potential beneficial effect of non-­invasive ventilation University of Toronto, Toronto, Ontario, Canada
2
after early extubation in reducing total days spent on invasive Departments of Critical Care and Medicine, Unity Health Toronto – St. Michael’s
Hospital, Toronto, Ontario, Canada
mechanical ventilation, though this was not associated with a 3
The Temerty Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
significant reduction in ICU mortality.87 4
Department of Clinical Epidemiology and Biostatistics, McMaster University,
Hamilton, Ontario, Canada
5
Strengths and limitations Keenan Research Centre for Biomedical Science and the Li Ka Shing Knowledge
Institute, St. Michael’s Hospital, Toronto, Ontario, Canada
This review was strengthened by a comprehensive trial search 6
The School of Medicine, Royal College of Surgeons, Dublin, Ireland
and standard systematic review methods to reduce risk of bias. 7
Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, Toronto,
Limitations of this work include this risk of bias in primary Ontario, Canada

Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993 759

 Critical care
8
Institute of Health Policy, Management, and Evaluation, University of Toronto, 14 Mehta S, Hill NS. Noninvasive ventilation. Am J Respir Crit Care Med
Toronto, Ontario, Canada 2001;163:540–77.
9
The Department of Critical Care Medicine, The Second Affiliated Hospital of 15 Rathgeber J, Schorn B, Falk V, et al. The influence of controlled mandatory ventilation
Zhengzhou University, Zhengzhou, China (CMV), intermittent mandatory ventilation (IMV) and biphasic intermittent positive
10
Department of Anesthesia and Physiology, University of Toronto, Toronto, Ontario, airway pressure (BIPAP) on duration of intubation and consumption of analgesics and
Canada sedatives. A prospective analysis in 596 patients following adult cardiac surgery. Eur J
Anaesthesiol 1997;14:576–82.
Correction notice  This article has been corrected since it was published Online 16 Criner GJ, Tzouanakis A, Kreimer DT. Overview of improving tolerance of long-­term
First. Some affiliations were listed incorrectly and a minor error was noted in Table 3. mechanical ventilation. Crit Care Clin 1994;10:845–66.
17 Burns KEA, Meade MO, Premji A, et al. Noninvasive ventilation as a weaning strategy
Twitter Karen E A Burns @karenburnsk for mechanical ventilation in adults with respiratory failure: a Cochrane systematic
Collaborators  Not applicable. review. CMAJ 2014;186:E112–22.
18 The Cochrane Collaboration. Review Manager (RevMan) Version 5.4] [Computer
Contributors  KEAB conducted the literature searches, resolved discrepancies program]. The Nordic Cochrane Centre, 2011.
between citation reviewers, selected studies meeting inclusion criteria, assessed 19 Cochran WG. The combination of estimates from different experiments. Biometrics
study quality, conducted risk of bias assessments, double checked data entry, 1954;10:101–29.
prepared initial and subsequent drafts of the manuscript and integrated comments 20 Berlin JA, Laird NM, Sacks HS, et al. A comparison of statistical methods for
into revised versions of the manuscript. JS and ML screened abstracts, selected combining event rates from clinical trials. Stat Med 1989;8:141–51.
studies meeting inclusion criteria, abstracted data, assessed study quality, conducted 21 Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-­analysis. Stat Med
risk of bias assessments and integrated comments into revised versions of the 2002;21:1539–58.
manuscript. YL, CL, ZL and LC assisted with screening articles for inclusion. YL, CL, 22 Higgins JPT, Green S, eds. Cochrane handbook of systematic reviews of interventions
XH, ZL, WW, LC and HZ translated articles, abstracted data, assessed study quality version 5.1.0. [updated March 2011] edition. The Cochrane Collaboration, 2011.
and conducted risk of bias assessments. NKJA provided methodologic guidance, 23 Borenstein N, Hedge LV, Higgins JPT, et al. Introduction to meta-­analysis. Chichester,
aided with drafting the manuscript and integrated comments into revised versions UK: John Wiley & Sons, 2008.
of the manuscript. JOF provided methodologic guidance, assessed study quality and 24 Guyatt GH, Oxman AD, Vist GE, et al. Grade: an emerging consensus on rating quality
conducted risk of bias assessments, hand searched conference proceedings and of evidence and strength of recommendations. BMJ 2008;336:924–6.
integrated comments into revised versions of the manuscript. All authors revised and 25 Nava S, Ambrosino N, Clini E, et al. Noninvasive mechanical ventilation in the weaning
approved the final version of the manuscript. of patients with respiratory failure due to chronic obstructive pulmonary disease. A
Funding  KEAB holds a career award from the Physician Services Incorporated randomized, controlled trial. Ann Intern Med 1998;128:721–8.
Foundation. 26 Girault C, Daudenthun I, Chevron V, et al. Noninvasive ventilation as a
systematic extubation and weaning technique in acute-­on-­chronic respiratory
Competing interests  None declared. failure: a prospective, randomized controlled study. Am J Respir Crit Care Med
Patient consent for publication  Not applicable. 1999;160:86–92.
27 Hill NS, Lin D, Levy M, et al. Noninvasive positive pressure ventilation (NPPV) to
Ethics approval  Not required. facilitate extubation after acute respiratory failure: a feasibility study. Am J Res Crit
Provenance and peer review  Not commissioned; externally peer reviewed. Care Med 2000;161:B18.
28 Chen J, Qiu D, Tao D. [Time for extubation and sequential noninvasive mechanical
Data availability statement  Data are available from KEAB. Data are available ventilation in COPD patients with exacerbated respiratory failure who received
from Dr. Burns (0000-­0002-9­ 967-­5424). invasive ventilation]. Zhonghua Jie He He Hu Xi Za Zhi 2001;24:99–100.
29 Ferrer M, Esquinas A, Arancibia F, et al. Noninvasive ventilation during persistent
ORCID iD
weaning failure. Am J Respir Crit Care Med 2003;168:70–6.
Karen E A Burns http://orcid.org/0000-0002-9967-5424
30 Rabie GM, Mohamed AZ, Mohamed RN. Noninvasive ventilation in the weaning of
patients with acute-­on-­chronic respiratory failure due to COPD. Chest 2004;126:755.
31 Wang X, Du X, Zhang W. Observation of the results and discussion on the timing
REFERENCES of transition form invasive mechanical ventilation to noninvasive ventilation
1 Melsen WG, Rovers MM, Groenwold RHH, et al. Attributable mortality of ventilator-­ in COPD patients with concomitant acute respiratory failure. Shandong Med
associated pneumonia: a meta-­analysis of individual patient data from randomised 2004;44:4–6.
prevention studies. Lancet Infect Dis 2013;13:665–71. 32 Wang C, Zhang Q, Cao Z. Collaborating Research Group for noninvasive mechanical
2 Farhan H, Moreno-­Duarte I, Latronico N, et al. Acquired muscle weakness in the ventilation of the Chinese respiratory Society. Pulmonary infection control window
surgical intensive care unit: nosology, epidemiology, diagnosis, and prevention. in the treatment of severe respiratory failure of chronic obstructive pulmonary
Anesthesiology 2016;124:207–34. diseases: a prospective, randomized controlled, multi-­centre study. Chinese Med J
3 Herridge MS, Tansey CM, Matté A, et al. Functional disability 5 years after acute 2005;118:1589–94.
respiratory distress syndrome. N Engl J Med 2011;364:1293–304. 33 Zheng R, Liu L, Yang Y. Prospective randomized controlled clinical study of sequential
4 Dowdy DW, Eid MP, Sedrakyan A, et al. Quality of life in adult survivors of critical non-­invasive following invasive mechanical ventilation in patients with acute
illness: a systematic review of the literature. Intensive Care Med 2005;31:611–20. respiratory failure induced COPD. Chinese J Emerg Med 2005;14:21–5.
5 Combes A, Costa M-­A, Trouillet J-­L, et al. Morbidity, mortality, and quality-­of-­life 34 Zou S-­hai, Zhou R, Chen P, et al. [Application of sequential noninvasive following
outcomes of patients requiring >or=14 days of mechanical ventilation. Crit Care Med invasive mechanical ventilation in COPD patients with severe respiratory failure by
2003;31:1373–81. investigating the appearance of pulmonary-­infection-­control-­window]. Zhong Nan Da
6 Cuthbertson BH, Hull A, Strachan M, et al. Post-­traumatic stress disorder after critical Xue Xue Bao Yi Xue Ban 2006;31:120–5.
illness requiring general intensive care. Intensive Care Med 2004;30:450–5. 35 Trevisan CE, Vieira SR, the Research Group in Mechanical Ventilation Weaning.
7 Jubran A, Lawm G, Duffner LA, et al. Post-­traumatic stress disorder after weaning Noninvasive mechanical ventilation may be useful in treating patients who fail
from prolonged mechanical ventilation. Intensive Care Med 2010;36:2030–7. weaning from invasive mechanical ventilation: a randomized clinical trial. Crit Care
8 Jubran A, Lawm G, Kelly J, et al. Depressive disorders during weaning from prolonged 2008;12:136
mechanical ventilation. Intensive Care Med 2010;36:828–35. 36 Prasad SBN, Chaudhry D, Khanna R. Role of noninvasive ventilation in weaning from
9 Ely EW, Gautam S, Margolin R, et al. The impact of delirium in the intensive care unit mechanical ventilation in patients of chronic obstructive pulmonary disease: an Indian
on hospital length of stay. Intensive Care Med 2001;27:1892–900. experience. Indian J Crit Care Med 2009;13:207–12.
10 Pandharipande PP, Girard TD, Jackson JC. BRAIN-­ICU study Investigators long-­term 37 Girault C, Bubenheim M, Abroug F, et al. Noninvasive ventilation and weaning in
cognitive impairment after critical illness. N Engl J Med 2013;369:1306–16. patients with chronic hypercapnic respiratory failure: a randomized multicenter trial.
11 MacIntyre NR, Cook DJ, Ely EW, et al. Evidence-­Based guidelines for weaning and Am J Respir Crit Care Med 2011;184:672–9.
discontinuing ventilatory support: a collective Task force facilitated by the American 38 Rabie Agmy GM, Metwally MM. Noninvasive ventilation in the weaning of patients
College of chest physicians; the American association for respiratory care; and the with acute-­on-­chronic respiratory failure due to COPD. Egypt J Chest Dis Tubercul
American College of critical care medicine. Chest 2001;120:375–95. 2012;61:84–91.
12 Nava S, Ambrosino N, Rubini F, et al. Effect of nasal pressure support ventilation and 39 Tawfeek MM, Ali-­Elnabtity AM. Noninvasive proportional assist ventilation may
external PEEP on diaphragmatic activity in patients with severe stable COPD. Chest be useful in weaning patients who failed a spontaneous breathing trial. Egypt J
1993;103:143–50. Anaesthesia 2012;28:89–94.
13 Appendini L, Patessio A, Zanaboni S, et al. Physiologic effects of positive end-­ 40 Vaschetto R, Turucz E, Dellapiazza F, et al. Noninvasive ventilation after early
expiratory pressure and mask pressure support during exacerbations of chronic extubation in patients recovering from hypoxemic acute respiratory failure: a single-­
obstructive pulmonary disease. Am J Respir Crit Care Med 1994;149:1069–76. centre feasibility study. Intensive Care Med 2012;38:1599–606.

760 Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993

 Critical care
41 Charra B, Hachimi A, Benslama A, et al. Contribution of noninvasive ventilation in the 65 Du L-­ling, Han H, Zhang X-­jun, et al. [Randomized control study of sequential
precocious extubation in the medical ICU. Crit Care 2009;13:P12. non-­invasive following short-­term invasive mechanical ventilation in the treatment
42 Rong F. Application of treating chronic obstructive pulmonary disease patients with of acute respiratory distress syndrome as a result of existing pulmonary diseases in
respiratory failure with the sequential noninvasive and invasive ventilation. J Bengbu elderly patients]. Zhongguo Wei Zhong Bing Ji Jiu Yi Xue 2009;21:394–6.
Med Coll 2012;37:442–4. 66 Venkatram S, Rachmale S, Kanna B. Non-­Invasive positive pressure ventilation
43 Mohamed AS, Ibrahim I. Elective early noninvasive ventilation as a weaning method of compared to invasive mechanical ventilation among patients with COPD
COPD patients. Europ Resp Soc 2012;40:P2041. exacerbations in an inner City MICU- predictors of NPPV use. Int J Pulmonary Med
44 Laiq N, Kahn RA, Malik A. Noninvasive positive pressure ventilation facilitates 2010;12:1531–2984.
early extubation in post-­operative cardiac surgery patients. J Postgrad Med Inst 67 Nava S, Grassi M, Fanfulla F, et al. Non-­invasive ventilation in elderly patients with
2013;27:361–5. acute hypercapnic respiratory failure: a randomised controlled trial. Age Ageing
45 El-­Shimy WS, Barima MA, Abo El-­Magd GH, et al. Non invasive ventilation versus 2011;40:444–50.
synchronized intermittent mandatory ventilation with pressure support in weaning of 68 Zheng D-­W, Wang C-­Z, Liu R-­S, et al. [The application of improved glasgow coma
COPD patients: comparative study. Egypt J Chest Dis Tuberc 2013;62:159–66. scale score of 15 as switching point for invasive noninvasive mechanical ventilation
46 Carron M, Rossi S, Carollo C, et al. Comparison of invasive and noninvasive positive in treatment of severe respiratory failure in chronic obstructive pulmonary disease].
pressure ventilation delivered by means of a helmet for weaning of patients from Zhongguo Wei Zhong Bing Ji Jiu Yi Xue 2011;23:224–7.
mechanical ventilation. J Crit Care 2014;29:580–5. 69 Vargas F, Clavel M, Sanchez P. Sequential and early use of noninvasive ventilation
47 Mishra M, Chaudhri S, Tripathi V, et al. Weaning of mechanically ventilated chronic after extubation in patients with chronic respiratory disorders. Am J Respir Crit Care
obstructive pulmonary disease patients by using non-­invasive positive pressure Med 2012;185:A6487.
ventilation: a prospective study. Lung India 2014;31:127–33. 70 Duan J, Guo S, Han X, et al. Dual-­Mode weaning strategy for difficult-­weaning
48 Wang X, Xu S, Liu G, et al. [Study of timing of invasive and noninvasive sequential tracheotomy patients: a feasibility study. Anesth Analg 2012;115:597–604.
ventilation in patients with acute respiratory distress syndrome]. Zhonghua Wei Zhong 71 Gao F. Comparison of non-­invasive positive pressure ventilation for extubated patients
Bing Ji Jiu Yi Xue 2014;26:330–4. who fail a single spontaneous breathing trial vs conventional weaning ISRCTN
49 Guo F, Xu S, Liu G, et al. Innovative effect of noninvasive sequential mechanical 32810409.
ventilation on prognosis of elderly patients with severe community acquired 72 Wang H, He X, Wu Q, et al. Invasive-­noninvasive sequential mechanical ventilation
pneumonia. Chinese Crit Care Med 2015;27:595–600. for treatment of acute exacerbation of COPD. Chinese J Respir Crit Care Med2009;1
50 Perkins GD, Mistry D, Gates S, et al. Effect of Protocolized weaning with early 8:290–1.
extubation to noninvasive ventilation vs invasive weaning on time to liberation 73 Mabrouk AA, Mansour OF, El-­Aziz AAA, et al. Evaluation of some predictors
from mechanical ventilation among patients with respiratory failure: the breathe for successful weaning from mechanical ventilation. Egypt J Chest Dis Tuberc
randomized clinical trial. JAMA 2018;320:1881–8. 2015;64:703–7.
51 Vaschetto R, Longhini F, Persona P, et al. Early extubation followed by immediate 74 Liang G-­P, Zeng Y-­H, Chen B-­X, et al. Prophylactic noninvasive positive pressure
noninvasive ventilation vs. standard extubation in hypoxemic patients: a randomized ventilation in the weaning of difficult-­weaning tracheotomy patients. Ann Transl Med
clinical trial. Intensive Care Med 2019;45:62–71. 2020;8:300.
52 Chen J, Wang W, Qiu W. Application of acute physiology and chronic health evaluation 75 Pu X-­X, Wang J, Yan X-­B, et al. Sequential invasive-­noninvasive mechanical
Ⅱ score in the timing of noninvasive ventilation in patients with acute exacerbation of ventilation weaning strategy for patients after tracheostomy. World J Emerg Med
chronic obstructive pulmonary disease. China Crit Care Med 2020;32:581–4. 2015;6:196–200.
53 Ishikawa S, Ohtaki A, Takahashi T, et al. Noninvasive nasal mask BiPAP management 76 Tan D, Walline JH, Ling B, et al. High-­flow nasal cannula oxygen therapy versus
for prolonged respiratory failure following cardiovascular surgery. J Card Surg non-­invasive ventilation for chronic obstructive pulmonary disease patients after
1997;12:176–9. extubation: a multicenter, randomized controlled trial. Crit Care 2020;24:489.
54 Radojevic D, Vuk LJ, Radomir B, et al. PIPAP and PSV in the weaning of cardiosurgical 77 Ayazoğlu TA, Baysal A, Özensoy A, et al. The comparison of two different weaning trial
patients from mechanical ventilation. Br J Anaesthesia 1997;78:A161. methods for patients with chronic obstructive lung disease (COPD) in a cardiothoracic
55 Kruger M, Walther T, Falk V. Prospective randomization of pressure-­controlled (BiPAP) intensive care unit (ICU). Intensive Care Med Exp 2015;3:A167.
and volume-­controlled (SIMV) ventilation after cardiac surgery. Intensive Care Med 78 Perkins GD, Mistry D, Lall R, et al. Protocolised non-­invasive compared with invasive
1998;24:5. weaning from mechanical ventilation for adults in intensive care: the breathe RCT.
56 Jiang JS, Kao SJ, Wang SN. Effect of early application of biphasic positive airway Health Technol Assess 2019;23:1–114.
pressure on the outcome of extubation in ventilator weaning. Respirology 79 Chaudhri S, Mishra M, Verma AK. Utility of noninvasive positive pressure ventilation
1999;4:161–5. (NPPV) for weaning COPD patients from invasive mechanical ventilation. Eur Resp J
57 Wang C, Shang M, Huang K. [Sequential non-­invasive following short-­term invasive 2009;34:80.
mechanical ventilation in COPD induced hypercapnic respiratory failure]. Zhonghua Jie 80 Glasziou PP, Shepperd S, Brassey J. Can we rely on the best trial? A comparison of
He He Hu Xi Za Zhi 2000;23:212–6. individual trials and systematic reviews. BMC Med Res Methodol 2010;10:23.
58 Luo H, Cheng P, Zhou R. [Sequential BiPAP following invasive mechanical ventilation 81 Sivakumar H, Peyton PJ. Poor agreement in significant findings between meta-­
in COPD patients with hypercapnic respiratory failure]. Hunan Yi Ke Da Xue Xue Bao analyses and subsequent large randomized trials in perioperative medicine. Br J
2001;26:563–5. Anaesth 2016;117:431–41.
59 Rosinha S, Lobo SMA, Sanches HS. The use of noninvasive ventilation after weaning of 82 Ioannidis JP, Cappelleri JC, Lau J. Issues in comparisons between meta-­analyses and
acute respiratory failure prevents reintubation and decreases hospital mortality. Am J large trials. JAMA 1998;279:1089–93.
Respir Criti Care Med 2001. 83 Burns KEA, Raptis S, Nisenbaum R, et al. International practice variation in weaning
60 Wang C, Shang M, Huang K, et al. Sequential non-­invasive mechanical ventilation critically ill adults from invasive mechanical ventilation. Ann Am Thorac Soc
following short-­term invasive mechanical ventilation in COPD induced hypercapnic 2018;15:494–502.
respiratory failure. Chin Med J 2003;116:39–43. 84 Heunks L, Bellani G, Pham T, et al. The worldwide assessment of separation of patients
61 Matic I, Sakic-­Zdravcevic K, Jurjevic M. Comparison of invasive and noninvasive from ventilatory assistance (WEAN safe) ERS clinical research collaboration. Eur Respir
mechanical ventilation for patients with chronic obstructive pulmonary disease: J 2019;53:1802228.
randomized prospective study. Periodicum Biologorum 2007;109:137–45. 85 Ouellette DR, Patel S, Girard TD. Liberation from mechanical ventilation in critically
62 Celebi S, Köner O, Menda F, et al. Pulmonary effects of noninvasive ventilation ill adults: an official American College of chest Physicians/American thoracic Society
combined with the recruitment maneuver after cardiac surgery. Anesth Analg clinical practice guideline. CHEST 2017;151:166–80.
2008;107:614–9. 86 Yeung J, Couper K, Ryan EG, et al. Non-­Invasive ventilation as a strategy for weaning
63 Kilic A, Yapıci N, Bıcer Y, et al. Early extubation and weaning with bilevel positive from invasive mechanical ventilation: a systematic review and Bayesian meta-­analysis.
airway pressure ventilation after cardiac surgery (weaning with BiPAP ventilation after Intensive Care Med 2018;44:2192–204.
cardiac surgery). Southern African J Anaesthesia Analgesia 2008;14:25–31. 87 Vaschetto R, Pecere A, Perkins GD, et al. Effects of early extubation followed by
64 Yang S-­yue, Shen J-­li, Feng E-­zhi. [Study on the indication and time of the use of noninvasive ventilation versus standard extubation on the duration of invasive
non-­invasive positive pressure ventilation in the patients with acute exacerbation of mechanical ventilation in hypoxemic non-­hypercapnic patients: a systematic review
chronic cor pulmonale combined with type II respiratory failure at high altitude area]. and individual patient data meta-­analysis of randomized controlled trials. Crit Care
Zhongguo Wei Zhong Bing Ji Jiu Yi Xue 2009;21:440–1. 2021;25:189.

Burns KEA, et al. Thorax 2022;77:752–761. doi:10.1136/thoraxjnl-2021-216993 761