Professional Documents
Culture Documents
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Monro-Somerville et al
METHODS
The research question was formulated using the Population, Incidence, Comparator, Outcome model (Supplemental Table1,
Supplemental Content 1, http://links.lww.com/CCM/C176).
Search Strategy
The study protocol was published on the International Prospective Register of Systematic Reviews website (http://www.
crd.york.ac.uk). Ovid Medline, Embase, and Cochrane Database of Systematic Reviews were searched for suitable studies
using a structured search strategy: high-flow nasal oxygen.mp
or nasal high flow.mp or high-flow nasal oxygen therapy.mp
or high-flow nasal prongs.mp or nasal high-flow oxygen.mp
or high-flow nasal cannula.mp or humidified high-flow nasal
cannula.mp or heated and humidified high-flow nasal oxygen.
mp or Optiflow.mp and respiratory insufficiency/or respiratory failure.mp or oxygen inhalational therapy.mp or acute
respiratory distress syndrome.mp or respiratory distress syndrome, adult/or ventilation.mp or ventilation/or dyspnea.
Search results were limited to randomized controlled trials
(RCTs) in adults, reported in English language. Authors and
research teams from commercial suppliers were contacted to
ascertain if they were aware of other studies not captured by
the search strategy; conference proceedings and bibliographies
2
www.ccmjournal.org
were also reviewed. Four further studies met inclusion criteria for this systematic review or meta-analysis. The literature
search was conducted independently by two authors (T.M.S.,
M.S.). Disparities in the literature review were resolved by
consensus of all authors. Search strategy and analysis were carried out using the Preferred Reporting Items for Systematic
Review and Meta-Analysis statement 2009 (27).
Study Selection Criteria and Quality Assessment
Search results were reviewed and evaluated independently by
two authors (T.M.S. and M.A.G.). The following criteria were
used for inclusion in our meta-analysis: RCTs of adult patients
with respiratory failure of any cause, HFNC versus any other
oxygen delivery modality including COT or NIV (referred
to as usual care), and complete data available on any of the
outcomes. Studies were excluded if not published in English,
crossover studies, nonhuman experimental studies, or lack of
complete data in the outcomes of interest. The primary outcome for the meta-analysis was hospital mortality, and secondary outcome was tracheal intubation. The following subgroups
were analyzed: postoperative respiratory failure, postextubation
respiratory failure, ARF of medical origin, and intubation rates
when HFNC was compared with patients treated with COT only
or NIV only. In addition, any randomized trials reporting incidence of delirium and comfort scores were included in a qualitative assessment of patient comfort and tolerability. Studies
were screened for methodological quality using the Cochrane
Collaboration Risk of Bias tool, an established method of assessing methodological quality and internal validity of studies to be
included in meta-analysis (28). The overall risk of bias was considered low only if all components were rated as having a low
risk of bias. Disagreements on studies to be included in the final
analysis were resolved by consensus of the whole study group.
Grading of Quality of Evidence
We used the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) Working Group methodology to assess the overall quality of the evidence for the primary
and secondary outcomes in the following domains: inconsistency, indirectness, imprecision, and publication bias. This was
classified as very low, low, medium, and high (29).
Data Extraction
Data extraction was undertaken by two authors (M.A.G.,
T.M.S.) for each eligible study and included author, year of publication, patient group studied, number of subjects, modality of
comparator oxygen delivery, commercial support, hospital mortality, and incidence of reintubation. A qualitative assessment of
patient tolerability and comfort was made where reported.
Statistical Analysis
Statistical analysis was carried out using RevMan Review Manager, version 5.3, Nordic Cochrane Review Centre, Copenhagen,
Denmark) and Trial Sequential Analysis (TSA) Program version 0.9 (Copenhagen Trial Unit, Denmark; http://www.ctu.dk/
tsa). Evidence of statistical heterogeneity between studies was
XXX 2016 Volume XX Number XXX
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
assessed by chi-square and I2tests; 25%, 50%, and 75% indicated the presence of low, moderate, and high between-trial heterogeneity, respectively. A p value of 0.1 was used to denote the
statistical significance of heterogeneity. The funnel plot method
was used to assess for evidence of publication bias for any of
the outcomes, either primary or secondary. Dichotomous outcomes were expressed as odds ratio (OR) with 95% CIs. For all
analyses performed, if no significant heterogeneity was noted,
fixed effect model (FEM) analysis using the Mantel-Haenszel
method was used; otherwise, a random-effects model analysis
was undertaken using the DerSimonian-Laird method. To assess
for the possibility of random error due to paucity of available
data, we used TSA. In brief, TSA uses methodology developed
for repeated significance testing in large RCTs and estimates
that the meta-analysis is of adequate information size (IS),
that is, contains sufficient patients and events to draw reliable
conclusions. Traditional meta-analysis software (e.g., RevMan)
does not provide this information. If the cumulative z score for
included trials crosses the monitoring boundary, then no further
Table 1.
trials are needed. If this does not occur, the available evidence is
not sufficient to draw a firm conclusion. We included TSA for
the outcomes of mortality and intubation.
RESULTS
Study Selection
The process for literature search and study selection is presented
in Figure 1. One hundred forty-seven nonduplicate citations
were screened; of which, 32 were available in full text and underwent full screening. Fourteen trials were eligible for inclusion
(11, 12, 1417, 23, 3036), and nine trials were used in the metaanalysis, including 2,507 subjects (Fig.1) (11, 15, 16, 3035).
Thirteen trials were used in the qualitative assessment of patient
comfort and tolerability (11, 12, 1417, 23, 3034, 36).
Characteristics of Included Studies
The characteristics of the included studies are summarized in
Table 1. Of note, five trials were multicenter (11, 16, 30, 31, 35),
Study
Design
Setting
Population
Reports
Included
Reports Reports Comfort Commercial in MetaComparator Multicentre Mortality Intubation Score
Support
Analysis
Hernandez
etal (35)
RCT
General Postextubation
ICU
527
FM
Yes
Yes
Yes
No
Yes
Yes
Rittayami
etal(14)
RCT
ED
ARF
40
FM/nasal
cannula
No
No
Yes
Yes
Yes
Yes
ED
ARF
100
FM
Yes
No
Yes
Yes
Yes
Yes
Corley
et al (34)
RCT
Cardiac
ICU
Postextubation,
body mass
index > 30
155
FM
No
No
Yes
No
Yes
Yes
Lemaile
etal(32)
RCT
100
FM
No
Yes
Yes
Yes
Yes
Yes
Stephan
etal(16)
RCT
Cardiac
ICU
Postextubation
830
NIV
Yes
Yes
Yes
Yes
Yes
Yes
Medical
ICU
ARF
310
FM/NIV
Yes
Yes
Yes
Yes
Yes
Yes
Maggiore
et al (11)
General Postextubation
ICU
105
FM
Yes
Yes
Yes
Yes
Yes
Yes
RCT
ARF
14
FM/NIV
No
No
No
Yes
Yes
No
Rittayami
etal(33)
Crossover Medical
ICU
Postoperative
extubation
17
FM
No
No
Yes
Yes
Yes
Yes
Parke
et al (15)
RCT
Postoperative
extubation
340
FM
No
Yes
Yes
No
Yes
Yes
30
FM/NIV
No
No
No
Yes
Yes
No
Cuquemelle
et al (23)
Crossover Medical
ICU
ARF
30
FM
No
No
Yes
Yes
Yes
No
Tiruvoipati
et al (12)
Crossover Medical
ICU
Postextubation
50
FM
No
Yes
No
Yes
Yes
No
Cardiac
ICU
ARF=acute respiratory failure, ED=emergency department, FM=face mask, NIV, noninvasive ventilation, RCT=randomized controlled trial.
www.ccmjournal.org
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Monro-Somerville et al
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram detailing search
strategy.
seven European (11, 16, 23, 30, 32, 35, 36), four Australian
(12, 15, 31, 34), one American (17), and two conducted in
Thailand (14, 33). Five trials reported intubation as a primary
outcome (16, 3032, 35) and four as a secondary outcome
(11, 15, 33, 34). Patient comfort was recorded as a secondary
outcome in 11 studies (11, 12, 1416, 23, 3033, 36). Dyspnea
was reported as a primary outcome in three trials (14, 31, 33)
and as a secondary outcome in six studies (16, 17, 30, 32, 34, 36).
In three studies, subjects were postoperative cardiothoracic
patients (15, 16, 34), four trials concerned subjects extubated
after ventilation for acute illness (11, 12, 14, 35), and eight studies involved patients with new ARF (17, 23, 3033, 35, 36). All
were conducted within the ICU and were discontinued when
the patient was discharged from critical care. Funnel plots did
not demonstrate the evidence of publication bias for primary
or secondary outcomes (supplemental file, Supplemental Content 1, http://links.lww.com/CCM/C176). The GRADE quality
evidence was thought to be low, mainly as a result of lack of
4
www.ccmjournal.org
Primary Outcome
Mortality data were available in five studies included
in the meta-analysis, including a total of 2,112 subjects
(11, 15, 16, 30, 35). Of the
1,006 patients treated with
HFNC, 60 (5.9%) died, compared with 90 of the 1,106
(8.1%) treated with usual
care. No significant difference
was demonstrated between
groups (p=0.29; I2 25%; FEM:
OR, 0.83; 95% CI, 0.581.17)
(Fig. 2).
Secondary Outcome
Intubation data were available in nine studies included in the
meta-analysis, including a total of 2,507 subjects (11, 15, 16, 30
35). Of the 1,207 patients treated with HFNC, 119 (9.9%) were
intubated compared with 204 of the 1,300 (15.7%) receiving
conventional oxygen therapies or NIV. Overall, there was no significant difference in intubation rates between groups (n=2,507;
p=0.08; I2, 53%; FEM: OR, 0.63; 95% CI, 0.371.06) (Fig. 3).
Post Hoc Analyses
Subgroup analysis showed no difference in the intubation
rate when restricted to postoperative patients (n=1,325;
p=0.91; I2, 14%; FEM: OR, 1.02; 95% CI, 0.701.50) (15, 16,
34). Restriction to postextubation patients showed no difference in the intubation rate (n=1,957; p=0.16; I2, 70%; FEM:
OR, 0.52; 95% CI, 0.211.29) (11, 15, 16, 34, 35). There was
no evidence of a positive effect in the subgroup of patients
with medical respiratory failure. This was defined as new
XXX 2016 Volume XX Number XXX
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Table 2.
No. of
Studies
Study
Design
No. of Patients
Effect
High-Flow Standard
Risk of
Other
Nasal
Oxygen
Bias Inconsistency Indirectness Imprecision Considerations Cannulae Therapies
Relative
OR,
(95% CI)
Absolute
(95% CI) Quality
Mortality
5
Not serious
Seriousb
None
60/1,006 90/1,106
0.83
13 fewer
(6.0%)
(8.1%) (0.581.17) per 1,000 LOW
(from 13
more to
33 fewer)
Not serious
Not serious
None
119/1,207 204/1,300
0.64
50 fewer
(9.9%)
(15.7%) (0.31.07)
per 1,000 LOW
(from 9
more to
91 fewer)
Intubation
9
Randomized Seriousa
control
trials
Seriousc
OR=odds ratio.
a
Unblinded intervention.
b
Wide CI.
c 2
I , 51%.
Figure 2. Forrest plot comparing mortality rates in patients treated with high-flow nasal cannulae (HFNC) compared with usual care.
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Monro-Somerville et al
Figure 3. Forrest plot comparing intubation rates in patients treated with high-flow nasal cannulae (HFNC) compared with usual care. M-H = Mantel-Haenzel.
did not meet statistical significance (12, 14, 36). Five studies
showed significantly improved comfort scores compared with
FM (11, 23, 30, 31, 33) or NIV (30, 36); another showed a significant improvement in tolerability of HFNC compared with
conventional oxygen (12). One showed facemask to be superior to HFNC (15). No studies included in the final analysis
used delirium as an outcome.
TSA
TSA was undertaken for the outcomes of mortality and intubation based on a relative risk reduction of 0.26 and 0.37, respectively, a type 1 error of 0.05 and a type 2 error of 0.8. For each
outcome, neither required IS (mortality, n=6,899; intubation,
n=4,018) nor boundaries for benefit, harm, or futility were
reached (supplemental file, Supplemental Content 1, http://
links.lww.com/CCM/C176).
DISCUSSION
The principal finding of this systematic review and metaanalysis is that no significant difference in mortality or intubation rate was detected in adult patients with ARF treated with
HFNC, when compared with usual care defined as COT or
NIV. The required IS was not reached using TSA; hence, more
studies are needed to answer this question definitively. A qualitative analysis on dyspnea, tolerability, and comfort suggested
that HFNC is at least equivalent to usual care, with the majority of studies suggesting improved dyspnea and comfort scores.
In post hoc analyses, there was no difference in intubation rates
when restricted to the perioperative subgroup, post extubation,
or new respiratory failure subgroups, or when compared with
NIV. A significant difference in the intubation rate was demonstrated when HFNC was compared with conventional oxygen
delivery alone (p=0.0008), although as a post hoc analysis, this
finding must be interpreted with caution.
6
www.ccmjournal.org
To our knowledge, this is the first systematic review and metaanalysis to evaluate whether the use of HFNC in patients with
ARF has an effect on mortality or intubation rate. Despite the
lack of high-quality trial data, HFNC is considered by many to be
superior to conventional methods of oxygen delivery and enjoys
widespread use (50). Much of the current evidence for the use
of HFNC has been conducted in pediatric patients or after adult
cardiothoracic surgery, and these studies may lack external validity for patients with acute or postoperative respiratory failure in
the general population. Observational and RCT data support
the assumption that HFNC is at least equivalent to conventional
oxygen therapies or NIV and has shown improvement in physiologic outcomes, and there is no evidence of harm and improved
patient comfort (11, 30). The largest RCT in adults with ARF did
not demonstrate a reduction in the intubation rate when compared with NIV or COT in adult patients with ARF (30). The
study did show significant reduction in duration of mechanical
ventilation and 90-day mortality in the HFNC group as secondary outcomes. A post hoc analysis suggested reduced intubation
rate in the subgroup with severe hypoxemic respiratory failure
(Pao2/Fio2 ratio, < 200mm Hg) treated with HFNC. One study
included time to intubation as a secondary outcome and found
no difference in outcomes (35). Other studies have suggested that
time to intubation may be longer in patients treated with HFNC
compared with COT (15) but similar in patients treated with NIV
(16, 30). This could reflect the existing controversy regarding the
use of NIV in ARF; that prolonged therapy results in delayed
intubation, leading to increased mortality in some patients (51,
52). Similar results have been reported regarding the prolonged
use of HFNC (26). The Hernandez protocol limited HFNC to
24 hours post extubation, based on the suggestion that a delay in
intubation of more than 48 hours may be detrimental (26, 35).
Other commentators have raised the possibility of oxygen toxicity as another undesirable feature of therapy with HFNC (25).
XXX 2016 Volume XX Number XXX
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
CONCLUSIONS
A systematic review and meta-analysis investigating HFNC in
adult patients treated for ARF did not demonstrate a reduction
in mortality associated with its use. TSA suggests that required
IS was not reached, and thus, further trials are required to
answer this question definitively. A qualitative analysis suggests
that this therapy was well tolerated and may improve dyspnea
scores and patient comfort. Future trials should identify the
populations most likely to benefit from this treatment and
define safe limits of therapy.
REFERENCES
1. Sztrymf B, Messika J, Mayot T, et al: Impact of high-flow nasal cannula oxygen therapy on intensive care unit patients with acute respiratory failure: A prospective observational study. J Crit Care 2012;
324:324.e9e13
2. Rello J, Prez M, Roca O, et al; CRIPS Investigators: High-flow nasal
therapy in adults with severe acute respiratory infection: A cohort
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Monro-Somerville et al
24. Sztrymf B, Messika J, Bertrand F, et al: Beneficial effects of humidified
high flow nasal oxygen in critical care patients: A prospective pilot
study. Intensive Care Med 2011; 37:17801786
25 Helmerhorst HJ, Schultz MJ, van der Voort PH, et al: Bench-tobedside review: The effects of hyperoxia during critical illness. Critical
Care 2015; 19:284
26. Kang BJ, Koh Y, Lim CM, et al: Failure of high-flow nasal cannula
therapy may delay intubation and increase mortality. Intensive Care
Med 2015; 41:623632
27. Moher D, Liberati A, Tetzlaff J, et al; PRISMA Group: Preferred reporting items for systematic reviews and meta-analyses: The PRISMA
statement. Ann Intern Med 2009; 151:264269, W64
28. Higgins JP, Altman DG, Gtzsche PC, et al; Cochrane Bias Methods
Group; Cochrane Statistical Methods Group: The Cochrane
Collaborations tool for assessing risk of bias in randomised trials.
BMJ 2011; 343:d5928
29. Guyatt GH, Oxman AD, Kunz R, et al: Rating quality of evidence and
strength of recommendations: Going from evidence to recommendations. BMJ 2008; 336:10491051
30. Frat JP, Thille AW, Mercat A, et al; FLORALI Study Group; REVA
Network: High-flow oxygen through nasal cannula in acute hypoxemic
respiratory failure. N Engl J Med 2015; 372:21852196
31. Bell N, Hutchinson CL, Green TC, et al: Randomised control trial of
humidified high flow nasal cannulae versus standard oxygen in the
emergency department. Emerg Med Australas 2015; 27:537541
32. Lemiale V, Mokart D, Mayaux J, et al: The effects of a 2-h trial of highflow oxygen by nasal cannula versus Venturi mask in immunocompromised patients with hypoxemic acute respiratory failure: A multicenter
randomized trial. Critical Care 2015; 19:18
33. Rittayamai N, Tscheikuna J, Praphruetkit N, et al: Use of high-flow
nasal cannula for acute dyspnea and hypoxemia in the emergency
department. Respir Care 2015; 60:13771382
34. Corley A, Bull T, Spooner AJ, et al: Direct extubation onto high-flow
nasal cannulae post-cardiac surgery versus standard treatment in
patients with a BMI 30: A randomised controlled trial. Intensive
Care Med 2015; 41:887894
35. Hernndez G, Vaquero C, Gonzlez P, et al: Effect of postextubation high-flow nasal cannula vs conventional oxygen therapy on reintubation in low-risk patients: A randomized clinical trial. JAMA 2016;
315:13541361
36. Schwabbauer N, Berg B, Blumenstock G, et al: Nasal high-flow
oxygen therapy in patients with hypoxic respiratory failure: Effect on
functional and subjective respiratory parameters compared to conventional oxygen therapy and non-invasive ventilation (NIV). BMC
Anesthesiol 2014; 14:66
37. Ebert TJ, Novalija J, Uhrich TD, et al: The effectiveness of oxygen delivery and reliability of carbon dioxide waveforms: A crossover comparison of 4 nasal cannulae. Anesth Analg 2015; 120:342348
www.ccmjournal.org
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.