Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Corley, 2017

Study characteristics and results are extracted from the SR (unless stated otherwise)

SR and meta-analysis of RCTs (parallel or cross-over design, however for this purpose only parallel designs are included)

Literature search up to Dec 2016

A: Corley, 2014

B: Cuquemelle 2012

C: Frat 2015

D: Lemiale 2015

E: Maggiore 2014

F: Parke 2011

G: Parke 2013a

H: Stephan 2015

Cross-over studies excluded: Chanques

2013; Rittayamai 2014; Schwabbauer 2014

Study design:

All RCTs parallel

Setting and Country:

Setting: all ICU

A: Australia

B: France

C: France and Belgium

D: France

E: Italy

F: New Zealand

G: New Zealand

H: France

Source of funding:

Nine studies declared funding or provision of equipment from

manufacturers of the HFNC system (Fisher &Paykel Healthcare)

(Chanques 2013; Corley 2014; Cuquemelle 2012; Frat 2015;

Lemiale 2015; Maggiore 2014; Parke 2011; Parke 2013a;

Schwabbauer 2014). Six of these reported that the manufacturer

had not been involved in study design, management, or data analyses,

and we considered these studies to have low risk of bias (Corley

2014; Cuquemelle 2012; Frat 2015; Lemiale 2015; Parke 2013a;

Schwabbauer 2014); two of these six did not report involvement

of the manufacturer, and we were unclear whether this funding

represented risk of bias (Chanques 2013; Maggiore 2014); one study reported that the manufacturer had been involved in the study design and had paid for statistical analysis; we judged this

study to have high risk of bias (Parke 2011).

Inclusion criteria SR: all randomized, parallel, and quasi-randomized studies

(including cross-over studies) that compared HFNC versus other forms of non-invasive respiratory support for selected outcome

measures.

Exclusion criteria SR: retrospective studies and prospective cohort or observational

studies

11 studies included, of which 8 used (4 cross-over studies were excluded) in this analysis and 5 with relevant outcomes

Important patient characteristics at baseline:

N, mean age

A: 155 patients, 63 (± 11.4) years (I);

65 (± 11.1) years (C)

B: 30 patients, 39-77 years

C: 313 patient, 61 (± 16) years (I); 59 (± 17) years (C1); 61 (± 17) years (C2)

D. 102 patients, Age median (25th to 75th percentile): 59.3 (43 to 70) years (I); Age median (25th to 75th percentile): 64.5 (53.25 to 72) years

E. 105 patients, 65 (± 18) years (I); 64 (± 17) years (C)

F. 56 participants, mean (range): 64 (39 to 83) years (I); 64 (26 to 85) years (C)

G. 340 participants, Age median (range) 65 (19 to 88) years (I) and 66 (21 - 87) years (C)

H: 830 participants; 63.8 (95% CI 62.5 to 65.2) years (I) and 63.9 (95% CI 62.6 to 65.2) years (C)

Groups were mostly comparable at baseline. However, for study B participants in the intervention group were older and had higher rates of infectious pneumonia. Reasons for exclusion potentially related to treatment

Describe intervention: high flow oxygen

A: HFNC (optiflow) n=81

B: HFNC (optiflow)

C: HFNC (optiflow) n=106

D: HFNC

E: HFNC n=53

F: HFNC (optiflow) n=29

G: HFNC (optiflow) n=169

H: HFNC (optiflow) n=414

Describe control: low flow oxygen

A: Standard oxygen therapy (face mask); n = 74

B: Standard oxygen therapy

Use of a flow meter from wall oxygen without humidification. Randomized to receive therapy during first 24 hours, then crossed-over to alternative therapy for 4 hours to reduce drop-outs

C-1: Standard oxygen (nonrebreather face mask) n=94

C-2: Non-invasive ventilation (face mask connected to an ICU ventilator with pressure support applied in

NIV mode;) n=110

D: Venturi mask

E: Venturi mask n=52

F: Standard oxygen therapy (HFFM standard face mask) n=27

G: Simple face mask (n=171)

H: BiPAP (n=416)

End-point of follow-up:

NA

For how many participants were no complete outcome data available?

(intervention/control)

A: no losses

B: Of 37 participants randomized, 7 were excluded from analysis, as they were unable to complete the study (5 owing to deterioration

and 2 because of rapid improvement in respiratory status)

C: 3 losses (2 in standard oxygen group and 1

in NIV group) due to withdrawal of consent.

D: Loss of two participants after randomization

due to withdrawal of consent.

E: no losses

F: Of 60 enrolled in the study, 4 participants (1 from the HFNC group, and 3 from the

HFFM group) were excluded: 2 refused consent for all data collection, and 2 failed

the screening. Five of 27 participants in the high-flow facemask group were switched to nasal high flow - no mention of how these

data were treated.

G: Attrition fully reported. Small number of

losses

H: no losses

Outcome measure-1

Defined as Failure of treatment as indicated by the

need for non- invasive

positive-pressure ventilation or invasive ventilation

à we decided not to include this outcome measure, is different from intubation rate

Outcome measure-2

Defined as In-hospital mortality up to 90 days

Effect measure: RR (95% CI):

C-1: 0.59 ( 0.25, 1.40)

C-2: 0.46 ( 0.20, 1.05)

E: 1.18 ( 0.38, 3.62)

G: 1.01 ( 0.06, 16.05)

Pooled effect (random effects model): RR 0.63 (95% CI 0.38 to 1.06),p=0.08 favoring HFNC

Heterogeneity (I²):0%

H. RR: 1.22 (95% CI 0.72 – 2.09) à not included in Cohrane in meta-analyse (uncomparable control group), in our meta-anlaysis we included this in the NIV group.

Outcome measure-3

Defined as length of ICU stay

Effect measure: (95% CI):

C-1: 1.60 ( -2.54, 5.74 )

E: 1.30 ( -2.29, 4.89 )

A: 0.0 ( -0.39, 0.39 )

G: 0.19 ( -0.02, 0.40 )

Pooled effect (random effects model): 0.15 ( 95%CI -0.03 to 0.34 ) favoring low flow oxygen

Heterogeneity (I²):0%

H. LOS ICU:

(I) Median (IQR) 6 (4 to 10)

(C) Median (IQR) 6 (4 to 10)

P=0.77

H. LOS hospital:

(I) Median (IQR) 13 (9 to 22)

(C)Median (IQR) 14 (9 to 20)

P=0.59

Studies B, Cuquemelle 2012, D Lamiale 2014, and F Parke 2011 did not report on any of the outcomes of interest.

Brief description of author’s conclusion:

We were unable to demonstrate whether HFNC was a more effective or safe oxygen delivery device compared with other oxygenation

devices in adult ICU patients. Meta-analysis could be performed for few studies for each outcome. In addition, we identified some risks of bias among included studies, differences in patient groups, and high levels of statistical heterogeneity for some outcomes, leading to uncertainty regarding the results of our analysis. Consequently, evidence is insufficient to show whether HFNC provides safe and efficacious respiratory support for adult ICU patients.

Level of evidence according to authors: GRADE (per comparison and outcome measure) including reasons for down/upgrading

2. GRADE: Low (Evidence downgraded by two levels, as some studies had high or unclear risk of bias in some domains and some studies

showed indirectness including participants requiring respiratory support for dif ferent purposes)

3. GRADE: Low (Evidence downgraded by two levels, as some studies had high or unclear risk of bias in some domains and some studies

showed indirectness including participants requiring respiratory support for dif ferent purposes)

Zhao, 2017

Study characteristics and results are extracted from the SR (unless stated otherwise)

SR and meta-analysis of RCTs

Literature search up to 20 October 2016

A: Bell, 2015

B: Corley, 2015

C: Frat, 2015

D: Hernandez, 2016 (COT)

E: Jones, 2015

F: Lemiale, 2015

G: Maggiore, 2014

H: Parke 2013

I: Parke, 2011

J: Stephan 2015

K: Hernandez, 2016 (NIV)

Study design:

RCTs

Setting and Country:

A: ED, Australia

B: ICU, Australia

C: ICU, France

D: ICU, Spain

E: ED, New Zealand

F: ICU, France

G: ICU, Italy

H: CVICU, New Zealand

I: CVICU, New Zealand

J: CTVS ICU, France

K: ICU, Spain

Source of funding:

No funding was received fort he SR.

Commercial interference was reported. Seven of the articles stated that the trial design and data analysis was independent of commercial interference;

nevertheless, another three studies did not

make this clarification.

Inclusion criteria SR:

1. Type of study: randomized controlled trials;

2. Population: adult patients with respiratory failure

who received oxygen therapy;

3. Intervention: HFNC treatment compared either

COT or NIV;

4. Predefined outcomes: intubation, mechanical

ventilation (includes noninvasive mechanical

ventilation and invasive mechanical ventilation),

escalation of respiratory support (HFNC, NIV or

intubation) and mortality.

Exclusion criteria SR:

Without complete information

- Not English/Chinese

- Applying HFNC for bronchoscopy, not for treatment

- Not included relevant outcomes

11 studies included

Important patient characteristics at baseline:

N, age

Total enrolled 3459 patients (HFNC, n = 1681)

Age was not reported

Unknown wether groups were comparable at baseline.

Describe intervention: high flow nasal cannula oxygen (HFNC)

A: HNFC

B: HNFC

C: HNFC

D: HNFC

E: HNFC

F: HNFC

G: HNFC

H: HNFC

I: HNFC

J: HNFC

K: HNFC

Describe control: Convential oxygen therapy or non-invasive ventilation

FM=face mask

A: FM/nasal

prongs

B: FM/nasal

cannula

C: FM/NIV

D: FM/nasal

cannula

E: FM/nasal

prongs

F: FM

G: FM

H: FM/or nasal

Prongs

I: FM

J: NIV

K: NIV

End-point of follow-up:

NA

For how many participants were no complete outcome data available?

(intervention/control)

Not reported, however for all studies was incomplete outcome data (attrition bias) reported as low risk according tot the Cochrane RoB tool.

Outcome measure-1a

Defined as intubation rate COT versus HFNC

Effect measure: OR (95% CI):

A: 0.35 (0.01-8.9)

B: 0.18 (0.01-3.77)

C: 0.72 (0.41-1.26)

D: 0.37 (0.19-0.73)

E: 0.29 (0.03-2.78)

F: 1.17 (0.30-4.64)

G: 0.29 (0.10-0.81)

H: 5.09 (0.24-106.8)

COT

Pooled effect (random effects model): OR 0.52 (95% CI 0.34 to 0.79),p=0.002 favoring

Heterogeneity (I²):9%

Outcome measure-1b

Defined as intubation rate NIV versus HFNC

Effect measure: OR (95% CI):

C: 0.62 (0.36-1.06)

J: 1.25 (0.84-1.85)

K: 1.03 (0.69-1.52)

NIV

Pooled effect (random effects model): OR 0.96 (95% CI 0.66 to 1.39 ),p=0.84 favoring

Heterogeneity (I²):53%

Outcome measure-2a

Defined as In-hospital mortality COT versus HFNC

Effect measure: OR (95% CI):

C: 0.55 (0.25-1.22)

D: 1.00 (0.20-4.98)

E: 1.34 (0.76-2.38)

G: 1.20 (0.34-4.20)

H: 1.01 (0.06-16.21)

Pooled effect (random effects model): OR 1.01 (95% CI 0.67 to 1.73),p=0.52 favoring COT

Heterogeneity (I²):0%

Outcome measure-2b

Defined as In-hospital mortality NIV versus HFNC

Effect measure: OR (95% CI):

C: 0.40 (0.19-0.83)

J: 1.15 (0.59-2.24)

K: 1.24 (0.70-2.19)

Pooled effect (random effects model): OR 0.85 (95% CI 0.43 to 1.68 ),p=0.65 favoring HFNC

Heterogeneity (I²):69%

Brief description of author’s conclusion:

Compared to COT, HFNC reduced the rate of intubation, mechanical ventilation and the

escalation of respiratory support. When compared to NIV, HFNC showed no better outcomes. Large-scale

randomized controlled trials are necessary to prove our findings.

Level of evidence according to authors: GRADE (per comparison and outcome measure) including reasons for down/upgrading

Intubation rate: moderate (lack of blinding of participants and personnel)

Mortality: Low (lack of blinding and imprecision)

Lin, 2017

Study characteristics and results are extracted from the SR (unless stated otherwise)

SR and meta-analysis of RCTs

Literature search up to September 2016

A: Azevedo 2015

B: Bell 2015

C: Frat 2015

D: Jones 2015

E: Lemaile 2015

F: Maggiore 2014

G: Rittayami 2015à exclusion: cross-over trial

H: Stephan 2015

Study design:

Parallel RCTs, one cross-over trial we exclude for our literature analysis

Setting:

A: surgical ICU

B: ED

C: ICU

D: ED

E: General ICU

F: General ICU

G: ED

H: Cardiac ICU

Source of funding:

No funding bodies played any role in study design, data collection and analysis, decision to publish, or preparation this

manuscript.

Inclusion criteria SR:

RCTs that

compared administration of HFNC versus conventional oxygen therapy

(COT) and noninvasive ventilation (NIV) in acute hypoxemic respiratory

failure and not requiring immediate ventilatory support

patients (such as those admitted to an ICU or emergency department),

and which reported the incidence of the rate of intubation,

or mortality.

8 studies included

Important patient characteristics at baseline:

N

A: 30

B: 100

C: 310

D: 303

E: 100

F: 105

G: 40

H: 830

All with acute respiratory failure

E: immunocompromised patients

F: postextubation

H: postextubation

Unknown whether groups were comparable at baseline.

Describe intervention: high flow oxygen

A: HFNC (n=14)

B: HFNC (n=48)

C: HFNC (n=14)

D: HFNC (n=106)

E: HFNC (n=165)

F: HFNC (n=52)

G: HFNC (n=20)

H: HFNC (n=414)

Describe control: low flow oxygen

A: NIV (n=16)

B: COT (n=52)

C-1: COT: 95

C-2: NIV:110

D: COT (n=138)

E: COT (n=48)

F: COT (n=52)

G: COT (n=20)

H: NIV (n=416)

End-point of follow-up:

NA

For how many participants were no complete outcome data available?

(intervention/control)

Attrition bias was defined low for all studies except Jones (D) reported some severely hypoxic patients were excluded from the study because clinicians were not comfortable with delaying NIV even for 30 min to deliver the study interventions.

Outcome measure-1a

Defined as intubation rate HFNC versus COT

Effect measure: OR (95% CI):

B: 0.35 (0.01-8.90)

C-1: 0.64 (0.40-1.04)

D: 0.27 (0.03-2.67)

E: 1.92 (0.33-10.97)

F: 0.15 (0.03-0.70)

G: Not estimable (no events)

Pooled effect (random effects model): OR 0.58 (95% CI 0.36 to 0.92),p=0.02 favoring HFNC

Heterogeneity (I²):29%

Outcome measure-1b

Defined as intubation rate HFNC versus NIV

Effect measure: OR (95% CI):

A: 1.25 (0.29-5.41)

C-2: 0.64 (0.40-1.04)

H: 1.03 (0.69-1.52)

Pooled effect (random effects model): OR 0.87 (95% CI 0.64 to 1.19),p=0.27 favoring HFNC

Heterogeneity (I²):24%

Outcome measure-2

Defined as In-hospital mortality HFNC versus NIV and COT

Effect measure: OR (95% CI):

C: 0.45 (0.23-0.90)

D: 1.15 (0.51-2.60)

F: 1.20 (0.34-4.20)

H: 1.24 (0.70-2.19)

Pooled effect (random effects model): OR 0.89 (95% CI to),p=0.51 favoring HFNC

Heterogeneity (I²):47%

Brief description of author’s conclusion:

The use of HFNC showed a trend toward reduction in the intubation rate, which did not

meet statistical significance, in patients with acute respiratory failure compared with conventional oxygen

therapy (COT) and noninvasive ventilation (NIV). Moreover no difference in mortality. So, Large, well-designed, randomized, multi-center trials are needed to confirm the effects of HFNC in acute hypoxemic respiratory failure patients.

RoB according to authors:

Low risk on selection bias for all studies

High risk on performance bias (no blinding of participants and personnel) in all studies and all but one study also no blinding for outcome assessment.

One study had high risk for attrition bias and one study had high risk for reporting bias. All other studies had a low risk for attrition and reporting bias. No other biases were reported.

Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Doshi, 2018

Type of study: RCT

Setting: ED, multicenter

Country: US

Source of funding: Dr. Doshi reports

receiving funding from Zoll Inc for coordination of the clinical trial.

Dr. Miller and Mr. Dunlap were employed by Vapotherm. All authors

report receiving study-related support from Vapotherm, Inc. The trial was sponsored by Vapotherm, which participated in the design

of the protocol and the selection and management of the sites. The sponsor agreed a priori to allow publication of the findings of

the trial at the discretion of the principal investigators. A steering

committee composed of the principal investigators oversaw the

analysis and interpretation of the data, wrote the first and

subsequent drafts of the article, and made the decision to submit the article for publication.

Inclusion criteria:

-older than 18 years -with clinical judgment of the treating clinician of acute respiratory failure requiring escalation to noninvasive

positive-pressure ventilation or to maintain noninvasive

positive-pressure ventilation if the patient was delivered to

the ED while receiving either type of ventilation from the out-of-hospital setting.

Exclusion criteria:

suspected

drug overdose, cardiovascular instability (hypotension

requiring immediate intervention), end-stage cancer, life

expectancy less than 6 months, significant respiratory

depression on presentation (eg, drug overdose), Glasgow

Coma Scale score less than 9, cardiac or respiratory arrest on presentation, need for emergency intubation, known or suspected cerebrovascular accident, known or suspected ST-segment elevation myocardial infarction, and patients with increased risk of pulmonary aspiration, agitation, or

uncooperativeness.

N total at baseline:

Intervention: 116, 104 received allocated intervention

Control: 112, 100 received allocated intervention

Important prognostic factors²:

age ± SD:

I: 63.4 (13.6)

C: 63.3 (14.8)

Sex:

I: 42% M

C: 44 % M

Sepsis/pneumonia discharge diagnosis

HFNC: 15 (14%)

NIPPV: 20 (20%)

Groups were comparable at baseline

Describe intervention (treatment/procedure/test):

High-velocity nasal insufflation (Precision Flow;

Vapotherm, Inc, Exeter, NH) using a small-bore

nasal cannula was initiated with a flow rate set to 35 L/min, with a starting temperature between 35°C and 37°C and

FiO2 at 1.0. Adjustments in flow (up to 40 L/min) and

temperature (typically between 35°C and 37°C) were made to alleviate respiratory distress and optimize comfort.

The target for each

intervention was to decrease breathing rate to fewer than 25 breaths/min and optimize comfort, whereas FiO2 was

adjusted to maintain a pulse oximetry reading (SpO2)

greater than 88%. The study model provided for having a

respiratory therapist at bedside for the first 4 hours, which facilitated rapid changing of settings as needed.

Describe control (treatment/procedure/test):

Noninvasive positive-pressure ventilation (Respironics Vision V60; Philips Healthcare, Murrysville, PA) was

initiated with an oronasal mask, with inspiratory and

expiratory positive airway pressures (IPAP, EPAP) set at the lower end of the following settings and increased as necessary to alleviate respiratory distress: IPAP 10 to 20 cm H2O (or 5 to 15 cm H2O above EPAP), and EPAP 5 to 10 cm H2O. FiO2 was initiated at 1.0 for noninvasive

positive-pressure ventilation.

The target for each

intervention was to decrease breathing rate to fewer than 25 breaths/min and optimize comfort, whereas FiO2 was

adjusted to maintain a pulse oximetry reading (SpO2)

greater than 88%. The study model provided for having a

respiratory therapist at bedside for the first 4 hours, which facilitated rapid changing of settings as needed.

Length of follow-up:

NA

Loss-to-follow-up/incomplete outcome data:

The 24 patients randomized but not enrolled were

excluded for meeting exclusion criteria (10), consent not

obtained or withdrawn (6), bedside clinician not

comfortable with enrollment after randomization (2), and

patient identified to not need noninvasive positive-pressure

ventilation after initial evaluation, thus failing to meet

inclusion criteria (6).

Outcome measures and effect size (include 95%CI and p-value if available):

Need for intubation within 72 hours:

I: 7/104

C: 13/100

% Risk difference: -7 (95% CI -14 to 2)

Arm failure rate, defined as the decision for crossover to the alternate therapy, within 72 hours of initiation of assigned therapy.

I: 27/104

C: 17/100

% Risk difference: 9 (95% CI -2 to 20)

LOS ICU, mean days (sd)

I: 3.3 (3.7) n=48

C: 3.9 (4.1) n=47

Difference (95%CI): -0.6 (-2.2 to 1.0)

LOS ED, hours (sd)

I: 7.6 (5.0) n=104

C: 8.0 (8.5)n=400

Difference (95%CI): -0.4 (-2.3 to 1.5)

LOS hospital, mean days (sd)

I: 6.8 (5.7) n=104

C: 6.0 (4.4) n=100

Difference: 0.8 (95% CI -0.6 to 2.2)