Journal of Critical Care (2010) 25, 463–468

High-flow nasal oxygen vs high-flow face mask:

A randomized crossover trial in extubated patients
Ravindranath Tiruvoipati ⁎, David Lewis, Kavi Haji, John Botha
Department of Intensive Care Medicine, Frankston Hospital, Frankston, Victoria 3199, Australia

Keywords:
Abstract
Face mask;
Purpose: Oxygen delivery after extubation is critical to maintain adequate oxygenation and to avoid
Nasal prongs;
reintubation. The delivery of oxygen in such situations is usually by high-flow face mask (HFFM). Yet,
Oxygen;
this may be uncomfortable for some patients. A recent advance in oxygen delivery technology is high-
Gas exchange
flow nasal prongs (HFNP). There are no randomized trials comparing these 2 modes.
Methods: Patients were randomized to either protocol A (n = 25; HFFM followed by HFNP) or protocol
B (n = 25; HFNP followed by HFFM) after a stabilization period of 30 minutes after extubation. The
primary objective was to compare the efficacy of HFNP to HFFM in maintaining gas exchange as
measured by arterial blood gas. Secondary objective was to compare the relative effects on heart rate,
blood pressure, respiratory rate, comfort, and tolerance.
Results: Patients in both protocols were comparable in terms of age, demographic, and physiologic
variables including arterial blood gas, blood pressure, heart rate, respiratory rate, Glasgow Coma Score,
sedation, and Acute Physiology and Chronic Health Evaluation (APACHE) III scores. There was no
significant difference in gas exchange, respiratory rate, or hemodynamics. There was a significant
difference (P = .01) in tolerance, with nasal prongs being well tolerated. There was a trend (P = .09)
toward better patient comfort with HFNP.
Conclusions: High-flow nasal prongs are as effective as HFFM in delivering oxygen to extubated
patients who require high-flow oxygen. The tolerance of HFNP was significantly better than in HFFM.
Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved.

1. Introduction masks to be uncomfortable and claustrophobic, leading to

Optimal oxygen delivery after extubation is critical to
maintain adequate oxygenation and avoid reintubation. The
delivery of oxygen in such situations is usually by high-flow
face mask (HFFM). Although face masks are used
commonly and are effective, some patients find these
⁎ Corresponding author. Tel.: +61431279347.
E-mail address: travindranath@hotmail.com (R. Tiruvoipati).
frequent removal [1]. Furthermore, it is difficult to commu-
nicate and eat while receiving oxygen through face masks
[2]. Face masks are often displaced, particularly, while the
patients are sleeping [3,4]. All these factors can potentially
lead to inadequate delivery of the prescribed oxygen with the
associated complications.
The other commonly used oxygen delivery system is
nasal cannula. Nasal cannulas have several potential
advantages over a face mask. They are generally better
tolerated than face masks [2]. They do not interfere with

0883-9441/$ – see front matter. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.jcrc.2009.06.050
464
eating or drinking, allow patients to talk without removal,
and do not cause claustrophobia. Nasal cannulas have been
found to be more comfortable than face masks [5]. Given
these advantages, it is not surprising that most patients prefer
nasal cannulas than face masks [6]. In spite of this, there are
concerns that nasal cannulas may not deliver adequate
oxygen in “mouth breathers” [1]. Moreover, nasal cannulas
are traditionally used only in situations where the required
oxygen delivery is not more than 4 L/min because of the
potential drying of the nasal mucosa [7].
A recent advance in oxygen delivery technology is the
high-flow nasal prongs (HFNP; Fisher and Paykel Health-
care, Auckland, New Zealand). These can deliver up to
50 L/min of warm and humidified oxygen similar to HFFM.
Humidification prevents the nasal mucosa from drying up.
In addition, HFNP were also shown to generate a significant
positive airway pressure [8], which could further aid in
improving oxygenation. High-flow nasal prongs have been
Fig. 1 Summary of the flow of participants in the trial.
R. Tiruvoipati et al.
used and validated in the neonatal population [9-11]. A few
small studies have also reported on their usefulness in the
adult population [8,12], but, to date, there is no randomized
evaluation of their efficacy in adult patients.
The primary objective of this study was to compare the
relative efficacy of HFNP with HFFM in maintaining gas
exchange. Secondary objective included the comparison of
relative effects of the 2 delivery modes on heart rate, blood
pressure, respiratory rate, patient comfort, and tolerance.
2. Methods
2.1. Ethics and enrollment procedure
Ethical approval was obtained from the Human Research
Ethics Committee for Clinical Research at Peninsula Health.
High-flow nasal oxygen vs high-flow face mask 465

Table 2 Exit criteria

Respiratory
(1) Oxygen saturation b90%
(2) Respiratory rate N30 breaths/min
(3) pH b7.30
Fig. 2 Comfort score. A score of 0 indicates no discomfort; 1,
Cardiovascular
little bit discomfort; 2, little more discomfort; 3, even more
(1) Heart rate N120 beats/min
discomfort; 4, whole lot more discomfort; 5, worst discomfort.
(2) Systolic blood pressure b100 mm Hg
Central nervous system
(1) Altered level of consciousness
The family/next of kin of the potential participant was first
presented with an information leaflet. After the family had
read and understood the information pertaining to the study,
Glasgow coma score (GCS) were recorded. If the patients
researchers approached them for assent for the enrollment of
remained stable, they were randomized to either protocol A
patients into the trial. Further explanation and clarifications
(HFFM for 30 minutes followed by HFNP for 30 minutes) or
were provided at this stage.
protocol B (HFNP for 30 minutes followed by HFFM for
30 minutes; Fig. 1). During both protocols, gas flow rate
2.2. Participant eligibility and FIO2 were maintained at similar settings as used during
the stabilization period. Blood gases and other observations
All adult patients (N18 years) with family to provide (heart rate, blood pressure, respiratory rate, saturation,
assent were eligible for the trial. In addition, patients had to GCS, sedation score, comfort score, and tolerance score)
fulfill the following criteria before extubation to be eligible were recorded at the end of each protocol. The bedside
for randomization: pH 7.35 to 7.45, oxygen saturation higher nurse assessed the comfort and the tolerance of patients to
than 90%, FIO2 lower than 40%, and positive end-expiratory both devices by a visual analogue scale (Fig. 2 and
pressure (PEEP) of 5 cm H2O or less. Patients were excluded Table 1). Randomization was performed using sequentially
if they were known to have severe chronic obstructive airway numbered, sealed, opaque envelope to maintain allocation
disease and were carbon dioxide retainers. concealment [13].
Furthermore, patients were excluded if extubation was
part of the withdrawal of active treatment or death was 2.4. Safety of participants in the trial
imminent or if there was no family member to consent to
the trial.
To ensure patient safety while the patients were in the
study, a set of exit criteria was used. Patients were removed
2.3. Study design from the trial and treated according to the clinical condition if
they had any of the exit criteria (Table 2).
A randomized crossover study of ventilated patients
suitable for extubation was conducted (Fig. 1). Potential 2.5. Statistical analysis
patients were clinically assessed for suitability of extubation,
and arterial blood gases (PaO2 N70 mm Hg, PCO2 35-45 mm
Data were collected on an Excel spreadsheet and exported
Hg, pH N7.30, SaO2 N93% on a FIO2 b0.4, and PEEP of ≤5)
to SPSS (version 15.0 for Windows, SPSS Inc, Chicago, IL)
were performed before extubation to ensure the patients had
for analysis. The variables were then analyzed using Fisher
optimal gas exchange. If found to be suitable for extubation,
exact test or paired Student t test. P values reported are 2-
patients were extubated and were provided with a high-flow
blend of oxygen and air by HFFM at a rate of 30 L/min,
delivering an FIO2 of 30% to 40% to achieve an oxygen Table 3 Comparison of patients in protocols A and B at
saturation of 93% ± 1 for 30 minutes. At the end of this randomization a
30-minute stabilization, arterial blood gas was performed, Variable Protocol A, Protocol B,
and heart rate, blood pressure, sedation score [7], and mean (SD) mean (SD)
Age (y) 65.04 (19.52) 65.42 (15.80)
Sex (male-female) 9:13 11:9
Table 1 Tolerance score
Diagnosis (medical-surgical) 11:11 14:6
(1) Patient never removes apparatus and does not complain No. of hours ventilated 54.4 (58.5) 52.8 (57.7)
(2) Patient complains about keeping apparatus on Sedation score 1.7 (1.0) 2.0 (1.7)
(3) Patient request removal of apparatus GCS 11.4 (2.0) 11.1 (2.3)
(4) Removal of apparatus by patient once APACHE III score 70.55 (27.39) 72.95 (23.22)
(5) Noncompliance a
None of the comparisons are statistically significant.
466 R. Tiruvoipati et al.

Table 4 Comparison of patients in protocols A and B before care unit (ICU) in patients randomized to protocol A. Two
interventions a patients randomized to protocol B died in the ICU (P = .21).
Variable Protocol A, Protocol B, Two patients in protocol A and 3 patients in protocol B died
mean (SD) mean (SD) before being discharged from the hospital (P = .65).
PaO2 (mm Hg) 102.5 (49.0) 93.4 (25.8)
PCO2 (mm Hg) 38.0 (7.12) 37.9 (5.3)
pH 7.39 (0.06) 7.37 (0.05) 4. Discussion
HCO3 (mmol/L) 23.11 (3.01) 22.57 (2.96)
Oxygen saturation (%) 94.75 (6.51) 95.9 (2.72)
Our study reveals that HFNP are comparable with HFFM
Heart rate (beats/min) 85.10 (19.46) 84.46 (15.19)
SBP (mm Hg) 136.63 (19.39) 141.20 (24.7) in terms of providing adequate gas exchange. High-flow
DBP (mm Hg) 64.42 (13.55) 71.40 (14.93) nasal prongs were tolerated better than HFFM, and there was
Sedation score 1.61 (0.97) 1.40 (0.91) a trend toward better patient comfort with the use of HFNP
GCS 14.53 (0.96) 14.60 (1.05) that did not reach statistical significance.
SBP indicates systolic blood pressure; DBP, diastolic blood pressure. The use of high-flow nasal oxygen has been studied
a
None of the comparisons are statistically significant. mostly in neonatal patients [9-11]. Studies evaluating high-
flow humidified oxygen therapy in neonatal patients claim it
to be associated with improved respiratory function post-
tailed, and we assumed statistical significance if P is less extubation and a low rate of reintubation [9,11]. In adults,
than .05. there have been studies evaluating the physiologic parameters
with the use of high-flow humidified oxygen [8,12].
However, data on the use of HFNP in adult patients are
3. Results scant. Our study is the first randomized trial evaluating the
efficacy of use of HFNP.
Fifty patients were recruited with 25 patients each The results of our study are comparable with other studies
randomized to protocol A or B. where nasal prongs were shown to be more comfortable
Eight patients were excluded from the final analysis (4 [1,5]. In the study by Eastwood et al [5], a significant
patients met exit criteria, and complete data were missing in difference was noted in terms of patient comfort when nasal
4 patients). As shown in Table 3, both groups were devices were compared with face masks. In our study, there
comparable in terms of age, sex, diagnoses, numbers of was a trend toward better patient comfort with the use of
hours ventilated before extubation, GCS, sedation, and HFNP as compared with HFFM. This, however, did not
APACHE III scores at randomization. After stabilization of reach statistical significance, possibly because of a smaller
30 minutes, the gas exchange and other physiologic number of patients in our study.
parameters were comparable (Table 4) between patients Eastwood and colleagues [5] found that the face mask use
randomized to protocols A and B. was associated with higher oxygen requirements than nasal
There was no statistically significant difference in gas devices to maintain oxygen saturation of 95% or higher.
exchange when HFNP were compared with HFFM (Table 5). Their study, however, used a low oxygen flow (2.6 ± 1.0 L/
The heart rate, blood pressure, and respiratory rate were also min with nasal prongs and 6.1 ± 0.4 L/min with the face
comparable between HFNP and HFFM (Table 6). There was, masks). In our study, where high-flow oxygen was used, we
however, a significant difference (P = .01) in terms of did not find a difference in oxygen saturations with either of
tolerance, with HFNP being better tolerated (Table 6). the devices we compared.
Tolerance scores were 3 or higher in 4 patients with the use of Although our study indicates HFNP to be as efficacious as
HFFM and in 1 patient with HFNP. HFFM in providing gas exchange, it does not address other
There was a trend (P = .09) toward better patient comfort important issues such as the requirement of reintubation,
with HFNP (Table 6) that did not reach conventional noninvasive ventilation, duration of ICU, and hospital stay
statistical significance. There were no deaths in the intensive and survival.

Table 5 HFNP vs HFFM—gas exchange (primary objective)

Variable HFNP HFFM Mean difference 95% Confidence interval P
PaO2 (mm Hg; mean [SD]) 102.14 (40.25) 98.35 (38.54) 3.79 (23.69) −4.89 to 12.48 .38
PCO2 (mm Hg; mean [SD]) 37.53 (6.23) 37.91 (6.22) −0.38 (0.54) −1.49 to 0.72 .48
Saturation (%), mean (SD) 95.25 (6.12) 95.64 (5.32) −0.38 −12 to 0.46 .36
HCO3 (mmol/L; mean [SD]) 23.37 (3.00) 22.86 (2.86) 0.50 (1.92) −0.19 to 1.2 .15
pH, mean (SD) 7.39 (0.05) 7.39 (0.05) 0.00 (0.03) −0.01 to 0.01 .92
High-flow nasal oxygen vs high-flow face mask 467

Table 6 HFNP vs HFFM (secondary objectives)

Variable HFNP HFFM Mean difference 95% Confidence interval P
Heart rate (beats/min; mean [SD]) 78.86 (15.29) 112.56 (192.45) −33.70 (191.93) −105.37 to 37.96 .34
SBP (mm Hg; mean [SD]) 135.70 (21.89) 133.86 (25.69) 1.83 (15.20) −3.84 to 7.51 .51
DBP (mm Hg; mean [SD]) 66.16 (12.03) 63.33 (14.45) 2.83 (9.38) −0.66 to 6.33 .10
Respiratory rate (breaths/min; mean [SD]) 18.68 (5.51) 19.68 (6.50) −1.0 (3.93) −2.49 to 0.49 .18
Comfort score, mean (SD) 0.53 (1.04) 0.96 (1.42) −0.43 (1.35) −0.93 to 0.07 .09
Tolerance score, mean (SD) 0.20 (0.66) 0.66 (1.29) −0.46 (0.97) −0.83 to −0.10 .01

In our study, only 1 patient (in protocol A) required
noninvasive ventilation followed by reintubation 4 hours
after extubation. We believe a larger sample size is required
to further clarify these issues.
Provision of positive airway pressure is known to improve
oxygenation [14,15]. It has been shown in earlier studies that
the use of HFNP generates significant positive airway
pressure [8,10]. The amount of PEEP in patients adminis-
tered with high-flow oxygen may be dependent on whether
the patient is breathing with an open or closed mouth. The
study by Groves and Tobin [8] has demonstrated that high-
flow nasal oxygen generated positive expiratory pressures,
irrespective of whether the breathing was with an open or
closed mouth. However, the increase in airway pressure was
significantly more with a closed mouth (up to 7.4 cm of
water with mouth closed as compared with 2.7 cm of water
with mouth open). We have not recorded if patients were
breathing with an open or closed mouth, and this may have
had an impact on our results, especially regarding the effect
on oxygenation.
Tolerance and comfort scores were recorded by bedside
nurses, and it was not possible to blind them for the
intervention being tested. The scoring, therefore, to some
extent may have been subjective and influenced by the
personal preferences of the person recording the score.
4.1. Limitations
A washout period after each intervention may have been a
more certain way of eliminating the effects of the mode of
prior oxygen delivery. However, this would have increased the
duration of the study period with its associated difficulties.
Previous studies assessing the impact of changes in FIO2
have shown that 5 to 10 minutes is sufficient to measure the
effects of the changes of FIO2 on gas exchange [16].
Our measurement time after each intervention was 30
minutes. We, therefore, do not believe that the lack of a
washout period between interventions significantly affected
our results.
5. Conclusions
High-flow nasal prongs were found to be as effective as
HFFM in delivering oxygen in extubated patients who
require high flow. Patient tolerance of HFNP was signifi-
cantly better than that in HFFM.
Acknowledgments
The authors would like to thank Nina Fowler, Laurel
Walker, Jodi Vuat, Naomi Pratt, and all our ICU staff for their
help in this study and also Fisher and Paykel Healthcare for
supporting this study.
References
[1] Sasaki H, Yamakage M, Iwasaki S, et al. Design of oxygen delivery
systems influences both effectiveness and comfort in adult volunteers.
Can J Anaesth 2003;50:1052-5.
[2] Stewart AG, Howard P. Devices for low flow O2 administration. Eur
Respir J 1990;3:812-7.
[3] Nolan KM, Winyard JA, Goldhill DR. Comparison of nasal cannulae
with face mask for oxygen administration to postoperative patients. Br
J Anaesth 1993;70:440-2.
[4] Nolan KM, Baxter MK, Winyard JA, et al. Video surveillance of
oxygen administration by mask in postoperative patients. Br J Anaesth
1992;69:194-6.
[5] Eastwood GM, O'Connell B, Gardner A, et al. Evaluation of
nasopharyngeal oxygen, nasal prongs and facemask oxygen therapy
devices in adult patients: a randomised crossover trial. Anaesth
Intensive Care 2008;36:691-4.
[6] Costello RW, Liston R, McNicholas WT. Compliance at night with low
flow oxygen therapy: a comparison of nasal cannulae and Venturi face
masks. Thorax 1995;50:405-6.
[7] Kallstrom TJ. AARC Clinical Practice Guideline: oxygen therapy for
adults in the acute care facility—2002 revision & update. Respir Care
2002;47:717-20.
[8] Groves N, Tobin A. High flow nasal oxygen generates positive airway
pressure in adult volunteers. Aust Crit Care 2007;20:126-31.
[9] Woodhead DD, Lambert DK, Clark JM, Christensen RD. Comparing
two methods of delivering high-flow gas therapy by nasal cannula
following endotracheal extubation: a prospective, randomized,
masked, crossover trial. J Perinatol 2006;26:481-5.
[10] Sreenan C, Lemke RP, Hudson-Mason A, Osiovich H. High-flow nasal
cannulae in the management of apnea of prematurity: a comparison
with conventional nasal continuous positive airway pressure. Pedia-
trics 2001;107:1081-3.
[11] Shoemaker MT, Pierce MR, Yoder BA, DiGeronimo RJ. High flow
nasal cannula versus nasal CPAP for neonatal respiratory disease: a
retrospective study. J Perinatol 2007;27:85-91.
[12] Wettstein RB, Shelledy DC, Peters JI. Delivered oxygen concentra-
tions using low-flow and high-flow nasal cannulas. Respir Care 2005;
50:604-9.
468
[13] Doig GS, Simpson F. Randomization and allocation concealment: a
practical guide for researchers. J Crit Care 2005;20:187-91.
[14] Kindgen-Milles D, Buhl R, Gabriel A, Bohner H, Muller E. Nasal
continuous positive airway pressure: a method to avoid endotracheal
reintubation in postoperative high-risk patients with severe non-
hypercapnic oxygenation failure. Chest 2000;117:1106-11.
[15] Kurt M, Boeken U, Litmathe J, Feindt P, Gams E. Oxygenation
failure after cardiac surgery: early re-intubation versus treatment by
R. Tiruvoipati et al.
nasal continuous positive airway pressure (NCPAP) or non-invasive
positive pressure ventilation (NPPV). Monaldi Arch Chest Dis 2008;
70:71-5.
[16] Cakar N, Tuorul M, Demirarslan A, Nahum A, Adams A, Akyncy O,
et al. Time required for partial pressure of arterial oxygen
equilibration during mechanical ventilation after a step change in
fractional inspired oxygen concentration. Intensive Care Med 2001;
27:655-9.