Accepted Manuscript

High-flow nasal therapy vs standard oxygen during breaks off

noninvasive ventilation for acute respiratory failure: A pilot
randomized controlled trial

Giulia Spoletini, Chiara Mega, Lara Pisani, Mona Alotaibi, Alia

Khoja, Lori Lyn Price, Francesco Blasi, Stefano Nava, Nicholas
S. Hill

PII: S0883-9441(18)31068-2
DOI: doi:10.1016/j.jcrc.2018.10.004
Reference: YJCRC 53076
To appear in: Journal of Critical Care

Please cite this article as: Giulia Spoletini, Chiara Mega, Lara Pisani, Mona Alotaibi, Alia
Khoja, Lori Lyn Price, Francesco Blasi, Stefano Nava, Nicholas S. Hill , High-flow nasal
therapy vs standard oxygen during breaks off noninvasive ventilation for acute respiratory
failure: A pilot randomized controlled trial. Yjcrc (2018), doi:10.1016/j.jcrc.2018.10.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
 ACCEPTED MANUSCRIPT

High-flow nasal therapy vs standard oxygen during breaks off noninvasi ve ventilation for acute

respiratory failure: a pilot randomized controlled trial

Giu lia Spoletini M Da,1 , Chiara Mega M Da,2 , Lara Pisani MDa ,3 , Mona Alotaib i M Da,4 , A lia Khoja

MDa,5 , Lori Lyn Price MASb,c, Francesco Blasi MD PhD FERSd , Stefano Nava MD PhDe, Nicholas S

Hill, M Da,* nhill@tuftsmedicalcenter.org

P T
RI
SC
NU
MA
ED
PT
CE
AC

1T he Leeds Regional Adult CF Centre and Respiratory Department, St James’s University Hospital, Leeds T eaching Hospital
NHS T rust, Leeds, UK.
2Department of Anaesthesia and Intensive Care, Azienda Ospedaliero-Universitaria Careggi, Florence, Italy.
3Department of Clinical, Integrated and Experimental Medicine (DIMES), Respiratory and Critical Care Unit, Ospedale
Sant’Orsola Malpighi, Alma Mater University, Bologna, Italy.
4Division of Pulmonary, Critical Care and Sleep Medicine, University of California at San Diego, San Diego, CA, USA.
5Department of Medicine, George Washington University, Washington DC, USA.

1
 ACCEPTED MANUSCRIPT

a
Pulmonary, Critical Care and Sleep Medicine Division, T ufts Medical Center, Boston MA, USA
b
T ufts Clinical and T ranslational Science Institute, Tufts University, Boston MA, USA
c
T he Institute for Clinical Research and Health Policy Study, Tufts Medical Center, Boston MA, USA
d
Department of Pathophysiology and Transplantation, Università degli Studi di Milano, IRCCS Fondazione Ospedale Maggiore

Policlinico Ca’ Granda, Milan, Italy

e
Department of Clinical, Integrated and Experimental Medicine (DIMES), Respiratory and Critical Care Unit, Ospedale

T
Sant’Orsola Malpighi, Alma Mater University, Bologna, Italy

P
*

RI
Correspondi ng author at: Pulmonary, Critical Care and Sleep Division, Tufts Medical Center, 800

Washington st, 02111, Boston MA USA.

SC
Highlights:

 Pilot randomized trial assessing the role of HFNT as complementary therapy to NIV
NU
 HFNT does not affect the time on and off NIV compared to SO

 HFNT is well tolerated and leads to greater comfort than SO

MA

 HFNT prevents increase in RR and dyspnea observed with SO during breaks off NIV

 HFNT could be a suitable alternative to SO during breaks off NIV

ED
PT
CE
AC

2
 ACCEPTED MANUSCRIPT

ABSTRACT

Purpose

To assess the role of high-flow nasal therapy (HFNT) compared to standard oxygen (SO) as

complementary therapy to non-invasive ventilation (NIV).

Methods

Multicenter trial including patients (n=54) anticipated to receive NIV for ≥ 24 hours due to acute or

T
acute-on-chronic respiratory failure. Subjects were randomized (1:1) to SO or HFNT during breaks off

P
NIV. Primary outcome was total time on and off NIV. Secondary outcomes were comfort and dy spnea,

RI
respiratory rate (RR), oxygen saturation (Sp O2 ), tolerance and side effects.

SC
Results

Total time per patient on NIV (1315 vs 1441 min) and breaks (1362 vs 1196 min), and mean duration
NU
of each break (520 vs 370 min) were similar in the HFNT and SO arms (p>0.05).

Comfort score was higher on HFNT than on SO (8.3±2.7 vs 6.9±2.3, p=0.001). Dyspnea, RR and Sp O2
MA

were similar in the two arms, but the increase in RR and dyspnea seen with SO during breaks did not

occur with HFNT.

Conclusion
ED

Compared to SO, HFNT did not reduce time on NIV. However, it was more comfortable and the

increase in RR and dyspnea seen with SO did not occur with HFNT. Therefore, HFNT could be a
PT

suitable alternative to SO during breaks off NIV.

CE
AC

3
 ACCEPTED MANUSCRIPT

INTRODUCTION

Non-invasive ventilation (NIV 6) is the ventilatory modality of first choice for acute respiratory failure

(ARF 7 ) due to acute exacerbations of chronic obstructive pulmonary disease (COPD 8 ) and acute

cardiogenic pulmonary edema (A CPE 9 ) [1,2]. NIV avoids invasive ventilation by delivering

ventilation and oxygenation to the upper airway via a non-invasive tight-fitting interface. These

T
interfaces, however, are often uncomfo rtable for patients and are frequently removed to facilitate

P
interactions with family and caregivers or to eat. Thus, breaks fro m NIV are necessary in most patients

RI
to improve tolerance and are also used to evaluate the feasibility of weaning the patient off NIV.

SC
Little is known about the best way to manage breaks , which often last for periods of up to a few hours .

Standard oxygen therapy (SO 10), including conventional systems to deliver o xygen therapy such as
NU
nasal cannulas, simple face masks and Venturi-masks, is usually started to maintain target o xygen

saturation (Sp O2 11 ) during breaks. However, these techniques don’t provide a reliab le fraction of
MA

inspired o xygen (FI O2 12 ), wh ich depends on room air entrain ment, position of the interface and

breathing pattern, nor do they provide respiratory support. These limitations can predispose to
ED

increased respiratory rate (RR 13 ), dyspnea and oxygen desaturations during breaks and could

precipitate respiratory failure with need for endotracheal intubation.

PT

High-flo w nasal therapy (HFNT 14) delivers heated, humidified and oxygenated gas with flow rates up
CE

to 60 L/ min at FI O2 adjustable between 0.21 and 1.0 via soft, loose fitting, large bore nasal prongs that
AC

permit unimpeded speaking and eating. HFNT enhances oxygenation compared to SO by more closely

matching the patient’s inspiratory flo w rate and reducing roo m air entrain ment [3]. It washes out

nasopharyngeal dead space [4], imp roves the efficiency of ventilat ion[5], lo wers respiratory rate and

6 NIV: Noninvasive ventilation

7 ARF: Acute respiratory failure
8 COPD: Chronic obstructive pulmonary disease
9 ACPE: Acute cardiogenic pulmonary edema
10 SO: Standard oxygen therapy
11 S O : Oxygen saturation
p 2
12 F O : Fraction of inspired oxygen
I 2
13 RR: Respiratory rate
14 HFNT: High flow nasal therapy

4
 ACCEPTED MANUSCRIPT

provides a small positive end expiratory pressure (PEEP 15 ) effect, especially during closed-mouth

breathing [6–8]. These effects combined can lower work of breathing co mpared to SO [6,9,10]. The

warmed and humidified gas reduces cooling and dessication of the upper airway s, and enhances

comfo rt and tolerance. Furthermo re, it mo istens secretions and facilitates mucociliary clearance [9,11].

Studies on clinical applications of HFNT have reported its successful use in de novo ARF [12], in the

post-surgical [13] and post-extubation settings [14,15] and in do-not-intubate patients [16,17].

T
Most prior studies have examined HFNT as a rep lacement for NIV; in this study, we examine its use as

P
a co mplementary therapy. By v irtue of the demonstrable advantages over SO described above, we

RI
hypothesized that, compared to SO during breaks, HFNT would reduce the time on NIV by

SC
lengthening the duration of breaks, which would ult imately lead to weaning o ff NIV, and imp rove RR,

comfort and dyspnea.

NU
METHODS
MA

Study design

An unblinded, mu lticenter, randomized, parallel controlled trial was performed in the 5 ICUs and in the
ED

Intermediate Care Unit at Tufts Medical Center (Boston, MA, USA) and in the ICU at Winchester

Hospital (Winchester, MA, USA). The study was approved by the Institutional Review Board in both
PT

institutions, was conducted in accordance with the principle of the Declaration of Helsin ki and was

registered with ClinicalTrials.gov (NCT01925534). Written informed consent (ICF 16 ) was

CE

prospectively obtained from all patients or their legally authorized representative (LAR17) if a patient

was deemed unable to consent. Delayed consent was sought fro m patients who regained mental
AC

competency, when the ICF was initially signed by the LAR.

Patients

Consecutive patients were enrolled if they met the following inclusion criteria: age ≥18 years and on

acute NIV for an anticipated time ≥24 hours due to acute or acute on chronic hypercapnic respiratory

failure (pH <7.35 and p CO2 >45 mmHg ) o r acute hypoxemic respiratory failure (PaO2 /FIO2 <300 and

15 PEEP: Positive end-expiratory pressure

16 ICF: Informed consent form
17 LAR: Legally authorized representative

5
 ACCEPTED MANUSCRIPT

RR ≥24 breaths/min). NIV was init iated according to standard guidelines [2,18]. Exclusion criteria

were: usual contraindications to NIV [1][1] (Tab le S1, supplemental material), deliriu m, previous

participation in the study, NIV init iation ≥48 hours prior to screening and inability to obtain consent.

Patients who withdrew consent, underwent endotracheal intubation before receiving the first break,

were weaned off NIV and o xygen therapy before the first break or received a different device per

clinical team decision were secondarily excluded.

T
Randomization

P
Subjects were randomly assigned in a 1:1 ratio to receive either HFNT or SO during breaks off NIV,

RI
with no prespecified stratification criteria (i.e. hypoxemic vs hypercapnic, study site, etc).

SC
Randomization was done using a single computer-generated random-number sequence with

assignement made in a masked fashion by selection of opaque envelopes, kept in permuted blocks of
NU
10, each consisting of 5 allocations per treatment arm to better assure equal distribution of enrolees

throughout the study.

MA

Study outcomes

The primary outcome was total time per patient on and off NIV, and the mean duration of each NIV
ED

session and break off NIV. Total time per patient on NIV was co mputed as the cumulative t ime each

patient spent on NIV; similarly, total t ime per patient on breaks was the cumulat ive time each patient
PT

spent on breaks off NIV. Mean duration of each NIV session was the average length of time spent by a
CE

patient on NIV during a single session; similarly, mean duration of each break was the average length

of time spent by a patient off NIV between sessions (see Supplemental Material for further details).
AC

Secondary outcomes included subjective measurements of co mfort and dyspnea during breaks; mean

and final Sp O2 during breaks, RR at the end of breaks, ease of eating and side effects.

Study procedures

All subjects received NIV via a Vision or V60 ventilator (Philips Respironics, Murrysville, PA, USA)

set in pressure support (PS) mode with init ial PEEP of 4 cmH2 O and PS between 4 and 8 cmH2 O. NIV

settings were adjusted by experienced respiratory therapists in agreement with the ICU attending

physician to maintain a target tidal volu me (VT ) of 6-8 ml/ kg, RR ≤ 24/ min and FIO2 to maintain a

6
 ACCEPTED MANUSCRIPT

targeted oxygen saturation adequate gas exchange. Targeted Sp O2 was 88 to 92% in patients with acute

or acute on chronic hypercapnic respiratory failure or in subjects with combined hypercapnic and

hypoxemic respiratory failure and ≥92% in subjects with acute hypoxemic respiratory failure. Active

humid ification was provided by a MR850 and NIV was delivered via a RT219 heated circuit and a

Free Motion RT040 oro-nasal mask (all Fisher and Paykel Healthcare, NZ).

During breaks fro m NIV, patients randomized to SO received o xygen therapy humidified via a cold

T
water system through nasal cannulae, Venturi or non-rebreather face masks with flow-rate adjusted to

P
maintain adequate Sp O2 .

RI
SC
In the HFNT arm, patients were fitted with large bore nasal prongs (Optiflow cannula, OPT500 series,

Fisher & Paykel Healthcare, NZ), depending on the size o f the patients’ nostrils and received high-flo w
NU
therapy generated via a flow generator (MaxVenturi, A rmstrong Medical, UK). The air/o xygen blend

was actively heated and humid ified by a MR850, and supplied with a RT200 circu it (all Fisher and
MA

Paykel Healthcare, NZ). Init ial flow-rate was set at 35 L/ min; flow-rate and FIO2 were subsequently

adjusted to maintain oxygenation (see Supplemental Material).

ED

Criteria for restarting NIV and discontinuing breaks were: worsening dyspnea uncontrolled by
PT

adjusting SO/HFNT settings, increased RR or heart rate (HR 18 ) by 15%, increased or decreased
CE

systolic blood pressure (SBP 19) by 20%, drop in Sp O2 belo w target without recovery by increasing the

settings, or patient or clinician’s request to resume NIV.

AC

Nursing staff and respiratory therapist working in the study sites received train ing sessions on the

equipment, p rotocol to titrate the flow and criteria to start and d iscontinue breaks (see Supplemental

Material).

Data collection

18 HR: Heart rate

19 SBP: Systolic blood pressure

7
 ACCEPTED MANUSCRIPT

At enrollment, we collected demographics and clinical data at admission and prior to the start of NIV.

In addition, at randomization and at the end of each NIV session or break, we recorded Riker Sedation-

Agitation Scale (Riker-SAS 20) [19], Sp O2 , RR, HR, SBP, and tidal volu mes (VTs 21). Patients also rated

dyspnea using the Borg scale [20], and co mfo rt using a 10-cm visual analogue scale (VA S 22). They

also rated ease of eating on a scale fro m 0 (unable to eat) to 4 (no difficulty eating) and reported side

effects as dryness of nose and mouth, and eye irritation . Durations of NIV sessions and breaks were

recorded to the nearest minute. Data collection was stopped after 6 breaks, o r when the subject was

P T
weaned off NIV.

RI
Sample size and statistical analysis

SC
Assuming an average time on NIV o f 60 hours [21–24], a sample size of 70 patients (35 per arm)

would provide the study with 85% power to detect a 30% reduction in NIV hours in the HFNT g roup
NU
compared to SO at a two-sided 0.05 level of significance. The study was terminated early due to slow

enrollment related to increased routine use of HFNT during breaks, and consequent reluctance of
MA

clin icians to enroll, and loss of personnel for screening. Furthermore, had we eventually reached the

targeted 70 subjects (35 per arm), the estimated increas likelihood of finding a statistically significant
ED

difference in the major outcome variable would have been 7%.

PT

For the primary outcome and for between-arm secondary outcome co mparisons, generalized estimating

equations (GEE23) with repeated measures analysis when appropriate [25] were used to model the
CE

effect of treat ment, taking into account baseline differences (see Supplemental Material). The Chi-

square test was used to compare the frequency distribution of categorical variables between arms.
AC

Within arm analyses were performed using a paired t-test for normally d istributed variables or the

Wilcoxon signed-rank test for non-normally distributed data.

20 Riker-SA: Riker Sedation-Agitation Scale

21 VTs: Tidal volume
22 VAS: Visual analogue scale
23 GEE: Generalized estimating equations

8
 ACCEPTED MANUSCRIPT

All tests were t wo-sided and significance level was p≤0.05. We used IBM SPSS v21 (IBM Corp.,

Armonk, NY, USA) for all analyses. Data are reported as mean±SD o r as percentage; except for GEE

results, which are reported as marginal mean (SEM).

RESULTS

Patients

Fro m October 2013 through July 2015, a total of 624 patients admitted to the ICU or to the

T
Intermediate Care Unit were started on NIV due to ARF or respiratory acidosis; 127 patients were

P
elig ible fo r inclusion in the study, and 54 underwent randomization. After secondary exclusion of 7

RI
patients, 47 subjects were included in the analysis; 23 received HFNT and 24 SO (Figure 1).

SC
Baseline Characteristics
NU
At randomizat ion, vital signs and comorbidit ies were similar between the two arms except that subjects

in the HFNT arm had higher HR (99±11 vs 87±18, p<0.01), lower BMI (27.6±8.5 vs 33.7±9.7, p
MA

<0.05) and a lower proportion of OSA (13% vs 41.7%, p<0.05) than those in the control group. An

underlying diagnosis of pneumonia was mo re co mmon in the HFNT arm (60.9% vs 29.2%, p<0.05)

whereas other diagnoses were similarly distributed (Table 1). Twenty-two patients were enro lled due to
ED

having acute or acute on chronic hypercapnic respiratory failure, wh ile 25 due to having hypoxic

respiratory failure . Out of the 47 subjects, 43 had a P a O2 /FIO2 <300. There were no significant
PT

differences in baseline characteristics between the group of patients enrolled due to acute hypoxemic
CE

respiratory failure and those enrolled due to acute on chronic hypercapnic respiratory failure other than

the expected differences in gas exchange and pH (data not shown).

AC

Treatments

NIV settings were similar between the groups; across all NIV sessions mean EPAP and IPAP were

5.5±0.9 and 12.0±2.9 cmH2 O in the HFNT arm and 6.2±2.7 and 13.7±2.7 cmH2 O in the SO arm, and

mean VTs on NIV were 9.0±2.6 ml/ kg of pred icted body weight (PBW), and 8.1±2.0 ml/ kg PBW in the

HFNT and SO arms, respectively (all p=ns). During breaks, gas flow rate was 38.5±6.6 l/ min in the

HFNT g roup and 4.8±3.3 l/ min in the SO group and F I O2 was similar (39.4±17.5 vs 40.8±15.7%,

respectively).

9
 ACCEPTED MANUSCRIPT

Primary outcome

Total duration of NIV sessions per patient was 1315(225) and 1441(220) min and mean duration of

each NIV session was 491(67) and 455(63) min in the HFNT and SO g roups respectively (both p=0.7).

Total duration of breaks per patient was 1362(181) and 1196(177) min (p=0.5) and mean duration of

each break tended to be longer with HFNT (520(61) min ) than with SO (370(55) min) but did not reach

statistical significance (p=0.1) (Fig 2). Patient request was the most common reason for init iating a

T
break and caregiver decision for ending one (Tables S3 and S4).

P
No significant d ifferences were noted in t ime spent on and off NIV when considering hypercapnic and

RI
hypoxemic patients separately (data not shown).

SC
Secondary outcomes
NU
Clinical variables

During NIV, respiratory rate was higher in the HFNT group compared to the SO group (23.3±5.3 vs
MA

20.1±4.1, p =ns). During breaks, RR rose significantly on SO (21.8±5.2 vs 20.1±4.1, p<0.05) but not

HFNT co mpared to NIV (23.8±6.8 vs 23.3±5.3, p =ns) (Table 2, Figure 3A). Mean Sp O2 was similar

between the HFNT and SO groups during breaks and at the end of breaks, Sp O2 in both arms was lo wer
ED

than at the end of NIV sessions. GCS, Riker-SAS and SBP were similar between breaks and sessions in

both arms as well as between groups (Table 2). Conversely, HR was significantly higher in the HFNT
PT

arm both during NIV and HFNT, consistent with the higher baseline values.
CE

Subjective measurements
AC

Co mfort score during breaks was significantly better in the HFNT arm co mpared to SO (8.3±2.7 vs

6.9±2.3, p<0.05) and, within the HFNT arm, co mfort was greater on HFNT than on NIV (8.3± 2.7 vs

5.9±3.9, p<0.05) (Tab le 2, Figure 3C). Conversely, within the SO arm, there were no differences in

comfort between NIV sessions and breaks .

We observed no significant differences in dyspnea score between the HFNT and SO groups, but within

groups, dyspnea score rose during breaks on SO co mpared to NIV (2.4±2.2 vs 1.3±2.0, p=0.05) but not

in the HFNT group compared to NIV (2.1±2.8 vs 2.3±2.9, ns) (Table 2, Figure 3B).

10
 ACCEPTED MANUSCRIPT

Adverse effects and tolerance

Dryness of the nose and mouth and nasal discomfort were similar co mparing HFNT, SO and NIV

(48%, 47% and 65%). HFNT caused less eye irritation than SO (8% vs 21.6%, p=0.05). Significantly

more patients found it easier to eat on HFNT than on SO (13% vs 36%, p<0.05) (Table S5).

Two subjects underwent endotracheal intubation (ETI), both in the HFNT arm.

T
DISCUSSION

P
To the best of our knowledge, this is the first randomized controlled trial examining HFNT as a

RI
complementary therapy to NIV by co mparing it to SO during breaks off NIV in patients being treated

SC
for acute respiratory failure. The major outcome variab le, total t ime spent on NIV, was similar between

the HFNT and SO groups, as was the total time on breaks. However, HFNT d id tend to lengthen the
NU
duration of each break co mpared to SO, albeit not significantly. It also imp roved comfort co mpared to

both SO and NIV, and the rise in RR and worsening dyspnea scores observed during breaks on SO do
MA

not occur with HFNT. Finally, HFNT was well to lerated, causing less eye irritation than SO and NIV

and facilitating eating better than SO, without significant adverse side effects.
ED

Previous studies have shown that HFNT improves comfo rt, dyspnea and oxygenation when compared

to SO in patients with hypoxemic respiratory failu re, and follo wing extubation [14,26–28] and cardiac
PT

surgery [29]. More recently some [12], but not all [15], large randomized, controlled trials , HFNT
CE

compared to NIV has provided better comfort and alleviat ion of dyspnea. Our findings support and

extend this previous evidence showing that HFNT improves patients’ overall co mfort co mpared to NIV
AC

or SO when used during breaks fro m NIV. Previous trials have attributed the greater co mfort on

HFNT primarily to the delivery of heated, hu mid ified gas that avoids airway cooling and dessication

[30,31]. However, patients in our study received active in-line hu midification in both the HFNT and

NIV groups and those in the SO group used cold water hu mid ification, and yet co mfo rt was

significantly better on HFNT. In particular, in our study, a large proportion (50-65%) of subjects

reported upper airway dryness on HFNT, SO and NIV. These data are in keeping with previous

evidence on NIV demonstrating perceived airway dryness, even severe, in patients who received active

humid ification, associated with oral breathing[32,33]. It is conceivable that during HFNT, oral

11
 ACCEPTED MANUSCRIPT

breathing is a predisposing factor for dryness as well. The greater co mfo rt on HFNT despite similar

rating of subjective dryness suggests that other factors, such as interface comfort and efficiency of

oxygenation, as well as the dimin ished eye irritation and facilitated eating, might contribute to

perceived overall comfort.

In our study, dyspnea scores, respiratory rate and oxygen saturation were similar during breaks

between the HFNT and SO groups. However, when co mpared with NIV, RR and dyspnea increased

T
significantly during breaks in the SO group, whereas the HFNT g roup maintained a steady respiratory

P
rate and level of dyspnea (Figure 3). That o xygenation did not differ between the groups is probably

RI
explained by the independent adjustment of o xygen delivery to reach a target Sp O2 . The lower Sp O2 at

SC
the end of breaks in both arms compared to NIV may have been due to the use of reduced Sp O2 as a

criterion to end breaks and restart NIV.

NU
Interestingly, we d id not observe the reduction in RR observed by others when comparing HFNT to SO
MA

in patients with COPD or pulmonary fib rosis[34]. A previous study on healthy individuals [5] noted a

reduced RR and increased VT during wakefulness and diminished VT during sleep compared to

wakefulness. We did not analyze the effects of sleep, sleep-wake cycle or sedatives on ventilator
ED

responses during NIV sessions or use of HFNT o r SO during breaks, but this would be of interest for

future studies.
PT
CE

Also, our HFNT patients were begun on a lower flow rate (35 l/ min) than used in some mo re recent

studies [12,15] and, although our therapists were instructed to increase flow rate as tolerated, the
AC

average increase in flow rate was only 3.5 l/ min. Furthermore, based on the algorithm to use HFNT

during the study, flow-rate could be reduced to 20 l/ min, lo wer than the current defin ition of high-flow.

This could have conceivably led to lengthening of the time spent on HFNT by pro longing the weaning

fro m HFNT itself. Despite this, no differences in time on breaks between the arms were observed.

Considering that the effect of HFNT on breathing pattern appears to be at least partly related to flo w

rate [5], future studies should consider that a higher initial and minimal flow rate may be important in

achieving greater effects on breathing pattern and work of breathing and, thereby, desired outcomes.

12
 ACCEPTED MANUSCRIPT

Tidal volu mes during NIV in our study were high compared to the 6-8 ml/kg PBW range that is

recommended for lung protection in patients with hypoxemic respiratory failure; 9 and 8 ml/ kg PBW in

the HFNT and SO groups, respectively. Tidal volu mes in this range have been reported during use of

NIV prev iously[12] and they fall below the threshold (>9.5 ml/kg PBW) that has been associated with

worse outcomes during NIV [35]. Most importantly, though, the similarity between the two groups

suggests that VTs during NIV did not influence between group comparisons.

T
Strengths of our trial include the mult icenter design, pre-defined criteria to interrupt a break and re -start

P
NIV, and the randomized controlled design. Also, because we enrolled patients who had already been

RI
started on NIV, our study population, including both hypoxemic and hypercapnic patients, was likely

SC
to be generalizab le to patients treated with NIV in real life settings. Moreover, although blinding was

not possible, the research team was removed from the clinical decision -making process.
NU
Limitations include s mall nu mbers and early termination, which may have reduced the likelihood of
MA

detecting significant differences in secondary outcomes . Also, our sample size calculation assumed that

patients would receive NIV for 3600 (±1500) minutes, whereas actual time on NIV [1395 (±1042)

minutes] was shorter and statistical variability of the outcome was greater than anticipated, weakening
ED

our statistical power. In addition, wh ile we had pre-determined criteria to end breaks , these criteria

could be over-ridden by patient care considerations such as need for medication, patient requests, or
PT

relatives visiting, factors not easily modifiable with HFNT. Furthermore, the baseline differences in
CE

BMI and HR, and the higher proportion of patients with pneu monia in the HFNT g roup and OSA in the

SO group could have contributed to differing outcomes between the groups . In particu lar, these
AC

baseline differences may indicate a greater severity of illness in the HFNT group and therefore could

have predisposed to worse outcomes in that group. To assess the effect of baseline differences, the

GEE analysis was used, but allowed statistical compensation only for continuous variables and not for

categorical ones such as pneumonia or OSA (see Supplemental Material)..

In view of our results and limitations, we suggest that future studies could be helpful in extending the

assessment of the comp lementary ro le of HFNT during breaks off NIV. In particular, these could focus

on temporal outcomes (e.g. t ime to weaning off NIV) or on subjective variables, such as a relative

13
 ACCEPTED MANUSCRIPT

change in dyspnea and comfort, as well as the economic impact of using HFNT. Finally, in v iew of the

recent promising results of NIV delivered via helmet [36], it could be interesting to study HFNT as a

complementary therapy to helmet-ventilation, especially considering that this interface could

potentially be associated with specific side effects leading to requirement for breaks, such as neck and

axillary (from straps) discomfort or sores, difficulties in coping with noise and claustrophobia.

Summary and conclusion

T
Our study shows that, when used in a co mp lementary fashion with NIV, HFNT does not reduce the

P
time spent on NIV or lengthen time on break. Ho wever, HFNT p rovides greater co mfort and improves

RI
indices of tolerability such as eye irritation and ease of eating compared with SO, which has been the

SC
usual way of providing supplemental o xygenation during breaks from NIV in the past. In addition, we

show that the increase in respiratory rate and worsening of dyspnea seen with SO during breaks off
NU
NIV do not occur with HFNT. We conclude that HFNT is a well-tolerated alternative to standard

oxygen therapy during breaks off NIV.

MA

ACKNOWLEDGMENTS

Fisher & Paykel Healthcare LTD provided all the equip ment used in the study and partially funded the
ED

study through a grant received by Tu fts Medical Center. The sponsor reviewed the study design and

manuscript. The Investigators performed all analyses and made all final decision regarding all aspects
PT

of the study and of the manuscript.

CE

We thank all the patients, medical and nursing staff and respiratory therapists at Tufts Medical Center

and Winchester Hospital for their support and cooperation.

AC

Preliminary results of the study were p resented at the ERS International Congress 2015 and at the ATS

International Conference 2016.

The project described was supported by the National Center for Advancing Translational Sciences,

National Institutes of Health, A ward Nu mber UL1TR001064. The content is solely the responsibility

of the authors and does not necessarily represent the official views of the NIH.

CONTRIB UTION

GS – study design, data collection, statistical analysis, manuscript writing

14
 ACCEPTED MANUSCRIPT

CM, LP, MO, AK – data collection

LLP – study design, statistical and methodological revision, critical revision of manuscript

FB, SN – critical revision of manuscript

NSH – study design and analysis, manuscript writing and critical revision of manuscript

CONFLICT OF INTEREST

T
GS, CM, LP, MO, AK, LLP, FB and SN have no conflict of interest to disclose.

P
NSH: Consultant for Fisher & Paykel Healthcare, and Respironics . Received research grants fro m

RI
Fisher & Paykel Healthcare.

SC
REFERENCES

[1] Nava S, Hill N. Non-invasive ventilation in acute respiratory failure. Lancet (London, England)
NU
2009;374:250–9. doi:10.1016/S0140-6736(09)60496-7.

[2] Rochwerg B, Brochard L, Elliott MW, Hess D, Hill NS, Nava S, et al. Official ERS/ATS
MA

clinical practice guidelines: noninvasive ventilation for acute respiratory failure. Eur Respir J

2017;50:1602426. doi:10.1183/13993003.02426-2016.

[3] Sim MAB, Dean P, Kinsella J, Black R, Carter R, Hughes M. Performance of oxygen delivery
ED

devices when the breathing pattern of respiratory failure is simulated. Anaesthesia

2008;63:938– 40. doi:10.1111/ j.1365-2044.2008.05536.x.

PT

[4] Kumar H, Spence CJT, Tawhai MH. Modeling the pharyngeal pressure during adult nasal high
CE

flow therapy. Respir Physiol Neurobiol 2015;219:51–7. doi:10.1016/j.resp.2015.06.011.

[5] Mundel T, Feng S, Tatkov S, Schneider H, Mündel T, Feng S, et al. Mechanisms of nasal high
AC

flow on ventilation during wakefulness and sleep. J Appl Physiol 2013;114:1058–65.

doi:10.1152/ japplphysiol.01308.2012.

[6] Parke RL, McGuinness SP. Pressures delivered by nasal high flow oxygen during all phases of

the respiratory cycle. Respir Care 2013;58:1621– 4. doi:10.4187/ respcare.02358.

[7] Groves N, Tobin A. High flow nasal oxygen generates positive airway pressure in adult

volunteers. Aust Crit Care 2007;20:126–31. doi:10.1016/j.aucc.2007.08.001.

[8] Parke RL, Eccleston ML, McGuinness SP. The effects of flow on airway pressure during nasal

high-flow oxygen therapy. Respir Care 2011;56:1151–5. doi:10.4187/respcare.01106.

15
 ACCEPTED MANUSCRIPT

[9] Spoletini G, Alotaibi M, Blasi F, Hill NS. Heated Humidified High-Flow Nasal Oxygen in

Adults. Chest 2015;148:253–61. doi:10.1378/chest.14-2871.

[10] Mauri T, Turrini C, Eronia N, Grasselli G, Volta CA, Bellani G, et al. Physiologic Effects of

High-Flow Nasal Cannula in Acute Hypoxemic Respiratory Failure. Am J Respir Crit Care

Med 2017;195:1207–15. doi:10.1164/ rccm.201605-0916OC.

[11] Papazian L, Corley A, Hess D, Fraser JF, Frat J-P, Guitton C, et al. Use of high-flow nasal

cannula oxygenation in ICU adults: a narrative review. Intens ive Care Med 2016.

T
doi:10.1007/s00134-016-4277-8.

P
[12] Frat J-P, Thille AW, Mercat A, Girault C, Ragot S, Perbet S, et al. High-flow oxygen through

RI
nasal cannula in acute hypoxemic respiratory failure. N Engl J Med 2015;372:2185–96.

SC
doi:10.1056/NEJMoa1503326.

[13] Ansari BM, Hogan MP, Collier TJ, Baddeley RA, Scarci M, Coonar AS, et al. A Randomized
NU
Controlled Trial of High-Flow Nasal Oxygen (Optiflow) as Part of an Enhanced Recovery

Program After Lung Resection Surgery. Ann Thorac Surg 2015;101:459–64.

MA

doi:10.1016/ j.athoracsur.2015.07.025.

[14] Maggiore SM, Idone FA, Vaschetto R, Festa R, Cataldo A, Antonicelli F, et al. Nasal high -

flow versus Venturi mask oxygen therapy after extubation. Effects on oxygenation, comfort,
ED

and clinical outcome. Am J Respir Crit Care Med 2014;190:282– 8. doi:10.1164/ rccm.201402-

0364OC.
PT

[15] Stéphan F, Barrucand B, Petit P, Rézaiguia-Delclau x S, Médard A, Delannoy B, et al. High-

CE

Flow Nasal Oxygen vs Noninvasive Positive Airway Pressure in Hypoxemic Patients After

Cardiothoracic Surgery: A Randomized Clinical Trial. Jama 2015;313:2331– 9.

AC

doi:10.1001/ jama.2015.5213.

[16] Epstein AS, Hartridge-Lambert SK, Ramaker JS, Voigt LP, Portlock CS. Humidified high -flow

nasal oxygen utilization in patients with cancer at Memorial Sloan -Kettering Cancer Center. J

Palliat Med 2011;14:835–9. doi:10.1089/jp m.2011.0005.

[17] Lacroix G, Legodec J, D’Aranda E, Esnault P, Romanat P-E, Goutorbe P. Is non-invasive

ventilation the best ventilatory support for “do not intubate” patients? Crit Care 2012;16:442.

doi:10.1186/cc11435.

[18] British Thoracic Society Standards of Care Committee. Non -invasive ventilation in acute

16
 ACCEPTED MANUSCRIPT

respiratory failure. Thorax 2002;57:192–211.

[19] Riker RR, Picard JT, Fraser GL. Prospective evaluation of the Sedation -Agitation Scale for

adult critically ill patients. Crit Care Med 1999;27:1325–9.

[20] Borg G. Perceived exertion as an indicator of somatic stress. Scand J Rehabil Med 1970;2:92–

8.

[21] Thys F, Roeseler J, Reynaert M, Liistro G, Rodenstein DO. Noninvasive ventilation for acute

respiratory failure: a prospective randomised placebo-controlled trial. Eur Respir J

T
2002;20:545– 55.

P
[22] Jurjević M, Matić I, Sakić-Zdravcević K, Sakić S, Danić D, Buković D. Mechanical ventilation

RI
in chronic obstructive pulmonary disease patients, noninvasive vs. invasive method

SC
(randomized prospective study). Coll Antropol 2009;33:791–7.

[23] Alsous F, Amoateng-Adjepong Y, Manthous CA. Noninvasive ventilation: experience at a

NU
community teaching hospital. Intensive Care Med 1999;25:458–63.

[24] Celikel T, Sungur M, Ceyhan B, Karakurt S. Comparison of noninvasive positive pressure

MA

ventilation with standard medical therapy in hypercapnic acute respiratory failure. Chest

1998;114:1636–42.

[25] Hardin J, Hilbe J. Generalized estimating equations. London: Chapman and Hall/CRC; 2003.
ED

[26] Hernández G, Vaquero C, González P, Subira C, Frutos -Vivar F, Rialp G, et al. Effect of

Postextubation High-Flow Nasal Cannula vs Conventional Oxygen Therapy on Reintubation in

PT

Low-Risk Patients. JAMA 2016;315:1354–61. doi:10.1001/jama.2016.2711.

CE

[27] Rittayamai N, Tscheikuna J, Rujiwit P. High-Flow Nasal Cannula Versus Conventional

Oxygen Therapy After Endotracheal Extubation: A Randomized Crossover Physiologic Study.

AC

Respir Care 2014;59:485–90. doi:10.4187/respcare.02397.

[28] Tiruvoipati R, Lewis D, Haji K, Botha J. High-flow nasal oxygen vs high-flow face mask: a

randomized crossover trial in extubated patients. J Crit Care 2010;25:463– 8.

doi:10.1016/ j.jcrc.2009.06.050.

[29] Corley A, Caruana LR, Barnett AG, Tronstad O, Fraser JF. Oxygen delivery through high-flow

nasal cannulae increase end-expiratory lung volume and reduce respiratory rate in post-cardiac

surgical patients. Br J Anaesth 2011;107:998–1004. doi:10.1093/bja/aer265.

[30] Chanques G, Constantin J-M, Sauter M, Jung B, Sebbane M, Verzilli D, et al. Discomfort

17
 ACCEPTED MANUSCRIPT

associated with underhumidified high-flow oxygen therapy in critically ill patients. Intensive

Care Med 2009;35:996–1003. doi:10.1007/s00134-009-1456-x.

[31] Cuquemelle E, Pham T, Papon J-F, Louis B, Danin P-E, Brochard L. Heated and humidified

high-flow oxygen therapy reduces discomfort during hypoxemic respiratory failure. Respir

Care 2012;57:1571–7. doi:10.4187/respcare.01681.

[32] Oto J, Imanaka H, Nishimura M. Clinical factors affecting inspired gas humidification and oral

dryness during noninvasive ventilation. J Crit Care 2011;26:535.e9 -535.e15.

T
doi:10.1016/ j.jcrc.2010.10.005.

P
[33] Oto J, Nakataki E, Okuda N, Onodera M, Imanaka H, Nishimura M. Hygrometric properties of

RI
inspired gas and oral drynes s in patients with acute respiratory failure during noninvasive

SC
ventilation. Respir Care 2014;59:39– 45. doi:10.4187/respcare.02351.

[34] Braunlich J, Beyer D, Mai D, Hammerschmidt S, Seyfarth H-J, Wirtz H, et al. Effects of Nasal
NU
High Flow in Ventilation in Volunteers, COPD and Idiopathic Pulmonary Fibrosis Patients.

Respiration 2012;85:319– 25. doi:10.1159/000342027.

MA

[35] Carteaux G, Millán-Guilarte T, De Prost N, Razazi K, Abid S, Thille AW, et al. Failure of

Noninvasive Ventilation for De Novo Acute Hypoxemic Respiratory Failure. Crit Care Med

2016;44:282– 90. doi:10.1097/ CCM.0000000000001379.

ED

[36] Patel BK, Wolfe KS, Pohlman AS, Hall JB, Kress JP. Effect of Noninvasive Ventilation

Delivered by Helmet vs Face Mask on the Rate of Endotracheal Intubation in Patients With
PT

Acute Respiratory Distress Syndrome. JAMA 2016;315:2435. doi:10.1001/jama.2016.6338.

CE

[2]
AC

18
 ACCEPTED MANUSCRIPT

Figure legends

Figure 1 Patient flowchart: screening, enrollment and randomization.

Figure 2 Total ti me on and off NIV and Mean duration of each NIV session and break. Total time

T
on NIV (A) and on break off NIV (C), shown as mean (circle) ± 2 SE (whiskers) co mparing the

P
high-flo w nasal therapy (HFNT) arm (in dark red) and the standard oxygen therapy (SO) group

RI
(in dark b lue). Mean duration of NIV sessions (B) and breaks off NIV (D), also shown as mean

SC
± 2 SE in the two groups.
NU
Figure 3 Relati ve change in respiratory rate (A), dyspnea (B) and comfort (C) co mparing NIV

sessions and breaks during HFNT (dark red) and SO (dark b lue). An increase in both
MA

respiratory rate (Panel A) and dyspnea (Panel B) was observed during breaks on SO co mpared

to NIV sessions, whereas RR and dyspnea were stable on HFNT co mpared to NIV. Co mfort

score (Panel C) was significantly better on HFNT co mpared to both NIV sessions and breaks
ED

on SO.Asterisks over lines indicate p ≤ 0.05 co mpared to NIV and crosses with brackets

indicate p ≤ 0.05 comparing HFNT and SO during breaks.

PT
CE
AC

19
 ACCEPTED MANUSCRIPT

Table 1 Patients’ characteristics at baseline

HFNT arm SO arm (n=24) p
(n=23)
Age, yr 67.9±13.8 63.4±14.0 ns
Male sex, n (% ) 8 (34.8) 10 (41.7) ns
Smoking history
Current, n (%) 6 (26.1) 11 (47.8) ns
Past, n (%) 12 (52.2) 9 (39.1) ns
Body mass index 27.6±8.5 33.7±9.7 <0.05
APACHE II 15.9±5.2 16.5±8.4 ns

T
ABG
pH 7.31±0.15 7.31±0.10 ns

P
PaCO2, mmHg 53.4±22.4 61.6±20.2 ns
PaO2, mmHg 93.3±45.8 98.8±72.7 ns

RI
FiO2, % 52.2±26.2 45.6±22.9 ns
PaO2/FiO2, mmHg 200.7±89.4 212.4±74.3 ns

SC
SpO2, % 96.5±2.7 95.4±3.3 ns
Respiratory rate, breaths/min 27.2±5.6 25.6±7.3 ns
Heart rate, beats/min 99±11 87±18 <0.01
Mean arterial pressure, mmHg 87±24 92±18 ns
NU
Temperature, C 36.7±0.5 36.7±0.5 ns
Home respiratory therapy
LTOT, n (%)
Nocturnal O2, n (%) 9 (39.1) 7 (29.2) ns
MA

Ambulatory O2, n (%) 2 (8.7) 1 (4.2) ns

CPAP, n (%) 1 (4.3) 1 (4.2) ns
NIV, n (%) 2 (8.7) 4 (16.7) ns
2 (8.7) 5 (20.8) ns
Comorbi dities
ED

COPD, n (%) 13 (56.5) 16 (66.7) ns

CHF, n (%) 5 (21.7) 11 (45.8) ns
Hypertension, n (%) 15 (65.2) 16 (66.7) ns
Diabetes, n(%) 6 (26.1) 11 (45.8) ns
PT

CKD, n (%) 2 (8.7) 5 (20.8) ns

Liver disease, n (%) 1 (4.3) 0 (0) ns
Neoplasm, n (%) 7 (30.4) 5 (20.8) ns
CE

Cerebrovascular disease, n (%) 2 (8.7) 1 (4.2) ns

OSA, n(%) 3 (13.0) 10 (41.7) <0.05
Diagnosis
COPD exacerbation, n (%) 5 (21.7) 11 (45.8) ns
AC

ACPE, n (%) 7 (30.4) 5 (20.8) ns

Pneumonia, n (%) 14 (60.9) 7 (29.2) <0.05
ARDS, n (%) 4 (17.4) 2 (8.3) ns
OHS, n (%) 2 (8.7) 0 (0) ns
Dyspnoea score 2.1±2.7 2.5±2.7 ns
Comfort score 6.2±3.3 6.0±3.3 ns

Data are mean ± SD or number (%). Yr: years; LTOT: long term oxygen therapy; CPAP: continuous positive
airway pressure; CHF: congestive heart failure; CKD: chronic kidney disease; O SA: obstructive sleep apnea;
ARDS: acute respiratory distress syndrome; OHS: obesity hypoventilation syndrome.

20
 ACCEPTED MANUSCRIPT

Table 2 Clinical and subjective variables during breaks and NIV sessions

HFNT arm (n=23) SO arm (n=24)

Break NIV Break NIV
GCS 14.9±0.4 14.9±0.4 14.9±0.6 14.7±1.2
Riker-SAS 3.9±0.2 3.9±0.2 3.9±0.2 3.9±0.1
Heart rate, beats/min 98.6±15.7* 98.6±15.6 † 82.1±16.1* 81.8±17.8 †
MBP_mmHg 87.0±15.0 84.5±15.5 87.4±10.8 88.1±11.6
Respiratory rate, breaths/min 23.8±6.8 23.3±5.3† 21.8±5.2‡ 20.1±4.1†‡
SpO2, %

T
Mean 94.1±3.2 - 94.7±2.1 -
Final 93.7±5.0‡ 95.9±2.9‡ 95.4±2.7‡ 96.4±2.0‡

P
Comfort Score, VAS 8.3±2.7*‡ 5.9* 3.9‡ 6.9±2.3* 6.2±2.9
2.4±2.2‡ 1.3±2.0‡

RI
Dyspnea Score, Borg 2.1±2.8 2.3±2.9

SC
*
Data are mean ± SD. M ean are presented before GEE compensations are applied. indicates a difference with
†
p≤0.05 comparing breaks on HFNT and SO. indicates a difference with p≤0.05 comparing the NIV sessions
‡
between the two arms. indicates a difference with p≤0.05 comparing NIV sessions with breaks within each arm.
NU
GCS: Glas gow coma scale; Riker-SA S: Riker sedation-agitations scale; M BP: mean blood pressure; VAS: Visual
analogue scale.
MA
ED
PT
CE
AC

21
 ACCEPTED MANUSCRIPT

Table 3 Adverse events during breaks and NIV sessions

HFNT SO NIV
Dryness of nose and mouth 48% 47.1% 65%
Eye irritation 8%* 21.6%* 19.1%
Nasal discomfort 2% 8% 9.6%
Difficulty in eating † † -
13.3% 36.2%
*
indicates a difference with p≤0.05 comparing breaks on HFNT and NIV.

T
†
indicates a difference with p≤0.05 when comparing HFNT and SO .

P
RI
SC
NU
MA
ED
PT
CE
AC

22
 ACCEPTED MANUSCRIPT

TP
RI
SC
NU
MA
ED
PT
CE
AC

23
 ACCEPTED MANUSCRIPT

TP
RI
SC
NU
MA
ED
PT
CE
AC

24
 ACCEPTED MANUSCRIPT

TP
RI
SC
NU
MA
ED
PT
CE
AC

25
 ACCEPTED MANUSCRIPT

High-flow nasal therapy vs standard oxygen during breaks off noninvasive ventilation for

acute respiratory failure: a pilot randomized controlled trial

Giulia Spoletini MD, Chiara Mega MD, Lara Pisani MD PhD, Mona Alotaibi MD, Alia Khoja MD, Lori

Lyn Price MAS, Francesco Blasi MD PhD FERS, Stefano Nava MD PhD, Nicholas S Hill, MD

T
Highlights:

P
 Pilot randomized trial assessing the role of HFNT as complementary therapy to NIV

RI
 HFNT does not affect the time on and off NIV compared to SO

SC
 HFNT is well tolerated and leads to greater comfort than SO

 HFNT prevents increase in RR and dyspnea observed with SO during breaks off NIV
NU
 HFNT could be a suitable alternative to SO during breaks off NIV

hxgSHXH
MA
ED
PT
CE
AC

26
Figure 1
Figure 2
Figure 3