Cortegiani et al.

Crit Care (2020) 24:692

https://doi.org/10.1186/s13054-020-03409-0

RESEARCH Open Access

High flow nasal therapy versus noninvasive

ventilation as initial ventilatory strategy in COPD
exacerbation: a multicenter non‑inferiority
randomized trial
Andrea Cortegiani1,2* , Federico Longhini3, Fabiana Madotto4, Paolo Groff5, Raffaele Scala6, Claudia Crimi7,
Annalisa Carlucci8, Andrea Bruni3, Eugenio Garofalo3, Santi Maurizio Raineri1,2, Roberto Tonelli9,
Vittoria Comellini10, Enrico Lupia11, Luigi Vetrugno12, Enrico Clini13, Antonino Giarratano1,2, Stefano Nava10,
Paolo Navalesi14, Cesare Gregoretti1,15 and the H. F.-AECOPD study investigators

Abstract
Background: The efficacy and safety of high flow nasal therapy (HFNT) in patients with acute hypercapnic exacerba-
tion of chronic obstructive pulmonary disease (AECOPD) are unclear. Our aim was to evaluate the short-term effect of
HFNT versus NIV in patients with mild-to-moderate AECOPD, with the hypothesis that HFNT is non-inferior to NIV on
­CO2 clearance after 2 h of treatment.
Methods: We performed a multicenter, non-inferiority randomized trial comparing HFNT and noninvasive ventilation
(NIV) in nine centers in Italy. Patients were eligible if presented with mild-to-moderate AECOPD (arterial pH 7.25–7.35,
­PaCO2 ≥ 55 mmHg before ventilator support). Primary endpoint was the mean difference of P ­ aCO2 from baseline to
2 h (non-inferiority margin 10 mmHg) in the per-protocol analysis. Main secondary endpoints were non-inferiority
of HFNT to NIV in reducing ­PaCO2 at 6 h in the per-protocol and intention-to-treat analysis and rate of treatment
changes.
Results: Seventy-nine patients were analyzed (80 patients randomized). Mean differences for P ­ aCO2 reduction from
baseline to 2 h were − 6.8 mmHg (± 8.7) in the HFNT and − 9.5 mmHg (± 8.5) in the NIV group (p = 0.404). By 6 h,
32% of patients (13 out of 40) in the HFNT group switched to NIV and one to invasive ventilation. HFNT was statisti-
cally non-inferior to NIV since the 95% confidence interval (CI) upper boundary of absolute difference in mean P ­ aCO2
reduction did not reach the non-inferiority margin of 10 mmHg (absolute difference 2.7 mmHg; 1-sided 95% CI 6.1;
p = 0.0003). Both treatments had a significant effect on ­PaCO2 reductions over time, and trends were similar between
groups. Similar results were found in both per-protocol at 6 h and intention-to-treat analysis.
Conclusions: HFNT was statistically non-inferior to NIV as initial ventilatory support in decreasing P ­ aCO2 after 2 h of
treatment in patients with mild-to-moderate AECOPD, considering a non-inferiority margin of 10 mmHg. However,
32% of patients receiving HFNT required NIV by 6 h. Further trials with superiority design should evaluate efficacy
toward stronger patient-related outcomes and safety of HFNT in AECOPD.

*Correspondence: [email protected]
1
Department of Surgical, Oncological and Oral Science (Di.Chir.On.S.),
University of Palermo, Palermo, Italy
Full list of author information is available at the end of the article

© The Author(s) 2020. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat​iveco​
mmons​.org/publi​cdoma​in/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Cortegiani et al. Crit Care (2020) 24:692 Page 2 of 13

Trial registration: The study was prospectively registered on December 12, 2017, in ClinicalTrials.gov (NCT03370666).
Keywords: High flow nasal therapy, High flow nasal cannula, Noninvasive ventilation, Chronic obstructive pulmonary
disease, Acute respiratory failure

Introduction Care Units (ICU), or Respiratory Unit of 9 centers in

Chronic obstructive pulmonary disease (COPD) patients Italy. The study protocol was approved by “Comitato
may require respiratory support and hospitalization due Etico Sezione Area Centro” Ethics Committee (approval
to an acute exacerbation of their disease (AECOPD) [1, no. 245—October 24, 2017, Catanzaro, Italy) and by the
2]. To date, noninvasive ventilation (NIV) represents the local ethics committee of all the study centers. This study
cornerstone treatment for the management of patients was conducted in accordance with the amended Declara-
with AECOPD with associated respiratory acidosis [3]. tion of Helsinki. Written informed consent was obtained
However, several factors may determine NIV failure like from all patients or legal representatives. The study has
discomfort related to the interface, patient–ventilator been prospectively registered in ClinicalTrials.gov in
interaction, airway secretions, the severity of the disease, December 2017 (Identifier: NCT03370666), and the
and the skill of the team of caregivers [4–7]. study protocol has been published [24].
High flow nasal therapy (HFNT) [8] has been shown to We considered eligible adults (i.e., > 18 years/old)
provide potential beneficial effects for patients with sta- patients with a diagnosis of COPD exacerbation accord-
ble COPD: It creates a distending pressure generating a ing to GOLD criteria [25] admitted for a mild-to-moder-
positive end-expiratory pressure (PEEP) effect that may ate acute hypercapnic respiratory failure, with an arterial
counterbalance intrinsic PEEP [9], a washout of naso- pH between 7.25 and 7.35 and a P ­ aCO2 ≥ 55 mmHg.
pharyngeal dead space optimizing ventilatory efficiency Exclusion criteria were: (1) received HFNT or NIV before
and facilitating carbon dioxide removal [10], a reduced the study enrolment; (2) long-term domiciliary NIV;
inspiratory resistance providing adequate flow and warm (3) clinical cardiovascular instability, as defined by the
gases preventing bronchoconstriction response to dry air need for vasopressors, acute coronary syndrome or life-
[11, 12] enhancing lung mucociliary clearance [13, 14], threatening arrhythmias [26]; (4) treatment refusal; 5)
and finally decreasing diaphragmatic effort in a similar agitation, characterized by a Richmond Agitation Seda-
way to NIV [15]. Recent data also showed that HFNT tion Scale (RASS) ≥ 2 or lack of collaboration, defined by
might have a role in managing patients with AECOPD a Kelly Matthay score ≥ 5 [26]; (6) acute failure of more
[16]. It has been shown to reduce arterial partial pres- than two organs [26]; (7) cardiac arrest; (8) respiratory
sure of carbon dioxide (­PaCO2) [17, 18] and to improve arrest deeming immediate intubation; (9) recent facial or
inspiratory effort when used in the NIV interval periods neck trauma, burns or skin breakdown; (10) pregnancy;
as compared to conventional oxygen therapy [19, 20]. (11) consent withdrawal; and (12) enrolment in other
In non-randomized trials, HFNT showed to be equiva- research protocols.
lent to NIV in avoiding intubation in mild-to-moderate
AECOPD patients with respiratory acidosis with similar Interventions

flow and MR850 or AIRVO™ 2, Fisher & Paykel Health-

failure rates, but with better comfort and fewer complica- In the intervention group, patients received HFNT (Opti-
tions for the patients using HFNT [21, 22]. Thus, there is
a reasonable physiological and clinical rationale for using care, Auckland, New Zealand), initially set at 60 L/min, at
HFNT in AECOPD, but its efficacy and safety are unclear a temperature of 37 °C. In case of discomfort, flow and/or
[23]. temperature were down-regulated to the most tolerated
Our aim was to evaluate the short-term effect of HFNT setting.
versus NIV in patients with mild-to-moderate AECOPD, In the control group, patients received NIV through
with the hypothesis that HFNT is non-inferior to NIV on a total full-face or oro-nasal mask. The ventilator was
­CO2 clearance after 2 h of treatment. set in Pressure Support Ventilation (PSV) mode, with a
PEEP titrated between 3 and 5 cm ­H2O. The inspiratory
Methods pressure was titrated to achieve a measured or estimated
Study design and patients expiratory tidal volume equal to 6–8 mL k­ g−1 of ideal
This was an investigator-initiated randomized, unblinded, body weight [26]. The attending physician, based on local
multicenter, non-inferiority, controlled trial conducted availability, selected the ventilators used to deliver NIV.
from February 15, 2018, to March 25, 2020. The study In both groups, therapeutic management other than
was conducted in the Emergency Department, Intensive ventilatory support was according to current guidelines
Cortegiani et al. Crit Care (2020) 24:692 Page 3 of 13

[25]. Sedation was allowed to improve patients’ comfort compromised the pursuit of the treatment (i.e., subject
and tolerance of the interfaces (target RASS between refusal). The decision to intubate was based on the phy-
0 and -2). The inspired oxygen fraction (­FiO2) was set sician’s clinical judgment and on the presence of at least
to maintain a peripheral oxygen saturation ­(SpO2) tar- one of the following criteria: respiratory arrest, respira-
get between 88–92% [25]. During study interventions, tory apnea or pauses with loss of consciousness, severe
patients were monitored with continuous S ­ pO2, electro- agitation, bradycardia (< 50 beats/min) with loss of con-
cardiogram, and noninvasive blood pressure. sciousness, hemodynamic instability with systolic arterial
pressure < 70 mmHg, need for invasive mechanical ven-
Study endpoints tilation due to worsening in arterial blood gases and pH
The primary endpoint was the mean difference of P ­ aCO2 decline or pH < 7.25, management of abundant respira-
to evaluate the non-inferiority of HFNT to NIV from tory secretions, intolerance to all the interfaces (includ-
baseline to 2 h after the randomization. The secondary ing the eventual change of the treatment). Data were
endpoints were: (1) non-inferiority of HFNT to NIV in collected on a dedicated case report form.
reducing ­PaCO2 at 6 h after randomization; (2) treatment
change rates (switch to the other study intervention, to Randomization and statistical analysis
IMV, to no support or no change); (3) dyspnea score and We computed a sample size of 56 patients, given an alpha
proportion of patients who did not improve the dyspnea error of 5% (one-sided) and a power of 80%, with a stand-
score; (4) discomfort score and proportion of patients ard deviation for the primary outcome equal to 15 mmHg
showing poor tolerance to treatment; (5) the propor- and a non-inferiority limit of 10 mmHg. Thus, the non-
tion of patients who had ­PaCO2 worsening or reduc- inferiority would be demonstrated if the upper bound-
tion < 10 mmHg from baseline assessment, or worsening ary of the 95% CI for the mean difference was lower than
or no improvement of the dyspnea; (6) respiratory rate; 10 mmHg. A non-inferiority margin of 10 mmHg was
(7) change in arterial blood gases; 8) time spent under set as a clinically relevant difference in change of ­PaCO2
mechanical ventilation (both IMV and NIV); (9) hospital between the groups by consensus among the investiga-
length of stay; and (10) hospital mortality. tors, based on clinical judgment and available data at
the time of trial design [17, 26, 32, 33]. After consider-
Data collection and outcome assessment ing potential dropouts (30%) and an increase in sample
We collected anthropometric and clinical baseline char- size for nonparametric analysis (15%), the final computed
acteristics, i.e., the Simplified Acute Physiology Score sample size was 80 patients (40 per group). No imputa-
(SAPS II), the Kelly-Matthay Score [27], the Charlson tion was planned for missing data.
index [28], and the Richmond Agitation Sedation Scale Randomization was achieved using a computer-gen-
(RASS) [29]. Furthermore, soon before the randomiza- erated randomization sequence, generated by an inde-
tion, we collected the vital parameters, the presence of pendent investigator, not otherwise involved in the trial,
dyspnea (Borg scale) [30], and the arterial blood gases with an allocation ratio of 1:1 and with permuted block
at patients’ inclusion. All these relevant variables and method. A single randomization list for all participants
endpoints were evaluated at 2 and 6 h after randomiza- was created. Allocation concealment was maintained
tion. In particular, we recorded: the NIV and HFNT set- using sequentially numbered sealed opaque envelopes.
tings; the discomfort related to the interface, as assessed Each envelope contained the patient’s allocation to either
through a ten-point numeric rating scale [31], the pro- control (NIV) or intervention (HFNT), with a unique
portion of patients reporting poor tolerance to the treat- patient identifier code. The randomization was based
ment (defined as a patient-reported complaint to the on a centralized phone call system. Due to the research
assigned treatment that did not cause treatment inter- design, neither the individual collecting data nor the
ruption) due to flow, temperature, noise, claustrophobia, patient can be blinded to treatment allocation. The base-
sweating, tightness, airway dryness, vomiting gastric dis- line was defined as the time of randomization.
tension, ocular irritation, or skin breakdown [31]. An independent statistician (FM), blinded to treat-
The decision to change the assigned treatment was ment allocation, performed all statistical analyses on per-
left up to the clinical judgment of the attending physi- protocol and intention-to-treat bases as recommended
cian and was motivated (in unfavorable cases) by any [34]. The per-protocol analysis included only patients
of the following: failure to improve or worsening of who received the randomly assigned intervention till
clinical signs of respiratory failure or gas exchange (i.e., the 2-h (per-protocol analysis at 2 h) and the 6-h assess-
­PaCO2 > 20% baseline and/or pH < 7.25) in the case of ment (per-protocol analysis at 6 h), excluding those who
change from HFNT to NIV; intolerance to the assigned changed the treatment. The intention-to-treat analysis
intervention defined as a patient-reported complaint that included all patients according to the randomization,
Cortegiani et al. Crit Care (2020) 24:692 Page 4 of 13

whether they changed the intervention or not at the switched to NIV (n = 5 worsening/no improvement of
study timepoints [34]. respiratory failure; n = 1 intolerance to the intervention)
After checking the skewness of distribution, continu- while one patient switched to HFNT (intolerance to the
ous data will be presented as mean (standard devia- intervention) and one to IMV in the NIV group (wors-
tion) or median [25th–75th percentile]. Categorical data ening of respiratory failure). By 6 h, seven patients had
were expressed as counts and percentages. Differences switched to NIV (worsening/no improvement of res-
between treatments in continuous variables were evalu- piratory failure), one to IMV (worsening of respiratory
ated by the Mann–Whitney U-test or the Student t test failure) in the HFNT group. Due to the improvement of
according to Normal distribution. Categorical data were respiratory failure, one patient switched to no support in
compared with the Chi-square test or Fisher exact test. the HFNT group while two patients switched to HFNT
Paired sample t test or paired samples Wilcoxon test were and six to no support in the NIV group.
used to assess the difference between timepoints in con- Seventy-one patients (HFNT n = 34 [85%] vs. NIV
tinuous variables for each group. Change over time was n = 37 [95%]; p = 0.2633) and 53 patients (HFNT n = 24
modeled using mixed-effects linear regression with a ran- [60%] vs. NIV n = 29 [74.4%]; p = 0.1745) continued
dom intercept and slope (time) in order to account for the allocated interventions after 2 and 6 h, respectively
non-independence among measures. ­PaCO2 values were (Fig. 1, Table 2). Characteristics of interventions are
regressed on time (baseline, 2 h, 6 h), group treatment reported in Additional file 2: Table S1. During the 6 h
(HFNT, NIV), and the time per group interaction. of treatment, three patients in the HFNT (7.5%) and in
For the primary outcome, a one-sided two-sample the NIV (7.7%) received sedatives according to the study
Student t test was performed while taking into account protocol.
the non-inferiority margin (10 mmHg) to test whether
­PaCO2 reduction with HFNT was non-inferior to NIV. Per‑protocol analysis
The same approach was used to evaluate the non-inferi- At baseline, clinical characteristics and blood gases
ority of HFNT compare to NIV in the P ­ aCO2 decrease were similar between the two groups in patients who
after 6 h. completed the treatment initially allocated after 2 h
A p value < 0.05 will be considered significant. Statisti- (see Additional file 2: Table S2). Mean differences for
cal analyses were performed with SAS software, version ­PaCO2 from baseline to 2 h were − 6.8 mmHg (± 8.7)
9.4 (SAS Institute Inc., Cary, NC, USA) and R, version in the HFNT and − 9.5 mmHg (± 8.5) in the NIV group
3.5.2 (The R Foundation for Statistical Computing) was (p = 0.404) (Table 3). Figure 2a, b shows the differences
used for data visualization. Data monitoring was done and trends in P­ aCO2. Both treatments were able to sig-
by two investigators (LB, PC) querying the final database nificantly lower the P ­ aCO2 over the study timepoints
after collecting CRFs from enrolling centers. (Fig. 2a). As regards to the primary outcome, absolute
­PaCO2 difference was 2.7 mmHg (1-sided 95% CI − ∞;
Results 6.1) and HFNT was non-inferior to NIV since the upper
CONSORT checklist for this non-inferiority trial is boundary of 95% CI did not reach the non-inferiority
reported in the Additional file 1. margin of 10 mmHg (p = 0.0003, Fig. 3). Both treatments
had a significant effect on P ­ aCO2 reduction over time
Patients and interventions (time effect, p < 0.0001), and trends were similar between
From 235 eligible patients, we randomized 80 patients, groups (interaction term, p = 0.5864), (Fig. 2b).
40 in both groups (Fig. 1). The most frequent reason for Similar results were found in the per-protocol analysis
exclusion that we registered was the prior use of NIV at 6 h (see Fig. 2c, d; Table 3, Additional file 2: Table S3
or HFNT before enrollment (53 out of 155 patients and Figure S1). However, in this analysis, there was a
excluded). One patient withdrew consent after randomi- lower baseline SAPS II score in the HFNT group.
zation; therefore, data from 79 patients was analyzed.
Baseline characteristics and blood gases of patients Intention‑to‑treat analysis and secondary outcomes
were evenly distributed between the two groups In Additional file 2: Figure S2 (panel A and B) shows the
(Table 1). Mean age was 74 (± 13) and 77 (± 12), SAPS differences and trends in ­PaCO2 for this analysis. Both
II was 30 (± 9.0), and 33 (± 10), the mean arterial pH interventions significantly lower P ­ aCO2 over the study
was 7.30 (± 0.03) and 7.29 (± 0.03), and the mean ­PaCO2 timepoints in both groups. Criteria for the non-inferior-
was 73.7 mmHg (± 12.8) and 72.0 mmHg (± 13.0) in the ity of HFNT versus NIV were also met in this analysis at
HFNT and NIV group, respectively (Table 1). 2 and 6 h (Fig. 3, Additional file 2: Figure S1)
All patients received the assigned treatment after ran- The other secondary outcomes, including dyspnea
domization. By 2 h, six patients in the HFNT group had score, discomfort, length of mechanical ventilation,
Cortegiani et al. Crit Care (2020) 24:692 Page 5 of 13

Fig. 1 Flow diagram of the trial according to CONSORT

Cortegiani et al. Crit Care (2020) 24:692 Page 6 of 13

Table 1 Patients’ characteristics in the high flow nasal therapy (HFNT) and noninvasive ventilation (NIV) groups
at baseline
HFNT group NIV group p value

N 40 39
Females, n (%) 19 (47.5) 20 (51.3) 0.7368
Age (years), mean ± SD 74 ± 13 77. ± 12 0.2273
Weight (kg)a, mean ± SD 85 ± 23 76 ± 13 0.0964
Height (m)a, mean ± SD 1.7 ± 0.1 1.7 ± 0.1 0.5494
BMI (kg/m2)a, mean ± SD 30.5 ± 8.7 26.7 ± 5.5 0.0622
Ward of admission, n (%) 0.5807
Emergency room 24 (60) 21 (53.8)
ICU or respiratory unit 16 (40) 18 (46.1)
SAPS II, mean ± SD 30 ± 9 33 ± 10 0.1497
Charlson index, mean ± SD 4±2 5±3 0.4709
Systolic blood pressure (mmHg), mean ± SD 131 ± 27 137 ± 24 0.3629
Diastolic blood pressure (mmHg), mean ± SD 71 ± 17 70 ± 13 0.7089
Heart rate (per min), mean ± SD 91 ± 20 92 ± 19 0.9609
Respiratory rate (per min), mean ± SD 27 ± 7 28 ± 7 0.5544
Body ­temperatureb (C°), mean ± SD 36.6 ± 0.7 36.8 ± 0.5 0.0489
Kelly Matthay score, n (%) 0.0694
Alert, follows complex command (1) 19 (47.5) 25 (64.1)
Alert, follows simple commands (2) 9 (22.5) 8 (20.5)
Lethargie (3) 12 (30) 4 (10.3)
Stuporous (4) 0 (0) 2 (5.1)
Borg dyspnea ­scoreb, mean ± SD 5±2 5±2 0.4463
RASS, n (%) 0.4813
Light sedation (− 2) 2 (5) 2 (5.1)
Drowsy (− 1) 13 (32.5) 7 (17.9)
Alert and calm (0) 21 (52.5) 26 (66.7)
Restless (+ 1) 4 (10) 4 (10.3)
Secretion, n (%) 0.2163
Normal 23 (57.5) 17 (43.6)
Abundant 17 (42.5) 22 (56.4)
PaCO2 (mmHg), mean ± SD 73.7 ± 12.8 72.0 ± 13.0 0.5270
Arterial pH, mean ± SD 7.30 ± 0.03 7.29 ± 0.03 0.7450
PaO2 (mmHg), mean ± SD 64.3 ± 17.6 73.3 ± 27.9 0.1480
SpO2 (%), median [IQR] 90.1 [87.0–94.1] 92.0 [88.0–96.0] 0.1835
HCO3− (mmol ­L−1), mean ± SD 34.3 ± 5.9 33.1 ± 6.30 0.3680
PaO2/FiO2, mean ± SD 203.2 ± 45.5 222.4 ± 71.0 0.4801
c −1
Lactate (mmol ­L ), median [IQR] 1.1 [0.7–1.6] 1.1 [0.9–1.5] 0.7376
BMI body mass index, FiO2 fraction of inspired oxygen, HCO3− bicarbonate, HFNT high-flow nasal therapy, ICU intensive care unit, IQR interquartile range (first and third
quartile), NIV noninvasive ventilation, PaO2arterial partial pressure, PaCO2 partial pressure of carbon dioxide, RASS Richmond agitation-sedation scale, SAPS simplified
acute physiology score, SD standard deviation
a
Data was not available for five patients (three in HFNT and two NIV group)
b
Data was not available for one patient in NIV group
c
Data was not available for one patient in HFNT group

respiratory rate, the proportion of patients with wors- A higher proportion of patients in the NIV group
ening or no significant reduction in P
­ aCO2 values after showed poor tolerance to the intervention by 6 h (74%)
6 h, length of hospital stay, and hospital mortality are compare to HFNT (35%) (p = 0.0019). The discomfort
reported in Table 2. score was slightly higher in the NIV group at 2 and 6 h of
Cortegiani et al. Crit Care (2020) 24:692 Page 7 of 13

Table 2 Clinical outcomes in the high flow nasal therapy (HFNT) and noninvasive ventilation (NIV) groups
HFNT group NIV group p value

N 40 39
Treatment changes from baseline to 2 h, n (%)
Switching to NIV or HFNT 6 (15) 1 (2.6) 0.1084
IMV 0 (0 1 (2.6) 0.4937
No treatment change 34 (85) 37 (94.9) 0.2633
Treatment changes from baseline to 6 h, n (%)
Switch to NIV or HFNT 13 (32.5) 3 (7.7) 0.0061
Switch to IMV 1 (2.5) 1 (2.6) 1.0000
Switch to no support 2 (5) 6 (15.4) 0.1543
No treatment change 24 (60) 29 (74.4) 0.1745
Poor treatment tolerance/intolerance from baseline to 6 ­hoursa, n (%)
In patients switching to NIV or HFNT or IMV 5 (35.7) 3 (75) 0.2745
In patients switching to no support or with no treatment change 9 (34.6) 26 (74.3) 0.0019
Discomfort, median [IQR]
At 2 ­ha 1 [0–2] 3 [1–5] 0.0010
At 6 ­hb 0 [0–2] 2 [1–4] 0.0003
Borg dyspnea score, mean ± SD
At 2 ­ha 3±2 3±2 0.2509
At 6 ­hb 5±2 5±2 0.4865
No improvement of Borg dyspnea score at 6 h, n (%) 6 (15) 6 (15.4) 0.9620
Respiratory rate (per min), mean ± SD
At 2 ­hb 22 ± 5 22 ± 4 0.5789
At 6 ­hc 20 ± 4 21 ± 4 0.5573
PaCO2 worsening or reduction < 10 mmHg after 6 h, n (%) 23 (57.5) 14 (35.9) 0.0544
IMV during ­hospitalizationd
Subjects, n (%) 2 (5) 1 (2.6) 1.0000
Length of IMV (hours), median [IQR] 123 166 1.0000
[45.5–200] [166–166]
NIV during hospitalization
Subjects, n (%) 23 (57.5) 39 (100.0) < .0001
Length of NIV (hours), median [IQR] 70 48 0.3007
[14–142] [18–75]
Length of hospital stay (days), median [IQR]
All patients 10 [9–19] 13 [9–16] 0.6579
Survivors at hospital discharge 10 [9–19] 13 [9–16] 0.5510
Dead at hospital discharge 16 [9–22] 15 [3–19] 0.6150
In-hospital mortality, n (%) 2 (5) 6 (15.4) 0.1543
HFNT high flow nasal therapy, IQR interquartile range [first and third quartile], IMV invasive mechanical ventilation, NIV noninvasive ventilation
a
Poor tolerance was defined as patient-reported complaint to the assigned treatment that did not cause treatment interruption. Intolerance was defined as patient-
reported complaint that compromised the pursuit of the treatment (i.e., subject refusal)
b
Outcome evaluated on patients still receiving the assigned treatment at 2 h (34 HFNT, 37 NIV)
c
Outcome evaluated on patients still receiving the assigned treatment at 6 h (24 HFNT, 29 NIV)
d
IMV during hospitalization was calculated from baseline from hospital discharge or death

treatment. For the other secondary outcomes, no differ- in Additional file 2: Tables S4–S6. We found no dif-
ences were found between groups. ference in mean change in arterial pH from base-
Changes in arterial blood gases and vital param- line to 2 h (HFNT group: 0.04 ± 0.04, NIV group:
eters over the study timepoints for the per-protocol 0.05 ± 0.03, p = 0.39) in the per-protocol analysis at
analysis and intention-to-treat analysis were reported 2 h. In the same analysis, we found a significant differ-
ence in mean change of ­PaO2/FiO2 ratio (HFNT group:
Cortegiani et al. Crit Care (2020) 24:692 Page 8 of 13

Table 3 Modifications in ­PaCO2 values during follow-up period in study population stratified by intervention (high flow
nasal therapy and noninvasive ventilation)
HFNT group NIV group p value

Per-protocol 2 h
PaCO2 (mmHg) on patients who completed the treatment originally
allocated at 2 h, mean ± SD
Subjects, n 34 37 –
At baseline ( T0) 74.0 ± 13.5 72.2 ± 13.3 0.5845
After 2 h ( T2h ) 67.2 ± 16.4 62.7 ± 13.5 0.1933
After 6 h ( T6h ) 64.5 ± 15.8 57.9 ± 12.2 0.0630
ΔT2h − T0 − 6.8 ± 8.7 − 9.5 ± 8.5 0.4040
ΔT6h − T0 − 9.5 ± 13.0 − 14.3 ± 11.1 0.0962
ΔT6h − T2h − 2.7 ± 9.7 − 4.8 ± 7.1 0.1637
Per-protocol 6 h
PaCO2 (mmHg) on patients who completed the treatment originally
allocated at 6 h, mean ± SD
Subjects, n 24 29 −
At baseline ( T0) 72.7 ± 10.3 74.0 ± 13.7 0.7955
After 2 h ( T2h ) 64.7 ± 8.7 63.7 ± 14.5 0.7493
After 6 h ( T6h ) 61.4 ± 7.7 59.8 ± 12.6 0.5632
ΔT2h − T0 − 8.0 ± 6.5 − 10.3 ± 8.9 0.5200
ΔT6h − T0 − 11.3 ± 7.3 − 14.2 ± 12.0 0.4475
ΔT6h − T2h − 3.3 ± 6.9 − 3.9 ± 7.6 0.8163
Intention-to-treat analysis
PaCO2 (mmHg) of enrolled patients, mean ± SD
Subjects, n 40 39
At baseline ( T0) 73.7 ± 12.8 72.0 ± 13.0 0.5270
After 2 h ( T2h ) 68.2 ± 15.6 63.4 ± 13.6 0.1387
After 6 h ( T6h ) 64.0 ± 14.9 58.1 ± 12.4 0.0610
ΔT2h − T0 − 5.5 ± 9.3 − 8.6 ± 9.3 0.2940
ΔT6h − T0 − 9.7 ± 13.2 − 13.9 ± 11.3 0.1329
ΔT6h − T2h − 4.2 ± 11.1 − 5.3 ± 7.5 0.3670
HFNT high flow nasal therapy, NIV noninvasive ventilation, PaCO2 partial pressure of carbon dioxide, SD standard deviation, Δ difference in ­PaCO2 values between
timepoints

−16.5 ± 52.2 mmHg, NIV group: 7.2 ± 56.3 mmHg, HFNT in AECOPD have not been demonstrated so far
p = 0.0357) (see Additional file 2: Table S4). [16]; (2) NIV is considered the gold standard respiratory
support for managing patients with AECOPD with high-
quality evidence supporting its use [3]. However, HFNT
Discussion has shown several valuable effects in COPD patients
The main findings of this trial are that HFNT was non- [16] and there are some drawbacks of using NIV, such as
inferior to NIV as initial ventilatory support in mean reduced comfort and poor patient–ventilator interaction,
­PaCO2 reduction in patients with mild-to-moderate which is often challenging to recognize and manage [4, 5,
AECOPD considering a non-inferiority margin of 35]. We intended to determine whether HFNT was not
10 mmHg; HFNT was able to reduce ­PaCO2 over both inferior to NIV in achieving relevant changes in physi-
study timepoints significantly, but 32% (14/40) of patients ologic outcomes in AECOPD patients (i.e., ­PaCO2) and
were switched to NIV within 6 h. explore its safety as the first step for future superiority
To the best of our knowledge, this is the first multi- trials.
center randomized controlled non-inferiority trial com- Previously published studies had shown the promis-
paring NIV and HFNT in mild-to-moderate AECOPD. ing results on the efficacy and safety of HFNT use for
In our study, we included patients with mild-to-moderate treating AECOPD. A retrospective study performed in
AECOPD for two reasons: (1) the efficacy and safety of a single ICU compared HFNT and NIV in 82 patients
Cortegiani et al. Crit Care (2020) 24:692 Page 9 of 13

Fig. 2 a, c Report boxplots showing median, interquartile range and mean (full dot) for P ­ aCO2 differences (ΔPaCO2) between baseline (T0), 2 h (T2h)
and 6 h (T6h) in HFNT and NIV groups (*p value < 0.05, difference ≠ 0) in per-protocol analysis at 2 h and at 6 h, respectively; b, d report mean P
­ aCO2
(and 95% confidence interval) observed at T0, T2h, T6h in HFNT and NIV groups in per-protocol analysis at 2 h and at 6 h, respectively

with mild-to-moderate AECOPD. No difference was 30-day intubation rate (25% vs. 27%) and mortality rate
found in the proportion of patients who switched (16% vs. 18%) in 92 patients with moderate AECOPD.
treatment or received IMV (28% HFNT vs. 39% NIV). The investigators also compared blood gases (including
Of note, patients were assigned to study treatments ­PaCO2) at 6- and 24-h, finding no difference between
if they received HFNT or NIV for at least 4 h within groups [21]. The results included between-group analy-
the first 24 h from admission [22]. In an observational sis at each timepoint, and no comparison of trends or
study performed in one respiratory unit, no difference differences were reported. A randomized study per-
was found between HFNT and NIV with regard to the formed in one respiratory unit concluded that HFNT
Cortegiani et al. Crit Care (2020) 24:692 Page 10 of 13

Fig. 3 Absolute difference between HFNT and NIV treatment in mean ­PaCO2 reduction after 2 h (and 1-sided 95% confidence interval), according
to conducted analyses: per-protocol on patients who completed the treatment originally allocated after 2 h (PP 2 h) and intention-to-treat (ITT). The
black box indicates the mean ­PaCO2 reduction after 2 h, full lines indicate 95% confidence interval and dashed line the pre-planned non-inferiority
margin of 10 mmHg

and NIV were both effective in improving blood gases perspective, all the most because NIV is a true standard
in 168 patients with AECOPD and the HFNT group of care in this context.
had a lower rate of complications and higher comfort. The external validity of our findings is supported by
No sample size calculation and protocol registration the multicenter design, but the results may have changed
have been reported and the outcomes assessment was according to different expertise in the use of devices, set-
done after 12 h and 5 days [36]. tings and management protocols of AECOPD. Of note,
We showed that HFNT might be a feasible initial ven- all patients were continuously monitored with standard
tilator strategy in the management of mild-to-moderate tools and serial blood gases check, so our findings can-
AECOPD, finding a mean difference in ­PaCO2 reduction not be extrapolated to settings with different levels of
of 2.66 mmHg at 2 h, and the 95% CI upper boundary of care. Larger clinical trials comparing HFNT and NIV
absolute difference in mean ­PaCO2 reduction not reach- in this patient population are currently ongoing, focus-
ing the set non-inferiority margin of 10 mmHg. Never- ing on other outcomes such as endotracheal intubation
theless, we acknowledge that a different non-inferiority (NCT03014869) and treatment failure (NCT03466385)
margin and a larger sample size would have changed the or with crossover design (NCT03033251).
study conclusions. From a clinical point of view, a cli- Strengths of the study were the multicenter rand-
nician may prefer a technique allowing a greater and omized design, the prospective registration, and pro-
faster decarboxylation. In fact, 32% of patients treated tocol publication and the analyses, both per-protocol
with HFNT were switched to NIV, even if patients had and intention-to-treat, according to CONSORT rec-
a slightly lower severity of illness (SAPS II) than those in ommendations for non-inferiority trials. However, our
the NIV group (Table 1 and Additional file 2). Moreover, study has limitations. First, due to the nature of the
in our study cohort, patients randomized to HFNT wors- interventions, blinding was not possible. This study was
ened the oxygenation during the 6 h (Additional file 2: a non-inferiority trial with a primary physiologic out-
Table S4), and those who subsequently underwent NIV come, leaving uncertainty on stronger patient-related
during hospitalization had a longer length of NIV than outcomes. Moreover, due to the study design, the vari-
those originally allocated to the NIV group (Table 2). ables associated with the need for escalation of treat-
Altogether, these findings should incite caution in declar- ments cannot be adequately evaluated. Although we
ing that HFNT is non-inferior to NIV from a clinical found lower baseline oxygenation in HFNT patients
Cortegiani et al. Crit Care (2020) 24:692 Page 11 of 13

who escalated the treatment to NIV or IMV after 6 h Supplementary information

compared to patients who did not switch (see Addi- Supplementary information accompanies this paper at https​://doi.
tional file 2: Table S7), further studies should assess org/10.1186/s1305​4-020-03409​-0.
predictors for HFTN success in this patient population.
Our patient cohort was old and the mean BMI in the Additional file 1: Description of data: CONSORT (consolidated standards
of reporting trials) Checklist for Non-inferiority and Equivalence Trials
HFNT group was 30 kg/m2 (± 8.7). These factors may Checklist for Non-inferiority and Equivalence Trials.
limit the external validity of our findings. Furthermore,
Additional file 2: Table S1: Characteristics of interventions in the high
we cannot exclude the co-presence of obesity-related flow nasal therapy (HFNT) and noninvasive ventilation group (NIV);
hypoventilation. It is difficult to screen patients for this Table S2: Per-protocol 2 h. Patients’ characteristics in the noninvasive
condition during the first management AECOPD, but ventilation (NIV) and high flow nasal therapy (HFNT) groups at baseline;
Table S3: Per-protocol 6 h. Patients’ characteristics in the noninvasive
we excluded patients on domiciliary NIV. We did not ventilation (NIV) and high flow nasal therapy (HFNT) groups at baseline;
register the tidal volume generated or estimated by the Figure S1: Absolute difference between HFNT and NIV treatment in mean
ventilators in the NIV group. Thus, we cannot exclude ­PaCO2 reduction after 6 h (and 1-Sided 95% confidence interval), accord-
ing to conducted analyses: intention-to-treat (ITT) and per-protocol on
that the applied level of pressure support might have patients who completed the treatment originally allocated after 6 h (PP
affected patients’ comfort and NIV tolerance. Moreo- 6hs). Figure S2: Changes in P ­ aCO2 values during time (intention to treat
ver, the PEEP values applied were slightly higher than analysis). Table S4: Per-protocol 2 h. Differences during follow-up in clinical
characteristics in the noninvasive ventilation (NIV) and high flow nasal
the range stated in the protocol. therapy (HFNT) groups; Table S5: Per-protocol 6 h. Differences during
Our study had a parallel-group design with short-term follow-up in clinical characteristics in the noninvasive ventilation (NIV)
timepoints of assessment, so we cannot evaluate the effi- and high flow nasal therapy (HFNT) groups. Table S6: Intention-to-treat
analysis. Differences during follow-up in clinical characteristics in the
cacy and safety of sequential use of the study treatments. noninvasive ventilation (NIV) and high flow nasal therapy (HFNT) groups.
The findings that 32% of patients after 6 h and more than Table S7: reports patients’ characteristics at baseline in the high flow nasal
50% of patients during hospitalization needed NIV sug- therapy (HFNT) group stratified by success by 6 h.

gest that the two treatments may have a complementary

role in AECOPD that should be further studied [19]. We Abbreviations
ABGs: Arterial blood gases; AECOPD: Acute exacerbation of chronic obstruc-
collected data about the primary outcome, blood gases tive pulmonary disease; CI: Confidence interval; CONSORT: Consolidated
and secondary outcomes at 6 h as the longest follow-up standards of reporting trials; COPD: Chronic obstructive pulmonary disease;
without further assessments. This can limit the evalu- FiO2: Inspired oxygen fraction; HFNT: High flow nasal therapy; ICU: Intensive
care unit; NIV: Noninvasive ventilation; PaCO2: Arterial partial pressure of car-
ation of the long-term efficacy of HFNT. However, we bon dioxide; PEEP: Positive end-expiratory pressure; RASS: Richmond Agitation
registered patients’ safety-related outcomes until hospital Sedation Scale; SAPS II: Simplified Acute Physiology Score; SpO2: Peripheral
discharge or death. oxygen saturation.
At 6 h, the number of patients actually treated as ran- Acknowledgements
domly allocated was slightly lower (n = 53) than the cal- The list of HF-AECOPD study investigators is reported in the acknowledgment
culated sample size (n = 56) for the primary outcome at section. List of The HF-AECOPD study investigators: Lorenzo Ball (Anesthesia
and Intensive Care, Ospedale Policlinico San Martino; Department of Surgical
2 h. The power of our per-protocol analysis after 6 h was Sciences and Integrated Diagnostic, University of Genoa, Genoa, Italy); Tiziana
76.5%, and therefore, the estimate of the compared effect Bove (Anesthesia and Intensive Care Clinic, Department of Anesthesia and
at this timepoint should be considered with caution. Intensive Care, Azienda Sanitaria Integrata di Udine; Department of Medicine,
University of Udine, Udine, Italy); Raffaele Campisi (Respiratory Medicine Unit,
Finally, it has to take into account also the documented A.O.U. “Policlinico-Vittorio Emanuele”, Catania, Italy); Paola Chirco (Department
clinical heterogeneity of COPD exacerbations [37]; there- of Surgical, Oncological and Oral Science University of Palermo, Italy); Maria
fore, although we used strict criteria to define mild-to- Stella Dionisi (Emergency Department, “S. Maria della Misericordia” Hospital,
Perugia, Italy); Mariachiara Ippolito (Department of Surgical, Oncological and
moderate AECOPD, we may have enrolled patients with Oral Science, University of Palermo, Italy); Riccardo Fantini (Respiratory Dis-
different trajectories of the diseases and responses to eases Unit, University Hospital of Modena, Italy); Luca Guidelli (Pulmonology
ventilator support and medical treatments. and Respiratory Intensive Care Unit, S. Donato Hospital, Arezzo, Italy); Uberto
Maccari (Pulmonology and Respiratory Intensive Care Unit, S. Donato Hospital,
Arezzo, Italy); Luca Tabbì (Respiratory Diseases Unit, University Hospital of
Modena, Italy); Maria Rita Taliani (Emergency Department, “S. Maria della
Conclusions Misericordia” Hospital, Perugia, Italy).
In this trial, HFNT was statistically non-inferior to Authors’ contributions
NIV as initial ventilatory support in decreasing ­PaCO2 AC, FL, FM, PN, and CG had full access to all the data in the study and take
after 2 h of treatment in patients with mild-to-moder- responsibility for the integrity of the data and the accuracy of the data analy-
sis, including and especially any adverse effects. AC, FM, CC, and CG wrote the
ate AECOPD, considering a non-inferiority margin of manuscript; AC, FL, FM, PG, RS, CC, ACa, AB, EG, SMR, RT, VC, EL, LV, EC, AG, SN,
10 mmHg. However, 32% of patients receiving HFNT PN, and CG contributed substantially to the study design, data analysis and
required NIV by 6 h. Further trials with superiority interpretation, and the writing of the manuscript. All authors approved the
final version of the manuscript and agreed to be accountable for all aspects of
design should evaluate efficacy toward stronger patient- the work in ensuring that questions related to the accuracy or integrity of any
related outcomes and safety of HFNT in AECOPD.
Cortegiani et al. Crit Care (2020) 24:692 Page 12 of 13

part of the work are appropriately investigated and resolved. All authors read References
and approved the final manuscript. 1. Singh D, Agusti A, Anzueto A, Barnes PJ, Bourbeau J, Celli BR, et al. Global
strategy for the diagnosis, management, and prevention of chronic
Funding obstructive lung disease: the GOLD science committee report 2019. Eur
No funding has been received for this trial. Some centers received material Respir J Engl. 2019;2019:53.
support (devices and cannula) by Fisher and Paykel that has no role in proto- 2. Celli BR, Barnes PJ. Exacerbations of chronic obstructive pulmonary
col design, data management, interpretation and analysis. disease. Eur Respir J Engl. 2007;29:1224–38.
3. Rochwerg B, Brochard L, Elliott MW, Hess D, Hill NS, Nava S, et al. Official
Availability of data and materials ERS/ATS clinical practice guidelines: noninvasive ventilation for acute
The dataset consisting of de-identified participants’ data is available from the respiratory failure. Eur Respir J. 2017;50:1602426.
corresponding author upon reasonable requests. 4. Gregoretti C, Pisani L, Cortegiani A, Ranieri VM. Noninvasive ventilation in
critically ill patients. Crit Care Clin US. 2015;31:435–57.
Ethics approval and consent to participate 5. Bruni A, Garofalo E, Pelaia C, Messina A, Cammarota G, Murabito P, et al.
The study protocol was approved by “Comitato Etico Sezione Area Centro” Patient-ventilator asynchrony in adult critically ill patients. Minerva Anest-
Ethics Committee (approval no. 245—October 24, 2017, Catanzaro, Italy) and esiol Italy. 2019;85:676–88.
by the local ethics committee of all the study centers. 6. Cortegiani A, Russotto V, Antonelli M, Azoulay E, Carlucci A, Conti G, et al.
Ten important articles on noninvasive ventilation in critically ill patients
Consent for publication and insights for the future: a report of expert opinions. BMC Anesthesiol.
Written informed consent for publication was obtained from each participant 2017;17:122.
enrolled or legal representatives. 7. Garofalo E, Bruni A, Pelaia C, Liparota L, Lombardo N, Longhini F, et al.
Recognizing, quantifying and managing patient-ventilator asynchrony
Competing interests in invasive and noninvasive ventilation. Expert Rev Respir Med Engl.
Dr. Cortegiani, Prof. Giarratano and Prof. Gregoretti declare a patent pending, 2018;12:557–67.
in association with the University of Palermo—Italy (No. 102019000020532— 8. Spoletini G, Cortegiani A, Gregoretti C. Physiopathological rationale of
Italian Ministry of Economic Development), not discussed in the present using high-flow nasal therapy in the acute and chronic setting: a narra-
study. Prof. Gregoretti received honoraria for lectures or consultancies from tive review. Trends Anaesth Crit Care. 2019;26–27:22–9.
Philips, Resmed, Vivisol, OrionPharma, Origin (not relevant to this protocol). 9. Groves N, Tobin A. High flow nasal oxygen generates positive airway
Prof. Enrico Clini received fees for lectures from Chiesi, Astra-Zeneca, Menarini, pressure in adult volunteers. Aust Crit Care. 2007;20:126–31.
Guidotti_Malesci, and AlfaSigma (not relevant to this protocol). Prof. Stefano 10. Ricard JD, Roca O, Lemiale V, Corley A, Braunlich J, Jones P, et al. Use of
Nava’s institution received unrestricted research grant from Fisher and Paykel. nasal high flow oxygen during acute respiratory failure. Intensive Care
Prof. Paolo Navalesi’s institution received funding from Maquet Critical Care, Med. 2020. https​://doi.org/10.1007/s0013​4-020-06228​-7.
Draeger, and Intersurgical S.p.A.; he received honoraria/speaking fees from 11. Fontanari P, Zattara-Hartmann MC, Burnet H, Jammes Y. Nasal eupnoeic
Maquet Critical Care, Orionpharma, Philips, Resmed, Merck Sharp & Dome, and inhalation of cold, dry air increases airway resistance in asthmatic
Novartis; he disclosed that he contributed to the development of the helmet patients. Eur Respir J. 1997;10:2250–4.
Next, whose license for patent belongs to Intersurgical S.P.A., and receives 12. On LS, Boonyongsunchai P, Webb S, Davies L, Calverley PMA, Costello
royalties for that invention. Prof. Federico Longhini and Prof. Paolo Navalesi RW. Function of pulmonary neuronal M2 muscarinic receptors in stable
contributed to the development of a device not discussed in the present chronic obstructive pulmonary disease. Am J Respir Crit Care Med.
study whose patent is in progress (European Patent application number 2001;163:1320–5.
EP20170199831). Dr. Paolo Groff received honoraria for lectures from Aspen 13. Kilgour E, Rankin N, Ryan S, Pack R. Mucociliary function deteriorates in
Pharmaceuticals and Menarini Pharmaceuticals (not relevant to this study). Dr. the clinical range of inspired air temperature and humidity. Intensive Care
Luigi Vetrugno received travel support for congress lecture by Cook Medical. Med. 2004;30:1491–4.
The remaining authors have disclosed that they do not have any conflicts of 14. Crimi C, Noto A, Cortegiani A, Campisi R, Heffler E, Gregoretti C, et al.
interest. High flow nasal therapy use in patients with acute exacerbation of COPD
and bronchiectasis: a feasibility study. COPD J Chronic Obstr Pulm Dis.
Author details 2020;17:184–90.
1
Department of Surgical, Oncological and Oral Science (Di.Chir.On.S.), Uni- 15. Pisani L, Fasano L, Corcione N, Comellini V, Musti MA, Brandao M, et al.
versity of Palermo, Palermo, Italy. 2 Department of Anesthesia, Intensive Care Change in pulmonary mechanics and the effect on breathing pattern
and Emergency, Policlinico Paolo Giaccone, University of Palermo, Palermo, of high flow oxygen therapy in stable hypercapnic COPD. Thorax.
Italy. 3 Intensive Care Unit, Department of Medical and Surgical Sciences, 2017;72:373–5.
University Hospital Mater Domini, Magna Graecia University, Catanzaro, Italy. 16. Pisani L, Astuto M, Prediletto I, Longhini F. High flow through nasal can-
4
Value‑Based Healthcare Unit, IRCCS MultiMedica, Sesto San Giovanni, Milan, nula in exacerbated COPD patients: a systematic review. Pulmonology.
Italy. 5 Emergency Department, “S. Maria Della Misericordia” Hospital, Perugia, 2019;25:348–54.
Italy. 6 Pulmonology and Respiratory Intensive Care Unit, S. Donato Hospital, 17. Pilcher J, Eastlake L, Richards M, Power S, Cripps T, Bibby S, et al. Physi-
Arezzo, Italy. 7 Respiratory Medicine Unit, A.O.U. “Policlinico-Vittorio Emanuele”, ological effects of titrated oxygen via nasal high-flow cannulae in COPD
Catania, Italy. 8 Pulmonary Rehabilitation Unit, Department of Medicina exacerbations: a randomized controlled cross-over trial. Respirology.
E Chirurgia, Istituti Clinici Scientifici Maugeri, Università Insubria Varese, 2017;22:1149–55.
Pavia, Italy. 9 Clinical and Experimental Medicine PhD Program, University 18. Bräunlich J, Wirtz H. Nasal high-flow in acute hypercapnic exacerbation of
of Modena and Reggio Emilia, Modena, Italy. 10 Department of Clinical, COPD. Int J COPD. 2018;13:3995–3897.
Integrated and Experimental Medicine (DIMES), Respiratory and Critical Care 19. Longhini F, Pisani L, Lungu R, Comellini V, Bruni A, Garofalo E, et al. High-
Unit, S. Orsola‑Malpighi Hospital, Alma Mater University, Bologna, Italy. 11 Unit flow oxygen therapy after noninvasive ventilation interruption in patients
of Emergency Medicine, Department of Medical Sciences, University of Turin, recovering from hypercapnic acute respiratory failure: a physiological
Turin, Italy. 12 Department of Medicine, Clinic of Anesthesia and Intensive Care, crossover trial. Crit Care Med. 2019;47:e506–11.
University of Udine, Udine, Italy. 13 Respiratory Diseases Unit, Department 20. Rittayamai N, Phuangchoei P, Tscheikuna J, Praphruetkit N, Brochard L.
of Medical and Surgical Sciences SMECHIMAI, University Hospital of Modena Effects of high-flow nasal cannula and non-invasive ventilation on inspir-
Policlinico, University of Modena Reggio Emilia, Modena, Italy. 14 Section atory effort in hypercapnic patients with chronic obstructive pulmonary
of Anesthesiology and Intensive Care, Department of Medicine ‑ DIMED, Uni- disease: a preliminary study. Ann Intensive Care. 2019;9:122.
versity of Padova, Padova, Italy. 15 Fondazione ‘Giglio’, Cefalù, Palermo, Italy. 21. Lee MK, Choi J, Park B, Kim B, Lee SJ, Kim SH, et al. High flow nasal can-
nulae oxygen therapy in acute-moderate hypercapnic respiratory failure.
Received: 31 August 2020 Accepted: 24 November 2020 Clin Respir J. 2018;12:2046–56.
22. Sun J, Li Y, Ling B, Zhu Q, Hu Y, Tan D, et al. High flow nasal cannula
oxygen therapy versus non-invasive ventilation for chronic obstructive
Cortegiani et al. Crit Care (2020) 24:692 Page 13 of 13

pulmonary disease with acute-moderate hypercapnic respiratory failure: 31. Maggiore SM, Idone FA, Vaschetto R, Festa R, Cataldo A, Antonicelli F, et al.
an observational cohort study. Int J COPD. 2019;14:1229–39. Nasal high-flow versus venturi mask oxygen therapy after extubation:
23. Pantazopoulos I, Daniil Z, Moylan M, Gourgoulianis K, Chalkias A, Zakyn- effects on oxygenation, comfort, and clinical outcome. Am J Respir Crit
thinos S, et al. Nasal high flow use in COPD patients with hypercapnic Care Med. 2014;190:282–8.
respiratory failure: treatment algorithm & review of the literature. COPD J 32. Bräunlich J, Wirtz H. NHF and hypercapnia: how brief can you look?
Chronic Obstr Pulm Dis. 2020;7:101–11. Respirology. 2017;22:1049–50.
24. Cortegiani A, Longhini F, Carlucci A, Scala R, Groff P, Bruni A, et al. High- 33. Brochard L, Mancebo J, Wysocki M, Lofaso F, Conti G, Rauss A, et al.
flow nasal therapy versus noninvasive ventilation in COPD patients with Noninvasive ventilation for acute exacerbations of chronic obstructive
mild-to-moderate hypercapnic acute respiratory failure: study protocol pulmonary disease. N Engl J Med. 1995;333:817–22.
for a noninferiority randomized clinical trial. Trials. 2019;20:450. 34. Piaggio G, Elbourne DR, Pocock SJ, Evans SJW, Altman DG. Reporting
25. Vogelmeier CF, Criner GJ, Martinez FJ, Anzueto A, Barnes PJ, Bourbeau J, of noninferiority and equivalence randomized trials: extension of the
et al. Global strategy for the diagnosis, management, and prevention of CONSORT 2010 statement. JAMA - J Am Med Assoc. 2012;308:2594–604.
chronic obstructive lung disease 2017 report. Am J Respir Crit Care Med. 35. Longhini F, Liu L, Pan C, Xie J, Cammarota G, Bruni A, et al. Neurally-
2017;195:557–82. adjusted ventilatory assist for noninvasive ventilation via a helmet in
26. Pisani L, Mega C, Vaschetto R, Bellone A, Scala R, Cosentini R, et al. Orona- subjects with COPD exacerbation: a physiologic study. Respir Care US.
sal mask versus helmet in acute hypercapnic respiratory failure. Eur Respir 2019;64:582–9.
J. 2015;45:691–9. 36. Cong L, Zhou L, Liu H, Wang J. Outcomes of high-flow nasal cannula
27. Kelly BJ, Matthay MA. Prevalence and severity of neurologic dysfunc- versus non-invasive positive pressure ventilation for patients with acute
tion in critically ill patients: Influence on need for continued mechanical exacerbations of chronic obstructive pulmonary disease. Int J Clin Exp
ventilation. Chest. 1993;104:1818–24. Med. 2019;12:10863–7.
28. Quan H, Li B, Couris CM, Fushimi K, Graham P, Hider P, et al. Updating and 37. Lopez-Campos JL, Agustí A. Heterogeneity of chronic obstructive pul-
validating the charlson comorbidity index and score for risk adjustment monary disease exacerbations: a two-axes classification proposal. Lancet
in hospital discharge abstracts using data from 6 countries. Am J Epide- Respir Med Engl. 2015;3:729–34.
miol. 2011;173:676–82.
29. Sessler CN, Gosnell MS, Grap MJ, Brophy GM, O’Neal PV, Keane KA, et al.
The Richmond Agitation-Sedation Scale: validity and reliability in adult Publisher’s Note
intensive care unit patients. Am J Respir Crit Care Med. 2002;166:1338–44. Springer Nature remains neutral with regard to jurisdictional claims in pub-
30. Kendrick KR, Baxi SC, Smith RM. Usefulness of the modified 0–10 Borg lished maps and institutional affiliations.
scale in assessing the degree of dyspnea in patients with COPD and
asthma. J Emerg Nurs. 2000;26:216–22.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission

• thorough peer review by experienced researchers in your field
• rapid publication on acceptance
• support for research data, including large and complex data types
• gold Open Access which fosters wider collaboration and increased citations
• maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions