Abstract
BACKGROUND: Oxygen therapy via high-flow nasal cannula (HFNC) has been extensively used during the COVID-19 pandemic. The number of devices has also increased. We conducted this study to answer the following questions: Do HFNC devices differ from the original device for work of breathing (WOB) and generated PEEP?
METHODS: Seven devices were tested on ASL 5000 lung model. Compliance was set to 40 mL/cm H2O and resistance to 10 cm H2O/L/s. The devices were connected to a manikin head via a nasal cannula with FIO2 set at 0.21. The measurements were performed at baseline (manikin head free of nasal cannula) and then with the cannula and the device attached with oxygen flow set at 20, 40, and 60 L/min. WOB and PEEP were assessed at 3 simulated inspiratory efforts (−5, –10, –15 cm H2O muscular pressure) and at 2 breathing frequencies (20 and 30 breaths/min). Data were expressed as median (first-third quartiles) and compared with nonparametric tests to the Optiflow device taken as reference.
RESULTS: Baseline WOB and PEEP were comparable between devices. Over all the conditions tested, WOB was 4.2 (1.0−9.4) J/min with the reference device, and the relative variations from it were 0, −3 (2–4), 1 (0–1), −2 (1–2), −1 (1–2), and −1 (1–2)% with Airvo 2, G5, HM80, T60, V500, and V60 Plus devices, respectively, (P < .05 Kruskal-Wallis test). PEEP was 0.9 (0.3–1.5) cm H2O with Optiflow, and the relative differences were −28 (22–33), −41 (38–46), −30 (26–36), −31 (28–34), −37 (32–42), and −24 (21–34)% with Airvo 2, G5, HM80, T60, V500, and V60 Plus devices, respectively, (P < .05 Kruskal-Wallis test).
CONCLUSIONS: WOB was marginally higher and PEEP marginally lower with devices as compared to the reference device.
Introduction
High-flow nasal cannula (HFNC) oxygen delivery has been shown to reduce intubation and mortality in patients with hypoxemic acute respiratory failure before the COVID-19 pandemic.1 HFNC can also improve weaning success in patients with low2 or high3 risk of extubation failure. During the COVID-19 pandemic, HFNC has been extensively used in the prehospital setting, in the emergency department, in dedicated high-dependence units, and in the ICU to support failing oxygenation and prevent intubation and, hence, to spare the ICU resources. Coupled with prone positioning, HFNC can reduce the rate of intubation as compared to patients kept in supine position.4 The recommended set oxygen flow is in the range 50–60 L/min depending on patient’s tolerance and effect on oxygenation.5 During this time a landmark study demonstrating the efficacy of HFNC was published and the early device improved,1 HFNC function was proposed as an option in several ventilators used in the ICU or in the step-down units soon after. HFNC settings include FIO2 0.21–1.0 and inspiratory flow up to 60 L/min or even higher with the most recent devices.
Early bench studies found that HFNC devices differed from each other regarding the achieved level of FIO2 and the quality of humidification of inspired air.6 In non-intubated patients with acute hypoxemic respiratory failure, the use of HFNC can decrease both the patient’s inspiratory effort and the work of breathing (WOB).7 However, no study has compared WOB between HFNC devices. Since many HFNC devices from different manufacturers are now available, such a comparison is logical. Indeed, it is important to verify whether WOB differs substantially between devices. Of note, the V60 Plus has been voluntarily recalled for a potential issue with the electrical circuit that controls the power supply to the ventilator and its alarm (https://www.fda.gov/safety/recalls-market-withdrawals-safety-alerts/philips-respironics-issues-voluntary-recall-notificationfield-safety-notice-v60-ventilator-product. Accessed May 16, 2022).
If WOB is similar between devices, using different HFNC brands would have a neutral clinical effect. Our hypothesis was that WOB may differ between the original HFNC device and other devices with different configurations. The mechanisms to deliver flow and adjust FIO2 used by these devices are unknown and may vary in many respects, including flow generation and oxygen blending. Another issue relates to the cost between devices. Certain ventilators include an HFNC mode, and if re-intubation is necessary, it can be convenient to have the ventilator readily available, although cost may be a factor. However, comparing HFNC devices would be difficult in patients; henceforth, we conducted a bench study to explore this question.
QUICK LOOK
Current Knowledge
High-flow nasal cannula (HFNC) oxygen therapy is increasingly used worldwide in patients with acute hypoxemic respiratory failure. There are also a growing number of available devices to deliver HFNC treatment. Whether these devices affect the work of breathing (WOB) differently has not been studied.
What This Paper Contributes to Our Knowledge
In a bench study, we measured the WOB among 7 HFNC devices. As compared to the original HFNC device, WOB did not differ significantly among the newer devices.
Methods
Seven HFNC devices were tested: Optiflow (Fisher & Paykel Healthcare, Auckland, New Zealand) equipped with a blender (Bio-Med Devices, Guilford, Connecticut), Airvo 2 (Fisher & Paykel), HM80 (BMC Medical, Beiji-ng, China), T60 (Air Liquide Medical Systems, Antony, France), V500 (Dräger, Lübeck, Germany), V60 Plus (Philips, Amsterdam, the Netherlands), and G5 (Hamilton Medical, Bonaduz, Switzerland). The first 3 devices listed above were specifically designed to deliver high-flow nasal oxygen; the last 4 are ventilators, which include an HFNC mode. The main characteristics of the devices are shown in Table 1.
Settings of the Devices for High-Flow Oxygen Tested
The same kind of nasal cannula (Optiflow 3S large size, Fisher & Paykel Healthcare) was used with each device. It was attached to the nose of a manikin head (Laerdal Medical, Stavanger, Norway) (Fig. 1). To minimize leaks, the manikin esophagus was clamped and the manikin mouth was occluded by a strap (Fig. 1). Each device was also connected to the ASL 5000 lung simulator (IngMar Medical, Pittsburgh, Pennsylvania). The ASL 5000 was set with a linear compliance of 40 mL/cm H2O and a resistance (inspiratory and expiratory being equal) of 10 cm H2O/Ls. To set the respiratory system compliance, we first took into consideration the current COVID-19 pandemic and used the value of 40 mL/cm H2O found by Grasselli et al8 in intubated subjects. We then attempted to determine the lung compliance in patients with acute hypoxemic respiratory failure placed on HFNC. In a study by Delorme et al,9 the vital capacity averaged 2.77 L in 12 subjects. The corresponding chest wall compliance computed as 4% of vital capacity10 was 111 mL/cm H2O. In a study by Mauri et al,7 the mean transpulmonary driving pressure averaged 4.3 cm H2O. At a mean tidal volume of 0.27 L in Delorme et al,9 lung compliance can be estimated to 63 mL/cm H2O, and hence respiratory system compliance to 40 mL/cm H2O, using the above value of the chest wall compliance. We set the resistance to 10 cm H2O/L/s according to the results found by Delorme et al9 in subjects with HFNC at < 60 L/min.
Schematic drawing up of the setup. In the bottom left, representative tracing recorded by ASL 5000 during baseline condition with the Optiflow device without cannula. Paw = airway pressure; Pmus = muscular pressure.
A sinusoidal half-wave inspiratory effort was simulated with the following settings: muscular pressure (Pmus) contraction during 16%, then pause during 2%, then relaxation during 20% of total breath duration, then passive expiration. Each effort was applied at 2 breathing frequencies of 20 and 30 breaths/min. The duration of inspiration and expiration was, therefore, 1.14 and 1.86 s and 0.76 and 1.24 s at 20 and 30 breaths/min, respectively. The 30 breaths/min rate was chosen because it was close to the mean value at the time of inclusion in the FLORALI trial of subjects with acute hypoxemic respiratory failure.1 The 20 breaths/min rate was chosen because it was far from the previous one and close to the mean value found by Mauri et al in subjects on HFNC.7 Therefore, these 2 breathing frequencies were clinically based and hence likely clinically relevant. Low, medium, and strong effort intensities were defined as −5, −10, and −15 cm H2O Pmus maximal amplitude (Fig. 2A), respectively. These levels were selected because a −10 cm H2O esophageal pressure (Pes) swing was found, on average, in clinical studies.7,9,11,12 The 2 other values were defined 50% below and above apart.
Method used to measure the work of breathing (WOB). A: Waveforms of flow and volume (right vertical axis) and airway (Paw), muscular (Pmus), chest wall elastic recoil (Pel, Pcw) pressures (left vertical axis) against time over one breath recorded with Optiflow device, strong effort, frequency 30 breaths/min without cannula and without oxygen flow, that is, baseline condition. B: Campbell diagram with volume on the y axis and pressure on the x axis. The area subtended by lung elastic recoil (Pel, L grey line) and Pmus-Patm (black curve) to the left defines the inspiratory WOB in the baseline condition. The red line is the elastic recoil of the chest wall (Pel, Pcw).
Three inspiratory flows set at the HFNC devices (20, 40, and 60 L/min) were tested in that order with each device (except for the V500, in which the highest inspiratory flow available was 50 L/min). We did not measure these set flows and assumed that the set flows were provided as such by the devices.
The experimental setup also included a pneumotachograph (3700 series, Hans Rudolph, Shawnee, Kansas) and a port to measure airway pressure inserted at the ASL inlet (Fig. 1). The airway pressure (Paw) port was connected to a pressure transducer (Gabarith PMSET 1DT-XX, Becton, Dickinson and Company, Franklin Lakes, New Jersey). This setup had a 0.79 cm H2O/L/s resistance.13 The pressure transducer and pneumotachograph were calibrated at ambient conditions before each experiment by using a pressure calibrator (717G, Fluke Biomedical, Everett, Washington) and a calibration pump of 1,000 mL ± 12 mL precision (Viasys Healthcare, Höchberg, Germany), respectively. This was used to ascertain that compliance and resistance set on the ASL were actually reached. Paw and flow signals were recorded separately by a data logger (Biopac 150, Biopac, Goleta, California).
Experiments were performed in a dedicated room at ambient air temperature and pressure. Each device was investigated in a single day. Heated humidifier was placed in the circuit (Fig. 1) but switched off and FIO2 set to 0.21. At first, without HFNC device and without nasal cannula in place, each combination of breathing frequency and simulated effort was run to define the baseline condition. The data logger and the ASL 5000 were started simultaneously in each condition at a sampling rate of 200 Hz and 512 Hz, respectively. After a 2-min recording, the data were stored for off-line analysis.
Data Analysis
The flow and airway pressure coming from the ASL 5000 of the last 30 breaths of each record were used for the off-line data analysis (Figure 1ESM, see related supplementary materials at http://www.rcjournal.com). This was automatically done via an application specifically designed in the MATLAB environment (MATLAB 2019b, MathWorks, Natick, Massachusetts). The WOB per breath was determined breath by breath from the Campbell diagram (Fig. 2). Combined WOB from the simulator, delivery device, and cannula was measured as the area under the curve subtended by the tidal volume in y axis and the Pmus-atmospheric pressure difference in x axis.
The primary end point was WOB. The secondary end points were the resistive and the elastic components of WOB and PEEP. WOB was expressed as J/min by multiplying the WOB per breath by the breathing frequency. We also provided the data of inspired tidal volume, peak inspiratory and expiratory flows, and WOB expressed as J/L measured on the same breaths as for WOB.
The values are expressed as median (first-to-third quartiles) unless otherwise stated. To make the summary of the results easier to follow, the relative variation of each device from the reference for the WOB and PEEP over all the conditions tested was also shown.
The cost of each device was estimated by using the data provided by and pertaining to our institution, which may not be representative to that in other hospitals. It includes the cost of the device (the cost of the ventilator, as an example) and of the ancillary components with the exception of the cost of the L/min oxygen flow. Because these costs are confidential, only the relative change from the reference was given. The normal distribution of the variables was tested using the Shapiro-Wilk test.
First, baseline WOB was compared between devices. It is expected that no difference should be found because no device was attached to the manikin. Second, WOB was compared across devices at each nominal high flow (ie, 20, 40, and 60 L/min [except for V500, which has a maximum flow of 50 L/min]) for each effort intensity and rate (Figure 1ESM, see related supplementary materials at http://www.rcjournal.com). Since a significant interaction between these 3 factors was anticipated, a series of Kruskal-Wallis or one-way analysis of variance test was planned; and, if significant, a pairwise comparison was done from the Optiflow device taken as reference by using Dunnett test. The Optiflow device was chosen as the reference because it was the first used in clinical practice and also because of its specific design. We anticipated that a bench design with a large number of highly reproducible breaths would make small differences between devices statistically significant though the clinical relevance of them would be meaningless. To deal with this issue, we applied a Bonferroni correction by dividing 0.05 by the number of comparisons, that is, 966 for the 7 variables mentioned above as the various end points, including the pairwise comparisons. Therefore, the P value deemed statistically significant was < .001. With such a stringent P value, the statistical significance of the differences would be closer to clinical relevance. Second, a 30% relative variation between devices was thought to reflect clinical importance because it was the mean difference in pressure-time product of Pes between standard oxygen therapy and HFNC found by Mauri et al7 in subjects with acute hypoxemic respiratory failure. The statistical analysis was conducted by using the R software version 4.3 (2020) (R Foundation for Statistical Computing, Vienna, Austria).
Results
WOB did not follow a normal distribution.
Work of Breathing
As expected, the baseline WOB was similar across devices for resistive, elastic, and total WOB. Across all devices, it was 0.85 (0.84−0.85) J/min versus 1.0 (1.01−1.02) J/min for low effort, 3.8 (3.8−3.8) J/min versus 4.3 (4.3−4.4) J/min for medium effort, and 8.5 (8.5−8.5) J/min versus 9.6 (9.6−9.6) J/min for strong effort at 20 versus 30 breaths/min effort rate, respectively. The same was true for its resistive and elastic components.
Across all the conditions tested, WOB was 4.2 (1.0−9.4) J/min with the reference device; and the relative variations from it were 0, 3 (2−4), 1 (0−1), 2 (1−2), 1 (1−2), and 1 (1−2) % with Airvo 2, G5, HM80, T60, V500, and V60 Plus devices, respectively, (P < .05 Kruskal-Wallis test).
Across all the efforts, the WOB was 4.9 (1.1−9.7), 4.3 (1.1−9.7), 4.3 (1.1−9.8), 4.2 (1.1−9.7), 4.6 (1.1−9.7), 4.9 (1.1−9.7), and 4.9 (1.1−9.7) J/min for Optiflow, Airvo 2, G5, HM80, T60, V500, and V60 Plus devices, respectively, at HFNC 20 L/min (Table 2), with a significant interaction between effort intensity and rate. The same was true at 40 and 60 L/min HFNC for total (Fig. 3) and resistive (Fig. 4) and elastic (Fig. 5) values of WOB. As of the 18 instances (3 effort intensities ×2 effort rates ×3 HFNC flows) for each device (except for V500 with 12 instances), the comparison to the reference device showed that the total WOB was significantly higher than Optiflow in 83% (15/18) Airvo 2, 83% (15/18) G5, 50% (9/18) HM80, 61% (11/18) T60, 66% (8/12) V500, and 50% (9/18) V60 Plus; and it was significantly lower than Optiflow in 0, 5.5 (1/18), 11 (2/18), 16 (3/18), 17 (2/12), and 12% (2/18) for the corresponding devices, respectively. The resistive WOB was lower than Optiflow in no instance for each device, and it was significantly higher than with Optiflow in 50 (9/18), 44 (8/18), 44 (8/18), 28 (5/18), 50 (6/12), and 17% (2/18) for the corresponding devices, respectively. The elastic WOB was not significantly lower than Optiflow with any device. It was significantly higher than Optiflow in 44 (8/18), 33 (6/18), 22 (4/18), 5.5 (1/18), 33 (4/12), and 39% (7/18) of the cases for the corresponding devices, respectively. The threshold of 30% difference between devices was never reached.
Total, Resistive, and Elastic Work of Breathing done by the lung, the High-Flow Nasal Cannula Device, and the Cannula, at Different Flows Merged Over Both the Intensity and the Rate of Effort
Box and whisker plots of the total work of breathing (WOB) (done by the lung, the device, and the nasal cannula) at 20 and 30 breaths/min and low, medium, and high respiratory efforts for 20 (A), 40 (B), and 60 L/min oxygen flow (C) across the devices. There are no data for the V500 device at 60 L/min as the highest flow achieved with it is 50 L/min. *P < .001 versus Optiflow device taken as the reference. The grey lines drawn along the median values of the Optiflow device.
Box and whisker plots of the resistive work of breathing (WOB) done by the lung, the device, and the nasal cannula at 20 and 30 breaths/min and low, medium, and high respiratory efforts for 20 (A), 40 (B), and 60 L/min oxygen flow (C) across the devices. There are no data for the V500 device at 60 L/min as the highest flow achieved with it is 50 L/min. *P < .001 versus Optiflow device taken as the reference. The grey lines drawn along the median values of the Optiflow device.
Box and whisker plots of the elastic work of breathing (WOB) done by the lung, the device, and the nasal cannula at 20 and 30 breaths/min and low, medium, and high respiratory efforts for 20 (A), 40 (B), and 60 L/min oxygen flow (C) across the devices. There are no data for the V500 device at 60 L/min as the highest flow achieved with it is 50 L/min. *P < .001 versus Optiflow device taken as the reference. The grey lines drawn along the median values of the Optiflow device.
PEEP
Across all conditions tested, PEEP was 0.9 (0.3–1.5) cm H2O with Optiflow; and the relative differences were −28 (22–33), –41 (38–46), –30 (26–36), –31 (28–34), –37 (32–42), and –24% (21–34) with Airvo 2, G5, HM80, T60, V500, and V60 Plus devices, respectively, (P < .05 Kruskal-Wallis test).
The PEEP generated by the high-flow oxygen increased with increasing oxygen flow as expected. This was the case with any device (Fig. 6). However, the level of that PEEP was consistently (100% of the occurrences for each device) higher with the Optiflow device than with any other, and the median difference between devices and reference was < 30% (Fig. 6). The difference in PEEP between Optiflow and other devices increased with increasing flow. PEEP never surpassed 2 cm H2O.
Box and whisker plots of the PEEP generated at 20 and 30 breaths/min and low, medium, and high respiratory efforts for 20 (A), 40 (B), and 60 L/min oxygen flow (C) across the devices. There are no data for the V500 device at 60 L/min as the highest flow achieved with it is 50 L/min. *P < .001 versus Optiflow device taken as the reference. The grey lines drawn along the median values of the Optiflow device. Values are median (first-third quartiles) in J/min. WOB = work of breathing.
Tidal Volume and Peak Flows
Tables 1–3 in the supplementary materials (see related supplementary materials at http://www.rcjournal.com) display the mean values (for the sake of clarity, the SD is not shown) of the nspired tidal volume and peak inspiratory and expiratory flows with the devices. The values of inspired tidal volume were higher with any device than with the reference. Even though these differences may not be clinically relevant, they explain why the WOB is slightly higher with these devices compared to Optiflow. The same was true regarding the values of peak inspiratory and expiratory flows, which were consistently higher with the devices than with the reference. The results of WOB expressed as J/L can be found in the supplementary materials (see related supplementary materials at http://www.rcjournal.com).
Cost Estimates
Setting to one the overall cost of the reference device, the cost amounted to 1.17, 10.70, 1.46, 6.10, 11.80, and 6.30 for Airvo 2, G5, HM80, T60, V500, and V60 Plus, respectively.
Discussion
The main findings of the present research, which is the first bench study to compare the effect of HFNC devices on WOB, can be summarized as follows: (1) the total WOB was higher with other HFNC devices compared to the Optiflow, (2) the differences were very small and may not be clinically relevant, and (3) the PEEPs generated by the devices were lower with these HFNC devices compared to the Optiflow.
In patients with acute hypoxemic respiratory failure, HFNC reduces patient respiratory effort and WOB compared to conventional oxygen therapy devices or no supplemental oxygen. By contrast, the bench setup, by nature, does not allow any interaction between simulated patient effort and oxygen flow delivery, (ie, no change in patient effort in response to higher oxygen flow compared to baseline). Therefore, after having checked that baseline conditions were equal before each HFNC device was tested, the comparisons between devices were performed at specific flows. Furthermore, with increasing simulated inspiratory effort, WOB increased.
It is reasonable to compare newer HFNC delivery devices to the first HFNC device developed, the Optiflow. To appraise WOB results in our study, we compared our results to data reported in the literature. For example, Delorme et al9 found WOB was almost 4.5 J/min prior to HFNC use. This level of WOB was almost reached in the present study at 40 L/min for a medium simulated effort (Fig. 3). In the study by Delorme et al,9 the total WOB went down to 3.5 and 2.0 J/min at 40 and 60 L/min HFNC, respectively.
Total WOB was similar between devices, including the reference, which implies that HFNC devices are interchangeable, and other features such as price, availability, and ease of use may take precedence. HFNC has been used in various patient populations, and indications have expanded with the pandemic.14 Our results suggest that the choice of HFNC delivery device will not adversely affect clinical outcomes, including WOB. Enrolling centers with various HFNC delivery devices should not affect research results. It should be noted, however, that the decreased elastic WOB resulting from increased compliance with higher flows in subjects9 can be offset by the higher elastic WOB with some devices according to present result (Fig. 5).
Of note, our study found between-device differences in generated PEEP. PEEP was lower with more recent HFNC delivery devices compared to the reference. These differences were small, < 2 cm H2O, but suggest that the regulation of flow delivery was varied between devices, with greater values provided by more recent HFNC equipment. However, HFNC generation of PEEP is one mechanism by which the modality improves oxygenation through increases in end-expiratory lung volume.15 Assuming a lung compliance of 60 mL/cm H2O, we would expect a 60 mL change in lung volume for each 1 cm H2O rise in PEEP. This value looks negligible. However, it is the mean increase in dependent or non-dependent end-expiratory lung volume at HFNC 45 L/min and hence half of the overall increase in end-expiratory lung volume, assessed by lung electrical impedance tomography in patients.11 Therefore, varying PEEP levels generated between HFNC devices may be clinically relevant regarding the change in oxygenation it may promote and warrants further investigations in patients.
Limitations and Strengths
In addition to the intrinsic limitation of a bench study to assess the relationship between inspiratory effort and WOB mentioned above, the present study did not investigate the difference in devices in terms of FIO2 and humidification performance. Another issue not herein covered is the risk of environmental contamination with high oxygen flow. This risk was raised at the onset of COVID-19 pandemic and may have contributed to an early intubation strategy during the first wave. Subsequent studies on the topic found that the risk of aerosol dispersion was limited.16-19 Finally, we tested an adult configuration, which differs from children and neonates. Our study is strengthened by rigorous and objective device evaluation under controlled conditions and assessment and stability of baseline conditions prior to HFNC testing.
Conclusions
As compared to Optiflow, the most recent HFNC devices were associated with negligible increased in WOB. These differences are most likely clinically insignificant. The PEEPs generated by the HFNC devices were lower than the reference device, which could have clinical relevance.
Acknowledgments
The authors would like to thank F. Charbon, biomedical engineer in Edouard Herriot Hospital, Lyon, France, for facilitating the present investigations in the dedicated room.
Footnotes
- Correspondence: Claude Guérin MD PhD, Médecine Intensive Réanimation, 5 Place d’Arsonval 69003 Lyon, France. E-mail: claude.guerin{at}chu-lyon.fr
The authors have disclosed no conflicts of interest.
Supplementary material related to this paper is available at http://www.rcjournal.com.
- Copyright © 2022 by Daedalus Enterprises