Comparative evaluation of three total full-face masks for delivering Non-Invasive Positive Pressure Ventilation (NPPV): a bench study
BMC Pulmonary Medicine volume 23, Article number: 189 (2023)
Historically, the oro-nasal mask has been the preferred interface to deliver Non-Invasive Positive Pressure Ventilation (NPPV) in critically ill patients. To overcome the problems related to air leaks and discomfort, Total Full-face masks have been designed. No study has comparatively evaluated the performance of the total Full-face masks available.The aim of this bench study was to evaluate the influence of three largely diffuse models of total Full -face masks on patient-ventilator synchrony and performance during pressure support ventilation. NPPV was applied to a mannequin, connected to an active test lung through three largely diffuse Full-face masks: Dimar Full-face mask (DFFM), Performax Full-face mask (RFFM) and Pulmodyne Full-face mask (PFFM).The performance analysis showed that the ΔPtrigger was significantly lower with PFFM (p < 0.05) at 20 breaths/min (RRsim) at both pressure support (iPS) levels applied, while, at RRsim 30, DFFM had the longest ΔPtrigger compared to the other 2 total full face masks (p < 0.05). At all ventilator settings, the PTP200 was significantly shorter with DFFM than with the other two total full-face masks (p < 0.05). In terms of PTP500 ideal index (%), we did not observe significant differences between the interfaces tested.
The PFFM demonstrated the best performance and synchrony at low respiratory rates, but when the respiratory rate increased, no difference between all tested total full-face masks was reported.
The main advantages of NPPV are: 1) avoiding the side effects and complications related to endotracheal intubation , 2) improving patient comfort and 3) preserving airway defense mechanisms. An important issue for NPPV success is the choice of an appropriate interface.
Currently, different types of interfaces for NPPV are available, including nasal masks, oro-nasal mask, total Full-face masks and helmets [9, 10]. The oro-nasal mask has historically been the preferred interface to deliver NPPV in critically ill patients [11, 12]. Despite an appropriate clinical indication, a relatively high percentage of patients fail NPPV delivered by oro-nasal mask due to air leaks, discomfort and claustrophobia [13, 14]. To overcome these problems, total Full-face masks have been designed in order to increase patient comfort by maintaining a tight seal around the face. In fact, these interfaces create an air seal trough a silicon gasket around the whole perimeter of the face , thus eliminating the discomfort generated by the application of pressure over the nasal bridge.
To our best knowledge, while several studies analyzed the performance of different models of helmets for NPPV [16, 17], no study has comparatively evaluated the performance of the total Full-face masks available.
The aim of this bench study was to evaluate the influence of three largely diffuse models of total full-face masks connected through a standard Y-piece to a double circuit on patient-ventilator synchrony and performance during pressure support ventilation with different breathing frequencies, times of pressurization, and cycling-off flow thresholds.
Material and methods
The study was performed at the Respiratory Mechanics Laboratory (Ventil@b) of the Fondazione Policlinico Universitario A. Gemelli IRCCS, Università Cattolica del Sacro Cuore in Rome, Italy.
NPPV was applied to a mannequin (Laerdal Medical AS, Stavanger, Norway), connected to an active test lung (ASL 5000; Ingmar Medical, Pittsburgh, PA, USA), through three largely diffuse total Full-face masks connected through a standard double circuit with a Y-piece: Dimar total Full-face mask (DFFM) (Mirandola, Italy), Performax total Full-face mask (RFFM) (Philips, Respironics, Murrysville, PA, USA) and Pulmodyne total Full-face mask (PFFM) (BiTrac MaxShield, Indianapolis, IN, USA). During the study, we tested the small adult size of all interfaces.
The interfaces tested are characterized by a total facial perimeter cushion made with soft silicone and an adjustable nuchal fixing system. The three models of interfaces tested presented slightly different inner volume: DFFM and PFFM have an inner volume of 660 ml, while RFFM has an inner volume of 700 ml.
Total Full-face mask NPPV was delivered, at simulator respiratory rates (RRsim) of 20 and 30/min, with a mechanical ventilator with NIV module (Puritan Bennet 840; Covidien Health Care, Mansfield, MA) set at two different inspiratory pressure support levels (iPS) 10 and 15 cmH2O, with Positive End-Expiratory Pressure (PEEP) 8 cmH2O. During the bench test, the ventilator was set to optimize patient-ventilator interaction by choosing the fastest pressurization ramp (100%), two values of expiratory cycling off (25% of peak inspiratory flow, slow setting and 50% of peak inspiratory flow, fast setting), checking for the absence of premature mechanical inspiratory termination, and a flow trigger set at the lowest value thus avoiding autotrigger phenomena.
Each test condition lasted for 20 min, recording continuously the last 5 min.
The ASL 5000 is a digitally controlled real-time breathing simulator, that allows the creation of various types of breaths as during spontaneous ventilation. A broad range of flow (from 1 to 180 l/min, with a rise time < 50 ms), tidal volume (from 2 to 2500 ml), respiratory rate (from 0 to 150 breaths/min), resistance (between 3 and 500 cmH2O/l/s), compliance (from 0.5 to 250 ml/cmH2O) and inspiratory muscle effort (from 0 to 100 cmH2O) can be reproduced.
For the purposes of this study, the ASL 5000 active test lung system was set using a single-compartment model, an active inspiration simulated by a semi-sinusoidal pressure waveform (Rise Time 15%, Pause 0% and Release Hold 25%). The lung simulator was set to mimic an adult patient of 70 kg Body weigh with mild restrictive acute respiratory failure, breathing with an inspiratory effort (Pmus) of 8 cmH2O, a respiratory system compliance of 40 ml/cmH2O and a respiratory resistance of 4 cmH2O/L/sec.
The airflow delivered by the ventilator to the total face mask (V’) during the inspiratory phase was measured with a pneumotachograph (Fleisch n.2; Metabo, Epalinges, Switzerland) positioned at the Y-connection of the ventilator circuit. The airway pressure (Paw) of the inspiratory limb of the circuit was measured by a pressure transducer with a differential pressure of ± 100 cmH2O (Digima Clic-1; KleisTEK, ICU-Lab System, Italy), placed distally to the pneumotachograph. All the signals were acquired, amplified, filtered, digitized at 100 Hz, recorded on a dedicated personal computer, and analyzed with a specific software (ICU lab 2.3; KleisTEK Advanced Electronic System, Italy and Analysis Plus; Novametrix Medical System, Wallingford, CT, USA). Ventilator inspiratory and expiratory time (Timec and Temec, respectively), and ventilator rate of cycling (RRmec) were all determined on the Flow (V’) tracing. The inspiratory duty cycle (Ti/Ttot) was calculated as the ratio between Timec and the sum of Timec and Temec (Ttot). Airflow (V’) and tidal volume (Vt) delivered to the simulator, airway opening pressure (Paw), and Pmus were displayed online on the computer screen. The signals obtained with the ASL were transmitted to a PC host via 10/100 MBit Ethernet, sampled, and processed in real time. by means of a specific software (Lab View; Ingmar Medical, Pittsburgh, PA, USA). The amount of tidal volume delivered to the simulator during its active inspiration (i.e., while Pmus is negative) (i.e., the simulator tidal volume, VTsim) was calculated as the volume generated from the onset of Pmus negative deflection to its return to baseline [17,18,19,20].
Trigger pressure drop (∆Ptrigger), defined as the pressure drop generated in triggering the ventilator.
Inspiratory pressure–time product (PTPt), defined as the area under Paw between the onset of inspiratory effort and of mechanical assistance.
Pressure time product 200 ms from the onset of the ventilator pressurization (PTP200), as an index of pure pressurization performance.
Pressure–time product at 300 and 500 ms (PTP 300 and PTP 500), variables defining the speediness of pressurization and the ventilator capacity to maintain the set pressure.
PTP 500 ideal index, expressed the percentage of ideal PTP, which is unattainable because it would imply a trigger pressure drop and an instantaneous pressurization of the ventilator.
Pressurization time (Timepress), defined as the time necessary to achieve the preset level of pressure support.
Inspiratory trigger delay (Delaytrinsp), calculated as the time lag between the onset of Pmus negative swing and the start of the ventilator support (i.e., Paw positive deflection).
Expiratory trigger delay (Delaytrexp), as assessed as the delay between the offset of the inspiratory effort and the offset of the mechanical insufflation.
Time of synchrony (Timesync), defined as the time during which inspiratory muscle effort and Paw are in phase.
Simulator VT/mechanical VT, intended as the percentage of Vt delivered during inspiratory muscle effort negative deflection.
The time during which simulator respiratory effort and ventilator assistance were synchronous, indexed to the simulator inspiratory time (Timesync/Tineu) was also computed.
The number of wasted (ineffective) efforts, defined as ineffective inspiratory efforts, not assisted by the ventilator.
Numbers of Autotrigger phenomena, namely a mechanical insufflation in absence of inspiratory effort.
Continuous data were expressed as mean and standard deviation (SD). Categorical data were presented as numbers and percentage in brackets. All variables were compared with each interface used. Comparisons were made by Student's t- test and Chi test, as appropriate. The analysis of variance (ANOVA) for repeated measures was used to detect significant differences between the different experimental conditions. When significant differences were detected, a post hoc analysis was performed using the Bonferroni test; p values < 0.05 were considered statistically significant. Statistical analysis was performed using MEDcalc version 18.6.
The analysis of the patient-ventilator interaction showed that, independently from the ventilator setting applied, at 20 and 30 breaths/min and PS 10 cmH2O, no significant difference was observed between the RFFM and the DFFM in terms of Delaytrinsp, while the PFFM showed significantly reduced values of this variable (p < 0.01) (Fig. 1).
The Pressurization Time (Timepress) showed that, at 20 breaths/min and iPS 10, PFFM presented a significantly faster Timepress compared to the other 2 masks (p < 0.05), while at 30 breaths/min with the same iPS values DFFM showed the shorter value of Timepress compared to the other two total face tested. In terms of Delaytrexp, at the study conditions above mentioned, RFFM presented the shorter Delaytrexp compared to the other two Full-face masks (p < 0.01), while with the fast setting and at 20 breaths/min this Full- face mask showed a premature termination of the mechanical insufflation (Fig. 1).
At iPS 15 cmH2O and 20 breaths/min, Delaytrinsp was shorter with PFFM compared both to DFFM and RFFM(p < 0.05), while at 30 breaths/min with the fast-setting, DFFM showed a significant prolongation of Delaytrinsp compared to the other two masks (p < 0.01) (Fig. 2) while with the slow setting no differences were observed between the three Full-face masks tested.
The Timepress analysis showed that, at both RRsim tested and independently from the setting applied, DFFM had a shorter Timepress compared to the other two Full-face masks tested (p < 0.05), while the Delaytrexp was significantly longer with this mask in almost all the study conditions (p < 0.01) (Fig. 2).
The Timesync analysis showed that, at both iPS levels tested, at 30 breaths/min no significant difference was found between the three interfaces, while, at 20 breaths/min and at both iPS levels applied, a significant prolongation of the Timesync with PFFM and DFFM compared to the RFFM (p < 0.05) was observed, even though PFFM showed, only at RR 20, the longer Timesync compared to the other two masks (p < 0.05) (Figs. 3 and 4).
At all ventilator settings, the RFFM delivered the lower value of mechanical VT in comparison with PFFM and DFFM (p < 0.05), while no difference was reported in terms of mechanical VT between PFFM and DFFM. Similar results were obtained in terms of simulator VT/mechanical VT. In fact, during all bench study conditions, DFFM showed the lower value of simulator VT/mechanical VT compared to the other full-face masks.
The three total Full-face masks tested did not present significant differences in terms of Timesynch/Tineu during the whole bench study.
The performance analysis showed that, independently from the setting, the ∆Ptrigger was significantly lower with PFFM (p < 0.05) at 20 RRsim at both iPS levels applied, while, at RRsim 30, DFFM had the longest ∆Ptrigger compared to the other 2 total full-face masks (p < 0.05) (Tables 1 and 2).
The PTPt analysis showed that, at RRsim 20, PFFM had a smaller PTPt than the other 2 total full-face masks at iPS 10, independently from the setting used, and at iPS 15 only during the fast setting. Finally, at RRsim 30 in all conditions tested, RFFM showed a shorter PTPt compared to the other two total Full-face masks (Tables 1 and 2).
At all ventilator settings, the PTP200 value was significantly shorter with DFFM than with the other two total full-face masks (p < 0.05). In terms of PTP 300, we observed that, at RRsim 20 and fast setting, PFFM presented the significantly longest value compared to the others two masks at both iPS and PEEP tested, while at RRsim 20 with slow setting the DFFM presented the significantly longest value (Tables 1 and 2).
At RRsim 30, in all conditions tested, the PFFM presented the shorter value of PTP 300 compared to the other total full-face masks (p < 0.05).
The results of this bench study show that the physical characteristics and the design of the total Full-face mask may influence patient–ventilator interaction during NPPV.
NPPV delivered through a total full-face mask can be a valid alternative to the oro-nasal mask for the treatment of patients with respiratory failure .
Several studies described the conditions associated to NPPV failure with the use of oro-nasal mask [25,26,27,28] as intolerance, attributed to mask discomfort or poor fit, excessively tightened straps, excessive air leaks, patient-ventilator asynchrony, or claustrophobia. The total Full-face mask has been designed to overcome the drawbacks related to the oro-nasal mask use .
Several studies have been conducted to explore the efficacy of total Full-face masks in comparison with the oro-nasal mask, both in terms of comfort and clinical outcomes, expressed as rate of endotracheal intubation, clinical evolution and gas exchange [29, 30]. In a recent study, Sadeghi and coll  compared the total full-face mask and the oro-nasal mask in terms of effectiveness and comfort in patients with acute respiratory failure treated with NPPV. During this study, the authors enrolled 48 patients with acute respiratory failure treated with NPPV (24 patients who applied the total full-face mask and 24 patients who applied the oro-nasal mask). The authors demonstrated the non-inferiority of total full-face mask in comparison with the oro-nasal mask in terms of clinical outcome and reported a lower score for cheeks pain in patients receiving NPPV via full face mask.
In the last years, several bench studies [16,17,18, 23] were conducted to evaluate the impact of interface physical characteristics and the role of the optimal choice of the circuit on interface performance and patient-ventilator interaction.
Moreover, in a recent study, Ferrone et al. demonstrated that the presence of 2 different connectors for inflow and outflow gases improves the interaction and performance of the only total full-face mask equipped with this kind of circuit present in the market . However, the aim of the present study was to perform a head-to-head comparison of three total full-face masks largely diffuse in the market and connected through a standard Y-piece to a double circuit.
To date, although different models of total face masks are available in Europe for clinical use, to our best knowledge, no study has evaluated the impact of their different physical characteristics on patient–ventilator interaction during NPPV.
In our study, despite a better patient-ventilator interaction with the PFFM in terms of Delaytrinsp and Timepress during iPS 10 and PEEP 8 cmH2O, with increased values of iPS also DFFM showed a better interaction respect to RFFM. The Time of Synchrony analysis showed that at high RR (30 breath/min) no significant differences were found between the three interfaces, while, at RR 20, PFFM showed a longer Timesync compared to the other two total full-face masks. However, it is worth to underline that the three interfaces were all able, at lower RR, to guarantee a Timesync above 400 ms, and thus to assist more than 50% of the duration of the simulator breath.
Concerning the performance, our results showed that PFFM presented the better performance at low RRsim, as demonstrated by the lower ∆Ptrigger and PTPt, while increasing the RRsim during iPS 10 the PFFM and RFFM showed a similar performance. At iPS 15 and high RRsim, the RFFM presented the better performance compared to the other two total full-face masks. Furthermore, RFFM showed the longest PTP 200 compared to the other two total full-face masks.
During this bench study, the DFFM presented a reduced performance increasing iPS and RRsim respect to the other interfaces tested, showing the longest values of ∆Ptrigger and PTPt.
These results can be partially explained by considering the different shape and material of the DFFM compared to the other two masks. In particular, the softer material of the flange, created to increase patient comfort, causes an initial pressure dissipation during the mechanical pressurization, with a consequently longer Delaytrinsp, ∆Ptrigger and PTPt, while, once the interface is well pressurized, the time to reach the preset level of iPS is faster with DFFM than with the other 2 masks .
This study has an important limitation, being a bench physiological study, and our results need confirmation in a clinical study performed in critically ill patients. In this bench study, we did not test a vented circuit configuration, as our aim was to evaluate patient ventilator interaction and performance of total full-face masks in an ICU settings, where this specific kind of circuit is widely used, while vented circuits are generally used outside the ICU in chronic respiratory failure settings and/or in stepdown units. Moreover, in our study, we did not perform a comparison between these total full-face masks with a standard circuit with the only model present on the market equipped with embedded double inflow and outflow circuits, that is probably more performant  but is very recent and not largely diffuse.
In conclusion, the results of this comparative bench study suggest that the physical characteristics and different design of the total Full-face masks can influence patient–ventilator interaction and performance. Moreover, different type of total face mask may perform diversely in delivering NPPV with various ventilator settings and breathing frequencies. The PFFM demonstrated a better performance and assistance at low respiratory rate compared to the others full-face masks tested, but when the respiratory rate increased all full-face masks showed a similar behaviour. This aspect should be considered when choosing a total Full-face mask for clinical use.
Availability of data and materials
The data generated or analyzed during current study are available from the corresponding author on reasonable request.
Ambrosino N, Vagheggini G. Noninvasive positive pressure ventilation in the acute care setting: where are we? Eur Respir J. 2008;31(4):874–86.
Nava S, Hill N. Non-invasive ventilation in acute respiratory failure. Lancet. 2009;374(9685):250–9.
Rochwerg B, Brochard L, Elliott MW, et al. Official ERS/ATS clinical practice guidelines: noninvasive ventilation for acute respiratory failure. Eur Respir J. 2017;50(2):1602426.
Esteban A, Frutos-Vivar F, Muriel A, et al. Evolution of mortality over time in patients receiving mechanical ventilation. Am J Respir Crit Care Med. 2013;188(2):220–30.
Nava S, Navalesi P, Carlucci A. Non-invasive ventilation. Minerva Anestesiol. 2009;75(1–2):31–6.
Official ERS/ATS clinical practice guidelines: noninvasive ventilation for acute respiratory failure. Rochwerg B, Brochard L, Elliott MW, Hess D, Hill NS, Nava S, Navalesi P Members Of The Steering Committee, Antonelli M, Brozek J, Conti G, Ferrer M, Guntupalli K, Jaber S, Keenan S, Mancebo J, Mehta S, Raoof S Members Of The Task Force.Eur Respir J. 2017;50(2):1602426. https://doi.org/10.1183/13993003.02426-2016. Print 2017 Aug.
Munshi L, Mancebo J, Brochard LJ. Noninvasive respiratory support for adults with acute respiratory failure. N Engl J Med. 2022;387(18):1688–98. https://doi.org/10.1056/NEJMra2204556.
De Pascale G, Antonelli M. Noninvasive ventilation: practical advice. Curr Opin Crit Care. 2013;19(1):1–8. https://doi.org/10.1097/MCC.0b013e32835c34a5.
Pingleton SK. Complications of acute respiratory failure. Am Rev Respir Dis. 1988;137:1463–93.
Antonelli M, Conti G, Bello G. Non-invasive ventilatory support modes. In: Civetta, Taylor, and Kirby’s Critical Care. 4th ed. In: Andrea Gabrielli, A. Joseph L, Mihae Y, editors. Philadelphia: Lippincott Williams & Wilkins Publishers; 2009. p. 1939–1957.
Kwok H, McCormack J, Cece R, et al. Controlled trial of oronasal versus nasal mask ventilation in the treatment of acute respiratory failure. Critical Care Med. 2003;31(2):468–73.
Navalesi P, Fanfulla F, Frigerio P, et al. Physiologic evaluation of noninvasive mechanical ventilation delivered with three types of masks in patients with chronic hypercapnic respiratory failure. Critical Care Med. 2000;28(6):1785–90.
Karnik AM. Noninvasive positive pressure ventilation: testing the bridge. Chest. 2000;117(3):625–7.
Antonelli M, Conti G, Moro M, et al. Predictors of failure of noninvasive positive pressure ventilation in patients with acute hypoxemic respiratory failure: a multi-center study. Intensive Care Med. 2001;27(11):1718–28.
Criner GJ, Travaline JM, Brennan KJ, et al. Efficacy of a new full face mask for noninvasive positive pressure ventilation. Chest. 1994;106(4):1109–15.
Costa R, Navalesi P, Spinazzola G, et al. Comparative evaluation of different helmets on patient–ventilator interaction during noninvasive ventilation. Intensive Care Med. 2008;34:1102–8. https://doi.org/10.1007/s00134-008-1041-8.
Ferrone G, Cipriani F, Spinazzola G, et al. A bench study of 2 ventilator circuits during helmet noninvasive ventilation. Respir Care. 2013;58(9):1474–81.
Olivieri C, Costa R, Spinazzola G, et al. Bench comparative evaluation of a new generation and standard helmet for delivering non-invasive ventilation. Intensive Care Med. 2013;39:734–8. https://doi.org/10.1007/s00134-012-2765-z.
Vignaux L, Vargas F, Roeseler J, et al. Patient–ventilator asynchrony during non-invasive ventilation for acute respiratory failure: a multicenter study. Intensive Care Med. 2009;35(840):21. https://doi.org/10.1007/s00134-009-1416-5.
Conti G, Gregoretti C, Spinazzola G, et al. Influence of different interfaces on synchrony during pressure support ventilation in a pediatric setting: a bench study. Respir Care. 2015;60(4):498–507. https://doi.org/10.4187/respcare.03309.
Vignaux L, Piquilloud L, Tourneux P, et al. Neonatal and adult ICU ventilators to provide ventilation in neonates, infants, and children: a bench model study. Respir Care. 2014;59(1463–75):22. https://doi.org/10.4187/respcare.02540.
Cammarota G, Longhini F, Perucca R, et al. New setting of neurally adjusted ventilatory assist during noninvasive ventilation through a helmet. Anesthesiology. 2016;125(6):1181–9. https://doi.org/10.1097/ALN.0000000000001354.
Olivieri C, Longhini F, Cena T, et al. New versus conventional helmet for delivering noninvasive ventilation: a physiologic, crossover randomized study in critically ill patients. Anesthesiology. 2016;124(1):101–8. https://doi.org/10.1097/ALN.0000000000000910.
Belchior I, Gonçalves MR, Winck JC. Continuous noninvasive ventilation delivered by a novel total face mask: a case series report. Respiratory Care. 2012;57(3):449–53. https://doi.org/10.4187/respcare.01275.
Soo Hoo GW, Santiago S, Williams AJ. Nasal mechanical ventilation for hypercapnic respiratory failure in chronic obstructive pulmonary disease: determinants of success and failure. Crit Care Med. 1994;22(8):1253–61.
Hill NS, Brennan J, Garpestad E, Nava S. Noninvasive ventilation in acute respiratory failure. Crit Care Med. 2007;35(10):2402–7.
Meduri GU, Abou-Shala N, Fox RC, et al. Noninvasive face mask mechanical ventilation in patients with acute hypercapnic respiratory failure. Chest. 1991;100(2):445–54.
Pennock BE, Kaplan PD, Carlin BW, et al. Pressure support ventilation with a simplify and ventilatory support system administered with a nasal mask in patients with respiratory failure. Chest. 1991;100(5):1371–6.
Chacur FH, Vilella Felipe LM, et al. The total face mask is more comfortable than the Oronasal mask in noninvasive ventilation but is not associated with improved outcome. Respiration. 2011;82:426–30. https://doi.org/10.1159/000324441.
Ozsancak A, Sidhom SS, Liesching TN, et al. Evaluation of the total face mask for noninvasive ventilation to treat acute respiratory failure. Chest. 2011;139:5.
Sadeghi S, Fakharian A, Nasri P, et al. Comparison of comfort and effectiveness of total face mask and oronasal mask in noninvasive positive pressure ventilation in patients with acute respiratory failure: a clinical trial. Canadian Respiratory J. 2017;2017(6):2048032. https://doi.org/10.1155/2017/2048032.
Ferrone G, Spinazzola G, Costa R, et al. Influence of total face masks design and circuit on synchrony and performance during pressure support ventilation: a bench study. Respir Med Res. 2022;82:100963.
Scala R, Accurso G, Ippolito M, et al. Material and technology: back to the future for the choice of interface for non-invasive ventilation – a concise review. Respiration. 2020;99:800–17.
No funding was used during the study.
Ethics approval and consent to participate
Consent for publication
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Spinazzola, G., Ferrone, G., Costa, R. et al. Comparative evaluation of three total full-face masks for delivering Non-Invasive Positive Pressure Ventilation (NPPV): a bench study. BMC Pulm Med 23, 189 (2023). https://doi.org/10.1186/s12890-023-02489-2