Li et al. Critical Care (2023) 27:78 Critical Care https://doi.org/10.1186/s13054-023-04361-5 REVIEW Open Access The effects of flow settings during high‑flow nasal cannula support for adult subjects: a systematic review Jie Li1*†, Fai A. Albuainain1,2†, Wei Tan3, J. Brady Scott1, Oriol Roca4,5,6 and Tommaso Mauri7,8 Abstract Background During high-flow nasal cannula (HFNC) therapy, flow plays a crucial role in the physiological effects. However, there is no consensus on the initial flow settings and subsequent titration. Thus, we aimed to systematically synthesize the effects of flows during HFNC treatment. Methods In this systematic review, two investigators independently searched PubMed, Embase, Web of Science, Scopus, and Cochrane for in vitro and in vivo studies investigating the effects of flows in HFNC treatment published in English before July 10, 2022. We excluded studies that investigated the pediatric population (< 18 years) or used only one flow. Two investigators independently extracted the data and assessed the risk of bias. The study protocol was prospectively registered with PROSPERO, CRD42022345419. Results In total, 32,543 studies were identified, and 44 were included. In vitro studies evaluated the effects of flow settings on the fraction of inspired oxygen ­(FIO2), positive end-expiratory pressure, and carbon dioxide ­(CO2) wash- out. These effects are flow-dependent and are maximized when the flow exceeds the patient peak inspiratory flow, which varies between patients and disease conditions. In vivo studies report that higher flows result in improved oxygenation and dead space washout and can reduce work of breathing. Higher flows also lead to alveolar overdis- tention in non-dependent lung regions and patient discomfort. The impact of flows on different patients is largely heterogeneous. Interpretation Individualizing flow settings during HFNC treatment is necessary, and titrating flow based on clinical findings like oxygenation, respiratory rates, ROX index, and patient comfort is a pragmatic way forward. Keywords High-flow nasal cannula, Oxygen therapy, Flow settings, Peak inspiratory flow, Oxygenation, Ventilation distribution, Patient self-inflicted lung injury † 5 Jie Li and Fai A. Albuainain are contributed equally Centro de Investigacion Biomedica en Red de Enfermedades Respiratorias (CIBERES), Instituto de Salud Carlos III, Madrid, Spain *Correspondence: 6 Departament de Medicina, Universitat Autònoma de Barcelona, Jie Li Bellaterra, Spain Jie_Li@rush.edu 7 1 Department of Anesthesia, Critical Care and Emergency, Fondazione Division of Respiratory Care, Department of Cardiopulmonary Sciences, IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy Rush University, 600 S Paulina St, Suite 765, Chicago, IL 60612, USA 8 2 Department of Pathophysiology and Transplantation, University Department of Respiratory Care, Imam Abdulrahman Bin Faisal of Milan, Milan, Italy University, Dammam, Saudi Arabia 3 Department of Respiratory and Critical Care Medicine, The First Affiliated Hospital of China Medical University, Shenyang, China 4 Servei de Medicina Intensiva, Parc Taulí Hospital Universitari, Sabadell, Spain © The Author(s) 2023. 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. Li et al. Critical Care (2023) 27:78 Page 2 of 20 Introduction RCTs. The Newcastle–Ottawa Scale was used to assess The use of high-flow nasal cannula (HFNC) in critical non-randomized trials. Any disagreement regarding care areas has increased over the past few years, par- study selection, data extraction, or quality assessments ticularly during the COVID-19 pandemic [1]. HFNC was resolved by a consensus discussion with the third has been shown to effectively reduce intubation rates for investigator. patients with acute hypoxemic respiratory failure (AHRF) A total of 32,543 studies were identified, and 32,395 [2] and prevent post-extubation respiratory failure [3]. It studies were excluded for reasons displayed in Fig. 1. One may also be non-inferior to noninvasive ventilation to hundred and forty-eight full-text articles were assessed prevent reintubation for patients with chronic obstruc- for eligibility, and 44 studies were finally included, of tive pulmonary disease (COPD) [4, 5]. Improved patient which 11 were in vitro studies [26, 28–37], 2 combined outcomes associated with HFNC are due to its physiolog- in vitro and in vivo studies to investigate patients with ical effects, such as improvement in oxygenation [6–11], AHRF [6, 14], 13 studies investigated healthy individu- efficiency of ventilation [6–8, 11–24], reduction of work als [13, 15–20, 38–43], 9 investigated patients with AHRF of breathing (WOB) [7, 11, 12, 24], avoidance of patient [7–10, 27, 44–46, 50], 5 studies examined patients with self-inflicted lung injury, and improvement in patient COPD [11, 12, 21–23], 1 study investigated both AHRF comfort and tolerance [25]. HFNC washes out upper and COPD patients [24], and 3 investigated patients airway dead space, and its effects are maximized when during procedural sedation [47–49]. Among the in vivo the delivered gas flow meets or exceeds the patient peak studies, only four were randomized controlled trials [27, inspiratory flow, resulting in a stable fraction of inspired 47–49], while 18 were randomized crossover studies [7, oxygen ­(FIO2) and a level of positive end-expiratory pres- 8, 10, 12, 13, 16, 17, 22–24, 38, 40–43, 45, 46, 50]. None of sure (PEEP) [6, 26]. Thus, flow settings play a vital role the included randomized trials had incomplete outcome during HFNC oxygen therapy [27]. data reporting but 10 did not have registration [7, 8, 10, In recent years, significant efforts have been made 13, 41, 43, 46, 48–50]. All of them had a clear description to investigate the effects of flow settings during HFNC of random sequence generation, but only five explained therapy for various patient populations. However, no the study allocation concealment [13, 16, 22, 27, 48]. consensus has been reached on the most effective initial Due to the nature of HFNC flows in conscious patients, flow setting and its subsequent titration. Therefore, we blinding participants and/or the treating clinicians was systematically reviewed the available evidence regarding not possible. No obvious publication bias was observed the physiological and clinical effects of different flow set- among the randomized controlled or crossover trials tings during HFNC therapy for adult subjects, aiming to (Additional file 1: Figs. S1, appendix p2) and non-rand- provide evidence-based guidance on optimal HFNC flow omized trials (Additional file 1: Table S1, appendix p3). settings for various clinical conditions. Subject peak inspiratory flow during tidal Literature search strategy and results breathing A literature search was conducted independently by two HFNC aims to provide a flow that meets or exceeds investigators in PubMed, Embase, Web of Science, Sco- the patient peak tidal inspiratory flow (PTIF) [51]. Sev- pus, and Cochrane for articles published before July 10, eral studies have reported PTIF in different popula- 2022, using the following keywords: (“high-flow nasal tions (Additional file 1: Table S2, appendix p4) [6, 13, 44, cannul*” OR “high flow cannul*” OR “high flow oxygen 52–54]. therapy” OR “high flow oxygen” OR “high flow therapy” OR “HFNC” OR “nasal high flow” OR “NHF”) AND Healthy individuals (“flow”) AND (“adult”). The search was limited to papers Healthy adult volunteer PTIF has been reported in two published in English. Original studies investigating studies [13, 52]. Ritchie et al. [13] reported PTIFs while more than one HFNC flow setting were included. Stud- study participants were at rest and during exercise. PTIF ies that only included pediatric populations, used only during exercise was higher (119.9 ± 20.0 vs 27.9 ± 9.2 L/ one flow during HFNC treatment, review articles, let- min). Moreover, PTIF increased as exercise intensity ters, abstracts, and editorials were excluded. Study titles increased [52]. and abstracts were initially screened, and full texts were subsequently reviewed to select studies included in this Patients with pulmonary disease review. The review protocol was prospectively registered The PTIF in adult patients with pulmonary disease, with PROSPERO, CRD42022345419. Two investigators especially for those with respiratory failure, is slightly independently extracted the data and assessed the risk of higher than PTIF in healthy individuals. Small vari- bias using the Cochrane collaboration risk of bias tool for ations in PTIF have been noted between different Li et al. Critical Care (2023) 27:78 Page 3 of 20 Fig. 1 Study flow diagram. HFNC, high-flow nasal cannula diseases (Additional file 1: Table S2, appendix p4). The In vitro evidence of flow settings during HFNC mean PTIF in patients with AHRF was reported to be treatment 34 ± 9 L/min [6]. Similar median PTIF values were Multiple bench studies have been conducted to evalu- observed in patients with stable asthma and COPD ate the effects of HFNC flows on tracheal F ­ IO2, [6, 26, [53]. In stable tracheostomized patients, the mean PTIF 28, 29] PEEP [6, 28–36], and dead space clearance [14, was 30 (27, 32) L/min when measured inside the tra- 32, 34, 35, 37]. When HFNC flow is lower than PTIF, chea [54]. Interestingly, Butt et al. [44] measured PTIF the ­FIO2 in the trachea is lower than the set F ­ IO2 (Fig. 2) in intubated patients on a mechanical ventilator with [6, 26, 28, 29]. This can be explained by air entrain- pressure support of zero and zero-PEEP. Their meas- ment occurring in the upper airway, which dilutes the ured PTIF was 60 (40, 80) L/min, significantly higher concentration of delivered oxygen [26]. When HFNC than the PTIF found in non-intubated spontaneously flow exceeds PTIF, studies have shown that a certain breathing patients. This difference might be caused by level of PEEP is indeed generated [6], with a quad- the need to overcome the resistance of the endotra- ratic correlation between HFNC flows and PEEP levels cheal tube or by a more severe respiratory condition. [33, 35]. That said, multiple factors affect PEEP level Li et al. Critical Care (2023) 27:78 Page 4 of 20 Fig. 2 Relationship between ­FIO2 and flow ratio of HFNC flow to peak inspiratory flow during tidal breathing. ­FIO2, fraction of inspired oxygen; HFNC, high-flow nasal cannula besides flow settings and include mouth status (open- lungs, where PEEP is needed most [31]. Lighter gases, vs closed-mouth breathing) (Fig. 3A, B) [31, 34], lung such as heliox [35], and smaller nasal prongs have also compliance (Fig. 3B) [31], gas type (Fig. 3C) [35], and been reported to generate lower PEEP [14, 35]. Finally, nasal prong size (Figs. 3C) [28, 30, 33, 35]. Of these, it high HFNC flows appear to clear C ­ O2 in less time, even appears that mouth status is the most impactful vari- when respiratory rates remain constant (Fig. 4) [32, 34, able on PEEP generation [31, 32]. During closed-mouth 35]. Overall, when flows are set to exceed PTIF, HFNC breathing, PEEP can be as high as 14.31 ± 1.33 ­cmH2O can produce a certain amount of PEEP, stabilize F ­ IO 2 when flows of 80 L/min are set in adults [29]. When the delivery, and wash out anatomic dead space. However, mouth is open, however, the PEEP level drops to almost mouth open/close status, lung compliance, gas type, zero [31]. Additionally, PEEP levels are lower in stiffer and prong size also impact HFNC effects. Fig. 3 Effects of mouth status (open- vs closed-mouth breathing) (A, B), lung compliance (B), gas type (C), and nasal prong size (C) on PEEP levels. PEEP, positive end-expiratory pressure; HFNC, high-flow nasal cannula Li et al. Critical Care (2023) 27:78 Page 5 of 20 Fig. 4 Relationship between ­CO2 clearance and HFNC flow settings. ­CO2, carbon dioxide; HFNC, high-flow nasal cannula Different flow settings in adult subjects nasopharyngeal pressure dropped from 6.8 to 0.8 c­ mH2O During HFNC oxygen therapy, flow settings have been with HFNC set at 60 L/min [17]. Caution must be taken shown to have a significant impact on short-term clini- while using very high flows, such as 100 L/min, as it can cal outcomes in 32 studies [6–24, 38–50] and long-term provide nasopharyngeal pressure as high as 11.9 ± 2.7 outcomes in one study [27]. ­cmH2O [18], and the tolerability is concerning. Healthy individuals End‑expiratory lung impedance Ventilation Besides the potential for significantly elevated airway Similar to in vitro studies, the physiological effects of pressure, an uneven distribution of delivered gas across HFNC are found to be flow-dependent (Table 1) [13, lung regions from different HFNC flows may also pose a 15–20, 38–40]. As HFNC flows increase from 0 to 40 L/ risk for regional overdistension. Three studies evaluat- min, tidal volume (Vt) increases [15–17, 38] and respira- ing ventilation distribution across lung regions from vari- tory rate (RR) decreases [13, 15–20]. However, Okuda ous HFNC flows using electrical impedance tomography et al. reported no change in minute ventilation between (EIT) report that global end-expiratory lung impedance HFNC flows of 0 and 50 L/min [15], and Parke et al. [18] (EELI) increases as HFNC flow increases [17–19]. How- reported that RR plateaued with HFNC flows > 40 L/min. ever, increases in EELI mainly occur in non-dependent regions of the lung. Plotnikow et al. reported an increase in EELI by 35% from baseline (room air) to HFNC set at 30 L/ Airway pressure min and by 22% from 30 to 50 L/min in the non-depend- The flow-dependent PEEP effect from the in vitro studies ent regions [19]. In the lung-dependent regions, the EELI has also been confirmed in healthy individuals, as end- only increased by 18% and 7.7%, respectively [19]. Since expiratory esophageal pressures or hypopharyngeal pres- the non-dependent lung regions are most likely open nor- sure gradually increase when HFNC flows are increased mally, these findings suggest a potential risk of over-dis- (Fig. 5) [13, 16–18, 20, 39]. However, maintaining a con- tending the non-dependent region, resulting in lung injury. stant PEEP with HFNC is challenging because it can sig- nificantly decrease with open-mouth breathing [13, 17, Swallow function 20, 39]. When subjects opened their mouth, hypopharyn- geal pressure dropped from 5.2 (3.5, 7.0) ­cmH2O to 1.1 Three studies have investigated the effects of flows (− 0.9, 2.4) ­cmH2O with HFNC set at 50 L/min [13], and on swallow function among healthy volunteers Li et al. Critical Care Table 1 Physiological effects of flow settings during HFNC treatment for healthy volunteers Author/year Study design N Position Breathing Flow (L/min) RR (breaths/ EELIglob VT (ml) PEEP ­(cmH2O) PTP ­(cmH2O s/ Others pattern min) (units) min) Okuda et al., Randomized 10 Supine 30° Not specified 0 15.1 ± 2.1 NA 685.6 ± 236.6 NA NA MV: 10.1 ± 3.0 2017 [15] crossover (2023) 27:78  Pes:4.2 ± 1.9 30 11.6 ± 3.1 NA 929.8 ± 434.7 NA NA MV: 10.1 ± 3.9  Pes: 6.9 ± 2.9 50 11.1 ± 2.3 NA 968.8 ± 451.1 NA NA MV: 10.1 ± 4.0  Pes: 7.48 ± 3.1 Moigne et al. Prospective 10 Supine 30° RMO 0 NA NA 526 (503, 604) NA NA NA 2021 [38] RMC NA NA 529 (480,726) NA NA NA HMO NA NA 540 (356, 663) NA NA NA HMC NA NA 417 (331, 553) NA NA NA RMO 30 NA NA 688 (629, 844) NA NA NA RMC NA NA 815 (511, 959) NA NA NA HMO NA NA 633 (538, 678) NA NA NA HMC NA NA 516 (405, 698) NA NA NA RMO 60 NA NA 753 (650, 1208) NA NA NA RMC NA NA 913 (680, 1166) NA NA NA HMO NA NA 664 (634, 1046) NA NA NA HMC NA NA 696 (351, 896) NA NA NA Plotnikow et al. Prospective 16 Supine RMC 0 16 ± 2.6 Baseline NA NA NA NA 2018 [19] Setting 45° 15 ± 3.6 1.05 (0.72, 1.34) NA NA NA NA 30 11 ± 3.6 1.12 (0.8, 2.01) NA NA NA NA 50 9 ± 2.8 1.44 (1.05, 2.16) NA NA NA NA Ritchie et al. Prospective 10 Not specified RMC 30 8.4 (4,15) b NA NA 2.5 (0.8, 4.7)b NA NA 2011 [13] b 40 6.8 (4,7) NA NA 4.0 (2.4, 7.8)b NA NA 50 7.6 (4.9, 8.0)b NA NA 5.2 (3.5, 7.0)b NA NA RMO 30 10 (6.9, 19)b NA NA 0.1 (− 0.2, 1)b NA NA 40 9.9 (5.5, 19)b NA NA 0.8 (− 0.8, 1.2)b NA NA 50 11.3 (7, 21.5)b NA NA 1.1 (− 0.9, 2.4)b NA NA Page 6 of 20 Table 1 (continued) Author/year Study design N Position Breathing Flow (L/min) RR (breaths/ EELIglob VT (ml) PEEP ­(cmH2O) PTP ­(cmH2O s/ Others pattern min) (units) min) Parke et al. 2015 Prospective 15 Not specified RMC 0 13.8 (9.8, 17.9)b Baseline NA 0.5 ± 0.3a NA NA [18] Li et al. Critical Care b a,c a 30 8.9 (5.8,11.9) 0.5 ± 0.3 NA 2.7 ± 0.7 NA NA b a,c a 40 7.7 (4.5, 11.1) 0.5 ± 0.4 NA 3.8 ± 0.8 NA NA b a,c a 50 7.3 (4.4, 10.3) 0.7 ± 0.5 NA 4.9 ± 1.1 NA NA 60 7.4 (3.8, 11.1)b NA NA NA (2023) 27:78 0.9 ± 0.6a,c 6.1 ± 1.4a 70 6.9 (3.2,10.7)b 1.0 ± 0.8a,c NA 7.6 ± 1.5a NA NA b a,c a 80 7.7 (3.6, 11.8) 1.2 ± 0.8 NA 9 ± 1.9 NA NA b a,c a 90 6.9 (3.3, 10.4) 1.4 ± 0.8 NA 10.1 ± 2.1 NA NA b a,c a 100 7.5 (2.6,12.5) 1.6 ± 0.8 NA 11.9 ± 2.7 NA NA Delorme et al. Randomized 10 Semi-recum- RMC 0 16 (15, 18) NA 337 (272, 443) NA NA Comfort: 10 (10, 2020 [16] crossover bent position 10) VA: 3,343; VD/ VT: 45 Dead space venti- lation: 2431 Dead space washout: 0 5 13 (12, 15) NA 358 (287, 458) NA NA Comfort: 9 (6, 10) VA: 3,343; VD/ VT: 39 Dead space venti- lation: 1823 Dead space washout: 863 20 10 (10, 13) NA 448 (345, 580) NA NA Comfort: 10 (9, 10) VA: 3,343; VD/ VT: 38 Dead space venti- lation: 1546 Dead space washout: 1140 60 8 (7, 10) NA 520 (470, 626) NA NA Comfort: 8 (8, 9) VA: 3,343; VD/ VT: 31 Dead space venti- lation: 1276 Dead space washout: 1160 Page 7 of 20 Li et al. Critical Care Table 1 (continued) Author/year Study design N Position Breathing Flow (L/min) RR (breaths/ EELIglob VT (ml) PEEP ­(cmH2O) PTP ­(cmH2O s/ Others pattern min) (units) min) Garofalo et al. Prospective 14 Semi-recum- RMC 0 15 (14,16) NA NA 0b NA Comfort: 10 (2023) 27:78 2019 [20] bent position (10,10 PEIP: ­0b 30 15 (14,16) NA NA 2.8 (0.8, 7.8)b NA Comfort: 10 (10,10) PEIP: 0.7 (0.3, 0.7)b 40 14 (12,14) NA NA 6 (3.2, 11)b NA Comfort: 10 (10,10) PEIP: 0.8 (0.6, 1.3)b 50 12 (12, 13) NA NA 7.6 (4.1, 12)b NA Comfort: 9 (9,10) PEIP: 1.1 (0.7, 1.5)b RMO 0 NA NA NA 0b NA PEIP: ­0b 30 NA NA NA 0.7 (0.5, 2.5)b NA PEIP: 0.1 (0, 0.4)b b 40 NA NA NA 1.7 (0.3, 5) NA PEIP: 0.3 (0, 0.3)b 50 NA NA NA 2.3 (0.7, 3.9)b NA PEIP: 0.3 (0, 0.6)b Vieira et al. 2022 Randomized 10 Not specified RMC 20 9.0 (7.3, 11.7) 90 (43, 160) 458 (364, 557) 1.2 (0.8, 1.3) 94.1 (79.3, 114.5) MV: 4.4 (3.2, 4.9 [17] crossover ­Ti:Ttot: 0.31 ± 0.05 40 7.2 (5.9, 9.9) 160 (96, 178) 447 (368, 787) 3.6 (2.8, 4.6) 112.5 (86, 127) MV: 3.9 (2.7, 4.2) Ti:Ttot: 0.26 ± 0.06 60 8.5 (5.0, 12.4) 230 (165, 344) 431 (229, 760) 6.78 (5.0, 7.9) 124.8 (72, 138) MV: 3.2 (2. 7,4.6) Ti:Ttot: 0.23 ± 0.07 RMO 20 11.6 (9.4, 14.2) 70 (15, 138) 421 (341, 517) 0.13 (0.06, 0.2) 70.0 (50.6, 103) MV: 5.2 (4.3, 6.1) Ti:Ttot: 0.33 ± 0.09 40 11.6 (8.9, 12.6) 92 (36, 151) 459 (388, 502) 0.55 (0.3, 0.7) 78.5 (66, 102) MV:5.2 (4.1–6.7) Ti:Ttot: 0.3 ± 0.07 60 12.7 (10.4,15.8) 202 (115, 233) 504 (385, 657) 0.83 (0.6, 1.2) 90.4 (70, 111) MV: 6.0 (4.8, 9.1) Ti:Ttot: 0.3 ± 0.07 Page 8 of 20 Table 1 (continued) Author/year Study design N Position Breathing Flow (L/min) RR (breaths/ EELIglob VT (ml) PEEP ­(cmH2O) PTP ­(cmH2O s/ Others Li et al. Critical Care pattern min) (units) min) Groves et al. Prospective 10 Setting position RMC 0 NA NA NA 0.8 (0.5, 1.3) NA MIP: − 1.1 2007 [39] (− 2.0, − 0.6) 10 NA NA NA 1.7 (1.2, 2.3) NA MIP: − 0.8 (2023) 27:78 (− 1.1, − 0.3) 20 NA NA NA 2.9 (2.2, 2.7) NA MIP: − 0.2 (− 0.9, 0.2) 40 NA NA NA 5.5 (4.1, 7.2) NA MIP:1.1 (− 0.1, 1.9) 60 NA NA NA 7.4 (5.4, 8.8) NA MIP: 1.6 (0.8, 2.7) RMO 0 NA NA NA 0.3 (0.3, 0.5) NA MIP: − 0.6 (− 1.1, − 0.4) 10 NA NA NA 0.7 (0.6, 0.9) NA MIP: − 0.2 (− 0.8, 0.1) 20 NA NA NA 1.4 (1.3, 1.8) NA MIP: − 0.2 (− 0.9, 0.1) 40 NA NA NA 2.2 (2.0, 2.5) NA MIP:0.1 (− 0.2, 0.4) 60 NA NA NA 2.7 (2.4, 3.1) NA MIP:0.5 (0.2, 0.7) Möller et al. Randomized 10 Not specified RMC 15 NA NA NA NA NA Clearance half- 2017 [40] crossover time in trachea: 23.73 ± 6.63 30 NA NA NA NA NA Clearance half-time in tra- chea:14.3 ± 13.43 45 NA NA NA NA NA Clearance half-time in tra- chea:10.53 ± 9.85 RR respiratory rate,  EELIglob global change in end-expiratory lung impedance measured by electrical impedance tomography, VT tidal volume, PEEP positive end-expiratory pressure measured by pharyngeal or tracheal catheter, MV minute ventilation (L/min), PTP esophageal pressure-time product per minute (­cmH2O·s/min),  Pes esophageal pressure swings ­(cmH2O), RMO regular respiratory rate breathing with mouth opened, RMC regular respiratory rate breathing with mouth closed, HMO high respiratory rate breathing with mouth opened, HMC high respiratory rate breathing with mouth closed. Ti:Ttot duty time (sec), VA alveolar ventilation (ml/ min), VD/VT percentage of dead space (%), Dead space ventilation (ml/min), Dead space washout (ml/min), Clearance half-time in trachea (sec), Comfort respiratory comfort level measured by a scale, PEIP positive end- inspiratory pressure, MIP mean inspiratory pressure ­(cmH2O); NA not available a Values are calculated b Values extracted from graph using graph reader Website c Values are reported as % of change from baseline Page 9 of 20 Li et al. Critical Care (2023) 27:78 Page 10 of 20 Fig. 5 The relationship between airway pressures and HFNC flow settings. HFNC, high-flow nasal cannula during HFNC therapy [41–43]. Sanuki et al. [41] reported while Zhang et al. [9] found no significant changes in reduced latency time of the swallow reflex with HFNC ROX index between room air and HFNC flow of 60 L/ flows being increased from 15 to 45 L/min when healthy min in patients with mild hypoxemia. Likewise, Mauri volunteers swallowed 5 mL of distilled water over 3 s. et al. reported that 30% (17/57) of AHRF patients had an Thus, they concluded that HFNC might enhance swal- unchanged or decreased ROX index when HFNC flows lowing function [41]. However, Arizono et al. reported were increased from 30 to 60 L/min. Their further anal- the opposite findings, as choking was observed when ysis revealed that the 17 patients had a higher SF ratio HFNC flows were ≥ 40 L/min in the 30 mL swallow test and ROX index at 30 L/min, compared to the other 40 [42]. Additionally, they noted that swallowing efforts patients who presented an increase in ROX index with were greater with HFNC flows ≥ 20 L/min than 10 L/min increasing flow [10]. Interestingly, the same authors [42]. Allen and Galek found a flow-dependent influence implemented a study 2 years earlier on similar patient on the duration of laryngeal vestibule closure (dLVC) populations, and they found that 30% of patients had among their healthy volunteers [43]. Since LVC is a pro- decreased PF ratios after increasing flows from 30 or 45 tective reflex that helps to prevent aspiration, the authors to 60 L/min [7]. suggest that dLVC modulation from HFNC flows might help prevent aspiration [43]. Notably, a large variation End‑expiratory lung impedance of dLVC between HFNC flows of 50 and 60 L/min was Similar to the findings in healthy individuals, increasing found, underscoring the reality that further research on flows also improves global EELI in patients with AHRF the impact of HFNC flows on swallow function is needed (Additional file 1: Table S3, appendix p5) [7–9]. Increas- [43]. Oral feeding during HFNC therapy should be closely ing HFNC flow generates a greater end-expiratory lung monitored, especially in severe hypoxemic patients who volume and PEEP [45, 46], which may cause recruitment might need treatment escalation and those with dyspha- that mainly occurs in dependent lung regions. However, gia or at high risk for aspiration. it may also generate overdistension that is more pro- nounced in non-dependent lung regions. It appears that Patients with acute hypoxemic respiratory failure changes in oxygenation, that correlate with changes in Oxygenation EELI [9], depend on the balance of alveolar recruitment During HFNC therapy for patients with AHRF, oxygena- and overdistension. tion has been assessed using ­SpO2/FIO2 (SF) ratio [6, 50], The regional distribution of the aeration depends ­PaO2/FIO2 (PF) ratio [7, 8], and the ROX index (= SF/ on HFNC flows and patients [7–9]. When flows were RR) [6, 9, 10]. Three studies have reported that oxygena- increased from 30 to 60 L/min, Mauri et al [7] reported tion improves as HFNC flows increase (Table 2) [6–8], that EELI increased, but not by a significant amount. Li et al. Critical Care Table 2 Physiological effects of flow settings during HFNC treatment for hypoxemic patients Author/year Study N Patients Position Flow (L/ FiO2 or RR  ­EELIglob PF or SF ROX PEEP PTP Others design min) ­SpO2 (breaths/ (ml) ­(cmH2O) ­(cmH2O s/ (2023) 27:78 min) min) Mauri et al. Prospective 17 AHRF Semi- 12 SpO2 24 ± 8 Baseline 151 ± 60 NA NA 254.3 (160.2, VT: 443 ± 302 2017 [7] recumbent 90–96% 359.5) MV:.1 ± 4.09 position  Pes: 9.4 (6.8,12.2) 30 20 ± 7 74 ± 174 177 ± 74 NA NA 173.5 (126.4, VT: 437 ± 314 256.4) MV: 7.0 ± 2.8  Pes: 7.9 (5.9,11.8) 45 19 ± 7 115 ± 142 187 ± 67 NA NA 168.9 (110.3, VT: 435 ± 307 217.2) MV: 7.0 ± 2.9  Pes: 8.1 (5.7, 9.5) 60 18 ± 7 230 ± 237 205 ± 61 NA NA 151.4 (111.8, VT: 429 ± 301 195.6) MV: 6.9 ± 2.1  Pes: 6.8 (5.1, 9.3) Pinkham Prospective 8 Decannu- Not speci- 0 SpO2 18 ± 5 NA NA NA 0.1 (0, 1.3) NA I:E: 0.75 ± 0.14 et al. 2022 lated from fied 93–97%c VT: 410 ± 170 [14] tracheos- MV: 6.8 ± 2.8 tomy tube 15 16 ± 3 NA NA NA 0.7 (0.2, 1.1) NA I:E:0.7 ± 0.18 VT: 420 ± 170 MV: 6.2 ± 2.2 30 14 ± 4 NA NA NA 1.0 (0.3, 3.0) NA I:E:0.6 ± 0.15 VT: 460 ± 220 MV: 6.4 ± 3.5 45 14 ± 4 NA NA NA 1.7 (0.7, 7.1) NA I:E: 0.5 ± 0.17 VT: 450 ± 210 MV: 5.7 ± 1.7 Page 11 of 20 Table 2 (continued) Author/year Study N Patients Position Flow (L/ FiO2 or RR  ­EELIglob PF or SF ROX PEEP PTP Others Li et al. Critical Care design min) ­SpO2 (breaths/ (ml) ­(cmH2O) ­(cmH2O s/ min) min) Delorme Randomized 7 AHRF Not speci- 5 ± 4 (base- SpO2 NA NA NA NA NA 152 (113, WOB: 4.1 (3.0, et al. 2017 crossover fied line) 90–94% 171) 4.7) [24] study  Pes: 8.6 (5.3, (2023) 27:78 9.9) 20 NA NA NA NA NA 127 (95, 139) WOB: 3.2 (2.1, 5.6)  Pes: 7.7 (4.9, 9.8) 40 NA NA NA NA NA 87 (74, 145) WOB: 3.1 (2.5, 4.2)  Pes: 5.7 (4.0, 8.4) 60 NA NA NA NA NA 65 (34, 99) WOB: 1.9 (1.0, 2.6)  Pes: 2.1 (1.9, 4.5) Zhang et al. Prospective 24 AHRF or high Semi- 0 FiO2 0.30 22 ± 6 Baseline NA 22.4 ± 5.4 NA NA NA 2020 [9] risk for AHRF recumbent 20 20 ± 4 7.2 (− 5.1, NA 23.9 ± 5.2 NA NA NA after extuba- position 18.7)d tion 40 20 ± 3 10.6 (− 3.4, NA 23.5 ± 4.4 NA NA NA 42)d 60 21 ± 5 26.3 NA 22.2 ± 4.3 NA NA NA (10.7,60.8)d Li et al. 2021 Prospective 49 AHRF Not speci- Matching FiO2 ≥ 0.40 23 ± 7 NA 161 ± 51 7.5 ± 3.2 NA NA NA [6] fied PTIF and ­SpO2 of 10 L/min 90–97% 23 ± 7 NA 181 ± 61 8.5 ± 4.0 NA NA NA above PTIF 20 L/min 22 ± 6 NA 201 ± 67 9.8 ± 4.7 NA NA NA above PTIF 30 L/min 23 ± 6 NA 207 ± 70 10.0 ± 4.7 NA NA NA above PTIF Basile et al. Prospective 12 AHRF Semi- 0.5 L/kg/min SpO2 20 ± 6 Baseline 194 ± 96 NA NA NA NA 2020 [8] recumbent 1.0 L/kg/min 90–96% 17 ± 5 366.7 ± ­1264c 211 ± 106 NA NA NA NA position 1.5 L/kg/min 18 ± 6 760.8 ± ­1096c 219 ± 118 NA NA NA NA Page 12 of 20 Table 2 (continued) Author/year Study N Patients Position Flow (L/ FiO2 or RR  ­EELIglob PF or SF ROX PEEP PTP Others design min) ­SpO2 (breaths/ (ml) ­(cmH2O) ­(cmH2O s/ min) min) Li et al. Critical Care Butt et al. Retrospec- 19 Post-extuba- Not speci- 20 SpO2 NA NA NA NA NA NA Comfort: 6.9 2021 [44] tive tion fied 92–97% (6.2, 7.7)ab 30 NA NA NA NA NA NA Comfort: 8.6 (8.0,9.0)ab (2023) 27:78 40 NA NA NA NA NA NA Comfort: 8.58 (8.0, 9.0)ab 50 NA NA NA NA NA NA Comfort: 6.8 (6.3, 7.2)ab 60 NA NA NA NA NA NA Comfort: 3.0 (2.3, 3.7)ab Parke et al. Prospective 15 Cardiac Not speci- 30 Not speci- NA NA NA NA 3.01 ± 1.18 NA MAP: 1.5 ± 0.6 2013 [45] surgery fied fied MPP:1.71 ± 0.73 MEP: 2.1 ± 0.83 MIP: 0.55 ± 0.38 40 NA NA NA NA 3.81 ± 1.45 NA MAP 2.2 ± 0.8 MPP:2.48 ± 0.94 MEP: 2.88 ± 1.04 MIP: 1.11 ± 0.51 50 NA NA NA NA 4.86 ± 1.79 NA MAP: 3.1 ± 1.2 MPP: 3.41 ± 1.24 MEP: 3.81 ± 1.33 MIP: 1.77 ± 0.69 Mauri et al. Retrospec- 57 AHRF Semi- 30 SpO2 NA 437 ± 314 NA 10.2 (7.2, NA NA NA 2019 [10] tive recumbent 90–96% 13.3) position 60 NA 429 ± 301 NA 11.1 (8.8, NA NA NA 13.9) Natalini et al. Prospective 5 Decannu- Not speci- 10 SpO2 NA NA NA NA Nasal 3.2 NA MEP nasal: 2.6 2019 [46] lated from fied 92–98% (2.3, 5.9)b (1.6, 5.5)b MEP tracheos- Tracheal 0.8 tracheal: 0.5 (0.3, tomy tube (0.5, 1.2)b 0.7)b 30 NA NA NA NA Nasal 3.3 NA MEP nasal: 2.6 (2.8, 5.5)b (1.9, 4.2)b Tracheal 1.3 MEP tracheal: (0.9,1.6)b 0.9 (0.6, 1.1) b 50 NA NA NA NA Nasal 5.1 NA MEP nasal: 3.9 (4.2, 7.7)b (3.1, 6)b MEP Tracheal 1.8 tracheal: 1.2 (1.4,2.2)b (1,1.5)b Page 13 of 20 Li et al. Critical Care (2023) 27:78 Table 2 (continued) Author/year Study N Patients Position Flow (L/ FiO2 or RR  ­EELIglob PF or SF ROX PEEP PTP Others design min) ­SpO2 (breaths/ (ml) ­(cmH2O) ­(cmH2O s/ min) min) Mauri et al. Prospective 40 AHRF Semi- 30 (Temp SpO2 22 (18, 29) NA 238 (192, NA NA NA Comfort: 3 (2, 5) 2018 [50] recumbent 31) 92–98% 246) position 30 (Temp 23 (19, 29) NA 240 (190, NA NA NA Comfort: 4 (2, 5) 37) 248) 60 (Temp 22 (19, 25) NA 240 (196, NA NA NA Comfort: 3 (1, 5) 31) 248) 30 (Temp 23 (19, 26) NA 240 (196, NA NA NA Comfort: 3 (2,5) 37) 248) FiO2 Fraction of inspired oxygen (%),SpO SpO2 Saturation of oxygenation using pulse oximeter (%), RR respiratory rate (breath/min),  EELIglob global change in end-expiratory lung impedance measured by electrical impedance tomography, PF ­PaO2/FiO2, SF ­SpO2/FiO2, ROX index [(SpO2/FiO2)/RR], PEEP positive end-expiratory pressure measured by pharyngeal or tracheal catheter, PTP esophageal pressure-time product per minute ­(cmH2O·s/min), VT tidal volume (ml), MV minute ventilation (L/min),  Pes esophageal pressure swings ­(cmH2O), I:E inspiratory to expiratory ratio, Comfort respiratory comfort level measured by a scale, MPP mean plateau pressure ­(cmH2O), MEP mean expiratory pressure ­(cmH2O), MIP mean inspiratory pressure ­(cmH2O), MAP mean airway pressure ­(cmH2O), Temp temperature (°C), AHRF acute hypoxemic respiratory failure, PTIF peak tidal inspiratory flow, NA not available a Values reported as 95% CI b Values extracted from graph using graph reader Website c Values reported directly from the author d Values are reported as % of change from baseline Page 14 of 20 Li et al. Critical Care (2023) 27:78 Page 15 of 20 Fig. 6 Effects of flow settings on  Pes (A), PTP (B), and WOB (C). HFNC, high-flow nasal cannula;  Pes: esophageal pressure swings ­(cmH2O); PTP: esophageal pressure–time product per minute ­(cmH2O·s/min); WOB, work of breathing (J/min) Interestingly, when compared to EELI with a facemask, and closed- vs open-mouth breathing. Regardless, close EELI in dependent lung regions significantly increased monitoring of individuals’ responses in regional lung vol- with HFNC at 60 L/min, while EELI in non-dependent umes to different flows might help avoid overdistension regions remained stable. These findings suggest more and lung injury. recruitment in dependent lung regions than overdisten- sion in non-dependent regions [7]. In a follow-up study Inspiratory efforts that included 12 patients with AHRF [8], the same group Beyond the regional distribution of volume, dynamic of authors compared the effect of different flows that were transpulmonary pressure reflects the patient inspira- set based on the patient’s predicted body weight (0.5, 1.0, tory effort and lung stress, which is associated with lung and 1.5 L/Kg/min). They utilized median flows of 35, 65, injury. Changes in esophageal pressure (Pes) are a sur- and 100 L/min, respectively [8]. Compared to EELI at 0.5 rogate for inspiratory effort [55]. When HFNC flows L/Kg/min, EELI in non-dependent lung regions increased were increased, inspiratory effort (Fig. 6A), pressure– at 1.0 L/Kg/min and 1.5 L/Kg/min (p = 0.01), with signifi- time product (Fig. 6B), and WOB (Fig. 6C) decreased [7, cance reached at 1.5 L/Kg/min (p < 0.05), while EELI in 24]. Mauri et al. described an exponential decay correla- dependent lung regions remained constant (p = 0.548). tion between HFNC flows and patient inspiratory effort Both studies suggest that HFNC flows at 60–65 L/min [7]. The reduction in the patient effort was caused by may cause more recruitment than overdistention, while several factors, such as recruitment of atelectatic regions, high flows (such as 100 L/min) may result in lung over- an increase in dead space washout, a decrease in nasal distention, especially in non-dependent lung regions [8]. resistance, an improvement in secretion clearance, and The large variability between patients in these two studies an increase in dynamic lung compliance [8, 24]. However, should be noted, suggesting that personalized flow titra- they also found that 43% of patients had increased Pes tion based on its physiological effects may be a pragmatic when HFNC flows were increased from 30 or 45 to 60 L/ approach to be used at the bedside. For example, Mauri min [7]. The patients that demonstrated an increase in  et al. [7] reported that 37% of patients had improvement Pes might have had a compliance decrease due to alveo- in EELI in dependent regions with HFNC at 30 or 45 L/ lar overdistention, particularly in the previously relatively min, but not at 60 L/min. Similarly, Zhang et al. [9] com- well-aerated regions of their lungs with 30 L/min [7]. pared EELI at baseline (room air) versus 60 L/min and Thus, due to the concerns that lung injury might occur in used the regional recruitment (recruited pixels) to define patients who have no recruitment with increasing HFNC the potential of lung recruitment, in which recruited pix- flows, it has been suggested to titrate flow based on the els > 10% pixels at 60 L/min than at baseline was defined inspiratory effort [9]. as the high potential of recruitment. They found that 13 in 24 patients (54%) had a high potential for recruitment. Dead space and respiratory rates For these patients, they noted that recruitment mainly Another important flow-dependent effect is the reduc- occurred in dependent lung regions when HFNC flow tion in dead space. Pinkham et al. [14] reported that was increased from 0 to 60 L/min [9]. For the rest of the exhaled gas rebreathing volume increased as RR patients included in the study, seven had unchanged EELI increased [14]. Moreover, when RR was ≥ 25 breaths/ and four had overdistension without lung recruitment, min, rebreathing volume with HFNC at 20 L/min was occurring mainly in the non-dependent lung regions greater than when flows were 40 and 60 L/min. However, [9]. The difference in regional volume distribution from there were no differences in rebreathing volume among various flows in the three studies might be due to the the three flows when RR was 15 breaths/min. Thus, the factors that cause different responses to PEEP, including authors proposed that RR could be used as an indicator disease severity, etiology, duration of pulmonary disease, that HFNC flows should increase when RR is high [14]. Li et al. Critical Care (2023) 27:78 Page 16 of 20 Five studies to date have shown an overall reduction in However, the ROX index plateaued when the flow was RR when HFNC flows are increased for patients with set at 1.67 times PTIF [6]. Currently, there is no commer- AHRF [6–9, 14], and two of them reported significant cially available device to monitor PTIF breath-by-breath, reductions in minute ventilation [7, 14]. so this approach may not be feasible. Butt et al. measured PTIF during mechanical ventilation before extubation Patient comfort and set HFNC flows based on the patient’s comfort after Mauri et al. [50] compared 30 vs 60 L/min and 31 °C vs extubation which ranged between 30 and 50 L/min [44]. 37 °C for 40 patients with AHRF. Patient comfort was They reported a significant correlation between PTIF and lower at 37 °C than at 31 °C, but comfort was not differ- HFNC flows [44]. Thus, PTIF measured on a mechanical ent between flows set at 30 and 60 L/min. Despite large ventilator before extubation might be a reference for flow interindividual variability, they reported a higher comfort settings during post-extubation HFNC treatment. How- score with the lowest temperature and the highest flow ever, more studies are needed to validate the accuracy of in the subgroup of patients whose F ­ IO2 was ≥ 0.45 [50]. these methods and explore the effects of these methods However, this does not necessarily mean that the opti- on regional lung aeration and inspiratory efforts. Addi- mal temperature to achieve better comfort should be tionally, the clinical benefits, such as the need for reintu- 31 °C. As temperature differences were only assessed by bation, of applying this method to set flows during HFNC comparing 20-min periods, tolerance may improve over treatment are unknown. time. Similarly, three other studies evaluating comfort in Before the aforementioned methods are clearly demon- patients with AHRF reported no significant differences strated and devices are commercially available, clinicians between flows set below 60–65 L/min [6, 8, 44]. Patient may use pragmatic assessments that can be easily meas- comfort was significantly lower when HFNC flows ured at the bedside, such as SF ratio, RR, ROX index, and were set at ~ 100 L/min [8]. Interestingly, Butt et al. [44] comfort to titrate HFNC flows. One possible strategy found that patient comfort significantly decreased when could be that when oxygenation starts to plateau, clini- HFNC flow was set at ≥ 50 L/min among post-extubation cians might stop the upward flow titration and return to patients. Thus, for patients with AHRF who are more the previous flow when the plateau is first recognized. hypoxemic and present higher inspiratory demands, Patient comfort is also a key consideration during HFNC higher flows may be associated with better comfort. flow titration. However, it should be noted that the Once again, variability in patient response to different changes in oxygenation, RR, and comfort to HFNC flows flows and temperatures highlights the importance of per- may not be sensitive in patients with mild hypoxemia. sonalization of HFNC settings. Patients with COPD Treatment failure The main effects of HFNC flows for COPD are improved Only one study compared the effect on clinical outcomes ventilation efficiency [11, 12, 21–24], pressure–time of using different flows, which was conducted among car- product (Fig. 6B), [11, 12] WOB (Fig. 6C), [24] and com- diac surgery patients with post-extubation hypoxemia fort [12, 21, 22]. When HFNC flows were increased, with HFNC at 40 versus 60 L/min. The authors reported seven studies reported reductions in RR [11, 12, 21, 22], clinically meaningful differences in treatment failure rate and two studies reported increases in Vt (Table 3) [12, between the two groups (30.3% vs 12.1%, p = 0.11), with 21]. Additionally, longer exhalation times were observed reintubation rates of 15.2% and 6.1%, respectively [27]. [12], which might help alleviate air-trapping and WOB. With increased flows, minute ventilation was the same or Individualization of HFNC flow settings lower [21], but P ­ aCO2 or ­PtCO2 was lower [11, 12, 21– For patients with AHRF, it appears that higher flows 24]. This interesting finding implies improved alveolar improve oxygenation and lung compliance, and reduce ventilation. WOB. These clinical benefits are, at least in part, due to the increase in PEEP and end-expiratory lung volume. Patients during procedural sedation However, it should be noted that not all patients respond Three RCTs compared the effectiveness of different flows to higher flows, and some patients might have uneven during procedural sedation (Additional file 1: Table S4, volume distribution in their lungs which might lead to appendix p6) [47–49], with two RCTs comparing HFNC alveolar overdistention in non-dependent lung regions. at 40 vs 60 L/min [47, 48] and one comparing HFNC at To individualize flow settings based on patient needs 30 vs 50 L/min [49]. Compared to HFNC at low flows (30 (PTIF), Li et al. [6] set HFNC flow to match the PTIF or 40 L/min), HFNC at higher flows (50 or 60 L/min) had or 10, 20, and 30 L/min above PTIF. They found that greater oxygenation at the end of the procedure (Addi- the SF ratio and ROX index increased as flow increased. tional file 1: Fig. S2, appendix p7) [47–49] and required Table 3 Physiological effects of flow settings during HFNC treatment for patients with COPD Author/year Study design N Patients Activity/ Flow (L/min) FiO2 or ­SpO2 RR PaCO2 or ­PtCO2 VT (ml) MV (L/min) PTP ­(cmH2O x Others Li et al. Critical Care Breathing (breaths/ Sec/min) pattern min) Bräunlich et al. Prospective 54 Stable COPD Not specified 20 Not specified 11.6 ± 3.6 91 ± 6.7 534 ± 215 6.0 ± 2.8 NA MAP: 0.92 ± 0.5 2016 [21] gold C or D RSBI: 27.4 ± 17.9 30 NA (2023) 27:78 11.1 ± 3.6 87.4 ± 6.2 524 ± 228 5.6 ± 2.6 MAP: 1.44 ± 0.8 RSBI: 28 ± 18.9 40 10.3 ± 3.3 NA 562 ± 249 5.5 ± 2.5 NA MAP: 2.14 ± 1 RSBI: 25.1 ± 19.9 50 9.9 ± 2.7 NA 559 ± 260 5.4 ± 2.5 NA MAP: 3.01 ± 1 RSBI:25.4 ± 18.3 Pisani et al. Randomized 14 COPD with MC 0 Not specified 24.8 ± 2.3 61.2 ± 9.2 314 ± 84 NA 238.3 ± 82.1 PEEPi, dyn: 2017 [12] controlled trial CHRF 2.12 ± 0.9 Comfort: 7 (5,8)  Pes: 13.5 ± 6.7 MC 20 19.0 ± 5.2 57.2 ± 11.7 391 ± 106 NA 164.2 ± 51.3 PEEPi, dyn: 1.48 ± 0.7 Comfort:5.5 (5,8)  Pes: 8.7 ± 4.1 MO 20.8 ± 5.8 NA NA NA 172.7 ± 45.4  Pes: 12 ± 5.8 MC 30 18.7 ± 3.6 55.7 ± 10.6 364.2 ± 66 NA 143.2 ± 48.9 PEEPi, dyn: 1.03 ± 0.6 Comfort: 5.5 (2,8)  Pes: 8.2 ± 3.7 MO 19.6 ± 2.8 NA NA NA 157.3 ± 56.9  Pes: 10.2 ± 5.2 Bräunlich et al. Prospective 36 Stable COPD Not specified 20 Not specified NA 94.2 ± 8.3a NA NA NA PaO2: 2018 [23] with CHRF 99.2 ± 18.9a MAP: 0.57 ± 3.8 40 NA 93.5 ± 4.4a NA NA NA PaO2: 93.1 ± 15.7a MAP: 2.3 ± 1.6 Mckinstry et al. Randomized 48 Stable COPD Not specified 0 FIO2 0.21 17.5 ± 4.8 38.8 ± ­5b NA NA NA NA 2018 [22] controlled trial 15 16.0 ± 5.7 38.0 ± 5.3b NA NA NA Comfort:11 (4.5,17,4) 30 13.4 ± 5.2 37.3 ± 6.0b NA NA NA Comfort:22 (13.8,26.7) 45 13.3 ± 4.8 36.3 ± 5.6b NA NA NA NA Page 17 of 20 Li et al. Critical Care Table 3 (continued) (2023) 27:78 Author/year Study design N Patients Activity/ Flow (L/min) FiO2 or ­SpO2 RR PaCO2 or ­PtCO2 VT (ml) MV (L/min) PTP ­(cmH2O x Others Breathing (breaths/ Sec/min) pattern min) Rittayamai Prospective 12 Hypercapnic Not specified 10 SpO2 ≥ 92%, 24 (20,29) 41 ± ­7b NA NA 220 ± 100 SF:271 ± ­3c et al. 2019 [11] COPD ­FIO2 20 0.30–0.40 23 (18,28) 41 ± ­8b NA NA 211 ± 90 SF:274 ± ­3c b 30 21 (18,27) 41 ± ­7 NA NA 187 ± 84 SF:277 ± ­3c 40 21 (18,27) 41 ± ­7b NA NA 189 ± 87 SF:277 ± ­2c 50 21 (18,26) 41 ± ­7b NA NA 201 ± 86 SF: 280 ± ­2c Delorme et al. Randomized 5 Hypercapnic Not specified 5 ± 4 (base- SpO2 90–94% NA NA NA 15.3 (11.2, 192 (161, 245) WOB: 7.6 (4.0, 2017 [24] crossover trial line) 20.0) 8.4) 20 NA NA NA 12.1 (7.4, 15.0) 173 (105, 252) WOB: 4.2 (2.4, 5.2) 40 NA NA NA 9.6 (7.6, 20.4) 173 (147, 228) WOB: 3.7 (3.0, 6.7) 60 NA NA NA 12.2 (6.6, 16.3) 145 (79, 235) WOB: 5.1 (1.8, 5.4) FiO2 fraction of inspired oxygen (%), SpO2 saturation of oxygenation using pulse oximeter (%), PaCO2 arterial partial pressure of carbon dioxide (mmHg), PtCO2 transcutaneous partial pressure of carbon dioxide (mmHg), PcCO2 capillary partial pressure of carbon dioxide (mmHg), RR respiratory rate (breath/min),  EELIglob global change in end-expiratory lung impedance measured by electrical impedance tomography, PF ­PaO2/FiO2, SF ­SpO2/FiO2, ROX index [ ­(SpO2/FiO2)/RR], PEEP positive end-expiratory pressure measure by pharyngeal or tracheal catheter, PTP esophageal pressure-time product per minute ­(cmH2O·s/min), VT tidal volume (ml), MV minute ventilation (L/min),  Pes esophageal pressure swings ­(cmH2O), MAP mean airway pressure (mbar), PEEPi dyn dynamic intrinsic positive end-expiratory pressure (­ cmH2O), WOB work of breathing (J/min), Comfort respiratory comfort level measured by a scale, COPD chronic obstructive pulmonary disease, CHRF chromic hypercapnic respiratory failure, MO mouth opened, MC mouth closed, RSBI rapid shallow breathing index, HFNC high-flow nasal cannula, NA not available a Value reported as change from baseline (%) b Value reported as ­PtCO2 c Values are calculated Page 18 of 20 Li et al. Critical Care (2023) 27:78 Page 19 of 20 fewer interventions, such as jaw lifting, during the proce- LP. OR discloses a research grant from Hamilton Medical and speaker fees from Hamilton Medical, Ambu, Fisher & Paykel Ltd., and Aerogen Ltd., and dure [49]. non-financial research support from Timpel. TM discloses personal fees from Fisher and Paykel, Draeger Medical, Hamilton and Mindray. FA and WT have no Conclusion conflict of interest to disclose. The physiological effects of HFNC oxygen therapy are flow-dependent and are maximized when the flow Received: 3 January 2023 Accepted: 15 February 2023 exceeds PTIF. However, PTIF varies between patients and disease conditions. Higher flows result in improved oxygenation and dead space washout and can reduce work of breathing. Notably, higher flows can also lead to References alveolar overdistention in non-dependent lung regions 1. Papoutsi E, Giannakoulis VG, Xourgia E, et al. Effect of timing of intubation on clinical outcomes of critically ill patients with COVID-19: a systematic and to patient discomfort. The impact of flows on differ- review and meta-analysis of non-randomized cohort studies. Crit Care. ent patients is largely heterogeneous. Individualizing flow 2021;25(1):121. settings during HFNC treatment is necessary, and titrat- 2. Rochwerg B, Granton D, Wang DX, et al. High flow nasal cannula com- pared with conventional oxygen therapy for acute hypoxemic respiratory ing flow based on clinical findings like oxygenation, RR, failure: a systematic review and meta-analysis. Intensive Care Med. and patient comfort is a pragmatic way forward, at least 2019;45(5):563–72. for now. 3. Fernando SM, Tran A, Sadeghirad B, et al. Noninvasive respiratory sup- port following extubation in critically ill adults: a systematic review and network meta-analysis. Intensive Care Med. 2021;48(2):137–47. Supplementary Information 4. Feng Z, Zhang L, Yu H, et al. High-flow nasal cannula oxygen therapy The online version contains supplementary material available at https://​doi.​ versus non-invasive ventilation for AECOPD patients after extubation: org/​10.​1186/​s13054-​023-​04361-5. a systematic review and meta-analysis of randomized controlled trials. COPD. 2022;17:1987. 5. Tan D, Walline JH, Ling B, et al. High-flow nasal cannula oxygen therapy Additional file 1. Assessments of risk of bias for included trials (Figure versus non-invasive ventilation for chronic obstructive pulmonary disease S1 and Table S1); report of peak tidal inspiratory flow (Table S2); global patients after extubation: a multicenter, randomized controlled trial. Crit and regional end-expiratory lung volume with different HFNC flows Care. 2020;24(1):489. (Table S3); RCTs (Table S4) and gas exchanges (Figure S2) of different 6. Li J, Scott JB, Fink JB, et al. Optimizing high-flow nasal cannula flow set- flows during procedural sedation. tings in adult hypoxemic patients based on peak inspiratory flow during tidal breathing. Ann Intensive Care. 2021;11(1):164. 7. Mauri T, Alban L, Turrini C, et al. Optimum support by high-flow nasal Acknowledgements cannula in acute hypoxemic respiratory failure: effects of increasing flow We thank Patricia Chavez, MLIS, from the Library of Rush University for her help rates. Intensive Care Med. 2017;43(10):1453–63. in the literature search and screening. We also thank Jian Luo, PhD, MD, from 8. Basile MC, Mauri T, Spinelli E, et al. Nasal high flow higher than 60 L/ the University of Oxford for his consultancy of the methodology. min in patients with acute hypoxemic respiratory failure: a physiological study. Crit Care. 2020;24(1):654. Author contributions 9. Zhang R, He H, Yun L, et al. Effect of postextubation high-flow nasal can- JL designed this project. FA, JL, and WT performed the literature search and nula therapy on lung recruitment and overdistension in high-risk patient. screening. FA and WT extracted the data for the systematic review. FA and Crit Care. 2020;24(1):82. WT had full access to the data and verified the data. JL and FA drafted the 10. Mauri T, Carlesso E, Spinelli E, et al. Increasing support by nasal high flow manuscript. JBS, OR, and TM interpreted the data and provided critical edits acutely modifies the ROX index in hypoxemic patients: a physiologic and comments on the manuscript. All authors reviewed the manuscript for study. J Crit Care. 2019;53:183–5. important intellectual content. JL and FA equally contributed to the overall 11. Rittayamai N, Phuangchoei P, Tscheikuna J, et al. Effects of high-flow nasal project described in this article. JL was responsible for the decision to submit cannula and non-invasive ventilation on inspiratory effort in hypercapnic the manuscript. All authors read and approved the final manuscript. patients with chronic obstructive pulmonary disease: a preliminary study. Ann Intensive Care. 2019;9(1):1–8. Funding 12. Pisani L, Fasano L, Corcione N, et al. Change in pulmonary mechanics and None. the effect on breathing pattern of high flow oxygen therapy in stable hypercapnic COPD. Thorax. 2017;72(4):373. Availability of data and materials 13. Ritchie JE, Williams AB, Gerard C, et al. Evaluation of a humidified Data will be available 36 months after article publication to researchers who nasal high-flow oxygen system, using oxygraphy, capnography and provide a methodologically sound and ethically approved proposal, for any measurement of upper airway pressures. Anaesth Intensive Care. purpose of analysis. 2011;39(6):1103–10. 14. Pinkham MI, Domanski U, Franke K, et al. Effect of respiratory rate and Declarations size of cannula on pressure and dead-space clearance during nasal high flow in patients with COPD and acute respiratory failure. J Appl Physiol. Ethics approval and consent to participate 2022;132(2):553. Not applicable. 15. Okuda M, Tanaka N, Naito K, et al. Evaluation by various methods of the physiological mechanism of a high-flow nasal cannula (HFNC) in healthy Competing interests volunteers. BMJ Open Respir Res. 2017;4(1):e000200. JL discloses research funding from Fisher & Paykel Healthcare Ltd., Aerogen 16. Delorme M, Bouchard P, Simon M, et al. Physiologic effects of high-flow Ltd., and Rice Foundation, and speaker fees from American Association for nasal cannula in healthy subjects. Respir Care. 2020;65(9):1346–54. Respiratory Care, Aerogen Ltd., Heyer Ltd., and Fisher & Paykel Healthcare Ltd. 17. Vieira F, Bezerra FS, Coudroy R, et al. High-flow nasal cannula compared JL also serves as section editor for Respiratory Care. JBS discloses research with continuous positive airway pressure: a bench and physiological funding from Teleflex and speaker fees from Aerogen and Medline Industries, study. J Appl Physiol. 2022;132(6):1580–90. Li et al. Critical Care (2023) 27:78 Page 20 of 20 18. Parke RL, Bloch A, McGuinness SP. Effect of very-high-flow nasal therapy 43. Allen K, Galek K. The influence of airflow via high-flow nasal cannula on on airway pressure and end-expiratory lung impedance in healthy volun- duration of laryngeal vestibule closure. Dysphagia. 2020;36(4):729–35. teers. Respir Care. 2015;60(10):1397–403. 44. Butt S, Pistidda L, Floris L, et al. Initial setting of high-flow nasal oxygen 19. Plotnikow GA, Thille AW, Vasquez DN, et al. Effects of high-flow nasal can- post extubation based on mean inspiratory flow during a spontaneous nula on end-expiratory lung impedance in semi-seated healthy subjects. breathing trial. J Crit Care. 2021;63:40–4. Respir Care. 2018;63(8):1016–23. 45. Parke RL, McGuinness SP. Pressures delivered by nasal high flow oxygen 20. Garofalo E, Bruni A, Pelaia C, et al. Evaluation of a new interface combin- during all phases of the respiratory cycle. Respir Care. 2013;58(10):1621–4. ing high-flow nasal cannula and cpap. Respir Care. 2019;64(10):1231–9. 46. Natalini D, Grieco DL, Santantonio MT, et al. Physiological effects of 21. Bräunlich J, Köhler M, Wirtz H. Nasal highflow improves ventila- high-flow oxygen in tracheostomized patients. Ann Intensive Care. tion in patients with COPD. Int J Chron Obstruct Pulmon Dis. 2019;9(1):1–9. 2016;11(1):1077–85. 47. Yi P, Li Q, Yang Z, et al. High-flow nasal cannula improves clinical efficacy 22. McKinstry S, Pilcher J, Bardsley G, et al. Nasal high flow therapy and PtCO2 of airway management in patients undergoing awake craniotomy. BMC in stable COPD: a randomized controlled cross-over trial. Respirology. Anesthesiol. 2020;20(1):1–156. 2018;23(4):378–84. 48. Lucangelo U, Vassallo FG, Marras E, et al. High-flow nasal interface 23. Bräunlich J, Mauersberger F, Wirtz H. Effectiveness of nasal high flow in improves oxygenation in patients undergoing bronchoscopy. Crit Care hypercapnic COPD patients is flow and leakage dependent. BMC Pulm Res Pract. 2012;2012:506382–6. Med. 2018;18(1):14. 49. Sago T, Harano N, Chogyoji Y, et al. A nasal high-flow system prevents 24. Delorme M, Bouchard P, Simon M, et al. Effects of high-flow nasal cannula hypoxia in dental patients under intravenous sedation. J Oral Maxillofac on the WOB in patients recovering from acute respiratory failure. Critical Surg. 2015;73(6):1058–64. Care Med. 2017;45(12):1981–8. 50. Mauri T, Galazzi A, Binda F, et al. Impact of flow and temperature on 25. Goligher EC, Slutsky AS. Not just oxygen? mechanisms of benefit from patient comfort during respiratory support by high-flow nasal cannula. high-flow nasal cannula in hypoxemic respiratory failure. Am J Respir Crit Crit Care. 2018;22(1):120. Care Med. 2017;195(9):1128–31. 51. Nishimura M. High-flow nasal cannula oxygen therapy in adults: physi- 26. Duprez F, De Terwangne CC, Bellemans VV, et al. High-flow nasal cannula ological benefits, indication, clinical benefits, and adverse effects. Respir therapy, factors affecting effective inspired oxygen fraction: an experi- Care. 2016;61(4):529–41. mental adult bench model. J Clin Monit Comput. 2022;36(5):1441–8. 52. Anderson N, Cassidy P, Janssen L, et al. Peak inspiratory flows of adults 27. Theologou S, Ischaki E, Zakynthinos SG, et al. High flow oxygen therapy exercising at light, moderate and heavy work loads. J Int Soc Respir Prot. at two initial flow settings versus conventional oxygen therapy in cardiac 2006;23:53–63. surgery patients with postextubation hypoxemia: a single-center, 53. Li J, Chen Y, Ehrmann S, et al. Bronchodilator delivery via high-flow nasal unblinded, randomized, controlled trial. J Clin Med. 2021;10(10):2079. cannula: a randomized controlled trial to compare the effects of gas 28. Chikata Y, Onodera M, Oto J, et al. ­FiO2 in an adult model simulating high- flows. Pharmaceutics. 2021;13(10):1655. flow nasal cannula therapy. Respir Care. 2017;62(2):193–8. 54. Chanques G, Riboulet F, Molinari N, et al. Comparison of three high flow 29. Sun Y, Dai B, Peng Y, et al. Factors affecting FiO2 and PEEP during oxygen therapy delivery devices: a clinical physiological cross-over study. high-flow nasal cannula oxygen therapy: a bench study. Clin Respir J. Minerva Anestesiol. 2013;79(12):1344–55. 2019;13(12):758–64. 55. Vaporidi K, Soundoulounaki S, Papadakis E, et al. Esophageal and trans- 30. Hebbink RHJ, Duiverman ML, Wijkstra PJ, et al. Upper airway pressure dis- diaphragmatic pressure swings as indices of inspiratory effort. Respir tribution during nasal high-flow therapy. Med Eng Phys. 2022;104:103805. Physiol Neurobiol. 2021;284:103561. 31. Luo J, Lu M, Zhao Z, et al. Positive end-expiratory pressure effect of 3 high-flow nasal cannula devices. Respir Care. 2017;62(7):888–95. 32. Nielsen KR, Ellington LE, Gray AJ, et al. Effect of high-flow nasal cannula Publisher’s Note on expiratory pressure and ventilation in infant, pediatric, and adult Springer Nature remains neutral with regard to jurisdictional claims in pub- models. Respir Care. 2018;63(2):147–57. lished maps and institutional affiliations. 33. Adams CF, Geoghegan PH, Spence CJ, et al. Modelling nasal high flow therapy effects on upper airway resistance and resistive work of breath- ing. Respir Physiol Neurobiol. 2018;254:23–9. 34. Onodera Y, Akimoto R, Suzuki H, et al. A high-flow nasal cannula system with relatively low flow effectively washes out ­CO2 from the anatomical dead space in a sophisticated respiratory model made by a 3D printer. Intensive Care Med Exp. 2018;6(1):7. 35. Moore CP, Katz IM, Pichelin M, et al. High flow nasal cannula: influence of gas type and flow rate on airway pressure and ­CO2 clearance in adult nasal airway replicas. Clin Biomech. 2019;65:73–80. 36. Guérin C, Cour M, Degivry F, et al. A bench comparison of the effect of high-flow oxygen devices on work of breathing. Respir Care. 2022;67(9):1129–37. 37. Möller W, Celik G, Feng S, et al. Nasal high flow clears anatomical dead space in upper airway models. J Appl Physiol. 2015;118(12):1525–32. Ready to submit your research ? Choose BMC and benefit from: 38. Le Moigne G, Nazir S, Pateau V, Courtois E, Lher E. Noninvasive tidal vol- ume measurements, using a time-of-flight camera, under high-flow nasal • fast, convenient online submission cannula: a physiological evaluation, in healthy volunteers. Crit Care Med. 2021;50(1):61. • thorough peer review by experienced researchers in your field 39. Groves N, Tobin A. High flow nasal oxygen generates positive airway • rapid publication on acceptance pressure in adult volunteers. Aust Crit Care. 2007;20(4):126–31. • support for research data, including large and complex data types 40. Möller W, Feng S, Domanski U, et al. Nasal high flow reduces dead space. J Appl Physiol. 2017;122(1):191–7. • gold Open Access which fosters wider collaboration and increased citations 41. Sanuki T, Mishima G, Kiriishi K, et al. Effect of nasal high-flow oxygen • maximum visibility for your research: over 100M website views per year therapy on the swallowing reflex: an in vivo volunteer study. Clin Oral Invest. 2016;21(3):915–20. At BMC, research is always in progress. 42. Arizono S, Oomagari M, Tawara Y, et al. Effects of different high-flow nasal cannula flow rates on swallowing function. Clin Biomech. Learn more biomedcentral.com/submissions 2021;89:105477.