Setting of High-Flow Nasal Oxygen Post Extubation based on Inspiratory Flow during a Spontaneous Breathing Trial

Background: High ow nasal cannula (HFNC) is commonly used post-extubation in intensive care (ICU). Patients’ comfort during HFNC is affected by ow rate: too low ow may limit the benecial effects of HFNC on gas exchange and work of breathing; whilst excessive ow may reduce comfort and adherence to therapy. Currently, there is no consensus on how to set the ow rate of HFNC post-extubation. The study aims to describe the relationship between pre-extubation inspiratory ow requirements and the post-extubation ow rates on HFNC that maximises patient’s level of comfort. Methods: This was an observational, retrospective, single-site study conducted in a tertiary, university-aliated ICU. We included all patients extubated following a successful, standardised spontaneous breathing trial (SBT) during a four-month study period. During a 30-minute SBT we recorded haemodynamic and respiratory variables including inspiratory ow, presence of any signs of respiratory distress and level of comfort using a visual analogue scale (VAS). Patients who passed the SBT were extubated onto HFNC – as per standard clinical practice. HFNC was titrated starting from a ow of 20 L/min and increased in steps of 10L/min, up to 60 L/min or maximum tolerated ow. At each step, patient’s level of comfort was assessed using a VAS. Fraction of inspired oxygen (FiO 2 ) was titrated to maintain oxygen saturation measured by pulse oximetry (SpO 2 ) 92-97%. Results: Nineteen participants were enrolled in the study with a mean (SD) age of 62.5 ± 13.1 years. There was a signicant positive correlation between mean inspiratory ow pre-extubation and the ow setting on HFNC which achieved the best comfort post-extubation (r 2 0.88; p <0.001). The greatest comfort was observed for HFNC ows between 30 and 40 L/min, while above 40 L/min patients’ comfort decreased. Conclusions: Measuring mean inspiratory ow during an SBT allows for individualised setting of HFNC ow rate immediately post-extubation

Results: Nineteen participants were enrolled in the study with a mean (SD) age of 62.5 ± 13.1 years. There was a signi cant positive correlation between mean inspiratory ow pre-extubation and the ow setting on HFNC which achieved the best comfort post-extubation (r 2 0.88; p <0.001). The greatest comfort was observed for HFNC ows between 30 and 40 L/min, while above 40 L/min patients' comfort decreased.
Conclusions: Measuring mean inspiratory ow during an SBT allows for individualised setting of HFNC ow rate immediately post-extubation and achieves the greatest comfort and interface tolerance.

Background
High ow nasal cannula (HFNC) is commonly used to administer supplemental oxygen in intensive care (ICU). Clinical trials have shown that the use of HFNC post-extubation decreases the incidence of reintubation, improves gas exchange and reduces the work of breathing 1,2,3,4 . In contrast with more traditional low-ow systems, which deliver non-heated and non-humidi ed air/oxygen mixture, HFNC allows the administration of warmed and humidi ed gas with FiO 2 up to 1 at a ow rate up to 60 L/min 5 through a nasal interface. The high ow rate presents physiological advantages over low ow 6 , allowing a better matching between the delivered gas ow and patient's spontaneous inspiratory ow, creating a positive end-expiratory pressure (PEEP) effect (generally between 2 and 8 cmH 2 O 7,8 ) and generating a "CO 2 washout" effect from the upper airways 8,9 . Furthermore, the humidi ed warm gas favours the mucociliary function and reduces upper airway resistance 10,11 . The synergistic combination of these mechanisms leads to improved oxygenation 10 , decrease in neuro-ventilatory drive and work of breathing 3,12,13 .
Despite growing evidence supporting the bene ts of HFNC therapy, it is still unclear how to set the ow rate to achieve the optimal comfort and e cacy. Some authors, such as Hernandez 4 , Corley 15 and Parke 16 , suggested titrating the ow in order to reach good level of oxygenation (SpO 2 > 92-95%) or 'titrate to comfort', but in practice a xed ow is often used for every patient, and titrations are not dictated by physiological parameters 17 . Titrating the ow rate according to measures of comfort and respiratory mechanics may increase tolerance and minimise the premature discontinuation of HFNC.
We hypothesised that the "optimal" post-extubation HFNC ow rate may be correlated to the inspiratory ow measured during a spontaneous breathing trial (SBT), and therefore may allow rapid establishment of the best HFNC setting immediately post-extubation.

Study design
This is was an observational, retrospective single-centre study conducted at Guy's & St Thomas' NHS Foundation Trust ICU, London (UK).

Patients
We collected data from all consecutive mechanically ventilated adult patients who underwent a SBT using a 'T-piece' (i.e., using PEEP and pressure support both set at 0 cmH 2 O), and who then underwent tracheal tube extubation onto HFNC from 01 January through 01 April 2017. We did not include patients ventilated for <24 hours, those who did not undergo formal SBT and patients who received non-invasive ventilation (NIV) post extubation. All data were retrospectively collected, fully anonymised from our database and from the ICU electronic patient record (CareVue, Philips Medical Systems UK Limited). The study had institutional approval. The need for individual informed consent was waived as this was an analysis of data collected as part of a prospective audit for usual clinical care, with no breach of privacy or anonymity. The study quali ed as a service evaluation as de ned by the UK NHS Health Research Authority and therefore did not require review by the Research Ethics Committee (http://www.hra.nhs.uk).

Study protocol
Pre-extubation The protocol was based on the Departmental guidelines at Guy's & St Thomas' NHS Foundation Trust.

Post-extubation
Post-extubation, all patients were commenced on HFNC therapy as per our standard of care and FiO 2 was adjusted in order to maintain SpO 2 92-97%. The HFNC device included an air-oxygen blender, which allowed the accurate adjustment of FiO 2 between 0.21 and 1.0 and delivery of gas ow up to 60 L/min through a heated humidi er (MR850, Fisher & Paykel Healthcare). The gas mixture was routed through a circuit to the subject at a temperature of 37°C and an absolute humidity of 44 mgH 2 O/L via large-bore nasal prongs. HFNC was initially administered at a gas ow of 20 L/min and then increased every 10-15 minutes by 10 L/min up to 60 L/min or maximum tolerated rate. At each step, patient's comfort was assessed using a VAS. The HFNC nal ow was set according to patient's best comfort (online supplement EFigure 2).
For every day of HFNC therapy post-extubation (from day 0 to the end of therapy), HFNC setting ( ow rate and FiO 2 ) was titrated and recorded in the patients' electronic medical record. VAS of comfort was reassessed and nal ow rates set accordingly.

Statistical analysis
Comparisons between pre-extubation ventilatory parameters and SBT-related variables versus postextubation HFNC ow at optimum VAS were performed using multiple comparisons ANOVA with pairwise post-hoc comparisons. Correlation between PIF on SBT and best comfort HFNC ow rate, and MIF on SBT and best comfort HFNC ow rate were performed using Spearman's Rank correlation test. A p-value <0.05 was considered statistically signi cant. All analyses were performed using GraphPad Prism version 7.0 (GraphPad Software, La Jolla California, USA) and STATA v16.1 (StataCorp -USA).

Results
Baseline and pre-extubation variables A total of 19 patients were included in the study. Baseline and demographic characteristics, reason for ICU admission and mechanical intubation are shown in the Table 1. Pre-extubation ventilatory setting and SBT-related variables are shown in Table 2. All patients were in a supported spontaneous breathing mode of ventilation before extubation: nine patients were ventilated on CPAP (47.3%), ten on CPAP/PS (52.7%).  HFNC settings, comfort and pre-extubation inspiratory ow The highest mean VAS of comfort was 8.58 seen for HFNC ow rates of both 30L/min (SD 1.57) and 40L/min (SD 1.35). Above 50 L/min of HFNC ow rate, average patients' comfort decreased signi cantly ( Fig. 1). We found a signi cant, positive correlation between both pre-extubation PIF (r 2 0.53, p < 0.001; Fig. 2A) and MIF (r 2 0.88, p < 0.001; Fig. 2B), and the post-extubation HFNC ow rate associated with the patient's greatest level of comfort as de ned by the highest VAS of each individual patient.
All the 19 patients enrolled in the study were followed up until the discontinuation of HFNC therapy. The duration of the HFNC therapy ranged between 1 day and 7 days post-extubation. During this time, VAS for comfort and HFNC settings ( ow and FiO 2 ) were collected and ow re-adjusted according to the level of comfort. Flow setting ranged from a 20 L/min to 50 L/min. Flow requirement was not always stable during the day, with variations of 0-20 L/min (Fig. 3).

Discussion
This study showed that the "optimal" post-extubation HFNC ow rate could be predicted by the inspiratory ow measured during the pre-extubation SBT. We found that the pre-extubation variable that best correlated with HFNC ow rate setting at which the patient was most comfortable (i.e. "optimal" post-extubation HFNC ow rate), was the mean inspiratory ow (MIF). In our study, HFNC ow rate setting associated with highest mean VAS of comfort was 30-40 L/min (Fig. 1). Comfort signi cantly and progressively decreased at ow setting on HFNC of 50 and 60 L/min.
The majority of studies on HFNC chose ow settings of 30-50 L/min, that were associated in our study with a mean VAS of comfort 8.02 (SD 1.64). An exception was the OPERA trial, where Futier et al 18 chose ow settings post-extubation between 50 and 60 L/min. In our patients, at these relatively high ow rates, we reported a lower mean VAS of comfort compared with a ow rate of 30-40 L/min (Fig. 3). It is possible that this could be explain the 7.4% premature discontinuation of HFNC in the OPERA trial. Moreover, the OPERA trial did not show improved pulmonary outcomes in patients who received HFNC compared with standard oxygen therapy. The choice of ow setting associated with lower comfort may have contributed to this outcome.
We also found a wide variation in the HFNC ow settings criteria among the previous studies. Most of the time a xed ow was used for every patient, whilst some authors decided to titrate ow to reach good level of oxygenation with SpO2 92-95%. Those who decided to use a xed ow had a higher rate of discontinuation than those who opted for a titration of ow. Stephan 3 , Tiruvoipati 17 and Maggiore 2 chose a xed ow with 1.45%, 6% and 7.4% of HFNC premature discontinuation, respectively (online supplement ETable 3. Hernandez 4,19,20 , who titrated HFNC ow to comfort, using clinical criteria had no discontinuation. These criteria are not speci ed and therefore not reproducible in all settings. In this context, more standardised criteria based on physiological parameters may lead to more consistent application of optimal settings in the immediate post-extubation period. Establishing comfort immediately post extubation may improve early extubation success rate and increase patient acceptance of the interface.
Overtime, titration will require clinical assessment of comfort and respiratory effort. Indeed, in our cohort, measurements of comfort throughout the duration of HFNC therapy showed that most of the patients required changes in ow rate to achieve the optimal comfort. Variations of ow rate up to 20L/min were required. These data suggest, that frequent ow rate adjustments could be required to optimize and maintain patient comfort during the HFNC therapy.
Limitations of these ndings include that our sample size was relatively small and may not give an indication of the relationship for values outside our range of PIF and MIF. Furthermore, we have not directly considered the impact on e cacy of HFNC and respiratory function when titrating ow to best comfort.

Conclusion
We found a signi cant correlation between the pre-extubation MIF and the post-extubation HFNC ow rate associated with the greatest level of patient comfort. Our data suggest that the mean patient's inspiratory ow rate could be a convenient parameter for individual titration of HFNC ow rate to match patients' best comfort. The study had institutional approval. The need for individual informed consent was waived as this was an analysis of data collected as part of a prospective audit for usual clinical care, with no breach of privacy or anonymity. The study quali ed as a service evaluation as de ned by the UK NHS Health Research Authority and therefore did not require review by the Research Ethics Committee (http://www.hra.nhs.uk).

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests Funding Not applicable Authors' contributions SB made substantial contributions to the design of the project, analysed and interpreted the data, and drafted the manuscript. LP made substantial contributions to the design of the project, analysed and interpreted the data, and drafted the manuscript. LF made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. CL made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. FV made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. GG made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. NB made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. BS made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. SG made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. MS-H made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. LC made substantial contributions to the design of the project, analysed and interpreted the data, and contributed to revising the nal manuscript. The authors read and approved the nal manuscript.

Figure 1
Relationship between mean VAS and HFNC ow setting during post-extubation HFNC.Line represents the mean and shaded area the 95% CI based on a fractional polynomial prediction model Page 13/14

Figure 2
Graph of relationship between peak inspiratory ow (A) and mean inspiratory ow (B) and HFNC setting which achieved best comfort. The regression line is shown with shaded area representing the 95% CI