Patient and public involvement

No patients were involved in the study, all results are based on modelling of data from previously published studies.

Modelling a cohort of healthy subjects receiving HFNC

To investigate the relationship between functional lung volume and airway pressures produced by HFNC, we adapted a multi-compartmental computational simulator, previously employed to simulate patients with different conditions, such as COVID-19 [15], chronic obstructive pulmonary disease (COPD) [16], ARDS [17, 18] and AHRF [19]. The simulator offers several advantages including the ability to define multiple alveolar compartments with individually configurable mechanical characteristics such as alveolar collapse, consolidation and stiffening, gas-exchange disruption, pulmonary vasoconstriction, vasodilation, and airway obstruction. This allows for the representation of various clinical features including acute lung injury, ventilation-perfusion mismatch, physiological shunt, alveolar gas trapping with intrinsic positive end-expiratory pressure (iPEEP), and reopening of collapsed alveoli [20, 21]. Key mechanisms involved in the application of HFNC therapy, including carbon dioxide clearance from dead space, gas leakage, a friction factor for turbulent flow, and increases in airway resistance at high flow rates are included in the model presented here – full details are provided in Sect. 4 of the Supplementary Material.

We initially adapted this computational simulator to create a virtual cohort of ten subjects whose characteristics were representative of participants in four previous studies reporting mean airway pressures in healthy subjects at multiple HFNC flow rates [9, 10, 22, 23]. This virtual cohort consists of an equal number of males and females, with an average age of 33 years, height of 170 cm, and weight of 74 kg, all falling within the normal BMI range. Detailed information regarding the cohort is presented in Table S2 in the supplementary material. Healthy lung physiology and respiratory effort are simulated in all cases, with compliance and airway resistance values varying within normal ranges.

Modelling reductions in functional lung volume in patients receiving HFNC therapy

To simulate alveolar consolidation/collapse (i.e., compartments with no participation in ventilation), the inlet resistance of a specified number of the alveolar compartments in the model is increased to a large value to preclude any airflow to the alveolus. In this part of the investigation, all other parameters pertaining to the subjects including FiO2 were left unaltered, ensuring that the results are focused exclusively on elucidating the influence of the size of the functional lung volume on the pressures generated by HFNC.

Modelling a cohort of AHRF patients receiving HFNC therapy

To estimate the PEEP produced by HFNC in actual AHRF patients, data were extracted from two studies [24, 25] reporting data on 60 non-COVID-19 patients with moderate-to-severe AHRF, conducted within a respiratory intensive care unit at the University Hospital of Modena, Italy, from 2016 to 2021. These comprised a comprehensive set of physiological measurements for each patient during their HFNC therapy, including gender, age, height, weight, fraction of inspired oxygen, and HFNC flow rate, all of which served as inputs for the cardiopulmonary simulator. Using global optimization algorithms, the simulator’s parameters were calibrated so that its outputs matched as closely as possible the responses of individual patients to HFNC therapy, encompassing partial pressures of oxygen (PaO2) and carbon dioxide (PaCO2) in the arterial blood, oesophageal pressure swings (ΔPes, measured by a dedicated oesophageal pressure transducer), and expiratory tidal volume (VT, measured by numerical integration of respiratory flow measured by a pneumotachograph) – for full details of the model matching process, see the supplementary material. Two patients were excluded from the analysis due to concerns regarding abnormally high VT measurements, potentially stemming from erroneous integration of a portion of the flow directed to the patient into the integrated flow signal.

Calculating PEEP and mean airway pressure during HFNC therapy

Previous experimental studies [9, 10, 22, 23], have measured mean airway pressure (mPaw) produced by HFNC at the participants’ pharynx. In our model, tracheal pressure serves as the reference for calculating both PEEP and mPaw. Specifically, PEEP is determined as the positive end-expiratory pressure calculated at the trachea, while mPaw represents the mean tracheal pressure throughout one complete breathing cycle.