This study is the first to report the breathing patterns of non-intubated patients with AHRF. Patient oxygenation improved as HFNC flow increased up to two times of their PTIF in all patients. Interestingly, the results of the in vitro study suggested that the oxygenation improvement observed above the PTIF was more likely due to the increase in the airway pressure as FIO2 remained unchanged.
Peak inspiratory flow during tidal breathing
Our findings showed that PTIF (34 [9] L/min) was higher in patients with AHRF compared to adult healthy volunteers (PTIF of 28 [9] L/min) [
17] and was similar when compared to patients with tracheostomy (PTIF of 30 [27–32] L/min) [
19]. Therefore, our findings align with the clinician assumption that patients with hypoxemia have higher PTIF than patients without hypoxemia [
22]. However, we did not find a significant correlation between PTIF and the severity of hypoxemia.
In two recently published studies in intubated patients, PTIF varied from 25–65 L/min [
20] to 40–80 L/min [
21], values that were higher than PTIF in our patients. This might be explained by the need to overcome the resistance of an endotracheal tube. Butt and colleagues utilized the PTIF measured with intubation to guide HFNC flow settings after extubation. They found a significant correlation between the PTIF pre-extubation and the HFNC flow settings that patients had the greatest comfort after extubation [
21]. Although PTIF measured during the spontaneous breathing trial while intubated slightly overestimated patient PTIF post-extubation, this strategy allows personalized flow titration immediately after extubation.
It should also be noted that, despite the obvious limitations in breathing pattern measurement, the breathing patterns described in the present study may provide a reference to establish settings for future in vitro studies that simulate spontaneous breathing of hypoxemic patients, such as studies on HFNC [
15,
16,
18], noninvasive ventilation, or aerosol therapy [
24].
Patient responses to different HFNC flows
We observed oxygenation improvement as HFNC flow increased, which agrees with others’ observations [
10,
12]. When HFNC flow was increased from 30 to 60 L/min in the study by Mauri and colleagues [
10] or from 20 to 60 L/min in the study by Delorme and coworkers [
12], both groups found significant improvement in oxygenation, lung aeration, dynamic compliance, and work of breathing. Based on their results, both groups of investigators recommended HFNC flow of 60 L/min for adult patients with AHRF [
10,
12]. However, it is worth noting that there was a high heterogeneity in patient’s response to different flows and, therefore, individualizing flow settings during HFNC therapy seems to be a reasonable approach. It was also hypothesized that the oxygenation improvement observed may be due to the increase of oxygen delivery. However, our in vitro study showed that when the flow ratio was ≥ 1, tracheal peak inspiratory and peak expiratory pressures increased as flow ratio increased with no additional increase in the F
IO
2. Thus, these results suggest that the oxygenation improvement observed with flow that exceed the PTIF could be, at least in part, explained by the increased airway pressure generated by these higher flows. [
16‐
18].
Respiratory rate decreased significantly at HFNC flows set at 1.34–1.67 times of PTIF and no further improvements in ROX index were found when HFNC flows were set at ≥ 1.68 times of PTIF. Similarly, Basile and co-investigators [
13] set HFNC flow based on patient predicted body weight (PBW) for 12 patients with AHRF. According to their protocol of 0.5, 1.0, and 1.5 L/min/kg of PBW, they utilized median flows of 35, 65, and 100 L/min. They found that HFNC flow at 1.5 L/min/kg of PBW was worse tolerated and did not improve homogeneity of ventilation or increase in end-expiratory lung volume (EELV) compared with HFNC flow at 1.0 L/min/kg of PBW. Moreover, the change in ROX measured at 30 L/min and 60 L/min has been correlated with a change in EELV [
25]. Importantly, in 30% of the patients, the ROX index and EELV decreased after increasing the flow. These findings support our observation that an arbitrary or maximum flow setting, such as 60 L/min, might exceed the individual plateau level in some patients but might be insufficient for other patients who have high PTIF. It should be noted that the increase in EELV and lung homogeneity associated at certain flows may reduce the likelihood of patient self-inflicted lung injury (P-SILI) [
26]. This is noteworthy, as P-SILI may be associated with HFNC failure and need for mechanical ventilation. Thus, like for an intubated patient, personalizing the flow settings to minimize the risk of P-SILI may be a strategy to potentially improve outcomes in AHRF patients treated with HFNC.
Currently, there is no commercially available device to measure patient PTIF, and different HFNC flows may alter PTIF given that RR and inspiratory effort are affected by the flows used [
11]. Therefore, it is unlikely that baseline PTIF could be used to optimize flow settings at the bedside during HFNC therapy. The present study highlights the reality that one size (or HFNC flow, in this case) does not necessarily fit all. Thus, a pragmatic solution to set HFNC flow would be to initiate HFNC at a flow of 40 L/min then rapidly titrate upwards based on response in ROX index and respiratory rate, as well as patient tolerance/comfort. Specifically, when the improvement in ROX index begins to plateau, then optimal HFNC flow has been achieved. It is worth noting that during this titration process, F
IO
2 needs to be adjusted to maintain a SpO
2 between the target range (preferably 90–97%) at different HFNC flow settings [
27,
28].
This study has certain limitations. First, the maximum studied flow was 60 L/min, and not all patients received HFNC flows of 20 and 30 L/min above their PTIF. Therefore, in patients whose PTIF was 40 L/min or higher, whether their responses to higher flow were the same as patients whose PTIF was 30 L/min or lower is unknown. Second, this was a short-term non-randomized study that might not reveal any long-term effects. Future studies are needed to understand the long-term benefits of the individualized HFNC flow settings with more frequent measurements and flow titration. Third, breathing pattern measurements were done while the patients were not using the HFNC device. However, we maintained the same oxygenation levels during the measurement, minimizing the effect that hypoxemia may have on the respiratory pattern. Fourth, we only assessed oxygenation, respiratory rate and patient comfort at different flows, which might not reflect the lung homogeneity during tidal ventilation. Similarly, we did not measure inspiratory effort and, therefore, significant improvements in terms of reducing P-SILI might be possible with higher flows despite no associated oxygenation improvement. Indeed, better oxygenation may not be necessarily related with better outcomes. Fifthly, the in vitro study was performed with the mouth closed, and a large-size cannula, thus, the pressures achieved might not reflect the ones during daily clinical practice. Finally, the effects on oxygenation were assessed using SpO
2/F
IO
2 instead of partial pressure of oxygen (PaO
2)/F
IO
2. That said, many AHRF patients treated by HFNC are monitored non-invasively and SpO
2/F
IO
2 has been shown to be a convenient, noninvasive, and practical substitute for PaO
2/F
IO
2 [
27,
28]. Therefore, this noninvasive assessment on oxygenation is clinically useful and represents what is currently done in daily clinical practice during treatment with HFNC.