Introduction
Prone positioning reduces mortality in intubated and mechanically ventilated patients with moderate to severe acute respiratory distress syndrome (ARDS) [
1,
2]. Awake prone positioning (APP) in non-intubated, spontaneously breathing patients with hypoxemic respiratory failure has gained wide-spread use in health care systems overwhelmed by patients with Coronavirus disease 2019 (COVID-19) [
3‐
5] although previously rarely reported [
6‐
9].
Prone positioning improves respiratory mechanics and gas exchange owing to several mechanisms in non-intubated spontaneously breathing and intubated mechanically ventilated patients. It increases lung volume [
10,
11], improves ventilation-perfusion ratio [
12‐
14] and distributes pleural pressure more evenly [
15]. Several studies report transient improvement in oxygenation during APP in a majority of patients with hypoxemic respiratory failure due to COVID-19 pneumonia [
3,
16‐
23]. However, translating physiological improvement into clinically relevant outcomes has not been supported by ARDS-studies [
24] and there remains a gap in the current knowledge for the use of APP [
25‐
28]. To date, the effect of APP on intubation rates in patients with hypoxemic respiratory failure has not been studied in a randomized clinical trial.
The primary aim of this trial was to determine if a protocol for APP and standard care reduces the rate of endotracheal intubation compared to standard care alone among COVID-19 patients with hypoxemic respiratory failure supported with high-flow nasal oxygen (HFNO) or noninvasive ventilation (NIV).
Discussion
This is to the best of our knowledge the first randomized clinical trial investigating prolonged prone positioning in non-intubated spontaneously breathing patients with COVID-19. The main finding was that implementation of a protocol for APP increased the duration of prone positioning but did not affect the rate of intubation in patients with moderate to severe hypoxemic respiratory failure compared with standard care. However, only a minority of patients in the prone group complied with the protocol target of 16 h APP duration per day. Furthermore, there were no statistically significant differences in the use of other supportive treatments, 30-day mortality or time to recovery, although these analyses may have been underpowered.
The results of this study were consistent also in exploratory post-hoc analyses subgrouping patients according to the duration of APP irrespective of group allocation. Further, no benefit of prolonged APP was found in patients with PaO2/FiO2 ratio < 15 kPa at inclusion between the prone and control group.
Prone positioning in mechanically ventilated patients with COVID-19 improves oxygenation and is associated with reduced mortality [
33]. Although APP similarly improves oxygenation in non-intubated patients with COVID-19 [
3,
16‐
23], reports have failed to show benefits on patient-centered outcomes [
25,
26]. A multicenter observational study, investigating a cohort of 199 patients with COVID-19 found no difference in intubation rates in patients with duration of APP for more than 16 h per day compared with shorter duration of APP [
25]. They reported similar baseline characteristics, degree of respiratory failure and mortality but higher intubation rates (41% in the control group and 40% in the prone group) compared with our investigation. Further corroborating our results, a single center observational study including 166 patients with COVID-19 with respiratory rate ≥ 24/min who required oxygen supplementation ≥ 3 L min
−1 found no difference in intubation rates or ICU admission in patients who were treated with APP compared to those who were not [
26]. Although the patients in this study were younger and had less severe respiratory failure at inclusion compared to our population, the authors reported higher overall intubation rates (58% in the prone group and 49% in the control group) compared with our trial.
There are several possible explanations for the neutral result of our investigation. Due to observed beneficial physiological effects, patients with COVID-19 were increasingly treated with APP as part of standard care during the study period at the participating study hospitals, resulting in longer APP duration than expected in the control group. The optimal duration of prone positioning is unknown. However, the mean duration of prone positioning was 17 h per day in the prone group compared to 0 h in the supine group in the first study that reported mortality benefit in mechanically ventilated patients [
1]. Although the median duration of APP per day was 9.0 h in the prone group compared with 3.4 h in the control group in our study, this difference may not have been large enough to decrease the rate of intubation. Intubated patients are often heavily sedated to tolerate prone positioning and it may be difficult to reach a similar duration of prone positioning in awake patients. Our protocol targeted an APP duration of 16 h in the prone group, but only two (6%) patients were able to reach this target. This is similar to a previous study [
23] and indicates that treatment adherence is a major limitation of APP.
Reduction in lung injury associated with mechanical ventilation may in part explain the mortality benefit in mechanically ventilated ARDS [
1] and COVID-19 patients [
33] undergoing prone positioning [
34]. In non-intubated critically ill COVID-19 patients, APP may delay intubation due to temporary improvements in oxygenation [
25] which could paradoxically lead to patient self-inflicted lung injury (P-SILI) [
35,
36]. Although not statistically significant, there were more deaths in the prone group compared with controls raising concerns of harm associated with awake prone positioning. However, time to intubation was similar between groups, and P-SILI thus may not be a relevant mechanism contributing to mortality in our population. A recently published meta-analysis of non-randomized cohort studies found no difference in mortality or duration of invasive mechanical ventilation in critically ill patients with COVID-19 who were intubated early compared with patients who were intubated late [
37]. Although further research is warranted, this could suggest that harm associated with early intubation outweighs harm associated with P-SILI in delayed intubation or that P-SILI may not be an important mechanism in COVID-19 [
38].
Among intubated patients in our study, there were more ventilator-free days in the prone group compared with the control group and among non-intubated patients there were more days free from NIV/HFNO in the prone group compared with the control group. These differences were not statistically significant, possibly due to low statistical power. Reducing duration of invasive and noninvasive respiratory support is important, in particular in settings with resource shortage. Ventilator-free days or days free from any respiratory support including NIV and HFNO [
39] may thus be an appropriate primary outcome of future studies.
Contrary to intubated patients, where prone position increases the risk of pressure sores [
40], patients in the prone group had fewer pressure sores than patients in the control group in our study. Frequent changes in body position may have reduced the risk of lower back and gluteal pressure sores in the prone group. Pressure sores have been associated with higher mortality in ICU patients and constitute a considerable burden to healthcare systems [
41]. The large number of patients during the pandemic combined with the high proportion of pressure sores in the control group indicates that this may be a substantial problem and the present investigation highlights the need for protocolized mobilization in critically ill patients.
Strengths of the present study included the randomized multicenter design and the well-defined protocol for APP increasing generalizability and reproducibility. This trial was conducted during the second pandemic wave, and physicians, nurses and physiotherapists at the participating ICUs and wards gained extensive experience of prone positioning in non-intubated patients during the first wave, ensuring high-quality APP for included patients. No patients were lost to follow up and there was minimal missing data. As the first randomized clinical trial of prolonged APP in COVID-19, this trial provides important new information to bedside clinicians and for future studies.
There are also limitations to this trial. First, the trial was halted early resulting in limited statistical power to detect differences between groups. In particular, analyses of subgroups that may benefit from APP and analyses of secondary outcomes with few events may have been hampered and the results should therefore be interpreted with caution. Second, due to the nature of the intervention, blinding was not possible, increasing the risk of bias. Third, we included patients with moderate to severe respiratory failure and there was a liberal use of NIV in both groups early after enrollment. Our results may therefore not be generalizable to patients with less severe degrees of respiratory failure and settings where prolonged respiratory support with HFNO is standard care. Fourth as all study sites became overwhelmed by severely ill patients with COVID-19, and research staff was relocated for clinical service, we were not able to identify all patients eligible for inclusion. Fifth, APP was increasingly considered standard of care in COVID-19-related hypoxemic respiratory failure attenuating the difference in duration of APP between groups. Finally, we did not pre-define criteria for intubation or de-escalation from NIV/HFNO to low flow oxygen therapy. This was a pragmatic choice to minimize extra workload on clinicians and decrease risk of protocol compliance issues.
Acknowledgements
The authors wish to thank Elin Söderman, Joanna Wessbergh, Anna Granström, Anna Schening, Ola Friman, Pia Zetterqvist, Viveca Hambäck Hellkvist, Olivia Sand and David Stenstad for excellent technical and administrative assistance. We also thank the collaborators of the PROFLO Study Group: Anna Gradin, Mustafa Ali, Ulrica Lennborn, Darko Bogdanovic, Andreas Roos, Matilda Modie and Julia Giesecke.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.