Self-proning in COVID-19 patients on low-flow oxygen therapy: a cluster randomised controlled trial.

RATIONALE AND
OBJECTIVES: Prone positioning as a complement to oxygen therapy to treat hypoxaemia in coronavirus disease 2019 (COVID-19) pneumonia in spontaneously breathing patients has been widely adopted, despite a lack of evidence for its benefit. We tested the hypothesis that a simple incentive to self-prone for a maximum of 12 h per day would decrease oxygen needs in patients admitted to the ward for COVID-19 pneumonia on low-flow oxygen therapy.
METHODS: 27 patients with confirmed COVID-19 pneumonia admitted to Geneva University Hospitals were included in the study. 10 patients were randomised to self-prone positioning and 17 to usual care.
MEASUREMENTS AND MAIN RESULTS: Oxygen needs assessed by oxygen flow on nasal cannula at inclusion were similar between groups. 24 h after starting the intervention, the median (interquartile range (IQR)) oxygen flow was 1.0 (0.1-2.9) L·min-1 in the prone position group and 2.0 (0.5-3.0) L·min-1 in the control group (p=0.507). Median (IQR) oxygen saturation/fraction of inspired oxygen ratio was 390 (300-432) in the prone position group and 336 (294-422) in the control group (p=0.633). One patient from the intervention group who did not self-prone was transferred to the high-dependency unit. Self-prone positioning was easy to implement. The intervention was well tolerated and only mild side-effects were reported.
CONCLUSIONS: Self-prone positioning in patients with COVID-19 pneumonia requiring low-flow oxygen therapy resulted in a clinically meaningful reduction of oxygen flow, but without reaching statistical significance. ©The authors 2021.

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-associated pneumonia is associated with severe hypoxaemic respiratory failure requiring treatment in high-dependency or intensive care units (ICUs) in ∼5–10% of hospitalised patients [1, 2]. Given the rapid increase of cases during the recent pandemic, many high-dependency units and ICUs have been overwhelmed in their capacity to provide care [1, 3]. In addition, several pharmacological agents for the treatment of SARS-CoV-2-associated pneumonia remain of uncertain benefit or have been associated with potentially life-threatening side-effects [4]. In patients hospitalised in a medical ward with a diagnosis of coronavirus disease 2019 (COVID-19) pneumonia, any simple intervention to limit the progression of hypoxaemia and avoid transfers of patients to critical care units for mechanical ventilation may be of benefit for the management of hospital resources.

Lung-protective mechanical ventilation and intermittent prone positioning with neuromuscular blockade are standard care and evidence-based strategies in the management of severe acute respiratory distress syndrome (ARDS) [5-7]. Use of low tidal volume ventilation (4–8 mL·kg−1 of predicted weight) targeting a plateau pressure <30 cmH2O, with high positive end-expiratory pressure and prone mechanical ventilation for 12–16 h·day−1 has been integrated into the Surviving Sepsis Campaign guidelines for the management of critically-ill adults with COVID-19 [8]. The rationale behind the prone position is to reduce ventilation/perfusion mismatch and thus hypoxaemia. The prone position decreases the pleural pressure gradient between dependent and nondependent lung regions, which is believed to generate a more homogeneous lung ventilation in ARDS patients [9]. As the prone position does not appear to alter blood flow distribution, a subsequent reduction in shunting might be observed [10].

At present, no published trials have documented the effect of the prone position in awake patients with COVID-19 pneumonia. Case series suggest that the prone position in awake patients treated with high-flow nasal oxygen therapy or noninvasive ventilation is feasible, easier to perform than in heavily sedated, more severely ill patients, and is not associated with major side-effects [11-16]. However, it remains unknown whether prolonged periods of prone position in patients admitted for COVID-19 pneumonia on low-flow oxygen therapy are associated with a persistent improvement in peripheral oxygen saturation (SpO) and lower needs of oxygen. We designed this single-centre, cluster randomised controlled trial to test the hypothesis that the prone position is associated with lower needs of oxygen in patients admitted to the medical ward for COVID-19 pneumonia.

Methods

Study design and participants

We conducted a single-centre cluster randomised controlled trial in six medical wards in Geneva University Hospitals (Geneva, Switzerland). As the intervention (incentive to self-prone) was not blinded and delivered by physicians and nurses involved in patient monitoring during the COVID-19 pandemic, a cluster randomised controlled trial design was chosen to minimise contamination between groups (i.e. to prevent patients in the control group from receiving the intervention if admitted to the same ward as those in the intervention group). Inclusion criteria were patients aged ≥18 years admitted to a medical ward for treatment of COVID-19 pneumonia with low-flow oxygen therapy (defined as 1–6 L·min−1) through nasal cannulas to obtain a SpO level of 90–92%. Exclusion criteria were patients initially treated in the ICU or high-dependency unit and recovering from ARDS; those with oxygen needs >6 L·min−1 using a nasal cannula or with >40% inspiratory oxygen fraction (FiO) using a Venturi mask to obtain a SpO level of 90–92%; pregnant women; terminally ill patients; and those unable to self-prone. Patients were screened by a daily review of admissions to each ward.

Randomisation

The randomisation unit was a medical ward in the division of internal medicine of our hospital with a 15-bed capacity. Six clusters were selected and a computer-generated randomisation scheme was used to assign each medical ward randomly in a 1:1 ratio to either the intervention or usual care. After April 14, 2020, most wards dedicated to the care of COVID-19 pneumonia gradually closed because of effective COVID-19 containment measures and a favourable evolution of the epidemic in our region. Four more patients were individually randomised by the computer-generated programme in the wards which remained open. From 25 April to 29 May 2020, no further eligible patients were admitted to the ward for COVID-19 pneumonia and we decided to close enrolment, despite not having reached the number required by our sample size calculation.

Intervention

We compared an add-on to usual care versus usual care alone. Usual care consisted of 1) oxygen titration with nasal cannula according to our institutional recommendations to target SpO values between 90% and 94%. Nurses carried out at least six routine rounds per 24 h to monitor oxygen needs and adapt oxygen flow to the prescribed SpO target; 2) empirical antibiotics for community-acquired pneumonia; 3) an association of hydroxychloroquine and lopinavir/ritonavir as proposed by our institutional guidelines; and 4) a restrictive fluid strategy. Regarding the intervention, an intern (CC) and a resident (AK) from the division of lung diseases promoted self-proning for 12 h per day as an addition to usual care for 24 h. After an initial demonstration with the study investigators, all patients were given an explanatory brochure with photographs of the prone position and it was suggested that they use their mobile phone “timer” function to alternate their body position every 4 h. Nurses regularly visited patients to encourage them to change their bed position during their rounds. Vital signs were recorded after 24 h and patients answered a brief survey on tolerance and estimated time of prone positioning.

Data collection and study outcomes

Oxygen flow (L·min−1), estimated FiO (%), SpO, respiratory rate and heart rate were retrieved directly from the institutional electronic patient health record. Transfers to critical care units or home discharge were recorded. Time spent in the prone position was self-reported in a diary. SpO and other vital signs were recorded at 24 h when the patient was supine at rest for 1 h. SpO was recorded after its value had stabilised for ≥1 min. The pre-specified primary outcome was oxygen needs assessed by nasal cannula oxygen flow at 24 h. Secondary outcomes were the SpO/FiO ratio (defined as SpO percentage divided by the FiO) at 24 h [17], respiratory and heart rate at 24 h, patient trajectory (transfer to critical care unit) and potential intervention-related adverse effects as defined by neck pain, position-related discomfort and gastro-oesophageal reflux.

Statistical analyses

Continuous variables were summarised as medians and interquartile ranges (IQR) and categorical variables as numbers and percentages. Differences between groups were assessed using the Mann–Whitney–Wilcoxon test for continuous outcomes.

Sample size estimate

We based our sample size estimation on a preliminary unpublished observation in 20 patients admitted to the respiratory wards for COVID-19 pneumonia on low-flow oxygen therapy. In these patients, prone position for 15 min was associated with an immediate improvement in SpO, allowing a decrease in oxygen flow by 1 L·min−1 with a standard deviation of 1 L·min−1. Flow meters used in our institution for oxygen therapy allow oxygen flow to be read with a precision of 0.5 L·min−1. Additionally, we considered that a treatment effect of 1 L·min−1 would be clinically relevant for triage strategies in an overwhelmed healthcare system. To show a difference of 1 L·min−1 of oxygen flow with a standard deviation of 1 L·min−1 in an individually randomised trial with a two-sided significance level of 0.05 and a power of 0.8, enrolment of 32 patients would be needed. To take into account the correlation between patients of the same medical ward, the sample size was multiplied by a design effect of 2.4 corresponding to an intraclass correlation coefficient of 0.1 and a number of patients per ward equal to 15. Therefore, enrolment of 76 patients would have been required.

Analyses were performed with R statistical language [18].

Ethics

The institutional ethics review committee approved the trial (CCER 2020–00705). The study was registered on the Swiss National Clinical Trial portal (SNCTP000003718). All participants provided written informed consent before screening.

Results

Seven medical wards were approached to participate in the trial and six wards were randomised in a 1:1 ratio to the intervention or usual care. From April 6 to April 25, 2020, 54 patients were screened and 27 were enrolled in the trial. Causes for noninclusion were 1) refusal to participate (n=19) and 2) impossibility of self-proning due to morbid obesity, hemiplegia or cervical minerva (n=5); and 3) end-of-life support care (n=3). 10 patients were randomised to self-prone and 17 to usual care (figure 1). Baseline characteristics are described in table 1. Mean±sd age of participants was 58±12 years; 10 (37%) out of 27 were female. Among the participants, 12 (44%) out of 27 had hypertension, five (19%) out of 27 had diabetes, and one patient had chronic kidney disease. Time from first symptoms to inclusion was 10.5±5.1 days.

FIGURE 1

Study flowchart. ICU: intensive care unit; COVID-19: coronavirus disease 2019.

TABLE 1

Clinical characteristics of the study population

	Whole population	Self-proning	Usual care
Patients	27	10	17
Male	17 (63)	6 (60)	11 (65)
Age years	58±12	54±14	60±11
Body mass index kg·m−2	28.2±4.7	29.7±5.3	27.3±4.2
Comorbidities
Hypertension	12 (44)	3 (30)	9 (53)
Diabetes	5 (19)	2 (20)	3 (18)
Chronic kidney disease	1 (4)	0	1 (6)
Self-reported heart disease	0	0	0
COPD	0	0	0
Time onset of symptoms until inclusion days	10.5±5.1	10.6±5.1	10.5±5.3
Treatment received
Azithromycin	2 (7)	1 (10)	1 (6)
Hydroxychloroquine	19 (70)	6 (60)	13 (77)
Lopinavir/ritonavir	15 (56)	5 (50)	10 (59)

Data are presented as n, n (%) or mean±sd.

Estimated self-prone time was 295±216 min in the self-prone group and 7±29 min in the control group (due to a single patient who spent an estimated time of 120 min in the position). At baseline, median (IQR) oxygen flow on a nasal cannula was 2.5 (2.0–3.0) L·min−1 in the self-prone group and 2.0 (1.0–3.0) L·min−1 in the control group. At 24 h, median (IQR) oxygen flow was 1 (0.1–2.9) L·min−1 in the self-prone position group and 2.0 (0.5–3) L·min−1 in the control group (p=0.507). This corresponded to a median (IQR) SpO/FiO2 ratio of 390 (303–432) in the self-prone group at 24 h compared to 336 (294–423) in the control group (p=0.633) (figure 2). Changes of oxygen flow and SpO/FiO ratio for individual patients are shown in supplementary figure 1A and B). Main and secondary physiological end-points are presented in table 2. Median respiratory rate decreased with the intervention, whereas no effect was observed for heart rate. One patient randomised to the self-prone position was admitted to the high-dependency unit because of increased oxygen needs versus none in the usual care group. This patient was a 45-year-old male with a body mass index of 27.8 kg·m−2 without known comorbidities. He had an estimated prone position time of 6 min over 24 h and a reported side-effect of mild discomfort. Five (50%) other patients in the intervention group reported intervention-related adverse events, mainly mild position-related discomfort. No other intervention-related side-effects were reported.

FIGURE 2

a) Oxygen (O2) flow on nasal cannula in the self-proning group and in the control group; b) peripheral oxygen saturation (SpO)/inspiratory oxygen fraction (FiO) ratio in the self-proning group and in the control group. Data are presented as median, interquartile range and 90th and 10th percentile.

TABLE 2

Primary and secondary outcomes

	Self-proning	Usual care	Difference between groups (95% CI)
Patients	10	17
O2 nasal flow L·min−1
At baseline	2.5 (2.0–3.0)	2.0 (1·0–3·0)
At 24 h	1 (0.1–2.9)	2.0 (0.5–3.0)	−1 (−2.75–2)
SpO2/FiO2 ratio
At baseline	318 (284–341)	336 (303–388)
At 24 h	390 (303–432)	336 (294–422)	54 (−91.6–133.0)
Respiratory rate breaths·min−1
At baseline	22.0 (20.0–25.8)	20.0 (16.0–26.0)
At 24 h	20.0 (17.3–22.8)	20.0 (18–24.0)	0 (−6.5–3.5)
Heart rate beats·min−1
At baseline	83 (71–96)	82 (75–89)
At 24 h	83 (72–89)	80 (70–86)	3 (−13–15)

Data are presented as n or median (interquartile range), unless otherwise stated. The difference between medians of the two randomised groups have been computed with their 95% confidence interval obtained by bootstrap using 1000 replications. O2: oxygen; SpO: peripheral oxygen saturation; FiO: inspiratory oxygen fraction.

Discussion

In this cluster randomised trial, self-prone positioning in patients admitted for COVID-19 pneumonia requiring low-flow oxygen therapy appeared to be effective in decreasing oxygen needs at 24 h. A clinically meaningful reduction of oxygen flow and an improved SpO/FiO ratio were observed, although they did not reach statistical significance. With an unprecedented number of ill patients in a small geographical area and the risk of overwhelming local health resources, a reduction of oxygen flow by 1 L·min−1 could be of importance to select stable patients for home discharge with an oxygen supply or to prevent unnecessary or premature transfers to intermediate care units.

The intervention consisted of a simple incentive to self-prone for 12 h over a period of 24 h. Invitation to self-prone was easy to implement after an initial demonstration and distribution of an explanatory brochure and resulted in a substantial time spent in this position. The intervention was well tolerated and only mild adverse events were reported. Our results are in line with published case series and expand current knowledge on the prone position in awake patients with hypoxaemic respiratory failure associated with COVID-19 pneumonia [12-16]. Prone positioning is believed to improve hypoxaemia by generating a more homogeneous lung ventilation without altering blood flow distribution [9, 10], as illustrated by data from our trial.

In this unique pandemic situation, health professionals have often been forced to provide immediate medical assistance rather than generating reliable data from randomised trials to inform clinical practice. Awake prone positioning has been widely adopted by physicians around the globe [19] and proposed in conscious COVID-19 patients by the UK Intensive Care Society, but without strong evidence [20]. Such a recommendation may discourage the scientific community to run trials, although most professional bodies emphasise the need for higher quality evidence [21, 22]. Therefore, we specifically focused this randomised trial on a selected population of nonsevere COVID-19 patients with no therapeutic limitations who could all be admitted at any time to the ICU for mechanical ventilation in the event of clinical deterioration. The main explanation for not reaching statistical significance is a small sample size, probably related to the early interruption of study enrolment. Indeed, a very sharp decrease in COVID-19-related admissions was observed from mid-April 2020 as a result of effective containment measures in Switzerland. The results of this trial are promising, but adequately powered trials are still needed. Our data are in agreement with previous physiological studies and observational reports on prone positioning [11–16, 23].

Our study has some additional limitations. The intervention and assessments of end-points were limited to a 24-h time frame. Therefore, it is not possible to assess medium-term effects on outcomes and follow-up of self-prone positioning. Moreover, according to recent published reports on prone positioning, the effect on oxygenation is transient [14, 15]. As assessment at 24 h was performed in the supine position, the effect of the intervention on oxygen needs could have been minimised, although our data suggest that alternating supine and prone position over 24 h may be associated with lower oxygen needs at 24 h, even in the supine position. Finally, follow-up time in the medical ward was very short and the oxygen needs of patients with acute respiratory failure related to COVID-19 pneumonia should be closely monitored for >24 h, as rapid clinical deterioration is well described in a time window of 7–10 days after the onset of first symptoms [2, 24].

In summary, self-prone positioning in patients with COVID-19 pneumonia requiring low-flow oxygen therapy showed a reduction of oxygen needs in our study, which did not reach statistical significance, probably due to a small sample size and insufficient statistical power. However, the observed reduction of oxygen needs at 24 h is clinically promising without any reported major side-effects. Our findings need to be corroborated by larger randomised trials to confirm the potential beneficial effects of self-prone positioning on oxygen needs. This information would be of particular interest for healthcare systems in low-income countries with a limited access to ICUs.

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Figure S1a 00692-2020.FigureS1a

Figure S1b 00692-2020.FigureS1b