Introduction
In March 2020, the numbers of hypoxic patients with COVID-19 pneumonitis admitted to Sheffield Teaching Hospitals began to rise at pace. Local audit data for the infectious diseases (ID) unit (with level 1 care capabilities) showed the number of patients requiring oxygen went from 0 to 13 then 22 patients (out of a 24 bed capacity), from early to mid then late March respectively. Optimal management with regards to other medical treatments were still being explored via research trials at this time. Finding efficient and effective treatment methods for patients with COVID-19 pneumonitis in order to prevent an intensive care unit (ICU) admission became paramount.
At this time adult respiratory distress syndrome (ARDS) was increasingly recognised as a major complication of COVID-19 pneumonitis, affecting up to 40% of patients (Wang et al. 2020; Wu et al. 2020). There was pre-existing data on the benefit of proning, where patients are assisted to lie on their front, in patients with ARDS arising from other conditions (Ding et al. 2020; Henderson et al. 2014; Pérez-Nieto et al. 2020); these studies were conducted in high acuity settings where patients were ventilated and sedated. According to the Intensive Care Society, the suggested physiological benefits of proning include improved ventilation and perfusion (V/Q) matching; reduced hypoxaemia; reduced shunting; recruitment of the posterior lung segments due to reversal of atelectasis and improved secretion clearance (Bamford et al. 2020).
Historically there was limited data on the efficacy of proning in patients who were not sedated and ventilated (Scaravilli et al. 2015), and this treatment had not previously been routinely used outside of ICU settings. However, towards the end of March there were increasing anecdotal reports on respiratory medical and physiotherapy social media forums of its use in people with COVID-19. A single article Sun et al. (2020) studying a Chinese cohort described the beneficial use of awake-proning – supporting awake patients to lie on their front – their rationale being that awake-proning reduced incidence of alveolar collapse and the ARDS-like picture emerging in COVID-19 pneumonitis. Importantly this awake-proning was commenced under the supervision of intensivists. They attributed their lower mortality rates compared to neighbouring provinces, in part, to this practice.
At this time there was high pressure on resources due to demand on hospital beds, staffing, equipment, personal protective equipment, oxygen supply and critical care beds, with an unknown timescale of how long the peak of the pandemic would last. Using the prone position for awake patients would be simple, cheap and, if effective, could potentially improve patient outcomes and decrease hospital length of stay.
After reviewing the available evidence across the multi-disciplinary team, it was decided to undertake a trial of awake-proning of patients with COVID-19 pneumonitis on the ID ward, supervised by physiotherapists and supported by ID medics rather than intensivists. Formal guidance from the Intensive Care Society regarding awake-proning was issued part way through this local trial (Bamford et al. 2020) – however a recommended length of time was not issued – just a suggested time of ‘as long as possible’ with timed position changes. By introducing awake-proning as an adjunct to the treatment of patients with COVID-19 on the ward, the aim was to reduce the number of ICU admissions and length of stay (LOS) by improving overall patient oxygenation. The objectives were to effectively initiate an intensive proning regime in a level 1 setting with a minimum of additional resources, given the constraints at the time. This paper describes a retrospective evaluation of this intervention.
After reviewing the available evidence across the multi-disciplinary team, it was decided to undertake a trial of awake-proning of patients with COVID-19 pneumonitis on the ID ward, supervised by physiotherapists and supported by ID medics rather than intensivists. Formal guidance from the Intensive Care Society regarding awake-proning was issued part way through this local trial (Bamford et al. 2020) – however a recommended length of time was not issued – just a suggested time of ‘as long as possible’ with timed position changes. By introducing awake-proning as an adjunct to the treatment of patients with COVID-19 on the ward, the aim was to reduce the number of ICU admissions and length of stay (LOS) by improving overall patient oxygenation. The objectives were to effectively initiate an intensive proning regime in a level 1 setting with a minimum of additional resources, given the constraints at the time. This paper describes a retrospective evaluation of this intervention.
Method
Setting
The evaluation took place in the ID unit of a large urban teaching hospital. It has a maximum capacity of 33 beds but at the time of the study, functional capacity was limited by staffing to 24 beds. To cope with the high rates of patients admitted with hypoxia, physiotherapy staffing was increased from normal levels (1 band 5), to 1 band 5 physiotherapist plus a further band 6 and band 7 physiotherapist.
Recruitment
Recruitment began over a 30-day period commencing 30th March 2020. Admissions were reviewed daily by the physiotherapy team to identify suitability for awake-proning.
Inclusion and exclusion
Inclusion criteria
All patients with confirmed or clinically suspected COVID-19 were considered. Priority was given to those with worsening early warning score (EWS); bilateral chest radiograph changes and severe lymphopaenia. Hypoxia was defined as an oxygen saturation of ≤94% (off supplemental oxygen) in patients with no background of CO2 retention.
Exclusion criteria
Patients with a respiratory rate >35; immediate need for intubation; agitation; systolic blood pressure <60mmHg; cardiac arrhythmia; unstable spine; recent thoracic or abdominal surgery or an inability to self-prone.
Intervention
Medical care was given as per standard of care during that time which included medication as part of the RECOVERY trial (The RECOVERY Collaborative Group 2021). For the first 7 days of recruitment, awake-proning was commenced on selected patients for 3 sessions of 30 – 90 minutes depending on tolerance. After successful outcomes for initial patients and local ICU units nearing capacity, regimes were intensified to include three dedicated time slots 10:30–12:00, 15:00–16:30, and 19:00–07:00 (next day).
All patients had a 30-minute direct 1-to-1 physiotherapist supervision at initiation. In this, initial observations including fraction of inspired oxygen (FiO2) respiratory rate (RR) and oxygen saturations (SpO2) were measured. The process was explained to the patient and they were assisted to find an initial comfortable prone position. Patients were monitored for 30 minutes and repositioned if acute clinical deterioration was observed. Observations were repeated after 30 minutes. At this point the patient was given a proning regime to follow, with additional support available from the ward nurses if required, or the decision made to stop if clinically appropriate. Patients were reviewed daily by physiotherapists – their adherence to the regime was noted and further recommendations made. If the patient was unable to tolerate the regime, modifications were either made to their positioning (including the introduction of additional pillows or use of side-lying) or to the time spent in the prone position.
Proning was continued until resolution of hypoxia, defined by maintaining SpO2 >94% on room air.
Cessation
Reasons for early cessation of proning within the first 72 hours (independent of clinical improvement) were recorded under 4 categories.
• Acute clinical deterioration (for example, witnessed deterioration during initial supervised 30 minutes).
• General clinical deterioration requiring escalation to level 2/3 care.
• Lack of patient engagement (not complying with regime when unsupervised).
• Patient discomfort or other side effects.
Data collection
Data were collected retrospectively using patient case notes and online e-observations charts in 4 categories:
• Patient demographics (age, sex, ethnicity).
• Clinical Characteristics (COVID-19 swab results, nadir lymphocyte count and co-morbidity). Patients were defined as co-morbid if they had a Charlston co-morbidity score of 1 or greater.
• Observations (FiO2, SpO2 and RR were measured on admission and immediately prior to commencement of awake-proning. FiO2 and SpO2 were repeated 30 minutes into proning).
• Outcomes ( minutes spent in prone position in first 72 hours, admission to ICU, LOS in days).
Analysis
Comparison of regime tolerance
Patients were divided into two groups based on their tolerance of the proning protocol. Patients who maintained full compliance with the regime, ceasing proning only due to clinical improvement, were assigned to group A. Patients who ceased proning early (within 72 hours) due to one of the four reasons given above were assigned to group B.
Statistical analysis was performed using the analysis ToolPak within Microsoft Excel (Excel version 14.0.7266.5000, 32 bit). Skewness of data was assessed using Microsoft Excel summary statistics function. Clinical characteristics and observations between groups A and B assuming a normal distribution were analysed using the student unpaired t-test. Given the sample size, Fisher’s exact test was used to compare proportions where appropriate between the two groups.
Two outcomes were measured for these groups – hospital length of stay in days (LOS) and admission to ICU. One patient with complex post-COVID-19 rehabilitation needs who was still an inpatient at the time of analysis was excluded from the length of stay analysis.
An ICU admission was necessary if the patient required non-invasive ventilation or intubation. ICU admission rates for the two groups are presented as percentages for comparison. Patients who underwent awake-proning following an ITU admission, without subsequent readmission to ITU, were not included in figures for admission to ITU post-proning.
Comparison with non-intervention
Given the nature of the study there was no fully matched comparator group or patients who did not receive the intervention. We therefore undertook additional analysis of patients admitted to the ID unit in the two weeks preceding this study, when awake-proning was not offered. Baseline characteristics, length of stay and ITU admissions of this group of patients were compared to those of the intervention group. In addition to the decision to prone, 30th March 2020 also marked the commencement of a geriatric pathway for COVID-19 admissions (so not all elderly COVID-19 patients were admitted to the ID ward). Given the potential for confounding due to this operational change and post-hoc nature of this analysis, gross figures are presented for interest.
Results
Over a period of 30 days, 104 patients were referred for physiotherapy of which 36 met the study inclusion criteria.
The proning regime was tolerated by 21 patients (58%), who were thus allocated to group A, and was not tolerated by 15 patients (42%) who were thus allocated to group B. The clinical and demographic characteristics of the 2 groups are shown in Table 1.
Of the 36 patients, median age across groups A and B groups was 54 years (IQR 48–62) and 25 (69%) were male. A third of patients were from BAME background. There was no significant difference in clinical and demographic characteristics between the 2 groups with regards to clinical status on admission, comorbidity and observations on hospital admission, see Table 1.
Patients selected for proning were tachypnoeic on admission – with mean respiratory rate (RR) 25, maintaining SpO2 of 96% on FiO2 36%. The mean time from symptom onset to commencement of regime was nine days.
Table 1: Clinical and demographic characteristics.
| |
Group A
(tolerated
proning)
(n = 21)
|
Group B
(did not
tolerate)
(n = 15)
|
All
(n = 36)
|
Significance
|
|
Demographics
|
|
Age, median (IQR), y
|
52 (42–61)
|
59 (50–62)
|
54 (48–62)
|
|
|
Male Sex – no. (%)
|
15 (71%)
|
10 (67%)
|
25 (69%)
|
|
|
BAME – no. (%)
|
8 (38%)
|
5 (33%)
|
13 (36%)
|
|
|
Clinical characteristics
|
|
Positive COVID-19 swab – no. ( %)
|
11 (52%)
|
12 (80%)
|
23 (64%)
|
p = 0.159*
|
|
Nadir lymphocytes – mean (range), ×109/L
|
0.82 (0.28–1.71)
|
0.76 (0.24–1.68)
|
0.68 (0.24–1.71)
|
p = 0.378**
|
|
Charlson co-morbidity index ≥1, no. (%)
|
10 (48%)
|
9 (60%)
|
19 (53%)
|
p = 0.516*
|
|
Observations on admission to hospital
|
|
SpO2, mean (range), %
|
96 (91–100)
|
96 (88–100)
|
96 (88–100)
|
p = 0.826**
|
|
FiO2, mean (range), %
|
37 (21–100)
|
32 (21–100)
|
36 (21–100)
|
p = 0.571**
|
|
Respiratory rate, mean (range), bpm
|
25 (19–49)
|
24 (18–36)
|
25 (18–49)
|
p = 0.563**
|
|
Observations at onset of proning regime
|
|
Day since symptom onset, mean (range)
|
8 (1–15)
|
9 (4–17)
|
9 (1–17)
|
p = 0.378**
|
|
SpO2, mean (range), %
|
95 (92–98)
|
96 (91–99)
|
95 (91–99)
|
p = 0.323**
|
|
FiO2, mean (range), %
|
41 (21–100)
|
44 (24–100)
|
42 (21–100)
|
p = 0.650**
|
|
Respiratory rate, mean (range), bpm
|
23 (18–30)
|
24 (18–32)
|
24 (18–32)
|
p = 0.457**
|
* – 2 tailed Fisher’s Exact test; ** – unpaired student t-test.
BAME: black and minority ethnic; SpO2 – oxygen saturations; FiO2 – fraction of inspired oxygen; bmp – breaths per minute.
Table 2 shows the outcomes of the proning regimes. Over a 72-hour period the range of time patients tolerated awake-proning was 10–2790 minutes. The regime was tolerated by 21 patients (58%) who achieved a mean of 1898 minutes (31.6 hours) in a 72-hour period. The 15 patients in group B of who ceased the regime within 72 hours tolerated a mean of 971 minutes (16.2 hours) in that period.
Table 2: Proning outcomes.
| |
Group A
(tolerated
proning)
(n = 21)
|
Group B
(did not
tolerate)
(n = 15)
|
All
(n = 36)
|
Significance
|
|
Minutes proned/72 hour mean (range)
|
1898 (420–2790)
|
971(10–1980)
|
1512 (10-2790)
|
*p = 0.0007‡
|
|
Length of stay mean (range), days
|
7.2 (3–16)
|
17.1 (4–54)**
|
10.8 (3–54)
|
*p = 0.050‡
|
|
LOS/days (excluding patients admitted to ITU)
|
6.8 (3–16)
|
15.3 (4–54)
|
9.26 (3–54)
|
*p = 0.049‡
|
|
ITU admission – no. (%)
|
1 (5)
|
6 (40)
|
7 (19)
|
|
|
SpO2, mean (range), %
|
97 (95–100)
|
97 (88–100)
|
97 (88–100)
|
p = 0.754‡
|
|
FiO2, mean (range), %
|
37 (21–60)
|
40 (24–60)
|
38 (21–60)
|
p = 0.630‡
|
† – 2 tailed Fisher’s Exact test; ‡ – unpaired student t-test.
BAME: black and minority ethnic; SpO2 – oxygen saturations; FiO2 – fraction of inspired oxygen; bmp – breaths per minute.
**1 patient still inpatient at time of writing.
Reasons for failure to tolerate the proning regime are shown in Figure 1. Clinical deterioration (acute and general) accounted for 6 patients, 5 patients did not comply with the regime when not under direct supervision. Finally of the 4 patients who stopped proning early due to side effects, 2 stated intolerable back pain, 1 developed nosebleeds and 1 patient had exacerbation of migraines.
Figure 1: Outcomes of all patients proned with reasons for early cessation.
ICU admission was required in 5% of patients in group A required compared to 40% in group B. Statistically significantly greater length of stay was noted in group B compared to group A (17.1 compared to 7.2 days, p = 0.05). Given the possible confounding of ICU admission on this figure, even when patients admitted to ITU were excluded from the analysis, length of stay was statistically significantly lower in group A (15.3 compared to 6.8 days, p = 0.049) compared to group B.
There was no significant difference between SpO2 and FiO2 in group A and group B after the proning regime. However across both groups, mean SpO2 increased by 2 percentage points (from 95%–97%) between the onset of proning and when SpO2 was recorded after 30 minutes, and mean FiO2 fell from 42% to 38% across the same period. Figure 2a and 2b shows the change in oxygen saturation for individual patients in group A and group B across the intervention.
Comparison with non-intervention
When compared with a group of 14 patients admitted to the ward before the awake-proning regime was introduced, the group who underwent proning had a reduced length of stay (10.8 days compared to 18.6) and reduced ICU admission rates (19% compared to 29%). Data for these two groups is shown in Table 3.
Table 3: Characteristics of patients before and after introduction of awake-proning.
| |
Admitted prior to
proning regime
(n = 14)
|
Included in proning regime
(n = 36)
|
|
Demographics
|
|
Age, median (IQR), y
|
74 (63–83)
|
54 (48–62)
|
|
Male sex – no. (%)
|
9 (64%)
|
25 (69%)
|
|
Clinical characteristics
|
|
Positive COVID-19 swab – no. ( %)
|
13 (93%)
|
23 (64%)
|
|
Nadir lymphocytes – mean (range), ×109/L
|
0.75 (0.08–1.75)
|
0.68 (0.24–1.71)
|
|
Observations on admission to hospital
|
|
SpO2, mean (range), %
|
94 (90–97)
|
96 (88–100)
|
|
FiO2, mean (range), %
|
27 (21–40)
|
36 (21–100)
|
|
Respiratory Rate, mean (range), bpm
|
21 (17–28)
|
25 (18–49)
|
|
Longer term outcomes
|
|
Length of stay mean (range), days
|
18.6 (6–49)
|
10.8 (3–54)
|
|
ICU admission – no. (%)
|
29%
|
19%
|
Discussion
An intensive awake-proning regime in hypoxic COVID-19 patients was tolerated by just over half the patients. In the first 72 hours of initiation, patients who were able to tolerate the intensive regime achieved 31.6 hours in the prone position (1898 minutes). Only 1 patient demonstrated acute deterioration in SpO2 on proning and required almost immediate intubation (see Figure 2 – group B).
There was an association between reduced length of stay for those who tolerated an intensive proning regime, together with lower rates of admission to ICU. Awake-proning was shown to successfully increase mean oxygen saturations by 2 percentage points after 30 minutes with a small reduction in FiO2, regardless of how long proning was eventually sustained. The following month Thompson et al. (2020) followed a cohort of 29 patients in which oxygen saturations were measured one hour post proning – they demonstrated a higher increase in SpO2 post proning (7%) although similar to our study, data as to whether there was a sustained improvement in SpO2 were unavailable. In addition, they did not report on length of time spent in the prone position, in their study 13 out of the 29 patients went on to be intubated. We postulate the increase in SpO2 is to be physiologically expected due to better recruitment of the posterior lung segments. However similar to the Thompson et al. (2020) study it is unclear whether this improvement in oxygen saturations is sustained and/or contributory to the lower ICU admission rates observed. Given patients in group B were in part, defined by their clinical deterioration, it is difficult to unpick the role awake-proning played as opposed to other factors such as ethnicity, nature of co-morbidity (for example, diabetes) and other medications given including steroids and antivirals.
A systematic review conducted by Anand et al. (2021) reviewed published data on awakeproning in COVID-19 up to July 2020 and reviewed 210 cases. All studies were case reports, case series or prospective cohort studies. Duration of proning length varied in studies, with intense regimes (>10 hours daily), limited to case reports only. Intubation rates across all cases were 23%. Our findings correlated with others with regards to improvement in oxygen saturations whilst proning – our general intubation rates were lower although overall small sample sizes in all reported studies make it difficult to draw any conclusions regarding this.
Limitations
This was an evaluation of service and therefore there was no predesigned control group. Monitoring and patient observations were limited to those routinely collected during ward-based care. This data were collected at a time when optimal medical management for COVID-19 patients was still under investigation, and treatments which would now be recognised as the standard of care (dexamethasone, remdesivir, tocalizumab) were given to only certain patients as part of the RECOVERY trial.
Conclusion
Awake-proning on a level 1 ward is an effective intervention that was deliverable with a minimum of additional specialist staff input and resulted in an initial increase of oxygen saturations. We have demonstrated that an intensive regime of proning (with a morning, evening and overnight session) is achievable in the majority of patients. We have demonstrated that ICU admission rates and length of stay were significantly lower in patients who tolerated an intensive proning regime. Subsequent to this evaluation, our trust has issued guidance to allow other healthcare professionals (for example, doctors and nurses) to initiate awake-proning on the wards. Further work, ideally through a randomised control trial, needs to be undertaken in a level 1 setting to fully assess the true benefit of awake-proning.
Key points
• Intensive (>10 hours/day) awake-proning is achievable in over 50% of hypoxic COVID-19 patients in a level 1 setting.
• This intervention can be effectively administered in a level 1 setting, but requires additional dedicated physiotherapist support.
• There is an association between tolerating intensive awake-proning and reduced length of stay which needs further exploration.
Funding
This project received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Ethical approval
This evaluation was conducted under the service evaluation framework of Sheffield Teaching Hospitals. Clinical consent was obtained from patients for all interventions.
