Table 1: Baseline characteristics of Patients with COVID-19 under Prone Position or Usual Care
Characteristics
|
Prone position
N = 135
|
Usual care
N = 132
|
SMD
|
|
Age, mean (SD), y
|
58.4 (12.1)
|
59.2 (11.0)
|
0.065
|
|
Men
|
96 (71)
|
93 (71)
|
0.014
|
|
Body mass index, mean (SD), kg/m2
|
28.6 (4.2)
|
28.6 (4.7)
|
0.014
|
|
Medical history
|
Hypertension
|
26 (19.3)
|
35 (26.5)
|
0.173
|
|
Coronary heart disease
|
9 (6.7)
|
5 (3.8)
|
0.130
|
|
Type 1 diabetes
Type 2 diabetes
|
1 (0.7)
11 (8.1)
|
1 (0.8)
18 (13.6)
|
0.002
0.177
|
|
Chronic obstructive pulmonary disease
|
6 (4.4)
|
1 (0.8)
|
0.233
|
|
Asthma
|
11 (8.1)
|
9 (6.8)
|
0.1051
|
|
Cerebral vascular disease
|
2 (1.5)
|
1 (0.8)
|
0. 069
|
|
Dialysis
|
0 (0)
|
5 (3.7)
|
<0.0001
|
|
Liver cirrhosis
|
1 (0.7)
|
0 (0.0)
|
0.122
|
|
Immunosuppressiona
|
8 (5.9)
|
4 (3.0)
|
0.140
|
|
Treatment at inclusion for COVID-19
|
Corticosteroids
|
134 (99.3)
|
130 (98.5)
|
0.073
|
|
Preferential sleeping position as declared by patients
Prone position
Other
Don’t know
Missing data
|
13 (9.6)
107 (79.3)
8 (5.9)
7 (5.2)
|
14 (10.6)
105 (79.5)
4 (3.0)
9 (6.8)
|
0.032
0.007
0.140
0.069
|
|
SpO2 at inclusion, median (IQR), %b
|
96 (94–98)
|
96 (94–98)
|
0.019
|
|
Oxygen therapy at inclusion
|
Standard oxygen therapy, No. (%)c
Nasal high-flow oxygen therapy, No. (%)d
|
129 (96)
6 (4)
|
126 (95)
6 (5)
|
|
|
Flow of oxygen if standard oxygen therapy, median (IQR), L/min
|
3 (2–5)
|
4 (2–6)
|
|
|
PaO2, median (IQR), mmHge
|
71 (64–83)
|
71 (62–83)
|
|
|
PaCO2, median (IQR), mmHge
|
34 (32–37)
|
34 (32–37)
|
|
|
PaO2 to FiO2 ratio, median (IQR), mmHgf
|
178 (151–226)
|
173 (131–226)
|
|
|
Stratificationc
|
Body mass index ≥ 30 kg/m2 and SpO2 ≥ 95%
PaO2 to FiO2 ratio, median (IQR), mmHg
|
29 (21.5)
200 (177–244)
|
28 (21.2)
188 (159–218)
|
_
|
|
Body mass index ≥ 30 kg/m2 and SpO2 < 95%
PaO2 to FiO2 ratio, median (IQR), mmHg
|
13 (9.6)
162 (114–190)
|
13 (9.8)
139 (99–157)
|
_
|
|
Body mass index < 30 kg/m2 and SpO2 ≥ 95%
PaO2 to FiO2 ratio, median (IQR), mmHg
|
63 (46.7)
185 (162–235)
|
61 (46.2)
194 (149–250)
|
_
|
|
Body mass index < 30 kg/m2 and SpO2 < 95%
PaO2 to FiO2 ratio, median (IQR), mmHg
|
30 (22.2)
152 (114–185)
|
30 (22.7)
133 (113–215)
|
_
|
|
Abbreviations: SMD, standardized difference of means or proportions; SpO2, oxygen saturation measured by pulse oximetry; PaO2, partial pressure of oxygen in arterial blood; PaCO2, partial pressure of carbon dioxide in arterial blood; FiO2, fraction of inspired oxygen; SD, standard deviation; IQR, interquartile range
a Patients were considered immunosuppressed if they used long-term (>3 months) corticosteroids or other immunosuppressant drugs or had solid organ transplantation, a solid tumor requiring chemotherapy in the last 5 years, hematologic malignancy, or primary immune deficiency.
b SpO2 was measured after inclusion during a standardized test, see Methods.
c Supplemental oxygen was given via standard nasal prongs or face mask at a flow rate up to 15 L/min.
d Nasal high flow oxygen therapy was administered via large bore nasal cannula with a range of flow from 40 to 70 L/min in the prone position plus usual care group and from 30 to 60 L/min in the usual care alone group. The FiO2 range was from 0.4 to 1.0 in the prone position plus usual care group and from 0.8 to 1.0 in the usual care alone group.
e Blood gases were those sampled within 24 hr before inclusion; data were missing for 11 and 8 patients in the intervention and usual care groups.
f The FiO2 was estimated according to Wettstein et al.(32)
Table 2: Primary and Secondary Outcomes in the Prone Position and Usual Care Groups
|
No. of events/
total No. of patients
|
|
|
|
Prone Position
|
Usual
Care
|
Adjusted OR (95%CI) a
|
P-value
|
Primary outcome
|
|
|
|
|
Non-invasive ventilation or tracheal intubation, or death within 28 days
|
19/132
|
17/132
|
0.43 (0.14 to 1.35)
|
0.15
|
Secondary outcomes
|
|
|
|
|
Non-invasive ventilation within 28 days of enrolment b
|
12/135
|
8/132
|
4.86 (0.95 to 24.87)
|
0.057
|
Tracheal intubation within 28 days of enrolment
|
10/135
|
13/132
|
0.11 (0.01 to 0.89)
|
0.038
|
Tracheal intubation or death within 28 days of enrolment
|
10/135
|
14/132
|
0.09 (0.01 to 0.76)
|
0.027
|
Death within 28 days
|
0/135
|
4/132
|
_
|
_c
|
Rate of transfer to the ICU within 28 days of enrolment
|
21/135
|
20/135
|
1.04 (0.53 to 2.05)
|
0.91
|
Weaning from oxygen in hospital ward within 28 days
|
117/135
|
110/132
|
1.50 (0.71 to 3.17)
|
0.29
|
Death during hospitalisation
|
2/135 d
|
4/132
|
0.49 (0.09 to 2.74)
|
0.41
|
Do-not-intubate order after inclusion within 28 days e
|
1/135
|
2/132
|
_
|
_ c
|
Readmission in hospital after hospital discharge and within 28 days
|
2/135 f
|
0/132
|
_
|
_ c
|
|
Median (IQR)
|
Median difference (95%CI)
|
|
Duration of oxygen therapy in patients not transferred to the ICU or undergoing non-invasive ventilation or intubation, days
|
5 (3–8)
|
5 (3.5–8)
|
0 (-1 to 1)
|
.95
|
Length of hospitalisation, days
|
7 (5–11)
|
7 (7–12)
|
0 (-1 to 1.5)
|
.79
|
Abbreviations: OR, odds ratio; CI, confidence interval; ICU, intensive care unit
a: Odds-ratios were adjusted for stratification
b: Among the 20 patients who underwent non-invasive ventilation, 7 (5 in the prone position group and 2 in the usual care group) were not transferred to an ICU. Among the remaining 13 patients, 2 (1 in each group) underwent non-invasive ventilation the day before ICU admission, and 11 within the first 2 days of ICU admission.
c: There were too few events to allow for meaningful logistic regression analysis
d: Two patients died during their hospitalisation after day 28
e: All 3 patients died withing 28 days of inclusion
f: The 2 patients were readmitted to an ICU
Table 3: Logistic Regression Analyses of Prespecified Subgroups
|
No. of events/
total No. of patients
|
|
|
|
|
Prone Position
|
Usual
care
|
Adjusted OR (95%CI)
|
P-value
|
P for interaction
|
Primary outcome: Non-invasive ventilation or intubation, or death within 28 days a
|
Initial SpO2 ≥95%
|
5/92
|
10/89
|
0.44 (0.14 to 1.36)
|
0.15
|
0.019
|
Initial SpO2 <95%
|
14/43
|
7/43
|
3.09 (0.97 to 9.87)
|
0.057
|
|
BMI <30 kg/m2
|
13/93
|
10/91
|
1.39 (0.54 to 3.60)
|
0.47
|
0.51
|
BMI ≥30 kg/m2
|
6/42
|
7/41
|
0.81 (0.24 to 2.69)
|
0.73
|
|
Secondary outcomes
|
|
|
|
|
|
Noninvasive ventilation within 28 days of enrolment b
|
Initial SpO2 ≥95%
|
5/92
|
10/89
|
0.92 (0.21 to 4.09)
|
0.90
|
0.33
|
Initial SpO2 <95%
|
8/43
|
4/43
|
2.51 (0.65 to 9.72)
|
0.18
|
|
BMI <30 kg/m2
|
8/93
|
2/91
|
9.85 (1.29 to 75.00)
|
0.027
|
0.049
|
BMI ≥30 kg/m2
|
4/42
|
6/41
|
0.61 (0.16 to 2.39)
|
0.48
|
|
Tracheal intubation within 28 days of enrolment c
|
Initial SpO2 ≥95%
|
1/92
|
8/89
|
0.11 (0.01 to 0.90)
|
0.040
|
0.016
|
Initial SpO2 <95%
|
9/43
|
5/43
|
2.11 (0.62 to 7.18)
|
0.23
|
|
BMI <30 kg/m2
|
7/93
|
9/91
|
0.70 (0.23 to 2.11)
|
0.53
|
>0.99
|
BMI ≥30 kg/m2
|
3/42
|
4/41
|
0.72 (0.15 to 3.47)
|
0.67
|
|
Tracheal intubation or death within 28 days of enrolment d
|
Initial SpO2 ≥95%
|
1/92
|
9/89
|
0.09 (0.03 to 0.27)
|
0.027
|
0.011
|
Initial SpO2 <95%
|
9/43
|
5/43
|
2.11 (0.62 to 7.19)
|
0.23
|
|
BMI <30 kg/m2
|
7/93
|
9/91
|
0.71 (0.24 to 2.12)
|
0.53
|
0.79
|
BMI ≥30 kg/m2
|
3/42
|
5/41
|
0.55 (0.12 to 2.51)
|
0.44
|
|
Rate of transfer to ICU within 28 days of enrolment e
|
Initial SpO2 ≥95%
|
8/92
|
12/89
|
0.61 (0.23 to 1.57)
|
0.31
|
0.11
|
Initial SpO2 <95%
|
13/43
|
8/43
|
1.91 (0.69 to 5.25)
|
0.21
|
|
BMI <30 kg/m2
|
2/93
|
0/91
|
_f
|
_
|
_
|
BMI ≥30 kg/m2
|
0/42
|
0/41
|
_f
|
_
|
_
|
Abbreviations: OR, odds ratios; CI, confidence interval; SpO2, oxygen saturation measured by pulse oximetry; BMI, body mass index.
a There was no 3-way interaction (P=.72). Among the 3 possible 2-way interactions, only the interaction “intervention by initial SpO2 < or ≥ 95%” was significant (P=.021) and was kept in the model.
b There was no 3-way interaction (P>.99). Among the 3 possible 2-way interactions, only the interaction “intervention by BMI < or ≥ 30 kg/m2” was significant (P=.049) and was kept in the model.
c There was no 3-way interaction (P=.97). Among the 3 possible 2-way interactions, only the interaction “intervention by initial SpO2 < or ≥ 95%” was significant (P=0.02) and was kept in the model.
d There was no 3-way interaction (P=.93). Among the 3 possible 2-way interactions, only the interaction “intervention by initial SpO2 < or ≥ 95%” was significant (P=.01) and was kept in the model.
e There was no 3-way interaction (P=.76). There was no significant interaction among the 3 possible 2-way interactions (all P>.05). None of those interaction terms were introduced in the logistic model.
f There were too few events to allow for meaningful logistic regression analysis.