Study Selection
The study selection process is illustrated in Fig. 1. A total of 1,426 records were identified using the search strategy on July 9, 2021, and 334 duplicate records were removed before the screening. Ninety-seven records were updated through monthly searches until February 10, 2022. Of the 1,179 records, 1,066 were excluded after screening using titles and abstracts. Subsequently, the full texts of the 113 reports were retrieved. After reviewing the eligibility of the original texts, 4 RCTs [18-21] and 16 cohort studies [5-9, 22-32] were included in our review. The list of excluded studies and reasons for exclusion are presented (Additional File 3).
The characteristics of the included studies are summarized in Table 1. Four RCTs [18-21] included 1,283 patients (sample size range, 27–1,121), six prospective cohort studies [5, 7, 22-25] 729 patients (sample size range, 38–335), and 10 retrospective cohort studies [6, 8, 9, 26-32] 2,865 patients (sample size range, 20–827). Nine studies [6, 8, 20, 21, 23, 26, 30-32] originated in Europe, four [5, 19, 24, 25] from Asia, three [22, 28, 29] from South America, two [9, 27] from the US, one [7] from Africa, and one [18] from multiple countries. Eight studies [5, 7, 9, 20, 25-28] were conducted at single centers and others at multiple centers. In most of the studies, including the four RCTs [18-21], patients were provided through the COT, HFNC, or NIV, and in two cohort studies [9, 31], through mechanical ventilation. The proning protocols varied in terms of time and frequency of sessions, such as at least 2–12 hours per day or no restrictions in time and frequency. In half of the studies, the prone position was used only in the ICU setting. [5-7, 9, 19, 22, 23, 26, 31, 32] The reported proning durations varied. The average proning time per day (3-15 hours per day), [18-22, 24, 30, 31] the total number of proning session (3-4 sessions), [6, 26] or days in proning (2.5-6.9 days) [6, 21, 22, 30, 31] were reported.
Risk of Bias in Studies
All RCTs were assessed as having a low risk of bias in all the dimensions. In more than half of the cohort studies, the domains of the possibility of target group comparison and selection were rated as having a high risk of bias (Additional File 4 Fig. S1). However, serious problems did not occur because the domains of exposure measurement, blinding of assessors, outcome assessment, and selective outcome reporting were assessed as having a low risk of bias in most cohort studies.
Primary outcomes
Mortality
Three RCTs [18, 19, 21] reported mortality rates. Mortality was similar between the prone and non-prone groups (n=1,256, RR 0.94, 95% CI 0.68 to 1.32, P=0.73, I2=6%, moderate certainty of evidence; Fig. 2).
Fifteen cohort studies [6-9, 22-32] reported mortality. The prone position reduced mortality compared to the non-prone position (n=3,514; RR 0.67, 95% CI 0.52 to 0.85, P=0.001, I2=74%, very low certainty of evidence; Fig. 3).
Mortality of intubated group
Mortality did not differ between the prone and non-prone groups (Fig. 3).
Mortality of Non-intubated group
The prone position had a significant advantage in the non-intubated patient group (13 studies, n=2,519; RR 0.59, 95% CI 0.47 to 0.76; P<0.0001; I2=48%). In the subgroup of non-intubated patients, prone reduced mortality compared to non-prone in the nasal cannula or facial mask group (five studies, n=1,261; RR 0.57, 95% CI 0.48 to 0.68, P<0.00001, I2=0%) and the HFNC or NIV group (five studies, n=1,019; RR 0.52, 95% CI 0.31 to 0.86, P=0.01, I2=46%) (Additional File 4 Fig. S2).
Although the funnel plot for mortality in cohort studies was asymmetric, we observed no evidence of publication bias in Egger's linear regression test (p= 0.0813, Additional File 4 Fig. S3).
Need for Intubation
Three RCTs [18, 19, 21] reported intubation rates. The intubation rate was significantly lower in the prone position than in the prone position (n=1,256; RR 0.83, 95% CI 0.71 to 0.97, P=0.02; I2=0%; high certainty of evidence; Fig. 4).
Twelve cohort studies [6, 7, 22-30, 32] reported intubation rates. There was no difference in intubation rate (n=2,384, RR 0.76, 95% CI 0.56 to 1.04, P=0.09, I2=77%; Fig. 5, very low certainty of evidence). In the subgroup analysis for the oxygen delivery method (the nasal cannula or facial mask group and the HFNC or NIV group), the prone position showed a significant advantage in the HFNC or NIV group (five studies, n=1,047; RR 0.64, 95% CI 0.44 to 0.94, P=0.02, I2=71%; Additional File 4 Fig. S4).
Although the funnel plot for the intubation rate of cohort studies was asymmetric, we observed no evidence of publication bias in Egger's linear regression test (p=0.4806, Additional File 4 Fig. S5).
Adverse events
All RCTs [18-21] reported adverse events. The incidence of cardiac arrest (at any time) was similar between the prone and non-prone positions (prone vs. non-prone 3/564 vs. 1/557, p value not reported) [18], and skin breakdown and vomiting were also similar between the two groups (moderate certainty of evidence, Additional File 4 Fig. S6). Six cohort studies [5, 9, 24, 28, 30, 32] reported adverse events in the prone group, which were mainly mild (very low certainty of evidence, Additional File 4, Table S1).
The GRADE summary of findings table of primary outcomes is reported in Table 2.
Secondary outcomes
Length of stay in hospital or ICU
Two RCTs [19, 21] (n=120, MD -2.05 days, 95% CI -9.45 to 5.35 days, P=0.59, I2=82%) and six cohort studies [7, 22, 23, 26, 31, 32] (n=1,440, MD 0.69 day, 95% CI -2.38 to 3.76 days, P=0.66, I2=85%) reported LOS in ICU. Two RCTs [18, 21] (n=1,196, MD -0.23 day, 95% CI -1.37 to 0.92 days, P=0.70, I2=0%) and six cohort studies [5, 8, 9, 22, 25, 32, 33] (n=985, MD 3.63 days, 95% CI -3.09 to 10.35 days, P=0.29, I2=93%) reported LOS in hospital. The prone position was not associated with shorter LOS in the ICU or hospital than in the non-prone position (Additional File 4 Fig S7-10).
ICU-free days and Ventilator-free days
One RCT [21] with 75 patients reported ICU-free days and ventilator-free days, which were similar between prone and non-prone positions (prone vs. non-prone (median, IQR) ICU-free days 25 (14-28) days vs. 26 (8-30) days, p=0.56; ventilator-free days 30 (12–30) days vs. 30 (11–30) days, p=0.69). Three cohort studies [22, 28, 31] reported ventilator-free days. There was no difference between prone and non-prone positions (n=1,235, MD -1.91days, 95% CI -5.51 to 1.69 days, p=0.3, I2=90%). However, in the subgroup of intubated patients, ventilator-free days in the prone position group were shorter than in the non-prone position group (one study, n=734, MD -4.49, 95% CI -6.23 to -2.75 days, p<0.00001; Additional File 4 Fig. S11).