The search yielded a total of 2527 studies prior to the removal of 742 duplicates. After screening the titles and abstracts, another 1642 out of the
remaining 1785 studies were excluded. Another 143 studies were subsequently excluded following full-text review. The remaining 9 studies
underwent data extraction and quantitative analysis (Fig. 1).
Characteristics of the Studies Analyzed
A total of 896 patients were enrolled in 9 RCTs, 4 of which were multicenter RCTs, and 5 studies were blinded. The main intervention methods
were early in-bed cycling (n = 2), EGDM (n = 2), and early combined rehabilitation (n = 5). The majority of the studies were from the general ICU (n = 7), followed by the CCU (n = 1), SICU (n = 1) (Table 1)[15–17, 21–26]. There were no statistically significant differences in age (RR, 0.99; 95% CI [0.88, 1.11]; I2 = 40%; P = 0.85), gender(MD, − 0.18; 95% CI [− 1.51, 1.16]; I2 = 20%; P = 0.80), sepsis (RR 1.05; 95% CI [0.87, 1.25]; I2 = 0%; P = 0.63) or APACHE II (MD, 0.07 95% CI [− 0.55, 0.69]; I2 = 0%; P = 0.83) (Additional file 1: Fig. 2–5).
Table 1
Baseline characteristics of patients in the trials included in the meta-analysis. CCU = Coronary Care Unit, SICU = Surgical ICU, MV = Mechanical Ventilation, EGDM = Early Goal-Directed Mobilisation, APACHE II = Acute Physiology and Chronic Health Evaluation II.
Author | Study design | Seting | Population | Treatment regimen | Region | Year |
Hodgson et al [17] | A Multi-center Double-blinded RCT | ICU | MV ༜48 h | EGDM Usual care | Australia New Zealand | 2016 |
Kho et al [21] | A Multi-center Blinded RCT | ICU | MV ≥ 96 h | Early in-bed cycling Usual care | Canadian | 2019 |
Nickels et al [15] | RCT | ICU | MV༞48 h | Early in-bed cycling Usual care | Australia | 2020 |
Patel et al [22] | A Multi-center RCT | ICU | MV | Early combined rehabilitation Usual care | America | 2014 |
Schaller et al [23] | A Multi-center Double-blinded RCT | SICU | 24 ≤ MV༜72 h | EGDM Usual care | Austria Germany America | 2016 |
Schweickert et al [16] | Double-blinded RCT | ICU | 24 ≤ MV༜72 h | Early combined rehabilitation Usual care | America | 2009 |
Song et al [24] | RCT | CCU | MV and Cardiac disease | Early combined rehabilitation Usual care | China | 2019 |
Yu et al [25] | RCT | ICU | MV ≥ 24 h and Respiratory failure | Early combined rehabilitation Usual care | China | 2019 |
Zhu et al [26] | RCT | ICU | APACHE Ⅱ༞8 and SIRS | EGDM Usual care | China | 2018 |
The risk of bias assessment of 9 studies showed that 4 studies were low-risk, 5 studies had some possible risks of bias, and there were no high-risk studies. A breakdown of each individual study’s risk of bias per domain is shown in a risk of bias summary plot (Supplemental Digital Content-Fig. 6–7).
Early mobilization or rehabilitation effect on the prevalence of ICUAW (main outcome)
Nine RCT analyses showed that early mobilization or rehabilitation was associated with a reduced prevalence of ICU-AW (RR, 0.73; 95% CI [0.61, 0.87]; I2 = 44%; P = 0.0006) (Fig. 2). The heterogeneity could not be significantly reduced by performing a sensitivity analysis by removing each study one by one. A subsequent subgroup analysis based on intervention modalities revealed: early combined rehabilitation (RR, 0.56; 95% CI [0.43, 0.74]; I2 = 34%; P = 0.0001) could reduce the prevalence of ICUAW, while early in-bed cycling (RR, 1.25; 95% CI [0.73, 2.13]; I2 = 0%; P = 0.41) and EGDM (RR, 0.85; 95% CI [0.65, 1.09]; I2 = 26%; P = 0.20) could not reduce the prevalence of ICUAW (Fig. 3). Tests for subgroup differences I2 = 76.3%, and the mode of intervention was the source of heterogeneity. The publication bias had no statistical significance by Egger’s test (P = 0.193) and the funnel plot was basically symmetrical (Additional file 1: Figs. 10–11, Table 2).
Early mobilization or rehabilitation effect on the length of stay in the ICU (ICU-LOS)
An analysis of eight RCTS showed that early mobilization or rehabilitation did not reduce the ICU-LOS (MD, − 2.22; 95% CI [4.57, 0.14]; I2 = 92%; P = 0.07). Sensitivity analysis found that the study of Song et al. was a source of heterogeneity. After excluding this study, early activity or rehabilitation was associated with a shortening of the ICU-LOS (MD, − 1.47; 95% CI [2.83, 0.10]; I2 = 56%; P = 0.04) (Additional file 1: Figs. 12–13). Further subgroup analysis showed that EGDM (MD, − 2.27; 95% CI [3.86, 0.68]; I2 = 0%; P = 0.005) and early combined rehabilitation (MD, − 2.21; 95% CI [3.28, 0.97]; I2 = 23%; P = 0.0003) could reduce the ICU-LOS. Early in-bed cycling (MD, 2.27; 95% CI [0.27, 4.80]; I2 = 0%; P = 0.08) did not reduce the ICU-LOS (Fig. 4). Tests for subgroup differences I2 = 80.8%, and the mode of intervention was the source of heterogeneity. The publication bias had no statistical significance by Egger’s test (P = 0.494) (Additional file 1༚Figs. 15,Table 3).
Early mobilization or rehabilitation effect on the length of mechanical ventilation (MV)
Analysis of 8 RCTS showed that early mobilization or rehabilitation was associated with the length of MV (MD, − 2.28; 95% CI [3.81, 0.75]; I2 = 90%; P = 0.003). In the sensitivity analysis, it was found that the study by Song et al. was the source of heterogeneity, and after excluding this study, it was found that early mobilization or rehabilitation could reduce the length of MV (MD, − 1.96; 95% CI [2.41, 1.51]; I2 = 0%; 0.00001). The publication bias had no statistical significance by Egger’s test (P = 0.191) (Additional file 1: Figs. 16–18,Table 4).
Early mobilization or rehabilitation effect on the other outcomes
Early mobilization or rehabilitation was not related to mortality in the ICU (n = 5; RR, 0.90; 95% CI [0.62, 1.32]; I2 = 3%; P = 0.60) but is was related to an increase in the MRC sum score (n = 4; MD, 2.88; 95% CI [1.09, 4.67]; I2 = 36%; P = 0.002) and ICU mobility scale (n = 3; MD, 1.07; 95% CI [0.355, 1.79]; I2 = 0%; P = 0.004) (Additional file 1: Figs. 19–21).
Adverse events
None of the included studies reported serious adverse events related to the intervention that threatened the life of the patients. Five studies reported detailed adverse events, among which hypotension and fatigue were common, but the criteria were not uniform and further statistical analysis was not possible(Additional file 1: Table 5).