Physiotherapy 107 (2020) 1–10

Systematic review

Early rehabilitation reduces the likelihood of developing

intensive care unit-acquired weakness: a systematic review
and meta-analysis
David E. Anekwe a,b,c , Sharmistha Biswas d,e , André Bussières a,c ,
Jadranka Spahija a,b,c,∗
a School of Physical and Occupational Therapy, McGill University, Montreal, Quebec, Canada
b Research Center, CIUSSS du Nord-de-l’Ile-de-Montréal, Sacré-Coeur Hospital, Université de Montréal, Montréal, Quebec,
Canada
c Center for Interdisciplinary Research in Rehabilitation in Montreal, CISS du Nord-de-l’Île-de-Montréal, Jewish Rehabilitation

Hospital, Laval, Quebec, Canada

d Respiratory Epidemiology and Clinical Research Unit, Research Institute of the McGill University Health Centre, Montreal,

Quebec, Canada
e Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada

Abstract
Background Intensive care unit-acquired weakness (ICUAW) is associated with significant impairments in body structure and function,
activity limitation, and participation restriction. The etiology and management of ICUAW remain uncertain.
Objective To estimate the extent to which early rehabilitation interventions (early mobilization [EM] and/or neuromuscular electrical
stimulation [NMES]) compared to usual care reduce the incidence of ICUAW in critically ill patients.
Data sources We searched MEDLINE, EMBASE, CINAHL, Cochrane Central and Physiotherapy Evidence Database databases from
inception to May 1st, 2017.
Eligibility criteria Randomized controlled trials of EM and/or NMES interventions in critically ill adults.
Data extraction and data synthesis Data on the incidence of ICUAW and secondary outcomes were extracted. Both odds and risk ratios for
ICUAW were pooled using the random-effects model.
Results We identified 1421 reports after duplicate removal. Nine studies including 841 patients (419 intervention and 422 usual care) were
included in the final analysis. The interventions involved EM in five trials, NMES in three trials, and both EM and NMES in one trial. Early
rehabilitation decreased the likelihood of developing ICUAW: odds ratio of 0.63 (95% CI: 0.43 to 0.92) in the screened population, and 0.71
(95% CI: 0.53 to 0.95) in the randomized population.
Conclusion, implications of key findings Early rehabilitation was associated with a decreased likelihood of developing ICUAW. Our
findings support early rehabilitation in the ICU. While results were consistent in both the screened and randomized populations, the wide
confidence intervals suggest that well-conducted trials are needed to validate our findings.

Systematic review registration number PROSPERO registration ID: CRD42017065031.

© 2019 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.

Keywords: Early mobilization; Neuromuscular electrical stimulation; Intensive care unit acquired weakness; Mortality; Mechanical ventilation duration;
Discharge location

∗ Corresponding author at: CIUSSS du Nord-de-l’Ile-de-Montréal, Hôpital du Sacré-Cœur de Montréal, Axe de recherche en maladies chroniques,

5400 boul. Gouin Ouest, Montréal, Quebec, H4J 1C5, Canada.

E-mail address: jadranka.spahija@mcgill.ca (J. Spahija).

https://doi.org/10.1016/j.physio.2019.12.004
0031-9406/© 2019 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.
2 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10
Introduction
Intensive care unit (ICU) survivors often develop muscle
weakness which is unrelated to the primary pathology for
ICU admission [1]. This weakness may progress to a clini-
cal syndrome known as ICU-Acquired Weakness (ICUAW),
a “clinically detected weakness in critically ill patients in
whom there is no plausible etiology other than critical illness”
[2]. This syndrome includes critical illness polyneuropathy,
myopathy, and neuromyopathy [2,3].
ICUAW is associated with significant impairments in body
structure and function [4,5], activity limitation [4–6], and par-
ticipation restriction [4]. It has a long-lasting impact [4–6],
persisting for months or years following ICU discharge,
results in a decline in quality of life for ICU survivors [5,7],
and places a high burden on the healthcare system [8]. There is
currently no effective treatment for ICUAW [9,10]. Although
it is often recommended that patients with ICUAW be referred
for rehabilitation, a 2015 Cochrane review [11] failed to
determine whether physical rehabilitation improves activi-
ties of daily living, muscle strength and quality of life in
these ICU survivors. Thus, preventing the development of
ICUAW through the control of associated risk factors [9]
such as immobility [12] is key. Immobility predisposes to
critical illness myopathy through protein loss due to altered
protein metabolism [13] and to critical illness polyneuropa-
thy through hyperglycemia as a result of insulin resistance
[13]. Therefore, reducing the level of immobility during
ICU admission may decrease the risk of developing ICUAW
[9,10].
Early mobilization (EM) reduces the duration of immo-
bility, while neuromuscular electrical stimulation (NMES)
reduces muscle atrophy [14] which is involved in the patho-
physiological process of ICUAW [15,16]. Both interventions
are sometimes used together as part of an early rehabilitation
intervention in the ICU, however, few robust studies have
evaluated the effectiveness of EM and NMES to reduce the
likelihood of developing ICUAW. To date, the few systematic
reviews [17–19] that have evaluated the effectiveness of EM
and NMES to reduce the risk of ICUAW, have included only
two to three RCTs. Additional trials are currently available
[20–24]. Moreover, the meta-analysis in last two systematic
reviews [18,19] analyzed patients with missing data (in some
of the included RCTs) as not having ICUAW. In these stud-
ies, ICUAW was assessed with the MRC sum score, which is
prone to having missing data in very weak and uncooperative
patients. Given this context, it seems more appropriate to err
conservatively on the side of inclusion and impute the worst
values (= ICUAW) for patients who were not evaluable. Fur-
thermore, the forest plot in the review by Castro-Avila et al
[18] also had errors for the incidence of ICUAW in both the
intervention and control groups for the study by Routsi et al.
[25]. The few trials included in two previous reviews also lim-
ited the exploration of sources of heterogeneity. This review
is, therefore, necessary to revise and update the knowledge
that is currently available. The specific research questions
Box 1: Inclusion criteria.
Design
• Randomized controlled trials
Participants
• Adults
• Critically ill participants patients not already diag-
nosed with intensive care unit acquired weakness
(ICUAW)
Intervention
• Evaluated the effect of the early rehabilitation
interventions of early mobilization (EM) and/or
neuromuscular electrical stimulation (NMES) inter-
ventions
Outcome measures
• Reported the incidence of ICUAW or muscle strength
assessed using the Medical Research Council Scale
(MRC)
Comparisons
• Usual care
were: to what extent do the early rehabilitation interven-
tions of EM and NMES, compared to usual care, reduce the
incidence of ICUAW among patients in the ICU, and alter
other outcomes that may be associated ICUAW (i.e. length
of time on mechanical ventilator, discharge location, ICU and
hospital length of stay, and acute mortality).
Method
Identification and selection of studies
We searched MEDLINE, EMBASE, CINAHL, Cochrane
Central Register of Controlled Trials and Physiotherapy Evi-
dence Database databases, from inception to May 1st, 2017,
without language restriction (Supplementary Digital Content
[SDC] I). Studies were included if they: (i) were conducted in
the ICU, (ii) were RCTs, (iii) involved adult participants, (iv)
evaluated the effect of EM or NMES interventions, and (v)
reported the incidence of ICUAW or assessed muscle strength
using the Medical Research Council Scale (MRC) (Box 1).
Studies that evaluated rehabilitation interventions in patients
already diagnosed with ICUAW were excluded.
Two reviewers independently screened the titles and
abstracts of the retrieved articles, and the full texts of poten-
tially eligible articles were obtained and further assessed for
final inclusion. Reviewers formally met at each step to reach a
consensus, and a senior author resolved differences if needed.
 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10 3

Data extraction and assessment of characteristics of the

studies

Two reviewers participated in the data extraction and qual-

ity assessment of the individual studies using the Cochrane
Risk of Bias Tool [26]. Population characteristics, treat-
ment intervention(s), control/comparators, and outcomes
were described for each included study. Quality assessment
was performed in duplicate by the reviewers; items were cate-
gorized as ‘Low, High or Unclear’ based on the criteria shown
in the on-line supplement (SDC II). All disagreements were
tracked and resolved by discussions and consensus or by a
senior author if needed. Authors of the primary studies were
contacted to request any important missing information.

Data analysis

The primary outcome was the incidence of ICUAW mea-

sured at any time point after the initiation of the intervention.
The secondary outcomes were length of time spent on a
mechanical ventilator (ventilator-free days and duration of
mechanical ventilation), discharge location, ICU and hospi-
tal LOS, and acute mortality (defined as death in the ICU
or hospital). For binary outcomes, we extracted the number
of events and total number in each group. For continuous
outcomes, we extracted the mean (standard deviation) or the
median (range) values. Fig. 1. PRISMA flow diagram.
Data were analyzed using Stata 14 (StataCorp., 2015).
We expected some differences in the true population effects
given the differences in the patients studied and the types of assessment (ICU awakening, 7th day post-awakening, ICU
rehabilitation interventions used across studies (heterogene- discharge, hospital discharge). The latest measuring time-
ity). We, therefore, used the random-effect model to pool the point was used for ICUAW in the main meta-analysis. We
effect sizes from the different studies in a meta-analysis and assessed publication bias using a funnel plot.
performed a sensitivity analysis using the fixed effect model
[27]. Only dichotomous data were pooled together. A priori,
we planned to only report the risk ratios (RR), but during
our review, we discovered that the true control group risk Results
(without exposure to the intervention) was unknown as the
control group in six of the studies included EM as part of The search yielded 1594 articles. After removing 173
their standard care. Since, the RR are bound by the control duplicates, we screened 1421 articles for eligibility (Fig. 1).
group risks, while odds ratios (OR) are not [28,29], we judged Of these, 41 articles were selected for full-text review.
that using the OR was mathematically more appropriate for Ten articles [20–25,32–35] were selected after the full-text
this meta-analysis [28,30,31]. We, therefore, reported both review, but only nine [20–25,33–35] were included in the
measures (RR and OR) but with a focus on the OR (95% critical appraisal because data could not be obtained for one
confidence intervals). For all the studies, we obtained data study [32], (see reasons for exclusions: SDC III).
for patients who were evaluated for ICUAW (the screened
population) and the total randomized population, and calcu-
lated the OR for both (with the imputation of worst values for Characteristics of the studies
patients who had missing data [intention-to-treat analysis] –
as was done in a previous Cochrane review [17]). Statistical Of the 841 patients included in the studies (419 hav-
heterogeneity was assessed with the I2 statistic, while clinical ing received an intervention and 422, the usual care), most
heterogeneity was explored with subgroup analysis based on received mechanical ventilation during their ICU stay, and
(i) short vs long ICU LOS (with short ICU LOS defined as the primary reason for ICU admission as well as the exist-
≤7 days in both the intervention and control group), (ii) the ing co-morbidities varied from study to study. Details of the
timing of intervention, (iii) the type of rehabilitation inter- patient characteristics and interventions delivered are shown
vention (EM or NMES), (iv) different time points of ICUAW in the online supplement (SDC II).
 4
Table 1
Summary of the results for the primary outcome-ICUAW.
Study Time point Proportion of events reported Proportion of events in Proportion of events in
screened population total randomized
population (reanalysis*)

D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10

Intervention Control P-value Intervention Control Intervention Control
Schweickert 2009 [35] Hospital discharge 15/49 27/55 0.09 6/40 13/41 Same as reported Same as reported
Routsi 2010 [25] ICU awakening 3/24 11/28 0.04 3/24 11/28 47/68 55/72
Dantas 2012 [20] Sedation cessation Not reported Not reported Not reported 7/14 8/14 19/26 27/33
ICU discharge Not reported Not reported Not reported 4/14 7/14 16/26 26/33
Denehy 2013 [33] 7th-day post-awakening 16/74 13/76 Not reported 16/57 13/49 33/74 40/76
Kho 2015 [23] ICU awakening 8/12 5/15 0.128 8/12 5/15 12/16 8/18
ICU discharge 3/12 4/16 1 3/12 4/16 7/16 6/18
Hospital discharge 1/12 5/17 0.354 1/12 5/17 5/16 6/18
Kayambu 2015 [22] ICU discharge Not reported Not reported Not reported 9/19 14/23 16/26 15/24
Fischer 2016 [21] ICU awakening Not reported Not reported Not reported
ICU discharge Not reported Not reported Not reported 6/21 4/20 12/27 11/27
Hospital discharge Not reported Not reported Not reported 1/14 1/10 14/27 18/27
Hodgson 2016 [34] ICU discharge 7/25 10/20 0.13 7/25 10/20 11/29 11/21
Schaller 2016 [24] ICU discharge 50/104 51/96 0.95 50/76 51/77 78/104 70/96
 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10
Fig. 2. Forest plot of comparison: incidence of ICUAW between early rehabilitation versus usual care in both the screened population and the randomized
population (timepoint: last ICUAW assessment).
Effect of the intervention
Table 1 shows the number of people with ICUAW in the
intervention and control groups as reported in the articles, in
the screened population, and in the total randomized popula-
tion. The pooled OR using the random effect model was 0.63
(95% CI: 0.43 to 0.92) in the screened population, and 0.71
(95% CI: 0.53 to 0.95) in the total population randomized
in favor of early rehabilitation (Fig. 2). Sensitivity analysis
using the fixed effect model showed similar results (plots not
shown).
Details of subgroup analyses are presented in Table 2 and
Figs. A–D of the online supplement—SDC II. The impact of
rehabilitation on the odds of developing ICUAW was more
profound in a subgroup of studies in which patients had longer
ICU LOS. Sub-analysis, by time point of ICUAW assessment,
favored assessment of ICUAW at hospital discharge com-
pared to earlier time points. The sub-analysis also showed
lower odds of developing ICUAW in studies where early
rehabilitation was started within 72 hours of ICU admission
compared to studies in which it was 72 hours after admission.
These results were consistent in both the screened population
and in the total randomized population.
The fixed effect model of RR analysis showed the same
results as the OR analysis, whereas with the random effect
model there was only a tendency for a reduced RR in favor
of early rehabilitation: 0.75 (0.57, 1.00) in the screened pop-
ulation and 0.91 (0.82, 1.01) in the randomized population.
Six studies reported ICU mortality [20–23,25,34], while
five studies reported hospital mortality [21,23,24,34,35].
There was no difference in the pooled OR for acute mor-
tality between the two groups (OR 1.19; 95% CI: 0.79 to
5
1.80) (results shown in Fig. E of the online supplement—SDC
II).
Though ICU LOS was numerically shorter with rehabili-
tation in six studies [20,21,24,25,34,35], only one involving
EM was statistically significant (2 days less for ‘LOS until
ICU discharge readiness’) [24] (Table 3). The same study
found a significantly shorter hospital LOS (6.5 days less,
P = 0.01) in favor of early rehabilitation [24]. Six studies
[21,22,24,33–35] reported median changes in LOS, making
it impossible to pool the study results in a meta-analysis.
Length of time on a mechanical ventilator was reported
as ‘duration of mechanical ventilation’ in eight studies
[20–23,25,33–35] and as ‘ventilator-free days’ in four
studies [22,24,25,35] (Table 3). Duration of mechanical
ventilation was statistically shorter in one study (favoring
intervention) [35] and was not significantly different in five
studies [20,22,23,25,34]. Ventilator-free days, was statisti-
cally longer in one study (favoring intervention) [35], shorter
in another study [25] and showed no significant difference in
two studies [22,24].
Five studies reported discharge location [23,24,33–35].
Only two studies compared it statistically, favoring discharge
home in the intervention group (P = 0.06 [35] and 0.0007
[24]). A meta-analysis of discharge location in the random-
ized population showed a pooled OR of 1.69 (95% CI: 1.04
to 2.75) in favor of rehabilitation for being discharged home
(SDC II: Fig. F).
Statistical heterogeneity (I2 statistic) showed values ‘that
might not be important’ [36] (I2 < 40%) for ICUAW (Fig. 2),
mortality, and discharge location (SDC II: Figs. E- & F-online
supplements, respectively). In contrast, subgroup analysis for
ICUAW showed clinical heterogeneity that may be explained
 6
Table 2
Table of odds ratios from forest plots (Details of the forest plots are given in the online supplemental digital content II)
A. Forest plot of comparison in the screened population

Analysis (OR (95% CI)

Incidence of ICUAW between early rehabilitation versus usual care ICU LOS ≤ 7days for either the ICU LOS not ≤ 7days for either
in both the screened and randomized populations (Subgroup intervention or control group the intervention or control group
analysis by ICU LOS, Fig. A-supplement) 0.96 (0.50 to 1.85) 0.51 (0.32 to 0.81)
Incidence of ICUAW between early rehabilitation versus usual care ICU Awakening 7th day post Sedation cessation ICU discharge Hospital discharge
in the total randomized population (Subgroup analysis at different awakening
assessment timepoints, Fig. B-supplement) 0.92 (0.05 to 1.08 (0.46 to 2.55) 0.75 0.17 to 3.33) 0.78 (0.49 to 1.24) 0.37 0.15 to 0.94)

D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10

15.68)
Incidence of ICUAW between early rehabilitation versus usual care ≤72 hours >72 hours
in the total randomized population (Subgroup analysis by timing of 0.57 (0.37 to 0.88) 0.70 (0.17 to 2.84)
intervention, Fig. C-supplement)
Incidence of ICUAW between early rehabilitation versus usual care Early mobilization Electrical Both early
in the total randomized population (Subgroup analysis by type of Stimulation mobilization &
rehabilitation intervention, Fig. D-supplement) electrical stimulation
0.71 (0.45 to 1.12) 0.26 (0.09 to 0.80) 0.58 (0.17 to 1.98)
B. Forest plot of comparison in the randomized population
Analysis (OR (95% CI)
Incidence of ICUAW between early rehabilitation versus usual care ICU LOS ≤ 7days for either the ICU LOS not ≤ 7days for
in both the screened population and the randomized population intervention or control group either the intervention or
(Subgroup analysis by ICU LOS, Fig. A-supplement) control group
0.90 (0.47 to 1.72) 0.64 (0.45 to 0.90)
Incidence of ICUAW between early rehabilitation versus usual care ICU Awakening 7th day post Sedation cessation ICU discharge Hospital discharge
in total randomized population (Subgroup analysis at different awakening
assessment timepoints, Fig. B-supplement) 1.43 (0.28 to 7.35) 0.72 (0.38 to 1.38) 0.60 (0.17 to 2.08 0.92 (0.62 to 1.38) 0.54 (0.30 to 0.97)
Incidence of ICUAW between early rehabilitation versus usual care ≤72 hours >72 hours
in total randomized population (Subgroup analysis by timing of 0.70 (0.50 to 0.97) 0.75 (0.42 to 1.35)
intervention, Fig. C-supplement)
Incidence of ICUAW between early rehabilitation versus usual care Early mobilization Electrical Both early mobilization & electrical stimulation
in total randomized population (Subgroup analysis by type of Stimulation
rehabilitation intervention, Fig. D-supplement) 0.70 (0.49 to 1.00) 0.68 (0.38 to 1.19) 0.96 (0.31 to 3.01)
 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10 7

Table 3
Summary of results for secondary outcomes: length of stay & length of time on mechanical ventilator.
A. ICU and Hospital Length of Stay
Length of stay ICU, days Length of stay hospital, days
Study Rehab. Usual care P-value Rehab. Usual care P-value
Schweickert 2009 [35] 6 (5 to 13)c 8 (6 to 13)c 0.08 14 (8 to 23)c 13 (9 to 20)c 0.93
Routsi 2010 [25] 14 (4 to 62)b 22 (2 to 92)b 0.11 Not Not Not
reported reported reported
Dantas 2012 [20] 19 (11)a 21 (17)a 0.77 25 (24)a 22 (25)a 0.25
Denehy 2013 [33] 8 (6 to 12)c 7 (6 to 11)c Not 24 (16 to 42)c 20 (13 to 31)c Not
reported reported
Kho 2015 [23] 22 (17)a 20 (17)a 0.72 36 (22)a 35 (20)a 0.85
Kayambu 2015 [22] 12 (4 to 45)c 9 (3 to 36)c 0.43 41 (9 to 158)c 45 (14 to 308)c 0.80
Fischer 2016 [21] 6 (3 to 23)d 7 (3 to 213)d 0.46 22 (4 to 84)d 19 (9 to 213)d 0.60
Hodgson 2016 [34] 9 (6 to 17)c 11 (8 to 19)c 0.28 19 (14 to 30)c 29 (16 to 34)c 0.33
Schaller 2016 [24] 7 (5 to 12)d 10 (6 to 15)d 0.005 15 (11 to 27)d 22 (15 to 30)d 0.01

B. Length of Time on Mechanical Ventilator

Mechanical ventilation duration, days Ventilator Free Days

Study Rehab. Usual care P-value Rehab. Usual care P-value

Schweickert 2009 [35] 3 (2 to 7)c 6 (4 to 10)c 0.002 24 (7 to 26)c 21 (0 to 24)c 0.05
Routsi 2010 [25] 7 (2 to 41)d 10 (1 to 62)d 0.07 4 (0 to 16)d 6 (0 to 41)d 0.003
Dantas 2012 [20] 10 (9)a 11(13)a 0.60 Not Not Not
reported reported reported
Denehy 2013 [33] 4 (2 to 9)c,e 4 (2 to 7)c,e Not Not Not Not
reported reported reported reported
Kho 2015 [23] 20 (18)a 16 (15)a 0.49 Not Not Not
reported reported reported
Kayambu 2015 [22] 8 (4 to 64)c 7 (2 to 30)c 0.22 20 (0 to 24)c 21 (0 to 26)c 0.71
Fischer 2016 [21] 2 (1 to 7)d 2 (1 to 15)d Not Not Not Not
reported reported reported reported
Hodgson 2016 [34] 5 (4 to 10)c 7 (5 to 12)c 0.18 19 (7)a 17 (9)a 0.40
Schaller 2016 [24] Not Not Not 23 (18 to 25)d 23 (16 to 25)d 0.31
reported reported reported
a Mean (±SD).
b Mean (range).
c Median (interquartile range).
d Median (range).
e Converted from hours to days.

by the type and the timing of rehabilitation interventions and Discussion

by ICU LOS. Funnel plot analysis (SDC II: Fig. G-online
supplement) revealed no significant publication bias, which
was also confirmed by Egger’s test (P = 0.364 for the null
hypothesis of ‘no small-study effects’).
Details of the risk of bias assessment results are shown
in Table D-online supplement (SDC II). With selection bias,
performance bias and detection bias judged as key indica-
tors for our primary outcome, three studies had high risk
of bias, one had an unclear risk, and five studies showed
low risk. Across studies, the risk of bias is summarized as
unclear.
Our study showed that rehabilitation in the ICU is associ-
ated with a 37% and 29% reduction in the odds of developing
ICUAW in screened and randomized populations, respec-
tively. This effect was more pronounced in a subgroup of
patients with longer ICU LOS. Our study also showed that
rehabilitation was associated with an increased likelihood of
being discharged home. There was inconsistent evidence that
early rehabilitation is associated with a shorter length of time
spent on a mechanical ventilator, and shorter ICU as well
as hospital LOS. It also showed that rehabilitation is not
associated with odds of acute mortality.
8 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10
A recent systematic review [19] showed that EM is associ-
ated with a 58% reduction in the odds of developing ICUAW.
However the robustness of this evidence is not clear as the
meta-analysis in the review pooled only 2 trials excluding
the two largest EM trials due on the review’s exclusion cri-
teria [33] and search date [24]. Two other earlier systematic
reviews [17,18] did not find an association between reha-
bilitation and reduced incidence of ICUAW which may be
attributed to the limited number of publications [17,18] (2 and
3 publications, respectively) included in the reviews. In con-
trast, this review obtained data from authors of many primary
studies and the meta-analysis pooled nine trials to provide the
most roboust evidence to date.
Our sub-analysis showed that a subgroup of studies, in
which patients spent a long time in the ICU, had 49% and
36% lower odds of developing ICUAW with rehabilitation in
the screened and randomized population, respectively. This
finding is supported by previous evidence that showed a
greater likelihood of developing ICUAW [33] and greater
benefits with exercise in patients with longer ICU LOS [37].
Our sub-analysis also showed that the effect of rehabili-
tation on ICUAW is best assessed at hospital discharge,
which is supported by evidence of increased quadriceps force
identified at hospital rather than ICU discharge following
a rehabilitation intervention implemented in the ICU [37].
Exercise improves health through neurological, metabolic
and morphological adaptation mechanisms [38–41] which
yield measurable effects at a future time past the point of
administration. Assessment at earlier time points may there-
fore not accurately reflect the effect of rehabilitation delivered
in the ICU. The sub-analysis based on the time of onset
of rehabilitation suggest that earlier onset of rehabilitation
(≤72 hours after ICU admission) is protective against the
development of ICUAW when compared to starting later than
72 hours. However, it may simply imply that patients who are
able to participate in rehabilitation within the first 72 hours
of ICU admission, are likely to have less ICUAW. But the
later is less likely as the analysis is based on the timing of
rehabilitation across individual studies and not for individ-
ual patients. Earlier studies have shown decreased skeletal
muscle synthesis and muscle thickness loss within the first
72 hours of critical illness [42,43]. Therefore, the timing of
the intervention appears to be an important variable to con-
sider.
Our findings also indicated that rehabilitation is safe in the
ICU, as has been shown in all the primary studies included
in this review, other primary studies [37,44–47], and system-
atic reviews [17,48–53]. Our study showed some evidence
of shorter ICU and hospital LOS in one EM study which
is consistent with the results of previous non-RCT studies
[45,47,54–56]. No RCT included in this review was powered
to detect a difference in ICU or hospital LOS. A previous
systematic review [48] pooled the results of several studies
together using the Hedges’ g statistic and reported shorter
ICU and hospital LOS in favor of rehabilitation; however,
clinical interpretation of the Hedges’ g statistic in relation to
LOS is unknown. Zhang et al.’s review [53] showed similar
results, but it is unclear how the authors pooled means and
medians together in the meta-analysis. There was also a ten-
dency towards a shorter duration of mechanical ventilation in
studies where EM was not part of usual care [25,35] compared
to studies in which the control group received rehabilitation
as part of usual care [20,22–24]. The presence of rehabil-
itation in both the intervention and control groups may be
a possible confounder differentiating studies with a positive
outcome [25,35] from those without [20,22,24,33]. In the cur-
rent review, rehabilitation compared to usual care was also
associated with a higher odds of patients being discharged
home. This finding is contrary to that reported in a previous
systematic review which only showed a tendency towards
an increased likelihood of being discharged home [52]. This
difference may be explained by the differences in the studies
included in the review. Contrary to the current review, which
included only rehabilitation studies of EM and NMES rela-
tive to our primary outcome (ICUAW), the previous review
[52] included both RCTs and controlled study designs that
evaluated only active mobilization interventions.
The rigorous nature of the search strategy, the inclusion
of only RCTs, the conservative nature of data imputation, the
consistency of the results (in the screened and randomized
populations, and with fixed and random effects models for
OR) and the ‘absence of evidence of statistical heterogeneity’
support the strength of the evidence generated in this review.
Nonetheless, a number of limitations exist. First, our search
strategy identified only RCTs which provided information on
our primary outcome. Second, sub-analyses may be subject
to limitations of observational investigations, therefore the
interpretations of the results of the sub-analysis are limited.
Third, control group interventions varied across studies (with
and without EM). We limited the effect of this by focusing on
the results of our OR analysis. Finally, ICUAW was measured
with the MRC sum score in all the primary studies which
may have underestimated the real incidence of ICUAW as
compared to electrophysiological studies [57].
In conclusion, our results suggest that beginning reha-
bilitation early in the course of critical illness may reduce
the odds of developing ICUAW, ICU and hospital LOS, and
duration of mechanical ventilation. This is the first robust sys-
tematic review to show that rehabilitation in the ICU reduces
the odds of developing ICUAW. Wide-scale adoption of reha-
bilitation in the ICU is therefore encouraged. Though such
adoption may involve extra cost, potential savings associ-
ated with the prevention of ICUAW may imply net savings
for the health care system and improved quality of life for
ICU survivors [45,56,58]. Studies exploring the return on
such investments are needed. Future studies should consider
evaluating ICUAW at hospital discharge, and should also be
adequately powered to detect differences in ICUAW, length
of time on mechanical ventilation, discharge location, as well
as ICU and hospital LOS.
Key messages
 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10
• This systematic review and meta-analysis provides the first
evidence that early rehabilitation in the ICU is associated
with lower odds of developing ICUAW.
• Our results imply that beginning rehabilitation early in the
course of critical illness reduces the odds of developing
ICUAW.
Ethics approval: None required.
Source of funding: This research did not receive any specific
grant from funding agencies in the public, commercial, or
not-for-profit sectors.
Conflict of interest: None declared.
Appendix A. Supplementary data
Supplementary material related to this article can be
found, in the online version, at doi:https://doi.org/10.1016/
j.physio.2019.12.004.
References
[1] Griffiths RD, Hall JB. Intensive care unit-acquired weakness. Crit Care
Med 2010;38(3):779–87.
[2] Stevens RD, Marshall SA, Cornblath DR, Hoke A, Needham DM,
de Jonghe B, et al. A framework for diagnosing and classifying
intensive care unit-acquired weakness. Crit Care Med 2009;37(10
Suppl):S299–308.
[3] Appleton R, Kinsella J. Intensive care unit-acquired weakness. Contin
Educ Anaesth Crit Care Pain 2012;12(2):62–6.
[4] Fan E, Dowdy DW, Colantuoni E, Mendez-Tellez PA, Sevransky
JE, Shanholtz C, et al. Physical complications in acute lung injury
survivors: a two-year longitudinal prospective study. Crit Care Med
2014;42(4):849–59.
[5] Sidiras G, Gerovasili V, Patsaki I, Karatzanos E, Papadopoulos E,
Markaki V, et al. Short and long term outcomes of ICU acquired
weakness. Health Sci J 2013;7(2).
[6] Hermans G, Van Mechelen H, Clerckx B, Vanhullebusch T, Mesotten
D, Wilmer A, et al. Acute outcomes and 1-year mortality of intensive
care unit-acquired weakness. A cohort study and propensity-matched
analysis. Am J Respir Crit Care Med 2014;190(4):410–20.
[7] Herridge MS, Tansey CM, Matte A, Tomlinson G, Diaz-Granados N,
Cooper A, et al. Functional disability 5 years after acute respiratory
distress syndrome. N Engl J Med 2011;364(14):1293–304.
[8] Rudis MI, Guslits BJ, Peterson EL, Hathaway SJ, Angus E, Beis S,
et al. Economic impact of prolonged motor weakness complicating
neuromuscular blockade in the intensive care unit. Crit Care Med
1996;24(10):1749–56.
[9] de Jonghe B, Lacherade JC, Sharshar T, Outin H. Intensive care
unit-acquired weakness: risk factors and prevention. Crit Care Med
2009;37(10 Suppl):S309–15.
[10] Fan E. Critical illness neuromyopathy and the role of physical
therapy and rehabilitation in critically ill patients. Respir Care
2012;57(6):933–44, discussion 44-6.
[11] Mehrholz J, Pohl M, Kugler J, Burridge J, Mückel S, Elsner
B. Physical rehabilitation for critical illness myopathy and neu-
ropathy. Cochrane Database Syst Rev 2015;(3). Available from:
http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD010942.
pub2/abstract.
9
[12] Chen YW, Gregory CM, Scarborough MT, Shi R, Walter GA, Van-
denborne K. Transcriptional pathways associated with skeletal muscle
disuse atrophy in humans. Physiol Genomics 2007;31(3):510–20.
[13] Booth FW. Effect of limb immobilization on skeletal muscle. J Appl
Physiol Respir Environ Exerc Physiol 1982;52(5):1113–8.
[14] Wageck B, Nunes GS, Silva FL, Damasceno MCP, de Noronha M.
Application and effects of neuromuscular electrical stimulation in
critically ill patients: systematic review. Med Intensiva 2014;38(7):
444–54.
[15] Helliwell TR, Wilkinson A, Griffiths RD, McClelland P, Palmer TEA,
Bone JM. Muscle fibre atrophy in critically ill patients is associated with
the loss of myosin filaments and the presence of lysosomal enzymes
and ubiquitin. Neuropath Appl Neuro 1998;24(6):507–17.
[16] Stibler H, Edstrom L, Ahlbeck K, Remahl S, Ansved T. Electrophoretic
determination of the myosin/actin ratio in the diagnosis of critical illness
myopathy. Intensive Care Med 2003;29(9):1515–27.
[17] Hermans G, De Jonghe B, Bruyninckx F, Van den Berghe G. Interven-
tions for preventing critical illness polyneuropathy and critical illness
myopathy. Cochrane Database Syst Rev 2014;(1). Available from:
http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD006832.
pub3/abstract.
[18] Castro-Avila AC, Seron P, Fan E, Gaete M, Mickan S. Effect of early
rehabilitation during intensive care unit stay on functional status: sys-
tematic review and meta-analysis. PLoS One 2015;10(7):e0130722.
[19] Fuke R, Hifumi T, Kondo Y, Hatakeyama J, Takei T, Yamakawa K, et al.
Early rehabilitation to prevent postintensive care syndrome in patients
with critical illness: a systematic review and meta-analysis. BMJ Open
2018;8(5):e019998.
[20] Dantas CM, Silva PF, Siqueira FH, Pinto RM, Matias S, Maciel
C, et al. Influence of early mobilization on respiratory and periph-
eral muscle strength in critically ill patients. Rev Bras Ter Intensiva
2012;24(2):173–8.
[21] Fischer A, Spiegl M, Altmann K, Winkler A, Salamon A, Themessl-
Huber M, et al. Muscle mass, strength and functional outcomes in
critically ill patients after cardiothoracic surgery: does neuromuscular
electrical stimulation help? The Catastim 2 randomized controlled trial.
Crit Care 2016;20(1):1.
[22] Kayambu G, Boots R, Paratz J. Early physical rehabilitation in intensive
care patients with sepsis syndromes: a pilot randomised controlled trial.
Intensive Care Med 2015;41(5):865–74.
[23] Kho ME, Truong AD, Zanni JM, Ciesla ND, Brower RG, Palmer
JB, et al. Neuromuscular electrical stimulation in mechanically ven-
tilated patients: a randomized, sham-controlled pilot trial with blinded
outcome assessment. J Crit Care 2015;30(1):32–9.
[24] Schaller SJ, Anstey M, Blobner M, Edrich T, Grabitz SD,
Gradwohl-Matis I, et al. Early, goal-directed mobilisation in the
surgical intensive care unit: a randomised controlled trial. Lancet
2016;388(10052):1377–88.
[25] Routsi C, Gerovasili V, Vasileiadis I, Karatzanos E, Pitsolis T, Tripo-
daki E, et al. Electrical muscle stimulation prevents critical illness
polyneuromyopathy: a randomized parallel intervention trial. Crit Care
2010;14(2):R74.
[26] Higgins JP, Altman DG, Gotzsche PC, Juni P, Moher D, Oxman AD,
et al. The Cochrane Collaboration’s tool for assessing risk of bias in
randomised trials. BMJ 2011;343:d5928.
[27] Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. A basic intro-
duction to fixed-effect and random-effects models for meta-analysis.
Res Synth Methods 2010;1(2):97–111.
[28] Cummings P. The relative merits of risk ratios and odds ratios. Arch
Pediatr Adolesc Med 2009;163(5):438–45.
[29] van Rhee H, Suurmond R, Available from: SSRN: Meta-
analyze dichotomous data: Do the calculations with Log
Odds Ratios and report Risk Ratios or Risk Differences; 2015
https://doi.org/10.2139/ssrn.3241369.
10 D.E. Anekwe et al. / Physiotherapy 107 (2020) 1–10
[30] van Rhee H, Suurmond R. Meta-analyze dichotomous data: Do the
calculations with Log Odds Ratios and report Risk Ratios or Risk
Differences; 2015.
[31] Fleiss JL, Berlin JA. Effect sizes for dichotomous data. In: Valentine JC,
Hedges LV, Cooper HM, editors. The handbook of research synthesis
and meta-analysis. New York: Russell Sage Foundation; 2009.
[32] Abu-Khaber HA, Abouelela AMZ, Abdelkarim EM. Effect of electrical
muscle stimulation on prevention of ICU acquired muscle weakness and
facilitating weaning from mechanical ventilation. Alexandria J Med
2013;49(4):309–15.
[33] Denehy L, Skinner EH, Edbrooke L, Haines K, Warrillow S, Hawthorne
G, et al. Exercise rehabilitation for patients with critical illness: a
randomized controlled trial with 12 months of follow-up. Crit Care
2013;17(4):R156.
[34] Hodgson CL, Bailey M, Bellomo R, Berney S, Buhr H, Denehy L,
et al. A binational multicenter pilot feasibility randomized controlled
trial of early goal-directed mobilization in the ICU. Crit Care Med
2016;44(6):1145–52.
[35] Schweickert WD, Pohlman MC, Pohlman AS, Nigos C, Pawlik AJ,
Esbrook CL, et al. Early physical and occupational therapy in mechan-
ically ventilated, critically ill patients: a randomised controlled trial.
Lancet 2009;373(9678):1874–82.
[36] Higgins J, Green S, Available from: Cochrane handbook for sys-
tematic reviews of interventions The Cochrane Collaboration; 2011
http://handbook-5-1.cochrane.org/.
[37] Burtin C, Clerckx B, Robbeets C, Ferdinande P, Langer D, Troosters
T, et al. Early exercise in critically ill patients enhances short-term
functional recovery. Crit Care Med 2009;37(9):2499–505.
[38] Adkins DL, Boychuk J, Remple MS, Kleim JA. Motor training induces
experience-specific patterns of plasticity across motor cortex and spinal
cord. J Appl Physiol 2006;101(6):1776–82.
[39] Matta Mello Portugal E, Cevada T, Sobral Monteiro-Junior R, Teix-
eira Guimarães T, da Cruz Rubini E, Lattari E, et al. Neuroscience of
exercise: from neurobiology mechanisms to mental health. Neuropsy-
chobiology 2013;68(1):1–14.
[40] Rivera-Brown AM, Frontera WR. Principles of exercise physiology:
responses to acute exercise and long-term adaptations to training. PM
R 2012;4(11):797–804.
[41] Folland JP, Williams AG. The adaptations to strength training : mor-
phological and neurological contributions to increased strength. Sports
Med 2007;37(2):145–68.
[42] Puthucheary ZA, Rawal J, McPhail M, Connolly B, Ratnayake G,
Chan P, et al. Acute skeletal muscle wasting in critical illness. JAMA
2013;310(15):1591–600.
[43] Schepens T, Verbrugghe W, Dams K, Corthouts B, Parizel PM, Jorens
PG. The course of diaphragm atrophy in ventilated patients assessed
with ultrasound: a longitudinal cohort study. Crit Care 2015;19(1):
422.
Available online at www.sciencedirect.com
ScienceDirect
[44] Bailey P, Thomsen GE, Spuhler VJ, Blair R, Jewkes J, Bezdjian L,
et al. Early activity is feasible and safe in respiratory failure patients.
Crit Care Med 2007;35(1):139–45.
[45] Morris PE, Goad A, Thompson C, Taylor K, Harry B, Passmore L,
et al. Early intensive care unit mobility therapy in the treatment of
acute respiratory failure. Crit Care Med 2008;36(8):2238–43.
[46] Zafiropoulos B, Alison JA, McCarren B. Physiological responses to
the early mobilisation of the intubated, ventilated abdominal surgery
patient. Aust J Physiother 2004;50(2):95–100.
[47] Muehling B, Schelzig H, Steffen P, Meierhenrich R, Sunder-Plassmann
L, Orend KH. A prospective randomized trial comparing traditional
and fast-track patient care in elective open infrarenal aneurysm repair.
World J Surg 2009;33(3):577–85.
[48] Kayambu G, Boots R, Paratz J. Physical therapy for the critically ill
in the ICU: a systematic review and meta-analysis. Crit Care Med
2013;41(6):1543–54.
[49] Li Z, Peng X, Zhu B, Zhang Y, Xi X. Active mobilization for mechan-
ically ventilated patients: a systematic review. Arch Phys Med Rehabil
2013;94(3):551–61.
[50] Silva VS, Pinto JG, Martinez BP, Camelier FWR. Mobiliza-
tion in the intensive care unit: systematic review. Fisioter Pesqui
2014;21(4):398–404.
[51] Stiller K. Physiotherapy in intensive care: an updated systematic review.
Chest 2013;144(3):825–47.
[52] Tipping CJ, Harrold M, Holland A, Romero L, Nisbet T, Hodg-
son CL. The effects of active mobilisation and rehabilitation in ICU
on mortality and function: a systematic review. Intensive Care Med
2017;43(2):171–83.
[53] Zhang G, Zhang K, Cui W, Hong Y, Zhang Z. The effect of early
mobilization for critical ill patients requiring mechanical ventilation: a
systematic review and meta-analysis. J Emerg Crit Care Med 2018;2(1).
[54] Malkoc M, Karadibak D, Yildirim Y. The effect of physiotherapy on
ventilatory dependency and the length of stay in an intensive care unit.
Int J Rehabil Res 2009;32(1):85–8.
[55] Needham DM, Korupolu R, Zanni JM, Pradhan P, Colantuoni E, Palmer
JB, et al. Early physical medicine and rehabilitation for patients with
acute respiratory failure: a quality improvement project. Arch Phys
Med Rehabil 2010;91(4):536–42.
[56] Engel HJ, Tatebe S, Alonzo PB, Mustille RL, Rivera MJ. Physical
therapist-established intensive care unit early mobilization program:
quality improvement project for critical care at the University of Cali-
fornia San Francisco Medical Center. Phys Ther 2013;93(7):975–85.
[57] Appleton RT, Kinsella J, Quasim T. The incidence of intensive care
unit-acquired weakness syndromes: a systematic review. J Intensive
Care Soc 2014;16(2):126–36.
[58] McWilliams D, Weblin J, Atkins G, Bion J, Williams J, Elliott C,
et al. Enhancing rehabilitation of mechanically ventilated patients in
the intensive care unit: a quality improvement project. J Crit Care
2015;30(1):13–8.