Awake extracorporeal life support and physiotherapy in adult patients: A systematic review of the literature

Abstract

Objective

The Awake Extracorporeal Life Support (ECLS) practice combined with physiotherapy is increasing. However, available evidence for this approach is limited, with unclear indications on timing, management, and protocols. This review summarizes available literature regarding Awake ECLS and physiotherapy application rates, practices, and outcomes in adults, providing indications for future investigations.

Methods

Four databases were screened from inception to February 2021, for studies reporting adult Awake ECLS with/without physiotherapy. Primary outcome was hospital discharge survival, followed by Extracorporeal Membrane Oxygenation (ECMO) duration, extubation, Intensive Care Unit stay.

Results

Twenty-nine observational studies and one randomized study were selected, including 1,157 patients (males n = 611/691, 88.4%) undergoing Awake ECLS. Support type was reported in 1,089 patients: Veno-Arterial ECMO (V-A = 39.6%), Veno-Venous ECMO (V-V = 56.8%), other ECLS (3.6%). Exclusive upper body cannulation and femoral cannulation were applied in 31% versus 69% reported cases (n = 931). Extubation was successful in 63.5% (n = 522/822) patients during ECLS. Physiotherapy details were given for 676 patients: exercises confined in bed for 47.9% (n = 324) patients, mobilization until standing in 9.3% (n = 63) cases, ambulation performed in 42.7% (n = 289) patients. Femoral cannulation, extubation and V-A ECMO were mostly correlated to complications. Hospital discharge survival observed in 70.8% (n = 789/1114).

Conclusion

Awake ECLS strategy associated with physiotherapy is performed regardless of cannulation approach. Ambulation, as main objective, is achieved in almost half the population examined. Prospective studies are needed to evaluate safety and efficacy of physiotherapy during Awake ECLS, and suitable patient selection. Guidelines are required to identify appropriate assessment/evaluation tools for Awake ECLS patients monitoring.

Graphical Abstract.

Introduction

Over the last decade, medical and scientific communities have dedicated their effort from not only promoting effective disease treatments but also to the patient’s subsequent quality of life and functional recovery.^1–3 Critically ill patients face multiple challenges which might significantly impact their recovery process and future quality of life. A substantial contributor to such an impairment is the Intensive Care Unit Acquired Weakness (ICU-AW).^4,5 This post-intensive care status directly affects the patient´s ability to achieve full recovery, complete daily tasks, and maintain mental wellbeing. This is particularly true for patients requiring long duration of Intensive Care Unit (ICU) stay with protracted mechanical ventilation or extracorporeal life support (ECLS), such as those supported as bridge to recovery (BTR) or to transplant (BTT). ⁶ Most of these patients are usually managed with conservative ECLS strategies based on invasive mechanical ventilation (IMV), sedation, immobilization, or passive range of movements (ROM) during bed rest and unconscious cognitive state. This allows for safe and standardized patient management, with a postulated low risk of ECLS-related complications.^7,8 On the other hand, reduced sedation time and an awake patient with controlled and active participation to the functional recovery have represented new strategies being developed under the name of Awake ECLS: a practice based on the concept that allowing active patient participation will favor a rapid recovery, less chronic comorbidities, reduced psychological as well as psychiatric complications, and improved survival. ⁹ The Awake ECLS strategy adopted so far is mainly based on liberation from IMV, reduction of sedation and analgesia, ¹⁰ and early rehabilitation until ambulation. ¹¹ Despite the initial enthusiasm for this practice, literature still lacks of a standardized definition of Awake ECLS strategy, a defined population selection and objective scales to evaluate patients’ status, feasibility, and safety outcomes.

Based on the discrepancy between the increasing interest towards Awake ECLS and the lack of evidence as well as the current limited use in adult patients, we hypothesized that Awake ECLS strategy associated with physiotherapy might positively influence survival at discharge and/or 30-day survival with low incidence of complications and achievement of active physiotherapy and liberation from IMV. Therefore, this systematic review investigated the definition, common practices, feasibility, safety, and outcomes of Awake ECLS and physiotherapy implementation.

Material and methods

Data sources and search strategies

The protocol for the systematic review was completed in line with the PRISMA statement and registered with PROSPERO (Registration No. CRD42021245872) before the start of the literature screening. PubMed/Medline and Cochrane Database of Systematic Reviews were screened from inception to February 2021. We further searched databases of ongoing trials such as the Cochrane Central Register of Controlled Trials and Clinical Trials (clinicaltrials.gov). Further details of the search strategy are provided in Supplemental Materials. After removing duplicates, the remaining titles and abstracts were assessed for inclusion by two independent reviewers (M.C. and S.M.) using a free, open-source citation screening program. ¹² Full texts of relevant articles were retrieved and independently assessed by two authors (M.C. and S.M.). Disagreements over articles appropriateness for inclusion were resolved by consensus. Reference lists of assessed full texts were screened for further relevant studies. ¹³

Population, intervention and outcome

We included randomized clinical trials, controlled before-and-after studies, prospective and retrospective cohort studies, cross-sectional studies, case–control studies, and case series. Conference abstracts, books or grey literature, articles not written in English, reviews, and animal studies were excluded.

The population of interest included hospitalized adult patients (age ≥ 18 years old), supported with ECLS as bridge to recovery, to transplant or to next therapy and undergoing one or more elements of the defined Awake ECLS strategy: weaned sedation, reduction or liberation from IMV, early mobilization or active range of movements aiming to ambulation. ¹⁴ The exclusion criteria were identified as follows: pediatric populations, central cannulation with open chest and/or sternotomy and patients supported with durable left ventricular assist device. Studies presenting only interventions referred to as “early rehabilitation” or focused on chest physiotherapy without proven Awake ECLS strategy were excluded. Finally, case series reporting less than 10 patients were excluded.

Primary outcome measure was survival at discharge. Secondary outcome measures included safety outcomes (adverse events during mobilization while supported on ECLS: cannula dislodgement, bleeding, hemodynamic instability, respiratory deterioration requiring re-intubation, circuit malfunction, delirium, local trauma, neurological sequelas, wound infections), feasibility outcomes (active physical therapy executed, extubation achieved, tracheostomy performed), current common practices, ICU length of stay, hospital length of stay, ECLS cannulation strategies, bridging strategy and evaluation scales. Definitions from Extracorporeal Life Support Organization (ELSO) Registry ¹⁵ were used to characterize ECLS-related complications.

Data extraction and risk-of-bias assessment

Using a standardized electronic report form, data were extracted into a dedicated database by the primary reviewer (M.C.) and checked for accuracy by a second reviewer (S.M.). Extracted variables are detailed in Supplemental Materials. Two reviewers (M.C. and S.M.) independently assessed risk-of-bias in individual studies; disagreements were resolved by consensus. For observational non-randomized studies, the Newcastle Ottawa Scale (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp) was used (Supplementary Table 1).

Data synthesis

Available evidence was summarized using systematic review methodology and standard summary statistics overall. ¹⁶ An additional analysis was performed to estimate the pooled median values for available continuous variables. The quantile estimation (QE) method for pooling median was based on interquartile ranges and minimum to maximum values. The sampling variance of the effect size for each study was estimated via the QE method. After estimating the sampling variances for all studies, studies were meta-analyzed using the restricted maximum-likelihood estimator (REML) in a random-effects model. ¹⁷ The analysis was performed with R 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria). Given the significant clinical, ¹⁸ methodological and statistical heterogeneity among published studies, further meta-analytic methods were not applied.

Results

Literature search

A total of 1,265 abstracts were identified through PubMed search (Figure 1). From reference list analysis, no additional publications were found. After exclusion of duplicates and irrelevant publications, 153 potential articles were identified. Subsequently, 30 studies were selected for qualitative analysis (Table 1). Twenty-seven studies had a retrospective design,^14,19–44 two studies were prospective and observational,^45–47 and one study was a Phase II pilot randomised control trial. ⁴³ Only one study was multicenter. ²⁶ Distribution over time of the included articles showed a growing number of publications in the most recent years and a subsequent reduction during the COVID-19 pandemic (Figure 2). Data collection of patients in the included studies covered a wide range of study period from 2002 to 2019.

Figure 1.

Study flow diagram according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) Protocol recommendations.

Table 1.

List of articles included in the systematic review: population characteristics.

Author	Country	Year	Study design (n centres)	Period range	Patients (n)	Age (years)	Males n (%)	V-A ECMO	V-V ECMO	Other ECLS	BTR	BTT	BTNT
Wells et al.	USA	2018	Retrospective (1)	2014–2015	167	50.52 ± 15.2	103 (61.7)	69 (41.3)	98 (58.7)	0	167 (100)	0	0
Hayes et al.	Australia	2020	Sub study phase II Pilot RCT (1)	2002–2017	15	52.0 ± 13.3 (Intensive PT) 51.1 ± 16.0 (Standard PT)	12 (80)	10 (66.7)	5 (33.3)	0	15 (100)	0	0
Mori et al.	USA	2019	Retrospective (1)	2012–2015	19	50.1 ± 18.3	14 (74)	19 (100)	0	0	0	0	19 (100)
Youn et al.	Korea	2020	Retrospective (1)	2006–2018	24	52.8 ± 11.6	22 (92)	24 (100)	0	0	0	24 (100)	0
Braune et al.	Germany	2020	Prospective (1)	2014	43	53 (45.0–68)	28 (65)	14 (32.5)	12 (27.9)	17 (39.5)	41 (95.3)	2 (4.6)	0
Xia et al.	China	2019	Retrospective (1)	2013–2018	12	39.1 ± 16.4	7 (58)	0	12 (100)	0	12 (100)	0	0
Camboni et al.	Germany	2012	Retrospective (1)	2006–2011	36	48 ± 15	18 (50)	0	36 (100)	0	32 (88.9)	4 (11.1)	0
Deng et al.	China	2020	Retrospective (1)	2013–2019	12	43.58 ± 10.1	8 (66.7)	12 (100)	0	0	12 (100)	0	0
Boling et al.	USA	2016	Retrospective (1)	2011–2013	18	49 ± 15	12 (67)	0	18 (100)	0	6 (33.3)	12 (66.6)	0
Hoopes et al.	USA	2013	Retrospective (2)	2003–2012	31	Range: 15–68	23 (74)	15 (48.4)	13 (41.9)	3 (9.7%)	0	31 (100)	0
Ellouze et al.	France	2019	Retrospective (1)	2014–2016	15	54 ± 11.9	7 (46.6)	15 (100)	0	0	15 (100)	0	0
Bataillard et al.	France	2017	Retrospective (1)	2008–2011	41	51 ± 17	34 (82.9)	41 (100)	0	0	15 (36.6)	26 (63.4)	0
Pasrija et al.	USA	2019	Retrospective (1)	2015–2017	15	54 (48–60)	12 (80)	15 (100)	0	0	15 (100)	0	0
Munshi et al.	Canada	2017	Retrospective (1)	2010–2015	50	45 ± 14	39 (78)	0	47 (94)	3 (6)	50 (100)	0	0
Sommer et al.	Germany	2015	Retrospective (1)	2010–2014	23	46.1 ± 15.5	16 (69.5)	23 (100)	0	0	23 (100)	0	0
Tipograf et al.	USA	2019	Retrospective (1)	2009–2018	121	44 (30–58)	60 (50)	52 (43)	63 (52)	6 (4.9)	0	121 (100)	0
Fuehner et al.	Germany	2012	Retrospective (1)	2008–2011	26	Range: 23–62	11 (42)	13 (50)	13 (50)	0	0	26 (100)	0
Levin et al.	Switzerland	2021	Retrospective (1)	2014–2020	17	43.8 ± 4.1	14 (70)	0	17 (100)	0	17 (100)	0	0
Bonizzoli et al.	Italy	2019	Retrospective (1)	2009–2016	101	49.5 ± 53	N/R	0	101 (100)	0	101 (100)	0	0
Yanagida et al.	USA	2019	Retrospective (1)	2013–2015	15	46 ± 17	9 (60)	0	15 (100)	0	8 (53.3)	7 (46.7)	0
Downey et al.	USA	2019	Prospective (1)	2015–2019	17	50 (24–67)	9 (53)	17 (100)	0	0	0	17 (100)	0
Yeo et al.	Korea	2016	Retrospective (1)	2012–2015	10	52.1 ± 7	7 (70)	0	8 (80)	2 (20)	10 (100)	0	0
Benazzo et al.	Austria	2019	Retrospective (1)	2005–2017	33	N/R	N/R	N/R	N/R	N/R	0	33 (100)	0
Abrams et al.	USA	2014	Retrospective (1)	2009–2013	35	45.2 ± 18.7	15 (43)	N/R	N/R	N/R	16 (45.7)	19 (54.3)	0
Biscotti et al.	USA	2017	Retrospective (1)	2007–2016	72	42.2 ± 15.1	34 (47)	23 (32)	45 (62.5)	4 (5.5)	0	72 (100)	0
Taniguchi et al.	Japan	2021	Retrospective (1)	2015–2018	54	58.8 (14.6)	44 (81)	0	54 (100)	0	54 (100)	0	0
Hakim et al.	USA	2018	Retrospective (1)	2012–2015	30	Range: 22–65	21 (70)	5 (16.6)	24 (80)	1 (3.3)	0	30 (100)	0
Magunia et al.	Germany	2020	Retrospective (1)	2015–2016	43	59 ± 14	32 (74.4)	43 (100)	0	0	43 (100)	0	0
Chicotka et al.	Germany	2018	Retrospective (1)	2010–2016	37	N/R	N/R	24 (64.9)	13 (35.1)	0	0	37 (100)	0
Crotti et al.	USA	2018	Prospective (1)	2010–2013	25	38 ± 8 (BTT) 72 ± 6 (COPD) 49 ± 18 (ARDS)	N/R	0	25 (100)	0	14 (56)	11 (44)	0

Data are reported as mean ± SD, median (interquartile range), range of values from min to max, or n (% as valid percentage, excluding missing values).

ARDS: acute respiratory distress syndrome; BTNT: Bridge to Next Therapy; BTR: Bridge to Recovery; BTT: Bridget to Transplant; COPD: chronic obstructive pulmonary disease; ECMO: Extracorporeal Mechanical Oxygenation; ECLS: Extracorporeal Life Support; N/R: Not reported; PT: Physical Therapy; V-A: Veno Arterial; V-V: Veno Venous.

Figure 2.

Time distribution of included studies.

Patients and ECLS characteristics

A total of 1,157 adult patients undergoing Awake ECLS strategy, described in studies published between 2012 and 2020, were included in the final analysis (Table 1). Main indications for ECLS support were BTR (n = 666, 36.6%)^{19,22–25,27–31,33–35,37,39,41,43–46} and BTT (n = 472, 34.8%).^{14,21,23,24,26,28,32,35–38,40,42,45–47} Only 19 (13.5%) ²⁰ patients were specifically bridged to next therapy (BTNT), such as Left Ventricular Assisted Device. The reported overall median age was 48.97 years (95% CI: 45.83–52.11). Males accounted for 88.4% of the reported population (n = 611/691) and females for the 11.6% (n = 80/691). Data on median or mean body mass index were reported only in 6 articles^{14,34,38,39,41,46} and ranged 20–30.1.

Common practice, definition, evaluation scores

The ECLS modality support, when reported, was veno-arterial (V-V) in 56.8% (n = 619/1089)^{14,19–21,25–27,29,31,32,38,40–43,45,47} of patients, veno-venous (V-A) in 39.6% (n = 431/1089)^{14,19,22–24,26,30,32–35,38–40,42–46} of cases and only 3.9% (n = 39/1089)^{26,30,32,38,40,44,45} of patients underwent other ECLS modalities. Cannulation strategy was reported in 931 cases (80.5% of the total population reviewed). Femoral-femoral cannulation (n = 642/931) was received by 69% patients femoral-femoral cannulation (n = 642/931) with either unilateral or bilateral approach,^{14,19–23,25–31,33,34,37,39–41,43–46} of which 75.9% for the specific reported cohort of V-A (n=284/374) and 63.5% for the specific reported cohort of V-V support mode (n=290/457). The 31% (n=289/931) of population underwent upper body cannulation through jugular and axillary access,^{19,24,26,30,34,35,37,40,42,45,47} represented by 17.4% (n=65/374) of the specific reported V-A cases and 36.5% (n=167/457) of the V-V cases respectively. ECMO and IMV duration were reported in 64.5% (n = 747/1157)^{14,19,20,22,24–27,29–35,37–41,43,44,46,47} and in 26.5% (n = 306/1157)^{22,25,27–29,33,34,41,43,44,46} of the overall included population, and the median values were 11.17 days (95% CI: 9.61–12.72) and 12.85 days (95% CI: 6.32–19.37), respectively.

Feasibility: Awake ECLS strategy and physical therapy

Details on the Awake ECLS strategy applied, respiratory status and physiotherapy performed are reported in Table 2, and specific reported data related to either V-A and V-V is located in Tables 3 and 4. Overall, 87.5% (n = 864/987) of patients were explicitly reported as awake,^14,20–47 with a similar proportion for the results related to V-A ECMO (100%, n = 343/343) and V-V ECMO (84.7%, n = 425/502). Liberation from IMV during ECLS was achieved in 67.7% of patients with reported data on this variable (n = 522/772).^{14,20–29,31–33,36–41,44–47} However, it was completed in all reported cases supported on V-A ECMO (205/205), but only in 41.1% (95/231) of patient supported with V-V ECMO. Tracheostomy during ECLS was mentioned in 12 studies with clear description of a performed tracheostomy in only 40.1% (n = 151/376) of patients.^{20,22,23,32,33,35–38,40,41} Lack of details on the IMV weaning strategies, indications, and timing for tracheostomy and extubation practices was noticed. Details on physical therapy were reported in 58.4% of the population (n = 676/1157).^{19,24,26,28,30,34–40,42,43,47} Ambulation was achieved in 42.7% (n = 289/676)^{19,24,26,29,32,34,35,37,38,40,42,47} of patients, standing in 9.3% (n = 63/676), and physical activity confined in bed was executed in 47.9% (n = 324/676) of cases. In the total cohort of actively mobilised patients, four studies^24,26,29,47 specified that all their patients (n = 81) were extubated, eight^{23,30,32,36–38,40,45} that active patients were on IMV and in only one study ³⁸ patients were on non-invasive mechanical ventilation (NIMV). V-A ECMO configuration resulted in a much higher successful ambulation (58.5%, 55/109) compared to V-V ECMO (24.9%, 62/249), based on the reported cases respectively.

Table 2.

List of articles included in the systematic review: practice characteristics.

Author	Upper body cannula only	Including femoral cannula	Defined as awake by authors	Extubated	Tracheostomy	PT on IMV	Unspecified PT executed	Maximum PT in bed	Maximum PT standing	Ambulation
Wells et al.	33 (19.7)	134 (80.2)	N/R	N/R	N/R	N/R	N/A	129 (77.2)	30 (18)	8 (4.8)
Hayes et al.	0	15 (100)	15 (100)	N/R	N/A	N/R	N/R	12 (80)	3 (20)	0
Mori et al.	0	19 (100)	19 (100)	19 (100)	2 (10.5)	N/R	N/R	N/R	N/R	N/R
Youn et al.	0	24 (100)	24 (100)	24 (100)	N/R	N/R	N/R	N/R	N/R	N/R
Braune et al.	3 (7)	40 (93)	9 (20.9)	9 (20.9)	N/A	Yes (pt. N unknown)	Yes (pt. N unknown)	Yes (pt. N unknown)	Yes (pt. N unknown)	Yes (pt. N unknown)
Xia et al.	0	12 (100)	12 (100)	12 (100)	1 (8.3)	0	Yes (pt. N unknown)	N/R	N/R	N/R
Camboni et al.	0	36 (100)	21 (58.3)	4 (11.1)	Yes (pt. N unknown)	13 (36.1)	Yes (pt. N unknown)	N/R	N/R	N/R
Deng et al.	0	12 (100)	12 (100)	12 (100)	0	N/A	Yes (pt. N unknown)	N/R	N/R	N/R
Boling et al.	18 (100)	0	18 (100)	18 (100)	N/R	N/A	N/A	0	0	18 (100)
Hoopes et al.	17 (54.8)	14 (45.2)	31 (100)	31 (100)	31 (100)	N/A	Yes (pt. N unknown)	N/R	N/R	19 (61.3)
Ellouze et al.	0	15 (100)	15 (100)	15 (100)	N/R	N/R	N/R	N/R	N/R	N/R
Bataillard et al.	0	41 (100)	41 (100)	41 (100)	0	N/R	N/R	N/R	N/R	N/R
Pasrija et al.	0	15 (100)	15 (100)	15 (100)	N/R	0	N/A	0	0	15 (100)
Munshi et al.	26 (52)	24 (48)	50 (100)	N/R	N/R	50 (100)	N/A	42 (84)	8 (16)	0
Sommer et al.	0	23 (100)	23 (100)	23 (100)	N/R	N/A	Yes (pt. N unknown)	N/R	N/R	N/R
Tipograf et al.	N/R	N/R	121 (100)	84 (69.4)	38 (31.4)	41 (33.9)	Yes (pt. N unknown)	N/R	N/R	82 (67.8)
Fuehner et al.	0	26 (100)	26 (100)	26 (100)	N/R	N/A	N/R	N/R	N/R	N/R
Levin et al.	0	17 (100)	17 (100)	17 (100)	Yes (pt. N unknown)	N/A	Yes (pt. N unknown)	N/R	N/R	N/R
Bonizzoli et al.	82 (81.2)	19 (18.8)	101 (100)	N/R	N/R	N/R	N/A	81 (80.2)	10 (9.9)	10 (9.9)
Yanagida et al.	15 (100)	0	11 (73.3)	N/R	15 (100)	N/R	N/R	N/R	N/R	11 (73.3)
Downey et al.	17 (100)	0	17 (100)	17 (100)	0	N/R	N/R	1 (5.9)	0	16 (94.1)
Yeo et al.	0	10 (100)	10 (100)	10 (100)	N/R	N/R	Yes (pt. N unknown)	N/R	N/R	N/R
Benazzo et al.	N/R	N/R	33 (100)	15 (45.5)	Yes (pt. N unknown)	Yes (pt. N unknown)	Yes (pt. N unknown)	N/R	N/R	Yes (pt. N unknown)
Abrams et al.	27 (77.1)	8 (22.9)	35 (100)	23 (65.7)	11 (31.4)	12 (34.2)	N/A	14 (40)	3 (8.6)	18 (51.4)
Biscotti et al.	N/R	N/R	72 (100)	22 (30.6)	34 (47.2)	Yes (pt. N unknown)	Yes (pt. N unknown)	N/R	N/R	50 (69.4)
Taniguchi et al.	0	54 (100)	11 (20.4)	11 (20.4)	N/R	Yes (pt. N unknown)	Yes (pt. N unknown)	45 (83.3)	9 (16.7)	N/R
Hakim et al.	14 (46.7)	16 (53.3)	5 (16.7)	6 (20)	15 (50)	Yes (pt. N unknown)	Yes (pt. N unknown)	N/R	N/R	5 (16.6)
Magunia et al.	0	43 (100)	43 (100)	43 (100)	6 (13.9)	N/A	Yes (pt. N unknown)	N/R	N/R	N/R
Chicotka et al.	37 (100)	0	37 (100)	N/R	Yes (pt. N unknown)	N/R	No	N/R	N/R	37 (100)
Crotti et al.	0	25 (100)	20 (80)	25 (100)	N/R	N/A	Yes (pt. N unknown)	N/R	N/R	N/R

Data are reported as n (%).

IMV: Invasive Mechanical Ventilation; N/A: Not Available; N/R: Not Reported; PT: Physical Therapy.

Table 3.

List of articles included in the systematic review: outcomes uniquely presented in V-A ECMO configuration.

Author	Total (n)	Only ECMO upper body cannulation	Including ECMO femoral cannulation	Defined as awake by authors	Extubated	PT on IMV	Unspecified PT executed	Maximum PT in bed	Maximum PT standing	Ambulation	Survival at hospital discharge
Wells et al.	69	24	45	N/R	N/R	N/R	––	Unknown	Unknown	Unknown	41
Hayes et al.	10	0	10	10	N/R	N/R	––	7	3	0	Unknown
Mori et al.	19	0	19	19	19	N/R	N/R	N/R	N/R	N/R	16
Youn et al.	24	0	24	24	24	N/R	N/R	N/R	N/R	N/R	22
Braune et al.	14	0	12	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	N/R
Xia et al.
Camboni et al.
Deng et al.	12	0	12	12	12	N/R	Yes	––	––	––	9
Boling et al.
Hoopes et al.	12	0	12	12	N/R	0	Yes	N/R	N/R	Unknown	12
Ellouze et al.	15	0	15	15	15	N/R	N/R	N/R	N/R	N/R	11
Bataillard et al.	41	0	41	41	41	N/R	N/R	N/R	N/R	N/R	31
Pasrija et al.	15	0	15	15	15	0	––	0	0	15	15
Munshi et al.
Sommer et al.	23	0	23	23	23	0	Yes	––	––	––	21
Tipograf et al.	52	N/R	N/R	52	Unknown	Yes	Yes	––	––	––	Unknown
Fuehner et al.	13	0	13	13	13	0	N/R	N/R	N/R	N/R	Unknown
Levin et al.
Bonizzoli et al.
Yanagida et al.
Downey et al.	17	17	0	17	N/R	N/R	––	1	0	16	15
Yeo et al.
Benazzo et al.	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown
Abrams et al.	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown
Biscotti et al.	23	N/R	N/R	23	Unknown	Yes	––	Unknown	Unknown	Unknown	Unknown
Taniguchi et al.
Hakim et al.
Magunia et al.	43	0	43	43	43	0	Yes	––	––	––	38
Chicotka et al.	24	24	0	24	N/R	N/R	––	0	0	24	19
Crotti et al.

Data are reported as n (%).

ECMO: Extracorporeal Mechanical Oxygenation; IMV: Invasive Mechanical Ventilation; N/A: Not Available; N/R: Not Reported; PT: Physical Therapy; Unknown: data is present but impossible to differentiate the only V-A ECMO number of patients; Yes: data present in the content but not precise amount declared according to feature requested.

Table 4.

List of articles included in the systematic review: outcomes uniquely presented in V-V ECMO configuration.

Author	Total (n)	Exclusively ECMO Upper body cannulation	Including ECMO femoral cannulation	Defined as awake by authors	Extubated	PT on IMV	Unspecified PT executed	Maximum PT in bed	Maximum PT standing	Ambulation	Survival at hospital discharge
Wells et al.	98	9	89	N/R	N/R	N/R	––	Unkown	Unknown	Unknown	68
Hayes et al.	5	0	5	5	N/R	N/R	––	5	0	0	Unknown
Mori et al.
Youn et al.
Braune et al.	12	0	12	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	N/R
Xia et al.	12	0	12	12	12	0	12	N/R	N/R	N/R	8
Camboni et al.	36	0	36	21	4	13	Yes	––	––	––	19
Deng et al.
Boling et al.	18	18	0	18	18	0	––	0	0	18	12
Hoopes et al.	12	12	0	12	N/R	0	Yes	N/R	N/R	Unknown	11
Ellouze et al.
Bataillard et al.
Pasrija et al.
Munshi et al.	47	Unknown	Unknown	47	0	47	47	Unknown	Unknown	Unknown	46
Sommer et al.
Tipograf et al.	63	N/R	N/R	63	Unknown	Yes	Yes	––	––	––	Unknown
Fuehner et al.	13	0	13	13	13	0	N/R	N/R	N/R	N/R	Unknown
Levin et al.	17	0	17	17	N/R	Yes	Yes	––	––	––	17
Bonizzoli et al.	101	82	19	101	N/R	N/R	––	81	10	10	75
Yanagida et al.	15	15	0	11	N/R	N/R	Yes	N/R	N/R	11	9
Downey et al.
Yeo et al.	8	0	8	8	8	N/R	Yes	––	––	––	6
Benazzo et al.	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown
Abrams et al.	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown
Biscotti et al.	45	N/R	N/R	45	Unknown	Yes	––	Unknown	Unknown	Unknown	Unknown
Taniguchi et al.	54	0	54	9	9	Yes	––	N/R	N/R	7	29
Hakim et al.	18	18	0	5	6	Yes	––	N/R	N/R	5	18
Magunia et al.
Chicotka et al.	13	13	0	13	N/R	N/R	––	0	0	13	6
Crotti et al.	25	0	25	25	25	0	Yes	––	––	––	19

Data are reported as n.

ECMO: Extracorporeal Mechanical Oxygenation; IMV: Invasive Mechanical Ventilation; N/A: Not Available; N/R: Not Reported; PT: Physical Therapy; Unknown: data is present but impossible to differentiate the only V-V ECMO number of patients; Yes: data present in the content but not precise amount declared according to feature requested.

Scores for sedation and mobilization

Eleven studies^{19,22,27,30,33,34,37,43–46} specifically reported the evaluation of sedation and mobilization: four studies^19,30,43,45 implemented both sedation and mobilization scores, five studies^{22,27,33,44,46} only reported sedation assessment, and two studies^34,37 applied the mobilization scales (Table 5). The scores involved were: the ICU mobility scale (IMS), ⁴⁸ the Adapted scale from IMS and the BORG scale ⁴⁹ for the evaluation of mobilization. Sedation was assessed with Richmond Agitation Sedation Scale (RASS), ⁵⁰ Riker sedation-agitation-scale (SAS) ⁵¹ and Ramsay scale. ⁵² Finally, the functional capacity was evaluated with the Katz scale, ⁵³ functional comorbidity index (FCI), clinical frailty scale (CFS) ⁵⁴ and simplified acute physiology score (SAPS II). ⁵⁵

Table 5.

Evaluation tools for sedation, mobilization and functionality scores reported in the articles included in the systematic review.

Author	Sedation				Mobilization				Functionality
	RASS	Ramsey	SAS	N/R	IMS	Adapted from IMS	BORG	N/R	KATZ	FCI	CFS	SAPS II	N/R
Wells et al.	✓				✓								×
Hayes et al.	✓				✓		✓		✓	✓	✓
Mori et al.				×				×					×
Youn et al.				×				×					×
Braune et al.	✓				✓							✓
Xia et al.		✓						×					×
Camboni et al.				×				×					×
Deng et al.				×				×					×
Boling et al.				×				×					×
Hoopes et al.				×				×					×
Ellouze et al.	✓							×					×
Bataillard et al.				×				×				✓
Pasrija et al.				×				×					×
Munshi et al.			✓		✓								×
Sommer et a.l				×				×					×
Tipograf et al.				×				×				✓
Fuehner et al.				×				×				✓
Levin et al.			✓					×					×
Bonizzoli et al.				×	✓							✓
Yanagida et al.				×				×					×
Downey et al.				×				×					×
Yeo et al.		✓						×					×
Benazzo et al.				×				×					×
Abrams et al.				×		✓							×
Biscotti et al.				×				×				✓
Taniguchi et al.				×				×					×
Magunia et al.				×				×					×
Chicotka et al.				×				×					×
Crotti et al.	✓							×					×

BORG: Borg Rating of perceived Exertion; CFS: Clinical Frailty Scale; IMS: ICU Mobility Scale; FCI: Functional Clinical Index; KATZ: Katz Index of Independence in Activities of Daily living; N/R: Not Reported; Ramsey: Ramsey Sedation Scale; RASS: Richmond Assessment Sedation Scale; SAPS II: Simplified Acute Physiology Score; SAS: Riker Sedation Agitation Score.

Safety outcomes: Complications during ECLS

Complications related to ECLS were described either in number of episodes witnessed or number of patients involved (Table 6). Based on the variability in reports, data synthesis was performed considering only the population with available numerical patient data, which is indicated as denominator in brackets.^{8,19,20,22–24,26,29,31,32,34,35,38,40,41,43,44,46,47} Mechanical complications were observed in nine studies^{14,23,25,26,29,38–40,45} with a rate of 3.2% (n = 11/344, and 11 episodes) in case of femoral cannulation and none in cervical cannulation, and 4.2% (n = 19/447, and 7 episode) in cohorts including patients on NIMV irrespectively of the type of cannulation. Patients with femoral cannulation experienced haemorrhagic complications (3.8%, n = 6/344, and 17 episodes) as well. Similarly, 4.4% (n = 14/323, and 11 episodes) of fully extubated patients had haemorrhagic events. Both cardiovascular and unspecified infective complications prevailed in the V-A ECMO (11.3%, n = 19/168; 9.5%, n = 16/168), femoral cannulation (5.8%, n = 20/344; 6.1%, n = 21/344), and fully extubated (6.5%, n = 21/323) cohorts of patients. Neurological issues were rare, and mainly related to ambulation and population supported with NIMV (7.8%, n = 35/477).

Table 6.

Complications reported per patients affected. Total number of patients for each feature is derived only from cohorts with patients showing the same configurations (i.e. total cohort with V-V ECMO mode). Cohorts with more than one configuration were excluded due to the inability of defining features of the patient affected from certain complications. The percentages presented include only the number of patients that have been subjected to complications, however it doesn’t show how many times the event has occurred. The episodes reported are not calculated in the percentage presented, therefore it must be considered as additional events reported for each complication, without clear indications of how many patients were involved.

	Features	Mechanic	Haemorrh	Neuro	Cardiovas	Infection	Reintubation
ECMO modality	V-V	7 (2.5%) 6 additional episodes	4 (1.5%) 6 additional episodes	0	1 (0.4%)	1 (0.4%)	3 (1%)
	V-A	3 (1.8%) 5 additional episodes	6 (3.6%) 12 additional episodes	1 (0.6%) 1 additional episode	19 (11.3%) 4 additional episodes	16 (9.5%) 33 additional episodes	0 10 additional episodes
CANNULA/s configuration	Femoral	11 (3.2%) 11 additional episodes	13 (3.8%) 17 additional episodes	0	20 (5.8%) 4 additional episodes	21 (6.1%) 33 additional episodes	11 (3.2%) 10 additional episodes
	Only upper body	0	5 (3.9%)	1 (0.8%)	1 (0.8%)	1 (0.8%)	0
Physical activity achieved	Ambulation	3 (6%)	2 (1.7%)	1 (2%)	0	0	0
	Until standing	0 6 additional episodes	0 5 additional episodes	0	0	0	0
Respiratory status	Extubated	8 (2.5%) 6 additional episodes	14 (4.4%) 11 additional episodes	1 (0.3%) 1 additional episode	21 (6.5%) 4 additional episodes	21 (6.5%) 33 additional episodes	11 (3.4%) 10 additional episodes
	Partial cohort on NIMV	19 (4.2%) 7 additional episodes	12 (2.7%) 13 additional episodes	14 (3.1%)	14 (3.1%)	19 (4.2%)	19 (4.2%)

Survival and hospital stay

As summarized in Table 7, Intensive Care Unit (ICU) length of stay (LoS) was reported in 35.4% (n = 410/1157)^{20–22,27–29,34,38–41,43,44} of the total population with a median value of 16.94 days (95% CI: 13.76–20.13) while hospital LoS was reported only for 24.6% (n = 285/1157)^{14,21,22,28,29,31,33,38,40,43,44} of patients and the median was 49.79 days (95% CI: 16.99–82.59). Overall survival at discharge was 70.8% (n = 789/1114) with a survival rate of 83.5% (n = 269/322) in patients without IMV. 83.6% (n = 87/104) of ambulatory patients alive at discharge. V-A ECMO configuration presented a higher survival rate (79.6%, n = 250/314) compared to V-V ECMO (72.4%, n = 343/474), based on the reported cases respectively (Tables 3 and 4). Ambulatory patients with only upper body cannula configuration had an in-hospital survival of 75.6% (n = 62/82),^24,35,42,47 while all ambulatory patients with femoral cannulation were discharged successfully at home (100%, n = 15/15). ²⁹ Ambulatory V-A ECMO patients had a survival at discharge of 89% (n = 49/55), while ambulatory V-V ECMO alive at hospital discharge resulted in 58% (n = 18/31).

Table 7.

List of articles included in the systematic review: outcomes.

Author	ECMO time (days)	IMV time (d)	ICU LoS (d)	Hospital LoS (d)	Survival hospital at discharge (n)	30-day survival (n)	1-year survival (n)	BTR achieved (n = survivors)	BTT achieved (n = survivors)	BTNT achieved (n = survivors)
Wells et al.	16.4 ± 17.4	N/R	N/R	N/R	109 (65.3)	N/R	N/R	109	0	0
Hayes et al.	8.1 ± 4.9 (Intensive PT) 10.9 ± 5.5 (Standard PT)	6.2 ± 2.5 (Intensive PT) 9.2 ± 3.8 (Standard PT)	12.9 (7.2–16.7) (Intensive PT) 21.4 (15.5–38.5) (Standard PT)	41.9 (34.3–56.4) (Intensive PT) 34.4 (29.3–87.2) (Standard PT) only survivors	11 (73.3)	N/R	N/R	11	0	0
Mori et al.	2.7 ± 1.4	N/R	24.8 ± 26.3	N/R	16 (84.2)	N/R	N/R	0	2	14
Youn et al.	N/R	N/R	21.5 (16.3–40.3)	22.0 (15.8–33.5)	22 (91.7)	N/R	20 (83.3)	0	22	0
Braune et al	N/R	N/R	N/R	N/R	N/R	N/R	N/R	N/R	N/R	N/R
Xia et al.	26.0 (15.5–64.8)	14.0 (12.0–37.3)	33.0 (22.3–65.5)	46.5 (27.3–84.8)	8 (66.7)	N/R	N/R	8	0	0
Camboni et al.	N/R	N/R	N/R	N/R	19 (52.8)	N/R	N/R	17	2	0
Deng et al.	5.35 (1.75–31)	4.49 ± 1.47	N/R	N/R	9 (88.9)	N/R	8 (66.7)	9	0	0
Boling et al.	18 ± 16	N/R	N/R	N/R	12 (66.7)	N/R	N/R	4	8	0
Hoopes et al.	13.7 (2–53)	N/R	N/R	N/R	30 (96.8)	N/R	29 (93.5)	0	30	0
Ellouze et al.	7 ± 3.1	3.3 ± 1.6	16 ± 14.2	N/R	11 (73.3)	11	N/R	11	0	0
Bataillard et al	N/R	3.7 ± 4.3	N/R	11.8 ± 9	31 (75.6)	N/R	N/R	N/R	N/R	N/R
Pasrija et al.	8 (6–10)	3 (2–7)	12 (10–19)	21 (13–34)	15 (100)	15 (100)	15 (100)	9	4	2
Munshi et al.	13 (10–19)	N/R	N/R	N/R	48 (96)	N/R	N/R	48	0	0
Sommer et al.	11.9 ± 12.9	N/R	N/R	31 (5–140)	21 (91.3)	20	N/R	3	3	15
Tipograf et al.	12 (6–24)	N/R	N/R	N/R	64 (52.9)	N/R	51 (42.1)	0	64	0
Fuehner et al.	9 (1–45)	N/R	18 (1–69) only survivors	38 (20–87)	16 (61.5)	N/R	N/R	0	16	0
Levin et al.	14.7 ± 2.3	23.9 ± 5.8	N/R	31.0 ± 5.0	17 (100)	N/R	N/R	17	0	0
Bonizzoli et al.	9.5 (11–26) only survivors	16 (11–26) only survivors	12 (7–18) only survivors	N/R	75 (74.2)	N/R	N/R	75	0	0
Yanagida et al.	18.2 (1–60)	N/R	N/R	N/R	9 (60)	N/R	N/R	2	7	0
Downey et al.	24.8 (1.9–83.5)	N/R	N/R	N/R	15 (88.2)	15 (88.2)	N/R	0	15	0
Yeo et al.	9.13 ± 2.2 all survivors 3.6 ± 1.7 “Awake” survivors	6.8 ± 4.7 all survivors	14.6 ± 6.8 all survivors	50.3 ± 31.5 all survivors	8 (80)	N/R	N/R	8	0	0
Benazzo et al.	N/R	N/R	N/R	N/R	30 (90.9)	N/R	N/R	0	30	0
Abrams et al.	18.7 ± 13.2 (BTT) 9.1 ± 2.6 (BTR)	N/R	N/R	N/R	23 (65.7)	N/R	N/R	14	9	0
Biscotti et al.	12 (6.25–18.75) (transplanted) 12 (7.5–23) (death/delist)	N/R	19 (11.25–34.25) (transplanted) 20 (12.25–35.25) (death/delist)	25 (13–40) (transplanted) 30 (14–49.75) (death/delist)	37 (51.4)	N/R	28 (38.9)	0	28	0
Taniguchi et al.	16 ± 14	N/R	36 ± 53	N/R	29 (53.7)	N/R	N/R	29	0	0
Hakim et al.	Range: 3–126	N/R	Range: 8–170	Range: 8–125	22 (73.3)	N/R	N/R	0	22	0
Magunia et al.	11 (8–19)	8 (4–18)	22 (14–41)	N/R	38 (88.4)	N/R	N/R	N/R (Partially Yes)	N/R	N/R (Partially Yes)
Chicotka et al.	N/R	N/R	N/R	N/R	25 (67.6)	N/R	N/R	0	25	0
Crotti et al.	8 ± 4 (BTT) 10 ± 6 (COPD) 11 ± 9 (ARDS)	0.1 ± 0.3 (BTT) 1.6 ± 2.4 (COPD) 2.6 ± 3.4 (ARDS)	N/R	N/R	19 (76)	N/R	N/R	N/R	N/R	N/R

Data are reported as mean ± SD, median (interquartile range), range of values from min to max, or n (%) as valid percentage, excluding missing values).

N/R: Not reported; N/A: not available; ECMO: Extracorporeal Mechanical Oxygenation; IMV: Invasive Mechanical Ventilation; ICU: Intensive Care Unit; LoS: Length of Stay; BTR: Bridge to Recovery; BTT: Bridget to Transplant; BTNT: Bridge to Next Therapy; PT: Physical Therapy; ARDS: acute respiratory distress syndrome; BRIDGE: bridge to lung transplant; COPD: chronic obstructive pulmonary disease.

Discussion

In all the studies reviewed, the definition of Awake ECLS strategy was achieved if one of the following statements was recorded: patients in spontaneous breathing without any mechanical ventilatory support, weaned from analgesia and sedation in order to be alert and responsive, treated with early aggressive rehabilitation to allow ambulation during ECLS care.

Main findings

Awake ECLS strategy in relation to physiotherapy was described and performed regardless cannulation approach and type of ECLS. Liberation from IMV and ambulation as the main goal was feasible to be achieved in about half of the population (63.5% and 43.8%). Complications related to the circuit were mostly associated with the ECMO mode (V-V ECMO), femoral cannulation, during walking exercises, and in cohorts with patients either on IMV support and without. Bleeding issues, relatively of minor entity, were predominantly found in V-V ECMO type, only upper body cannulation configuration, during ambulation, and in cohorts with fully extubated patients. Neurological issues were rare, and mainly related to ambulation and population not fully liberated from IMV. Both cardiovascular and infectious complications were reported in those on V-A ECMO, with femoral cannulation, and in fully extubated cohorts of patients. Re-intubation was mainly reported in V-V ECMO type, femoral cannulation, and in cohorts partially supported on IMV. Specific standardized evaluation tools are needed to objectively quantify safety and efficacy of physiotherapy and Awake ECLS strategies.

Common practice in Awake ECLS strategy

Traditionally, the conventional ECLS strategy requires the patient to be under sedation, at times paralysed, and fully supported with IMV while physiotherapy is performed occasionally with passive ROM. The benefits for this approach include a reduced likelihood of local trauma, bleeding, and cannula movements, maintaining the highest control over hemodynamic stability. However, this strategy leads to other kind of complications such as withdrawal syndrome, muscles loss and neuropathy, infections, longer rehabilitation, and decreased quality of life after discharge.

More recently, the Awake ECLS strategy has gained popularity, especially during the COVID-19 pandemic. A universally accepted Awake ECLS strategy definition is yet to be reached, considering that this strategy is continuously evolving (Table 8). Generally, the Awake ECLS approach consists in weaning the sedation and extubating from IMV support, in order to allow the patient in participating to decision making and active physiotherapy. This strategy aims to reduce delirium from sedation and analgesia, ¹⁰ muscles weakness and neuropathy from prolonged immobilization, ⁵⁶ maintain skin integrity and mental wellbeing, as well as active participation to muscle conditioning and respiratory physiotherapy. ⁵⁷ The main challenges faced by both healthcare staff and the patient are linked to cannula dislodgement or circuit malfunction during mobilization, increased agitation caused by being awake during the treatment, bleeding and/or local trauma, respiratory distress requiring re-intubation and the resource need of a large multidisciplinary team that is able to perform the intervention safely.

Table 8.

Criteria of “Awake” ECLS strategy definition described in each article included in the systematic review.

Author, YearN. total population	Cohorts (Int/Control)Survival at hospital discharge (N, %)	Definition features/criteria
Wells, 2018 N. 254	ECMO with PT Survival at HD 109 (65.3%)	RASS > - 2 (light sedation) Ambulatory status achieved
	ECMO without PT Survival at HD 36 (41.4%)	RASS < - 2 (moderate sedation, no eye contact)
Hayes, 2020 N. 15	ECMO with intensive PT Survival at HD 4 (57.1%)	RASS between +1 and −1 IMS score Rehab intensity based on BORG level 3 to 5
	ECMO with standard PT Survival at HD 7 (87.5%)	RASS between +1 and −1 IMS score
Mori, 2019 N. 19	AWAKE on ECMO Survival at HD 16 (84.2%)	Unspecified assessment to follow instructions and mental wakefulness
	N/A
Youn, 2020 N. 102	Early extubation on ECMO Survival at HD 22 (91.7%)	Extubated within 48 h from ECMO cannulation CAM-ICU score for delirium
	Deferred extubation on ECMO Survival at HD 19 (75.6%)	Extubated after 48 h from ECMO cannulation to protect airway in situations of altered mental status resulting in cardiopulmonary resuscitation or elective procedures
Braune, 2020 N. 115	Mobilized on ECMO Survival at HD (N/A)	RASS > - 2 (light sedation)
	Non-mobilized on ECMO Survival at HD (N/A)	RASS < - 2 (moderate sedation, no eye contact)
Xia, 2020 N. 12	AWAKE on ECMO Survival at HD 10 (66.7%)	Clear consciousness Strong airway protection capability Extubated on RA/NIMV Hemodynamic stability Ramsey sedation score
	N/A
Camboni, 2012 N. 36	AWAKE on ECMO Survival at HD 19 (52.8%)	Sedation stopped = declared awake Extubation in 11.1% Mobilisation performed in 25%
	N/A
Deng, 2020 N. 40	AWAKE on ECMO Survival at HD 9 (75%)	Extubated Oral intake Mobilization performed
	Asleep and sedated on ECMO Survival at HD 3 (10.7%)	Sedated Haemodynamically unstable
Boiling, 2020 N. 18	AWAKE on ECMO Survival at HD 12 (66.7%)	Extubated Ambulatory status achieved (100%)
	N/A
Hoopes, 2013 N. 31	AWAKE on ECMO Survival at HD 30 (96.8%)	Extubated Ambulatory status achieved (61.3%), after weaned of vasopressor completed
	N/A
Ellouze, 2019 N. 57	NIMV on ECMO Survival at HD 11 (73.3%)	RASS > - 2 (light sedation) Extubated
	IMV on ECMO Survival at HD 17 (40.5%)	RASS < - 2 (moderate sedation, no eye contact)
Bataillard, 2017 N. 177	NIMV on ECMO Survival at HD 11 (26.8%)	RASS > - 2 (light sedation) Extubated
	IMV on ECMO Survival at HD 106 (77.9%)	RASS <-2 (moderate sedation, no eye contact)Multi organ failureNeuro impairment
Pasrija, 2019 N. 15	Early extubation on ECMO Survival at HD 15 (100%)	Extubated Ambulatory status achieved (100%)
	N/A
Munshi, 61 N. 177	PT on ECMO Survival at HD 49 (98%)	SAS score (maximum achieved score 4, “calm and collaborative,” correlated with IMS ≥ 2) IMS score Including also intubated patients on active physiotherapy
	No-PT on ECMO Survival at HD 4 (36.4%)	SAS score (maximum achieved score 3, “sedated,” correlated with IMS ≤ 1) IMS score
Sommer, 2019 N. 23	AWAKE on ECMO Survival at HD 21 (91.3%)	Extubated
	N/A
Tipograf, 2019 N. 121	AWAKE on ECMO Survival at HD 64 (52.9%)	Extubation and sedation weaning after ECMO cannulation in operating room. Ambulatory status achieved (67.7%)
	N/A
Fuehner, 2012 N. 26	AWAKE on ECMO Survival at HD 16 (61.5%)	Extubated
	IMV on ECMO Survival at HD 12 (33.3%)	Intubated on IMV
Levin, 2021 N. 65	AWAKE, off Positive pressure on ECMO Survival at HD 17 (100%)	Extubated or tracheotomy off positive pressure
	IMV Positive pressure on ECMO Survival at HD 22 (45.8%)	Intubated on IMV on positive pressure
Bonizzoli, 2021 N. 101	AWAKE on ECMO Survival at HD 75 (74.2%)	IMS score Ambulatory status achieved (9.9%)
	N/A
Yanagida, 2019 N. 15	Mobilized on ECMO Survival at HD (N/A)	Ambulatory status achieved (73%)
	N/A
Downey, 2019 N. 17	AWAKE on ECMO Survival at HD 15 (88.2%)	Extubated Ambulatory status achieved (94.1%) (Technique of minimally invasive central VA ECMO that spare sternum)
	N/A
Yeo, 2016 N. 10	AWAKE on ECMO Survival at HD 8 (80%)	Extubated Ramsey score ≥ 2 Antipsychotics or alpha-2 agonist administered intermittently to monitor agitation or anxiety
	N/A
Benazzo, 2019 N. 120	AWAKE on ECMO Survival at HD 30 (90%)	Extubated (recently institutional guidelines that patients with increasing PCO2 >70 mmHg in arterial blood, despite continuous positive airway pressure therapy, are candidates for initiation of awake ECLS implementation.)
	Asleep and sedated on ECMO Survival at HD 75 (86.2%)	Sedated
Abrams, 2014 N. 35	AWAKE on ECMO Survival at HD 23 (65.7%)	IMS score Either IMV and NIMV participated in active physiotherapy Ambulatory status achieved (51.4%)
	N/A
Biscotti, 2017N. 72	AWAKE on ECMO Survival at HD 37 (51.4%)	Early tracheostomy for pulmonary toilet or HF nasal cannula Criteria for physiotherapy: Hemodynamic stability, secure cannulas without active bleeding, cooperative status
	N/A
Taniguchi, 2021 N. 54	AWAKE on ECMO Survival at HD (N/A)	Extubated Participating in physiotherapy
	N/A
Hakim, 2018 N. 19	AWAKE on ECMO Survival at HD 9 (100%)	Extubated Participating in physiotherapy
	N/A
Magunia, 2020 N. 43	AWAKE on ECMO Survival at HD 38 (88.4%)	Extubated
	N/A
Chicotka, 2010 N. 37	AWAKE on ECMO Survival at HD 25 (67.6%)	Ambulatory status achieved (100%)
	N/A
Crotti, 2010 N. 25	AWAKE on ECMO Survival at HD 19 (76%)	Spontaneously breathing Clear consciousness
	N/A

BORG: Borg Rating of perceived Exertion; CAM-ICU: Confusion Assessment Method for the Intensive Care Unit; CFS: Clinical Frailty Scale; ECMO: Extracorporeal Membrane Oxygenation; FCI: Functional Clinical Index; IMV: Invasive Mechanical Ventilation; Int: Interval; IMS: ICU Mobility Scale; KATZ: Katz Index of Independence in Activities of Daily living; NIMV: Non Invasive Mechanical Ventilation; N/A: Not available; N/R: Not Reported; PT: Physical Therapy; RA: Room Air; Ramsey: Ramsey Sedation Scale; RASS: Richmond Assessment Sedation Scale; SAPS II: Simplified Acute Physiology Score; SAS: Riker Sedation Agitation Score; Survival at HD: Survival at Hospital Discharge.

Beyond these “technical” problems, another important task is patient selection criteria, timing and speed of initiation. ⁵⁶ Through our analysis we found that the exclusion criteria for Awake ECLS were sedation score RASS ≤ 2 or agitation score SAS <3, coma without sedation, lack of manpower, hemodynamic instability, respiratory instability, unstable cannula, multi organ failure, neurological impairment, major bleeding; while femoral cannulation and V-A ECMO specifically for the ambulation exclusion.^24,26,35

Biscotti et al. ³⁸ mentioned, instead, a defined list of inclusion criteria to proceed with physiotherapy (hemodynamic stability, secure cannula, no active bleeding, cooperative patient), and further criteria to advance the patient up to ambulation (bed sitting and standing successfully achieved without airway or hemodynamic complications). Moreover, they highlighted their purpose to liberate patients from IMV either with early tracheostomy or NIMV such as high flow nasal cannula.

Evaluation tools

The assessment and evaluation of level of mobilization and sedation in patients supported with Awake ECLS are fundamental to monitor safety and efficacy of this practice. There was an evident heterogeneity of assessment scale for either mobilization and sedation (Table 5): three different scales adopted for assessing patient’s level of mobilization, and three for the sedation assessment. A similar level of heterogeneity was found in the assessment of patient’s functionality and mortality prediction on ICU admission. The accurate description of mobilization intensity and spontaneous breathing status allows to have an insight in patient’s level consciousness and collaboration. There is no consciousness level evaluation scale of active patients yet. Patient’s readiness for Awake ECLS strategy was based mainly on sedation assessment scale (RASS, SAS, Ramsey) as an initial screening, potentially followed by pain evaluation scale which, however, were not mentioned in the studies analysed in this review.

Physiotherapy during ECLS

Once the patient has achieved a good degree of consciousness, a physiotherapeutic strategy can be planned. The first emphasis on physiotherapy started with Wells et al. ¹⁹ and Bonizzoli et al. ³⁴ who witnessed a better ICU Mobility Scale value after early physiotherapy, which also led to a shorter ECMO duration, IMV time and in-hospital stay. Similar outcomes were supported by studies investigating the ambulatory status of ECLS patients as determinant for a better outcome.^35–37,40 Interestingly, Abrams et al. ³⁷ and Hayes et al. ⁴³ specifically stressed the absence of changes in ECMO blood flow or sweep gas flow in ambulatory patients. Moreover, Hayes et al. ⁴³ observed an association between higher mobility, reduced inotropic dosage and longer exercise duration. Braune et al. ⁴⁵ reported sedation as the main cause for patient to not be mobilized. Bonizzoli et al. ³⁴ reported additional benefits of ambulation during ECLS: reduction of sedation and muscle wasting, as well as improved mental confidence. Youn et al. ²¹ presented positive results in infection and delirium reduction when early extubation ECMO strategy was performed. Finally, Abrams et al. ³⁷ presented an insightful perspective of cost-effectiveness of Awake ECLS strategy based on early high intensity mobilization and its directly associated weaning of medications and invasive mechanical supports.

Despite the difficulty to retrieve all information regarding cannulation approaches and the disproportion in the different groups of support, we overcame two important barriers: the site of cannulation and the severity of the disease. For the first one, we found that overall survival at hospital discharge for femoral cannulation was similar to the exclusively upper body cannulation indicating that, apparently, type of ECLS cannulation does not affect the success of early physiotherapy. ²⁹ For the second, we observed that patients bridging to lung transplant were the ones reviewed with more consideration, mostly because of their longer hospital stay. Positive remark was addressed to physiotherapy given the improvements seen in reduced muscle loss, neuropathy and nutritional optimization, which are all likely to preserve the patient vitality, ⁴⁰ extend transplantation candidacy ⁴² and lead to better transplant outcomes.^22,23,38 Chicotka et al. ⁴² showed that the implementation of ambulation during ECLS bridging to transplant led to 80% reduction in mortality risk while awaiting transplantation.

Criteria of inclusion for physiotherapy are mandatory to establish patient safety standards: secured cannulas without active bleeding, ¹⁹ hemodynamic stability, cooperative patient, ⁴⁵ ambulation attempted if patient performing safely the standing, ³⁸ appropriate physical strength ²⁹ and respiratory stability. ⁴⁵ Moreover, with the aim of facilitate safe mobilization, it is recommended to ensure enough tubing length from the patient to circuit for movement, tighten connections and assess for leaks and thrombi.

Pasrija et al. ²⁹ reported a few changes to medically prepare patients on femoral V-A ECMO for safe ambulation, such as withdrawn of intra-aortic balloon pump when present, and surgical atrial septostomy in patient who required left ventricular decompression. Likewise, Hoopes et al. ²⁶ described the conversion from a pumpless extracorporeal lung assist to V-V ECMO with double lumen cannulation, and the conversion from centrifugal pump and inotropic requirement to V-AV ECMO or upper body cannulation, in preparation for ambulation. Despite the conventional prejudice that prevents mobilization of patients on invasive mechanical ventilator support, ³⁷ Braune et al. ⁴⁵ demonstrated the feasibility of active physiotherapy during ECMO on a population mostly supported with femoral cannulation and not spontaneously breathing.

Finally, to achieve a successful early rehabilitation program, a well-trained multidisciplinary team is required to complement the work of the advanced ICU ECMO-trained physiotherapist.^{19,30,37,40,42} An experienced team was deemed necessary to identify potential barriers, optimal timing, dosage, and safety profile, but also to allow a higher number of patients to perform ambulation and achieve successfully spontaneously breathing.^{19,30,37,40,42} A daily input from the physiotherapy team was reported to be an important determinant for accurate patient screening.^19,34,37

Safety

Several studies reported the occurrence of complications during the Awake ECLS strategy and mobilization. Among all the adverse events mentioned, the majority occurred in patients with femoral cannulation and in cohorts with mixed respiratory status (partially intubated and extubated). However, the cohort of fully extubated patients reported a high number of patients affected, and an even higher number of episodes compared to the cohorts with mixed respiratory status. Interestingly, among the comprehensive list of complications, the presence of infection resulted in the highest incidence of patients affected and episodes reportedmainly linked to V-A ECMO mode, femoral cannulation, and non-intubated status.

Survival

The success of the Awake ECLS strategy, mainly reported by the authors of articles included in this systematic review as free from invasive mechanical ventilation and alert level of consciousness, is expressed by the survival at hospital discharge of nearly 80% of patients who were non-intubated during ECMO support. In the extended meaning of Awake ECLS approach which includes early physical rehabilitation until ambulation, the positive outcome was confirmed by the 73.4% (n = 94/128) of ambulatory patients surviving at hospital discharge.

Limitations

Several limitations apply to this systematic review which should be considered for its interpretation. Most included articles were observational studies, and the paucity of randomized trials increases the risk of bias. Nineteen studies comprised a population of less than 50 patients each with a high likelihood of type II error and 21 references were lacking a conventional ECLS control group.^{8,14,20–24,26,29,31,32,34,35,38–42,44,46,47} Despite the attempt to highlight differences in the use of different ECMO configurations and their potential benefits related to various features (Tables 3 and 4), the large missing data constitutes a powerful obstacle to an accurate representation of the real occurrences. These studies are probably subject to further confounding and additional risks of recall bias, selective outcome reporting and selective analysis reporting biases. The rate of missing data was considerable as well as the variability in outcomes definition. This might reflect a general difficulty in clinical evaluation of Awake ECLS strategy outcome, safety, and feasibility. Due to the observational nature of most studies, it was not possible to identify survival at discharge for all patients who successfully completed physical therapy sessions out of bed.^{19,24,30,34,36–39,43} This systematic review wasn’t deemed eligible for a suitable meta-analysis as the observational studies with both interventional and control groups were not comparable with regards to similar interventions and therefore a suitable population was not granted as sufficient for the in-depth analysis. However, we would like to highlight the observational studies, which could have potentially been considered for a meta-analysis, presenting valuable comparison groups between conservative and awake ECLS (with variable such as liberation from IMV and/or physiotherapy): Wells et al. (ECMO with/without physiotherapy), Youn et al. (early and deferred liberation from IMV), Braune et al. (ECMO with and without mobilization), Deng et al. (awake and sedated on ECMO), Bataillard et al. (ECMO with NIMV or IMV), Munshi et al. (ECMO with and without mobilization), Fuehner et al. (IMV and extubated on ECMO), Levin et al. (IMV and extubated on ECMO), Benazzo et al. (awake and sedated on ECMO).

Conclusion

In conclusion, our study shows that Awake ECLS strategy is safe and feasible with different cannula approaches, different type of ECLS, different destination therapy, varied level of physiotherapy achieved until ambulation and a complete liberation from invasive mechanical ventilation support. This review supports the exploration for a deeper and multifaced definition of the concept “Awake ECLS,” since the effects of active mobilization has been witnessed to be more beneficial irrespective of the respiratory status tolerated by the patient. However, a more comprehensive and consistent evaluation protocol for mobilization, including the involvement of a multidisciplinary team, sedation, delirium, and pain score is strongly recommended and required for future randomized controlled multicentre trials.

Glossary of Abbreviations

APACHE

acute physiology and chronic health evaluation

BORG

Borg rating of perceived exertion

BTR

bridge to recovery

BTT

bridge to transplant

BTNT

bridge to next therapy

CAM-ICU

confusion assessment method Intensive Care Unit

CFS

clinical frailty scale

ECLS

extracorporeal life support

ECMO

extracorporeal membrane oxygenation

FCI

functional comorbidity index

ICU

intensive care unit

IMS

ICU mobility scale

PRISMA

preferred reporting items for systematic review and meta-analyses

RASS

Richmond agitation-sedation scale

RESP

respiratory ECMO survival prediction

RoM

range of movements

SOFA

sequential organ failure assessment

SAPS

simplified acute physiology score

SAS

sedation-agitation scale

V-A ECMO

veno-arterial extracorporeal membrane oxygenation

V-V ECMO

veno-venous extracorporeal membrane oxygenation

ORCID iDs

Marta Cucchi https://orcid.org/0000-0003-3221-8795

Emma Shkurka https://orcid.org/0000-0001-7344-6925

Roberto Lorusso https://orcid.org/0000-0002-1777-2045