How are you taking part in this consultation?

You will not be able to change how you comment later.

You must be signed in to answer questions

    Evidence summary

    Population and studies description

    This interventional procedure overview is focused on VA ECMO in postcardiotomy cardiogenic shock. Two additional overviews have been developed focusing on VA ECMO use in severe acute heart failure and as extracorporeal cardiopulmonary resuscitation (ECPR). Some of the evidence includes a mix of indications and has been presented in more than one overview.

    This interventional procedures overview is based on approximately 49,500 people from 4 systematic reviews (Biancari 2018, Wang 2018, Kowalewski 2020, Alba 2021), 3 retrospective registry studies (4 publications; Kowalewski 2021, Loungani 2021, Biancari 2020, 2021), 1 multicentre retrospective study (Mariani 2023) and 4 single centre retrospective studies (Chen 2017, Danial 2023, Aboud 2024, Rubino 2017). There were 29 overlaps accounting for 3,830 people in primary studies included across 4 systematic reviews (Biancari 2018, Wang 2018, Kowalewski 2020, Alba 2021). A UK study (Rubino 2017) was included in the systematic review by Kowalewski (2020). This is a rapid review of the literature, and a flow chart of the complete selection process is shown in figure 1. This overview presents 12 studies (13 publications) as the key evidence in table 2 and table 3, and lists 56 other relevant studies in table 5.

    The 4 systematic reviews of observational studies included in the key evidence included studies from Asia, Australia, Europe, North America and South America (Wang 2018, Alba 2021), however 2 systematic reviews did not report study location (Biancari 2018, Kowalewski 2020). Three registry studies in the key evidence were done from the Extracorporeal Life Support Organisation (ELSO) which collates data worldwide (Kowalewski 2021), from the RESCUE registry collating data from 3 centres across the US (Loungani 2021), and from the postcardiotomy extracorporeal membrane oxygenation (PC-ECMO) registry gathering data from Asia and Europe (Biancari 2020, 2021). The included propensity matched retrospective study was done at a single centre in Taiwan (Chen 2017), and the 3 other single-centre studies included were done in Germany, France and the UK (Aboud 2024, Danial 2023, Rubino 2017). A multicentre retrospective study from Europe and Asia (Mariani 2023) was also included.

    All key evidence studies included people who needed VA ECMO after cardiac surgery. Two systematic reviews (Wang 2018, Kowalewski 2020), 3 studies (Kowalewski 2021, Aboud 2024, Biancari 2021) specifically reported on people with postcardiotomy cardiogenic shock (PCS). One systematic review (Alba 2021), and 1 single centre retrospective study (Danial 2023) included people with cardiogenic shock of multiple aetiologies, and 1 registry study included people who had VA ECMO for several aetiologies (Loungani 2021).

    The systematic review by Biancari et al. (2018) included 31 observational studies reporting on 2,986 adults needing VA ECMO after cardiac surgery. Most primary studies included in this study had populations with a mix of cardiac surgery procedures (29 studies) and 2 studies included isolated coronary surgery patients. The mean age was 58 years and 31% of the population were female. Meta-analyses of the studies pooled survival outcomes from studies with follow-ups of 30 days and hospital discharge.

    The systematic review by Wang et al. (2018) included 20 observational studies reporting on 2,877 people with postcardiotomy cardiogenic shock who had ECMO treatment. Risk of bias across studies included in the review was considered high as all studies were retrospective in nature. The baseline characteristics (age and percentage male) of the people in the included studies was not reported. Meta-analyses of the studies reported pooled survival outcomes at hospital discharge, at 1-year, and midterm survival (defined as 3 to 5 years).

    The systematic review by Kowalewski et al. (2020) included 54 observational studies reporting on 4,421 people with postcardiotomy refractory cardiogenic shock. It included people who had CABG, valvular surgery and combined surgery at specialist heart transplant and non-heart transplant centres. Studies were considered to have a moderate to severe risk of bias. The age of people included in the studies ranged from 41 to 77 years, and 49% to 93% of the population were female. Meta-analyses of the studies pooled survival outcomes from studies with follow-up to hospital discharge.

    The systematic review by Alba et al. (2021) included 306 observational studies reporting on 29,289 people with cardiogenic shock of any aetiology. This included 8,231 people with postcardiotomy cardiogenic shock. Risk of bias across studies was considered low in 219 (72%), moderate in 81 (26%), and high in 6 (2%) studies. The age of people included in the studies ranged from 47 to 61 years, and 22% to 59% of the population were female. Meta-analyses of the studies pooled short-term outcomes from studies with follow-ups of 30 days and hospital discharge.

    The single centre retrospective study by Chen et al. (2017) was the only comparative study included in the key evidence. It used propensity score matching (PSM) to compare outcomes between people admitted for cardiac surgery (CABG or valve surgery) who had PCS and ECMO (n=1,137) and those who did not have PCS (or ECMO) following cardiac surgery (n=5,685). The mean age was 64 years and 71% of the population were male. Outcomes were reported for a follow-up period until hospital discharge and up to 10 years.

    The retrospective ELSO registry study by Kowalewski et al. (2021) reported efficacy and safety outcomes for 7,185 adults having VA ECMO for intra-operative failure to wean from CPB due to right, left or biventricular failure, and post-operative refractory cardiogenic shock or cardiac arrest during the hospitalisation after the surgical procedure. This included people whose primary procedure was CABG, valvular surgery, heart transplant and combined surgery. The mean age was 56 years and 68% of the population were male. Outcomes were reported for a follow-up period until hospital discharge.

    The retrospective RESCUE registry study by Loungani et al. (2021) reported efficacy and safety outcomes for 723 adults treated with VA ECMO, including those with persistent circulatory failure postcardiotomy (31%). The mean age was 57 years and 70% of the population were male. Outcomes were reported for a follow-up period until hospital discharge.

    The single centre retrospective study done in France by Danial et al. (2023) included 1,253 adults treated with peripheral VA ECMO for cardiogenic shock, 297 of which were postcardiotomy patients (excluding primary graft dysfunction [PGF] following heart transplant). The mean age was 55 years and 30% of the population were female. Outcomes were reported for a follow-up period until hospital discharge and 5 years.

    The single-centre retrospective study done in Germany by Aboud (2024) included 576 people having ECMO for PCS. The mean age was 65 years and 37% of the population were female. Outcomes included both short-term and long-term (up to 15 years) results. The retrospective international multicentre observation PELS-1 study by Mariani (2023) included 2,058 people having postcardiotomy VA ECMO. The median age was 65 years, with women accounting for 41% of the population. The median follow up was 2.5 years.

    The multicentre (19 centres) retrospective registry study by Biancari (2020) included 781 people having postcardiotomy VA ECMO for refractory cardiopulmonary failure. The mean age was 63 years and 32% of the population were female. Of the 19 centres, 17 agreed to collect data on late all-cause mortality. So, 665 people having VA ECMO for PCS were included in the Biancari (2021) study which reported 5-year survival.

    The single-centre retrospective cohort study by Rubino (2017) included 101 people having central VA ECMO after cardiac surgery. The mean age was 57 years, with female accounting for 37% of the population.

    Table 2 presents study details.

    Figure 1 Flow chart of study selection

    Table 2 Study details

    Study no.

    First author, date

    country

    Characteristics of people in the study (as reported by the study)

    Study design

    Inclusion criteria

    Intervention

    Follow up

    1

    Biancari, 2018

    Countries not reported

    n=2,986

    Mean age=58.1 years

    Female= 30.9%

    Procedure types prior to ECMO:

    • Isolated coronary surgery (2 studies)

    • Mixed cardiac surgery procedures (29 studies)

    • Proportion of HTx patients in included studies: 4.4% (28 studies, n=2,879)

    Systematic review and meta-analysis of 31 studies.

    Search date: Sept 2016

    Adults who required VA ECMO after cardiac surgery procedure.

    VA ECMO

    1 year

    2

    Wang 2018,

    USA, Taiwan, Germany, Italy, China

    n=2,877

    Mean age not reported

    Male % not reported

    Procedure types prior to ECMO:

    • CABG: 18 studies

    • Valve procedure: 14 studies

    • Aortic surgery: 6 studies

    • Heart transplant: 5 studies

    • Other: 9 studies

    Systematic review and meta-analysis of 20 retrospective studies.

    People after cardiac surgery with postcardiotomy cardiogenic shock (PCS).

    VA ECMO

    In-hospital, 1 year

    3

    Kowalewski 2020

    Countries not reported

    n=4,421

    Age (years): Range 41 to 77

    Male %: Range 49 to 93

    Procedure types prior to ECMO:

    • CABG

    • Valvular surgery

    • Combined surgeries

    Systematic review and meta-analysis of 54 retrospective studies.

    Search date: March 2018

    Postcardiotomy refractory cardiogenic shock

    VA ECMO

    In-hospital

    4

    Alba, 2021

    Europe, Asia, North America, South America, Australia

    n=29,289

    Age (years): Range 47 to 61

    Female %: Range 22 to 59

    Indication

    • ECPR: 7,814 (113 cohorts)

    • Post-AMI: 7,774 (80 cohorts)

    • Postcardiotomy: 8,231 (64 cohorts)

    • Post-HTx: 771 (25 cohorts)

    • Heart failure: 3,567 (33 cohorts)

    • Myocarditis: 906 (13 cohorts)

    Pulmonary embolism: 221 (10 cohorts)

    Systematic review and meta-analysis of 306 observational studies.

    Search date: June 2019

    Adults (aged 18 and over) with cardiogenic shock of any aetiology, with VA ECMO implantation.

    VA ECMO

    Concomitant IABP: Range 20 to 67%

    30 day or in-hospital

    5

    Chen, 2017

    Taiwan

    n=1,137 ECMO (5,685 propensity matched cohort)

    Mean age (years)=63.8 (SD 13.2)

    Male=71.2%

    Procedure types prior to ECMO:

    • CABG alone: 63.9% (728)

    • Valve alone: 24.2% (275)

    • CABG + Valve: 11.8% (134)

    Propensity score-matched retrospective single centre study.

    Adults (aged over 18 years) admitted for cardiac surgery (CABG or valve surgery)

    Intervention: VA ECMO (people with PCS)

    Comparator: No VA ECMO (people without PCS) (propensity score matched)

    In-hospital, 10 years

    6

    Kowalewski, 2021

    Worldwide

    n=7,185

    Mean age (years)=56.3 (range 18 to 86)

    Male=67.5%

    Procedure types prior to ECMO:

    • CABG: 26.8%

    • Valvular surgery: 25.6%

    • Heart transplant: 20.7%

    • CABG with valve: 13.4%

    • CABG with VAD: 8.5%

    Retrospective ELSO Registry study

    Search date: 2010 to 2018

    Adults over 18 years old undergoing a single run VA ECMO for refractory PCS.

    People with pre-operative ECMO were excluded.

    VA ECMO initiated for intra-operative failure to wean from CPB due to right, left or biventricular failure, and post-operative refractory cardiogenic shock or cardiac arrest during the hospitalisation after the surgical procedure.

    In-hospital

    7

    Loungani, 2021

    US

    n=723

    Median age (years)=57

    Male=69.6%

    Indication

    • Postcardiotomy (30.7%)

    • Cardiomyopathy (26.1%)

    • MI (16.9%)

    • Non-cardiogenic shock (11.3%)

    • HTx/graft dysfunction (8.2%)

    • Other cardiogenic shock (6.8%)

    Retrospective RESCUE Registry study

    Search date: 2007 to 2017

    Adult patients (over 18 years old) treated with ECMO.

    VA ECMO

    In-hospital

    8

    Danial, 2023

    France

    n=1,253 (n=297 postcardiotomy excluding PGF)

    Mean age (years)=54.8 (SD:14.9)

    Female=30%

    Indication

    • Postcardiotomy excluding PGF (n=297)

    • PGF (n=245)

    • AMI (n=233)

    • Cardiomyopathy (n=171)

    • Fulminant myocarditis (n=47)

    • Massive PE (n=41)

    • Sepsis induced cardiogenic shock (n=29)

    • Refractory vasoplegia shock (n=9)

    • Drug overdose (n=25)

    • Arrhythmic storm (n=30)

    • Other/unknown aetiology (n=126)

    Single centre retrospective study

    Search date: 2015 to 2018

    Adult patients (over 18 years old) treated with peripheral VA ECMO for cardiogenic shock.

    VA ECMO

    In-hospital, 5 year

    9

    Aboud A, 2024

    Germany

    n=576

    Mean age=65 years

    Female gender=37.3%

    Mean EuroSCORE 2=14.4%

    21.7% had isolated coronary bypass, 16.5% single valve surgery, 34.3% combined cardiac surgery and 13.2% had heart transplantation.

    Retrospective single-centre cohort study

    Procedures were done between 2008 and 2017

    All people who had ECLS for PCS were included. All types of cardiac operations were included, except people who had primary surgery for left ventricular assist device implantation.

    Postcardiotomy ECLS

    The median duration on ECLS was 7.4 days. A peripheral ECLS implantation was done in 350 cases (61%). In 36 people (6%), an IABP was implanted before surgery.

    Up to 15 years

    98.6% completeness of follow-up.

    10

    Mariani S, 2023

    Austria, Australia, Belgium, Chile, China, Colombia, Czech Republic, France, Germany, Italy, Lithuania, Singapore, South Korea, Thailand, The Netherlands, US

    n=2,058

    Median age=65.0 years

    Women=41%

    Race or ethnicity

    • Asian=8.8%

    • Black=0.8%

    • Hispanic=4.1%

    • White=77.1%

    • Other or unknown=9.2%

    Median EuroSCORE2=7.53 (IQR 3 to 18.5)

    Retrospective international multicentre observational PELS-1 (Postcardiotomy Extracorporeal Life

    Support) study

    2000 to 2020

    Adults (aged 18 years and over) were included if they had postcardiotomy ECMO between January 2000 and December 2020. Inclusion criteria required cardiac surgery before ECMO (including VA ECMO and veno-venous ECMO). Exclusion criteria comprised ECMO support after discharge or before surgery, ECMO support after noncardiac surgical procedures, and ECMO implantation not strictly related to cardiac surgery hospitalisation.

    Postcardiotomy VA ECMO

    (An additional 33 people had venovenous ECMO and were excluded from the analysis)

    Median ECMO duration was 118 hours (IQR 60 to 192).

    Cardiogenic shock was reported as the indication for ECMO in 25% (n=506) of people, and failure to wean was the indication in 39% (n=788). Other indications included acute pulmonary embolism, arrhythmia, cardiac arrest, pulmonary haemorrhage, right ventricular failure, respiratory failure and biventricular failure.

    Median 2.5 years (in hospital survivors)

    11

    Biancari F, 2020

    Belgium, Czech Republic, Finland, France, Italy, Germany, Saudi Arabia, Sweden, and UK

    n=781

    Mean age=63.1 years

    Female gender=31.9%

    Complex cardiac surgery=40% (surgery on more than 1 heart valve, aortic surgery, repair of ventricular wall or septal defect, and repair of complex congenital defects)

    Coronary artery bypass graft=50%

    Retrospective multicentre cohort study (postcardiotomy extracorporeal membrane oxygenation

    [PC-ECMO] registry)

    January 2010 to March 2018

    All participating centres were tertiary referral hospitals.

    9 centres treated more than 30 people, and 6 centres treated more than 50 with postcardiotomy VA ECMO.

    Aged more than 18 years, VA ECMO needed for refractory cardiopulmonary failure occurring during the index hospitalisation after any surgical procedures on the heart valves, coronary arteries, ascending aorta, aortic arch or ventricular wall and septum, grown-up congenital heart diseases, and chronic thromboembolic pulmonary hypertension were considered for this analysis. Cardiopulmonary failure in these people was considered not treatable with inotropes and intra-aortic balloon pump.

    Exclusion criteria: people who were on any ECMO before cardiac surgery or who needed VA ECMO after implantation of a ventricular assist device (VAD) or heart transplantation.

    Postcardiotomy VA ECMO

    Mean VA ECMO duration=6.9 days

    • Inserted at primary surgery=60.7%

    • After weaning attempt with inotropes only=45.3%

    • After weaning attempt with intra-aortic balloon pump=15.2%

    • After weaning attempt with Impella=0.1%

    Arterial cannulation site

    • Ascending aorta=31.4%

    • Femoral artery=59.8%

    Other peripheral artery=8.8%

    Mean or median follow-up not reported

    Biancari F, 2021

    Belgium, Czech Republic, Finland, France,

    Germany, Italy, Saudi Araba, Sweden, and the UK

    n=665

    Mean age=62.5 years

    Female=32.6%

    Retrospective multicentre cohort study (postcardiotomy extracorporeal membrane oxygenation

    [PC-ECMO] registry)

    January 2010 to March 2018

    17 of the 19 participating centres agreed to collect data on late all-cause mortality.

    People with PCS following cardiac surgery. Additional detailed information in Biancari (2020)

    Please see Biancari (2020)

    Mean follow-up of overall cohort=1.7 years (median, 0.04 years, IQR 0.1 to 3.2). Mean follow-up of post-discharge patients=4.6 years (median, 4.4 years, IQR 2.9 to 6.5).

    12

    Rubino A, 2017

    UK

    n=101

    Mean age=57.1 years

    Female=36.6%

    Surgical groups:

    • Pulmonary thromboendarterectomy, n=29

    • Transplant, n=22

    • CABG and valvular, n=16

    • CABG and other, n=13

    • Valvular, n=12

    Other, n=9

    Single-centre retrospective cohort study

    Tertiary centre

    March 2008 to July 2016

    All consecutive patients who had been supported with central VA ECMO after cardiac surgery.

    Central VA ECMO

    Median duration of ECMO=5 days

    ECMO was started at the end of surgery in 63 people (62.4%), postoperatively to within 24 hours of surgery for 14 (13.9%), between the first and fourth postoperative days for 18 (17.8%), and after the fourth postoperative day for 6 (5.9%).

    Mean or median follow-up not reported.

    Table 3 Study outcomes

    First author, date

    Efficacy outcomes

    Safety outcomes

    Biancari, 2018

    Pooled hospital survival

    Meta-analysis random effects model, 31 studies (n=2,986)

    • 36.1% (95% CI 31.5 to 40.8), I2=84%

    Pooled 1-year survival (Kaplan-Meier estimate)

    Meta-analysis random effects model, 11 studies (n=1,290)

    • 30.9% (95% CI 24.3 to 37.5), I2=82%

    Pooled weaning from VA ECMO

    Meta-analysis random effects model, 24 studies (n=2,049)

    • 59.5% (95% CI 54.6 to 64.3), I2=77%

    Pooled rate of post-ECMO HTx

    Meta-analysis random effects model, 21 studies (n=1,685)

    • 1.9% (95% CI 1.0 to 2.8), I2=50%

    Pooled hospital survival of post-ECMO HTx recipients

    Meta-analysis random effects model, 7 studies (n=18)

    • 66.2% (95% CI 48.2 to 84.1), I2=0%

    Pooled rate of post-ECMO VAD implantation

    Meta-analysis random effects model, 21 studies (n=1,685)

    • 2.3% (95% CI 1.3 to 3.4), I2=57%

    Pooled hospital survival of post-ECMO VAD recipients

    Meta-analysis random effects model, 9 studies (n=45)

    • 45.6% (95% CI 28.0 to 63.1), I2=43%

    Pooled rate of reoperation for bleeding

    Meta-analysis random effects model, 18 studies (n=1,779)

    • 42.9% (95% CI 34.2 to 51.5), I2=93%

    Pooled rate of major neurological events

    Meta-analysis random effects model, 16 studies (n=1,736)

    • 11.3% (95% CI 7.8 to 14.8), I2=79%

    Pooled rate of limb ischaemia

    Meta-analysis random effects model, 16 studies (n=1,909)

    • 10.8% (95% CI 8.0 to 13.5), I2=70%

    Pooled rate of lower limb amputation

    Meta-analysis random effects model, 5 studies (n=330)

    • 1.1% (95% CI 0.0 to 2.3), I2=0%

    Pooled rate of deep sternal wound infection/mediastinitis

    Meta-analysis random effects model, 4 studies (n=490)

    • 14.7% (95% CI 4.0 to 25.4), I2=92%

    Pooled rate of renal replacement therapy

    Meta-analysis random effects model, 19 studies (n=1,979)

    • 47.1% (95% CI 38.9 to 55.2), I2=92%

    Wang 2018

    Pooled survival to hospital discharge

    Meta-analysis random effects model, 20 studies (n=2,877)

    • 34% (95% CI 30 to 38), I2=71.8%

    Pooled 1-year survival rate

    Meta-analysis random effects model, 6 studies (n=1,860)

    • 24% (95% CI 19 to 30), I2=75.6%

    Pooled midterm survival rate (3- to 5-year)

    Meta-analysis random effects model, 4 studies (n=742)

    • 18% (95% CI 11 to 27), I2=77.3%

    Pooled rate of leg ischaemia

    Meta-analysis random effects model, 11 studies (n=945)

    • 14% (95% CI 10 to 20), I2=74.8%

    Pooled rate of reoperation for bleeding

    Meta-analysis random effects model, 10 studies (n=1,268)

    • 50% (95% CI 32 to 68), I2=96.6%

    Pooled rate of renal failure

    Meta-analysis random effects model, 12 studies (n=1,279)

    • 57% (95% CI 47 to 66), I2=87.1%

    Pooled rate of neurological complications

    Meta-analysis random effects model, 12 studies (n=1,341)

    • 16% (95% CI 13 to 20), I2=60.5%

    Pooled rate of systemic infection

    Meta-analysis random effects model, 9 studies (n=598)

    • 31% (95% CI 22 to 41), I2=78.9%

    Kowalewski 2020

    Pooled survival to hospital discharge

    Meta-analysis random effects model, 53 studies (n=4,367)

    • 35.3% (95% CI 32.5 to 38.2)

    Pooled rate of bridge to HTx

    • 3.5% (95% CI 1.8 to 6.6)

    Pooled rate of bridge to short or long term VAD

    • 4.3% (95% CI 2.8 to 6.5)

    Successful weaning from ECMO

    • 55.3% (95% CI 31.4 to 100%)

    Pooled limb complications

    Meta-analysis random effects model, 30 studies (n=2,766)

    • 13.0% (95% CI 32.5 to 38.2)

    Pooled rate of reoperations for bleeding

    Meta-analysis random effects model, 33 studies (n=2,832)

    • 41.2% (95% CI 35.6 to 47.1)

    Pooled neurological complications

    Meta-analysis random effects model, 33 studies (n=2,730)

    • 14.1% (95% CI 11.8 to 16.8)

    • Included 7.9% brain deaths: n=88

    Pooled rate of sepsis

    Meta-analysis random effects model, 29 studies (n=1,860)

    • 20.7% (95% CI 17.0 to 24.9)

    Pooled rate of acute kidney injury

    Meta-analysis random effects model, 34 studies (n=3,199)

    • 47.3% (95% CI 41.5 to 53.1)

    Alba, 2021

    Pooled short-term mortality (30 day and in-hospital)

    • Overall: 61% (95% CI 59 to 63) 306 studies n=29,289

    • ECPR OHCA: 76% (95% CI 69 to 82), I2=94%, 41 studies n=2,974

    • ECPR IHCA: 64% (95% CI 59 to 69), I2=81%, 46 studies n=2,987

    • Post AMI: 60% (95% CI 59 to 64), I I2=87%, 80 studies n=7,774

    • Postcardiotomy: 59% (95% CI 56 to 63), I2=87%, 64 studies n=8,231

    • AHF: 53% (95% CI 46 to 59), I2=89%, 33 studies n=3,567

    • Post-HTx: 35% (95% CI 29 to 42), I2=64%, 25 studies n=771

    • Myocarditis: 40% (95% CI 33 to 46), I2=65%, 13 studies n=906

    • PE: 52% (95% CI 38 to 66), I2=75%, 10 studies n=221

    Probability of HTx

    Meta-analysis

    • Post AMI: 2.8%, 95% CI 0.8 to 5.5, 19 studies

    • Postcardiotomy: 0.4%, 95% CI 0.0 to 1.1, 34 studies

    • Post-HTx: 0.0%, 95% CI 0.0 to 0.5, 5 studies

    • AHF: 13.1%, 95% CI 5.5 to 23.7, 16 studies

    • Myocarditis: 4.5%, 95% CI 0.3 to 11.7, 5 studies

    • PE: 0.0%, 95% CI 0.0 to 22.8, 1 study

    Probability of VAD

    Meta-analysis

    • Post AMI: 9.0%, 95% CI 4.2 to 15.1, 22 studies

    • Postcardiotomy: 0.8%, 95% CI 0.2 to 1.8, 35 studies

    • Post-HTx: 2.4%, 95% CI 0.0 to 6.8, 5 studies

    • AHF: 29.0%, 95% CI 17.3 to 42.1, 17 studies

    • Myocarditis: 2.3%, 95% CI 0.2 to 5.6, 5 studies

    • PE: 0.0%, 95% CI 0.0 to 22.8, 1 study

    No safety outcomes were reported

    Chen, 2017

    In-hospital mortality

    • ECMO for PCS: 61.7% (701/1,137)

    • Non-ECMO (without PCS): 6.8% (385/5,685)

    OR 22.34 (95% CI 19.06 to 26.18), p<0.001

    1-year survival (Kaplan-Meier estimate)

    • ECMO for PCS: 24.1% (95% CI 21.6 to 26.6)

    • Non-ECMO (without PCS): 83.4% (95% CI 82.4 to 84.4)

    Log rank test p<0.001

    5-year survival (Kaplan-Meier estimate)

    • ECMO for PCS: 17.7% (95% CI 14.7 to 20.7)

    • Non-ECMO (without PCS): 66.0% (95% CI 64.3 to 67.6)

    10-year survival (Kaplan-Meier estimate)

    • ECMO for PCS: 9.7% (95% CI 4.0 to 15.5)

    • Non-ECMO (without PCS): 50.2% (95% CI 46.7 to 53.7)

    Re-exploration for bleeding

    • ECMO for PCS: 11.3% (129/1,137)

    • Non-ECMO (without PCS): 2.5% (141/5,685)

    OR 5.04 (95% CI 3.93 to 6.45), p<0.001

    Massive blood transfusion (PRBC >8 Units)

    • ECMO for PCS: 79.1% (899/1,137)

    • Non-ECMO (without PCS): 15.3% (870/5,685)

    OR 21.25 (95% CI 18.09 to 24.96), p<0.001

    New onset ischaemic stroke

    • ECMO for PCS: 3.2% (36/1,137)

    • Non-ECMO (without PCS): 3.5% (201/5,685)

    OR 0.89 (95% CI 0.62 to 1.28), p=0.534

    New onset haemorrhagic stroke

    • ECMO for PCS: 1.1% (12/1,137)

    • Non-ECMO (without PCS): 0.4% (23/5,685)

    OR 2.63 (95% CI 1.30 to 5.29), p=0.007

    Acute renal failure and need for haemodialysis

    • ECMO for PCS: 32.9% (374/1,137)

    • Non-ECMO (without PCS): 7.4% (418/5,685)

    OR 6.26 (95% CI 5.34 to 7.35), p<0.001

    Postoperative infection

    • ECMO for PCS: 13.2% (150/1,137)

    • Non-ECMO (without PCS): 4.5% (256/5,685)

    OR 3.23 (95% CI 2.61 to 4.00), p<0.001

    Fasciotomy or amputation

    • ECMO for PCS: 2.3% (26/1,137)

    • Non-ECMO (without PCS): 0.8% (47/5,685)

    OR 2.81 (95% CI 1.73 to 4.56), p<0.001

    Kowalewski 2021

    Successful weaning from ECMO

    • 56.4% (4,051/7,185)

    Survival to hospital discharge

    • Overall: 41.7% (2,997/7,185)

    Mortality by primary surgery type

    • CABG: 65.4%

    • Vascular aortic: 69.6%

    • Heart transplant: 46.0%

    Limb complications: 6.3% (456/7,185)

    • Ischaemia 4.3% (312)

    • Limb compartment syndrome 1.5% (106)

    • Fasciotomy 2.0% (143)

    • Amputation 0.6% (43)

    Haematological complications: 42.5% (3,052/7,185)

    • Disseminated intravascular coagulation: 2.8% (200)

    • Haemolysis: 4.0% (290)

    • Surgical site bleed: 26.4% (1,897)

    • Cannulation site bleed: 15.7% (1,130)

    • Mediastinal cannulation bleeding: 1.4% (98)

    • Cardiac tamponade: 7.6% (547)

    • GI bleeding: 4.1% (298)

    Neurological complications: 9.1% (654/7,185)

    • Diffuse ischaemia confirmed by US/CT/MRI: 0.1% (7)

    • Haemorrhage confirmed by US/CT/MRI: 1.7% (122)

    • Infarction confirmed by US/CT/MRI: 4.5% (326)

    • Intra/extra parenchymal haemorrhage confirmed by US/CT/MRI: 0.3% (19)

    • Intraventricular haemorrhage confirmed by US/CT/MRI: 0.1% (7)

    • Neurosurgical intervention performed: 0.0% (1)

    • Seizures confirmed by EEG: 0.4% (32)

    • Seizures clinically determined: 1.1% (78)

    • Brain death: 2.5% (18)

    Sepsis: 12.1% 871/7,185

    • Culture proven infection: 10.7% (771)

    Kidney failure: 48.9% 3,510/7,185

    • Serum creatinine 1.5 to 3: 22.1% (1,591)

    • Serum creatinine >3: 10% (715)

    • Continuous renal replacement therapy: 36.1% (2,593)

    Cardiovascular complications: 54.2% 3,894/7,185

    • Cardiac arrhythmia: 15.9% (1,141)

    • CPR required >3 times: 2.9% (206)

    • Hypotension requiring vasodilators: 3.1% (222)

    • Inotropes on ECMO: 44.5% (3,196)

    Metabolic complications: 26.9% 1,934/7,185

    • Glucose <40: 1.4% (104)

    • Glucose >240: 10.5% (758)

    • Hyperbilirubinemia: 13.1% (941)

    • pH <7.2: 8.6% (620)

    • pH >7.6: 2.9% (208)

    Pulmonary complications: 3.8%271/7,185

    • Pneumothorax: 1.3% (91)

    • Pulmonary haemorrhage: 2.6% (187)

    Loungani, 2021

    Overall survival to discharge

    • 40% (290/723)

      • Survival without need for permanent cardiac support (n=235)

      • Survival with HTx (n=7)

      • Survival with LVAD (n=48)

    Survival to discharge by aetiology

    • Postcardiotomy: 36.0%

    • HTx/PGD: 57.6%

    • MI: 39.3%

    • Cardiomyopathy: 40.7%

    • Other cardiogenic shock: 42.9%

    • Non-cardiogenic shock: 36.6%

    Death during ECMO or hospitalisation by aetiology

    • Postcardiotomy: 64.0% (142)

    • HTx/PGD: 42.4% (25)

    • MI: 60.7% (74)

    • Cardiomyopathy: 59.3% (112)

    • Other cardiogenic shock: 27.1% (28)

    • Non-cardiogenic shock: 63.4% (52)

     

    Complications on ECMO (n=723)

    • Infection: 21.3% (154)

    • Acute renal dysfunction: 35.5% (257)

    • Major bleeding: 36.1% (261)

    • Clinically significant coagulopathy: 14.2% (103)

    • Disseminated intravascular coagulopathy: 2.2 (16)

    • Deep venous thrombosis: 2.6% (19)

    • Pulmonary embolism: 0.4% (3)

    • Haemothorax:3.5% (25)

    • Pneumothorax: 3.0% (22)

    • Diffuse cerebral oedema/hypoxic encephalopathy: 3.9% (28)

    • Intracranial haemorrhage/haemorrhagic stroke: 2.4% (17)

    • Ischaemic stroke/embolisation: 2.4% (17)

    • Seizures: 0.4% (3)

    • Limb ischaemia: 12.2% (88)

    • Fasciotomy: 3.5% (25)

    • Peripheral wound: 1.7% (12)

    • Hyperperfusion: 0.4% (3)

    • Air embolism: 0.1% (1)

    • Cannula dislodgement: 0.8% (6)

    • Oxygenator failure: 1.1% (8)

    • Pump malfunction: 0.8% (6)

    • Thrombosis: 1.1% (8)

    • Tubing rupture: 0.1% (1)

    Danial, 2023

    In-hospital survival

    • Postcardiotomy excluding PGF: 34.6%

    • PGF: 73.3%

    • Drug overdose: 58.6%

    • Cardiomyopathy: 53.2%

    • Arrhythmic storm: 51.6%

    • Massive PE: 46.8%

    • Sepsis induced cardiogenic shock: 44.4%

    • Fulminant myocarditis: 37.9%

    • AMI: 37.3%

    • Refractory vasoplegia shock: 11.1%

    • Other/unknown aetiology: 25.7%

    5-year survival

    • Postcardiotomy excluding PGF: 33.3%

    • PGF: 57.3%

    • Drug overdose: 54.0%

    • Arrhythmic storm: 50.0%

    • Cardiomyopathy: 45.3%

    • Sepsis induced cardiogenic shock: 42.4%

    • Massive PE: 38.3%

    • Fulminant myocarditis: 32.9%

    • AMI: 31.5%

    • Refractory vasoplegia shock: 0.0%

    • Other/unknown aetiology: 22.8%

    Complications (entire cohort) n=1,253

    • Site infection: 19% (240)

    • Limb ischaemia: 9% (118)

    • Limb amputation: 0.9% (11)

    • Vascular cannulation adverse event: 3% (34)

    • Vascular decannulation adverse event: 9% (71)

    • Sensory-motor deficit: 4% (34)

    • General bleeding: 25% (316)

    • Neurological adverse event: 16% (194)

    • Ischaemic stroke: 7% (81)

    • Intracranial bleeding: 4% (53)

    • Brain oedema: 2% (22)

    • Brain death: 9% (107)

    • Renal failure requiring haemodialysis: 52% (630)

    • Hydrostatic pulmonary oedema: 9% (11)

    Complications (postcardiotomy) n=297

    • Site infection: 13% (37)

    • Limb ischaemia: 11% (34)

    • Limb amputation: 0.3% (1)

    • Vascular cannulation adverse event: 3% (9)

    • Vascular decannulation adverse event: 9% (16)

    • Sensory-motor deficit: 3% (5)

    • General bleeding: 34% (101)

    • Neurological adverse event: 14% (41)

    • Ischaemic stroke: 6% (18)

    • Intracranial bleeding: 4% (13)

    • Brain oedema: 1% (2)

    • Brain death: 5% (16)

    • Renal failure requiring haemodialysis: 58% (170)

    • Hydrostatic pulmonary oedema: 6% (17)

    Aboud, 2024

    Reasons for ECLS termination

    • Weaning=47.2% (272/576)

    • End of unsuccessful therapy=47.6% (274/576)

    • Switch to VAD=4.7% (27/576)

    • Heart transplant=0.5% (3/576)

    In-hospital mortality=66.0% (380/576)

    Actuarial cumulative overall survival

    • 1 year=22.3% (95% CI 19.1 to 26.1)

    • 2 years=20.1% (95% CI 16.9 to 23.8)

    • 5 years=15.7% (95% CI 12.5 to 19.8)

    • 10 years=11.0% (95% CI 3.6 to 18.4)

    People who survived the in-hospital period had a 10-year survival rate of 32.4% (95% CI 12.3 to 52.5)

    Exclusion of people who had a VAD (n=27) implant or a heart transplantation (n=3) resulted in 1-year, 5-year and 10-year survival rates of 21.7% (95% CI 18.2 to 24.2), 15.4% (95% CI 12.3 to 18.5%) and 10.6% (95% CI 7.9 to 13.4%), respectively.

    Multivariable analysis suggested that severe aortic stenosis, previous cardiac surgery and IABP support were risk factors for in-hospital mortality (OR 1.71, 95% CI 1.04 to 2.83, p=0.04; OR OR 1.62, 95% CI 1.08 to 2.42, p=0.018; OR 2.46, 95% CI 1.05 to 5.87, p=0.043, respectively). Peripheral cannulation reduced the risk of mortality both in hospital and in the long follow-up.

    Advanced age, insulin dependent diabetes and severe mitral regurgitation were all strong negative predictors for long-term mortality, and peripheral cannulation was protective.

    In-hospital outcomes

    • Stroke=18.8% (108/576)

    • Re-thoracotomy for bleeding=60.2% (347/576)

    • Myocardial infarction=6.3% (36/576)

    • Laparotomy=6.8% (39/576)

    • Arterial vascular complication=18.6% (107/576)

    • Cardiac arrest=26.0% (150/576)

    • Dialysis=83.9% (483/576)

    • Low output syndrome=82.1% (473/576)

    Mariani, 2023

    In-hospital mortality=60.5%

    In-hospital survivors were discharged after a median of 38 (IQR 26 to 60) days, whereas in-hospital death happened at a median of 11 (IQR 4 to 22) days after surgery.

    Independent variables associated with in-hospital mortality were age (HR 1.02, 95% CI 1.01 to 1.02) and preoperative cardiac arrest (HR 1.41, 95% CI 1.15 to 1.73).

    Overall survival probabilities

    • 1 year=32.4% (95% CI 30.3 to 34.6)

    • 2 years=30.9% (95% CI 28.8 to 33.1)

    • 5 years=27.8% (95% CI 25.7 to 30.1)

    • 10 years=19.5% (95% CI 16.7 to 22.8)

    Survival in hospital survivors

    • 1 year=89.5% (95% CI 87.0 to 92.0)

    • 2 years=85.4% (95% CI 82.5 to 88.3)

    • 5 years=76.4% (95% CI 72.5 to 80.5)

    • 10 years=65.9% (95% CI 60.3 to 72.0)

    Variables associated with post-discharge mortality included older age, atrial fibrillation, emergency surgery, type of surgery, postoperative acute kidney injury, and postoperative septic shock.

    Postoperative outcomes (percentages were calculated excluding missing values)

    • Postoperative bleeding=57.2% (n=1,156)

      • Needing re-thoracotomy=39.7% (n=765)

      • Cannulation site bleeding=12.2% (n=246)

      • Diffuse non-surgery related=25.4% (n=472)

    • Brain oedema=4.3% (n=84)

    • Cerebral haemorrhage=3.4% (n=66)

      • Minor=43.8% (n=21)

      • Disabling=31.3% (n=15)

      • Fatal=25.0% (n=12)

    • Seizure=2.1% (n=41)

    • Stroke=10.6% (n=217)

      • Minor=46.9% (n=83)

      • Disabling=32.2% (n=57)

      • Fatal=20.9% (n=37)

    • Arrhythmia=33% (n=624)

    • Cardiac arrest=16.1% (n=304)

    • Pacemaker implantation=3% (n=56)

    • Bowel ischaemia=5.7% (n=107)

    • Right ventricular failure=21% (n=389)

    • Heart transplant=7.2% (n=111)

    • Acute kidney injury=56.7% (n=1,069)

    • Pneumonia=22.2% (n=411)

    • Septic shock=16.8% (n=310)

    • Vasoplegic syndrome=9.5% (n=176)

    • Acute respiratory distress syndrome=5.5% (n=104)

    • Multiorgan failure=34.3% (n=697)

    • Embolism=6.1% (n=113)

    Biancari, 2020

    In-hospital mortality=64.4% (503/781)

    Death on VA ECMO=46.1% (360/781)

    Survival at 1 year=32.8%

    Survival at 5 years=28.5%

    Hospital mortality after VA ECMO therapy for more than 7 days was 60.5%.

    3.6% (28/781) of people had VAD insertion or heart transplantation after postcardiotomy VA ECMO, and their hospital and 1-year mortality were 42.9% and

    53.6%, respectively.

    14 people had heart transplantation, and their hospital and 1-year mortality after heart transplantation were 21.4% and 28.6%, respectively. Among 14 people who had insertion of a VAD without heart transplantation, hospital mortality was 64.3% and 1-year mortality was 78.6%.

    28 octogenarians had hospital and on VA ECMO mortality rates of 82.1% (p=0.046) and 71.4% (p=0.009), respectively, which were statistically significantly higher than in those younger than 80.

    Logistic regression identified advanced age (crude rates: less than 60 years, 52.2%; 60 to 69 years, 64.4%; 70 years or above, 76.1%), female gender (crude rates: 69.5% versus 62.0%), stroke or unconsciousness immediately before surgery (crude rates: 88.9% versus 63.5%), prior cardiac surgery (crude rates: 71.5% versus 62.2%), aortic arch surgery (crude rates: 82.1% versus 63.5%), and arterial lactate level 6 mmol/l or greater at start of VA ECMO (crude rates: 71.6% versus 57.9%) as independent predictors of hospital death.

    Centres that had treated more than 50 people with postcardiotomy VA ECMO during the study period had a statistically significantly lower hospital mortality rate than those with lower volume of postcardiotomy VAECMO (60.9% versus 70.2%, p=0.009).

    A postcardiotomy ECMO score was derived by assigning a weighted integer to each independent pre-VA ECMO predictor of hospital mortality as follows: female gender (1 point), advanced age (60 to 69 years, 2 points; 70 years or older, 4 points), prior cardiac surgery (1 point), arterial lactate 6.0 mmol/l or greater before VA ECMO (2 points), aortic arch surgery (4 points), and preoperative stroke or unconsciousness (5 points).

    Hospital mortality rates according to the postcardiotomy ECMO score:

    • 0 point=45.6%

    • 1 point=40.5%

    • 2 points=51.1%

    • 3 points=57.8%

    • 4 points=70.7%

    • 5 points=68.3%

    • 6 points=77.5%

    7 points or more=89.7% (p<0.0001)

    • Arterial complications

    • Aortic rupture=0.3% (2/781)

    • Type A aortic dissection=1.0% (8/781)

    • Type B aortic dissection=0.4% (3/781)

    • Peripheral artery dissection=1.2% (9/781)

    • Vascular perforation=0.9% (7/781)

    • Arterial thrombosis=5.5% (43/781)

    • Major lower limb amputation=1.3% (10/781)

    • Tracheostomy=23.0% (180/781)

    • Pancreatitis=1.5% (12/781)

    • Liver failure=34.0% (265/781)

    • Gastrointestinal complication needing surgery=5.5% (42/781)

    • Multiorgan failure=49.9% (390/781)

    • Major neurological complications=18.9% (147/781)

    • Stroke, non-disabling=3.6% (28/781)

    • Stroke, disabling=7.8% (61/781)

    • Global brain ischaemia=7.4% (58/781)

    • Infectious complications

      • Deep sternal wound infection or mediastinitis=3.7% (29/781)

      • Vascular access site infection=8.6% (67/781)

      • Pneumonia=36.5% (285/781)

      • Bloodstream infection=22.9% (179/781)

    • Renal replacement therapy=53.4% (409/781)

    • Red blood transfusion (mean units)=23.4

    • Red blood transfusion of 10 or more units=70.1% (547/781)

    • Reoperation for intrathoracic bleeding=42.1% (328/781)

    • Reoperation for peripheral arterial bleeding=8.5% (66/781)

    Biancari, 2021

    Survival to hospital discharge=36.1% (240/665)

    5-year survival=27.7%

    5-year survival according to postcardiotomy ECMO score:

    • 0 point=50.9%

    • 1 point=44.9%

    • 2 points=40.0%

    • 3 points=34.7%

    • 4 points=21.0%

    • 5 points=17.6%

    • 6 points or more=10.7%, p<0.0001

    The EuroSCORE 2 was statistically significantly higher in people who had died at 5-years (p<0.0001).

    Factors independently associated with 5-year mortality were advanced age, female sex, recent myocardial infarction, active endocarditis, increased pre-VA ECMO arterial lactate level, and participating centres.

    Kaplan-Meier estimate of 5-year survival was 12.2% in people older than 70 years and 34.4% in younger people (log-rank test, p<0.0001, adjusted HR 1.840, 95% CI 1.522 to 2.224). This difference persisted among post-discharge people (52.0% versus 83.0%, log-rank test, p<0.0001, adjusted HR 3.080, 95% CI 1.686 to 5.627).

    In people who had implantation of a ventricular assist device or heart transplant (3.2%), the 5-year survival was 42.9% and 63.6%, respectively.

    In-hospital safety outcomes for the whole cohort are described in Biancari (2020).

    Rubino, 2017

    Hospital survival=33.7% (34/101)

    1-year survival=27.7% (28/101)

    2-year survival=20.8% (21/101)

    Regression analysis showed that an age older than 70 years was the strongest mortality predictor (OR 7.82; 95% CI 1.71 to 35.65; p=0.001).

    Lactate level after central VA ECMO was the strongest post-

    operative predictor for hospital mortality (OR 3.20; 95% CI 1.35 to 7.57; p=0.008). 63 people (62.4%) had lactate levels greater than 4 mmol/l during the first 48 hours of ECMO support, and only 23.8% (n=15) of those were discharged from the hospital; the survival rate was 50% (n=19) for those with lower lactate levels (n=38).

    Successful weaning from VA ECMO support=57.4% (58/101)

    9 people (8.9%) were bridged to peripheral veno-venous ECMO, 2 (2%) were bridged to peripheral VA ECMO and 2 (2%) were converted to a temporary biventricular assist device.

    Six people (5.9%) on central VA ECMO support were bridged either to a heart transplantation or long-term mechanical cardiovascular support. Three patients (3%) received a heart transplantation, and 3 (3%) had central VA ECMO converted to a long-term VAD.

    Hospital survival by surgical group, OR for in-hospital mortality

    • Pulmonary thromboendarterectomy=37.9% (OR 0.76, 95% CI 0.31 to 1.88, p=0.362; n=11)

    • Transplantation=63.6% (OR 0.19, 95% CI 0.07 to 0.52, p=0.001, n=14)

    • CABG and valvular=12.5% (OR 4.22, 95% CI 0.9 to 19.82, p=0.04, n=2)

    • CABG and other=23.1% (OR 1.81, 95% CI 0.46 to 7.07, p=0.298, n=3)

    • Valvular=16.7% (OR 2.81, 95% CI 0.57 to 13.61, p=0.158

    • Other=22.2% (OR 1.86, 95% CI 0.26 to 9.51, p=0.360, n=2)

    1-year survival by surgical group

    • Pulmonary thromboendarterectomy=34.5% (10/29)

    • Transplantation=54.5% (12/22)

    • CABG and valvular=12.5% (2/16)

    • CABG and other=15.4% (2/13)

    • Valvular=8.3% (1/12)

    • Other=11.1% (1/9)

    Complications during ECMO support

    • Bleeding=97.3% (98/101)

    • Lactate more than 4 millimoles/litre=62.5% (63/101)

    • Renal failure=64.4% (65/101)

    • Hepatic failure=27.7% (28/101)

    • Bilirubin more than 17 micromoles/litre=80.2% (81/101)

    • Rethoracotomy=25.7% (26/101)

    • Stroke=13.9% (14/101)

    • Tamponade=11.9% (12/101)

    • Limb ischaemia=5.9% (6/101)

    • Left ventricle or aortic valve thrombosis=4.9% (5/101)

    • Ventricular tachycardia or ventricular fibrillation=4.9% (5/101)

    • Mesenteric ischaemia=7.9% (8/101)

    Hepatic failure happened in 23.7% (OR 4.18; p=0.015) of hospital mortalities and renal failure was present in 49.5% of hospital mortality cases (OR 3.72; p=0.003).

    Mean units of red blood cells used=17.7

    Procedure technique

    Of the 12 studies, none detailed the ECMO device or combination of devices used. VA ECMO was inserted during the initial cardiac surgery in the cases of circulatory instability during or immediately after weaning from CPB in 2 systematic reviews (54% Biancari 2018, 43% Kowalewski 2020). One study noted that the exact timing of ECMO implantation was unavailable, so the study authors presumed that most ECMO implantation occurred after cardiac surgery (Chen 2017). Three studies noted the location of ECMO initiation as either the operating room, intensive care unit, catheterisation laboratory, emergency department, or transferred from other institutions already on ECMO support (Kowalewski 2020, Loungani 2021, Danial 2023).

    Peripheral cannulation was preferred and most common strategy for VA ECMO in the 7 studies that detailed cannulation procedure (Biancari 2018, Mariani 2023, Aboud 2024, Danial 2023, Kowalewski 2020, Kowalewski 2021, Loungani 2021). However, 39% of people included in Aboud (2024) and 46% of people included in Kowalewski (2021) were noted to be centrally cannulated. Left ventricular unloading using concomitant IABP was used in 31% (Kowalewski 2021) and 62% of people (Biancari 2018). Of the 12 studies, 7 detailed the length of time on ECMO ( Kowalewski 2020, Kowalewski 2021, Loungani 2021, Aboud 2024, Biancari 2020, Mariani 2023, Rubino 2017). The median duration of ECMO ranged from 4.9 to 7.4 days, and the mean duration of ECMO was 6.6 days in Rubino (2017) and 6.9 days in Biancari (2020).

    Efficacy

    Survival

    In-hospital survival

    Of the 12 key evidence studies, 7 reported the in-hospital survival of people having ECMO postcardiotomy.

    In meta-analyses from 3 systematic reviews, pooled in-hospital survival ranged from 34 to 36% (Biancari 2018, Wang 2018, Kowalewski 2020). Meta-regression analysis by Biancari (2018) showed a trend toward lower hospital survival in studies with higher mean age (p=0.064). The pooled analysis of 12 studies showed that hospital survivors (n=387) were significantly younger than people who died after VA ECMO (pooled mean age, 56 versus 64 years; mean difference, -7.223 years, 95% CI -9.777 to - 4.669, I2=53%, p=0.015)

    In the registry study of 7,185 people with refractory PCS, in-hospital survival was 42% (Kowalewski 2021). In the registry study of 723 adults treated with VA ECMO (31% postcardiotomy), the survival in the overall population was 40% and 36% in postcardiotomy patients (Loungani 2021).

    In the single centre retrospective study of people treated with VA ECMO for cardiogenic shock, among those with PCS (n=297), in-hospital survival was 35% (Danial 2023). Hospital survival was 34% in the single-centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery (Rubino 2017). Regression analysis showed that an age older than 70 years was the strongest hospital mortality predictor (OR 7.82; 95% CI 1.71 to 35.65; p=0.001). Lactate level after central VA ECMO was the strongest post-operative predictor for hospital mortality (OR 3.20; 95% CI 1.35 to 7.57; p=0.008).

    In the multicentre retrospective registry study of 665 people having VA ECMO for postcardiotomy cardiogenic shock, survival to hospital discharge was 36% (Biancari 2021).

    1-year survival

    Of the 12 key evidence studies, 7 reported the 1-year survival of people having ECMO postcardiotomy.

    In the systematic review of 31 studies of people who required VA ECMO following cardiac surgery, the pooled 1-year survival in a meta-analysis was 31% (95% CI 24.3 to 37.5), I2=82% (11 studies [n=1,290]; Biancari 2018). In the systematic review of 20 studies of people having ECMO for PCS following cardiac surgery, the pooled 1-year survival in a meta-analysis was 24% (95% CI 19 to 30), I2=76% (6 studies [n=1,860]; Wang 2018).

    In the registry study of 576 people having ECMO for PCS, the actuarial cumulative 1-year survival was 22% (95% CI 19.1 to 26.1). Exclusion of people who had a VAD implant or a heart transplantation resulted in a 1-year survival rate of 22% (95% CI 18.2 to 24.2; Aboud 2024). In the multicentre retrospective registry study of 781 people having postcardiotomy VA ECMO for refractory cardiopulmonary failure, the survival rate at 1 year was 33% (Biancari 2020).

    In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, the 1-year overall survival probability was 32% (95% CI 30.3 to 34.6). In the subgroup of hospital survivors, the 1-year survival rate was 90% (95% CI 87.0 to 92.0; Mariani 2023).

    In the propensity score-matched study of people admitted for cardiac surgery who had VA ECMO (n=1,137), the cumulative 1-year survival using Kaplan-Meier estimate was 24% (95% CI 21.6 to 26.6) in those who had ECMO for PCS (Chen 2017).

    In the single-centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, the 1-year survival rate was 28% (Rubino 2017).

    Mid-term survival

    Of the 12 key evidence studies, 6 studies reported mid-term survival rates in people having ECMO postcardiotomy, with a 5-year survival available in 5 studies. .

    In the systematic review of 20 studies of people with PCS following cardiac surgery, the pooled 3- to 5- year survival in a meta-analysis was 18% (95% CI 11 to 27), I2=77% (4 studies [n=742]; Wang 2018). In the single centre retrospective study of 1,253 people treated with VA ECMO for cardiogenic shock (297 with PCS), the 5-year survival for people postcardiotomy (excluding PGF) was 33% (Danial 2023). The authors also reported that the 5-year survival rate was 57% for PGF.

    In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, the overall survival probability was 31% (95% CI 28.8 to 33.1) at 2 years and was 28% (95% CI 25.7 to 30.1) at 5 years (Mariani 2023). In the subgroup of hospital survivors, the 2-year survival rate was 85% (94% CI 82.5 to 88.3) and the 5-year survival rate was 76% (95% CI 72.5 to 80.5).

    In the propensity score-matched study of people admitted for cardiac surgery who had VA ECMO (n=1,137), the cumulative 5-year survival using Kaplan-Meier estimate was 18% (95% CI 14.7 to 20.7) (Chen 2017). The authors note that although the risk of all-cause mortality was greater in the group receiving ECMO for PCS than in the group without PCS (non-ECMO) (p<0.001) in the first year of follow-up, no difference was observed after the first year of follow-up (p=0.209; Chen 2017).

    In the registry study of 576 people having ECMO for PCS, the actuarial cumulative 2- and 5-year survival rates were 20% (95% CI 16.9 to 23.8) and 16% (95% CI 12.5 to 19.8), respectively. Exclusion of people who had a VAD implant or a heart transplantation resulted in a 5-year survival rate of 15% (95% CI 12.3 to 18.5; Aboud 2024).

    In the multicentre retrospective registry study of 781 people having VA ECMO for refractory cardiopulmonary failure following cardiac surgery, the survival rate at 5 year was 29% (Biancari 2020). In the Biancari (2021) study of 665 people having VA ECMO for PCS, the 5-year survival rate was 28%. The EuroSCORE 2 was statistically significantly higher in people who had died at 5 years (p<0.0001). Factors independently associated with 5-year mortality were advanced age, female sex, recent myocardial infarction, active endocarditis, increased pre-VA ECMO arterial lactate level, and participating centres (Biancari 2021).

    In the single-centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, the 2-year survival rate was 21% (Rubino 2017).

    Long-term survival

    Of the 12 key evidence studies, 3 reported the 10-year survival of people having ECMO postcardiotomy. In the propensity score-matched study of 6,822 people admitted for cardiac surgery with (n=1,137) or without (n=5,685) VA ECMO, the cumulative 10-year survival using Kaplan-Meier estimate was 10% (95% CI 4.0 to 15.5) in those who had ECMO for PCS following cardiac surgery compared to 50% (95% CI 46.7 to 53.7) in those who did not (Chen 2017). Again, the authors note that although the risk of all-cause mortality was greater in the ECMO for PCS group than in the non-PCS group (p<0.001) in the first year of follow-up, no difference was observed after the first year of follow-up (p=0.209; Chen 2017).

    In the registry study of 576 people having ECMO after cardiac surgery, the actuarial cumulative 10-year survival was 11% (95% CI 3.6 to 18.4; Aboud 2024). People who survived the in-hospital period had a 10-year survival rate of 32% (95% CI 12.3 to 52.5). Exclusion of people who had a VAD implant or a heart transplantation resulted in a 10-year survival rate of 11% (95% CI 7.9 to 13.4).

    In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, the 10-year overall survival probability was 20% (95% CI 16.7 to 22.8; Mariani 2023). In the subgroup of hospital survivors, the 10-year survival rate was 66% (94% CI 60.3 to 72.0).

    Successful weaning from ECMO

    Of the 12 key evidence studies, 5 reported the proportion of people successfully weaned from ECMO postcardiotomy. In the systematic review of 31 studies of people who required VA ECMO following cardiac surgery, the pooled proportion successfully weaned in a meta-analysis was 60% (95% CI 54.6 to 64.3), I2=77% (24 studies [n=2,049]; Biancari 2018). In the systematic review of 54 studies reporting on 4,421 people with refractory PCS, 55% (31 to 100%) of people were successful weaned from ECMO (Kowalewski 2020).

    In the registry study of 7,185 people with refractory PCS, 56% were successfully weaned from ECMO (Kowalewski 2021). In the registry study of 576 people with PCS, the proportion of people who successfully weaned from ECMO was 47% (Aboud 2024).

    In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, 57% (n=58) were successfully weaned from VA ECMO (Rubino 2017). Of the 58 people, 9 people were bridged to peripheral veno-venous ECMO, 2 were bridged to peripheral VA ECMO and 2 were converted to a temporary biventricular assist device.

    Bridged to heart transplant

    Of the 12 key evidence studies, 5 studies (6 publications) reported the proportion of people bridged to heart transplant following ECMO. The pooled rate of heart transplantation post-ECMO from a meta-analysis of 21 studies (n=1,685) was 2% (95% CI 1.0 to 2.8, I2=50%) in the systematic review of people who required VA ECMO following cardiac surgery (Biancari 2018). Of these heart transplant recipients, 66% (95% CI 48.2 to 84.1, I2=0%) survived until hospital discharge (Biancari 2018). In the systematic review of 54 studies reporting on 4,421 people with refractory PCS, the pooled rate of heart transplantation was 3.5% (95% CI 1.8 to 6.6) (Kowalewski 2020). The pooled rate of heart transplantation in those with PCS was 0.4% (95% CI 0.0 to 1.1) in a meta-analysis of 34 studies in the systematic review by Alba et al. (2021).

    In the registry study of 576 people with PCS, less than 1% (n=3) of people bridged to heart transplant following ECMO (Aboud 2024). In the registry study of 781 people having postcardiotomy VA ECMO for refractory cardiopulmonary failure, 2% (n=14) of people had heart transplantation, and their hospital and 1-year mortality after heart transplantation were 21% and 29%, respectively (Biancari 2020). In the Biancari (2021) study of 665 people having VA ECMO for PCS, 2% (n=11) of people had heart transplantation.

    In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, 3% (n=3) received a heart transplantation (Rubino 2017).

    Bridged to long term VAD

    Of the 12 key evidence studies, 5 studies (6 publications) reported the proportion of people bridged to a ventricular assist device (VAD) using ECMO. The pooled rate of VAD implantation post-ECMO from a meta-analysis of 21 studies (n=1,685) was 2% (95% CI 1.3 to 3.4, I2=57%) in the systematic review of people who required VA ECMO following cardiac surgery (Biancari 2018). Of these VAD recipients, 46% (95% CI 28.0 to 63.1, I2=43%) survived until hospital discharge (Biancari 2018). In the systematic review of 54 studies reporting on 4,421 people with refractory PCS, the pooled rate of heart transplantation was 4.3% (95% CI 2.8 to 6.5) (Kowalewski 2020). The pooled rate of heart transplantation in those with PCS was 0.8% (95% CI 0.2 to 1.8) in a meta-analysis of 35 studies in the systematic review by Alba et al. (2021).

    In the registry study of 576 people with PCS, 5% (27) of people switched to VAD following ECMO (Aboud 2024). In the registry study of 781 people having VA ECMO for refractory cardiopulmonary failure following cardiac surgery, 3% (n=21) of people had insertion of a VAD (Biancari 2020). Of these people, 14 people inserted a VAD without heart transplantation, and their hospital mortality was 64% and 1-year mortality was 79%. In the Biancari (2021) study of 665 people having VA ECMO for postcardiotomy cardiogenic shock, 21 people had implantation of a VAD or heart transplant, and 5-year survival was 43% of people who had a VAD device implanted.

    In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, 3% (n=3) were converted to a long-term VAD (Rubino 2017).

    Mortality

    Of the 12 key evidence studies, 7 reported on mortality. In the registry study of 7,185 people with refractory PCS, in-hospital mortality by primary surgery type was 65% for CABG, 70% for vascular aortic surgery and 46% for heart transplant surgery (Kowalewski 2021). Older age was significantly associated with in-hospital mortality. Of the patients aged over 70 years, 70% did not survive to discharge compared to 55% those younger than 70 years (p<0.001).

    In the systematic review of 306 studies of CS of any aetiology, the pooled overall short-term mortality (30-day and in-hospital) for those with PCS was 59% (95% CI 56 to 63, I2=87%, 64 studies). Univariate meta regression analysis stratified by aetiology also showed an 8% increase in mortality per 10-year increase in cohort's age (Alba 2021).

    In the registry study of 723 adults treated with VA ECMO (31% postcardiotomy), 64% postcardiotomy patients died during ECMO or hospitalisation (Loungani 2021). Multivariable regression analysis identified older age as a risk factor for mortality on ECMO (OR 1.26; 95% CI 1.12 to 1.42, p<0.001). Mortality rates while on ECMO support increased from 26% in those aged 35 to 44 years to 54% in those 75 years or older (Loungani 2021).

    In the propensity score-matched study of 6,822 people admitted for cardiac surgery with (n=1,137) or without (n=5,685) VA ECMO, in-hospital mortality was 62% in those who had ECMO for PCS following cardiac surgery compared to 7% in those who did not have PCS or ECMO (OR 22.34, 95% CI 19.06 to 26.18, p<0.001; Chen 2017).

    In the registry study of 576 people having ECMO for PCS, in-hospital mortality was 66% (Aboud 2024). Multivariable analysis suggested that severe aortic stenosis (OR 1.71, 95% CI 1.04 to 2.83, p=0.04), previous cardiac surgery (OR 1.62, 95% CI 1.08 to 2.42, p=0.018) and IABP support (OR 2.46, 95% CI 1.05 to 5.87, p=0.043) were risk factors for in-hospital mortality. Peripheral cannulation reduced the risk of mortality both in hospital and in the long follow up.

    In the multicentre retrospective registry study of 781 people having VA ECMO for refractory cardiopulmonary failure following cardiac surgery, in-hospital mortality was 64% and mortality on VA ECMA was 46% (Biancari 2020). When the duration of VA ECMO therapy was greater than 7 days, the crude hospital mortality was 61% compared with 66% in people in whom VA ECMO lasted 7 days or less (p=0.10; Biancari 2020). Centres that had treated more than 50 people with postcardiotomy VA ECMO during the study period had a statistically significantly lower hospital mortality rate than those with lower volume of postcardiotomy VA ECMO (61% versus 70%, p=0.009). Logistic regression identified advanced age, female gender, stroke or unconsciousness immediately before surgery, prior cardiac surgery, aortic arch surgery, and arterial lactate level 6 mmol/l or greater at start of VA ECMO as independent predictors of hospital death.

    In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, in-hospital mortality was 61% (Mariani 2023). Independent variables associated with in-hospital mortality were age (HR 1.02, 95% CI 1.01 to 1.02) and preoperative cardiac arrest (HR 1.41, 95% CI 1.15 to 1.73). Variables associated with post-discharge mortality included older age, atrial fibrillation, emergency surgery, type of surgery, postoperative acute kidney injury, and postoperative septic shock.

    Safety

    Bleeding

    Of the 12 key evidence studies, 10 reported bleeding adverse events or complications. In the propensity score-matched study of 6,822 people admitted for cardiac surgery with (n=1,137) or without (n=5,685) VA ECMO, re-exploration for bleeding was statistically significantly higher in those on ECMO for PCS (11.3% [129 of 1,137]) compared to those who did not have PCS or ECMO (2.5% [141 of 5,685], OR 5.04, 95% CI 3.93 to 6.45, p<0.001; Chen 2017). Massive blood transfusion (PRBC more than 8 Units) was also statistically significantly higher in those on ECMO for PCS (79% [899 of 1,137]) compared to those who did not have PCS or ECMO (15% [870 of 5,685], OR 21.25, 95% CI 18.09 to 24.96, p<0.001; Chen 2017).

    In the systematic review of 31 studies of people who required VA ECMO following cardiac surgery, the pooled rate of reoperation for bleeding was 43% (95% CI 34.2 to 51.5, I2=93%) in the meta-analysis of 18 studies (n=1,779; Biancari 2018). In the systematic review of 20 studies of people with PCS following cardiac surgery, the pooled rate of reoperation for bleeding was 50% (95% CI 32 to 68, I2=97%, 10 studies, n=1,268; Wang 2018). The pooled rate of reoperations for bleeding was 41% (95% CI 35.6 to 47.1) in the meta-analysis of 33 studies (n=2,832) from the systematic review of people with refractory PCS (Kowalewski 2020).

    In the registry study of people with refractory PCS, haematological complications were reported in 43% of people (3,052 of 7,185), including surgical site bleed 26% (1,897), cannulation site bleed 16% (1,130), mediastinal cannulation bleeding 1% (98), cardiac tamponade 8% (547), GI bleeding 4% (298), and haemolysis 4% (290) (Kowalewski 2021). In the registry study of adults treated with VA ECMO, major bleeding was reported in 36% (261 of 723), clinically significant coagulopathy in 14% (103 of 723), and disseminated intravascular coagulopathy in 2% (16 of 723) of the overall population (Loungani 2021).

    In the registry study of 576 people having ECMO for PCS, the rate of re-thoracotomy for bleeding was 60% (347 of 576; Aboud 2024). In the multicentre registry study of 781 people, red blood transfusion of 10 or more units was reported in 70% (n=547) of people (Biancari 2020). The rates of reoperation for intrathoracic bleeding and reoperation for peripheral arterial bleeding were 42% (n=328) and 9% (n=66), respectively.

    In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, postoperative bleeding was reported in 57% (n=1,156) of people, including bleeding needing re-thoracotomy (40%, n=765), cannulation site bleeding (12%, n=246) and diffuse non-surgery related bleeding (25%, n=472) (Mariani 2023).In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, bleeding was reported in 98% (n=98) of people and re-thoracotomy in 26% (n=26) (Rubino 2017).

    Neurological events

    Of the 12 key evidence studies, 8 reported neurological adverse events or complications. The pooled neurological complication rates reported in 3 systematic reviews, were 11% (95% CI 7.8 to 14.8, I2=79%, 16 studies [n=1,736]; Biancari 2018), 16% (95% CI 13 to 20, I2=61%, 12 studies [n=1,341]; Wang 2018) and 14% (95% CI 11.8 to 16.8, 33 studies [n=2,730]; Kowalewski 2020).

    In the registry study of people with refractory PCS, the rate of neurological complications was 9% (654 of 7,185). This included clinically determined seizures 1% (78), brain death 3% (18), haemorrhage confirmed by US/CT/MRI 2% (122), and infarction confirmed by US/CT/MRI 5% (326; Kowalewski 2021). In the registry study of adults treated with VA ECMO, diffuse cerebral oedema or hypoxic encephalopathy occurred in 4% (28 of 723) of the overall population (Loungani 2021).

    In the single centre retrospective study of people with PCS, rates of neurological adverse events were 14% (41 of 297). This included sensory-motor deficit 3% (5), intracranial bleeding 4% (13), brain oedema 1% (2) and brain death 5% (16) (Danial 2023). In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, brain oedema was reported in 4% (n=84) of people, cerebral haemorrhage in 3% (n=66), seizure in 2% (n=41) and stroke in 11% (n=217) (Mariani 2023).

    In the multicentre registry study of 781 people, major neurological complications were reported in 19% (n=147) of people. These complications included non-disabling stroke (4%, n=28), disabling stroke (8%, n=61) and global brain ischaemia (7%, n=58; Biancari 2020).

    Limb complications

    Of the 12 key evidence studies, 10 reported limb adverse events or complications. The pooled limb complication rates reported in 3 systematic reviews, were 11% (95% CI 8.0 to 13.5, I2=70%, 16 studies [n=1,909]; Biancari 2018), 14% (95% CI 10 to 20, I2=75%, 11 studies [n=945]; Wang 2018) and 13% (95% CI 32.5 to 38.2, 30 studies [n=2,766]; Kowalewski 2020).

    In the registry study of people with refractory PCS, rates of limb complications were 6% (456 of 7,185), including ischaemia 4% (312) and limb compartment syndrome 2% (106) (Kowalewski 2021). In the registry study of adults treated with VA ECMO 12% (88 of 723) of the overall population were reported with limb ischaemia (Loungani 2021).

    In the single centre retrospective study of people with PCS, rates of limb ischaemia were 11% (34 of 297; Danial 2023). This was 7% (11 of 169) in the multicentre retrospective study of adults having ECMO following cardiac surgery (Bonacci 2020).

    Rates of limb fasciotomy were 2% (143 of 7,185) in the registry study of people with refractory PCS, and 4% (25 of 723) in the overall population in the registry study of adults treated with VA ECMO (Loungani 2021), and 3% (4 of 169) in the single centre retrospective study of people undergoing CABG who had PCS (Bonacci 2020).

    In the propensity score-matched study of 6,822 people admitted for cardiac surgery with or without VA ECMO, statistically significantly more people were reported with limb fasciotomy or amputation on ECMO for PCS 2% (26 of 1,137), than those not on ECMO without PCS 1% (47 of 5,685), OR 2.81 (95% CI 1.73 to 4.56, p<0.001; Chen 2017). The pooled rate of lower limb amputation was 1% (95% CI 0.0 to 2.3, I2=0% in a meta-analysis of 5 studies (n=330) in the systematic review of people who required VA ECMO following cardiac surgery (Biancari 2018), and the registry study of people with refractory PCS by Kowalewski et al. (2021) also reported amputation rates of 1% (43 of 7,185).

    In the multicentre registry study of 781 people, major lower limb amputation was reported in 1% (n=10) of people (Biancari 2020).

    Limb ischaemia was reported in 6% (6 of 101) of people in the single centre retrospective cohort study of people having central VA ECMO after cardiac surgery (Rubino 2017).

    Infection and sepsis

    Of the 12 key evidence studies, 9 reported infection or sepsis events or complications. In the propensity score-matched study of 6,822 people admitted for cardiac surgery with or without VA ECMO, significantly more people were reported with post-operative infection on ECMO for PCS 13% (150 of 1,137), than those not on ECMO without PCS 5% (256 of 5,685), OR 3.23 (95% CI 2.61 to 4.00, p<0.001; Chen 2017). In the systematic review of people who required VA ECMO following cardiac surgery, the rate of deep sternal wound infection or mediastinitis was 15% (95% CI 4.0 to 25.4, I2=92%) in a meta-analysis of 4 studies (n=490; Biancari 2018). Pooled systemic infection rates were 31% (95% CI 22 to 41, I2=79%) in the systematic review of people with PCS following cardiac surgery (9 studies [n=598]; Wang 2018). In the registry study of adults treated with VA ECMO (31% postcardiotomy), infection rates were 21% (154 of 723) (Loungani 2021). Site infection occurred in 13% (37 of 297) of people with PCS in the French single centre retrospective study (Danial, 2023).

    In the systematic review of people with refractory PCS, pooled rates of sepsis were 21% (95% CI 17.0 to 24.9) in a meta-analysis of 29 studies (n=1,860; Kowalewski 2020). The rate of sepsis was reported as 12% (871 of 7,185) in the registry study of people with refractory PCS (Kowalewski 2021). Septic shock was reported in 17% (310 of 2,058) of people in the retrospective multicentre observational study (Mariani 2023)..

    In the multicentre registry study of 781 people, infectious complications included deep sternal wound infection or mediastinitis (4%, n=29), vascular access site infection (9%, n=67), and bloodstream infection (23%, n=179; Biancari 2020).

    Renal complications

    Of the 12 key evidence studies, 10 reported renal adverse events or complications. In the propensity score-matched study of 6,822 people admitted for cardiac surgery with or without VA ECMO, statistically significantly more people were reported with acute renal failure and need for haemodialysis on ECMO for PCS 33% (374 of 1,137), than those not on ECMO 7% (418 of 5,685), OR 6.26 (95% CI 5.34 to 7.35, p<0.001; Chen 2017).

    The pooled rates of RRT, renal failure, or acute kidney injury were reported in 3 systematic reviews. Rates were 47% (95% CI 38.9 to 55.2, I2=92%, 19 studies [n=1,979]; Biancari 2018), 57% (95% CI 47 to 66, I2=87%, 12 studies [n=1,279]; Wang 2018) and 47% (95% CI 41.5 to 53.1, 34 studies [n=3,199]; Kowalewski 2020), respectively.

    In the registry study of people with refractory PCS, rates of kidney failure were 49% (3,510 of 7,185), and rates of RRT were 36% (2,593 of 7,185; Kowalewski 2021). Acute renal dysfunction was reported as 36% (257 of 723) in the registry study of adults treated with VA ECMO (31% postcardiotomy) (Loungani 2021).

    Renal failure requiring haemodialysis was reported in 58% (170 of 297) in the single centre retrospective studies of people with PCS in France (Danial 2023).. The rate of in-hospital dialysis was 84% (483 of 576) in the registry study of 576 people having ECMO for PCS (Aboud 2024). The rate of renal failure was 64% in the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery (Rubino 2017). RRT was needed in 53% (409 of 781) of people in the multicentre registry study (Biancari 2020).

    Acute kidney injury occurred at a rate of 57% (1,069 of 2,058) in the retrospective international multicentre observation PELS-1 study of people having postcardiotomy VA ECMO (Mariani 2023).

    Stroke

    Of the 12 key evidence studies, 4 reported stroke events. In the propensity score-matched study of 6,822 people admitted for cardiac surgery with or without VA ECMO, rates of new onset ischaemic stroke were 3% (36 of 1,137) for those on ECMO for PCS, compared to 4% (201 of 5,685) in those not on ECMO (OR 0.89, 95% CI 0.62 to 1.28, p=0.534; Chen 2017). Rates of new onset haemorrhagic stroke for those on ECMO for PCS were 1% (12 of 1,137), compared to less than 1% (23 of 5,685) in those not on ECMO without PCS (OR 2.63, 95% CI 1.30 to 5.29, p=0.007; Chen 2017).

    Intracranial haemorrhage or haemorrhagic stroke and ischaemic stroke or embolisation were both reported as 3% (17 of 723) in the registry study of adults treated with VA ECMO (31% postcardiotomy) (Loungani 2021).

    Ischaemic stroke was reported in 6% (18 of 297) in the single centre retrospective studies of people with PCS in France (Danial 2023). Stroke occurred at a rate of 14% (14 of 101) in the single centre retrospective cohort study of people having central VA ECMO after cardiac surgery (Rubino 2017). The rate of in-hospital stroke was 19% (n=108) in the registry study of 576 people having ECMO for PCS (Aboud 2024).

    Cardiovascular complications

    Of the 12 key evidence studies, 5 reported cardiovascular adverse events or complications. In 1 registry study, cardiovascular complications occurred in 54% (3,894 of 7,185) of people with refractory PCS (Kowalewski 2021). These included cardiac arrhythmia 16% (1,141), CPR required more than 3 times 3% (206), hypotension requiring vasodilators 3% (222), and inotropes on ECMO 45% (3,196) (Kowalewski 2021). In the retrospective international multicentre observation PELS-1 study of 2,058 people having postcardiotomy VA ECMO, arrhythmia was reported in 33% (n=624) of people, cardiac arrest in 16% (n=304), pacemaker implantation in 3% (n=56), right ventricular failure in 21% (n=389), vasoplegic syndrome in 10% (n=176) and embolism in 6% (n=113) (Mariani 2023)..

    In the registry study of 576 people having ECMO for PCS, the rates of in-hospital cardiac arrest, myocardial infarction, low output syndrome and arterial vascular complication were 26% (n=150), 6% (n=36), 82% (n=473) and 19% (n=107), respectively (Aboud 2024).

    The multicentre registry study of 781 people reported arterial complications, including aortic rupture (0.3%, n=2), type A aortic dissection (1%, n=8), type B aortic dissection (0.4%, n=3), peripheral artery dissection (1%, n=9), vascular perforation (1%, n=7), and arterial thrombosis (6%, n=43) (Biancari 2020).

    In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, left ventricle or aortic valve thrombosis was reported in 5% (n=5) of people, and ventricular tachycardia or ventricular fibrillation in 5% (n=5) (Rubino 2017).

    Metabolic complications

    Of the 12 key evidence studies, 1 registry study reported metabolic adverse events or complications in 27% (1,934 of 7,185) of people with refractory PCS. These included glucose levels below 40 (1%, n=104), glucose levels greater than 240 (11%, n=758), hyperbilirubinemia (13%, n=941), pH lower than 7.2 (9%, n=620), and pH higher than 7.6 (3%, n=208) (Kowalewski 2021).

    Pulmonary complications

    Of the 12 key evidence studies, 5 reported pulmonary adverse events or complications. In one registry study, pulmonary complications occurred in 4% (271 of 7,185) of people with refractory PCS, including pneumothorax 1% (91), and pulmonary haemorrhage 3% (187) (Kowalewski 2021). In other registry study of 723 adults treated with VA ECMO (31% postcardiotomy), complications included pulmonary embolism less than 1% (3), haemothorax 4% (25), and pneumothorax 3% (22) (Loungani 2021).

    Hydrostatic pulmonary oedema was reported in 6% (17 of 297) of people with PCS, in the single centre retrospective study done in France (Danial 2023). Pneumonia was reported in 22% (411 of 2,058) of people and acute respiratory distress syndrome in 6% (104 of 2,058) in the retrospective multicentre observational study (Mariani 2023). Tracheostomy was reported in 23% of people (180 of 781) and pneumonia in 37% (285 of 781) in the multicentre registry study (Biancari 2020).

    GI complications

    Of the 12 key evidence studies, 2 reported GI complications. One multicentre registry study reported GI complication needing surgery in 6% (42 of 781) of people (Biancari 2020). In the single centre retrospective cohort study, mesenteric ischaemia was reported in 8% (8 of 101) of people having central VA ECMO after cardiac surgery (Rubino 2017).

    Hepatic complications

    Of the 12 key evidence studies, 2 studies reported hepatic complications. In the multicentre registry study of 781 people, the rates of pancreatitis and liver failure were 2% (n=12) and 34% (n=265), respectively (Biancari 2020). In the single centre retrospective cohort study of 101 people having central VA ECMO after cardiac surgery, hepatic failure was reported in 28% (n=28) of people (Rubino 2017).

    Multiorgan failure

    Of the 13 key evidence studies, 2 studies reported multiorgan failure. The rate of multiorgan failure was 50% (n=380) in the multicentre registry study of 781 people having VA ECMO for refractory cardiopulmonary failure following cardiac surgery (Biancari 2020) and was 34% (n=697) in the retrospective multicentre observational study of 2,058 people having postcardiotomy VA ECMO (Mariani 2023).

    Technical complications

    Of the 12 key evidence studies, 2 reported technical adverse events or complications. In 1 registry study of adults treated with VA ECMO (31% postcardiotomy), oxygenator failure rates were 1% (8 of 723), and air embolism, cannula dislodgement, pump malfunction and tubing rupture were reported in less than 1% of the overall population (Loungani 2021). In the single centre retrospective study of 297 people with PCS, vascular cannulation and decannulation adverse event rates were 3% (9) and 9% (16), respectively (Danial 2023).

    Anecdotal and theoretical adverse events

    Expert advice was sought from consultants who have been nominated or ratified by their professional society or royal college. They were asked if they knew of any other adverse events for this procedure that they had heard about (anecdotal), which were not reported in the literature. They were also asked if they thought there were other adverse events that might possibly occur, even if they had never happened (theoretical).

    They listed the following anecdotal and theoretical adverse events:

    • Left ventricle overloading

    • Deep vein thrombosis

    • Arteriovenous fistula

    • Pseudoaneurysm

    • Harlequin syndrome 

    • Haemolysis

    • Intra-cerebral haemorrhage

    • Major pulmonary bleed

    • Failure to cannulate during cardiac arrest

    • Malposition of the cannula

    • Device clotting

    • Differential oxygenation

    • Lower body hyperoaxemia/hypocapnia

    • Air entrapment

    • Oxygenator failure

    • Consumption coagulopathy

    • Acquired Von Willebrand syndrome

    • Systemic inflammatory response syndrome (SIRS).

    Sixteen professional expert questionnaires were submitted. Find full details of what the professional experts said about the procedure in the https://www.nice.org.uk/guidance/indevelopment/gid-ipg10432/documents.

    Validity and generalisability

    • Most studies included in the key evidence had a large number of participants from a variety of countries, although only 1 UK-specific study (Rubino 2017) was included.

    • Due to the nature of the procedure, randomised controlled trials in the postcardiotomy population are not possible. There was therefore a lack of comparative studies included in the key evidence. Chen et al. (2017) was the only comparative study. This study compares those who had ECMO for PCS following cardiac surgery, to a propensity matched sample with the same cardiac surgery who did not have PCS or ECMO. This comparison is a clinically lower risk group compared to those who had ECMO.

    • Some studies did not include definitions of PCS or qualifying clinical reasons for requiring ECMO postcardiotomy.

    • The studies included people with a mix of primary surgery types and only 1 study (Rubino 2017) stratified outcomes by cardiac surgery type.

    • Many studies lacked pre-, intra-, and postoperative information including differences between institutions in terms of patient selection, volume and expertise, treatment strategy as well as availability of ventricular assist devices and heart transplantation, which may impact outcomes.

    • Follow-up for most studies was short, reporting key efficacy outcomes at hospital discharge. Five-year survival was reported in 5 studies and 10-year survival in 3 studies.