Landmark Trial Database

Coronary angiography delayed until after neurological recovery

22

Open-label; conducted in the Netherlands; the delayed arm still received angiography, so this is a trial of timing and not of angiography itself.

Survival at 90 days

Immediate coronary angiography

In a resuscitated patient without ST elevation, immediacy of angiography buys nothing — and it costs time to temperature control. Chapter 22 uses the second half of that finding: the trip to the catheterisation laboratory has an opportunity cost measured in the other things that then do not happen on time.

Adults successfully resuscitated after out-of-hospital cardiac arrest WITHOUT signs of ST-segment elevation myocardial infarction

Anything about ECPR. Every participant had a circulation, which is the one feature that changes the argument: in ECPR the pump has already restored coronary perfusion, so the case for angiography is diagnostic rather than a reperfusion-time case. COACT also excluded ST elevation, whereas the only mortality signal in the ECPR angiography literature is in ST-elevation patients from whom angiography was withheld.

176 of 273 (64.5%) alive in the immediate group versus 178 of 265 (67.2%) in the delayed group; OR 0.89 (0.62 to 1.27), P = 0.51. No significant difference in any secondary endpoint. Median time to target temperature was LONGER with immediate angiography, 5.4 versus 4.7 hours, ratio of geometric means 1.19 (1.04 to 1.36).

Targeted normothermia, 37 degrees

22

Open-label for treating clinicians; the primary endpoint interval barely excludes zero; mortality unchanged; conducted in 25 French ICUs and may not transfer.

Survival with a favourable neurological outcome (Cerebral Performance Category 1 or 2) at day 90, blinded assessment

Moderate therapeutic hypothermia, 33 degrees for the first 24 hours

The one modern temperature trial that was positive, and it studied the deepest cerebral insult in the randomised literature. It is the empirical anchor for the insult-severity argument used in Chapter 22 Controversy 3: temperature management may work where the injury is large enough for it to act on, and the neutral modern trials were conducted in populations with bystander CPR rates near 80% and arrests of about a minute before compressions began.

Comatose adults admitted to intensive care after resuscitation from cardiac arrest with a NON-SHOCKABLE initial rhythm (asystole or pulseless electrical activity)

Nothing about ECPR. These patients had a return of spontaneous circulation and no extracorporeal circuit, so their temperature was not being set by a heat exchanger. The favourable-outcome difference is 4.5 points with a lower bound of 0.1, and mortality was unchanged — so the result is fragile and is about disability among survivors rather than about survival.

29 of 284 (10.2%) in the hypothermia group versus 17 of 297 (5.7%) in the normothermia group. Absolute difference 4.5 percentage points (0.1 to 8.9), P = 0.04. Mortality at 90 days did not differ (81.3% versus 83.2%). Prespecified adverse events did not differ.

Femoral arterial access without routine ultrasound

21

Underpowered for the composite; open-label; operator ultrasound experience varied; the population is the opposite of the ECPR population in the one respect that matters most. [VERIFICATION REQUIRED] — the enrolled sample size is not recorded here because it was not confirmed from a retrieved source.

Composite of major bleeding or vascular complication at 30 days

Routine ultrasound guidance for femoral arterial access

Ultrasound reliably improves the mechanics of femoral arterial access in a randomised design. The mechanical gain is real and reproducible.

Adults undergoing coronary angiography or intervention via femoral arterial access, all with a spontaneous circulation

Whether that mechanical gain converts into an outcome gain. It does not in an elective catheter laboratory, where a failed pass costs seconds. This chapter argues it does in ECPR, where a failed pass costs minutes of no-flow, but that argument is reasoning, not a finding of this trial. It also does not speak to an arresting patient at all: every participant had a circulation, so pulsatility and blood colour both still identified the vessel.

No difference on the primary composite. Every mechanical endpoint favoured ultrasound: higher first-pass success, fewer attempts, fewer inadvertent venipunctures.

Continued standard advanced cardiac life support

19, 20

Two trials only, both single-centre, both stopped early (one for superiority, one for futility). Modified intention-to-treat analysis. No funding declared.

180-day survival with favourable neurological outcome (CPC 1–2), modified intention to treat

Invasive approach — intra-arrest transport with in-hospital ECPR initiation

In centres able to deliver it, the invasive approach significantly improves 30- and 180-day neurologically favourable survival. The subgroup finding that benefit was LARGER in patients resuscitated for more than 45 minutes is the strongest published argument against a hard 60-minute cut-off.

Pooled individual patient data from the two randomised trials that compared intra-arrest transport with in-hospital ECPR initiation against continued standard ALS. 147 invasive, 139 standard. Median age 57 and 58; median resuscitation duration 58 vs 49 minutes (P = 0.17). INCEPTION is deliberately excluded because it did not test the same strategy.

It answers ARREST and PRAGUE-OHCA's question, not INCEPTION's. It does not generalise to systems unlike these two dedicated single centres, and it cannot separate the circuit from the transport and invasive-assessment components.

45 (32.4%) vs 29 (19.7%); absolute difference 12.7% (2.6–22.7), P = 0.015. 180-day survival 47 (33.8%) vs 33 (22.4%), HR 0.59 (0.43–0.81), P = 0.0009. 30-day favourable neurological outcome 44 (31.7%) vs 24 (16.3%), absolute difference 15.4% (5.6–25.1), P = 0.003. 30-day cardiac recovery 60 (43.2%) vs 46 (31.3%), absolute difference 11.9% (0.7–23), P = 0.05. Shockable rhythms: absolute difference 18.8% (7.6–29.4), P = 0.01; HR 2.26 (1.23–4.15), P = 0.009. CPR longer than 45 minutes: HR 3.99 (1.54–10.35), P = 0.005.

Conventional CPR (standard advanced cardiac life support)

19, 20, 21

A quarter to a third of the ECPR arm never received ECPR. No ECPR protocol was imposed, so practice varied. Very low per-centre volume. Underpowered for the observed effect size. The Red Book and the trialists describe the resuscitation trigger differently (three rounds of ALS versus 15 minutes).

30-day survival with favourable neurological outcome (CPC 1–2)

Extracorporeal CPR

This is the effectiveness trial, and it is the one most hospitals should read first. Pragmatic implementation of ECPR across ten ordinary centres, without an imposed ECPR protocol, produced no detectable benefit. Critically, the excess low-flow time was generated INSIDE the hospitals — the ambulance services were as fast as those in the positive trials.

Adults 18–70 with witnessed out-of-hospital cardiac arrest, bystander CPR, initial ventricular arrhythmia (99% VF/VT), no ROSC within 15 minutes of ALS; randomised pre-hospital. 10 centres, Netherlands; about 800,000 catchment per centre; 0.29 attempted cannulations per centre per month. 134 analysed of 160 randomised (26 excluded for not meeting criteria at hospital admission).

It does not show that ECPR is ineffective. It shows that implementing ECPR does not reproduce the results of dedicated single centres. The per-protocol and Bayesian analyses both favour ECPR without reaching conventional significance.

14/70 (20%) vs 10/64 (16%); odds ratio 1.4 (0.5–3.5); P = 0.52. Serious adverse events per patient similar. Mean low-flow 74.7 min — the LONGEST of the three trials despite a mean on-scene time of 13 min, the shortest; hospital arrival to start of cannulation 16 min plus 20 min of cannulation. Delivery of ECPR in the ECPR arm is reported as 74% (52/70) by the trialists' comparison and 66% (46/70) by the trial group's own secondary analysis — BOTH RECORDED, NEITHER ASSERTED. Per-protocol analysis (n=81): 15% vs 9%, adjusted OR 1.9 (0.4–9.3), P = 0.393, with 84% posterior probability of any benefit and 61% of a 5% absolute risk reduction. Bayesian ITT re-analysis: absolute risk difference 3.6% (95% CrI −9.5 to 16.7), median RR 1.22 (0.59–2.51), 42% probability of a minimal clinically important difference at both 30 days and 6 months.

Standard advanced cardiac life support continued on hospital arrival

19, 20, 21

30 patients. Single centre. Open label. Stopped early for superiority, which systematically inflates effect size. Primary endpoint was survival, not neurologically favourable survival.

Survival to hospital discharge — note this differs from the other two ECPR trials, which used neurologically favourable survival

Early ECMO-facilitated resuscitation on hospital arrival, with percutaneous coronary intervention as needed

ECPR has efficacy under near-ideal conditions: one high-volume centre, one team, one rhythm, mandatory mechanical compressions, a short transfer ceiling, and in-hospital randomisation so that the only difference between arms is the addition of ECPR. This is the strongest single effect estimate in the ECPR literature.

Adults 18–75 with out-of-hospital cardiac arrest and refractory ventricular fibrillation or pulseless VT; no ROSC after three shocks; mechanical CPR device applied; estimated transfer time under 30 minutes. Single centre, University of Minnesota; catchment 3.5 million; 18 ECPR cases per year.

Nothing about non-shockable rhythms, nothing about lower-volume or multicentre settings, and nothing about long transfer distances. It cannot be used to predict what a new programme will achieve — INCEPTION is the trial for that question.

6/14 (43%) vs 1/15 (7%); risk difference 36.2% (95% credible interval 3.7–59.2); posterior probability of ECMO superiority 0.9861. Stopped early at the first preplanned interim analysis on DSMB recommendation after 30 of a planned 174 patients. Mean low-flow time 59 min; mean on-scene time 23 min; door-to-ECMO 12 min. 80% (12/15) of the ECPR arm actually received ECPR; no crossover in the control arm.

Continued standard advanced cardiac life support on scene

19, 20, 21, 22

Single centre. Stopped for futility, so underpowered for the primary endpoint. Bundle intervention. A third of the ECPR arm never received ECPR.

Survival with good neurological outcome (CPC 1–2) at 180 days

Invasive strategy — mechanical compressions, intra-arrest transport to a cardiac centre, ECPR and immediate invasive assessment and treatment. The randomised unit is the whole strategy, not the circuit.

Two findings matter more than the headline. The control arm did far better than designed — 22% against a presumed 10% — partly because building the invasive pathway improved on-scene logistics as well. And widening entry to non-shockable rhythms diluted the effect: about 5% survival in that subgroup against about 40% CPC 1–2 in shockable rhythms on post-hoc analysis. Among patients with neurologically intact survival after more than 45 minutes of CPR, 24 of 26 had received ECPR.

Adults 18–65 with witnessed out-of-hospital cardiac arrest of presumed cardiac origin without ROSC; all initial rhythms (61% VF/VT, 22% asystole, 17% PEA); randomised pre-hospital after 5 minutes of unsuccessful ALS. Single centre, Prague; catchment 1.25 million; 12 ECPR cases per year in the trial arm across 92 months.

It does not isolate the effect of the circuit from the effect of intra-arrest transport and immediate invasive assessment. It does not tell a low-volume centre what to expect.

39 (31.5%) vs 29 (22.0%); odds ratio 1.63 (0.93–2.85); difference 9.5% (−1.3 to 20.1); P = 0.09. Stopped early for futility at 256 of a planned 285. 30-day neurological recovery 38 (30.6%) vs 24 (18.2%); OR 1.99 (1.11–3.57); P = 0.02. Bleeding 31% vs 15%. Mean low-flow 62 min (given as 61 min in the same source's text); on-scene 13 min; door-to-ECMO 12 min. 66% (82/124) of the ECPR arm received ECPR; 8% (10/132) crossover in the control arm.

Conventional approach, permitting rescue transseptal left atrial cannulation if clinical signs of increased left ventricular afterload appeared

15

Single-centre, open-label, 116 patients - powered only for a large effect. 50% crossover in the control arm dilutes any true difference. Transseptal left atrial cannulation reduces preload but not afterload, so a null result may reflect the modality rather than the concept. Read via the published abstract and a peer-reviewed review; full text of the Circulation paper not retrieved (inaccessible from this environment). Volume, issue and page numbers unverified. ClinicalTrials.gov NCT04775472.

All-cause mortality within 30 days

Early routine left ventricular unloading by intracardiac-echo-guided transseptal left atrial cannulation, Y-connected into the ECMO drainage limb, within 12 hours of randomisation (achieved at a median of 1.1 hours)

Routine early left atrial venting in unselected VA ECMO patients does not reduce mortality at 30 days or at 1 year. It does clear the lungs about two days faster. The trial is best read as a TIMING and SELECTION trial: unloading everybody early was compared with unloading the half who developed a clear indication, and the two were equivalent.

Adults with cardiogenic shock undergoing VA ECMO, single centre (Chonnam National University Hospital, Gwangju, South Korea), March 2021 to September 2022. Mean age 67.6 plus or minus 13.5 years; 34 (29.3%) women. 58 early versus 58 conventional; all 116 completed the trial.

It does not test unloading against no unloading - half the control arm was unloaded. It does not test the intra-aortic balloon pump or a micro-axial pump, which are the devices the observational mortality signal actually belongs to. It cannot address whether a preload-only strategy that leaves the aortic valve shut is the right physiological target. Single centre, single operator group, single modality.

30-day death 27 (46.6%) early versus 26 (44.8%) conventional; hazard ratio 1.02 (95% CI 0.59 to 1.74), P=0.942. At 1 year (Eur Heart J Acute Cardiovasc Care 2025;14(4):203-211, DOI 10.1093/ehjacc/zuae150): 33/58 (56.9%) versus 32/56 (57.1%), HR 0.97 (0.60 to 1.58), P=0.887, with no difference in cardiac or non-cardiac mortality, heart-failure rehospitalisation (HR 1.17, 0.43 to 3.24) or the composite. CROSSOVER: 29 of 58 (50%) conventional patients received rescue transseptal cannulation on a clear indication, at a median of 21.8 hours (IQR 12.4 to 52.2). Only significant secondary difference: faster resolution of pulmonary congestion in the early group, median 3 days (IQR 2 to 6) versus 5 (3 to 7), P=0.027.

Conventional strategy, unloading only when clinically indicated (n=30); median time from ECMO insertion to unloading 48.4 hours (IQR 47.8 to 96.5)

15

Very small (60 patients), two centres, open-label, and a process endpoint (weaning) rather than a patient-centred one. High control-arm unloading rate. Read via the published abstract and a peer-reviewed review; full text not retrieved. Volume, issue, pages and full author list unverified - first author reported as Park H.

Rate of successful weaning from VA ECMO

Early left ventricular unloading - percutaneous transseptal left atrial cannulation via the femoral vein, incorporated into the ECMO venous circuit, performed at the time of VA ECMO insertion (n=30)

The second randomised test of early routine left atrial venting, using the same modality as EARLY-UNLOAD and a weaning rather than mortality endpoint, and equally null. It reproduces the same dissociation: congestion clears faster, outcomes do not change. Pooled with EARLY-UNLOAD (176 patients, Eur J Heart Fail 2024, DOI 10.1002/ejhf.3178) there is no difference in in-hospital mortality, weaning or bridging, and no difference in stroke, limb ischaemia, tamponade or bleeding.

Adults receiving VA ECMO for cardiac arrest or shock at two institutions, December 2018 to August 2022. 60 patients randomised 1:1. Unloading was actually delivered in 29 (96.7%) of the early group and 23 (76.7%) of the conventional group.

77% of the conventional arm was eventually unloaded, so this is not a comparison against no unloading. It does not test afterload-reducing devices. With 60 patients it is powered only for very large effects.

Weaning 70.0% versus 76.7%; relative risk 0.91 (95% CI 0.67 to 1.24), P=0.386. Survival to discharge 53.3% versus 50.0%, P=0.796. The pulmonary congestion score index improved significantly only in the early group (2.0 plus or minus 0.7 at baseline to 1.7 plus or minus 0.6 at 48 hours, P=0.008). Authors' conclusion: the results do not support routine application of early unloading after VA ECMO insertion.

Liberal oxygen strategy: SaO2 97-100%, oxygenator FbO2 1.0, ventilator FiO2 minimum 0.5

14, 13, 11, 30

Open-label. Modest size for a mortality endpoint. The 29.5% vs 1.3% deviation rate means achieved separation between arms was imperfect, so a true difference would have been harder to detect — the null is a little weaker than it appears. Primary endpoint (ICU-free days) had a median of 0 in both arms, which limits its discriminating power. Full author list beyond the first three and the precise definition of 'major protocol deviation' not independently checked.

ICU-free days to day 28

Conservative oxygen strategy: SaO2 92-96%, oxygenator FbO2 0.5-1.0, ventilator FiO2 minimum 0.21

THE FIRST RANDOMISED TRIAL OF HOW TO SET VA ECMO rather than whether to use it — it corrects Chapter 13's statement that no VA target has randomised support. A conservative oxygen target confers no benefit on any measured outcome. Equally important and more practical: the conservative target is HARD TO DELIVER — nearly a third of that arm could not be held in band, because the oxygenator returns blood at near-maximal saturation and VA mixing physiology offers little fine control. A target a trial team could hold in only 70% of patients should not be enforced rigidly on a ward round.

Adults over 18 commenced on VA ECMO in ICU. Registry-embedded, open-label, multicentre, parallel-group RCT; enrolled at 12 hospitals, drawn from 934 VA ECMO patients reported to the registry across 26 hospitals. 149 conservative, 151 liberal. Exclusions: randomisation more than 6 h after ECMO initiation, pregnancy, unwillingness to receive blood products, concurrent oxygen titration study, clinician-specified oxygen targets.

It does NOT license hyperoxia. The liberal arm targeted SaO2 97-100%, not deliberate hyperoxia, so the trial says nothing about running FbO2 1.0 and ignoring the resulting PaO2 — and the observational literature associating hyperoxia with harm after arrest is not overturned by 300 patients. It says nothing about any other VA target: flow, MAP, pulse contour, DO2:VO2, transfusion threshold and venous saturation all remain untested by randomisation.

Median 0 days (IQR 0-13.7) conservative vs 0 days (IQR 0-13.3) liberal; median treatment effect 0 days (95% CI -3.1 to 3.1). Day 28 mortality 59/149 (39.6%) vs 59/151 (39.1%); day 60 64/149 (43%) vs 62/151 (41.1%). Duration of ECMO and mechanical ventilation, ICU and hospital length of stay, and 6-month functional outcomes all similar. Subgroups consistent (cardiogenic shock, ECPR, age 50 or under vs over 50, SAVE score above/below median, randomisation timing). Critical hypoxaemia, CPR and seizures similar. MAJOR PROTOCOL DEVIATIONS: 44 (29.5%) conservative vs 2 (1.3%) liberal, p less than 0.001. Authors: 'a conservative compared to a liberal oxygen strategy, did not affect the number of ICU-free days to day 28.'

Early conservative strategy permitting downstream VA ECMO on haemodynamic worsening

11, 14

Small and underpowered for realistic effect sizes. Composite primary endpoint including 'implementation of another MCS', which is partly a treatment decision rather than a patient outcome. Very high control-arm crossover. Long recruitment period spanning changing practice. READ VIA SECONDARY SUMMARY (The Bottom Line) because Circulation is inaccessible from this environment; full author list, volume and pages unverified. A 1-year follow-up publication exists and has NOT been retrieved.

Composite of death from any cause, resuscitated cardiac arrest, or implementation of another mechanical circulatory support at 30 days

Immediate VA ECMO

Immediate VA ECMO in SCAI D-E cardiogenic shock is feasible but does not improve outcomes compared with an early conservative approach that keeps ECMO available as rescue. Notably it selected on SEVERITY - the sickest end of the spectrum - and still found nothing.

Severe or rapidly deteriorating cardiogenic shock, SCAI stage D or E, defined as progressive haemodynamic instability requiring repeated vasopressor boluses to maintain MAP above 50 mmHg plus impaired LV function, or multiple haemodynamic and metabolic derangements including raised lactate and reduced mixed venous saturation. Four Czech tertiary ICUs, September 2014 to January 2022. 122 randomised; ITT 58 immediate ECMO vs 59 conservative.

With 39% crossover it cannot show whether ECMO saves individual lives; it compares timing strategies, not the device against nothing. Small (122 patients), powered only to detect a 50% reduction in the composite, so a modest benefit cannot be excluded. Single-country, four centres, recruitment spanning eight years.

63.8% vs 71.2%; risk difference -7.4% (95% CI -24.3 to 9.5) - not significant. No significant difference in 30-day all-cause mortality, neurological outcome, clinically significant bleeding, leg ischaemia, pneumonia or sepsis. 23 control patients (39%) required downstream VA ECMO, of whom 52.2% died.

Usual care alone, with rescue VA ECMO permitted

11, 1, 14, 17

Unblinded. Crossover in the control arm dilutes the comparison. Population deliberately narrow. READ VIA SECONDARY SUMMARY (The Bottom Line) because NEJM is inaccessible from this environment — full author list, volume, pages and the precise endpoint definitions are unverified.

30-day all-cause mortality

Early VA ECMO plus usual care

Early, unselective VA ECMO in infarct-related cardiogenic shock does not reduce 30-day mortality, roughly doubles moderate-to-severe bleeding and nearly triples limb ischaemia requiring intervention. The largest single trial in this field and the anchor of the modern VA evidence base.

Adults with acute myocardial infarction complicated by cardiogenic shock undergoing revascularisation. Inclusion required ALL of: systolic BP under 90 mmHg for over 30 min or catecholamine dependence; signs of impaired organ perfusion; arterial lactate over 3 mmol/L; planned PCI or CABG. Excluded: over 45 min CPR pre-enrolment; mechanical or non-cardiac shock; shock onset over 12 h; severe peripheral vascular disease; age over 80; life-limiting illness. 877 screened, 420 enrolled, 417 in the ITT analysis.

It does NOT compare ECMO with no ECMO: controls could receive rescue ECMO, so it tests an early-unselective STRATEGY against a wait-and-rescue strategy. It says nothing about VA ECMO outside infarct-related shock — myocarditis, poisoning, massive PE, arrhythmic storm, post-cardiotomy failure, graft failure, hypothermia, or bridging to transplant or a durable device were excluded or never studied. Nor does it address patients over 80, those with severe peripheral vascular disease, or shock present for more than 12 hours.

48% vs 49%; relative risk 0.98 (95% CI 0.80 to 1.19), p = 0.81. No difference in poor neurological outcome, renal replacement therapy, repeat revascularisation, time to stabilisation, or ICU/hospital stay. HARMS: limb ischaemia requiring intervention 11% vs 4%; moderate or severe bleeding 23% vs 10%. 28 control patients crossed over to VA ECMO.

Control / usual care

11, 15

Subgroup finding vulnerable to multiplicity. READ VIA SECONDARY SUMMARY (ACC journal scan) because Lancet and ScienceDirect are inaccessible; volume, pages and DOI unverified. IMPORTANT: the summary's account of the device split (611 loading / 503 unloading) alongside its statement that IABP and microaxial pumps were not specifically compared appears internally inconsistent for a nine-trial analysis - VERIFY THE DEVICE BREAKDOWN against the primary before quoting it elsewhere.

6-month all-cause mortality

Temporary mechanical circulatory support (VA ECMO and LV unloading devices)

The largest and longest-follow-up synthesis in this field. Extends the negative finding to 6 months and across device types, and quantifies harm again. Its single positive subgroup is defined by SUBSTRATE and NEUROLOGICAL INTEGRITY rather than by shock severity - the key conceptual finding for patient selection, and the basis of Chapter 11's Controversy 2.

Infarct-related cardiogenic shock, pooled individual patient data from nine randomised trials of temporary mechanical circulatory support of all types. Reported split: 611 loading devices (VA ECMO), 503 left ventricular unloading devices. The nine constituent trials are NOT named in the summary read.

The STEMI-without-hypoxic-brain-injury result is one subgroup among many in a pooled analysis: hypothesis-generating, not practice-defining, and it should not be quoted as licence to cannulate. Still confined to infarct-related shock. Device categorisation into 'loading' and 'unloading' is coarse and pools heterogeneous devices.

Overall HR 0.87 (95% CI 0.74 to 1.03), p = 0.10. LV unloading devices HR 0.80 (0.62 to 1.02), p = 0.075. Loading devices HR 0.93 (0.75 to 1.17), p = 0.55. ONE positive subgroup: STEMI without risk of hypoxic brain injury, HR 0.77 (0.61 to 0.97), p = 0.024. HARMS: major bleeding increased 2.64-fold; vascular complications increased 4.43-fold. Authors' conclusion: 'MCS use should only be reserved for selected patients, i.e., patients with STEMI and CS without risk of hypoxic brain injury.'

Control (usual care, with rescue ECMO available in the constituent trials)

11, 1

IPD meta-analysis of trials sharing a common design weakness (crossover). Constituent trials not identified in the source read. READ VIA SECONDARY SUMMARY (ACC journal scan) because Lancet and ScienceDirect are inaccessible from this environment; volume, pages and DOI unverified.

30-day all-cause mortality

Early VA ECMO

Pooling individual patient data across the randomised evidence confirms no mortality signal for early VA ECMO in infarct-related cardiogenic shock, and quantifies the harm: roughly 2.4-fold major bleeding and 3.5-fold peripheral vascular complications. The absence of ANY positive subgroup is the strongest single argument against unselected use.

Patients with infarct-related cardiogenic shock pooled from four randomised trials: 284 allocated to VA ECMO, 283 to control. The four constituent trials are NOT named in the summary read for this entry.

Confined to infarct-related shock. Inherits the crossover problem of its constituent trials, so it still compares strategies rather than device against nothing. Subgroup analyses were prespecified but the population was too homogeneous to test substrate-based selection - myocarditis, poisoning and PE are simply absent from the pooled dataset.

OR 0.93 (95% CI 0.66 to 1.29) - no significant reduction. Prespecified subgroup analyses consistent and showing no benefit in any subgroup (p for interaction 0.079 or greater). HARMS: major bleeding OR 2.44 (95% CI 1.55 to 3.84); peripheral ischaemic vascular complications OR 3.53 (95% CI 1.70 to 7.34). Authors' conclusion: 'These data do not justify unselected use of VA-ECMO in patients with cardiogenic shock.'

Standard ACLS

1, 19, 20

Single centre, phase 2, 30 patients, stopped early at an efficacy boundary. Highly selected population within a mature, high-volume ECPR system. Early stopping inflates effect size.

Survival to hospital discharge

Early ECMO-facilitated resuscitation

Within a mature single-centre system, an ECMO-facilitated pathway for refractory shockable OHCA produced a large survival difference.

Out-of-hospital cardiac arrest with refractory ventricular fibrillation or pulseless VT, no ROSC after three shocks, age 18-75

It does not establish generalisability to other systems, and the effect estimate should be regarded as an upper bound.

6/14 (43%) vs 1/15 (7%); posterior probability of ECMO superiority 0.9861, crossing the prespecified 0.986 efficacy boundary

Usual care alone

1, 11, 14, 15

Open-label; crossover to ECMO in the control arm; confined to infarct-related shock in revascularised patients.

All-cause death at 30 days

Early VA ECMO plus usual care

Routine early VA ECMO in infarct-related cardiogenic shock does not reduce 30-day mortality and roughly doubles major bleeding and limb ischaemia.

Acute myocardial infarction complicated by cardiogenic shock, planned early revascularisation

It says nothing about non-infarct cardiogenic shock, ECPR, rescue cannulation in refractory shock, or LV unloading strategies.

48% vs 49%; RR 0.98 (95% CI 0.80-1.19), p=0.81. Moderate/severe bleeding 23% vs 10%; limb ischaemia requiring intervention 11% vs 4%.

Early conservative strategy with ECMO permitted as rescue

1, 11, 14

Small; 23/59 (39%) of the conservative arm received downstream VA ECMO, of whom 52% died. Composite endpoint includes MCS escalation, which favours the immediate-ECMO arm by construction.

Composite of death from any cause, resuscitated circulatory arrest, or implementation of another mechanical circulatory support at 30 days

Immediate VA ECMO

An immediate VA ECMO strategy did not improve a 30-day composite outcome compared with early conservative management with rescue ECMO available.

Rapidly deteriorating or severe cardiogenic shock, SCAI stage D-E

It is not powered for mortality and does not address non-infarct phenotypes or the timing of rescue cannulation.

63.8% (37/58) vs 71.2% (42/59); risk difference -7.4% (95% CI -24.3 to 9.5)

Conventional CPR

1, 19, 20, 21

Only 46/70 (66%) of patients allocated to ECPR were successfully cannulated. Multicentre with variable institutional ECPR experience; long low-flow times.

Survival with favourable neurological outcome (CPC 1-2) at 30 days

Extracorporeal CPR

In a multicentre setting with mixed institutional experience, an ECPR strategy did not significantly improve 30-day neurologically favourable survival.

Refractory out-of-hospital cardiac arrest with an initial shockable rhythm, witnessed, age 18-70

With one third of the intervention arm never cannulated, it tests an ECPR pathway rather than ECPR itself, and cannot exclude benefit in systems with higher cannulation success and shorter time to flow.

20% (14/70) vs 16%; OR 1.4 (95% CI 0.5-3.5), p=0.52. 160 randomised, 134 in the primary analysis.

Conventional management with ECMO permitted as rescue

1, 4, 7

Stopped for futility at 75% of planned enrolment. 35/125 (28%) of controls crossed over to rescue ECMO; 20/35 (57%) of those died.

All-cause mortality at 60 days

Immediate VV ECMO

Early ECMO in very severe ARDS produced an 11 percentage point absolute mortality difference that did not reach conventional significance in an underpowered trial.

Very severe ARDS: PaO2/FiO2 <50 for >3 h, or <80 for >6 h, or pH <7.25 with PaCO2 >60 for >6 h

It does not compare ECMO with no ECMO. With 28% crossover the trial compares early ECMO against rescue ECMO, and therefore cannot exclude a clinically important benefit.

35% (44/124) vs 46% (57/125); RR 0.76 (95% CI 0.55-1.04), p=0.09

Standard resuscitation on scene

1, 19, 20, 23

Stopped for futility at a prespecified interim analysis. Tests a whole bundle rather than ECPR alone; single high-performing system.

Survival with CPC 1-2 at 180 days

Invasive bundle: intra-arrest transport, ECPR if no ROSC within 60 minutes, immediate invasive assessment and treatment

An invasive intra-arrest bundle including ECPR produced a directionally favourable but statistically non-significant improvement in 180-day neurologically favourable survival.

Refractory out-of-hospital cardiac arrest of presumed cardiac cause

It cannot attribute any of the observed difference specifically to the extracorporeal circuit rather than to intra-arrest transport or immediate invasive assessment.

31.5% vs 22.0%; absolute difference 9.5% (95% CI -1.3 to 20.1), p=0.09

Optimal medical therapy without routine early VA ECMO

1, 11, 17

Four trials, 284 VA-ECMO and 283 control patients; all open-label; restricted to infarct-related shock.

All-cause death at 30 days

Early VA ECMO

At individual patient data level there is no mortality signal for routine early VA ECMO in infarct-related cardiogenic shock, and a consistent harm signal for bleeding and limb ischaemia.

Pooled individual patient data from randomised trials of early VA ECMO in infarct-related cardiogenic shock

It cannot identify a responder subgroup that was not represented in the constituent trials, and it does not address rescue use or non-infarct phenotypes.

OR 0.93 (95% CI 0.66-1.29); no benefit in any prespecified subgroup (p for interaction >=0.079). Major bleeding OR 2.44 (1.55-3.84); peripheral ischaemic vascular complications OR 3.53 (1.70-7.34).

Continued conventional management at the referring centre

1, 4, 7

Randomised to referral, not to ECMO: only 68/90 (76%) of the referral arm actually received ECMO. Control arm had no protocolised lung-protective ventilation. Single ECMO centre.

Death or severe disability at 6 months

Referral and transfer to a single specialist ECMO centre

A pathway of transfer to a specialist ECMO centre improved 6-month death-or-severe-disability compared with conventional management as delivered in 2001-2006.

Adults with severe but potentially reversible respiratory failure (Murray score >3 or pH <7.20)

It does not isolate the effect of the ECMO circuit itself, and it does not describe a comparison against modern protocolised lung-protective ventilation.

37% (33/90) vs 53% (46/87); RR 0.69 (95% CI 0.05-0.97), p=0.03