Part I — Foundations · Chapter 1
Evidence search date: 6 September 2026. Guideline inventory verified against the ELSO guidelines index on that date. Trial results transcribed from published reports of the primary trials.
Clinical Question
When a patient's heart or lungs can no longer meet metabolic demand despite optimised conventional therapy, what exactly does an extracorporeal circuit replace — and what does it leave untouched?
Why This Matters
Almost every avoidable error in ECMO comes from a category mistake. A clinician increases the sweep gas because the saturation is low. A team escalates pump speed because the lactate is rising. A patient is cannulated because the ventilator settings look frightening, without anyone having named what the circuit is a bridge to.
Each of these is a failure of framing, not of knowledge. ECMO is not a treatment for respiratory or cardiac failure; it is a mechanical substitute for two specific functions — gas exchange and blood flow — that buys time for something else to work. If you cannot say what that something else is, the circuit has no endpoint.
This chapter builds the mental model that the remaining 91 chapters assume.
Definitions
Terminology in this field is used loosely in conversation and precisely in the literature. This book uses the precise sense throughout.
Term | Meaning as used in this book | Common misuse |
ECLS (extracorporeal life support) | The umbrella term for any prolonged extracorporeal circuit supporting gas exchange, circulation, or both. | Used interchangeably with VA ECMO in cardiology literature. |
ECMO | ECLS delivered through a membrane oxygenator for days to weeks in an ICU setting. The term derives historically from the spiral-wrapped silicone membrane lung. | Used as if it named a single therapy rather than a family of configurations. |
VV ECMO | Venous drainage, venous return. Supports gas exchange only. Provides no circulatory support. | Assumed to raise blood pressure. It does not, except by correcting hypoxaemic vasoplegia or hypercapnia. |
VA ECMO | Venous drainage, arterial return. Supports gas exchange and systemic blood flow. With retrograde peripheral return, this comes at the cost of increased LV afterload. | Described as "unloading" the left ventricle. It unloads preload and increases afterload — see Chapter 15. |
ECPR | VA ECMO instituted during ongoing cardiopulmonary resuscitation, before return of spontaneous circulation. | Used for VA ECMO started after ROSC, which is post-arrest VA ECMO, not ECPR. |
ECCO₂R | Low-blood-flow extracorporeal CO₂ removal. Clears CO₂ effectively; contributes little oxygenation. | Called "mini-ECMO" and assumed to help hypoxaemia. It largely does not. |
Cardiopulmonary bypass | Operating-theatre circuit with an open reservoir, full anticoagulation, cardiotomy suction and typically cardioplegic arrest. Hours, not days. | Treated as the same thing as VA ECMO. The anticoagulation intensity and circuit design differ fundamentally. |
The Core Idea: ECMO Is a Parallel Circuit
Physiology
The patient's own cardiopulmonary system does not stop when ECMO starts. The circuit is placed in parallel with it. Every observed variable — arterial saturation, PaCO₂, mean arterial pressure, mixed venous saturation — is therefore the summed output of two systems, weighted by how much of the venous return each one handles.
This single fact explains most of what is confusing about ECMO. The arterial blood gas does not describe the circuit; it describes a mixture. Interpreting it requires knowing the proportions.
An extracorporeal circuit can do exactly two things:
- Exchange gas — add oxygen to and remove carbon dioxide from the blood that passes through the membrane lung.
- Move blood — generate flow, and therefore pressure, when returned to the arterial side.
VV ECMO does the first. VA ECMO does both. Nothing else the circuit appears to do is a primary effect; it is a consequence of one of these two, or of the injury the circuit itself causes.
Clinical Pearl
The most useful opening question on any ECMO round is not "what are the settings?" but "what fraction of this patient's venous return is going through the circuit, and what is the rest of the cardiac output doing?" Every subsequent number becomes interpretable once you can answer that.
Five Quantities That Must Never Be Conflated
These five are routinely used as if they were one. They are not, and the distinctions are the backbone of every troubleshooting chapter in this book.
Quantity | What it is | Primarily set by | What it is not |
ECMO flow | Volume of blood per minute through the circuit (L/min) | Pump RPM, drainage cannula size and position, patient volume status, circuit resistance | Not a measure of oxygen delivered. A high flow through a failing oxygenator delivers little. |
Oxygen delivery (DO₂) | Oxygen presented to tissues per minute (mL/min) | Total cardiac output × arterial oxygen content — and content is dominated by haemoglobin | Not the same as saturation. A saturation of 100% with Hb 6 g/dL is poor delivery. |
Oxygenation | Arterial oxygen saturation and tension in the patient | Circuit flow relative to cardiac output, recirculation, native lung function, FdO₂ | Not evidence of adequate delivery, and not a measure of circuit performance. |
CO₂ clearance | Carbon dioxide removed by the membrane lung per minute | Sweep gas flow, membrane surface area and integrity | Not meaningfully changed by FdO₂, and only modestly by blood flow. CO₂ clearance is efficient at low flows — which is why ECCO₂R works. |
Ventilation | What the patient's own lungs and the ventilator are doing | Ventilator settings, respiratory drive, lung mechanics | Not the same as CO₂ clearance once a membrane lung is in circuit. The two run in parallel and must be assessed separately. |
Pitfall
Turning up the sweep for hypoxaemia and turning up the FdO₂ for hypercapnia. Both are common, both are physiologically inverted, and both waste time during a deteriorating episode. Sweep gas governs CO₂. Blood flow and FdO₂ govern oxygen. See Chapter 6 and Chapter 9.
Oxygen Content and Delivery: The Equation That Reframes the Ward Round
The conceptual framework becomes concrete through one equation.
Arterial oxygen content
Variables and units
- — arterial oxygen content, mL O₂ per dL of blood
- — haemoglobin concentration, g/dL
- — arterial oxygen saturation, expressed as a fraction
- — arterial oxygen tension, mmHg
- 1.34 — mL O₂ carried per gram of fully saturated haemoglobin (Hüfner constant; values between 1.34 and 1.39 appear in the literature)
- 0.003 — mL O₂ dissolved per dL per mmHg
Assumptions. Normal haemoglobin binding without significant carboxyhaemoglobin or methaemoglobin; saturation measured by co-oximetry rather than inferred from PaO₂.
Oxygen delivery
where is total cardiac output in L/min and the factor of 10 converts dL to L. On ECMO, the oxygen a tissue bed sees is delivered by the mixture, not by the circuit alone.
Worked example
A 70 kg adult on VV ECMO. Native cardiac output 6 L/min. Hb 9 g/dL. SaO₂ 88%. PaO₂ 55 mmHg. Estimated VO₂ 250 mL/min.
Step | Calculation | Result |
Bound oxygen | 1.34 × 9 × 0.88 | 10.61 mL/dL |
Dissolved oxygen | 0.003 × 55 | 0.17 mL/dL |
CaO₂ | 10.61 + 0.17 | 10.78 mL/dL |
DO₂ | 10.78 × 6 × 10 | 647 mL/min |
Oxygen extraction ratio | 250 ÷ 647 | 39% |
Now compare two interventions that a team might argue about at the bedside.
Intervention | New CaO₂ | New DO₂ | Gain |
Raise SaO₂ from 88% to 92% (Hb unchanged at 9) | 11.26 mL/dL | 676 mL/min | +29 mL/min (+4.5%) |
Raise Hb from 9 to 12 g/dL (SaO₂ unchanged at 88%) | 14.32 mL/dL | 859 mL/min | +212 mL/min (+33%) |
Clinical interpretation. Chasing the last few percentage points of saturation on VV ECMO is, in oxygen-delivery terms, close to noise. Haemoglobin and cardiac output are the levers that move delivery. This is why a saturation of 85–88% is frequently tolerated on VV ECMO when delivery is adequate and lactate is falling — and why the reflex to escalate flow for a saturation number, rather than for a delivery deficit, is a recurring error.
Limitations of this reasoning. The arithmetic above says nothing about regional delivery. A global DO₂ of 647 mL/min may be ample for the whole body and inadequate for the brain or a limb distal to an arterial cannula. It also assumes a stable VO₂; fever, shivering, agitation and work of breathing can raise VO₂ substantially and shift the same numbers from comfortable to marginal. Finally, the optimal transfusion threshold on ECMO is not established by this equation — see Chapter 46, where the evidence is genuinely weak.
What ECMO Does Not Do
This list is short and worth committing to memory.
- It does not treat the underlying disease. Pneumonia still needs antimicrobials; infarction still needs revascularisation; pulmonary embolism still needs a decision about reperfusion.
- It does not reverse established structural injury. Fibrotic lung and necrotic myocardium do not recover because a circuit is attached. This is why the reversibility question precedes the cannulation decision, not the other way round.
- It does not guarantee oxygen delivery. Flow is not delivery. See above.
- It does not protect the brain. Neurological injury is among the most consequential complications of ECMO and is caused partly by the therapy itself — embolism, haemorrhage, abrupt PaCO₂ correction, differential perfusion. See Chapters 23 and 28.
- It does not create time indefinitely. Circuit-related complications accumulate with duration. A run without a destination becomes a slow accumulation of harm.
Danger
The most dangerous moment in an ECMO programme is not a circuit emergency. It is the decision to cannulate a patient whose organ failure is irreversible and who is not a candidate for transplantation or durable support. That decision commits a family, a team and a bed to a course with no exit, and it is made in minutes. Chapters 69 to 77 exist because of it.
The Configuration Decision
The choice between configurations is one physiological question, asked before any technical question about cannulae.
The common failure is to answer the second question after cannulation rather than before it. Configuration detail is developed in Chapter 3; selection criteria in Chapters 4, 11 and 20.
Every Run Is a Bridge — to Something You Can Name
ECMO has no therapeutic endpoint of its own. Every run is a bridge, and the destination should be recorded in the notes on the day of cannulation.
Bridge to | What must happen for it to succeed | Where it is covered |
Recovery | The organ recovers enough to sustain life without the circuit | Chapter 71 |
Decision | Information not available at cannulation becomes available — neurological status, candidacy, family goals | Chapter 70 |
Transplant | The patient is, or becomes, a listed candidate and an organ arrives in time | Chapters 72 and 82 |
Durable support | The patient is a candidate for LVAD or equivalent | Chapter 73 |
Nothing | Recognised early, this is the indication not to cannulate; recognised late, it becomes withdrawal | Chapters 74 and 75 |
The Evidence Landscape
The randomised evidence for ECMO is smaller, more conflicted and more heavily qualified than the volume of published ECMO literature suggests. The table below summarises the trials that define current practice; each is analysed in full in its own chapter.
Trial | Year | Population | N | Primary endpoint | Result | Key limitation |
CESAR | 2009 | Severe potentially reversible adult respiratory failure | 180 | Death or severe disability at 6 months | 37% vs 53%; RR 0.69 (95% CI 0.05–0.97), p=0.03 | Randomised to referral, not to ECMO; only 68/90 (76%) of the referral arm received ECMO; control arm had no protocolised lung-protective ventilation |
EOLIA | 2018 | Very severe ARDS | 249 | 60-day mortality | 35% vs 46%; RR 0.76 (95% CI 0.55–1.04), p=0.09 | Stopped for futility at 75% of target; 28% of controls crossed over to rescue ECMO, of whom 57% died — the comparison is early vs rescue ECMO, not ECMO vs none |
ECMO-CS | 2023 | Rapidly deteriorating or severe cardiogenic shock (SCAI D–E) | 117 analysed | Composite of death, resuscitated arrest or other MCS at 30 days | 63.8% vs 71.2%; risk difference −7.4% (95% CI −24.3 to 9.5) | Small; 39% of the conservative arm received rescue VA ECMO |
ECLS-SHOCK | 2023 | Infarct-related cardiogenic shock undergoing revascularisation | 417 | 30-day all-cause death | 48% vs 49%; RR 0.98 (95% CI 0.80–1.19), p=0.81 | Bleeding 23% vs 10%; limb ischaemia requiring intervention 11% vs 4%. Does not address non-infarct shock or ECPR |
IPD meta-analysis (infarct-related shock) | 2023 | Pooled individual patient data | 567 (4 trials) | 30-day death | OR 0.93 (95% CI 0.66–1.29); no benefit in any prespecified subgroup | Major bleeding OR 2.44 (1.55–3.84); peripheral ischaemic vascular complications OR 3.53 (1.70–7.34) |
ARREST | 2020 | OHCA with refractory VF/pVT | 30 | Survival to discharge | 6/14 (43%) vs 1/15 (7%); posterior probability of superiority 0.9861 | Single centre, 30 patients, stopped early at a Bayesian efficacy boundary; a highly selected, mature system |
Prague OHCA | 2022 | Refractory OHCA | 256 | Survival with CPC 1–2 at 180 days | 31.5% vs 22.0%; absolute difference 9.5% (95% CI −1.3 to 20.1), p=0.09 | Stopped for futility; tests a bundle (intra-arrest transport, ECPR, immediate invasive assessment), not ECPR alone |
INCEPTION | 2023 | Refractory OHCA, shockable rhythm | 160 randomised; 134 analysed | 30-day survival with CPC 1–2 | 20% vs 16%; OR 1.4 (95% CI 0.5–3.5), p=0.52 | Only 66% of those allocated to ECPR were successfully cannulated; multicentre, variable institutional experience |
What this evidence establishes
- In infarct-related cardiogenic shock, routine early VA ECMO does not reduce 30-day mortality and materially increases bleeding and limb ischaemia. This is the strongest and most consistent negative signal in the field, and it comes from randomised data pooled at the individual patient level.
- In very severe ARDS, the effect of early ECMO is compatible with a clinically important benefit but was not demonstrated at conventional statistical thresholds, in a trial where a quarter of the control arm received the intervention as rescue.
- In refractory out-of-hospital cardiac arrest, effect estimates favour ECPR in every randomised trial but reach conventional significance in only the smallest and most selected one. System factors — time to flow, cannulation success rate, case volume — plausibly explain more of the variance between trials than the intervention itself.
What this evidence does NOT establish
- It does not establish that ECMO is ineffective. Trials of routine early application in unselected populations answer a different question from rescue application in a selected one.
- It does not license extrapolation between phenotypes. ECLS-SHOCK studied infarct-related shock; it is not evidence about fulminant myocarditis, pulmonary embolism, poisoning or post-cardiotomy shock, in which the reversibility profile is entirely different.
- It does not tell us who benefits. No trial has prospectively validated a selection rule. Selection remains the single largest determinant of outcome and the least evidence-based part of the process.
- It does not settle configuration, timing, anticoagulation, unloading, ventilation or weaning — none of which were the randomised variable in any of these trials.
Evidence — registry context
The ELSO Registry reported 154,568 ECMO runs between 2009 and 2022 across 780 centres (557 reporting in 2022), with the median annual adult run volume per centre rising from 4 to 15 over that period. By 2022 the Registry had enrolled its 200,000th patient and its 100,000th patient discharged alive. Survival to hospital discharge was 68.5% for neonatal respiratory support and 29.5% for adult ECPR. (ELSO Registry International Report 2022, ASAIO J 2024)
Registry data describe what happened to patients who were selected for ECMO by clinicians. They cannot establish what would have happened without it, and centres that report are not a random sample. Use registry figures for calibration and case-mix context — never as evidence of effect.
Reading the Guidelines
ECMO guidance is not a single document, and the pieces are of very different ages and methodological strength. The inventory below reflects the ELSO guidelines index as at the evidence search date.
Guideline | Date | Note |
ELSO General Guidelines for all ECLS cases, v1.4 | August 2017 | The general document is now nine years old. Treat it as orientation, not as current practice guidance. |
Adult Respiratory Failure Managed with VV ECMO | June 2021 | Predates several relevant post-pandemic analyses |
Interim Guidelines for VA ECMO in Adult Cardiac Patients | August 2021 | Predates ECLS-SHOCK and the 2023 IPD meta-analysis — see the currency warning below |
ECPR in Adults — Interim Guideline Consensus Statement | March 2021 | Predates INCEPTION |
Adult and Paediatric Anticoagulation Guidelines | January 2022 | Chapter 42 covers the substantial practice variation that persists |
Transport Guideline | February 2022 | Chapter 60 |
Neurological Monitoring and Management for Adult ECMO Patients | December 2024 | Chapter 28 |
Early Rehabilitation or Mobilization of Adults on ECMO | February 2025 | Chapter 59 |
Pregnancy and Peripartum ECMO (narrative) | October 2025 | Chapter 61 |
Accidental Hypothermia (narrative) | October 2025 | Chapter 65 |
ECMO Training and Continuing Education (narrative) | June 2026 | Most recent ELSO guideline at the search date |
Potentially outdated recommendation — review required
The ELSO adult cardiac guidance is dated August 2021 and therefore predates ECLS-SHOCK (2023) and the individual patient data meta-analysis (2023), which together constitute the most important randomised evidence in VA ECMO. Any recommendation about routine early VA ECMO in infarct-related cardiogenic shock drawn from a pre-2023 document must be checked against that evidence before it is applied.
Evidence — how ELSO guidance is graded
ELSO guidance is produced by three methods: narrative (expert synthesis of the literature supplemented by expert judgment), consensus (formal voting to quantify agreement), and GRADE. As of the 2026 ELSO guidelines roadmap, no ELSO guideline uses GRADE methodology, and the organisation has set out plans for formal 3–5 year update cycles. (O'Neil ER, et al. Perfusion 2026)
This matters for how the reader should weight a recommendation: an ELSO narrative guideline is high-quality expert synthesis, not a graded evidence appraisal, and this book labels it accordingly.
The Ten Questions
When any physiological variable deteriorates on ECMO, work through these in order. They recur, adapted, in every troubleshooting chapter.
- Is the measurement correct? (Sampling site, transducer, calibration, cannula position relative to the sampling point.)
- Is the circuit functioning? (Pump, oxygenator, pressures.)
- Is the cannula in the right place?
- Has patient physiology changed?
- Has native cardiac output changed?
- Has oxygen consumption changed? (Fever, agitation, shivering, work of breathing, seizures.)
- Has metabolic demand or its distribution changed?
- Has haemoglobin changed?
- Is there new pathology? (Pneumothorax, tamponade, haemorrhage, sepsis, PE.)
- Is this configuration still the right one?
Clinical Pearl
Question 1 resolves more crises at 3 AM than questions 2 to 10 combined. Before acting on any alarming ECMO number, confirm that the number is real.
Controversies
Controversy — Is ECMO a therapy or a platform?
Clinical question: Should ECMO be evaluated in randomised trials as a discrete intervention, in the way a drug is?
Evidence supporting "therapy": It is a discrete, expensive, high-risk intervention with a defined start and stop. Randomisation is feasible, as EOLIA, ECLS-SHOCK, INCEPTION and Prague OHCA demonstrate, and the resulting data have changed practice — particularly the negative findings in infarct-related shock.
Evidence supporting "platform": Outcomes in every trial depend heavily on co-interventions and system factors — time to flow, cannulation success, ventilation strategy, unloading, anticoagulation, centre volume. INCEPTION cannulated only 66% of its ECPR arm; CESAR randomised to referral; EOLIA's control arm crossed over at 28%. In each case the randomised contrast is between care pathways, not between the presence and absence of a circuit.
Current consensus: Both framings are partly right, and the field is moving toward adaptive platform designs that randomise components of ECMO care rather than ECMO itself.
Practical approach: Read each trial for the pathway it tested, not the technology it names. Ask what the control arm actually received, and what proportion of each arm received the other arm's treatment.
Knowledge gap: No randomised evidence identifies which patients benefit. Selection is where the effect lives, and it is the part no trial has yet tested prospectively.
Controversy — What does a "negative" ECMO trial mean?
Clinical question: After ECLS-SHOCK and the IPD meta-analysis, should VA ECMO be offered in cardiogenic shock at all?
Evidence supporting restriction: Pooled individual patient data from four randomised trials show no mortality signal (OR 0.93, 95% CI 0.66–1.29) with a clear excess of major bleeding and limb ischaemia. Consistency across prespecified subgroups is unusual and argues against a hidden responder group within the studied population.
Evidence supporting continued use: The studied population was infarct-related shock in patients undergoing revascularisation. Rescue use in profound refractory shock, and use in non-infarct phenotypes with high intrinsic reversibility — fulminant myocarditis, massive pulmonary embolism, some poisonings, accidental hypothermia — was not tested and cannot be inferred from these data.
Current consensus: Routine early VA ECMO in infarct-related cardiogenic shock is not supported. Selective use in refractory shock and in high-reversibility phenotypes continues, on physiological rationale and observational data.
Practical approach: Treat "cardiogenic shock" as a heterogeneous set of phenotypes, not a single indication, and be explicit about which one you are treating.
Knowledge gap: Randomised data in non-infarct cardiogenic shock are essentially absent.
Evidence Summary
Statement | Certainty | Basis |
ECMO supports gas exchange and, in VA configuration, systemic blood flow; it does not treat the underlying disease | Physiological rationale — not contested | Mechanistic; universal in textbooks and guidelines |
Routine early VA ECMO does not reduce mortality in infarct-related cardiogenic shock and increases bleeding and limb ischaemia | High | Individual patient data meta-analysis of 4 RCTs, 567 patients (2023) |
Early ECMO in very severe ARDS may reduce mortality; the trial result was not statistically significant and was confounded by crossover | Moderate | EOLIA (2018); CESAR (2009) tested referral rather than ECMO |
ECPR may improve neurologically favourable survival in highly selected refractory OHCA within mature systems | Low to moderate | ARREST (positive, n=30, single centre); Prague OHCA and INCEPTION (both directionally favourable, neither significant) |
Oxygen delivery is dominated by haemoglobin and cardiac output rather than by the last few points of arterial saturation | High (arithmetic identity) | Oxygen content equation; the clinical threshold for transfusion on ECMO remains uncertain — Chapter 46 |
Selection determines outcome more than any managed variable | Low — consistent observational signal, no prospective validation | Registry and trial heterogeneity; explicitly a knowledge gap |
Key Takeaways
- ECMO is a parallel circuit, not a replacement. Every measured variable is a mixture of circuit and patient, and interpreting it requires knowing the proportions.
- A circuit does exactly two things: exchanges gas and moves blood. Everything else is a consequence.
- ECMO flow, oxygen delivery, oxygenation, CO₂ clearance and ventilation are five different quantities with different controllers. Conflating them produces most bedside errors.
- Haemoglobin and cardiac output move oxygen delivery; the last few points of saturation barely do.
- Every run is a bridge. If the destination cannot be named on day zero, the indication is questionable.
- The randomised evidence is mostly about pathways, not about the circuit. Read each trial for what the control arm actually received.
- Routine early VA ECMO in infarct-related cardiogenic shock is not supported by randomised evidence and carries a clear harm signal.
- When a number alarms you, confirm the number before treating it.
Key References
- Peek GJ, Mugford M, Tiruvoipati R, et al. Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet. 2009. [DOI/PMID — VERIFICATION REQUIRED]
- Combes A, Hajage D, Capellier G, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome (EOLIA). N Engl J Med. 2018. DOI: 10.1056/NEJMoa1800385. PMID: 29791822
- Ostadal P, Rokyta R, Karasek J, et al. Extracorporeal membrane oxygenation in the therapy of cardiogenic shock: results of the ECMO-CS randomized clinical trial. Circulation. 2023. DOI: 10.1161/CIRCULATIONAHA.122.062949. PMID: 36335478 — 1-year outcomes: PMID 39113628
- Thiele H, Zeymer U, Akin I, et al. Extracorporeal life support in infarct-related cardiogenic shock (ECLS-SHOCK). N Engl J Med. 2023. DOI: 10.1056/NEJMoa2307227. PMID: 37634145
- Zeymer U, Freund A, Hochadel M, et al. Venoarterial extracorporeal membrane oxygenation in patients with infarct-related cardiogenic shock: an individual patient data meta-analysis of randomised trials. Lancet. 2023. PMID: 37643628
- Yannopoulos D, Bartos J, Raveendran G, et al. Advanced reperfusion strategies for patients with out-of-hospital cardiac arrest and refractory ventricular fibrillation (ARREST): a phase 2, single centre, open-label, randomised controlled trial. Lancet. 2020. PMID: 33197396
- Belohlavek J, Smalcova J, Rob D, et al. Effect of intra-arrest transport, extracorporeal cardiopulmonary resuscitation, and immediate invasive assessment and treatment on functional neurologic outcome in refractory out-of-hospital cardiac arrest (Prague OHCA): a randomized clinical trial. JAMA. 2022. PMID: 35191923
- Suverein MM, Delnoij TSR, Lorusso R, et al. Early extracorporeal CPR for refractory out-of-hospital cardiac arrest (INCEPTION). N Engl J Med. 2023. DOI: 10.1056/NEJMoa2204511. PMID: 36720132
- Tonna JE, Boonstra PS, MacLaren G, et al. Extracorporeal Life Support Organization Registry International Report 2022: 100,000 survivors. ASAIO J. 2024;70(2):131–143. DOI: 10.1097/MAT.0000000000002128
- O'Neil ER, Peek GJ, Ontaneda A, et al. ELSO guidelines development: a roadmap. Perfusion. 2026;41(1_suppl). DOI: 10.1177/02676591261425548
- Extracorporeal Life Support Organization. ELSO Guidelines index. Guideline titles and dates as verified 6 September 2026. Available at elso.org
- Brodie D, Peek G, MacLaren G, et al. (eds). Extracorporeal Life Support: The ELSO Red Book, 6th edition. — background for the historical and definitional material in this chapter
Citation verification status. DOIs and PMIDs above were verified against source records during the 6 September 2026 search cycle, except where marked. The CESAR bibliographic identifiers could not be confirmed against the primary record in this cycle and are flagged accordingly; the reported CESAR results were transcribed from a published trial summary and should be re-checked against the original Lancet paper before this chapter is cited in a formal document.
The published CESAR confidence interval (0.05–0.97) is reproduced as printed in the source; its width is an artefact of the reported analysis and not a transcription error.
Educational use only. This chapter does not replace institutional ECMO protocols, local policy, specialist consultation, current guidelines or patient-specific clinical judgement.