<aside> 🧭
Chapter question: An echocardiogram on a patient supported by ECMO produces the same measurements as any other echocardiogram — an ejection fraction, a velocity–time integral, a tricuspid annular excursion. Every one of them was made while a machine was unloading, loading or bypassing the chamber being measured. What does the study actually tell you about the heart — and what has to change about how it is requested, performed and reported before the answer is usable?
</aside>
<aside> 📅
Evidence search date: 18 September 2026.
Primary sources: ELSO Red Book 6th edition — Chapter 48, Bedside Ultrasound (echocardiographic guidance for cannulation, monitoring on venovenous and venoarterial support, lung ultrasound, the paediatric and neonatal sections) and Chapter 29, Weaning and Decannulation in Adult Cardiac Failure (the echocardiographic weaning parameters, the arteriovenous bridge, the pump-controlled retrograde trial off). ELSO Ultrasound Guidance for Extracorporeal Membrane Oxygenation — General Guidelines (Nanjayya and Murphy, May 2015), with its editors listed and its expiry date noted. ISCCM Manual of RRT and ECMO in ICU — Chapter 33, the echocardiographic evaluation-by-stage table. ECMO: A Practical Guide to Management (Taha, Caridi-Scheible, Leiendecker).
External evidence: a 35-section narrative review of echocardiography for ECMO in a cardiac-imaging journal; a 2026 narrative review of diagnostic imaging challenges during ECMO; a European consensus statement on multimodality imaging in mechanical circulatory support; a prospective load-manipulation study of 46 patients imaged at three circuit flows; a prospective weaning cohort of 38 patients; a randomised unloading trial protocol; a retrospective transthoracic series of 11 dual-lumen cannula studies; two cannula-malposition case reports; a pre-cannulation pulmonary vascular study of 21 patients; a resuscitative transoesophageal case report and a systematic review of transoesophageal echocardiography in cardiac arrest; and a case report of delayed tamponade after extracorporeal cardiopulmonary resuscitation.
Every external reference was read as an indexed full-text passage returned by a passage-retrieval service, not as a complete article. Digital object identifiers are reproduced from the search record and have not been independently resolved. Every database row seeded from this chapter is marked Verified = No.
</aside>
Chapter 25 established that every haemodynamic instrument at the ECMO bedside was calibrated in a circulation with one pump, in series, generating a pulse — and that the instruments go on producing numbers without announcing that their assumptions have failed. Echocardiography is the apparent exception and is not one. It looks directly at the chamber rather than inferring from a waveform, which is why it is reached for whenever the other instruments disagree. But it measures loading conditions, and on ECMO the loading conditions are set by a dial.
This chapter is about echocardiography as an instrument: what it can see that nothing else can, what it measures that is not what its label says, and how to request, perform and report it so that the result survives contact with the circuit.
| This chapter owns | Owned elsewhere — cite, do not re-argue |
|---|---|
| Echocardiography as a modality on a circuit — what each measurement means once a pump is running, and the load-dependence that governs all of it | The physiology of the heart–circuit interaction itself: afterload, series-versus-parallel flow, the mixing point — Chapter 13, Chapter 16 |
| The pre-cannulation study, and the findings that change the configuration or stop the cannulation | Choosing the configuration — Chapter 3; indications and contraindications — Chapter 4, Chapter 11 |
| Cannula position as an imaged variable — how it is confirmed, why it drifts, and what a drifted cannula looks like | The technique of cannulation, including real-time guidance during insertion — Chapter 5, Chapter 12, Chapter 21 |
| The structured surveillance study on venovenous and venoarterial support: what is looked at, in what order, how often | Cannula malfunction as a circuit problem, and its correction — Chapter 34 |
| The turndown (flow-reduction) study: protocol, physiology, what it can and cannot establish | The weaning decision and the decannulation procedure — Chapter 10, Chapter 18 |
| The predictive performance of echocardiographic weaning parameters, and why three cohorts disagree | Recirculation estimation as a calculation — Appendix 5; recirculation physiology — Chapter 8 |
| Echocardiographic cross-check of a derived cardiac output (the obligation Chapter 25 created) | Left ventricular distension: mechanism, consequences and the decision to unload — Chapter 15 |
| Diagnosing tamponade, thrombus, new regurgitation, shunt and air on a circuit | Circuit thrombosis and oxygenator failure — Chapter 31, Chapter 32; bleeding — Chapter 40 |
| Who performs the study, how often, and what the report must state | Training and accreditation programmes as a whole — Chapter 88 to Chapter 90 |
The boundary that matters most is with Chapter 15 and Chapter 18. This chapter owns the image and the measurement. Chapter 15 owns what to do about a distended ventricle and Chapter 18 owns the decision to come off. A worked turndown study appears here because it is an echocardiographic examination with a protocol; the decision it feeds belongs to Chapter 18.
This is the organising idea of the chapter, and the ELSO Red Book states it in a single sentence that is almost universally ignored in practice.
<aside> 📖
Evidence · The Red Book's own instruction
From Chapter 48, on echocardiographic monitoring during venoarterial support: "Complex interactions between ventricular preload and afterload make it challenging to accurately measure intrinsic ventricular contractility; all parameters should be interpreted with caution and reported with the ECMO flows at the time of assessment."
Certainty: expert consensus, and mechanically compelling. It is an instruction about reporting, not about physiology, and it is the only sentence in the ECMO echocardiography literature that would change most units' practice if it were followed literally.
</aside>
The reason is straightforward once stated. Ejection fraction, velocity–time integral, annular systolic velocity and longitudinal strain are all load-dependent. Every one of them rises when afterload falls and falls when afterload rises; every one of them rises with preload within the ascending limb of the Frank–Starling relation. Peripheral venoarterial ECMO moves both variables simultaneously and in opposite directions: the drainage cannula removes preload, the return cannula adds afterload. The number produced is a joint function of contractility and two machine settings, and only one of the three is a property of the patient.
<aside> 🧠
Physiology · Why the same heart gives different numbers at different flows
Consider a left ventricle with fixed intrinsic contractility on peripheral venoarterial support.
Raise circuit flow. Retrograde aortic flow raises aortic root pressure, so systolic wall stress rises. Simultaneously, drainage removes venous return, so right ventricular output and hence transpulmonary flow fall, and left ventricular preload falls. Stroke volume falls from both ends. The measured velocity–time integral falls, ejection fraction falls, and annular velocities fall. Nothing has happened to the myocardium.
Lower circuit flow. Afterload falls, preload rises, stroke volume rises. The same heart now produces a better-looking study. Again nothing has happened to the myocardium.
The magnitude of that swing is the informative quantity. A heart with no contractile reserve barely moves between the two conditions because it cannot generate more work at lower afterload. A heart with reserve moves a great deal. The difference between two measurements is therefore a property of the patient in a way that either measurement alone is not.
The corollary for venovenous support is different and easier. Venovenous ECMO does not directly change left ventricular loading; it changes gas exchange, and through gas exchange it changes pulmonary vascular resistance and therefore right ventricular afterload. On venovenous support the load-dependence problem is a right-heart problem, not a left-heart one (§26.13).
</aside>
There is direct measurement of the swing. A prospective study of 46 adults on peripheral venoarterial ECMO imaged each patient by transthoracic echocardiography at 100%, 120% and 50% of target circuit flow, on day 1 and again around day 4 (Ng et al. 2022, DOI 10.1002/ehf2.13959). Left ventricular global longitudinal strain improved from −6.1% (interquartile range −8.9 to −4.0) at 120% of target flow to −8.8% (−11.5 to −6.0) at 50% of target flow, with p under 0.001.
The second finding is the one worth carrying to the bedside.
<aside> 📖
Evidence · The flow-sensitivity of the ventricle changes with the age of the injury
In the same 46-patient cohort, the percentage change in global longitudinal strain between 50% and 120% of target flow was 72.7% (26.8–100.0) in the acute phase (mean day 1) and 22.5% (14.9–43.8) in the delayed phase (mean day 4), p under 0.001. Ejection fraction behaved the same way (43.0% versus 22.8% change, p = 0.012), as did left ventricular outflow tract velocity–time integral (45.8% versus 24.2% change, p = 0.001). In the 28 patients considered potentially weanable, and in the 25 with ischaemic cardiomyopathy, the same pattern held.
Certainty: low to moderate. Single-centre, 46 patients, no outcome endpoint attached to the sensitivity measure itself. Read as a physiological observation, it is well made and internally consistent across three independent parameters.
Source: Ng PY, Ma TS, Ip A, et al. ESC Heart Failure 2022;9(5):3241–3253. DOI 10.1002/ehf2.13959.
</aside>
This has an interpretation that the paper does not spell out and that this book states as its own reasoning. Early after an acute myocardial insult the ventricle is operating on a steep part of its afterload–performance relation: small changes in load produce large changes in measured function. Days later, in survivors, that sensitivity falls by roughly two-thirds. A ventricle whose measured function no longer swings much with flow is behaving differently from one that does — and the direction of that difference is not obvious. It may mean recovery of an intrinsically more robust contractile state; it may mean a ventricle so damaged it can no longer respond to unloading at all. The study cannot distinguish those two, because it reports the sensitivity and not what happened to the patients grouped by it. That is an open question of real practical importance and §26.17 records it as such.
<aside> ⚠️
Danger · An echocardiogram report on ECMO that does not state the circuit flow is uninterpretable and should be returned
This is not a stylistic preference. Two reports on the same patient on the same day reading "LVEF 20%" and "LVEF 40%" are entirely compatible with each other if one was performed at 4.5 L/min and the other at 2 L/min — and the 20% study is the one that looks worse while describing the better-supported patient. Serial trends assembled from studies performed at undocumented and varying flows are not trends; they are noise with a plausible shape.
Minimum data on every ECMO echocardiogram report: circuit blood flow in L/min at the time of imaging, sweep gas flow, pump speed, whether the patient was on the same vasoactive infusions as the previous study, and — for a turndown study — the flow at each stage and the duration held.
</aside>
<aside> 💡
Clinical pearl · Ask for a comparison, not a number
The useful request to the echocardiographer on a patient supported by venoarterial ECMO is almost never "what is the ejection fraction". It is one of:
All three are answerable. "What is the ejection fraction" is answerable only with a flow attached, and is rarely the question that was meant.
</aside>