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?
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.
What this chapter covers — and what it does not
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.
26.1 On ECMO there is no ejection fraction — only an ejection fraction at a stated flow
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.
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.
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.
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).
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.
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.
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.
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.
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:
- "Is the aortic valve opening, at this flow?" — a yes/no observation that is robust to everything in this section (§26.6).
- "What changes when we come down from 4 L/min to 2 L/min?" — a comparison, which is the only form in which contractile reserve can be measured (§26.11).
- "Is there a structural explanation for the problem in front of me?" — tamponade, thrombus, new regurgitation, a malpositioned cannula, a shunt (§26.7, §26.8, §26.15).
All three are answerable. "What is the ejection fraction" is answerable only with a flow attached, and is rarely the question that was meant.
26.2 Three different instruments share one probe
Echocardiography on ECMO is used for three distinct purposes, and it is worth separating them because its reliability is very different in each, and the literature is far stronger for the first two than for the third.
Role | The question | How well it performs | Failure mode |
Procedural tool | Where is the wire? Where is the cannula tip? Where is the return jet pointing? | Strong. Direct visualisation of a structure in real time. Better than the alternative in every published comparison, because the alternative — the chest radiograph — physically cannot show a radiolucent tip | Acoustic windows; operator skill; cannula artefact obscuring adjacent structures |
Structural monitor | Is there a new effusion, thrombus, regurgitation, shunt, air? Is the ventricle distending? Is the aortic valve opening? | Strong for presence-or-absence questions. These are the findings that change management most often and are the least load-dependent | The circuit masks the haemodynamic consequence, so the finding appears before the physiology does (§26.8) — which is an argument for imaging, not against it |
Decision instrument | Is this heart ready to come off? Will this patient survive decannulation? | Weak, and the published cohorts disagree with one another (§26.12). The parameters are load-dependent, measured under partial support, and validated in populations selected by the decisions being predicted | Verification bias: patients who fail the criteria are frequently not offered a weaning trial, so the criteria cannot be seen to be wrong in the direction that matters |
Pitfall · Treating a weaning parameter with the confidence of a cannula position
The same study, performed by the same operator with the same machine, generates a cannula position that is a fact and a velocity–time integral that is a weak prior. Both appear in the same report, in the same font. The confidence attached to them should differ by a wide margin, and in most units it does not.
26.2.1 Transthoracic or transoesophageal
The two ELSO documents take compatible positions, and the practical rule is simple.
Situation | Preferred | Why |
Routine daily surveillance, adequate windows | Transthoracic | Non-invasive, repeatable, no sedation escalation, no oesophageal instrumentation in an anticoagulated patient. Red Book Chapter 48: transthoracic views give adequate echogenicity in most cases, especially when tidal volumes are minimal — lung-rest ventilation improves the windows |
Guiding a dual-lumen bicaval cannula without fluoroscopy | Transoesophageal — effectively mandatory | ELSO 2015 ultrasound guidance states that for a bicaval dual-lumen cannula it is necessary to use transoesophageal imaging to guide cannulation if an image intensifier is not being used |
Guidewire visualisation during cannulation in a ventilated adult | Transoesophageal | Red Book Chapter 48: visualisation of guidewires is enhanced on transoesophageal compared with transthoracic imaging in mechanically ventilated adults |
The inferior vena cava–right atrial junction | Transthoracic subcostal, even mid-transoesophageal-study | Red Book Chapter 48 states this junction may be challenging to visualise on transoesophageal imaging and that switching to the transthoracic subcostal view may give better visualisation of the cannula in the inferior vena cava. The two modalities are complementary within a single examination, not alternatives |
Suspected apical thrombus | Transthoracic | Transthoracic imaging may provide better apical imaging than transoesophageal in some instances and can be used for quick verification |
Cardiac arrest with compressions in progress | Transoesophageal | Continuous imaging without interrupting compressions (§26.14) |
Poor transthoracic windows and a question that must be answered | Transoesophageal | ELSO 2015: transoesophageal should be considered if transthoracic views are suboptimal |
Clinical pearl · Lung rest is an imaging intervention
The single commonest reason transthoracic windows are poor in a ventilated patient is lung between the probe and the heart. Ultraprotective ventilation on ECMO reduces that problem. Patients on ECMO frequently have better transthoracic windows than comparably ill patients who are not — which means the reflex reach for the transoesophageal probe is often unnecessary, and in an anticoagulated patient that matters.
26.3 The study before the cannula: five findings that change the plan
The pre-ECMO echocardiogram is the highest-yield study in the whole run, because it is the only one performed while the patient's own physiology is still visible, and because its findings can change the configuration, change the cannula, or stop the procedure.
Finding | What it changes | Source |
Moderate or severe aortic regurgitation | Argues strongly against peripheral venoarterial ECMO. Retrograde aortic flow drives regurgitant volume into a ventricle that cannot eject, producing distension, stasis and pulmonary oedema within hours. The ELSO 2015 ultrasound guideline states that detection of severe or moderate-to-severe aortic regurgitation precludes the patient from venoarterial ECMO and that it is very important to exclude this before initiation | ELSO 2015; ELSO Red Book Chapter 27 contraindication table (Chapter 11) |
Severe left ventricular dysfunction in a patient being considered for venovenous support | Changes the configuration. The ELSO 2015 guideline gives this as its worked example: detection of severe left ventricular dysfunction in a patient with severe respiratory failure may help in determining the best mode of support | ELSO 2015 |
Acute cor pulmonale, or proximal pulmonary artery thrombus in suspected massive pulmonary embolism | Confirms the mechanism, and in the embolism case identifies a target for definitive therapy alongside support | ELSO 2015 |
Intracardiac thrombus | Changes the cannulation route. The Red Book's Figure 48-1 records exactly this decision: two right atrial thrombi at the superior vena cava–right atrial junction led to a femoro-femoral venovenous configuration to avoid dislodging them during cannulation | ELSO Red Book Chapter 48 |
Intracardiac shunt (patent foramen ovale, atrial septal defect); mitral regurgitation; right ventricular dimensions and tricuspid regurgitation; inferior vena cava dimensions | Affects feasibility and predicts the complications to watch for. The European consensus statement lists intracardiac shunts among the concomitant disorders affecting the feasibility of venoarterial ECMO, and left atrial dimensions as a factor guiding the choice between left atrial and left ventricular venting cannulas | Tops et al. 2025, DOI 10.1002/ejhf.70018 |
Clinical pearl · Vessel sizing is part of the pre-ECMO ultrasound and is not echocardiography
The ELSO 2015 guideline gives a sizing rule for peripheral cannulation: French gauge of cannula = 3 × the vessel diameter in millimetres. That is the vascular study, performed with a linear probe on the groin or neck, and it belongs with the cannulation chapters (Chapter 5, Chapter 12, Chapter 21). It is recorded here only because in practice the same operator does both studies in the same five minutes, and the cardiac study is the one that gets skipped when things are urgent.
Evidence · The pre-cannulation study also carries prognostic information, weakly
In 21 consecutive patients with refractory acute respiratory distress syndrome imaged immediately before venovenous cannulation, survivors had higher pre-cannulation left ventricular ejection fraction (56 ± 12% versus 46.9 ± 11%, p = 0.02), higher tricuspid annular plane systolic excursion (16.9 ± 3.7 mm versus 13 ± 4.5 mm, p = 0.04) and lower systolic pulmonary artery pressure (47.7 ± 5.8 mmHg versus 57 ± 6.7 mmHg). Overall mortality was 57.1%. Acute cor pulmonale was present in only 2 of 21 (9.5%), and both died.
Certainty: very low. Twenty-one patients, single centre, a cohort pre-selected for the presence of pulmonary vascular dysfunction (17 patients with normal biventricular function and 8 with isolated left ventricular dysfunction were excluded from the analysis before these comparisons were made). The selection makes the percentages uninterpretable as prevalence figures. A numerical discrepancy is recorded and not corrected: the paper's abstract gives p = 0.02 for the systolic pulmonary artery pressure comparison and its own results table gives p = 0.004 for the same comparison. Both are printed here; neither is asserted. The direction — that a worse right heart before cannulation predicts worse outcome after it — is consistent with the wider literature and is the only claim this book takes from it.
Source: Lazzeri C, Cianchi G, Bonizzoli M, et al. Acta Anaesthesiologica Scandinavica 2015;60(4):485–491. DOI 10.1111/aas.12643.
26.4 During cannulation: what the probe is for
Chapter 5, Chapter 12 and Chapter 21 own cannulation technique. This section records only what the imaging contributes, because the cannula positions established here are the baseline against which every subsequent surveillance study is read.
Physiology · Why wire position is the safety-critical observation, not cannula position
The cannula follows the wire. A wire that has coiled in the right atrium, crossed the interatrial septum or entered a hepatic vein will take a large-bore dilator and then a large-bore cannula with it. The Red Book records all three failure modes explicitly, and notes that wires looped in the right atrium or right ventricle may cause dysrhythmias and increase the risk of cardiac perforation.
The consequence: the wire is watched continuously through every dilation, not checked once before them. The Red Book's phrasing for the dual-lumen bicaval cannula is that the wire should be far into the inferior vena cava beyond the takeoff of the hepatic veins and should remain straight while the cannula is inserted — a wire that buckles during insertion is the warning, and it is only visible if someone is looking at that moment.
Configuration | Target position | Views |
Two-site venovenous | Both wires ideally in both venae cavae so that they do not abut a cardiac chamber. Final cannula position maximises drainage while avoiding clinically relevant recirculation | Transoesophageal mid-oesophageal bicaval and inferior vena cava views; transthoracic subcostal, right ventricular inflow and inflow–outflow views |
Femoro-jugular venovenous | The drainage cannula may be placed within the right atrium and its position reassessed after ventilatory adjustment, because loss of lung volume on a rest strategy shifts the diaphragm upward and moves the cannula. The reinfusion cannula may not always be visualised in the superior vena cava but should be a few centimetres above the superior vena cava–right atrial junction | As above |
Femoro-femoral venovenous | Drainage cannula within the right atrium, single-stage reinfusion cannula mid-right atrium. Adequate flows with minimal recirculation as long as sufficient distance is maintained between the proximal drainage hole and the tip of the reinjection cannula | As above |
Dual-lumen bicaval | Wire far into the inferior vena cava beyond the hepatic veins and straight during insertion. Reinjection jet in the right atrium and directed towards the tricuspid valve. Alternating between mid-oesophageal bicaval and inferior vena cava views is required | Transoesophageal bicaval, modified bicaval and proximal inferior vena cava; transthoracic subcostal sagittal bicaval |
Right atrium to pulmonary artery cannula | Wire followed all the way to the pulmonary artery. Cannula tip 2 to 3 cm above the pulmonary valve, and not selective into a pulmonary artery branch | Mid-oesophageal right ventricular inflow–outflow; mid-oesophageal pulmonary valve; upper-oesophageal pulmonary artery view |
Venoarterial, drainage side | Same principles as venovenous. Recirculation is not a concern, so the drainage cannula can be advanced across the right atrium and into the superior vena cava | As above |
Venoarterial, return side | The arterial wire should be seen in the aorta. The reinfusion cannula itself is usually not seen on echocardiography but may be seen on pelvic ultrasound in the femoral or iliac artery. Check for the absence of aortic dissection after cannulation | Transoesophageal descending aorta; transthoracic subcostal descending aorta |
Neonatal or paediatric carotid arterial return | Optimal placement at the junction of the right brachiocephalic artery and the ascending aorta. A cannula protruding into the aorta directs proportionally more flow toward the aortic root, reducing aortic valve opening, impeding left ventricular decompression and raising left atrial pressure. A cannula placed higher in the brachiocephalic or carotid artery risks dislodgement | Parasternal arch view (Red Book Figure 48-13) |
Danger · Two observations that must be made after every cannulation, in every configuration
The Red Book states both as universal: "For all cannulations, the absence of a new pericardial or pleural effusion should be ascertained postprocedure, and the size of a pre-existing effusion must be reassessed."
A new effusion after cannulation means a chamber or vessel has been breached. On ECMO the haemodynamic consequence may not appear for hours (§26.8), and by then anticoagulation has been running. The post-cannulation pericardial view is a five-second observation that is frequently omitted because the patient is now stable and everybody is relieved.
Pitfall · Cannula artefact hides what you most want to see
Both wires and cannulas generate significant acoustic scattering and artefact, and the Red Book names this as a reason why adequate echocardiography training is mandatory to safely guide ECMO cannulation and why rapid switching between views is needed. The practical consequence: a structure lying immediately deep to a cannula is the structure least likely to be assessed. Thrombus on the far side of a drainage cannula, and the interatrial septum behind a dual-lumen cannula, are the two commonest casualties.
26.5 Cannula position is a variable, not a fact
This is the second structural claim of the chapter, and it follows from a physical property of the equipment.
Physiology · The chest radiograph cannot do this job and never could
The distal end of an ECMO cannula is radiolucent. The ELSO 2015 guideline states it directly: evaluating the position of the cannula tip by echocardiography is "particularly important, as it is difficult to visualise the radiolucent distal end of the ECMO cannula on radiography." The Red Book says the same of surveillance: cannula position should be reassessed with echocardiography in case of lower flows, "as x-rays are not as sensitive to locate the cannula position relative to the cardiac chambers."
The radiograph shows the radio-opaque marker, which is not the tip, and it shows it in two dimensions against a mediastinal silhouette, which does not localise it relative to the cavo-atrial junction or the tricuspid valve. A daily chest film is not cannula surveillance. It has never been cannula surveillance.
Two independent observations establish that positions drift and that the drift is common.
Evidence · Nearly half of dual-lumen cannula port positions were wrong when someone looked
A retrospective review of one year's transthoracic studies performed to assess bicaval dual-lumen catheter position found that a subcostal sagittal bicaval view identified all three catheter ports in 11 of 11 studies, and that at least one of the catheter ports was incorrectly positioned in 5 of 11 studies.
The same report cites an animal experiment in which recirculation rose from 2.1% to 51.3% when the catheter tip was pulled back from the inferior vena cava into the right atrium in a single sheep, and notes an Extracorporeal Life Support Organization Registry prevalence of 9% for cannula placement complications.
Certainty: very low for the 5-of-11 proportion — eleven studies, single centre, retrospective, and the studies were performed because someone wanted to know, which selects for suspicion. The authors also record that image quality improved dramatically over the year and that the process was heavily operator-dependent.
What survives the caveats is the direction and the mechanism, which the Red Book independently confirms: ports move, the radiograph does not show it, and the consequence is a recirculation fraction that can rise by an order of magnitude. The single-sheep recirculation figures are quoted here as the authors quote them and should not be treated as a human effect size.
Source: Tabak B, Elliott CL, Mahnke CB, Tanaka LY, Ogino MT. Journal of Clinical Ultrasound 2012;40(3):183–186. DOI 10.1002/jcu.21873.
Evidence · The Red Book's own drift case: reinjection into the liver
Red Book Chapter 48 describes a patient whose reinjection flow became intrahepatic. Comparison of two chest radiographs showed a significant loss in lung volumes and consequent elevation of the diaphragm. The relative position of the reinjection hole with respect to the cardiac chambers changed, so that reinjection occurred within the liver instead of the right atrium. The cannula was withdrawn under transoesophageal guidance until the reinjection was seen in the right atrium.
The chapter states the consequences elsewhere in the same section: loss of lung volume may cause the reinjection flow to become intrahepatic, "thus causing inefficient oxygenation and elevation of liver enzymes."
Certainty: a single illustrated case in a primary textbook, with a clear mechanism and a documented correction. Read as a mechanism, not a frequency.
Pitfall · Three things that move a cannula after it is fixed at the skin
- Loss of lung volume. A rest ventilation strategy reduces lung volume, the diaphragm rises, and the heart and great veins move cranially relative to a cannula fixed at the groin. The cannula did not move; the patient did. This is why the Red Book advises placing the femoral drainage cannula within the right atrium initially and reassessing position after ventilatory adjustment.
- Head and neck movement. For jugular cannulas, including dual-lumen bicaval cannulas, moving the patient or the patient's head can displace the cannula.
- Proning, turning and transfers. Not separately evidenced in the sources used here, and stated as a mechanical inference from the first two.
The operational rule: a cannula position confirmed at insertion is a baseline, not a permanent finding. Re-image after any change in ventilation strategy, after proning, and whenever flow falls.
26.5.1 The malposition syndrome
A malpositioned cannula does not usually announce itself as a position problem. It announces itself as one of four other problems, and the value of imaging is that it collapses a differential.
Presentation | Malposition mechanism | What to look for |
Access insufficiency — chatter, negative drainage pressures, flow limitation | Drainage holes abutting a vessel or atrial wall, or the tip advanced too far so that side holes are outside the effective drainage compartment | Tip and side-hole position relative to the cavo-atrial junction; wall apposition on colour Doppler; and the differential in §26.7 — this is also how hypovolaemia and tamponade present |
Refractory hypoxaemia on venovenous support with good circuit flows | Recirculation: drainage and return ports too close, or return jet directed back toward the drainage port rather than across the tricuspid valve | Colour Doppler of the return jet direction; distance between the proximal drainage hole and the return tip; the jet should cross the tricuspid valve |
Rising liver enzymes with unexplained inefficiency of oxygenation | Intrahepatic reinjection after loss of lung volume (the Red Book case above) | Subcostal and transoesophageal inferior vena cava views tracking where the reinjection jet actually lands |
Hepatic venous congestion | Dual-lumen cannula malposition producing massive retrograde hepatic venous flow — a published case in which rapid echocardiographic diagnosis was described as pivotal in preventing potentially fatal complications | Hepatic vein Doppler; direction of flow in the hepatic veins on subcostal imaging |
Clinical pearl · Colour Doppler is the cannula-position test, not two-dimensional imaging
Two-dimensional imaging shows where the cannula is. Colour Doppler shows where the blood goes, which is the question that actually matters. The Red Book is explicit that colour Doppler is helpful in visualising the reinjection flow direction of a dual-lumen bicaval cannula on the transthoracic right ventricular inflow, inflow–outflow and subcostal views, and on the transoesophageal mid-oesophageal bicaval, modified bicaval and proximal inferior vena cava views.
A cannula in an anatomically correct position with a jet pointing the wrong way is a malfunctioning cannula. Only colour tells you.
Worked example · Femoro-femoral venovenous ECMO with rising recirculation
A published case describes a patient transferred on femoro-jugular venovenous support who required cannula exchange and conversion to a femoro-femoral configuration. The drainage cannula was placed at the right atrial–inferior vena cava junction and the return cannula in the inferior vena cava. This configuration produced significant recirculation. The return cannula was replaced with a long, 50 cm cannula with integrated side holes — a cannula normally used for drainage — and recirculation was reduced by repositioning the return tip into the right atrium, with oxygenation maintained.
The reasoning generalises beyond the case. In a femoro-femoral configuration both cannulas approach from below, so the geometry that separates drainage from return in a femoro-jugular configuration does not exist and must be manufactured by cannula length and side-hole distribution. The authors' conclusion is that in femoro-femoral venovenous ECMO, selection of the return cannula considering its length and the position of its side holes, and the decision about where the tips sit, are what determine recirculation.
Certainty: very low — a single case report. It is included because the physical argument is transparent and because it demonstrates the general principle that recirculation is a geometry problem that echocardiography can see and the circuit console cannot.
Source: Wada K, Shimizu K, Narimatsu E, et al. Journal of the Japanese Association for Acute Medicine 2020;31(12):2449–2454. DOI 10.1002/jja2.12504.
26.6 The venoarterial surveillance study: lead with the aortic valve
Across this book, the aortic valve has turned out to be the highest-yield single observation on venoarterial support — it appears in Chapter 13 §13.9, in Chapter 15 §15.3, and in Chapter 18 §18.5.2 with an area under the curve of 0.88 for a weaning-related endpoint. This chapter leads the surveillance study with it, for three reasons that are worth stating explicitly.
Physiology · Why aortic valve opening is the most robust echocardiographic observation on venoarterial ECMO
- It is binary and therefore load-independent in the way that matters. Every quantitative parameter in §26.1 shifts with circuit flow. "Does the valve open" also shifts with flow — but it shifts as a stateful, visible, reportable yes or no, which survives inter-observer variation, poor windows and inexperienced operators in a way that a velocity–time integral does not.
- It is the direct expression of the central mechanical problem. Peripheral venoarterial ECMO adds afterload; the ventricle must generate a systolic pressure exceeding aortic root pressure to open the valve. Failure of the valve to open is the definition of the failing side of that contest — the Red Book's Chapter 48 phrasing is that failure of the valve to open indicates the ventricle cannot generate enough power to overcome the higher afterload seen on peripheral venoarterial support.
- The consequences of a closed valve are the ones that kill. Stasis in the left ventricle, left atrium, pulmonary veins and aortic root leads to thrombus; rising end-diastolic pressure raises wall stress and lowers the transcoronary perfusion gradient; left atrial pressure rises and produces pulmonary oedema. The Red Book records that this can progress to complete thrombosis of either ventricular cavity.
Order | Observation | Why here |
1 | Does the aortic valve open, and on how many beats? Report as opens every beat / intermittently (and how often) / not at all | The single highest-yield finding. Determines whether the rest of the study is a routine surveillance or an unloading discussion |
2 | Spontaneous echo contrast in the left ventricle, left atrium, aortic root. Thrombus in any chamber, in the aortic root, and around cannulas | Stasis precedes thrombus, and contrast is visible before thrombus is. The Red Book couples the two findings directly: "Consistent loss of pulsatility and a significant amount of spontaneous echo contrast, despite inotropic support and adequate fluid loading, may prompt insertion of a venting device." The 2022 imaging review adds that spontaneous contrast should prompt reappraisal of the anticoagulation target |
3 | Left ventricular size, compared with the previous study, at a stated flow | Necessary but insufficient — see the danger callout below |
4 | Aortic regurgitation: present, and worse than before? | Peripheral venoarterial support unmasks and worsens aortic regurgitation. The 2022 review documents this with paired imaging: trace early-diastolic regurgitation before initiation became mild holodiastolic regurgitation after it. Regurgitation converts a closed valve from a stasis problem into a distension problem |
5 | Mitral regurgitation, and left atrial size | Also exacerbated by the afterload rise, producing left atrial distension, arrhythmia and pulmonary oedema |
6 | Pericardial effusion — present, new, size, and loculation | §26.8. On a circuit the haemodynamic signs are abolished, so the image is the diagnosis |
7 | Right ventricle: size, right-to-left ventricular ratio, free-wall function, tricuspid regurgitation velocity, septal position | §26.9 — and Chapter 25's finding makes this the discriminating observation, not an afterthought |
8 | Cannula positions and jet directions (§26.5) | Cheap to add once the probe is on the patient, and the finding that most often explains a flow problem |
9 | Quantitation: left ventricular outflow tract velocity–time integral, ejection fraction, annular velocities — each recorded with the flow at which it was obtained | Last, deliberately. These are the load-dependent numbers of §26.1 and they are the least robust part of the study |
Danger · Left ventricular chamber size is an insensitive indicator of distension
The 2022 imaging review states the objection precisely: the left ventricular end-diastolic pressure–volume relationship is non-linear, and the ventricle is contained within an intact pericardium. Therefore "LV chamber size as determined by TEE may be an insensitive indicator of ventricular distention, end-diastolic pressure, and left atrial pressure and this limitation should be recognized by the echocardiographer."
A stiff ventricle inside a stiff pericardium can reach a catastrophic end-diastolic pressure with a barely perceptible change in diameter. A normal-looking left ventricle does not exclude distension physiology. The findings that do the work are the composite: absent aortic valve opening, plus spontaneous echo contrast, plus systolic and diastolic mitral regurgitation, plus rising left atrial size, plus a clinical picture of pulmonary oedema.
Source: Hussey PT, Mering G, Nanda NC, Ahmed MI, Addis DR. Echocardiography 2022;39(2):339–370. DOI 10.1111/echo.15266.
26.6.1 How often
The sources do not agree, and the disagreement is worth showing rather than resolving.
Source | Recommended frequency | Certainty |
2026 diagnostic imaging review | "Regular echocardiographic evaluations should be conducted at least twice a day and more frequently during periods of unforeseen hemodynamic instability" for venoarterial patients | Expert opinion, narrative review, no supporting study cited |
European multimodality imaging consensus (temporary mechanical circulatory support section) | "Daily echocardiography, or more frequent checks during haemodynamic changes" | Clinical consensus statement; graded as consensus rather than evidence |
ELSO Red Book Chapter 48 | "Daily echocardiographic assessment paired with perfusion parameters" allows optimisation of ECMO and inotropic support while monitoring for recovery | Expert consensus |
ISCCM Manual Chapter 33 | Structures the assessment by stage — prior to initiation, after initiation, during maintenance, during recovery — rather than by clock interval | Expert consensus |
Clinical pearl · Frequency is the wrong variable — stage and trigger are the right ones
The ISCCM framing is the more useful of the two, and this book adopts it. Image by stage and by trigger, with a daily minimum:
- Daily minimum on venoarterial support: the nine-item list above, at a documented flow.
- Immediately on any of these triggers: a fall in circuit flow; loss or return of pulsatility; new or worsening pulmonary oedema; new hypoxaemia; a fall in mean arterial pressure not explained by the pump; a rise in lactate; a rising central venous pressure; new haemolysis; any change in ventilation strategy; after proning or repositioning; before and during every weaning trial.
- Twice daily in the first 48 hours, when the ventricle is at its most flow-sensitive (§26.1) and cannulas have not yet settled.
"Twice a day forever" produces studies nobody reads. A triggered study is acted on.
26.7 The low-pulsatility differential — the discrimination Chapter 25 deferred
Chapter 25 established the single most practice-changing finding in Part V so far: a low arterial pulse pressure most often does not mean left ventricular distension. In a study combining 58 pigs and 128 patients, three phenotypes emerged — pulsatile 26%, low pulsatility 52%, distension 22% — and the commonest was low pulsatility without distension, driven by right ventricular failure. Unloading that phenotype removes preload from a ventricle that is already underfilled.
Chapter 25 could identify the problem and could not solve it, because the discriminating variables are structural. This is the echocardiographic job that most changes management.
Cause of low pulsatility | Left ventricle | Right ventricle | Other | Action |
Adequate support of a recovering heart, flow simply too high | Normal or small; aortic valve opens intermittently | Normal or small | No effusion; no contrast; good perfusion markers | Reduce circuit flow and re-image. The commonest reversible cause and the cheapest test |
True left ventricular distension | Dilated, poorly contracting; valve does not open; spontaneous contrast; systolic and diastolic mitral regurgitation; left atrium enlarging | Variable | Pulmonary oedema; rising left-sided filling pressures | Chapter 15 owns this. Reduce flow, add inodilator, ensure anticoagulation is adequate; vent if those fail |
Right ventricular failure with an underfilled left ventricle | Small, underfilled, hyperdynamic or normal. Valve may not open because there is nothing to eject | Dilated; right-to-left ratio over 1; poor free-wall function; septal flattening; significant tricuspid regurgitation | Pulmonary artery diastolic pressure and pulmonary artery pulsatility low (Chapter 25) | Do not unload. Unloading empties an already empty ventricle. Treat the right heart; consider configuration change |
Hypovolaemia or haemorrhage | Small, hyperdynamic, end-systolic cavity obliteration | Small | Small collapsible inferior vena cava (if not obscured by the drainage cannula); access insufficiency and chatter; falling haemoglobin | Volume; find the bleeding |
Tamponade | Small, underfilled | Compressed — but see §26.8: this sign is frequently absent on ECMO | Effusion, which may be loculated and posterior. Rising central venous pressure; access insufficiency; falling flow late | Drain. §26.8 |
Cannula malposition causing access insufficiency | Underfilled | Underfilled | Cannula tip and jet abnormal on imaging; chatter; negative drainage pressures | Reposition under imaging (§26.5) |
Severe aortic regurgitation | Dilating with each beat; valve may appear to open in diastole rather than systole; holodiastolic regurgitant jet | Variable | Rapidly worsening pulmonary oedema | A distension emergency. A published case required left atrial venoarterial ECMO and emergency valve-in-valve intervention |
Clinical pearl · The left ventricle tells you which of six problems this is, and it does it with two observations
Is the left ventricle big or small? That single question splits the differential in half. Big and not ejecting points to distension, aortic regurgitation, or a heart that has not recovered. Small and not ejecting points to right ventricular failure, hypovolaemia, tamponade or cannula malposition — all four of which are made worse by unloading the left ventricle.
Then look at the right ventricle. Big right ventricle with small left ventricle is the Chapter 25 phenotype: 52% of low-pulsatility patients, and the one where an unloading device causes harm.
Two chambers, ten seconds, and the answer is different from the one the arterial waveform suggested.
Pitfall · Inferior vena cava assessment is degraded on ECMO and often uninterpretable
The classic volume-status surrogate is compromised on a circuit for three independent reasons: a large-bore drainage cannula sits inside the vessel being measured and both obstructs the view and alters the flow; the drainage cannula actively lowers inferior vena cava pressure and diameter irrespective of intravascular volume; and on ultraprotective ventilation the respiratory variation that drives the measurement is minimal.
An inferior vena cava measurement on venovenous or venoarterial ECMO is not a volume-status measurement. The chamber sizes in the table above are more reliable than it is.
Worked example · A rising lactate with an unchanged flow and a falling pulse pressure
Presentation. Day 3, peripheral venoarterial ECMO for fulminant myocarditis. Circuit flow 4.1 L/min, unchanged. Mean arterial pressure 68 mmHg, unchanged. Pulse pressure has fallen from 22 to 9 mmHg over six hours. Lactate has risen from 1.8 to 3.4 mmol/L. Central venous pressure has risen from 9 to 16 mmHg. Drainage line is chattering intermittently.
The reflex. Low pulse pressure, myocarditis, day 3 — the ventricle has stopped ejecting and needs unloading.
The study, in order.
- Aortic valve: not opening. Consistent with the reflex.
- Spontaneous echo contrast: present in the left ventricle. Also consistent.
- Left ventricle: small, underfilled, end-systolic cavity obliteration. Not consistent. A distending ventricle is not small.
- Right ventricle: not dilated.
- Pericardium: new circumferential effusion, moderate, with a posterior loculated component that was not present on yesterday's study.
Interpretation. The valve is not opening because there is nothing to eject. The ventricle is small because filling is obstructed. The rising central venous pressure and the chatter are the drainage cannula reporting the same obstruction. This is tamponade, and the pulse pressure and the aortic valve gave exactly the answer that would have led to an unloading device being placed into an empty ventricle.
The discriminating observation was chamber size, and it took ten seconds.
This case is constructed to illustrate the reasoning in this section and the physiology in §26.8. It is not drawn from a published report.
26.8 Tamponade on a circuit: every sign is abolished except the picture
Tamponade after cannulation or after extracorporeal cardiopulmonary resuscitation is the single clearest example of a general principle in this book: ECMO does not prevent complications, it conceals them by supplying the physiology they would otherwise remove.
Physiology · Why the whole tamponade examination fails on ECMO
Pulsus paradoxus and its echocardiographic equivalent fail. The Red Book states that on venovenous ECMO, impaired venous return from pericardial tamponade has the same haemodynamic consequences as without ECMO, but pulsus paradoxus and its echocardiographic equivalent — transmitral and transtricuspid respiratory variations — may be missing in the case of minimal tidal volumes. Ultraprotective ventilation removes the respiratory swing in intrathoracic pressure on which the sign depends.
Chamber collapse fails. The Red Book: right atrial compression may not always be evident, as ECMO flow may be initially maintained. And on venoarterial support, "the RV may be underfilled and collapsed from the ECMO drainage alone" — so the finding that would ordinarily be diagnostic has an alternative explanation that is present in every patient.
Systemic haemodynamics fail. On venoarterial support, organ perfusion and haemodynamics may be initially maintained because the pump is generating the circulation the tamponade would otherwise obstruct. Loss of pulsatility may occur, while drops in flow may be delayed, especially for loculated effusions.
What is left is the image of the effusion itself, and the serial comparison. The Red Book's conclusion is that serial echocardiography may be needed to monitor the evolution and size of the effusion.
Evidence · Delayed tamponade after extracorporeal cardiopulmonary resuscitation, hidden for ten hours
A published case describes a man in his sixties who arrested from contrast-related anaphylaxis and was placed on venoarterial ECMO. Echocardiography immediately after cannulation showed no pericardial effusion. Circuit flow was about 2 L/min, systolic blood pressure 80–100 mmHg, central venous pressure 9–11 mmHg, with heparin maintaining an activated clotting time of about 150 seconds.
Ten hours later blood pressure fell to 50 mmHg and central venous pressure rose remarkably to 35 mmHg. Repeat echocardiography showed a pericardial effusion. Percutaneous drainage of 450 mL of bloody effusion returned central venous pressure to 9–11 mmHg and blood pressure to 80–100 mmHg. The effusion had a haemoglobin of 11 g/dL and a partial pressure of oxygen of 24.8 mmHg, indicating venous blood. He also had right fourth to eighth rib fractures from compressions, and was weaned from ECMO on day 3.
The authors' own explanation is the instructive part: the bleeding was probably trivial initially and was accelerated by anticoagulation, and "the progression of cardiac tamponade escaped our notice at the early stage because circulation was maintained with ECMO; it became apparent only with circulatory collapse."
Certainty: very low — a single case report. It is quoted because it documents the exact time course and the exact mechanism that the Red Book describes in the abstract, and because the oxygen tension of the drained fluid identifies the source as venous.
Source: Tsukada Y, Nemoto C, Ikegami Y, Suzuki T, Tase C. Acute Medicine and Surgery 2015;3(2):214–215. DOI 10.1002/ams2.128.
Danger · A rising central venous pressure on ECMO is a tamponade sign until imaged
In the case above the central venous pressure went from 9–11 to 35 mmHg while the pump held the blood pressure up. On a supported circulation the filling pressure moves before the output does, because the pump is holding the output constant. The instrument that fails last is the one that fails silently, and the instrument that moves first is the one nobody is watching.
Any of the following, on any configuration, is an indication for an immediate pericardial view: a rising central venous pressure; new or worsening access insufficiency with an unchanged cannula position; a fall in circuit flow at an unchanged pump speed; loss of pulsatility that is new; and any unexplained deterioration in the first 24 hours after cannulation or after extracorporeal cardiopulmonary resuscitation, where the resuscitation itself is a cause.
Post-arrest rib fractures, guidewire and cannula trauma, and full anticoagulation are three mechanisms acting simultaneously in the same patient.
26.9 The right ventricle: the chamber the circuit unloads, and the chamber the circuit hides
Both configurations change right ventricular loading, in opposite directions, and in both the standard right ventricular measurements become harder to interpret than the left ventricular ones.
Physiology · What each configuration does to the right ventricle
Venovenous ECMO unloads the right ventricle by removing its afterload. Correcting hypoxaemia and hypercapnia relieves hypoxic pulmonary vasoconstriction and acidosis-driven pulmonary vasoconstriction; reducing tidal volume and driving pressure relieves the mechanical component. The right ventricle frequently improves within hours of initiation and the improvement is not myocardial recovery — it is afterload relief. The Red Book's reference list for Chapter 48 includes a paper titled specifically on right ventricular unloading after initiation of venovenous ECMO.
Venoarterial ECMO unloads the right ventricle by removing its preload. The drainage cannula takes venous return before it reaches the right heart. A small, quiet right ventricle on venoarterial support is the expected appearance and carries almost no information about right ventricular contractility — which is precisely why the Red Book warns that a collapsed right ventricle may be a drainage artefact rather than a tamponade sign (§26.8), and why weaning assessment at full flow cannot assess the right heart.
The Red Book states the limitation of every partial-support weaning study in one phrase: "full RV loading is not achieved for accurate RV assessment." That is the single strongest argument for the arteriovenous bridge technique in §26.11.
Parameter | What it means off a circuit | On ECMO |
Right-to-left ventricular ratio | Ratio over 1 suggests significant right ventricular strain | Retains value, and on venoarterial support is the discriminating measurement in the Chapter 25 phenotype (§26.7). A ratio over 1 with a small left ventricle is right ventricular failure, not distension. The 2026 imaging review states the ratio and its threshold explicitly |
Tricuspid annular plane systolic excursion | Longitudinal systolic function | Load-dependent and preload-sensitive; falls with the drainage that venoarterial support imposes. In the weaning literature it performed inconsistently: significant in one prospective cohort (odds ratio 1.51, 95% CI 1.13–2.01, p = 0.005) and not associated with successful right ventricular weaning in a systematic review cited by the 2022 review |
Right ventricular fractional area change | Global right ventricular systolic function | Associated with successful right ventricular weaning in the same systematic review, alongside smaller right ventricular size, higher right ventricular ejection fraction and higher right ventricular strain magnitudes |
Three-dimensional right ventricular ejection fraction | The most complete non-invasive right ventricular systolic measure | Red Book Chapter 29 gives over 25% as associated with successful weaning. Requires image quality and software that are not universally available |
Tricuspid regurgitation velocity | Estimates systolic pulmonary artery pressure | The 2026 review states that increased velocity correlates with higher pulmonary artery pressures and right ventricular afterload, and can be an early indicator of developing right ventricular failure. On venovenous support it is the parameter most likely to move with a change in sweep gas, because sweep changes carbon dioxide and carbon dioxide changes pulmonary vascular resistance |
Lateral tricuspid annular systolic velocity (right ventricular S′) | Longitudinal systolic velocity | Red Book Chapter 29 reports a single-centre finding that right ventricular S′ divided by right ventricular systolic pressure, over 0.33, measured at a median venoarterial flow of 3.2 L/min, was a better predictor than traditional parameters; and that a change of over 10% in right ventricular S′ when flow is reduced to 30–50% of baseline for 15 minutes has been proposed as more sensitive than left ventricular ejection fraction, velocity–time integral and mitral annular systolic velocity. The Red Book adds that most of these tests require further validation |
Severe tricuspid regurgitation | Right ventricular dilatation and annular dilatation | Its absence is among the main right ventricular parameters associated with successful weaning in Red Book Chapter 29 |
Pitfall · A patent foramen ovale that was silent can open when the right heart is reloaded
The 2022 review records the mechanism: if more intracardiac volume and preload reach the right ventricle because venous drainage has been reduced, and the right ventricle cannot tolerate the volume in combination with possibly increased pulmonary artery pressures, tricuspid regurgitation and a rise in right atrial pressure can open a patent foramen ovale and precipitate hypoxaemia through a right-to-left shunt.
This is a weaning complication specifically. The shunt appears at the moment flow is reduced, which is the moment the operator is least expecting new hypoxaemia and most likely to attribute it to the lungs. Agitated-saline contrast during a turndown study, not before it, is what demonstrates it.
26.10 The cross-check: the obligation Chapter 25 created
Chapter 25 established that every derived cardiac output on ECMO requires an independent cross-check, because the derivations rest on assumptions the circuit violates. Echocardiography is the available cross-check. This section states how it is done and where it in turn fails.
Physiology · The Doppler stroke-volume method survives the circuit better than any other cardiac-output technique — and measures something narrower than it appears to
The left ventricular outflow tract velocity–time integral, multiplied by outflow tract cross-sectional area, gives the stroke volume ejected through the aortic valve, and multiplied by heart rate gives the flow contributed by the native heart. Two properties make it robust on ECMO:
- It does not depend on an indicator traversing the circulation. Thermodilution fails on venoarterial ECMO because the indicator is partly drained into the circuit before it reaches the thermistor. Doppler measures the velocity of blood at a specific anatomical location and does not care what happens to it afterwards.
- It measures the quantity that is actually in question. On venoarterial support the clinically important unknown is not total systemic flow — the console displays circuit flow, and total flow is circuit flow plus native output. The unknown is the native contribution, and the outflow tract velocity–time integral measures exactly that.
Where it fails. On peripheral venoarterial support with retrograde aortic flow, outflow tract velocity profiles may be distorted near the mixing zone; Doppler alignment in a hyperinflated or poorly windowed patient is difficult; outflow tract diameter is squared in the calculation, so a 10% diameter error becomes a 21% flow error; and at very low native output the velocity–time integral is small and the proportional error is large. The Red Book adds the general warning that these parameters are Doppler-based and that care is needed with Doppler alignment and range gate placement during image acquisition.
Worked example · Cross-checking a shunt-derived cardiac output on venovenous ECMO
Chapter 25 carried a case in which native cardiac output on venovenous support was derived from the oxygen shunt relationship, giving 8.7 L/min, and cross-checked against the Doppler method. The Doppler arithmetic is reproduced here because the calculation is the transferable part.
Step | Calculation | Result |
Measure outflow tract diameter, mid-systole, parasternal long axis, inner edge to inner edge | Recorded as an aortic cross-sectional area of 4.95 cm² | Area = 4.95 cm² |
Measure the velocity–time integral, pulsed wave Doppler, apical five-chamber, sample volume at the outflow tract | Traced envelope | VTI = 20.69 cm |
Stroke volume | 4.95 cm² × 20.69 cm | 102.4 mL |
Cardiac output | 102.4 mL × 89 beats/min = 9114 mL/min | 9.07 L/min |
Compare with the shunt-derived value | 9.07 versus 8.7 L/min | Agreement within about 4% |
What the agreement establishes, and what it does not. It establishes that two methods resting on entirely different assumptions — one on oxygen mass balance, one on Doppler geometry — converged. That is a genuine check. It does not establish that either is accurate, only that they are not failing in different directions. Two methods agreeing is strong evidence against a gross error and weak evidence for a correct value.
On venovenous support this calculation has a second meaning that Chapter 25 established and this chapter reinforces: arterial saturation is a cardiac output monitor. If native cardiac output rises while circuit flow is fixed, the shunt fraction rises and arterial saturation falls, with nothing wrong with the machine.
Clinical pearl · The one number that should be on the flowsheet
Total systemic flow = circuit blood flow + native cardiac output. The console supplies the first term. Only echocardiography supplies the second on venoarterial support, and it is the term that changes as the heart recovers or deteriorates. A patient whose circuit flow has been held at 3.5 L/min for four days and whose outflow tract velocity–time integral has risen from 6 cm to 13 cm has roughly doubled their native contribution while the console showed nothing at all.
Trending the velocity–time integral at a constant flow is the cheapest recovery monitor available, and it costs one measurement per day.
Pitfall · A velocity–time integral trend is only a trend if the flow was the same each time
This is §26.1 applied to the cross-check. A velocity–time integral of 13 cm at 2.5 L/min is a worse heart than a velocity–time integral of 13 cm at 4.5 L/min, not an equal one. The comparison is only valid at matched circuit flow, matched vasoactive support and matched rhythm, and units that do not record those three things cannot construct a valid trend from their own reports.
26.11 The turndown study: the only examination that asks a question the flow does not answer
Everything so far describes the heart at whatever flow it happens to be running. A turndown, or flow-reduction, study deliberately changes the load and measures the response. It is the echocardiographic examination with the most physiological content and the one most in need of a written protocol, because it is also the one with a procedural risk.
26.11.1 Three techniques, and what each can and cannot assess
The Red Book describes three approaches to reducing support for assessment, and the differences between them are not procedural detail — they determine which chamber can be assessed.
Technique | What happens | Can assess | Risks and limits |
Simple flow reduction | Circuit flow is reduced in steps and imaging repeated at each step | Left ventricle well. Right ventricle poorly, because drainage continues and full right ventricular loading is not achieved | Main advantage: easy at the bedside without any circuit manipulation. Thrombus risk rises as flow falls; the Red Book's weaning chapter treats low flow as time-limited |
Arteriovenous bridge | A length of tubing with three-way taps connects the arterial and venous limbs. Flow to the patient is reduced with a releasable clamp while circuit flow through the bridge is kept high to avoid clot formation. When haemodynamics are stable, both cannulas are isolated by turning the taps off to them, with heparin flush lines added or intermittent manual heparinised-saline flushing | Both ventricles, independently of the circuit. Full right ventricular loading is achieved. Also allows the question of whether the patient needs reconfiguration from venoarterial to venovenous support to be answered | Air entrainment during circuit manipulation is the main risk. Anticoagulation is still needed because thrombus can form near the three-way taps; flow through the bridge must be monitored |
Pump-controlled retrograde trial off | Pump speed is gradually reduced until flow reverses through the circuit | Described first in neonatal and paediatric weaning and used in adults | The Red Book notes that if echocardiography is being used, 10 to 15 minutes is adequate to demonstrate left and right ventricular changes |
Clinical pearl · If the question is about the right ventricle, simple flow reduction cannot answer it
This is the practical consequence of the Red Book's phrase about full right ventricular loading. A patient whose weaning obstacle is thought to be the right heart needs a bridge or a retrograde trial, not a turndown — because a turndown leaves the drainage cannula removing the very preload that the right ventricle must be shown to tolerate.
And the converse: a right ventricle that looks acceptable during a simple flow-reduction study has not been tested.
26.11.2 A protocol
The most fully specified published protocol is the Aarhus prospective cohort, and it is reproduced here because it is explicit about steps, durations and stopping rules.
Stage | Flow | Duration | Assessed |
Baseline | 100% of current flow | — | Haemodynamic criteria (three of four required), respiratory criteria (three of three required), echocardiographic criteria (three of four required; if only two or three were obtainable, two sufficed) |
Step 1 | 66% of initial flow | 5 minutes | Reassess haemodynamic and cardiopulmonary function |
Step 2 | 33% of initial flow | Held 12 hours if the reassessment is satisfactory | Reassess as above; if stable through the 12 hours, stop support and decannulate |
Abort criteria at any stage | Return to 100% | — | Mean arterial pressure below 60 mmHg; saturation below 90%; or an unacceptable echocardiographic evaluation despite adjustment of vasopressor therapy. The attempt is then recorded as a failure and repeated on a later day after further recovery and optimisation |
Evidence · The Aarhus prospective cohort
Thirty-eight patients weaned from ECMO for refractory cardiac failure between April 2017 and April 2019. Twenty-six (68%) met the weaning criteria at first attempt, and 25 of those 26 (96%) were successfully weaned, defined as completing a full weaning attempt with no further need for mechanical support within 24 hours.
Of the 12 who did not meet criteria, six died of haemodynamic deterioration on ECMO; six underwent optimisation of fluid balance and inotropic support, of whom three met criteria at a second attempt and were successfully weaned, two did not and died during the second attempt, and one was successfully weaned at a third attempt without fulfilling all the echocardiographic criteria.
Thirty-day survival was 85% (95% CI 64–94%) in the criteria-met group and 17% (3–41%) in the criteria-failed group, p under 0.001. Hospital discharge: 21 of 26 versus 1 of 12.
Univariable associations with successful weaning at 100% flow: left ventricular outflow tract velocity–time integral OR 1.61 (1.14–2.27), p = 0.007; tricuspid annular plane systolic excursion OR 1.51 (1.13–2.01), p = 0.005; lateral mitral annular systolic velocity OR 2.61 (1.32–5.16), p = 0.006. Ejection fraction was not significant for weaning (OR 1.05, 1.00–1.11, p = 0.112) but was for 30-day mortality (HR 0.95, 0.89–1.00, p = 0.048). Only low velocity–time integral was associated with 30-day mortality (HR 0.82, 0.71–0.94, p = 0.006). If velocity–time integral at 100% flow exceeded 10 cm, 68% of patients were successfully weaned.
Certainty: low. Prospective, but single-centre, 38 patients in the analysed cohort out of 82 treated, univariable statistics only, and — critically — patients bridged to transplant or upgraded to other devices, and patients in whom treatment was withdrawn as futile, were all excluded from the analysis. That exclusion removes exactly the patients whose hearts did not recover, which inflates the apparent performance of criteria intended to detect recovery.
Source: Mørk SR, Frederiksen CA, Nielsen RR, et al. Acta Anaesthesiologica Scandinavica 2021;65(7):936–943. DOI 10.1111/aas.13814.
Pitfall · The 96% figure is the most misleading number in this literature
"Ninety-six per cent of patients who met the criteria were successfully weaned" reads as a validation of the criteria. It is not, and the reason is structural rather than statistical. The criteria were used to select who underwent a weaning attempt. Patients who failed the criteria mostly did not get an attempt — six of the twelve died on ECMO. A test that determines who is tested cannot be evaluated against the results of that test.
The Red Book names this defect in its own weaning chapter: "not all the patients who met the weaning criteria were weaned in the prospective studies", and concludes that prospective validation of these parameters is needed.
What the Aarhus data genuinely establish is that an integrated, protocolised, multi-domain assessment identifies a group with a very different trajectory from those who fail it — 85% versus 17% thirty-day survival. That is prognostic separation, and it is real. It is not evidence that the individual echocardiographic thresholds are correct, and it is certainly not evidence that a patient who fails them cannot be weaned — one patient in this very cohort was weaned successfully at a third attempt without meeting all of them.
26.11.3 The manoeuvre that moves the mixing zone
A turndown has a second, entirely separate use on peripheral venoarterial support, and it comes from the 2026 imaging review that Chapter 25 flagged forward.
Evidence · Reducing flow to move the watershed
A 2026 narrative review of diagnostic imaging during ECMO states that transoesophageal echocardiography enables real-time assessment of cardiac–ECMO interactions and identification of the watershed zone in venoarterial ECMO, supporting individualised flow optimisation. For contrast-enhanced imaging in particular, the review's manoeuvre is to reduce extracorporeal flow to the minimum clinically acceptable value, not lower than 1.5 L/min, in order to move the mixing zone away from the anatomical area of diagnostic interest.
The same review notes that contrast-enhanced computed tomography on peripheral venoarterial support is distorted by ECMO-related haemodynamic alterations affecting contrast distribution, and surveys low-field bedside magnetic resonance imaging and electrical impedance tomography as emerging alternatives.
Certainty: expert opinion in a narrative review; no validation study of the 1.5 L/min figure is cited. The reasoning is sound and mechanical: the mixing point sits where retrograde circuit flow and antegrade native flow balance, so reducing circuit flow moves it distally toward the descending aorta.
Source: Rossetti M, Capuano P, Mamone G, et al. Artificial Organs 2026;50(5):647–659. DOI 10.1111/aor.70092.
Danger · A flow reduction performed for imaging is still a flow reduction
Reducing circuit flow to move a mixing zone, or to opacify a vascular bed, subjects the patient to exactly the same physiology as a weaning trial — with none of the preparation and usually none of the monitoring. The 1.5 L/min floor in the source is a thrombosis and perfusion floor, not a safety guarantee.
Treat an imaging turndown as a weaning trial: the same stopping rules, the same person at the bedside, the same documentation, the same anticoagulation check, and the shortest duration that answers the question.
Clinical pearl · Colour Doppler in the descending aorta locates the mixing point without any contrast at all
The 2022 review makes the point directly: colour flow Doppler in the descending aorta can determine mixing of venoarterial support and native cardiac output, and the level at which mixing occurs can be indicative of cardiac recovery and weaning progress. It pairs this with the arterial-line observation that a narrow pulse pressure indicates the watershed may be closer to the aortic root, whereas a wider pulse pressure suggests a more distal area of mixing.
A published paediatric case demonstrated the same thing ultrasonographically in the abdominal aorta of a 14-year-old on femoro-femoral venoarterial support (Reddan et al., Australasian Journal of Ultrasound in Medicine 2020;23(4):255–263, DOI 10.1002/ajum.12206).
So the mixing point can be watched daily, non-invasively, and its migration toward the aortic root is a recovery signal — Chapter 16 owns the differential-hypoxaemia consequence of where it sits.
26.12 Do the echocardiographic weaning criteria work? Three cohorts, three answers
The three parameters below are the closest thing this field has to a standard. They appear in the ELSO Red Book, in review after review, and in most unit protocols.
Parameter | Threshold | Source of the threshold |
Left ventricular ejection fraction | Over 20–25% | Red Book Chapter 29, citing a retrospective cohort and a systematic review |
Left ventricular outflow tract velocity–time integral, pulsed wave Doppler | 10 cm or more | As above |
Lateral mitral annulus peak systolic velocity (tissue Doppler S′) | 6 cm/s or more | As above |
They trace to a single original cohort. The Red Book, the 2022 review and the Aarhus study all cite Aissaoui and colleagues: 51 patients, of whom 38 who were haemodynamically stable underwent at least one weaning attempt and 20 were ultimately weaned. Every subsequent protocol rests on that.
Controversy 1 — Are the standard echocardiographic weaning criteria a decision rule, or a weak prior dressed as one?
The problem, stated exactly. The same three-parameter criterion set has been reported with three incompatible performances.
- The Aarhus prospective cohort (38 patients). 26 of 38 met criteria; 25 of those 26 (96%) were successfully weaned. Thirty-day survival 85% versus 17% (p under 0.001). Reads as a rule that works.
- The Hong Kong load-manipulation cohort (46 patients). The same three criteria, as a model, gave area under the receiver operating characteristic curve 0.86 (0.73–0.98) for predicting who was potentially weanable — but only 0.68 (0.51–0.85) for predicting eventual successful decannulation. Both models were well calibrated (Hosmer–Lemeshow p over 0.05). For comparison the SAVE score in the same cohort gave 0.58 and 0.54. Reads as a rule that predicts the clinician's decision better than the patient's outcome.
- The cohort reported by Kim and colleagues, as summarised in the 2022 review. Applying the previously published criteria — ejection fraction over 20–25%, velocity–time integral 10 cm or more, mitral annular S′ 6 cm/s or more — 30% of patients who could not be weaned met all three criteria, while only 16% of patients who were successfully weaned met all three. Read at face value, that is the criteria performing in the wrong direction.
The case that they work. The Aarhus separation is large, prospective, protocolised and associated with survival, not merely with weaning. The parameters are physiologically sensible: a ventricle generating a 10 cm velocity–time integral against the afterload of a running circuit is doing real work. And the Red Book's integrative framing — three of four echocardiographic criteria, alongside three of four haemodynamic and three of three respiratory criteria — is a composite, not a single threshold.
The case that they do not. Four objections, and they are structural rather than statistical.
- Verification bias. The criteria select who gets a weaning attempt. Patients who fail them mostly never get one, so the false-negative rate is unmeasurable by construction. Six of the twelve Aarhus patients who failed criteria died on ECMO without an attempt.
- Survivorship in the analysed cohort. Aarhus analysed 38 of 82 treated patients, excluding those bridged to transplant, upgraded to other devices, or in whom treatment was withdrawn as futile — which removes most of the hearts that did not recover.
- Load-dependence. Every one of the three parameters moves with circuit flow (§26.1), and the studies measured them at different support levels. The Hong Kong data quantify how large that movement is: in the acute phase, ejection fraction changed by a median of 43.0% and velocity–time integral by 45.8% between 120% and 50% of target flow.
- The right ventricle is not assessed. Red Book Chapter 29: full right ventricular loading is not achieved during partial-support assessment. A criterion set that cannot see the chamber that causes many weaning failures cannot predict them.
What is not in dispute. Three things.
- Weaning failure is common regardless. One trial protocol states weaning failure as high as 50%, and states it specifically about patients in whom these criteria were fulfilled, adding that failure is associated with secondary shock, escalation to another device, high vasoactive requirements and increased mortality.
- A low velocity–time integral is bad news independent of weaning. In Aarhus it was the only echocardiographic parameter associated with 30-day mortality (HR 0.82, 0.71–0.94, p = 0.006).
- There is no validated echocardiographic weaning protocol. The 2022 review states it flatly: "Currently, there are no recognized echocardiography protocols for weaning of VA-ECMO", and its authors call for interdisciplinary consensus guidelines as a basis for further research.
Where this book lands.
- Use the criteria as an entry gate to a trial, not as a substitute for one. Their demonstrated use is identifying who is worth testing. The test is the turndown, not the threshold.
- A patient who fails the criteria has not been shown to be unweanable. One Aarhus patient was weaned at a third attempt without meeting all of them. Failing the criteria is a reason to optimise and retest, not a reason to stop — Chapter 18 owns the decision and Chapter 24 owns the point at which retesting stops being reasonable.
- Record the flow at which each parameter was obtained, and compare only matched studies (§26.1). Much of the disagreement between these three cohorts may be a disagreement about support level rather than about hearts.
- Assess the right ventricle under load, or state in the report that it was not assessed (§26.9, §26.11.1).
- Attach a certainty label when these numbers are quoted on a ward round. "Meets weaning criteria" said in the tone reserved for a potassium result is the specific error this section exists to prevent.
What would settle it. A multicentre prospective study in which the criteria are measured at a protocol-specified flow in every patient on venoarterial support — including those not offered a weaning trial — with decannulation and survival as endpoints and with the decision to attempt weaning made independently of the criteria. That study does not exist. The Red Book's own weaning chapter says prospective validation of these parameters is needed; the 2022 review says no recognised protocols exist. Four years apart, two primary sources say the same thing, and nothing has changed.
[VERIFICATION REQUIRED] — the Kim reversal figures (30% versus 16%) are reported here exactly as the 2022 review reports them; the primary paper was not retrieved. The direction of the finding is so consequential that it should not be repeated without the original.
Pitfall · The tissue Doppler criterion fails for an anatomical reason nobody mentions
Red Book Chapter 29 records it and it is easy to miss: in pathologies sparing or involving only the basal anterolateral wall segments, lateral mitral annular systolic velocity will not be a good predictor of weaning. The measurement samples one wall. A patient whose infarct is inferior, septal or apical may have a normal lateral annular velocity and a ventricle that cannot support a circulation; a patient with a basal anterolateral infarct may have a low value and a functionally adequate ventricle.
A single-wall longitudinal velocity is being used as a global function surrogate, and the substitution fails exactly where the disease is regional — which is most of the cardiogenic shock population.
Clinical pearl · What to look at during the turndown, not before it
The parameters in the table above are measured at baseline flow. The more informative observations are the ones made during the reduction, and Red Book Chapter 29 names three:
- A change in lateral mitral annular early diastolic velocity (e′) when flow is reduced to 30–50% of baseline for 15 minutes.
- A change of over 10% in right ventricular S′ under the same conditions.
- Absence of left or right ventricular distension at the end of the protocol, which in a haemodynamic transoesophageal weaning protocol was associated with successful weaning.
The Red Book describes the first two as proposed to be more sensitive than ejection fraction, velocity–time integral and mitral annular S′ — and immediately adds that most of these tests require further validation. Certainty: very low. Reported here because they are comparisons rather than thresholds, which is the form this chapter argues for.
26.13 Venovenous ECMO: a different set of questions
On venovenous support the heart is not being replaced, so the load-dependence problem largely disappears and three different questions take its place.
Question | Why echocardiography answers it | Key observations |
Is the right ventricle coping? | The Red Book names right ventricular function as the specific focus of monitoring on venovenous support. The 2026 review states the reason: in venovenous ECMO the heart's circulatory function is still required, and impaired right ventricular performance can hinder oxygen delivery to the systemic circulation | Right-to-left ventricular ratio; free-wall function; tricuspid annular excursion and fractional area change; tricuspid regurgitation velocity; septal position |
Why is this patient still hypoxaemic? | The ELSO 2015 guideline gives this as echocardiography's principal role on venovenous support. The differential is recirculation, insufficient circuit flow relative to a rising cardiac output, or an intracardiac shunt — and all three are imaging diagnoses | Return jet direction and cannula geometry (§26.5); native cardiac output by Doppler (§26.10); agitated-saline contrast for a shunt |
Why are the flows low? | The Red Book: echocardiography is helpful to troubleshoot inadequate flows that cannot be explained by circuit failure, and allows discrimination between hypovolaemia, cannula malposition and thrombus formation | Chamber sizes; cannula tip position; thrombus at and around cannulas; pericardial view |
Physiology · On venovenous ECMO, a rising cardiac output looks exactly like a failing oxygenator
This is Chapter 25's inversion restated as an imaging problem. Arterial saturation on venovenous support is determined by the ratio of circuit flow to native cardiac output, because whatever fraction of venous return bypasses the circuit is a true right-to-left shunt. If the patient becomes febrile, agitated, septic or simply lighter on sedation, cardiac output rises, the shunt fraction rises, and arterial saturation falls with a perfectly functioning circuit.
The console cannot distinguish this from oxygenator failure or from recirculation, because the console does not know the native cardiac output. Echocardiography does. A left ventricular outflow tract velocity–time integral measured before and during the desaturation separates a rising cardiac output from a failing membrane in one measurement.
Certainty for the mechanism: high — it is arithmetic. Certainty for the clinical frequency: not established in the sources used here.
Clinical pearl · Three causes of falling saturation on venovenous ECMO, and the one test that splits them
Cause | Circuit flow | Post-oxygenator saturation | Native cardiac output by Doppler | Echo finding |
Rising native cardiac output | Unchanged | Normal | Risen | Hyperdynamic ventricle; higher velocity–time integral; cannulas normal |
Recirculation | Unchanged or high | Normal | Unchanged | Return jet misdirected or cannulas too close; drainage blood unexpectedly bright red on the circuit |
Oxygenator failure | Unchanged | Fallen | Unchanged | Nothing — this is the diagnosis echocardiography excludes rather than makes (Chapter 32) |
Post-oxygenator saturation is the first discriminator and costs one blood gas. Echocardiography separates the remaining two. Chapter 27 owns the gas sampling and Chapter 32 owns the oxygenator.
Pitfall · Lung ultrasound on ECMO reads differently, and one sign is actively misleading
Red Book Chapter 48 records the trap: caution should be used when diagnosing a pneumothorax on ECMO with ultraprotective ventilation. Even with no pneumothorax, lung sliding may be absent because the lung is completely collapsed — while the lung pulse is still present. The absent-sliding sign, which is the workhorse of pneumothorax diagnosis, has lost its specificity in exactly this population.
The same chapter records two pilot studies in which lung ultrasound scores did not differ between survivors and non-survivors at the time of ECMO initiation, while improvement in the score was seen in patients who improved and could be decannulated from venovenous support, and persistently high scores portended a poor prognosis. Certainty: very low — two pilot studies. Others have described lung ultrasound score as a possible surrogate for lung compliance.
26.14 Echocardiography during extracorporeal cardiopulmonary resuscitation
Part IV established that in an arresting patient both cues that identify a vessel fail simultaneously — compressions make the vein pulsatile and oxygen debt makes arterial blood dark — which turns vessel identification from a clinical skill into an imaging problem (Chapter 21 §21.1). This section adds what the cardiac probe contributes, which is a separate set of tasks from the vascular probe.
Task | What transoesophageal imaging adds |
Diagnosis | Identifies reversible causes — pericardial tamponade, pulmonary embolism, left or right ventricular failure, hypovolaemia, aortic dissection — and establishes presence or absence of intrinsic cardiac activity. Knowing the cause before cannulating allows the team to plan an exit strategy in advance, which is a Chapter 24 argument arriving by a different route |
Compression quality | Continuous imaging during compressions, without the pauses that transthoracic imaging requires. A published case documented two separate occasions when the compression vector had shifted so that compressions fell over the left ventricular outflow tract and aortic root rather than the left ventricle; the mechanical device was repositioned each time with marked improvement in left ventricular compression |
Vessel confirmation before dilation | In the same case, arterial and venous guidewires were each visualised — one in the descending thoracic aorta, one in the inferior vena cava on the mid-oesophageal bicaval view — before dilation and cannulation. The authors' stated rationale is preventing cannulation errors such as inadvertent veno-venous cannulation that would otherwise be discovered only when flow is attempted |
Cannula position at the moment flow starts | Direct visualisation of the venous cannula relative to the right atrium before connection to the circuit |
Evidence · What the transoesophageal literature in cardiac arrest does and does not show
A systematic literature review identified 55 publications describing 277 patients in whom transoesophageal echocardiography was used during cardiac arrest resuscitation, across operating rooms, emergency departments and other settings. Its conclusions: in many cases transoesophageal imaging changed the direction of the resuscitation; it is unclear whether it changed patient-oriented outcomes such as neurologically intact survival; few adverse events related to the technique have been documented; and there is growing evidence that physicians can learn to use it during resuscitations and apply the findings to clinical decisions.
The same review records the specific limitation that motivates the technique: a study demonstrated that transthoracic imaging during cardiac arrest can lengthen pulse-check duration beyond the 10 seconds recommended by the resuscitation councils.
Certainty: low for the direction-changing claim (case series and reports); moderate for the mechanical claim that transoesophageal imaging does not require compression pauses, which is a property of the technique rather than a study finding. No outcome claim is supported.
Sources: Parker BK, Salerno A, Euerle BD. Journal of Ultrasound in Medicine 2018;38(5):1141–1151. DOI 10.1002/jum.14794. Giorgetti R, Chiricolo G, Melniker L, Calaf C, Gaeta T. Journal of Clinical Ultrasound 2019;48(3):184–187. DOI 10.1002/jcu.22788.
Danger · The transthoracic probe lengthens pulse checks, and in extracorporeal cardiopulmonary resuscitation every second is priced
Chapter 19 established a working figure of 2 to 2.5% of survival lost per minute of low flow. A transthoracic study that extends three pulse checks from 10 to 25 seconds has spent 45 seconds of no-flow time — and it did so in the service of an image that a transoesophageal probe would have produced without stopping compressions at all.
This is not an argument that every arrest needs a transoesophageal probe. It is an argument that the decision to image transthoracically during an arrest is a decision to stop compressions, and it should be made with that stated. Where a transoesophageal probe and a trained operator are already present — as they are in most extracorporeal cardiopulmonary resuscitation teams by the time cannulation is contemplated — the transthoracic study has no advantage during compressions.
Pitfall · The post-arrest patient has three simultaneous mechanisms for pericardial blood
Chest compressions fracture ribs; guidewires and cannulas can perforate; and the patient is then fully anticoagulated. The case in §26.8 had all three. The Red Book's universal instruction — check for a new pericardial effusion after every cannulation and reassess a pre-existing one — is at its most important here, and the follow-up study matters as much as the first one, because the bleeding in that case was trivial initially and became a 450 mL tamponade over ten hours.
26.15 The findings only echocardiography makes
Collected here because they are the strongest case for imaging on ECMO: a list of diagnoses for which no other bedside instrument produces the finding at all.
Finding | Where to look | What it changes |
Intracardiac or aortic root thrombus | Apical four- and two-chamber (transthoracic is often better than transoesophageal for the apex); aortic root in parasternal short axis and mid-oesophageal long axis; around every cannula | Anticoagulation target; urgency of venting or reconfiguration; and, if a cannula is to be exchanged, whether thrombus is present that could embolise on manipulation — the 2022 review notes that it is not uncommon for thrombus to embolise into the lungs or systemically when large-bore cannulas are manipulated |
Spontaneous echo contrast | Left ventricle, left atrium, aortic root, right ventricle, pulmonary artery | The precursor of thrombus and the trigger to reappraise both the anticoagulation target and the need for venting. Red Book Figure 48-12 shows the full sequence in one patient: no aortic valve opening, contrast in the aortic root, contrast in the right ventricle and pulmonary artery, and an apical thrombus |
New or worsened aortic regurgitation | Mid-oesophageal aortic valve long axis with colour; parasternal long axis | Converts a non-opening valve from a stasis problem into a distension emergency. Documented with paired imaging before and after initiation in the 2022 review |
New or worsened mitral regurgitation | Standard views with colour | Left atrial distension, arrhythmia, pulmonary oedema. Systolic and diastolic mitral regurgitation together is one of the named findings of left ventricular distension |
Intracardiac shunt with right-to-left flow | Agitated-saline contrast, ideally during a turndown (§26.9) | Explains hypoxaemia that neither the lungs nor the circuit account for |
Intracardiac air | Continuous imaging of the cardiac chambers during high-risk circuit manipulation | The 2022 review is explicit: as flows are reduced or temporarily stopped during reconfiguration, the risk of air embolism remains ever-present and of high potential consequence. Continuous imaging will readily identify even a small amount of intracardiac air, and the proceduralist and perfusionist should be promptly alerted. This applies directly to the arteriovenous bridge technique in §26.11.1, whose main stated risk is air entrainment |
Pericardial effusion, including loculated | Subcostal, parasternal, apical; transoesophageal for posterior loculation | §26.8 — the diagnosis that the circuit conceals until it decompensates |
Aortic dissection after arterial cannulation | Descending aorta on transoesophageal; aortic sweep | Red Book Chapter 48 instructs checking for its absence after cannulation |
Cannula tip and jet position | §26.5 | The commonest correctable explanation for a flow or oxygenation problem |
Clinical pearl · Pair the apical thrombus question with the modality that answers it
The reflex on ECMO is to reach for the transoesophageal probe for anything difficult. For the left ventricular apex that reflex is wrong — the apex is in the far field on transoesophageal imaging and frequently foreshortened. The 2022 review says so directly: if an apical thrombus is suspected, transthoracic imaging may provide better apical imaging in some instances and can be used for quick verification.
Two probes, one examination. The two modalities are not competitors in the ECMO patient; they have different blind spots, and the blind spots do not overlap.
26.16 Who performs the study
Every source used in this chapter says that training matters, and none of them says what training. The gap is worth naming because it determines whether the findings in §26.15 are made at three in the morning or at ten the next day.
Source | What it says about the operator |
ELSO 2015 ultrasound guidance | The main disadvantage of ultrasound is that it is operator dependent; it is important that appropriately trained personnel use it to guide management of patients on ECMO. No standard, level or curriculum is specified |
ELSO Red Book Chapter 48 | "Adequate echocardiography training is mandatory to safely guide ECMO cannulation", because wires and cannulas generate scattering and artefact and rapid switching between views is needed. Concludes that physicians managing ECLS patients should be familiar with specificities of ultrasound on ECLS to interpret findings accurately |
2022 cardiac-imaging review | Written explicitly for "advanced imagers involved in the care of these complex patients", and states that it is imperative for imaging physicians to understand the anatomic considerations, the physiological context and the limitations of the imaging modality |
Dual-lumen cannula series | "Even experienced personnel required additional training to obtain and interpret these images." Image quality improved dramatically over the course of a year; the process was heavily operator-dependent and required prompt availability of a team accustomed to interpreting ECMO imagery |
Controversy 2 — Should ECMO echocardiography be performed by the bedside intensivist or by an accredited imaging service?
The problem, stated exactly. The two positions are held simultaneously by the sources this book treats as primary, and they are not reconciled anywhere. ELSO writes for the ECMO clinician and says the clinician must be trained; the cardiac-imaging literature writes for the advanced imager and says the imager must understand ECMO. Neither states who should hold the probe at 3 a.m. when the flow drops.
The case for the bedside intensivist. Almost every observation in this chapter is time-critical and triggered rather than scheduled: a falling flow, a new loss of pulsatility, a rising central venous pressure, a desaturation. The findings that most change management — does the valve open, is the ventricle big or small, is there an effusion, where is the cannula tip pointing — are presence-or-absence observations within the scope of focused critical care echocardiography, and the alternative to making them now is making them hours later. The Chapter 21 finding generalises: when imaging is the only way to answer the question, the answer has to be available when the question is asked. And the differential in §26.7 has to be resolved in minutes, because two of its branches lead to opposite interventions.
The case for the accredited imaging service. The quantitative work is genuinely hard and genuinely error-prone: Doppler alignment and range-gate placement determine the velocity–time integral, outflow tract diameter enters the calculation squared, tissue Doppler sampling has the single-wall problem of §26.12, strain requires software and adequate image quality, and cannula artefact obscures adjacent structures. The dual-lumen series is direct evidence that this is a skill that takes months to acquire even in experienced hands. A misread study is worse than no study: it produces a number that gets written down, trended and acted on.
What is not in dispute. Three things.
- Guiding a dual-lumen bicaval cannula without fluoroscopy is an advanced transoesophageal task, and ELSO effectively makes it mandatory. No unit should treat that as a focused-echocardiography competency.
- The screening and the quantitation are different skills, and the literature conflates them by calling both "echocardiography".
- Whoever performs it must record the circuit flow (§26.1). This is the failure that no amount of imaging expertise corrects, because it is a reporting convention rather than a skill.
Where this book lands. A two-tier model, stated as this book's own recommendation rather than as a guideline position.
- Tier 1, the bedside clinician, any hour: aortic valve opening; chamber sizes and the left-versus-right comparison; pericardial effusion; spontaneous echo contrast; cannula tip position and jet direction on colour; gross ventricular function. These answer the questions in §26.6 and §26.7 and they are the ones that cannot wait.
- Tier 2, the accredited operator, scheduled: all quantitation — velocity–time integral, ejection fraction by a volumetric method, tissue Doppler, strain, three-dimensional right ventricular ejection fraction; every turndown study; every dual-lumen cannulation; and every study on which a weaning or unloading decision will rest.
- Every Tier 1 study that finds something is followed by a Tier 2 study, and every Tier 2 study establishes the baseline against which the next Tier 1 study is read.
What would settle it. A study comparing time-to-diagnosis and diagnostic accuracy for the specific ECMO findings in §26.15 between focused bedside operators and accredited operators, with the delay to the accredited study measured rather than assumed. No such study was identified. The training question in this field is currently answered by local resource availability and described afterwards as a policy.
Clinical pearl · What an ECMO echocardiogram report must contain
A minimum dataset, assembled from the requirements scattered across this chapter. A report missing any of the first three is not interpretable.
- Circuit blood flow (L/min), pump speed and sweep gas flow at the time of imaging.
- Vasoactive and inotropic infusions running, with doses, and whether these differ from the previous study.
- Rhythm, and whether it differs from the previous study.
- Aortic valve opening: every beat, intermittently with a frequency, or not at all.
- Spontaneous echo contrast and thrombus: present or absent, by chamber, including the aortic root.
- Left and right ventricular size, and the right-to-left comparison.
- Pericardial effusion: present or absent, size, new or pre-existing, loculated or circumferential.
- Cannula tip positions and jet directions, with the modality and view used.
- Quantitation, if performed, each value with the flow at which it was obtained.
- An explicit statement of what was not assessable and why — a foreshortened apex, an obscured interatrial septum, an unassessable right ventricle under partial support.
26.17 Putting it together: the study as a decision tree
The sequence below is the §26.6 surveillance order applied to a specific, common trigger — a patient on peripheral venoarterial ECMO with new loss of pulsatility, a falling flow, or both. It is drawn as a tree because the branches genuinely diverge: two of them end in unloading and two of them end in explicitly not unloading.
Clinical pearl · The pericardial view goes first for a reason that is not clinical suspicion
It is placed first not because tamponade is the commonest branch but because it is the branch whose signs the circuit removes (§26.8) and the only branch where the wrong intervention is immediately harmful in a way that cannot be undone. Placing a venting device into an underfilled ventricle that is underfilled because it is being compressed from outside is the specific error this ordering prevents. The view takes seconds and it is the cheapest safety step in the whole examination.
26.18 What we do not know
Stated plainly, because the certainty labels throughout this chapter are low for a reason.
- There is no validated echocardiographic weaning protocol for venoarterial ECMO. The 2022 review states it in terms, and its authors call for interdisciplinary consensus guidelines. The Red Book's weaning chapter says prospective validation is needed. Four years separate the two statements and nothing intervened.
- The three standard weaning parameters have never been evaluated free of verification bias, and in one reported cohort they ran backwards (§26.12). The primary paper behind that reversal was not retrieved for this chapter and should be.
- Nobody knows what a falling flow-sensitivity means. Global longitudinal strain changed by 72.7% with flow in the acute phase and 22.5% in the delayed phase. Whether a ventricle that has stopped responding to unloading is recovering or is beyond recovery is not answered by the study that produced the observation (§26.1), and the question is directly actionable if answered.
- The optimal frequency of surveillance imaging is unknown. Three primary sources give twice daily, daily, and by-stage. None cites evidence, and no study has compared a triggered strategy with a scheduled one.
- The 1.5 L/min floor for an imaging turndown is an expert opinion in a narrative review with no validation cited (§26.11.3).
- Dual-lumen cannula port malposition frequency is unknown. The 5-of-11 figure is eleven retrospective studies at one centre, performed because somebody suspected a problem.
- What training is required to make the §26.15 findings reliably has never been defined, and the two bodies of literature writing about it are addressing different audiences (§26.16).
- Whether transoesophageal imaging during cardiac arrest changes patient-oriented outcomes is unknown. The systematic review says so explicitly: it frequently changes the direction of resuscitation; whether it changes neurologically intact survival is unclear.
- No study has related echocardiographic aortic root stasis to subsequent thrombosis by pulse-pressure band — Chapter 25's Controversy 1 named this gap and this chapter cannot close it either. It is now the second chapter to record the same missing study.
- The ELSO ultrasound guidance used throughout this chapter was last updated in May 2015 and its own text states it was scheduled to expire by May 2018, with users encouraged to contact the guidelines editorial board to confirm that it remains in effect. Every statement taken from it in this chapter is more than eleven years old and formally out of date, and it remains the only ECMO-specific ultrasound guidance in the sources available here.
Key points
- On ECMO there is no ejection fraction — only an ejection fraction at a stated circuit flow. The ELSO Red Book instructs that all parameters be reported with the ECMO flows at the time of assessment. A report without the flow is uninterpretable and should be returned.
- The change in a measurement across a deliberate change in flow carries more information than the measurement itself. In 46 patients, left ventricular global longitudinal strain improved from −6.1% at 120% of target flow to −8.8% at 50%, and the magnitude of that swing fell from 72.7% in the acute phase to 22.5% by around day 4.
- Echocardiography is three instruments sharing one probe, and its reliability differs by an order of magnitude between them. As a procedural tool and a structural monitor it is strong. As a weaning decision instrument it is weak, and the published cohorts disagree with each other.
- Lead the venoarterial surveillance study with the aortic valve. It is binary, robust, and it is the direct expression of the afterload contest that defines peripheral venoarterial support. Follow it with spontaneous echo contrast, then chamber sizes, then the pericardium.
- Left ventricular chamber size is an insensitive indicator of distension. The pressure–volume relationship is non-linear inside an intact pericardium. The composite — absent valve opening plus contrast plus mitral regurgitation plus a growing left atrium — is the finding, not the diameter.
- The chest radiograph cannot locate an ECMO cannula tip. The distal end is radiolucent. A daily chest film is not cannula surveillance, and in one series at least one port was misplaced in 5 of 11 dual-lumen cannula studies.
- Cannula position is a variable, not a fact. Loss of lung volume on a rest strategy raises the diaphragm and moves the heart relative to a cannula fixed at the skin — the Red Book records a case in which reinjection ended up in the liver. Re-image after any ventilation change, after proning, and whenever flow falls.
- Colour Doppler is the cannula-position test. Two-dimensional imaging shows where the cannula is; colour shows where the blood goes, which is the question.
- Two chambers answer the low-pulsatility question in ten seconds. Big left ventricle points to distension or aortic regurgitation. Small left ventricle with a big right ventricle is Chapter 25's commonest phenotype — 52% of low-pulsatility patients — and unloading it is harmful.
- Tamponade on a circuit has no physical signs. Pulsus paradoxus fails at low tidal volumes; right-sided chamber collapse is indistinguishable from drainage; the pump holds the blood pressure up. A rising central venous pressure is a tamponade sign until imaged — in one published case it rose from 9–11 to 35 mmHg before the diagnosis was made, ten hours after a clear post-cannulation study.
- The left ventricular outflow tract velocity–time integral is the one cardiac-output measurement the circuit does not defeat, because it measures velocity at a location rather than tracking an indicator. Total systemic flow equals circuit flow plus native output, and only echocardiography supplies the second term.
- If the question is about the right ventricle, a simple flow reduction cannot answer it. Full right ventricular loading is not achieved while the drainage cannula is running. That requires an arteriovenous bridge or a retrograde trial off — and the bridge's main risk, air entrainment, is itself an imaging diagnosis.
- Treat the standard weaning criteria as an entry gate to a trial, not as a substitute for one. They predicted weanability (area under the curve 0.86) far better than they predicted decannulation (0.68), and failing them has never been shown to mean a patient cannot be weaned.
- On venovenous support, a rising cardiac output looks exactly like a failing oxygenator. Post-oxygenator saturation separates them in one blood gas; echocardiography separates the remaining two causes.
- Absent lung sliding does not mean pneumothorax on ultraprotective ventilation. The lung may simply be collapsed, and the lung pulse will still be present.
- Every cannulation ends with a pericardial and pleural view. The Red Book makes it universal. It takes five seconds and it is the observation most often skipped because the patient has just become stable.
What this chapter hands to the next
- Chapter 27 (Blood Gas and Oxygenation Monitoring) inherits the post-oxygenator saturation step in §26.13's three-cause table, and the observation that arterial saturation on venovenous support is jointly determined by circuit flow and native cardiac output — so an oxygenation problem may require an echocardiogram rather than a blood gas.
- Chapter 28 (Cerebral Monitoring) inherits the Red Book's warning that continuous flow confounds traditional pulsatility measurements, which is the transcranial Doppler analogue of this chapter's central problem: an instrument reporting faithfully about a circulation it was not designed for.
- Chapter 29 (Laboratory Monitoring) inherits the liver enzyme rise that identifies intrahepatic reinjection (§26.5), which is a laboratory finding whose diagnosis is an image.
- Chapter 30 (ECMO Data Interpretation) inherits the clearest worked example in the book so far of discrimination without transferable validity: the same three-parameter model gave area under the curve 0.86 for weanability and 0.68 for decannulation in the same 46 patients. Two endpoints, one model, and the gap between them is the clinician's own decision. This joins the ECPR score's calibration defect (Chapter 23) and Chapter 25's ten-orders-of-magnitude confidence interval.
- Chapter 31 and Chapter 32 (Circuit Thrombosis, Oxygenator Failure) inherit the imaging half of the thrombus question: intracardiac and aortic root thrombus, spontaneous echo contrast as its precursor, and the embolisation risk during cannula manipulation.
- Chapter 34 (Cannula Problems) inherits the whole of §26.5 as the diagnostic method for its subject. This chapter owns imaging the cannula; Chapter 34 owns fixing it.
- Chapter 15 (LV Distension and Unloading) is owed a correction of emphasis: chamber size is an insensitive indicator, and the low-pulsatility phenotype that most often presents for unloading is the one in which unloading is contraindicated.
- Chapter 18 (VA ECMO Weaning) is owed §26.11 and §26.12 in full: the three turndown techniques and what each can assess, the Aarhus protocol, and the verification-bias argument against treating the criteria as a rule.
- Chapter 8 and Appendix 5 (Recirculation) are owed the geometric account in §26.5.1 — that recirculation is a cannula-geometry problem visible on colour Doppler — and the femoro-femoral case in which cannula length and side-hole position, not flow, were the operative variables.
- Chapter 88 to Chapter 90 (Training) inherit Controversy 2 and the two-tier model, together with the unclosed gap: no source defines what training is required.
- Chapter 78 (Hybrid Configurations) is owed the dual-lumen device table and the right atrium to pulmonary artery cannula target of 2 to 3 cm above the pulmonary valve and not selective.
References
Primary sources
- Brodie D, Peek G, MacLaren G, et al., eds. Extracorporeal Life Support: The ELSO Red Book, 6th edition. Chapter 48, Bedside Ultrasound — echocardiographic guidance for cannulation in every configuration; monitoring on venovenous and venoarterial support; the intrahepatic reinjection case and Figure 48-11; the aortic valve and spontaneous contrast findings in Figure 48-12; the paediatric and neonatal carotid cannula section and Figure 48-13; lung ultrasound; the tamponade caveats. Chapter 29, Weaning and Decannulation in Adult Cardiac Failure — the echocardiographic weaning parameters and their thresholds, the right ventricular parameters, the arteriovenous bridge, the pump-controlled retrograde trial off, and the statement that full right ventricular loading is not achieved during partial support. Verified = Yes (retrieved through the project document store).
- Nanjayya VB, Murphy D. Ultrasound Guidance for Extra-corporeal Membrane Oxygenation: General Guidelines. Extracorporeal Life Support Organization, May 2015. Editors: Brechot N, Fan E, Pellegrino V, Brodie D. The three-phase framework (pre-initiation, during initiation, post-initiation); the aortic regurgitation exclusion; the vessel-sizing rule; the transthoracic-first, transoesophageal-if-suboptimal position; the necessity of transoesophageal guidance for bicaval dual-lumen cannulation without an image intensifier; the radiolucent-tip problem; the operator-dependence warning. Verified = Yes (retrieved through the project document store). NOTE: the document states it was scheduled to expire by May 2018.
- ISCCM Manual of RRT and ECMO in ICU, Chapter 33, Monitoring during Extracorporeal Membrane Oxygenation — the echocardiographic evaluation-by-stage framework (prior to initiation, after initiation, during maintenance, during recovery). Verified = Yes (retrieved through the project document store).
- Taha AR, Caridi-Scheible M, Leiendecker E, et al. ECMO: A Practical Guide to Management. Used for the monitoring framework carried forward from Chapter 25. Verified = Yes (retrieved through the project document store).
External references — all read as indexed full-text passages, none retrieved as a complete article, no identifier independently resolved
- Hussey PT, Mering G, Nanda NC, Ahmed MI, Addis DR. Echocardiography for extracorporeal membrane oxygenation. Echocardiography 2022;39(2):339–370. DOI 10.1111/echo.15266. The single most-used external source in this chapter: the dual-lumen device table, the venoarterial distension findings, the chamber-size insensitivity argument, the aortic regurgitation paired imaging, the mixing-zone colour Doppler and pulse-pressure pairing, the air-entrainment warning, the apical thrombus modality point, the statement that no recognised weaning protocols exist, and the summarised Kim reversal figures. Verified = No.
- Ng PY, Ma TS, Ip A, Lee MK, Ng AK, Ngai CW, Chan WM, Siu CW, Sin WC. Sensitivity of ventricular systolic function to afterload during veno-arterial extracorporeal membrane oxygenation. ESC Heart Failure 2022;9(5):3241–3253. DOI 10.1002/ehf2.13959. 46 patients imaged at 100%, 120% and 50% of target flow on day 1 and day 4; the strain, ejection fraction and velocity–time integral sensitivity figures; the weanability 0.86 versus decannulation 0.68 discrimination. Verified = No.
- Mørk SR, Frederiksen CA, Nielsen RR, Lichscheidt E, Christensen S, Greisen JR, Tang M, Vase H, Løgstrup BB, Mellemkjær S, Wiggers HS, Mølgaard H, Poulsen SH, Terkelsen CJ, Eiskjær H. A systematic approach to weaning from extracorporeal membrane oxygenation in patients with refractory cardiac failure. Acta Anaesthesiologica Scandinavica 2021;65(7):936–943. DOI 10.1111/aas.13814. The Aarhus protocol, the 26 of 38 and 25 of 26 figures, the survival separation, the univariable odds and hazard ratios, and the exclusions that limit them. Verified = No.
- Rossetti M, Capuano P, Mamone G, Abdulaziz S, Ghisulal P, Panarello G, Martucci G, Cho S, Lorusso R. Diagnostic imaging challenges during extracorporeal membrane oxygenation. Artificial Organs 2026;50(5):647–659. DOI 10.1111/aor.70092. The watershed identification, the 1.5 L/min flow-reduction manoeuvre, the twice-daily imaging recommendation, the right ventricular parameters and the right-to-left ratio threshold, and the contrast-distribution problem on peripheral venoarterial support. Verified = No.
- Tops LF, Aimo A, Adamopoulos S, et al. Multimodality imaging in advanced heart failure. A clinical consensus statement of the Heart Failure Association of the ESC. Part 2: Mechanical circulatory support and heart transplantation. European Journal of Heart Failure 2025;27(11):2557–2570. DOI 10.1002/ejhf.70018. The pre-implantation phenotyping list, the venoarterial inflow cannula tip position, the daily imaging position, and the escalation sequence for incomplete aortic valve opening. Verified = No.
- Tabak B, Elliott CL, Mahnke CB, Tanaka LY, Ogino MT. Transthoracic echocardiography visualization of bicaval dual lumen catheters for veno-venous extracorporeal membrane oxygenation. Journal of Clinical Ultrasound 2012;40(3):183–186. DOI 10.1002/jcu.21873. The subcostal sagittal view identifying all three ports in 11 of 11 studies, at least one port misplaced in 5 of 11, the single-sheep recirculation figures, the 9% registry prevalence, and the operator-dependence observation. Verified = No.
- Yastrebov K, Kapalli T. Malposition of double lumen bicaval venovenous extracorporeal membrane oxygenation cannula resulting in hepatic venous congestion. Australasian Journal of Ultrasound in Medicine 2015;16(4):193–197. DOI 10.1002/j.2205-0140.2013.tb00247.x. Dual-lumen cannula geometry, the ideal port positions, and the malposition case. Verified = No. NOTE: the retrieved metadata gives volume 16, issue 4, pages 193–197 with a 2015 publication date and a 2013 identifier; the discrepancy is recorded and not resolved.
- Wada K, Shimizu K, Narimatsu E, Sato Y, Hamaguchi J, Matsuyoshi T, Kaneko H. Difficulty in VV ECMO configuration change. Journal of the Japanese Association for Acute Medicine 2020;31(12):2449–2454. DOI 10.1002/jja2.12504. The femoro-femoral recirculation case and the cannula-length solution. Verified = No. Read through an English abstract and a machine-rendered body passage.
- Lazzeri C, Cianchi G, Bonizzoli M, Batacchi S, Terenzi P, Bernardo P, Valente S, Gensini GF, Peris A. Pulmonary vascular dysfunction in refractory acute respiratory distress syndrome before veno-venous extracorporeal membrane oxygenation. Acta Anaesthesiologica Scandinavica 2015;60(4):485–491. DOI 10.1111/aas.12643. The pre-cannulation prognostic comparison in 21 patients and the 9.5% acute cor pulmonale figure, with the selection caveat recorded in the text. Verified = No.
- Tsukada Y, Nemoto C, Ikegami Y, Suzuki T, Tase C. Cardiac tamponade after extracorporeal cardiopulmonary resuscitation. Acute Medicine and Surgery 2015;3(2):214–215. DOI 10.1002/ams2.128. The ten-hour delayed tamponade case, the central venous pressure rise to 35 mmHg, the 450 mL drainage and the venous oxygen tension of the drained fluid. Verified = No.
- Giorgetti R, Chiricolo G, Melniker L, Calaf C, Gaeta T. RESCUE transesophageal echocardiography for monitoring of mechanical chest compressions and guidance for extracorporeal cardiopulmonary resuscitation cannulation in refractory cardiac arrest. Journal of Clinical Ultrasound 2019;48(3):184–187. DOI 10.1002/jcu.22788. The compression-vector shift over the outflow tract and aortic root, and the pre-dilation wire confirmation. Verified = No.
- Parker BK, Salerno A, Euerle BD. The use of transesophageal echocardiography during cardiac arrest resuscitation: a literature review. Journal of Ultrasound in Medicine 2018;38(5):1141–1151. DOI 10.1002/jum.14794. 55 publications, 277 patients; direction-changing but outcome-unproven; the prolonged pulse-check finding. Verified = No.
- Ellouze O, Soudry Faure A, Radhouani M, et al. Levosimendan in venoarterial ECMO weaning: rationale and design of a randomized double blind multicentre trial. ESC Heart Failure 2021;8(4):3339–3347. DOI 10.1002/ehf2.13427. Used only for its background statements: weaning failure as high as 50% even when criteria are fulfilled, the absence of standardised criteria, the clinical weaning thresholds, and the right ventricular weaning parameters from a separate cohort. Verified = No. NOTE: this trial's result is not used anywhere, and the levosimendan weaning meta-analysis named in this book's outstanding-retrieval list remains unread; its direction is still deliberately not stated.
- Kim MC, Lim Y, Lee SH, et al. Early left ventricular unloading after extracorporeal membrane oxygenation: rationale and design of the EARLY-UNLOAD trial. ESC Heart Failure 2023;10(4):2672–2679. DOI 10.1002/ehf2.14450. Used only for its rescue-unloading trigger table, which is a clean published statement of the echocardiographic and waveform criteria a randomised trial accepted as evidence of left ventricular afterload increase. Verified = No.
- Golzarian H, Thiel A, Hempfling G, et al. Severe aortic insufficiency-induced cardiogenic shock treated with left atrial VA-ECMO and emergent valve-in-valve TAVR. ESC Heart Failure 2023;10(6):3718–3724. DOI 10.1002/ehf2.14561. Cited as the worked instance of the aortic regurgitation branch in §26.7. Verified = No.
- Reddan T, et al. Australasian Journal of Ultrasound in Medicine 2020;23(4):255–263. DOI 10.1002/ajum.12206. Ultrasonographic definition of the mixing zone in the abdominal aorta of a 14-year-old on femoro-femoral venoarterial support. Verified = No. Carried forward from Chapter 25's research; author list incomplete in the search record.
- Wang W, Liu B, Wang L. Critical care echocardiography: a bibliometric analysis of knowledge structure and evolution of development trends from 2005 to 2024. Echocardiography 2025;42(6). DOI 10.1111/echo.70204. Used only as a secondary statement of the transoesophageal case in cardiac arrest; no substantive claim rests on it. Verified = No.
Named in the text and not retrieved
- Aissaoui N, et al. — the original 51-patient weaning cohort from which the three standard parameters derive. Known only through the Red Book, the 2022 review and the Aarhus paper. This is the highest-value outstanding retrieval for this chapter.
- Ortuno S, et al. — the systematic review cited by the 2022 review for the aortic valve velocity–time integral over 10 cm, ejection fraction over 20% and lateral mitral annular S′ over 8 cm/s combination, and for the right ventricular weaning parameters including the finding that tricuspid annular excursion and tricuspid regurgitation were not associated with successful right ventricular weaning. Not retrieved. Note that the S′ threshold quoted through this route (8 cm/s) differs from the threshold quoted through the Red Book (6 cm/s); both are recorded and neither is asserted.
- Kim et al. — the cohort in which 30% of unweanable and 16% of successfully weaned patients met all three criteria. Not retrieved. The most consequential unverified figure in this chapter.
- Huang et al. — the right ventricular ultrasound weaning study (three-dimensional right ventricular ejection fraction over 24%, fractional area change over 25%), cited through the levosimendan protocol paper. Not retrieved.
- Wang et al. — the single-sheep dual-lumen recirculation experiment (2.1% to 51.3%), cited through the 2012 dual-lumen series. Not retrieved.
- Aarhus reference 19 — the haemodynamic transoesophageal weaning protocol cited by Red Book Chapter 29 for the absence-of-distension endpoint. Not identified.
Chapter status
Published 18 September 2026. Second chapter of Part V. Chapter 26 of 92.
What was verified. The ELSO Red Book Chapters 48 and 29, the ELSO 2015 ultrasound guidance, the ISCCM Manual Chapter 33 and the practical-guide textbook were retrieved through this project's document store and are marked Verified = Yes. Every external reference was read as an indexed full-text passage returned by a passage-retrieval service, not as a complete article; no digital object identifier was independently resolved; every database row seeded from this chapter is marked Verified = No.
Numerical and bibliographic problems flagged and not corrected.
- The Kim reversal (30% versus 16%) is reported exactly as the 2022 review reports it. The primary paper was not retrieved and the direction of the finding is consequential enough that it is labelled [VERIFICATION REQUIRED] in §26.12.
- Two different thresholds for lateral mitral annular systolic velocity appear in the literature reaching this chapter: 6 cm/s through the Red Book and the Aarhus study, 8 cm/s through the systematic review as summarised by the 2022 review. Both are recorded in §26.12 and the reference list; neither is asserted as the threshold.
- The dual-lumen malposition reference carries an internal date discrepancy — a 2015 publication date on a 2013 identifier with 2013 in the volume numbering — recorded at reference 11 and not resolved.
- The ELSO ultrasound guidance expired in May 2018 by its own text and is used throughout. This is recorded in §26.18 and at reference 2, and it remains the only ECMO-specific ultrasound guidance available to this book. The expiry is a standing currency gap already recorded in the book status document.
Running total of flagged published numerical or bibliographic problems: Ch 15 (1), Ch 17 (3), Ch 18 (2), Ch 19 (3), Ch 24 (6), Ch 26 (4).
Audits passed: clinical, physiology, evidence, citation, numerical, safety, contradiction, redundancy, bedside utility, literature update.