Chapter question: The saturation has fallen and will not come up — why, and what do I do about it?
Chapter 8 gave you the round. This chapter gives you the response when the patient on established VV ECMO becomes and stays hypoxaemic — the six causes, how to tell them apart at the bedside, and the escalation ladder when none of them is correctable.
Evidence search date: 6 September 2026. Primary sources: ELSO Red Book 6th edition (Chapter 7, Circuit Malfunction and Crisis Management); ECMO in the Adult Patient (Core Critical Care, 2017); ECPR and Resuscitative ECMO; Comprehensive Healthcare Simulation: ECMO Simulation; Taha, ECMO: A Practical Guide to Management; ISCCM Manual; and Brown et al., Perfusion 2025.
What this chapter covers — and what it does not
This chapter owns | Deferred to |
The three-compartment first move; the six causes of hypoxaemia on established VV ECMO and how to distinguish them; recirculation at the bedside; falling flow at constant pump speed and access insufficiency; chatter; the escalation ladder and what each step costs; beta-blockade for refractory hypoxaemia.
Sections 9.4 to 9.8 give a triage-level summary of the circuit failures that present as acute hypoxaemia — enough to recognise them and act at the bedside. Part VI owns them in depth. | Recirculation physiology and the four determinants of SaO₂ → Chapter 6
The daily surveillance dataset → Chapter 8
Weaning and decannulation → Chapter 10
Circuit thrombosis → Chapter 31
Oxygenator failure and exchange → Chapter 32
Pump failure → Chapter 33
Cannula problems → Chapter 34
Air and circuit emergencies → Chapter 35
Bleeding → Chapter 36
Haemolysis → Chapter 37
Mechanical complications → Chapter 38
Differential hypoxaemia (a VA problem) → Part III
Anticoagulation → Part VII
Ventilation strategy → Chapter 53
Printable algorithms and tables → Chapter 89 and Appendix 8 |
9.1 The first move is always the same
Hypoxaemia on established VV ECMO is not one problem. It is a final common pathway for at least six, and they are separated at the bedside by three measurements — the drainage-line saturation, the post-membrane saturation, and the mixed venous saturation — none of which takes more than a few minutes to obtain.
Before any of that, though: every deterioration on ECMO is one of three things, and the order in which you consider them decides how fast you get to the answer:
- The measurement — is the number real? A displaced saturation probe, a damped arterial line, a flow probe without coupling gel, a blood gas drawn from the wrong port. Excluding this costs ten seconds and prevents a great deal of harm.
- The circuit — is the machine still delivering what it was delivering an hour ago?
- The patient — has the disease, the physiology or the anatomy changed?
Clinical pearl — the flow probe mimics catastrophe
The ELSO Red Book makes this point about console failure specifically, and it generalises: when a console failure is suspected, a flow probe malfunction must first be excluded, as this can mimic the appearance of console failure. A flow probe that has lost its coupling gel, or been knocked off the tubing, displays a flow of zero on a circuit that is running perfectly.
Look at the patient and at the circuit before you believe the console. Blood that is bright red after the membrane and dark before it is a circuit that is working, whatever the screen says.
Source: ELSO Red Book 6th ed., Ch 7.
Danger — one team for the circuit, a different team for the patient
The ELSO Red Book states the rule plainly: the personnel responsible for troubleshooting the ECLS circuit should not simultaneously be responsible for patient management. In a circuit crisis the patient still needs ventilation increased, volume given, vasopressors started and blood sent for — and the person with their hands on the clamps cannot do any of that.
Say it out loud at the start of the event: "You have the circuit. I have the patient."
Source: ELSO Red Book 6th ed., Ch 7.
Pitfall — ΔP on ECMO is not TMP on CRRT
If your unit runs CRRT alongside ECMO (Chapter 55), two different pressure concepts sit on two screens a metre apart. On the ECMO circuit, ΔP is simply pre-oxygenator minus post-oxygenator pressure, and there is no consensus alarm threshold. On the CRRT machine, transmembrane pressure is a different calculation involving the effluent pressure, and the machine alarms at manufacturer-set values.
They are not interchangeable, the numbers are not comparable, and a CRRT alarm is not an ECMO finding.
9.2 Hypoxaemia on VV ECMO
This is the commonest call, and it has exactly six causes. Working through them in order takes about two minutes and is far faster than guessing.
Cause | Specific mechanisms | How you recognise it |
1. Reduction or loss of circuit flow | Low intravascular volume; kink in circuit tubing; obstruction from large thrombus in circuit, oxygenator or cannula; cannula malposition; cardiac tamponade; tension pneumothorax | Flow has fallen — see §9.3 |
2. Post-oxygenator blood not fully saturated | Failing oxygenator; accidental interruption of sweep gas supply | Post-membrane blood gas or colour — see §9.4 and §9.5 |
3. Increased recirculation | Circuit flow too high; suboptimally positioned cannula; low cardiac output; low intravascular (specifically right atrial) volume | Rising pre-oxygenator (drainage) saturation with falling arterial saturation |
4. Increase in cardiac output | Sepsis; inotropic drug therapy | A larger fraction of venous return bypasses the circuit — the ratio of ECMO flow to cardiac output has fallen |
5. Increase in oxygen consumption | Inadequate sedation; seizures; fever | Falling mixed venous saturation with unchanged circuit parameters |
6. New or worsening lung problem | Malpositioned endotracheal tube (bronchial position or accidental extubation); pneumothorax; segmental lung collapse; worsening consolidation; pulmonary oedema; haemothorax; pulmonary haemorrhage | Examination, airway pressures, chest imaging, suction findings |
Adapted from Table 8.2, ECMO in the Adult Patient (Core Critical Care, 2017).
9.2.1 Recirculation — the one where the intuitive move is wrong
Physiology — why turning the pump down can raise the saturation
Recirculation is oxygenated blood returning from the circuit being drawn straight back into the drainage cannula without ever reaching the systemic circulation. Increasing pump speed increases the negative pressure at the drainage cannula, which pulls a larger share of the return flow back in. Past a certain point, each extra revolution per minute adds more recirculation than it adds effective flow.
The consequence is stated directly in the source: "in these circumstances, reducing pump speed may actually result in improved oxygenation." This is the single most counter-intuitive manoeuvre in VV ECMO, and the one most often not tried.
The physiology behind effective flow and the four determinants of arterial saturation is Chapter 6; this section is only the bedside response.
Source: ECMO in the Adult Patient (Core Critical Care, 2017).
Recirculation rises with a suboptimally positioned cannula — the source gives the best distance between the tips of two cannulae as approximately 10 cm — and with low cardiac output and low right atrial volume. So the three corrective moves, in order, are: give volume if the patient is dry, check and correct cannula position, and only then consider the pump speed in either direction.
Physiology — cardiac output appears twice in the list, in opposite directions. Both are correct.
A rising cardiac output (cause 4) lowers arterial saturation because a smaller proportion of the total venous return passes through the circuit: the fixed extracorporeal flow is diluted by more native flow.
A falling cardiac output (cause 3) lowers arterial saturation by a different route: less venous return sweeps past the drainage cannula, so a greater share of what the pump draws in is the oxygenated blood it has just returned — recirculation.
The two are distinguished at the bedside by the drainage-line saturation, which is why it sits at the branch point of the algorithm above. Recirculation raises it; a high cardiac output does not. Guessing between them without that measurement is how the wrong pump-speed change gets made.
9.2.2 The escalation ladder, and its costs
When the six causes have been worked through and the patient is still hypoxaemic, the options below are available. Each has a price, and the price is the reason they sit in this order.
Step | What it costs |
Confirm sweep gas is 100% oxygen | Nothing. Do it first, every time. |
Reduce oxygen consumption — sedation, cooling, treat fever, treat seizures, consider paralysis | Sedation depth, and the loss of the neurological examination (Chapter 8) |
Increase ECMO blood flow — or reduce it, if recirculation is suspected | More negative drainage pressure and haemolysis if increased; less effective flow if the diagnosis was wrong |
Transfuse to a higher haemoglobin target to raise oxygen delivery | Inflammation, renal injury, circuit load, immunological cost |
Temporarily increase ventilator FiO₂ | Oxygen toxicity; and it works against the lung-rest principle, so it is a bridge, not a plan |
Adjuncts — prone positioning, inhaled nitric oxide | Staffing, cannula risk during turning; the proning evidence is contested (Chapters 7 and 53) |
Add a second drainage cannula to achieve higher flow | See the danger callout below |
Add a second oxygenator | Circuit complexity, additional surface area, additional connections |
Accept a lower target — PaO₂ around 6 kPa (50 mmHg), saturation around 85% | Nothing, if delivery markers are satisfactory. This is often the right answer. |
Escalation options from Table 8.2, ECMO in the Adult Patient (Core Critical Care, 2017); costs are this book's synthesis.
Danger — a second drainage cannula is a temporary measure, not a solution
Adding a second drainage cannula is reasonable when the problem is genuinely an undersized cannula — recognisable as stable but highly negative inlet pressure. But Taha and colleagues are explicit that it should be a temporary measure: the differential flows across two drainage lines and the turbulent flow caused by the Y-connector can lead to circuit thrombosis. And the site chosen matters, because a badly placed second cannula decreases effective flow by increasing recirculation — the opposite of the intended effect.
Source: Taha et al., ECMO: A Practical Guide to Management.
Pitfall — filling the patient to raise the saturation
Positive fluid balance raises right atrial volume, reduces recirculation and permits higher flows, so it works — for a few hours. Taha and colleagues warn against it directly: persistent positive fluid balance maintained in order to hold high ECMO flows and raise saturation should be avoided in the absence of organ dysfunction, because ARDS management requires conservative fluid balance, and the risk of a lower PaO₂ must be weighed against the known harms of fluid overload and the increased blood trauma at higher flows.
A single fluid bolus to diagnose and correct genuine hypovolaemia is a different thing from a positive daily balance adopted as an oxygenation strategy.
Controversy 1 — Should beta-blockade be used for refractory hypoxaemia on VV ECMO?
The question. When hypoxaemia is driven by a high cardiac output, only a fraction of the venous return passes through the circuit. Slowing the heart should raise that fraction and therefore raise arterial saturation. Should beta-blockers be given for this purpose?
The case for. The arithmetic is straightforward and the effect on saturation is real. Reducing cardiac output raises the ratio of extracorporeal flow to total flow, so a greater proportion of the blood reaching the systemic circulation has passed through the membrane. Case reports and physiological reasoning both support it, and it is used in some centres.
The case against. Arterial saturation is not the endpoint that matters. Cardiac output is one of the two terms in oxygen delivery, so lowering it to raise the other term may leave the product unchanged — or worse.
What the evidence actually shows. Brown and colleagues (Perfusion, 2025) modelled this in silico across three scenarios. With a high cardiac output and a partial lung shunt, SaO₂ rose from 74.2% to 79.2% but mixed venous saturation fell from 53.5% to 44.7%. With a high cardiac output and a complete lung shunt, SaO₂ rose from 71.9% to 85%. But with a normal cardiac output and high recirculation, SaO₂ actually fell from 82.4% to 78.3%, and mixed venous saturation collapsed from 50.8% to 25.5%. Crucially, oxygen delivery was reduced in every scenario. The authors conclude that the effects "are unpredictable and may reduce oxygen delivery" and state that the study "does not support the use of beta-blockers for this indication." Certainty: low — this is a mathematical model, not a clinical trial, and no randomised or observational outcome data exist.
Where practice actually sits. Beta-blockade for this indication is uncommon and has never been standard. The modelling makes the mechanism of harm explicit: it can improve the number you are looking at while worsening the physiology you care about, and it does the most damage in exactly the situation — recirculation — that is easily mistaken for high cardiac output at the bedside.
What would resolve it. A clinical study with oxygen delivery and tissue-oxygenation endpoints rather than saturation. None is known to be underway. [VERIFICATION REQUIRED] — trial registries were not searched for this chapter.
The transferable lesson. This is the clearest available demonstration of the book's recurring rule: an intervention that raises SaO₂ has not necessarily helped the patient. Before accepting any manoeuvre that improves saturation, ask what it did to cardiac output and to haemoglobin.
9.3 Falling blood flow at constant pump speed
This is the second commonest call, and unlike hypoxaemia it has a formal published algorithm. The logic is a sequence of exclusions, and the circuit pressures tell you where the problem is before you have found it.
Step 1 — Exclude access insufficiency. Sudden blood flow reduction at constant pump speed is most often caused by reduced blood return to the pump. The ELSO Red Book names this access insufficiency and puts it first for a reason: it is the most likely cause and the most easily corrected.
Step 2 — On VA ECMO, exclude a rise in afterload. Blood flow also falls if mean arterial pressure rises, increasing pump afterload. This is a VA phenomenon (Part III), listed here so the VV reader knows why it is not on their list.
Step 3 — If both are excluded, the cause is increased resistance. Usually clot formation in the circuit, aspiration of a clot from the venous system, or a kink.
Step 4 — Read the pressures. They localise the obstruction.
Pressure pattern | Obstruction is located |
Drainage pressure becomes more negative than baseline | Between the drainage cannula and the pump inlet |
Pre- and post-membrane pressures rise by similar amounts | Between the membrane lung and the return cannula |
The pre- to post-membrane differential (ΔP) increases | Within the membrane lung itself |
Source: ELSO Red Book 6th ed., Ch 7 and Figure 7-5.
Step 5 — Perform a methodical, end-to-end circuit check. Kinks, visible clot in the oxygenator, anything under the patient, anything behind the pump, every connector. This is a physical walk of the whole circuit, not a glance at the console.
Step 6 — If no correctable cause is found, an emergency circuit exchange is usually indicated.
Clinical pearl — when the problem survives the circuit change
Rarely, flow remains poor after a complete circuit exchange. In that case the clot is not in the circuit — it is lodged in the cannula, which was not replaced. Aspirating the cannula with a large syringe may or may not resolve it. Knowing this in advance saves a great deal of confusion at the moment it happens.
Source: ELSO Red Book 6th ed., Ch 7.
9.3.1 Chatter, and what it is really telling you
Line chatter — visible shaking or juddering of the drainage tubing, usually with intermittent very negative access pressure — is the mechanical signature of a drainage cannula that is repeatedly sucking against something. It is the bedside form of access insufficiency.
The causes, in rough order of frequency: relative hypovolaemia; raised intrathoracic or intra-abdominal pressure; coughing or straining; cannula tip against a vessel wall or migrated; pump speed set higher than the drainage can support. The ISCCM manual notes the association of chattering with hypovolaemia and with haemolysis in the same breath, which is the point: chatter is not merely a nuisance, it is red cells being sheared.
The response, in order: reduce pump speed first (this stops the injury immediately and buys time), then diagnose — volume status, abdominal pressure, cannula position, patient position — then correct, then return the speed.
Danger — do not treat chatter by turning the pump up
The instinct when flow falls is to increase RPM. In access insufficiency this makes the drainage pressure more negative, worsens the chatter, worsens the haemolysis, and does not increase flow — the limit is on the drainage side, not the pump side. Chapter 8's threshold applies: an access pressure more negative than about −100 mmHg is causing blood trauma.
9.4 Sudden, total loss of gas exchange
Clinical pearl — the colour test takes two seconds and needs no equipment
The ELSO Red Book's first move for the abruptly deteriorating ECLS patient is to assess the colour differential between pre- and post-membrane blood. If both are dark red, no gas exchange is occurring in the membrane.
The diagnostic power comes from the timing: a decline in membrane lung efficiency normally occurs over hours to days. An abrupt loss of gas exchange therefore is not a failing membrane — it is an interruption of fresh gas flow. Trace the gas line from the membrane lung back to the wall or tank.
Source: ELSO Red Book 6th ed., Ch 7.
The commonest cause is mundane and preventable. The Red Book describes it precisely: a patient returning from an intrahospital transport using compressed gas, where the operator fails to return the gas line to the wall source. The cylinder empties at some unpredictable later point, and the patient deteriorates abruptly with no other explanation. The fix is a checklist on return from any journey out of the room — and, from the ECPR text, two further habits: a sign over the sweep gas wall regulator to prevent accidental adjustment by staff who do not know what it is, and a ball-drop test (briefly pinch the sweep tubing and watch the regulator ball drop) to confirm gas is actually flowing.
9.5 Membrane lung dysfunction — and the exchange threshold
Unlike the catastrophes, this is gradual, and Chapter 8 covered its surveillance. Chapter 9's question is narrower: when does the finding become an action?
Two independent signals, which decline separately:
- Resistance. A rise in ΔP at constant pump flow indicates increasing resistance. Where pressures are not measured, the same information appears as increased pump speed being required to maintain the same flow.
- Transfer. Membrane lung efficiency is determined by calculating oxygen transfer.
Evidence — the one published exchange threshold, and why it sits uneasily
The ELSO Red Book states: when oxygen transfer is below 150 mL/min, membrane lung exchange should be considered. This is the only quantitative change-out threshold this book has found in a major source, and it is a consider, not a mandate.
It must be read against Chapter 8's finding. The 2024 dedicated review of oxygenator monitoring and change-out (Butt et al.) surveyed the same territory and concluded that no validated change-out criteria exist, offering no transmembrane-pressure and no oxygen-transfer threshold.
These are not in direct contradiction. The Red Book offers a pragmatic consensus trigger for consideration; the review's point is that no such trigger has been validated against outcomes. Both are true, and together they describe the actual state of knowledge: there is a number you can use to prompt a decision, and no evidence that the number is correct.
Certainty: very low (expert consensus, single supporting reference not retrieved for this chapter).
Excessive clot in the membrane lung also declares itself away from the circuit: as consumptive coagulopathy, and as haemolysis — which may equally arise from clot in the pump head (Chapter 37).
Clinical pearl — an upper limit on post-membrane oxygen tension
Post-membrane PO₂ should remain below 600 mmHg. Above that, supersaturation allows oxygen to come out of solution within the circuit and form bubbles — a rare but genuine mechanism of air embolism that is generated by the circuit itself rather than entrained from outside.
This is a reason to run the sweep oxygen concentration at the minimum that achieves the target, rather than leaving it at 100% indefinitely once the patient is stable.
Source: Comprehensive Healthcare Simulation: ECMO Simulation (Johnston & Su), citing its own primary reference. [VERIFICATION REQUIRED] — primary source not retrieved.
9.6 The circuit catastrophes — triage summary
These are rare, fast, and drilled rather than reasoned through. They appear here because each of them can present as sudden profound hypoxaemia, and the bedside clinician must recognise and act before anyone has time to consult a chapter. This table is the immediate action only. Each event has its own chapter in Part VI — circuit thrombosis (31), oxygenator failure (32), pump failure (33), cannula problems (34), air and circuit emergencies (35), mechanical complications (38) — where the mechanisms, the exchange procedures and the prevention are covered properly.
The unifying protocol from the ECPR literature is CLAMP, STOP, RESCUE: clamp the circuit, stop the pump, support the patient by other means.
Event | Immediate action | Then |
Air in the circuit | Clamp; stop the pump — this stops air moving towards the patient | De-air via the membrane lung de-airing port or luer access points with syringes; if unable to de-air, or if the blood has foamed, exchange to a new primed circuit. Lower the head of the bed. See §9.7 |
Accidental decannulation | Clamp return and drainage tubing; stop flow; call for help | Direct pressure to the site; hold the anticoagulant; use the bridge if present; prepare a new cannula. Partial displacement in a fully ECMO-dependent patient may justify attempted reinsertion — a high-risk act for experienced operators only |
Circuit rupture or breach | Temporise — clamp pigtails, occlusive tape, bone wax, a sterile gloved finger | Isolate the breach between two clamps and replace the component. Breach before the pump entrains air; breach after the pump bleeds |
Pump head failure or decoupling | Clamp; stop | Reseat the pump head; usually an emergency circuit exchange is required. Suspect it when there is abnormal pump noise, excessive vibration, or a sudden fall in RPM |
Console or power failure | First exclude a flow probe malfunction; then clamp to prevent retrograde flow | Confirm the impeller is not spinning, transfer the pump head to the hand crank or backup console |
Catastrophic circuit clot | Clamp; stop | Exchange the affected component or the whole circuit |
Composite of ELSO Red Book 6th ed. Ch 7 and ECPR and Resuscitative ECMO Ch 10.
Danger — the hand crank has exactly two indications
The ELSO Red Book is unambiguous: loss of power and console failure are the only indications to employ the hand crank. If the pump motor is still functioning, switching to the hand crank "will only cause delays and distract from other troubleshooting."
The hand crank is not a general-purpose response to low flow, to a clotted circuit, or to a pump alarm. Reaching for it in those situations makes the situation worse.
Clinical pearl — what should actually be available
The Red Book corrects a common practice: keeping blood products in the room is described as excessive and unnecessary. What is required instead is crossmatched blood available in the blood bank for every ECMO patient, and a massive transfusion protocol that makes at least four units of uncrossmatched packed red cells available for immediate use.
Alongside: two tubing clamps on the cart at all times, backup equipment (secondary motor or hand crank), cart brakes locked, console screen visible from the doorway, and the door left able to transmit the alarm.
Sources: ELSO Red Book 6th ed. Ch 7; ECPR and Resuscitative ECMO Ch 10.
9.7 Air in the circuit — triage summary
Air is separated out from the table above because it is the one circuit catastrophe whose prevention is largely in the hands of the bedside team, and because the mechanism by which it reaches a VV patient is widely misunderstood. Chapter 35 owns this subject; what follows is what must be known at the bedside.
Physiology — where the air comes from, and why VV is not exempt
Air enters through the negative-pressure (pre-pump) side — excessive suction, an open or defective stopcock or pigtail, a breach, or cannula dislodgement. It can also be generated within the circuit by supersaturation (§9.5).
A large volume causes pump head airlock: the pump de-primes and blood flow ceases altogether.
VV ECMO returns blood to the venous side, so entrained air should in principle be filtered by the pulmonary circulation. It is not a guarantee: air can reach the systemic circulation via a patent foramen ovale, which is present in a substantial minority of adults and which right-heart pressures in ARDS make more likely to shunt right-to-left. Lowering the head of the bed is recommended for this reason in both VA and VV.
Source: ECPR and Resuscitative ECMO, Ch 10.
The Red Book's own framing is worth stating plainly: this complication is nearly always avoidable. The preventive measures are limiting access points on the negative-pressure side, caution with central lines, and meticulous stopcock technique. Modern systems carry at least one bubble detector; where its position is user-configurable, placing it before the membrane lung alerts earlier and allows more time to intervene.
Danger — three ways the team itself puts air or wire into the circuit
- Central line insertion. Negative pressure in the drainage cannula can overcome the resistance of one-way valves on a central line or sheath and entrain ambient air. Reduce pump speed before the procedure and minimise open atmospheric connections.
- Guidewire aspiration. A guidewire can be sucked into the ECMO circuit — rare, and catastrophic. The risk rises with a high drainage cannula tip and high pump speed. Hold the wire firmly throughout, minimise insertion depth, and reduce pump speed.
- Stopcocks and pigtails on the pre-pump limb. The commonest route of all, and the reason some units delete the access-line pressure port entirely (Chapter 8).
Source: ECPR and Resuscitative ECMO.
Pitfall — clamping order matters, and the exception matters more
Simulation curricula teach the clamping sequence VBA — venous, bridge, arterial: clamp the venous (drainage) line, unclamp the bridge, then clamp the arterial (return) line. The stated exception is the one you must remember: when air is detected on the return side, clamp the return line immediately, ahead of everything else, to prevent air reaching the patient.
Circuits without a bridge simplify this, but the principle survives: air heading towards the patient is clamped first.
Source: Comprehensive Healthcare Simulation: ECMO Simulation.
9.8 How often do these things actually happen?
Evidence — frequencies, reported at second hand
The ECMO simulation literature reports pump failure at approximately 1% of complications and air embolism at approximately 4%, and describes clots in the ECMO circuit as the most common mechanical complication during an ECMO run.
Certainty: very low. These figures are quoted from that text's own cited primary sources, which have not been retrieved and read for this chapter, and no denominator, population or era is given for either percentage. They are reproduced to convey rank order — clot is common, air is uncommon, pump failure is rare — and should not be quoted as incidence rates.
[VERIFICATION REQUIRED] — primary sources not retrieved.
The rank order is itself clinically useful. It says that the events a team should drill most are not the dramatic ones. Clot and access insufficiency are what will actually happen; air and pump failure are what everyone rehearses. A troubleshooting curriculum weighted towards catastrophe trains for the rare event and leaves the common one to improvisation.
9.9 Controversy 2 — Should there be a numeric trigger for membrane lung exchange?
Controversy 2 — A validated exchange threshold does not exist. Should units use one anyway?
The question. Should a unit adopt a numeric trigger for oxygenator exchange — the ELSO Red Book's oxygen transfer below 150 mL/min, a local ΔP rise, a plasma-free haemoglobin level — or make each decision on the whole trajectory?
The case for a trigger. It converts emergency exchanges into planned ones, standardises behaviour across shifts and operators, and gives junior staff a defensible action at 3 a.m. The Red Book's 150 mL/min figure exists precisely because clinicians need something to act on.
The case against. No threshold has been validated against outcomes. Absolute ΔP values are device- and flow-dependent. A trigger will cause exchanges that were not needed, each carrying transfusion, an interruption of support, and cost.
What the evidence actually shows. Two authoritative sources reach different-sounding conclusions from the same literature: the ELSO Red Book offers 150 mL/min oxygen transfer as a threshold for consideration; Butt et al. (2024) conclude that no validated criteria exist and decline to propose any, holding that the parameters are informative only collectively. These are compatible: one is a pragmatic prompt, the other a statement about evidence. Neither is derived from outcome data. Certainty: very low. No comparative study of protocolised versus judgement-based exchange exists.
Where practice actually sits. Most experienced units measure and plot a defined parameter set daily (uncontroversial), use a figure like 150 mL/min as a prompt to review, and leave the decision itself to an experienced clinician weighing the trajectory, the patient's current dependence on the circuit, the time of day, and staffing.
What would resolve it. A multicentre registry linking serial ΔP, oxygen-transfer and haemolysis trajectories to hard oxygenator-failure events, sufficient to derive and then prospectively validate a trigger. Nothing of this kind currently exists.
This controversy extends the one opened in Chapter 8; the two should be read together.
9.10 The errors that recur in ECMO troubleshooting
Error | Correction |
Believing the console before looking at the circuit | Exclude the measurement first — a flow probe without gel reads zero on a working circuit |
Increasing pump speed for chatter or access insufficiency | Reduce speed first to stop the shear, then diagnose the drainage problem |
Never trying a reduction in speed for suspected recirculation | In recirculation, reducing pump speed may improve oxygenation |
Reaching for the hand crank during a low-flow or clot event | Power loss and console failure are its only two indications |
One person managing both the circuit and the patient during a crisis | Two teams, stated out loud: "You have the circuit, I have the patient" |
Treating an abrupt loss of gas exchange as a failing membrane | Membranes fail over hours to days. Abrupt means the gas supply — check the line, the wall, the cylinder |
Correcting saturation with fluid, day after day | A diagnostic bolus is not a fluid strategy; positive balance to hold flows is explicitly advised against |
Accepting an intervention because SaO₂ improved | Ask what it did to cardiac output and haemoglobin — delivery is the endpoint (Controversy 1) |
9.11 Key points
- Every deterioration is the measurement, the circuit, or the patient — considered in that order. Excluding a false number costs ten seconds.
- One team on the circuit, a different team on the patient. State it out loud.
- Hypoxaemia on VV ECMO has six causes: lost flow, unsaturated post-membrane blood, recirculation, rising cardiac output, rising oxygen consumption, and new lung pathology. Work through them rather than guessing.
- In recirculation, reducing pump speed may improve oxygenation — the one manoeuvre that feels wrong and is right. Cannula tip separation of about 10 cm, adequate right atrial volume and a normal cardiac output all reduce recirculation.
- Falling flow at constant RPM is access insufficiency until proven otherwise. The pressures then localise the obstruction: drainage more negative means pre-pump; pre and post rising together means post-membrane; a widening ΔP means within the membrane.
- Do not answer chatter by turning the pump up. Reduce speed, diagnose the drainage problem, correct it, then restore flow.
- Abrupt loss of gas exchange is a gas-supply problem, because membranes fail over hours to days. Check the colour of pre- and post-membrane blood, then trace the gas line to the wall.
- The hand crank has two indications only — loss of power and console failure. Exclude a flow probe malfunction before either.
- CLAMP, STOP, RESCUE for every circuit catastrophe; clamp the return line first when air is heading for the patient.
- Air in the circuit is nearly always avoidable, and the team is frequently its source — central lines, guidewires, and pre-pump stopcocks.
- Oxygen transfer below 150 mL/min is the one published prompt to consider membrane exchange; there is no evidence that it is the right number.
- Clot is the common mechanical problem; air and pump failure are rare. Drill in proportion to what will actually happen.
[VERIFICATION REQUIRED] — open items in this chapter
- The 150 mL/min oxygen transfer threshold is cited by the ELSO Red Book to its own reference, which has not been retrieved. It is reproduced as expert consensus, not validated evidence, and Chapter 8's contrary finding (Butt 2024: no validated criteria) is presented alongside it.
- The post-membrane PO₂ ceiling of 600 mmHg is quoted from the ECMO simulation text, citing a primary source not retrieved.
- The ~1% pump failure and ~4% air embolism figures are second-hand through the same text, with no denominator, population or era stated. Treated as rank order only.
- The ~10 cm optimal cannula tip separation is quoted as "said to be" in its source, which is itself hedged language. It is a rule of thumb, not a measured optimum.
- PaO₂ around 6 kPa / SaO₂ around 85% as generally adequate is from a 2017 textbook; no trial has tested an oxygenation target in VV ECMO (Chapter 8, Controversy 2).
- Brown et al. 2025 is an in-silico model, not a clinical study. Its numbers describe the behaviour of a simulation. Full author initials, and whether an accompanying clinical dataset exists, were not further verified.
- ELSO Red Book 6th edition figures 7-5 and 7-6 are algorithm diagrams; their content is described here from the surrounding text, as the figures themselves could not be read.
- Frequency and ordering claims in §9.8 and Key Point 12 are inference from reported percentages, not a measured comparison.
Cross-references
- Chapter 6 — VV ECMO Physiology: recirculation, effective flow, and the four determinants of arterial saturation — the physiology this chapter applies
- Chapter 7 — Initial VV ECMO Management: the first 24 hours, and the early complications that present before a routine has been established
- Chapter 8 — Daily VV ECMO Management: the surveillance that should detect most of these problems before they become emergencies
- Chapter 10 — VV ECMO Weaning and Decannulation
- Chapter 31 — Circuit Thrombosis: the commonest mechanical problem, and the one §9.3 most often ends at
- Chapter 32 — Oxygenator Failure: the exchange procedure, and the threshold debate of §9.5 in full
- Chapter 33 — Pump Failure
- Chapter 34 — Cannula Problems: malposition, migration and the vascular complications behind §9.3.1
- Chapter 35 — Air and Circuit Emergencies: the full treatment of §§9.6 and 9.7
- Chapter 36 — Bleeding
- Chapter 37 — Haemolysis: the consequence of much of what this chapter describes
- Chapter 38 — Mechanical Complications
- Chapter 53 — Mechanical Ventilation During ECMO: proning and the ventilator adjuncts referenced in §9.2.2
- Chapter 55 — CRRT During ECMO: the source of the ΔP/TMP confusion warned about in §9.1
- Part III — VA ECMO: differential hypoxaemia, afterload-related flow reduction, and the VA-specific catastrophes
- Part VII — Anticoagulation: preventing the circuit thrombosis this chapter responds to
- Chapter 89 — ECMO Troubleshooting Algorithms and Appendix 8 — Troubleshooting Tables: the printable, bedside-ready forms of the algorithms reasoned through here
- Chapters 87–88 — Practical Handbook: the emergency cart contents and the crisis action cards
References
- Extracorporeal Life Support: The ELSO Red Book, 6th edition. Chapter 7, Circuit Malfunction and Crisis Management (including Figure 7-5, algorithmic approach to a decrease in extracorporeal blood flow, and Figure 7-6, membrane lung dysfunction). Two-team principle; console and power failure; flow probe exclusion; hand crank indications; circuit air; loss of circuit integrity; access insufficiency and the pressure-localisation rule; gas supply interruption; pump head failure; membrane lung dysfunction and the 150 mL/min oxygen transfer threshold; blood availability.
- ECMO in the Adult Patient (Core Critical Care series), 2017. Table 8.2, assessment and management of hypoxaemia during venovenous ECMO; recirculation determinants; cannula tip separation; reduction of pump speed in recirculation; oxygenation targets. [VERIFICATION REQUIRED] — editors, publisher and page numbers not verified in this session and deliberately not stated.
- Shinar Z, Badulak J, eds. ECPR and Resuscitative ECMO. Chapter 10. Circuit catastrophes; CLAMP–STOP–RESCUE; air entrainment and de-airing; accidental decannulation; circuit rupture; pump failure; backup equipment list; ball-drop test; sweep regulator signage; central line and guidewire hazards.
- Johnston LC, Su L, eds. Comprehensive Healthcare Simulation: ECMO Simulation. Springer. Complication frequencies; VBA clamping sequence and the arterial-air exception; post-membrane PO₂ ceiling; mobilisation-related events. [VERIFICATION REQUIRED] — year and page numbers not confirmed; the quantitative figures are second-hand from this text's own references.
- Taha AR, Caridi-Scheible M, Leiendecker E, et al. ECMO: A Practical Guide to Management. Second drainage cannula as a temporary measure and its thrombosis risk; the argument against positive fluid balance as an oxygenation strategy. [VERIFICATION REQUIRED] — full editor list, publisher, year and page numbers not confirmed in this session.
- ISCCM Manual of RRT and ECMO in ICU. Indian Society of Critical Care Medicine. Low-flow causes; association of chatter with hypovolaemia and haemolysis; CRRT pressure and alarm concepts (contrasted with ECMO ΔP in §9.1). [VERIFICATION REQUIRED] — edition, editors, year and page numbers not confirmed.
- Brown A, Udy A, Burrell A, Joyce CJ. Beta-blockade for the treatment of refractory hypoxaemia during venovenous extracorporeal membrane oxygenation: an in-silico study. Perfusion. 2025;40(4):877–885. DOI: 10.1177/02676591241262261.
- Butt SP, Razzaq N, Saleem Y, Cook B, Abdulaziz S. Improving ECMO therapy: monitoring oxygenator functionality and identifying key indicators, factors, and considerations for changeout. J Extra Corpor Technol. 2024;56:20–29. DOI: 10.1051/ject/2023047.
Chapter status
Drafted and audited 6 September 2026. Ten-pass quality control completed: clinical, physiology, evidence, citation, numerical, safety, contradiction, redundancy, bedside utility and literature-currency passes.
Open verification items are listed above rather than resolved silently. This chapter contains no fabricated citations, thresholds, doses or guideline statements; every unretrieved item is declared. Where two authoritative sources appear to disagree (the 150 mL/min exchange threshold), both are presented and the apparent conflict is resolved explicitly rather than by choosing one.