Troubleshooting the ECMO circuit

ECMO is briefly covered in a short summary elsewhere on this site. One of these days, a more detailed exploration of this fascinating modality will become available. In the meantime, an excellent troubleshooting guide is offered by LITFL, and brilliant videos accompany a practical guide based at the venerable This page does not touch on the delicate art of manipulating the gas exchange via the ECMO circuit, because that is another vastly different topic area.

Problems of the ECMO circuit

ECMO problems can be arbitrarily divided into two categories:

Routine ECMO problems

  • Access issues:
    • Flow fluctuations
    • Kick-back of the venous line
    • Haemolysis and raised free haemoglobin
  • Limb ischaemia
  • Bleeding cannula sites

Catastrophic ECMO problems

  • Pump failure
  • "Suck-down"
  • Venous or arterial decannulation
  • Circuit rupture
  • Air embolism
  • Cardiac arrest on VV ECMO


  • LV distension
  • Cardiac chamber thrombosis

Access issues

Flow fluctuations in the access pump

  • This is when the venous access pump is unable to maintain a constant flow rate. The L/min number keeps fluctuating wildly.
  • The problem is usually with the venous access line or poor venous pressure.

Kicking line syndrome

  • This is when the venous access line keeps wiggling randomly, like a fire hose.
  • The reason this happens is the vein collapses intermittently around the access cannula tip, causing the intracannula pressure to briefly become very negative (which causes the atmospheric pressure to forcefully straighten the tube).

Haemolysis: anaemia and the rise of free haemoglobin

  • This happens because of the perverse environment of the circuit, and its unfriendlyness to the red cells within it.
  • Usually, the rate of haemolysis is kept under control. Free haemoglobin should be under 0.1g/L.
  • However, if the venous access is in trouble, the high suction at the cannula tip will shred more red cells. Generally an access pressure in excess of 250 mmHg will give rise to haemolysis.
  • One is alerted to this not by routine bloods, but by the presence of bright red haemoglobinuria.

Causes of access problems

  • Kinked cannula
  • Poor cannula tip position
  • Excessive pump suction for a given cannula diameter
  • Decreased venous pressure, due to
    • Poor intravascular volume
    • Increased intrathoracic pressure
    • Increased intraabdominal pressure

Management of access issues

Troubleshooting the circuit

  • First things first: check the cannula. Is it kinked?
  • Patient repositioning may be helpful (or more harmful, if you kink the cannula)
  • The oxygenator and pump head should be inspected for clots, as these may be contributing to haemolysis.

Troubleshooting the patient

  • Some fluid resuscitation may be helpful (as it refills the central venous spaces, hopefully giving the venous access pump more pumpable substrate).
  • One should think about abdominal compartment syndrome or some other reason as to why the big central vein where the cannula is might be underfilled.

Giving up

  • The pump RPM might need to be turned down. Perhaps this patient cannot tolerate such a vigorous suction. Generally speaking, a small cannula will misbehave with a pump speed in excess of 4000 RPM.
  • Ultimately the cannula may need to be repositioned, if it is in a profoundly stupid place.
  • An additional cannula may be required if access problems continue. A second cannula frequently puts an end to the access problems.

Limb ischaemia

    • Arterial embolism
    • Haematoma compression
    • Femoral arterial cannula is too large to allow retrograde perfusion
  • Either there has been an embolic complication, or a haematoma has formed around the femoral access site (putting pressure on the arterial structures), or the intraarterial cannula has completely occluded the end artery.
  • Whatever the cause, you need to urgently image the artery with Doppler.
  • This will ascertain where the occlusion is.
  • If an embolic complication has indeed taken place, the vascular surgical team will then reluctantly review the patient, and hopefully offer them some sort of revascularisation.
  • Any surgical approach here will be horribly dangerous, given the level of anticoagulation.
  • In the event of arterial occlusion by a femoral arterial VA-ECMO cannula, or in the event of a localised haematoma being the culprit, one may consider diverting some of the flow from the arterial line into the artery distal to the occlusion, in an anterograde fashion (i.e. using the ECMO circuit to perfuse the limb). In some cases this is done prophylactically.
  • Thankfully, probably because of vigilant monitoring the mortality for ECMO patients does not seem to be affected by the occurrence of limb ischaemia.

Cannula site blood loss

  • Most people will accept some sort of ooze. Cannula site bleeding - particularly after cut-down cannulation - is an expected complication.
  • Large amounts of ooze, however, are a source of ongoing blood loss.
  • Escalation of management strategies should resemble the following:
    • Firstly, ensure the cannula is properly positioned (i.e. inserted right up to the taper)
    • Thrombotic dressings are the next step (LITFL recommend caltostat
    • Direct manual pressure (or with a sandbag)
    • One may even decrease the ACT taget, or cease the heparin altogether, and maybe even administer blood products. Protamine can kill the circuit, particularly if it is a heparin-bonded circuit.
    • Local injections of adrenaline (1:200,000) or adrenaline-soaked gauze can occasionally be helpful
    • Ultimately, vascular surgical referral and repair with recannulation may be the only option, if the levels of ooze are becoming troublesome.

Circuit thrombosis

  • Frequent thrombosis monitoring is carried out, but for what? What do you do when you find a clot inside the circuit or the oxygenator?
    • Well, not much. Prevention is more important.
    • Small clots in the oxygenator dont tend to impair its function, and can be monitored.
    • Ensure adequate anticoagulation
    • Arrange staff and exchange the thrombosed circuit element.
  • Obviously, this issue is more important in a VA ECMO circuit. Systemic thromboembolism will trash the vital organs.

Catastrophic problems of the ECMO circuit

In contrast to the "routine" problems, which are also potentially lifethreatening, catastrophic ECMO problems are those which typically result in death within seconds or minutes, in some sort of highly visually effective way (eg. whole blood volume emptying out of the patient and onto the floor, or entire venous circulation filling with air). Obviously, it would be helpful to have an immediate response to these problems all ready to go inside your head, just waiting for the opportunity.

Pump failure

By a more forgiving general definition, "pump failure" is the failure of the circuit pump to generate an adequate amount of flow though the circuit. Usually the pump is not at fault; rather, the venous access is to blame.

By a more specific definition, "pump failure" is what happens when the pump suddenly stops.

There are three ways this can happen:

  • Motor failure
  • AC power failure
  • Pump head disengagement
  • Pump flow sensor failure

One's first signs of this may be cardiac arrest (on VA ECMO) or profound hypoxia (on VV ECMO).

Motor failure

  • The motor will usually alarm when this happens. Or it may not.
  • Management consists of hand-cranking the pump while getting a new motor console.

AC power failure

  • There will be no alarms with this one.
  • Management consists of hand-cranking while trying to reestablish AC power.

Pump head disengagement

the Vinnies protocol mentions that this will be accompanied by "an unusual grinding noise and vibration of the pump head". This happens when the pump head is improperly inserted into the pump. The grinding noise is the sound of the centrifugal rota rapidly destroying itself, and this usually calls for urgent circuit replacement. Not only will there be poor flow or no flow, but whatever flow remains may be enriched by shredded red cells and little particles of the disintegrating rota.

  • In VA ECMO, this resembles cardiac arrest.
  • In VV ECMO, this resembles profound hypoxia, which progresses into cardiac arrest.

How does one react to this?

  • Clamp the venous line
  • Re-engage the pump head, ensure that it is in properly
  • Turn the pump to 1000 RPM (a gentle slow setting) and unclamp the line
  • Gradually increase the RPM to the previous setting
  • If the hideous grinding sound persist, the rota has been damaged beyond rescue, and the circuit needs to be changed urgently.

Pump flow sensor failure

  • This is a flow alarm (or error message) which is accompanied by neither arrest, nor hypoxia, nor any other sorts of changes. The problem is with the ultrasonic flow sensor. The perfusionist will need to relubricate the sensor-tube interface.


This the complete obstruction of flow through the venous access circuit, resulting from the sudden collapse of the central vein around the access cannula tip. The result of this is a massively negative pressure in the access line, and failure of forward flow. Cardiac arrest or hypoxia may ensue.

The major contributor to this problem is actually the suction produced by the centrifugal pump.

Thus, you need to decrease the pump speed to decrease the suction (to about 1000 RPM).

Some flow should be restored by this manoeuvre.

One can then administer some fluid boluses to try to recondition the venous circulation, so as to prevent this from happening. The placement of an additional cannula may be called for on this occasion.

Accidental venous access decannulation

The accidental disconnection of the venous access cannula.

Immediate consequences of this:

  • Exsanguination of the patient from the cannula site
    • A peripherally cannulated patient stands some chance of haemostatic control; somebody can put pressure on the wound while everybody else scrambles to reinstitute access via another central vein.
    • If this is a centrally cannulated VA ECMO patient, the patient exsanguinates from the right atrium and this is not cool.
  • Aspiration of air into the circuit, with subsequent pump failure (as the rota will soon be impotently chewing foam)

Management consists of several simultanous measures:

  • Start CPR for the VA patient
  • For the VV patient, crank up the PEEP and FiO2 on the ventilator, or better yet grab the Laerdel bag and and try to oxygenate them that way.
  • Control bleeding
  • Reeestablish access
  • Get another circuit (as the existing circuit will doubtlessly be full of air and clot by the time you get it running again).

That is easier said than done. While everything is being organised, the VV ECMO patient will be trying to oxygenate via their own faulty lungs (and clearly that wasn't working for them, otherwise they wouldn't be on ECMO). The VA patient will be substantially worse off, being in a state of cardiac arrest with CPR being their only means for forward flow.

Accidental arterial access decannulation

The accidental disconnection of the arterial return cannula.

For the VA patient, this means cardiac arrest; for the VV patient, this means a sudden return to anaerobic metabolism. The disconnected cannula will also continue to empty the the patient's blood volume into the ambient bed space, at a rate equivalent to their cardiac output.

Thus, the immediate consequences are:

  • Cardiac arrest, or hypoxia followed by cardiac arrest
  • Exsanguination (over about 30 seconds the blood volume will drop by 1-2 litres)
  • Aspiration of air into the venous circulation, or into the aorta and cardiac chambers

Management consists of several simultanous measures:

  • Clamp the line
  • Stop the pump
  • Start CPR for the VA patient
  • For the VV patient, crank up the PEEP and FiO2 on the ventilator, or better yet grab the Laerdel bag and and try to oxygenate them that way.
  • Control bleeding:
    • If the patient is peripherally cannulated, one should make an attempt to control the bleeding by manual pressure.
    • If this is a centrally cannulated VA patient, the cannula has come out (traumatically) from the root of the aorta, and the chest will need to be opened in an urgent and unceremoneous fashion. A reasonably senior cardiothoracic person should probably do that.
  • Replace volume, as fast as possible
  • Reeestablish access
  • Get another circuit (as the existing circuit will doubtlessly be full of clot by the time you get it running again).

Circuit rupture

This term describes the breakdown of some random circuit component, be it tubing, oxygenator, pump head or what have you.

Depending on where this rupture is in relation to the pump, the circuit will either suck air into itself (resulting ultimately in mechanical pump failure and air embolism) or spray blood in a comically Pythonesque fashion from the fracture site (which is actually less rapidly lethal, and more obvious).

The solution, predictably, is to replace the circuit.

Management strategy suggested by the Vinnies protocol:

  • If the leak is small, get some sterile gloves on and block the hole with your finger.
  • If the rupture is unfingerable, clamp the circuit on either side of the rupture.
  • Start CPR (for the VA patient) or manually ventilate with 100% FiO2(for the VV patient)
  • Change the circuit
  • Replace the lost blood

Air embolism

Air embolism into the circuit

This tends to happen with circuit rupture or accidental access decannulation.

  • The circuit will rapidly de-prime (i.e. fill with air and froth)
  • The pump will become filled with foam, and pumping will not occur
  • The good news is, the loss of output will hopefully prevent air embolism
  • The bad news is, there will be loss of output.

Again, either arrest (VA) or profound hypoxia (VV) will be the overall result.

Additionally, air embolism into the patient's circulation may still occur.

Air embolism into the patient's venous circulation

Management strategy:

  • Clamp the circuit
  • Start CPR and/or manual bag oxygenation
  • Tilt the patient head down (so the air collects in the apex of the right ventrile, and the base fills with blood- this way, hopefully something other than air will get pumped into the pulmonary circulation)
  • The only way to get that air out is by aspirating it with a PA catheter (one advances it into the right ventricle until the pressure transducer no longer reads anything resembling a waveform, suggesting that it is in an air pocket)

Air embolism into the arterial circulation via the VA circuit

Management strategy:

  • Clamp the circuit
  • Start CPR and/or manual bag oxygenation
  • The arterial air emboli are likely to travel to the uppermost parts of the patient; thus, the brain and upper limbs are most at risk.
  • Thus, neuroprotective measures should be considered.
  • The St Vincents protocol mentions hypothermia to 34°, barbiturates, mannitol, and so forth.
  • There is little literature to guide one in this matter.

Rescuing the circuit

Unless the whole circuit is clotted up due to the blood-air interface, one might be able to de-air and reprime it. The arterial part of the circuit is clamped, and air is aspirated from the lines using a 60ml Luer lock syringe (the air can be guided to the three-way tap by manipulating the tubing, allowing to bubbles to percolate to the site of aspiration). The clamp can be taken off and pump restarted when the circuit is again full of blood.

Cardiac arrest from a non-circuit-related cause

This, in the VA ECMO patient, is a non issue. The circuit has already excluded the myocardium from any meaningful participation in the circulatory process. The monitor will be interesting to look at (VF, asystole, etc) but the VA circuit should provide enough flow to maintain organ perfusion. One can relax, have a coffee, and work out what caused the arrest in a calm and reflective fashion. The major risk from a totally immobile ventricle on VA ECMO is LV distension (see below) which takes a little while to develop.

In a VV patient, the sudden cessation of cardiac function will be rather disastrous, as venous access pressure will fail in the absence of cardiac output.

One should commence CPR according to the normal algorithm.

There is one notable difference with advanced life support in a VV ECMO patient. If ECMO flow can be established, the "A" and "B" parts of the algorithm can be safely neglected. The airway, as a conduit for gases, is superfluous if oxygenation is provided by the circuit. Additinally, the patient probably has some sort of massively gas-exchange-disabling lung problem, and trying to oxygenate them in a coventional fashion will probably meet with failure.

Thus, to get oxygen into the patient, one must reestablish satisfactory venous return to the access cannula. This can only be accomplished by CPR. Once good quality CPR is in progress, venous pressure should increase, allowing at least a low flow VV ECMO to continue (set the pump to 1000 RMP initially, and titrate up).

LV distension on VA ECMO

With a motionless arrested left venricle, the VA ECMO circuit acts as a total bypass of the pulmonary circulation. Under ideal circumstances, this means there is no flow in the pulmonary circulation.

However, that is not always the case; thus any pulmonary blood flow ends up distending this useless immobile ventricle.

If the LV is not ejecting blood into the aorta, pressure builds up in it, distending the mitral annulus and left atriu, and backfilling the pulmonary circulation until severe pulmonary oedema develops.

Not only that, but permanent LV damage can occur with such overdistension.

There is no conventient management strategy for this.

Options include the following:

  • Cardiac compressions to forcibly eject the LV blood are a temporising measure, and may cause more harm than good (you dont want to rupture an atrium).
  • Venting the left atrium or ventricle - in the olden days one would actually access the LV apex with a cannula, and connect that cannula to the venous access lines of the ECMO circuit, allowing the ECMO pump to decompress the left ventricle. It seems to work - at least in dogs
  • Atrial septostomy (creating an atrial septal defect) with the intention of allowing left-sided blood to mingle with the right side, thus equalising the pressures between the cardiac chambers.
  • IABP to improve forward flow has been used by some authors, even though the inflation of the balloon will certainly interfere with the arterial flow from the ECMO pump.

LV thrombus with VA ECMO

This is the unfortunate consequence of having a motionless chamber with blood stasis within it.

Frequent TTE surveillance is recommended to detect this complication.

Unfortunately, there is little that can be done about it once it forms. Certainly, the presence of a large LV thrombus will frustrate any attempts to wean from VA ECMO once the reversible LV failure resolves. If the LV failure is irreversible, and ECMO is a bridge to transplant, then we can safely say that the LV thrombus is less of an issue, as the thrombus and the heart are both going in the bin.


Sidebotham, David. "Troubleshooting adult ECMO." The Journal of extra-corporeal technology 43.1 (2011): P27-32.

St Vincents Hospital, the House of ECMO, have an ECMO protocol for house staff. So does Westmead Hospital. These are not freely available online, and one must rely on friendly locals to supply them.

These protocols are particularly helpful, and much of the practical information included above is derived from these evidence-based guidelines.

The Royal Adelaide Hospital ICU ECMO Guidelines are also an excellent resource, and very much available online. I have used them more for the routine care of the ECMO patient.

Allen, Steve, et al. "A review of the fundamental principles and evidence base in the use of extracorporeal membrane oxygenation (ECMO) in critically ill adult patients." Journal of intensive care medicine 26.1 (2011): 13-26.

Foley, Paul J., et al. "Limb ischemia during femoral cannulation for cardiopulmonary support." Journal of vascular surgery 52.4 (2010): 850-853.

Oliver, William C. "Anticoagulation and coagulation management for ECMO."Seminars in cardiothoracic and vascular anesthesia. Vol. 13. No. 3. SAGE Publications, 2009.

Davies, Andrew, et al. "Extracorporeal membrane oxygenation for 2009 influenza A (H1N1) acute respiratory distress syndrome." JAMA: the journal of the American Medical Association 302.17 (2009): 1888-1895.

Eugene, J., et al. "Cardiac assist by extracorporeal membrane oxygenation with in-line left ventricular venting." ASAIO Journal 30.1 (1984): 98-102.

Seib, Paul M., et al. "Blade and balloon atrial septostomy for left heart decompression in patients with severe ventricular dysfunction on extracorporeal membrane oxygenation." Catheterization and Cardiovascular Interventions 46.2 (1999): 179-186.

Russo, Claudio F., et al. "Veno-arterial extracorporeal membrane oxygenation using Levitronix centrifugal pump as bridge to decision for refractory cardiogenic shock." The Journal of thoracic and cardiovascular surgery 140.6 (2010): 1416-1421.

Platts, David Gerard, et al. "The role of echocardiography in the management of patients supported by extracorporeal membrane oxygenation." Journal of the American Society of Echocardiography 25.2 (2012): 131-141.