End of life care


Regardless of the mode of ECMO or underlying disease plans to palliate should prompt referral to Donate Life and discussion about organ donation as in any other patient in intensive care.


Palliation in the setting of VV ECMO is a rare occurrence. Even drastic changes in the lung parenchyma on CT can resolve over weeks to months and extreme caution should be taken in the decision to palliate even with a considerable degree of lung pathology. Exceptional long VV runs with recovery and successful weaning from VV ECMO have been reported.

Other factors however such as catastrophic CNS complications may lead to a palliative approach. In case the patient is not heavily dependent on the ECMO oxygen delivery, decannulation can be performed and desired respiratory support delivered by the ventilator. In patients heavily dependent on the ECMO support, the fresh gas flow is turned to zero and comfort care is provided as per usual practice with the ECMO cannulae left in situ. Near the time of death, the ECMO circuit which does not provide any support can simply be clamped, placed under the blanket and the ECMO console removed from the room.


Medical consensus is essential that no further advanced support, in particular, VAD/ heart transplantation is not a viable pathway nor one-way weaning and decannulation is deemed an acceptable pathway. Palliation can be carried out with the usual precautions of end of life care by clamping the circuit, simultaneous cessation of inotropic and vasopressor support and extubation if applicable. The circuit can be placed under the blanket and the ECMO console removed from the room. Donation after cardiac death would routinely be considered and ECMO support as such does NOT represent a contra-indication.

Diagnosis of brain death on ECMO support

Brain death may develop during ECMO support, either as a consequence of the initial insult (hypoxic encephalopathy) or as a complication of ECMO (such as intra-cerebral haemorrhage). Clinical determination of brain death is preferred, and is subject to the same preconditions as in other patients.

Brain death may be suspected on the basis of clinical examination of the cranial nerves and follows the same principle pathway as in any other ICU patient with the exception of the apnea test in the clinical diagnosis of brain death. An apnea test is possible on ECMO, however it requires careful control of the fresh gas flow to ensure an adequate increase in PaCO2, to >60 mmHg without concurrent hypoxia in body parts perfused by the ECMO return blood. The rise in PaCO2, needs to be demonstrated in both a right radial blood gas and the post-oxygenator blood gas.[1] If it is not possible to conduct clinical testing, a nuclear medicine brain scan should be undertaken if brain death is suspected and organ donation is planned.

Principles in apnoea testing on VA ECMO

  • Minimising CO2 exchange at the oxygenator
  • Calculation of safe fresh gas flow to provide oxygenation of ECMO blood flow
  • Following routine steps in apnea testing in addition to the below instructions
  • Demonstrating CO2 rise to >60 mmHg in both the systemic circulation and the post oxygenator gas

Performing an apnoea test on VA ECMO support

Reduce ECMO blood flow if good native cardiac function

  • CO2 clearance will be lower with lower blood flow
  • If the patient has very poor cardiac or pulmonary native function and is reliant on the oxygen delivery by the circuit, then do not reduce the blood flow

Baseline oxygenation and circuit setting titration

  • Check baseline arterial blood gas (from right radial artery if peripheral VA ECMO) to ensure patient SaO2≥ 88% and document baseline pH and PaCO2

Patient oxygenation and ventilator monitoring for apnoea test

  • Commence continuous positive airway pressure as appropriate or prior PEEP
  • Include a capnometer in the circuit to detect ventilatory waveform
  • Dedicate a staff member to observe for ventilatory effort with patient torso exposed

Calculate fresh gas flow at the oxygenator

  • Take a pre-oxygenator gas for Hb and pre-oxygenator oxygen saturation [SpO2preox]
  • Calculation 2x 1.34 x Hb (in g/L) x (1-SpO2preox) x ECMO blood flow in L = ml/min fresh gas flow* (minimum 200ml/min)
  • Set the fresh gas flow [FGF] to the nearest ml/min rounded up to at least 200ml/min, the FiO2 must be set to 1.0
  • Observe for preserved colour differential

  • Example1 pre-oxygenator blood gas shows Hb 80g/L, SpO2 65% and the ECMO blood flow is 2.5 L/min. Hence 2x 1.34 x 80 x (1-0.65) x 2.5 = 187 set the fresh gas flow to 200ml/min and observe for maintained colour differential

  • Example2 per-oxygenator SpO2 35%, Hb 125g/L, ECMO blood flow 4.3 L/min. 2x 125 x 1.34 x (1-0.35) x 4.3= 898 Set FGF at 900ml/min

*The formula calculates the oxygen uptake at the oxygenator with a safety factor 2x for suboptimal oxygenator function, potential drop in PaO2 in the oxygenator or delivery errors of oxygen. The visual observation of the colour differential remains important throughout the testing.

Ensure safe circuit oxygenation and adequate hypercapnia

  • If the colour differential is maintained there is no need for a post-oxygenator blood gas. If in doubt ensure SaO2≥ 90% on the post-oxygenator gas; if not, titrate FGF upwards until this target is met (ensure adequate circuit oxygenation)
  • In parallel ensure the patient’s (right arm) SpO2≥ 88% (adequate upper body oxygenation). If SaO2< 88% consider 1-2 rescue breaths or increase of ECMO blood flow (note need to recalculate the minimal FGF)

Ensure adequate PaCO2 for test completion

  • Continue to observe for signs of patient ventilatory effort
  • Check both patient ABG AND post-oxygenator gas to ensure both show PaCO2> 60 mmHg
  • Re-check the patient’s arterial blood gas after five (5) minutes or earlier if haemodynamic instability or desaturation <88% occur.
  • If PaCO2 has not risen adequately, generally wait longer with 1-2 ventilated breath and repeat the measurement which is generally sufficient.
  • Continue visual monitoring for adequate colour differential in the ECMO return limb.

Endpoints for apnoea testing

  • Patient’s blood gas AND post-oxygenator gas shows an adequate rise in PaCO2 with fall in pH < 7.30 : consistent with brain death
  • No adequate rise in PaCO2 with FGF at minimum flow tolerated by patient AND long enough period to allow CO2 to rise then the apnoea testing is not possible
  • If oxygenation is difficult but PaCO2 rises in both patient and post oxygenator gas – the ECMO flow could be increased and the process repeated to increase the overall oxygen delivery.
  • Haemodynamic instability (mean arterial pressure [MAP] < 60 mmHg) that is unsupportable with inotropes: apnoea testing is not possible
  • Respiratory effort noted: not brain dead

Safety is ensured by continuous visualisation of the colour differential at the oxygenator. The washout of oxygen at the oxygenator might be delayed hence the observed colour differential provides surrogate monitoring, post-oxygenator PaO2may assist when in doubt.

The fresh gas flow should never be turned below the calculated FGF.

If clinical testing cannot be performed, brain death should be determined radiologically. A nuclear medicine cerebral perfusion scan can be undertaken in the same way as for patients that are not on ECMO. The institution of ECMO has no consequences on the radio-chemical purity of Tc-99m HMPAO, nor on the uptake of tracer within the brain. Alternatively, a four-vessel cerebral angiogram can be considered. Logistics of transport to these departments need to be carefully planned and executed.

[1] Ihle, J. F. et al. A Protocol that Mandates Postoxygenator and Arterial Blood Gases to Confirm Brain Death on Venoarterial Extracorporeal Membrane Oxygenation. ASAIO J. 66, e23–e28 (2020). [PMID 31609793]

Performing an apnoea test on VV ECMO support

Analogous to VA ECMO the patient fresh gas flow is adjusted in the same way and observed for a maintained colour differential. No post-oxygenator gases are required and an adequate rise in CO2 is determined as usual in a peripheral arterial blood gas. If the patient is not ECMO dependent for oxygenation the fresh gas flow can be turned off altogether.

Organ donation

DCD after withdrawal of ECMO support is occasionally overlooked. Although these patients may have multi-organ dysfunction, this does not always preclude donation and it is advised that medical staff contact DonateLife to discuss suitability for donation.

Potential organ donors on ECMO should be managed with the same principles as other DCD cases outlined in the Alfred Hospital guidelines on ‘Organ and tissue Donation’ and ‘Overview ICU Withholding and Withdrawing Treatment Donation after Circulatory Death’

Initiation of ECMO with the sole intention of organ donation is NOT practiced at The Alfred.

Assessment of suitability of lungs for donation on VA ECMO

In order to assess the suitability of lungs for donation, it is necessary to determine the PaO2 of blood returning from the patient’s lungs. This is difficult because of the influence of highly oxygenated blood returning from the ECMO circuit, and requires careful planning.

When a patient has enough intrinsic cardiac output (relative to ECMO flow), an arterial blood gas sample drawn from the right radial artery will reflect oxygenation of blood passing through the lungs. However, when patients have a lower intrinsic cardiac output, highly oxygenated blood from the ECMO circuit is likely to perfuse the right arm.

The following guide aims to maximise the chance that an arterial blood gas sample from the right radial artery reflects native pulmonary function rather than ECMO function. It does this by increasing the patient’s own cardiac output and reducing ECMO support. The protocol also aims to guide the interpretation of blood gas samples where there is mixing of ECMO blood and intrinsic cardiac output.

The patient should be adequately anti-coagulated to prevent clot formation in the circuit during a reduction in ECMO blood flow. The addition of inotropic support may facilitate reduction in ECMO blood flow by increasing intrinsic cardiac output and thereby making assessment of lung function more successful.

  • Increase ventilatory support to maximise lung function: FiO2 of 1.0, respiratory rate ≥14, tidal volume 6 ml/kg, optimise PEEP.
  • After 30 minutes, reduce the ECMO FiO2 to 0.5 and then continue to titrate it downwards until the PaO2 on a post-oxygenator blood gas is between 100 and 150mmHg (PaO2 A).
  • Reduce ECMO blood flow as low as tolerated by the patient, but typically not lower than 1 L/min.
  • After 5 minutes perform an arterial blood gas sample from the right radial artery (PaO2 B).
  • Restore the ECMO FiO2 to 1.0 and re-check the PaO2 from a right radial artery sample after 5 minutes (PaO2 C).
  • Restore ECMO blood flow and ventilator to previous settings

This process results in three PaO2 recordings taken at low ECMO blood flow

  • PaO2 A: PaO2 from post-oxygenator blood on low ECMO FiO2
  • PaO2 B: PaO2 from radial artery with low ECMO FiO2
  • PaO2 C: PaO2 from radial artery on ECMO FiO2 1.0

Interpreting the results

  • If PaO2 B and C are identical, then the arterial blood gas sample represents the patient’s own cardiac output and is, therefore, representative of native lung function. The measured PaO2 therefore can be used to assess lung suitability for organ donation.
  • If PaO2 C is significantly higher than PaO2 B, then the PaO2 represents a mixture of ECMO blood (with a PaO2 equal to the post-oxygenator PaO2 ) and blood returning from the patient’s lungs (with an unknown PaO2).
  • If PaO2 B is greater than PaO2 A, then the PaO2 of blood returning from the lungs is no lower than PaO2 B. If PaO2 B is above the threshold for lung acceptability, then the assessment is complete.
  • If PaO2 B is greater than A but lower than that required for acceptance of lungs for transplantation, then the radial PaO2 cannot be used to assess lung function using this protocol and further assessment is required.
  • If PaO2 B is less than PaO2 A, then the PaO2 of blood returning from the lungs is no higher than PaO2 B. This means that the lungs are not suitable for transplantation as the PaO2 ‘generated’ by the lungs cannot be greater than 150mmHg.
  • If PaO2 C is significantly higher than PaO2 B, and PaO2 B is exactly the same as PaO2 A, then the PaO2 of blood returning from the patient’s lungs cannot be determined by this method because the radial arterial blood sample is likely to be reflective only of the ECMO oxygenation. In these patients (who have very little intrinsic cardiac output) the radial PaO2 cannot be used to assess lung function using this protocol.

Leave a comment

Feedback and comments are appreciated, although we cannot respond to each comment. Your email address will not be published. Required fields are marked *