Secondary Logo

Journal Logo

How to do it Article

Management of Circuit Air in Extracorporeal Membrane Oxygenation: A Single Center Experience

Chan, Kai Man*; Wan, Winnie Tsz Pan*; Ling, Lowell; So, Jack Mei Chun; Wong, Constance Hau Ling*; Tam, Sandy Bik Shan*

Author Information
doi: 10.1097/MAT.0000000000001494
  • Free


The incidence of air in Extracorporeal Membrane Oxygenation (ECMO) circuit or systemic air embolism has been reported to be 1.4–4.6%, and outcomes are frequently fatal.1–4 Prevention strategies, including meticulous circuit priming, elimination of unnecessary connections (especially on the venous limb), and use of air detection alarms, should always be adopted. When overwhelming circuit air occurs, the patient should be immediately separated from ECMO to prevent air embolism, source of air should be identified and rectified, and ECMO flow should be re-established as soon as possible. Options to restart ECMO include either deairing or changing the circuit. During deairing procedure, airlock in the centrifugal pump is resolved by shifting the air out of the pump head and up to nondependent position(s) where air can be removed. This is usually accomplished by aligning the pump-head inlet, outlet and oxygenator into a straight conduit from bottom to top to enable syringe aspiration of air from the oxygenator. However, deairing in miniaturized heart-lung support systems, such as Cardiohelp HLS (Maquet, Rastatt, Germany), is particularly challenging. While its integrated pump and oxygenator design facilitates rapid deployment and transport, in this configuration, the inlet of centrifugal pump is sandwiched between its outlets and its oxygenator. The pathway from the pump inlet to the oxygenator resembles an “umbrella fountain,” and it renders aspiration of air from oxygenator difficult.

Air in circuit is an emergency, and rapid response is needed. ECMO centers should have established protocols and training to prepare for this rare but fatal complication. However, the implementation of these protocols and the efficacies of crisis management are rarely reported. We report our institutions’ experience in the management of circuit air in Cardiohelp HLS system.



Our unit is a 23 bed adult mixed medical and surgical tertiary teaching hospital ICU in Hong Kong with more than 1,700 annual admissions. We provide both venovenous (VV) and venoarterial (VA) ECMO support and the annual volume ranges from 15 to 20 ECMO runs. ECMO training including deairing practice is provided to 48 ICU nurses, 13 intensivists, 5 cardiac surgeons, 10 anesthetists, and 8 perfusionists.

Patient Identification and Data Collection

Since October 2009, a database was set up in our ICU to collect information in all patients who received VV or VA ECMO. These data are submitted to ELSO registry biannually. In this study, we identified all patients who developed gas bubbles in circuit or systemic gas embolism from our database between October 2009 and July 2020. We describe patients’ clinical characteristics, their ECMO configurations, trigger events that lead to circuit gas bubbles, sources of air, presence of arterial air or pump airlock, cardiorespiratory status during the event, techniques employed to re-establish flow, ECMO downtime, neurologic, and other clinical outcomes. This study was approved by The Joint Chinese University of Hong Kong—New Territories East Cluster Clinical Research Ethics Committee with waiver of informed consent (2020.517).

Management Protocol of Air in HLS Circuit/systemic Air Embolism

The patient and circuit should be simultaneously managed with clear role allocation. The circuit management consists of three components, including separation of patient from ECMO circuit, identification and elimination of the source of air entry, and re-establishment of flow. In our institutional protocol, the decision tree depends on two factors, whether there is air beyond oxygenator in the arterial limb (arterial air) and whether the centrifugal pump is air locked (Figure 1).

Figure 1.:
Management of circuit air in Cardiohelp HLS ECMO system. ECMO, extracorporeal membrane oxygenation; HLS, heart-lung support.

In the presence of arterial air, the circuit is immediately clamped and changed. On the other hand, if the pump is stopped by air lock without arterial air, circuit should be clamped while the source of air should be identified and rectified. Subsequently, deairing is achieved by backflush technique to re-establish ECMO flow (Figure 2). Alternatively, if there is no airlock in pump head and ECMO flow is minimally interrupted, consideration may be given to continue the ECMO when the patient and the circuit have been carefully evaluated for source of air and ongoing microbubble spillage has been ruled out. Syringe aspiration of air from the top of oxygenator is then attempted.

Figure 2.:
Backflush deairing technique in Cardiohelp HLS. A: Overwhelming air in Cardiohelp HLS ECMO circuit. Spillage of air to arterial limb is prevented by airlock in the centrifugal pump leading to cessation of flow and by gas trapping in the oxygenator. B: Three clamps are applied to separate patient from ECMO and prevent air embolism. One liter of normal saline in collapsible bag is connected to the oxygenator via a rapid infuser line and a one-way valve. C: After sterilization of the tubing, the air effluent outlet is created by cutting the venous limb between the two clamps. D: The pump head is disengaged from the motor drive, and the inlet of the centrifugal pump is then turned to 12 o’clock position to facilitate shifting of air. The machine side of the venous limb is unclamped; the saline bag is manually pressurized to backflush the oxygenator, centrifugal pump, and the distal venous limb. Effluent, including blood and air, is flushed out the venous cut end until clear. E: The machine side of the venous limb is clamped again. A 3/8′′–3/8′′ straight connector is added, and the two ends of the venous limb are reconnected in a standard fashion. The rapid infuser line is clamped. F: The ECMO is re-established by unclamping the circuit and dialing up the pump speed. The circuit should be inspected for any recurrence of air entrainment. Once ECMO flow is stabilized, the remaining air at the top of the oxygenator can be aspirated by syringe. A video link showing the procedure can be found at ECMO, extracorporeal membrane oxygenation; HLS, heart-lung support.


Four patients were identified from a cohort of 116 (3.4%) patients receiving VV or VA ECMO. All patients were supported by Cardiohelp HLS ECMO system. The findings are summarized in Table 1 and Figure 3. The source of air was located and isolated in all cases. Three were on the venous side of the ECMO circuit. None had air in the arterial limb of the circuit or systemic air embolism. ECMO flow was stopped by centrifugal pump airlock in two patients, and the deairing protocol was not followed in one of these two patients. In this patient, syringe aspiration from oxygenator failed to resolve the airlock, and cardiac arrest occurred after 20 minutes. In the end, ECMO circuit change was performed. The ECMO downtime caused by pump airlock in the two patients was 30 and 23 minutes, respectively. In all cases, the ECMO flow was resumed without further circuit complications.

Table 1. - Description of ECMO Configuration, Source and Management of Air in Circuit and Patient Outcomes.
Patient and Diagnosis ECMO Configuration Trigger Event Source of Air Airlock in Centrifugal Pump Control Measures for Air Procedures to Re-establish ECMO Flow Hemodynamic and Respiratory Status During Clamped Circuit Time to Resume ECMO Neurologic Outcome Final Outcome
1. Acute myocardial infarct, post-CABG PCCS Femoral peripheral VA
23 Fr MS 55 cm
17 Fr Art 23 cm
DPC 7 Fr
LV vent via RSPV
Post-CABG cardiac tamponade needed resternotomy on postop day 7; clot formation in LV vent Air entrainment via three-way stopcock of LV vent during attempt of LV vent clot removal Yes LV vent clamping 1. Unable to aspirate air via pigtail on the oxygenator with syringe
2. Change of HLS circuit
BP 42/22, P 78, CVP 34, degenerate to VF after 20 minutes; 10 minutes of ECC before back on ECMO 30 minutes Fully conscious and movement in four limbs was observed 4 days after the index event Wean off ECMO on day 14
Death 6 days after ECMO decannulation due to pulseless VT
2. Acute myocardial infarct Femoral peripheral VA
23 Fr MS 55 cm
17 Fr Art 23 cm
DPC 7 Fr
Circuit change on ECMO day 13 for hemolysis A crack in the 3/8′–3/8′ connector, which was embedded in the venous cannula (from scalpel blade during circuit change) No; ECMO flow was not interrupted Connector change 1. ECMO circuit recirculation by end to end connection during connector change
2. Trimming of the venous cannula end to facilitate removal of the embedded connector
3. Circuit reconnection
BP 74/64, MAP 66, P 121
SpO2 98%
No CPR was required
3 minutes Pre-existing bilateral occipital infarcts 6 days before index event; fully conscious and MRC grade 4 motor power in four limbs Wean off ECMO on day 21
Discharge alive and awake to the referring hospital after 53 days
3. Tracheal rupture Femoral Jugular VV
23 Fr MS 55 cm
19 Fr SS 23 cm
Fibrin and clot in oxygenator leading to circuit change on ECMO day 11 Residual air confined in the oxygenator was not noticed during priming of new circuit No; ECMO flow was not interrupted Nil needed after careful circuit check Syringe aspiration of air through oxygenator pigtail SpO2 92% FIO2 1.0 PEEP 6;
BP 170/70 MAP 98
P 89;
ECMO flow 3.2 L/min
No CPR was required
<1 minute Fully conscious 8 days after index event; full recovery from tetraparesis due to CIPNM Wean off ECMO day 19
Discharge home alive and ambulatory after 45 days
4. Massive pulmonary embolism Jugular femoral VA
23 Fr MS 55 cm
17 Fr Art 23 cm
7 Fr DPC
Partial decannulation of venous cannula on ECMO day 1 Venous drainage cannula sideholes Yes Drainage cannula pushed in after sterilization Deairing by backflush technique (Figure 2) BP 76/58 MAP 60
P 103
No CPR was required
23 minutes Recurrent pre-ECMO cardiac arrest; full neurologic recovery Wean off ECMO on day 2
Discharge home alive and ambulatory after 54 days
Art, arterial; BP, blood pressure; CABG, coronary aorto bypass graft; CIPNM, critical illness polyneuromyopathy; CPR, cardiopulmonary resuscitation; DPC, distal perfusion catheter; ECC, external chest compression; MRC, Medical Research Council; MS, multistage; P, pulse; PCCS, postcardiotomy cardiogenic shock; RSPV, right superior pulmonary vein; SS, single stage.; VA, venous-arterial; VF, ventricular fibrillation; VT, ventricular tachycardia; VV, venous-venous.

Figure 3.:
Circuit air. A: In patient 2, small amount of air was found at the top of oxygenator (horizontal green arrow) after the circuit change. There was no pump airlock or interruption of ECMO flow. Bubble detector alarmed when the flow probe was placed on the venous but not the arterial limb. A crack in connector accidentally made by scalpel during circuit change was confirmed (vertical yellow arrows) as source of air. The bubble alarm resolved afterwards. B: The amount of air found immediately after the circuit change in patient 3. The air was confined to the oxygenator without interruption of ECMO flow. The bubble detector inside the flow probe placed on the arterial limb did not alarm. The air was simply aspirated by syringe through pigtail (blue arrow) with the circuit running. There was no air re-accumulation afterwards. C: Overwhelming air in HLS was noticed immediately after the jugular venous cannula was shifted outward in patient 4. The pump was airlocked, and ECMO flow ceased. The circuit was immediately clamped, and the venous cannula side holes were found to be outside skin. The cannula was scrubbed and pushed back in. Deairing of HLS circuit was then accomplished by using backflush technique. The cut ends of the venous limb were reconnected by straight connector (yellow arrow). D: Status of ECMO circuit after 23 minutes of downtime in patient 4. No clot was found in the circuit. ECMO, extracorporeal membrane oxygenation; HLS, heart-lung support

None of the four patients had immediate sequelae directly attributable to the index event of circuit air. However, air in circuit event probably affected patient 1’s ECMO duration due to prolonged downtime. All of the patients had ECMO weaned and decannulated. Three of the four patients were discharged alive. Patient 1 died from recurrence of pulseless ventricular tachycardia 6 days after ECMO was weaned and decannulated. The cause of the arrhythmia was presumed to be coronary ischemia.


To our knowledge, we are the first to report the results of protocolized management of air in ECMO circuit from a consecutive cohort. Our findings in patients 1 and 4 support the hypothesis that deairing during pump airlock in the complex pathway of HLS is not easily accomplished by simple aspiration and alternative methods, such as backflush technique or a change of circuit, is required. Our unfavorable experience in patient 1 prompted in vitro testing of this backflush technique in our water lab and its subsequent team-based simulation training. At the time when index event happened in patient 4, 67 of the 71 ICU doctors and nurses had undergone at least two sessions of simulation training, and the resultant favorable clinical outcome adds further support to the feasibility of this technique if staff are given appropriate training.

Although change of circuit is always an option, we make an argument in favor of backflush technique in our institution. First, although some data suggest preprimed ECMO circuits can maintain sterility for up to 65 days, bacterial colonization and expiration in low-volume centers are still concerns.5 Second, backflush technique might require less time compared to a new circuit prime. Our training record revealed that the time to accomplish deairing by backflush procedure ranges from 6 to 13 minutes, as opposed to 15–40 minutes for new circuit priming. Although backflush deairing may appear to be superior in training condition, the time required was much longer in our cases. The impact of human and environmental factors on team performance in stressful clinical setting remains to be further studied.

In our patients, no macro-bubbles (≥5 mm diameter) were detected in the arterial limb of HLS circuit. As the oxygenator serves as a gas filter, the likelihood of macro-bubble spillage to the arterial side of the circuit is low. However, the hazards of gaseous microemboli (microbubbles ≤500 μm diameter) should not be overlooked, especially when ECMO flow is minimally interrupted as in patients 2 and 3. It should be emphasized that only 45–75% of microbubbles are filtered by commonly used oxygenators.6 In fact, gaseous microemboli had been shown to pass through the oxygenator after intravenous drug injection during normal ECMO operation and is regarded as a cause of neurocognitive insults in VA ECMO.7,8 Albeit the risk may be less in VV ECMO due to lung filtering, vigilance for ongoing microbubbles spillage below the detection limit of ultrasound detectors (≥5 mm) should be maintained. Regular neurologic examination and subsequent brain imaging should be considered even if there is no interruption of flow after circuit air occurred.

In the presence of macro-bubbles in the arterial limb, we believe the circuit should be immediately clamped and changed for two reasons. First, the added complexity and the extra manipulation required to deair the arterial segment perpetuates the risk of systemic gas embolism. Second, once bubbles appear in arterial limb, elimination of microbubbles in the arterial limb cannot be guaranteed even after an apparent complete deairing.

There are several limitations in our study. First, since all of our patients had residual native cardiac output during the index event, outcomes may be different in patients who are totally ECMO dependent. Second, our study is retrospective with low number of patients. Despite these limitations, our results show that, with appropriate management, systemic gas embolism could be avoided during circuit air emergencies. Backflush technique is a safe and effective deairing method to overcome centrifugal pump airlock due to venous air in Cardiohelp HLS system.


1. Kumar A, Keshavamurthy S, Abraham JG, Toyoda Y: Massive ar embolism caused by a central venous catheter during extracorporeal membrane oxygenation. J Extra Corpor Technol. 51: 9–11, 2019.
2. Timpa JG, O’Meara C, McILwain RB, Dabal RJ, Alten JA: Massive systemic air embolism during extracorporeal membrane oxygenation support of a neonate with acute respiratory distress syndrome after cardiac surgery. J Extra Corpor Technol. 43: 86–88, 2011.
3. Yan S, Lou S, Zhao Y, Liu G, Ji B: Air in extracorporeal membrane oxygenation: can never be overemphasized. Perfusion. 36: 97–99, 2020.
4. Organization ELS: ECLS Registry Report: International Summary. 2017.
5. Tan VE, Evangelista AT, Carella DM, et al.: Sterility duration of preprimed extracorporeal membrane oxygenation circuits. J Pediatr Pharmacol Ther. 23: 311–314, 2018.
6. Jirschik M, Keyl C, Beyersdorf F: A clinical comparison of bubble elimination in Quadrox and Polystan oxygenators. Perfusion. 24: 423–427, 2009.
7. Jiao Y, Gipson KE, Bonde P, et al.: Quantification of postmembrane gaseous microembolization during venoarterial extracorporeal membrane oxygenation. ASAIO J. 64: 31–37, 2018.
8. Zanatta P, Forti A, Bosco E, et al.: Microembolic signals and strategy to prevent gas embolism during extracorporeal membrane oxygenation. J Cardiothorac Surg. 5: 5, 2010.

crisis; protocol; ECMO; outcomes; adult; deair

Copyright © ASAIO 2021