Secondary Logo

Journal Logo

ELSO Guidelines

Extracorporeal Life Support Organization (ELSO): Guidelines for Neonatal Respiratory Failure

Wild, K. Taylor; Rintoul, Natalie; Kattan, Javier; Gray, Brian*; Engle, Reviewers: William*; Keene, Sarah; Best, Derek††; Davis, Carl; DiGeronimo, Robert#; Raman, Lakshmi**

Author Information
doi: 10.1097/MAT.0000000000001153
  • Free


This guideline describes prolonged extracorporeal life support (ECLS), applicable to newborn infants with respiratory failure. These guidelines address technology and patient management during ECMO. Equally important issues such as personnel, training, credentialing, resources, follow-up, reporting, and quality assurance are addressed in other ELSO documents or are center-specific.


Figure 1 lists indications, contraindications, and modes of support for Neonatal ECMO.

Figure 1.
Figure 1.:
Indications, contraindications, and modes of support.

A. Cannulation

1. Methods. The decision for VA versus VV support for neonatal respiratory ECMO is dependent on the patient, surgeon, and center. Open cut down exposure for VA and VV cannulation and percutaneous VV cannulation are all reasonable options if performed safely with the proper expertise.

2. Cannulation technique (ELSO Red Book, 5th edition, Ch 12).1 In most situations, cannula selection occurs after cut down with direct inspection of the vessels. Some surgeons perform bedside ultrasound of the right neck vessels before cannulation to estimate the vessel size and guide selection. Some vasospasm can be expected, so a cannula with a diameter (size in Fr divided by 3.14) that is 80% of the diameter of the vessel can reliably be chosen.

Direct cut down cannulation. Cannulation is usually done in the ICU with full sterile preparation and an OR team. Which vessel to cannulate first depends on surgeon preference? Many prefer to cannulate the artery first as it can be deeper than the vein depending on neck position. Others argue that venous cannula should be inserted first as it is secured at 6–8 cm and the arterial cannula is inserted only 2–3 cm. Furthermore, given that the arterial cannula is clamped while the venous cannula is placed, there is an increased risk of developing an arterial clot, which would be more significant than a venous clot. Direct handling of the vessels is minimized as much as possible to avoid spasm. If the vessels are very small, if there is difficulty with cannulation, or if spasm occurs, then fine stay sutures placed in the proximal edge of the vessel are very helpful. Chest x-ray should be obtained to verify venous cannula tip location in either the right atrium or in the IVC for a bicaval dual-lumen cannula. Echocardiography (ECHO) can also be used to assess position of the cannula tip, particularly during dual-lumen cannula placement. Femoral cannulation is not appropriate for neonates.

Percutaneous cannulation. Some centers have employed percutaneous cannulation techniques for VV ECMO in neonates. This can be accomplished through ultrasound-guided right jugular vein access and gradual dilation of the vessel over a wire using the Seldinger technique. ECHO and/or fluoroscopy should be used to assure safe positioning of the wire during dilation and ultimate placement of the cannula tip to avoid vascular and cardiac injury.

3. Venous drainage. In most cases, there is no need for further venous drainage as suitably sized cannulae will give adequate support. If venous drainage is inadequate, a separate drainage cannula may be required. The cephalic end of the jugular vein is the most common location. If possible, a 10 Fr or 12 Fr arterial cannula with a distal hole should be used for the cephalic drain, as an 8 Fr cannula will likely thrombose.

Some centers routinely place a cephalad venous cannula, but this is an institutional preference and is not normally required. The femoral veins are usually insufficiently sized and should not be accessed in neonates. Central cannulation could be considered for inadequate venous drainage.

B. Pump Selection (ELSO Red Book Ch 5)1

There is variation among centers regarding the use of roller versus centrifugal pumps for neonates. Roller pumps control flow precisely in infants, but have the risk of circuit rupture. Centrifugal pumps avoid blowout, but are less precise and can cause hemolysis. Many centers have made the transition from roller pumps to centrifugal models, whereas other centers continue to use roller pumps because of the risk for hemolysis with centrifugal pumps. Hemolysis has been associated with neurologic injury, kidney injury, and higher mortality. Rigorous data regarding choice of pump type in neonates are lacking, so pump choice is based on expertise and preference of each center.2–5

C. Oxygentor Selection

Rated flow should be over 500 ml/min (see ELSO Red Book Ch 5).1


Figure 2 outlines blood flow during neonatal ECMO.

Figure 2.
Figure 2.:
Blood flow for neonatal ECMO.

A. Oxygenation

Oxygenation of the tissues is affected by blood flow, hemoglobin, and oxygen saturation. Oxygen delivery should be at least three times oxygen consumption (6 ml O2/kg/min for neonates). Support is typically initiated with a sweep gas of 100% FiO2. In VA ECMO, SvO2 is an excellent indicator of adequate end organ oxygen delivery, with a goal of 65–80%. SvO2 can be used as a trend in VV ECMO. Because of the negative effects of hyperoxia, if the patient’s arterial PaO2 is found to be over 100 mm Hg, the circuit flow or sweep gas FiO2 should be adjusted accordingly.

B. CO2 Clearance

CO2 clearance is controlled by the sweep gas flow rate. CO2 transfer across the membrane lung is very efficient and will exceed oxygen transfer. Blood gas values must be assessed soon after initiating flow to avoid hypocarbia. Sweep gas flow rates are initially set equal to blood flow and adjusted to keep the patient’s PaCO2 40–45 mm Hg.

C. Anticoagulation (ELSO Red Bood Ch 7)1

Anticoagulation is achieved with heparin or a direct thrombin inhibitor (DTI). In neonates, the most common anticoagulant is heparin, but some centers have started to use DTIs. However, experience with DTIs is limited in this patient population. As with other patient groups, there is no consensus or ELSO recommendation regarding best anticoagulation practice for neonatal respiratory ECMO patients, and there is no ideal choice for anticoagulation.

Table 1.
Table 1.:
Anticoagulation Monitoring
  1. Table 1 outlines anticoagulation monitoring.
  2. Anticoagulants
  3. Heparin (unfractionated heparin or UNFH) is given as a bolus (50–100 units/kg) to the patient at the time of cannulation, followed by a continuous infusion. Given that premature infants are at higher risk for IVH, the cannulation bolus dose is often 50 units/kg.

The neonatal ECMO circuit prime is generally anticoagulated with 100 units of heparin per unit of packed red blood cells (PRBCs) for the neonatal circuits with 1/4 inch tubing. UNFH infusion rates in neonates on ECMO usually start at 25–30 units/kg/hr. Neonates may require higher doses of UFH due to their low plasma concentrations of ATIII. Note: Heparin-induced Thrombocytopenia (HITT) is extremely rare in neonates.

Direct thrombin inhibitors (DTIs) include argatroban and bivalirudin. They do not require ATIII for action and unlike heparin, DTIs can inhibit clot-bound thrombin, which may be an advantage. DTI dose is titrated by PTT (1.5–2 times normal) or ACT. It may be difficult to monitor the effect of DTIs on anticoagulation if the patient has an abnormal PTT level before initiating ECMO. Furthermore, the PTT level tends to plateau over time, resulting in potentially unreliable levels.

D. Patient Management (ELSO Red Book Ch 14)1

1. Hemodynamics during VV support. Patients are dependent on their own cardiac output to maintain hemodynamic status. Appropriate medications and infusions are used to control cardiac output, blood pressure, and vascular resistance. Patients on inotropic support before ECMO will typically wean substantially after being placed on ECMO despite a lack of direct blood pressure support from the ECMO circuit.

During VA support, hemodynamics are controlled by the blood flow (pump flow plus native cardiac output) and vascular resistance. Because the pulse pressure is low, the mean systemic arterial pressure will be somewhat lower than normal. In addition, patients placed on ECMO for cardiac support are typically on significant inotropes before ECMO initiation. As these drugs are titrated down, resistance decreases and systemic pressure falls proportionately. If the systemic perfusion pressure is inadequate (low urine output and poor perfusion), pressure can be increased by increasing pump flow, transfusing blood products, and/or titrating vasopressor infusions. Systemic vasodilatation requiring pressors is common in patients in septic shock.

Systemic perfusion is best measured by mixed venous blood saturation (SvO2). In VV ECMO, the SvO2 value measured from the ECMO circuit is not accurate due to recirculation. Some centers choose to obtain an accurate measure by turning off the flow for 30 seconds, then measuring SvO2 over the next 10 seconds. If the patient has a cephalad cannula, some centers may also place a separate saturation probe on the cephalad cannula to continuously monitor cerebral SvO2. Assuming a SaO2 more than 95% and a SvO2 greater than 65% indicates adequate systemic oxygen delivery, even though the pressure may be low. If systemic oxygen delivery is not adequate (SvO2 less than 65%), pump flow can be increased until perfusion is adequate. If extra blood volume is required to gain extra flow, blood products and crystalloid solution can be considered. In VA ECMO, increasing ECMO flow will result in increased oxygen delivery to the systemic circulation.

In the case of VV ECMO, flow changes are less straightforward, as recirculation (newly oxygenated blood just delivered into the patient and taken back into the drainage lumen) can occur. Recirculation is dependent on cannula position, patient volume status (right atrial dimensions), native cardiac function, and flow rates, and it should be considered whether increasing ECMO flow does not result in a higher patient saturation. Elevated circuit flow may worsen the phenomenon of recirculation and is evidenced by decreased peripheral SpO2 and increased SVO2. In this scenario, a CXR or echocardiogram should be obtained to evaluate the cannula position. Potential treatments for recirculation include sedation, patient repositioning, cannula adjustment, volume administration, and decreasing circuit flow.

2. Ventilator management (Red Book Ch 14)1

Figure 3 shows ventilator management and gas exchange while on neonatal ECMO.

Figure 3.
Figure 3.:
Ventilator management on ECMO.

3. Fluid management: blood volume and fluid balance

The ECMO circuit for neonates is primed with red blood cells, which will equilibrate with the native blood volume during the first several minutes. The goal of fluid management is to return the extracellular fluid volume to normal (dry weight). During the acute inflammatory stage early in ECLS, capillary leak may occur and is exacerbated by excessive crystalloid infusion. When the patient is hemodynamically stable (typically 24–48 hours), ultrafitration, continuous renal replacement therapy (CRRT), and/or diuretics may be used to manage fluid status. Diuretics are often delivered as an infusion to avoid anticoagulation instability. If the diuretic response is not sufficient, SCUF (slow continuous ultrafiltration) or CRRT can be used to remove fluid. If renal failure occurs for any reason, it may be treated with CRRT.

4. Sedation

Minimal sedation is the overall goal. Neonates are initially managed on intermittent dosing of narcotics, but most will require continuous infusions. Given concern for adverse neurologic effects, benzodiazepine use should be minimized. Dexmedetomidine may be considered as a sedation adjunct. Muscle relaxants are given during cannulation, decannulation, and circuit changes because of the risk of air embolism. Some institutions hold sedation and analgesia for a neurologic examination (a daily drug holiday).

5. Nutrition management

Neonates may have significant catabolism and weight loss while on ECMO, but they are not typically fully enterally fed. Trophic feeds should be encouraged and advanced slowly as tolerated. Parenteral nutrition should be started within 24 hours of ECMO with a goal of providing approximately 80–120 kcal/kg/day. Some centers choose to pursue fluid restriction to limit fluid overload, so calorie intake should be maximized in light of fluid restrictions. Alternatively, early initiation of CRRT to control fluid balance has been shown to improve protein delivery.6 Some institutions run lipids through a peripheral intravenous line to avoid lipid precipitation and thrombosis in the circuit.

6. Temperature

The desired blood temperature within an ECMO circuit matches the normal average body temperature (typically 37°C). This temperature is a set point on the water bath of the heat exchanger. If there is a concern for hypoxic ischemic encephalopathy (HIE) and there is a desire to maintain mild hypothermia (33–34°C) during the first 24–72 hours, it is possible to decrease the water bath temperature. Case series show an increased risk of intracranial hemorrhage (ICH) for HIE patients on ECMO, but available evidence suggests this is from the ischemic injury combined with heparinization and not clearly worsened by hypothermia.7 Thus, many centers perform standard of care cooling for these patients. However, there is no benefit in initiating cooling after starting ECMO support if the patient has not been cooled pre-ECMO in a setting where HIE is a concern.8 Hyperthermia (from fever or inflammation) may be masked due to the water bath set point. Therefore, it is advantageous to closely assess for infection using laboratory values (WBC, CBC shifts, acidosis, and blood cultures) to avoid hypermetabolism and recognize potential infection.

7. Infection and antibiotics

There is no standard policy regarding prophylactic antibiotics or surveillance blood cultures during ECMO support. There is, however, no data to support prophylactic antibiotic use in neonatal ECMO, and antibiotics should be reserved for suspected or proven infection. Bacteremia during ECMO may be related to bacterial growth on a component of the circuit, but is usually related to another source in the patient. Unlike suspected “line sepsis” in the usual critically ill patient, it is often not possible to change ECMO cannulas. Changing the circuit should be considered if cultures remain positive while on antibiotics, although there are limited data on this practice.

8. Procedures

Procedures on ECMO require special consideration for hemostasis in anticoagulated patients. The heparin infusion is often decreased, platelet levels are increased to >80–100,000/mm3, and fibrinogen to >150 mg/dL. Surgical procedures on ECMO may also benefit from an antifibrinolytic therapy, that is, aminocaproic acid (amicar). Anticoagulation may be discontinued 1 hour before and during invasive procedures and restarted after the risk of rebleeding has decreased. If bleeding persists postprocedure despite decreased heparin and product replacement, it is reasonable to stop anticoagulation until the bleeding stops. This may stop the bleeding, but may also result in clotting in the circuit. Therefore, active surveillance of the circuit is important, particularly in neonatal circuits with relatively low flow.

9. Neuroimaging

Head ultrasound (HUS) is an essential neuroimaging technique in neonates. A common HUS imaging protocol includes obtaining a baseline pre-ECMO ultrasound followed by a daily monitoring HUS for the first 3–5 days of cannulation, the period during which the incidence of ICH is highest.9,10 HUS is then done at variable intervals for the duration of the ECMO run. Additional ultrasounds should be obtained based upon clinical indications, such as new onset seizures, sudden drop in hemoglobin, or other clinical concerns.

Head computed tomography (HCT) has been used for follow-up of abnormal ultrasounds for emergent evaluation that would impact immediate survival or continued ECMO candidacy. However, head CT is not sensitive or specific for gray and white matter injury.

Magnetic resonance imaging (MRI) is the gold standard for detection of intracranial injury and pathology, in particular stroke and white matter injury. MRI cannot logistically be performed on ECMO. However, MRI with contrast can be done after ECMO support or before discharge. Some centers report improved visualization of the white matter with 1 mm cuts.

10. Neuromonitoring

Electroencephalogram (EEG or aEEG). Critically ill infants are at high risk for seizures, and the majority of seizures tend to be electrographic only.11,12 The incidence of seizures for neonates on ECMO varies between 5% and 30%. Treating seizures is typically accomplished in a multidisciplinary approach using antiepileptics at the direction of the Neurology service. Please refer to the Neuromonitoring guideline for additional information.


Management of Common Complications (Table 2)

Table 2.
Table 2.:
Complications of Neonatal ECMO

A. Air in the Circuit Requires Immediate Response, Especially on VA ECMO

  1. Stop the blood pump and clamp the arterial (infusion) line.
  2. If possible, restart blood flow through a bridge line to shunt the air to the venous portion of the circuit.
  3. Examine the circuit for air and identify the source of air introduction.
  4. Remove any residual air and repair or replace the source of the entrainment. Potential causes should be examined from the venous cannula and progress through the entire circuit, looking for loose tubing connections, stopcocks and/or air in infusion lines preoxygenator.
  5. Much of this type of air intrusion is trapped by the oxygenator. Once visible air is extracted from the circuit, circulate flow through the bridge, examine the circuit reaccumulation, and resume support when corrected.

B. Bleeding into the Head or Brain Parenchyma

  1. Stable neonates: Head ultrasounds (HUS) should be performed every 24 hours for at least the first 3–5 days and then per institution protocol.
  2. HUS shows bleed: CT scan may provide additional information on severity and progression of hemorrhage.
  3. For a small bleed: coagulation status will need to be optimized and HUS repeated frequently to detect any extension.
  4. For extending bleeds or bleeds that are moderate: measures to optimize cardiorespiratory support should be undertaken to allow the patient to be weaned from ECMO.
  5. For severe intraparenchymal hemorrhage: withdrawal of ECLS may be indicated.

C. Bleeding Post Chest Tube Placement

  1. Accumulated blood should be evacuated by a posterior tube, as evacuation quantifies the rate of bleeding and decreases the risk of a hemothorax and later clot.
  2. A CT scan may be indicated to determine if the tube is in the lung parenchyma. If it is, then the tube should be removed, but thoracotomy will probably be needed to control the bleeding and air leak.
  3. If less invasive measures do not stop the bleeding, thoracotomy is indicated to find and control the source.


Figure 3 outlines ventilator management and weaning off of ECMO.

A. Decannulation

Cannulas placed by direct cutdown are removed by direct cutdown. The cannulas are removed, and the vessels are often ligated. Neck vessel reconstruction at the time of decannulation is done at the discretion of particular surgeons and centers, but there is no consensus on its utility.13 A percutaneously placed cannula is easily removed, and hemostasis is obtained by holding pressure over the area until the skin is sutured. No further pressure will be required once the skin is closed. Long-term patency of these vessels can be expected.

B. Stopping Support for Futility (Nonreversibility)

ECMO should be stopped if the disease process is nonreversible. This could include severe brain injury, diagnosis of an irreversible process, or organ failure with no possibility of transplant. During the consent process before placing a patient on ECMO, it should be made clear that continuation of ECMO will be reassessed frequently and that ECMO may be discontinued if the patient has not shown any improvement in a reasonable amount of time. The definition of a reasonable amount of time is patient and diagnosis dependent.


The overall survival to hospital discharge is 73% for patients treated with ECMO for neonatal respiratory disease. The overall survival rate to decannulation from ECMO support is slightly higher at 83%. The average ECMO run times have increased during the last 15 years to slightly more than 200 hours per ECMO run, as compared with approximately 150 hours in the 1990s.14 Longer ECMO runs and lower survival suggests that ECMO is being increasingly utilized as a therapy for more critically ill patients with more associated comorbidities.15


Table 3 outlines Neonatal Considerations for ECMO.

Table 3.
Table 3.:
Neonatal Considerations


We thank Elaine Cooley, MSN RN and Peter Rycus, MPH for the help in the overall process.


1. Brogan TV, Lequier L, Lorusso R, MacLaren G, Peek G. Extracorporeal Life Support: The ELSO Red Book. 2017, 5th ed. Ch. 5, 7, 12, 14, 17.
2. Shade BC, Schiavo K, Rosenthal T, Connelly JT, Melchior RW. A single center’s conversion from roller pump to centrifugal pump technology in extracorporeal membrane oxygenation. Perfusion 2016.31: 662–667.
3. Lou S, MacLaren G, Best D, Delzoppo C, Butt W. Hemolysis in pediatric patients receiving centrifugal-pump extracorporeal membrane oxygenation: Prevalence, risk factors, and outcomes. Crit Care Med 2014.42: 1213–1220.
4. O’Brien C, Monteagudo J, Schad C, Cheung E, Middlesworth W. Centrifugal pumps and hemolysis in pediatric extracorporeal membrane oxygenation (ECMO) patients: An analysis of extracorporeal life support organization (ELSO) registry data. J Pediatr Surg 2017.52: 975–978.
5. Delaplain PT, Zhang L, Nguyen DV, et al. Effect of pump type on outcomes in neonates with congenital diaphragmatic hernia requiring ECMO. Perfusion 2018.33(1_suppl): 71–79.
6. Murphy HJ, Cahill JB, Twombley KE, Kiger JR. Early continuous renal replacement therapy improves nutrition delivery in neonates during extracorporeal life support. J Ren Nutr 2018.28: 64–70.
7. Cuevas Guaman M, Lucke AM, Hagan JL, Kaiser JR. Bleeding complications and mortality in neonates receiving therapeutic hypothermia and extracorporeal membrane oxygenation. Am J Perinatol 2018.35: 271–276.
8. Field D, Juszczak E, Linsell L, et al.; NEST Study Collaborative Group: Neonatal ECMO study of temperature (NEST): A randomized controlled trial. Pediatrics 2013.132: e1247–e1256.
9. Khan AM, Shabarek FM, Zwischenberger JB, et al. Utility of daily head ultrasonography for infants on extracorporeal membrane oxygenation. J Pediatr Surg 1998.33: 1229–1232.
10. Biehl DA, Stewart DL, Forti NH, Cook LN. Timing of intracranial hemorrhage during extracorporeal life support. ASAIO J 1996.42: 938–941.
11. Abend NS, Dlugos DJ, Clancy RR. A review of long-term EEG monitoring in critically ill children with hypoxic-ischemic encephalopathy, congenital heart disease, ECMO, and stroke. J Clin Neurophysiol 2013.30: 134–142.
12. Lin JJ, Banwell BL, Berg RA, et al. Electrographic seizures in children and neonates undergoing extracorporeal membrane oxygenation. Pediatr Crit Care Med 2017.18: 249–257.
13. Duggan EM, Maitre N, Zhai A, et al. Neonatal carotid repair at ECMO decannulation: Patency rates and early neurologic outcomes. J Pediatr Surg 2015.50: 64–68.
14. International Summary of ELSO Registry Report. Extracorporeal Life Support Organization (ELSO). 2017. Ann Arbor, Michigan: Published by ELSO online at
15. Sharma J, Rimal A, Weiner J, Haney B, Pallotto EK. Neonatal pulmonary extracorporeal membrane oxygenation (ECMO) practice trends based on primary diagnosis - Two decade comparison. Poster session presented at: Pediatric Academic Society (PAS); May 6–9, 2017; San Francisco, CA.
16. Revenis ME, Glass P, Short BL. Mortality and morbidity rates among lower birth weight infants (2000 to 2500 grams) treated with extracorporeal membrane oxygenation. J Pediatr 1992.121: 452–458.
17. Church JT, Kim AC, Erickson KM, et al. Pushing the boundaries of ECLS: Outcomes in <34 week EGA neonates. J Pediatr Surg 2017.52: 1810–1815.
18. Backes CH, Nicholson L, Rivera BK, Swier N, Marshall W, Cua CL. Extracorporeal membrane oxygenation incidence, characteristics, and outcomes in neonatal down syndrome patients. ASAIO J 2016.62: 477–481.
19. Baerg JE, Thirumoorthi A, Hopper AO, Tagge EP. The use of ECMO for gastroschisis and omphalocele: Two decades of experience. J Pediatr Surg 2017.52: 984–988.
20. Agarwal P, Altinok D, Desai J, Shanti C, Natarajan G. In-hospital outcomes of neonates with hypoxic-ischemic encephalopathy receiving extracorporeal membrane oxygenation. J Perinatol 2019.39: 661–665.
21. Southgate WM, Annibale DJ, Hulsey TC, Purohit DM. International experience with trisomy 21 infants placed on extracorporeal membrane oxygenation. Pediatrics 2001.107: 549–552.
22. Bagdure D, Torres N, Walker LK, Waddell J, Bhutta A, Custer JW. Extracorporeal membrane oxygenation for neonates with congenital renal and urological anomalies and pulmonary hypoplasia: A case report and review of the extracorporeal life support organization registry. J Pediatr Intensive Care 2017.6: 188–193.
    23. Kunisaki SM, Barnewolt CE, Estroff JA, et al. Ex utero intrapartum treatment with extracorporeal membrane oxygenation for severe congenital diaphragmatic hernia. J Pediatr Surg 2007.42: 98–104; discussion 104.
    24. Allen KY, Allan CK, Su L, Mcbride ME. Extracorporeal membrane oxygenation in congenital heart disease. Semin Perinatol 2018.42: 104–110.
    Copyright © 2020 by ELSO