ECMO
General Concepts
The primary goal is to provide oxygenated/ventilated blood to the body, performing the work of the lungs (Veno-Venous ECMO), or both the lungs and the heart (Veno-Arterial ECMO) as a temporary measure until the heart/lungs can heal or potentially as a bridge toward transplantation. Hence, ECMO is NOT a therapy or a treatment, but rather a supportive measure to allow for definitive therapy, treatment, or time to heal.
General indications for ECMO include a reversible condition that requires pulmonary or cardiopulmonary support to either sustain life or avoid support that might otherwise be deleterious (very high ventilator settings/toxic FiO2 etc). In the neonatal population, an OI of 40 or more has been proposed as one cutoff.
There exists significant variability in whether ECMO program directors would offer ECMO to various children with cardiopulmonary disease (i.e. 50% of pediatric intensivists asked would offer ECMO to a patient with cerebral palsy/developmental delay who could smile and enjoy school with an acute pneumonia whereas 50% would not).
Relative contraindications included recent neurosurgical procedure or intracranial bleed, recent surgery or trauma, depleted bone marrow/refractory thrombocytopenia, severe neurological compromise, and >14 days of mechanical ventilation (generally associated with worse outcomes).
ELSO Pediatric Respiratory Guidelines, ELSO General Management Guidelines
ECMO accomplishes this goal by taking deoxygenated blood from the body, running it through an oxygenator which also works to remove CO2, and putting it back into the body.
Nomenclature: The first term is where the blood is drawn from the patient and is thus always V for venous. The second/third terms are where the blood is delivered and hence can be V (back to R atrium)-hence VV, A (into the aorta generally via the carotid or femoral artery)-hence VA, or AV (into the aorta and the right atrium via separate cannulae)- hence VAV
Flow in an ECMO circuit functions to deliver the oxygenated blood and can be thought of as cardiac output in VA ECMO and the primary driver for delivery of oxygenated blood in VV ECMO. Flow rate through a catheter is directly proportional to the internal diameter of the catheter and inversely proportional to its length. Resistance to flow decreases by the fourth power of the change in radius as the radius of the catheter increases. This is described by Poiseuille's Equation. Hence, catheter size is critical to achieve adequate flows, particularly in patients who may have significantly increased metabolic demands (ie profound septic shock)
Flow can also be limited despite adequate catheter size if the venous resevoir is not adequate (i.e. the patient is hypovolemic or there is impeded drainage to the right atrium)
Flow is generally started at ~50 cc/kg/min and titrated up based on the patient's oxygen and hemodynamic needs. Infants typically require 100-160 cc/kg/min, pediatric patients ~90 cc/kg/min, and adult patients 50-70 cc/kg/min to maintain adequate oxygen delivery. This can be increased dramatically in the setting of a very hyperdynamic or hypermetabolic state
Overall pediatric survival rates vary by diagnosis, but in general, the ECMO survival rate for respiratory and cardiac support is 56% and 49%, respectively.1
Cannulation
Venous cannulation is generally performed in the right internal jugular vein. It can also be performed in the femoral vein but due to size limitations, is generally limited to adolescents and adults (with some recommendations for femoral cannulation in children greater than 15 kg)
VV ECMO cannulation can occur using only one double lumen catheter (i.e. Avalon or Origen) placed in the R IJ vein.
Avalon has two drainage ports (intended for SVC and IVC) and one reinfusion port (intended at RA) and thus is technically more difficult to place, ideally utilizing fluoroscopy and echocardiography for cannulation. Percutaneously placed in general so that a wire can be placed to track into the IVC.
Origen has single drainage port (SVC) and reinfusion port (RA) and is technically more straightforward to place (can be done with cutdown)
Figure 1: Example of Double Lumen Veno-Venous (VV) Cannula -Avalon
Crescent Cannula Positioning and Radio-opacity
With central cannulation, the venous cannula is placed into the right atrial appendage and the arterial cannula is placed directly into the aorta under direct visualization. Central cannulation has been used by McLaren et al in Australia to provide increased flows for patients with profound septic shock from meningococcemia
Arterial cannulation generally occurs in the right common carotid artery with the cannula advanced to the aortic arch (and verification that it is not directed toward the aortic valve as this can cause severe AI). The femoral artery can also be used in larger patients (at least 15 kg but again, adequate flow for the particular patient has to be considered and most patients cannulated through the femoral artery are >30kg). With femoral cannulation, one also has to monitor for evidence of limb ischemia and often a reperfusion cannula in the posterior tibial artery directed retrograde needs to be inserted. Recent evidence has not shown a significant difference in neurologic complication rates (ie stroke or hemorrhage) with carotid vs alternative cannulation strategies.
Cannula Positioning:
VV: Based on inflow/outflow ports (ie SVC/IVC and towards tricuspid valve/RA respectively). This corresponds to about T4-5/T8-10/and T 6-7, respectively on CXR)
VA: Venous cannula with tip at RA/IVC junction (generally around T6-7) and arterial cannula at T2-T4 (RCC/Transverse arch bifurcation) on CXR. The arterial cannula should at least overlay the midline trachea.
If any questions about cannula position, echocardiogram can help identify
Figure 2: Example of VA ECMO Circuit
Figure 3: VV ECMO Circuit Components Schematic (From University of Iowa)
Figure 4: ECMO Circuit Components
In patients with profound cardiac dysfunction who cannot adequately eject blood from their left ventricle, blood can back up, causing left atrial hypertension and subsequent pulmonary venous hypertension that can lead to pulmonary edema and hemorrhage. Some signs of this can include pink frothy secretions from the ETT or diminished/absent pulse pressure on the arterial line tracing. These patients require emergent placement of a left atrial drainage catheter or balloon atrial septostomy in the cardiac catheterization lab to relieve this pulmonary venous congestion.
With femoral artery cannulation, if the patient regains adequate native cardiac function, but still has significant lung injury, oxygenated blood from the ECMO circuit may not reach the cerebral and coronary circulation due to the increased afterload from native cardiac output. Hence, the lower extremities may be well oxygenated with relative cyanosis and hypoxemia in the upper body. This is sometimes referred to as north-south syndrome. Monitor O2 sats at the level of the head (i.e. ear) to ensure adequate cerebral oxygenation.
Femoral artery cannulation can also impair limb perfusion distal to the cannula, requiring insertion of a reperfusion cannula directing blood to poster tibial artery to provide distal flow.
If possible, VV ECMO is generally preferred over VA ECMO due to the risk of thromboembolic phenomenon to the cerebral circulation given a cannula inserted directly into the arterial circulation. For example, patients with severe ARDS may have some myocardial dysfunction primarily due to hypoxemia and acidosis that may improve simply with adequate oxygenation/ventilation provided by VV ECMO rather than necessarily requiring cardiac support via VA ECMO. However, VA ECMO is generally more straightforward in terms of cannulation and if significant doubt re: VV or VA persists, it is reasonable to proceed with VA cannulation and consider conversion to VV (if necessary) at a later time
Echocardiography is the study of choice to clarify cannula position
Parts of the ECMO Circuit
Cannulas: site and size determined by patient's clinical status as well as size and flow requirements
Bladder: 30-50 ml reservoir at the lowest point of the ECMO circuit that serves as an air trap and typically has an access port to remove any air collected. Also serves as a "shock absorber" in that blood that drains into the bladder is gravity dependent and then gets drawn actively via the ECMO pump. If the negative pressure generated by the pump gets too high, the very compliant bladder collapses down, which sends a signal to the pump to slow down, relieving the high negative pressure.
Pump: Mechanism for drawing the blood into the circuit and then back into the patient. Typically, roller pumps with positive displacement of blood have been used although newer centrifugal pumps are being used more frequently.
Oxygenator: Oxygenates incoming blood via hollow fibers and countercurrent exchange with deoxygenated blood going one direction and oxygen the other direction. Typical postoxygenator PO2's are >350 mmHg. Also removes CO2 very efficiently when oxygen is connected. This oxygen flowing through is termed sweep and is measured in liters/min (ie 2 lpm of sweep) and can be increased/decreased to increase/decrease CO2 clearance. Comes in various sizes and can use more than one in line if needed.
Venous saturation monitor: used to monitor the oxygen saturation of venous drainage in the ECMO circuit. Can give an indication of recirculation if very high, or inadequate oxygen delivery if very low (or extraction very high)
Heat exchanger: Used to set/maintain the temperature of the blood as otherwise, the extracorporeal circuit acts like a huge radiator and leads to patient hypothermia. As the temperature is set, it is difficult for patients on ECMO to become febrile.
ECMO also offers the opportunity to perform CRRT or plasma exchange therapy without the need for other access as these therapies can occur with access to the ECMO circuit itself
Anticoagulation on ECMO
Blood entering the artificial ECMO circuit has a tendency to clot and thus the blood running through the circuit has to be anticoagulated. While anticoagulation helps to prevent thromboembolic complications, it also puts the patient at risk of bleeding. Unlike the regional anticoagulation using citrate with CRRT, the much higher flows (i.e. 100+ ml/kg/min for ECMO vs. 5 ml/kg/min for CRRT) with ECMO would lead to using so much citrate that it would overwhelm the liver's ability to clear it. Hence, patients require systemic anticoagulation which is achieved with heparin or bivalirudin
Heparin Bolus of 50units/kg followed by bival infusion immediately after initiation or if using heparin, 1 hour later (LPCH practice)
Bivalirudin:
Direct thrombin inhibitor, does not require AT3 for antiothrombotic activity (unlike heparin).
Approximately 20% renal excretion. Rapid onset of approximately 2 min via continuous IV infusion, half life of ~25 min with normal renal function, will prolong INR
Avoid stasis of blood as this will lead to thrombus formation
Must use C clamp on bridge
Pressure transducers are set up with
Generally ACT goals are targeted, with UM's goals of 210-230 seconds for standard patients. This can be downtitrated to 190-210 or even 170-190 seconds if there is concern for bleeding. ACT measurements vary by machine utilized and thus these goals may not be applicable across institutions. Heparin activity level (HAL) is sometimes used when utilizing heparin and aPTT (goal 70-90sec) is used when utilizing bivalirudin (see possible dose adjustment guidelines below)
Significant bleeding often requires surgical intervention. Other interventions such as aminocaproic acid infusion, topical thrombin, factor VII, and running the circuit briefly without heparin may be used as well
Platelets are generally kept >100K to avoid significant bleeding (although 50K is sometimes tolerated if there is significant ongoing platelet consumption)
Occasionally, heparin requirements may progressively increase. This can occur with consumption of clotting factors, as heparin requires AT3 to function; this can be treated with AT3 or FFP infusion
HIT can occur (generally 5-15 days after initial exposure to heparin) but is rare in the pediatric population. This requires transition to an alternative anticoagulant (ie lepirudin or argatroban)
Bleeding occurs in ~70% of pediatric ECMO patients and is independently associated with mortality. Hemolysis is associated with thrombosis but not associated with mortality.
Thromboelastography (TEG) is also used by some centers to provide further information regarding the clotting cascade and anticoagulation.
Courtesy of Richard Pierce, MD
Patient Selection
No definite criteria exist although in general, the disease process should be reversible and patients have generally failed to respond to other therapies (typically conventional mechanical ventilation, +/- HFOV, +/- iNO, +/- moderate/high dose vasoactive agents). In general, patients should not have significant neurologic damage or a significant ongoing bleeding diathesis
For example, in neonatal patients, IVH >grade 1 is a contraindication to ECMO
Previous selection criteria have used indices of severity such as the oxygenation index (OI) where OI= 100* [(MAP*FIO2)/PaO2] where a value greater than 40 was used. While useful as a serial measure of hypoxemic respiratory failure, OI has not been shown to be a reliable marker for ECMO eligibility
There exists significant variability in whether ECMO program directors would offer ECMO to various children with cardiopulmonary disease (i.e. 50% of pediatric intensivists asked would offer ECMO to a patient with cerebral palsy/developmental delay who could smile and enjoy school with an acute pneumonia whereas 50% would not.
VV ECMO
Provides respiratory support, requires the heart to pump the well oxygenated blood returning from the ECMO circuit
Allows the clinician to turn down the ventilator settings so as to avoid significant barotrauma and ventilator induced lung injury. Typically rest settings are used (i.e. 20-25/5-10 x a low physiological rate of 10-15 at 30-40% FIO2). Adequate ECMO support should allow the patient to have an oxygen saturation of 80% or above (see ELSO guidelines here)
Nonetheless, well oxygenated blood entering the pulmonary circulation may reduce pulmonary vascular resistance and thereby improve right heart function while well oxygenated blood reaching the coronaries may also improve left ventricular function. Hence, VV ECMO is a potentially viable option for primary respiratory disease with cardiac dysfunction as a consequence of respiratory impairment
Recirculation occurs when well oxygenated blood returning from the ECMO circuit gets "sucked in" by the venous drainage cannula rather than going forward through the pulmonary circulation. A very high ECMO circuit venous oxygenation reading may be indicative of recirculation. ELSO Recirculation Guidelines
Generally have lower systemic oxygenation than with VA ECMO due to proportionally less cardiac output captured; hence, more native blood goes through the lungs which contribute little/no oxygenation and thus the end result is some systemic hypoxemia. For instance, if the total cardiac output is estimated to be 5 L/min and you capture 2.5 L/min through ECMO, you only oxygenate 50% of the blood. Hence, if the mixed venous oxygen saturation is 60%, and assuming the oxygenator is working correctly and providing 100% saturated ECMO blood, then your systemic oxygen saturation would be around 80%
You can also estimate the proportion of cardiac output you are capturing in the ECMO circuit (assuming the lungs do not contribute) using the known variables of ECMO flow, ECMO arterial saturation of 100%, measured mixed venous oxygen saturation, and systemic oxygen saturation via pulse oximetry using the above principle. (ie if your ECMO flow is 2 L/min, ECMO arterial sat is 100%, ECMO venous saturation is 60%, and systemic oxygen saturation is 80% on pulse ox then 2L/min (100%) + X l/min (60%) = (80%)(2+x). Solving for X= 2 l/min meaning 2 l/min is not being captured by the ECMO circuit and going through the pulmonary circulation.
Inadequate oxygen delivery (a function of oxygen content and cardiac output) can be augmented by increasing content (via blood transfusion) or increasing ECMO flow (analagous to cardiac output). Hence, if there is evidence of inadequate oxygen delivery (i.e. rising lactates, high AVO2 difference, etc.), increasing the flow can be therapeutic.
Known as the Cilley Test at the University of Michigan (named after a former fellow), turning the ventilator oxygen FIO2 to 100% can give one an indication of how the patient's lungs are functioning as if there is a significant rise in SpO2, then the patient's lungs are clearly contributing whereas lack of any response indicates lack of significant lung recruitment/response, or alternatively, your ECMO flow is too high and capturing nearly all the cardiac output. Additional info re: Oxygen challenge
Trialing off VV ECMO simply involves placing the patient on more supportive ventilator settings (vs. rest setttings) and disconnecting the sweep gas, which then removes any oxygenation/ventilation occurring via the ECMO circuit. The ECMO circuit then becomes simply a large extracorporeal diversion taking blood from a vein and putting it back into the veins
Trialing off VV ECMO can continue for as long as needed to establish that the patient can successfully decannulate from ECMO (typically 12-24 hours but can be extended to days if needed).
VA ECMO
Provides respiratory and cardiac support, with flow being the main limiting factor in the amount of support that can be provided
Patients with severe myocardial dysfunction may develop backup of blood (as evidenced by poor pulse pressure, flattened arterial line tracing, pulmonary edema/hemorrhage, and echo findings of left atrial dilation) that causes left atrial hypertension->pulmonary venous hypertension->pulmonary edema and hemorrhage. Elevated LVEDP can also compromise myocardial coronary perfusion pressure (the difference between Diastolic blood pressure and RA pressure normally but DBP-LVEDP when LVEDP is > RA pressure. These patients require emergent insertion of a left atrial drain or balloon atrial septostomy in the cardiac catheterization lab
Avoids recirculation but due to the presence of a cannula going directly to the artery (vs. a vein with the subsequent natural filter of the human lung), there is an increased risk of thromboembolic phenomenon (ie CVA)
Trialing off VA ECMO is more complicated than with VV ECMO, as the circuit must be clamped off from reaching the patient. (Simply removing the sweep gas that provides oxygenation/ventilation like when trialing off VV ECMO would just lead to a huge venoarterial (right to left) shunt). However, the circuit cannot just be clamped off as it would develop stasis. Hence, the presence of a bridge between venous and arterial cannulas that allows the circuit to continue flowing while the patient is temporarily clamped off ECMO support. Nonetheless, the area of tubing from the cannulas in the patient to the bridge undergo stasis when the ECMO circuit is clamped so that portion of tubing/cannula has to be flashed (clamps opened and blood allowed to flow) every 15 minutes to prevent clotting. The sweep gas should be removed while trialing otherwise the patient will receive a bolus of very hypocarbic/alkalotic blood with each flash. This clearly demands more careful attention to possible clotting, patient response, and ECMO technician time than trialing off VV ECMO and thus is generally performed for ~ 4 hours.
ECPR
Resuscitative tool for patients in cardiac arrest with overall survival to discharge of ~34%
Requires circuit that is primed or can be rapidly primed and staff that can be readily mobilized both for cannulation as well as running the ECMO circuit
Essentially emergent placement onto VA ECMO generally while undergoing active CPR
Complications
Bleeding is a major complication. Occurs primarily at cannula or surgical sites (hence, lines/tubes are generally not removed while on ECMO and invasive procedures generally require electrocautery to ensure adequate hemostasis). Can be managed by surgical intervention, decreasing the rate of heparin/lowering ACT goals (or even discontinuing heparin for a limited period of time), using aminocaproic acid infusions (100 mg/kg load followed by infusion of 25-50 mg/kg/hr), factor VIIa etc.
Cannula displacement (echocardiography is the test of choice to evaluate proper positioning)
Cerebral infarction as a result of thromboembolic phenomenon (more common with VA ECMO)
Infection as a result of indwelling catheters, large extracorporeal circuit
Evidence
No clear evidence (ie RCT's) that ECMO is superior to conventional therapy (ie mechanical ventilation). However, the CESAR Trial (Lancet 2009) and Noah's study (Referral to ECMO Center H1N1) of the H1N1 epidemic (JAMA 2011) both demonstrated improved outcomes when adults were referred to ECMO centers (even though ~25% of those referred did not end up receiving ECMO). Thus, it is difficult to say whether cointerventions at the ECMO referral site vs. ECMO were the primary factors driving improved outcomes.
Propensity and case matched analysis by Barbaro et al using the RESTORE database also shows no significant difference in outcomes (mortality and secondary outcomes) between those supported by ECMO vs. those not supported by ECMO (pediatric ARDS patients).
Combes et al, in a randomized multicenter RCT of 249 adult patients with severe ARDS (P/F <50 for 3 hours, P/F <80 for 6 hours, or pH <7.25 and pCO2 >60 for 6 hours) demonstrated no significant difference in mortality utilizing VV ECMO vs. conventional mechanical ventilation with crossover (to ECMO) (RR 0.76, 0.55-1.04, p=0.09) (Combes et al, NEJM 2018)
Troubleshooting
Insufficient Flow
Flow is generally started at ~50 cc/kg/min and titrated up based on the patient's oxygen and hemodynamic needs. Infants typically require 100-160 cc/kg/min, pediatric patients ~90 cc/kg/min, and adult patients 50-70 cc/kg/min to maintain adequate oxygen delivery. This can be increased dramatically in the setting of a very hyperdynamic or hypermetabolic state
If there is insufficient flow (as manifested by inadequate oxygen delivery, hypotension (on VA ECMO), hypoxemia (on VV and/or VA ECMO), there could be several causes:
Venous side: This is the most common source of flow issues. Are your negative pressures on the venous side high (ie -100 mmHg or more depending on the circuit). This could be due to a number of factors including:
Cannula is too small and associated with excess resistance. May need additional venous drainage (ie femoral venous drainage) or upsizing of catheter
Cannula is malpositioned: Determine cannula position and reposition.
Venous reservoir is low- hypovolemia, third spacing, bleeding. Treat with volume administration or blood.
High intraabdominal (ie abdominal compartment syndrome) or high intrathoracic pressure (tamponade, tension pneumothorax): remove fluid/air from the associated compartment
Oxygenator: Has the gradient across the oxygenator increased? This could be reflective of increasing clot burden across the oxygenator and could potentially restrict flow although you should also see less oxygenator efficiency (ie post oxygenator blood gas with a lower PaO2 than prior)
Arterial side: Has your positive outlet pressure increased significantly? This could be due to a number of factors including:
Cannula is too small and associated with excess resistance
Cannula is malpositioned
Clot within the arterial limb/cannula
High SVR due to recovering myocardium, excessive vasopressors and peripheral vasomotor tone, seizures, inadequate sedation
Hypoxemia
VV ECMO
Are you recirculating? Check your SvO2 sat to ensure it is not spuriously elevated due to recirculation
Has your ECMO flow decreased?
Is your patient particularly hypermetabolic and consuming more O2? (ie their SvO2 has decreased, and more importantly, their AVO2 difference (extraction) has increased)
Is your oxygenator working (ie oxygen is attached, not excessively clotted, the blood post oxygenator is visibly red). This would be a rare contributing factor to hypoxemia.
Are your cannulas in appropriate position? In general, if your flows have not changed and you are not recirculating, it is likely your cannulas are in reasonable position
Does your patient have a higher cardiac output now? Remember, your ECMO flow, and thereby the amount of red (well oxygenated blood) your patient receives is relatively fixed. What determines your patient's systemic oxygen saturation is the relative mixture of ECMO flow with native cardiac output (that with sick lungs, is not getting oxygenated). If cardiac output goes up and relatively more native blood flow goes through the lungs without first going through ECMO and does not get oxygenated, your patient will become more hypoxemic.
Does it matter? If your patient has an SpO2 of 85% but has a SvO2 of 70% (extraction of 15%), normal hemodynamics, and normal perfusion and lactates, this can be perfectly acceptable (and even expected)
VA ECMO
Has your ECMO flow decreased?
Is your patient particularly hypermetabolic and consuming more O2? (ie their SvO2 has decreased, and more importantly, their AVO2 difference (extraction) has increased)
Is your oxygenator working (ie oxygen is attached, not excessively clotted, the blood post oxygenator is visibly red). This would be a rare contributing factor to hypoxemia.
Does your patient have a higher cardiac output now? Remember, your ECMO flow, and thereby the amount of red (well oxygenated blood) your patient receives is relatively fixed. What determines your patient's systemic oxygen saturation is the relative mixture of ECMO flow with native cardiac output (that with sick lungs, is not getting oxygenated). If cardiac output goes up (i.e. your sick heart is getting better) and relatively more native blood flow goes through the lungs and does not get oxygenated, your patient will become more hypoxemic.
Does it matter? If your patient has an SpO2 of 85% but has a SvO2 of 70% (extraction of 15%), normal hemodynamics, and normal perfusion and lactates, this can be perfectly acceptable
Hypotension
VV ECMO
VV ECMO is not going to help you with hypotension and outside of massive bleeding or some other catastrophe, ECMO should not be contributing to hypotension.
Normal management of hypotension (ie volume as needed, vasoactive agents as needed, etc)
VA ECMO
Increase flow (as this is equivalent to your cardiac output minus whatever contribution your native heart is doing). If you are maxed out on flow (ie due to excessively high negative presure or shearing in your circuit etc), usual hypotension management including fluid, vasoactive agents etc.
Bleeding
Prevention is key- do not insert new catheters or remove existing catheters (even peripheral IV's) if at all possible. Exercise extreme caution and be judicious if your patient needs an invasive procedure- surgical colleagues and electrocautery should generally be utilized.
Can you safely lower your ACT goals? If you are at 210-230s, consider decreasing them to 190-210 or even 170-190 depending on the severity of the bleeding.
Are your platelets, fibrinogen levels, etc ok? We generally aim for platelets >100, Fibrinogen >100, and transfuse FFP if we are giving other products frequently or have escalating heparin requirements without response in ACTs (ie suspicion of AT3 deficiency). One can also measure AT3 directly and give AT3 concentrate.
Surgical colleagues can often help with surgicell, gelfoam, topical thrombin, etc
Aminocaporic acid infusions can be utilized. See UM protocol here.
References
1. Paden ML, Conrad SA, Rycus PT, Thiagarajan RR; ELSO Registry. ExtracorporealLife Support Organization Registry Report 2012. ASAIO J. 2013
2. Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, Thalanany MM, HibbertCL, Truesdale A, Clemens F, Cooper N, Firmin RK, Elbourne D; CESAR trial collaboration. Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet. 2009 Oct 17;374(9698):1351-63.
3. Stanford ECMO Clinical Guidelines (internal)