ECMO should be considered for any patient with severe respiratory compromise that fails to respond to all other available therapeutic measures and is potentially reversible. For patients at the extremes of age and size, technical issues might preclude a successful application of the technology, but it is difficult to define absolute limits of applicability as they continue to change. For larger patients, arterial access is limited due to the adverse effects of ligating the vessel that is cannulated and venous return can be a significant issue, requiring sometimes two or three venous cannulas to support flow. A distal internal jugular cannula (“brain drain”) can provide substantial additional flow even in neonates.
Veno-venous ECMO has become the preferred approach except when there is significant primary cardiac disease and in most babies with CDH, who often have poor venous drainage due to cardiac hypoplasia and severe mediastinal displacement. Especially in older patients, percutaneous VV cannulation has also become available and should be used whenever possible. A significant problem with VV-ECMO has been the development of transient cardiac dysfunction (“stun”), presumably due to cytokine release or electrolyte abnormalities, which sometimes leads to early emergent conversion to VA-ECMO. This has been effectively avoided at The Children’s Hospital of Philadelphia since the introduction several years ago of a protocol that includes: low initial flow through the circuit (10 mL/kg/min) for 8 min, calcium gluconate 100 mg/kg (50 mg/kg in pediatric patients) given as an intravenous bolus during this initial period, and a very slow increase in flow (10 mL/kg/min every 2–3 min) up to the goal of 100–120 mL/kg/min.
Cannulation can be harrowing in neonates and should not be taken lightly. The internal jugular vein tears easily, sometimes requiring ligation and a second or third attempt. One should always use the smallest cannula expected to provide sufficient flow as a large cannula is more likely to injure the vein and can be surprising difficult to place even when the vein appears to be huge. In extremely rare cases, a portion of the clavicle can be resected to expose the vein in the chest or a sternotomy with direct placement of a cannula in the atrium is required. Unfortunately, the right jugular vein is essentially the only vein available in the neonate for safe cannulation and it should therefore be carefully protected. A common problem is for the cannula to pass for a short distance and then for there to be resistance to passage of the last few centimeters into the atrium. Firm pressure might be all that is required but perforation of the SVC or right atrium is a significant concern. It is useful in these situations to: gently spin the cannula as it is being advanced, lift the baby’s thorax off the bed for a short distance, and change the degree of traction applied to the vein in an attempt to straighten it (more traction) or prevent the natural narrowing that occurs when a tubular structure is stretched (less traction). Rather than forcing it and risking a catastrophe, it is usually best to remove the cannula, regroup, and try again with a smaller cannula. Sterile water-soluble lubricant is also useful.
Cannulation of the artery is usually much more straightforward, but there is a risk of placing the cannula in a subintimal plane, which is potentially disastrous. One should be very careful to place the tip of the cannula within the true lumen. If an intimal flap has occurred, it might be prudent to tack it down with fine polypropylene stitches prior to attempting cannulation. Probably the most common mistake in arterial cannulation is in the positioning of the tip of the cannula too far into the aortic arch, which can limit flow into the right arm and coronary arteries. The tip should be advanced no more than about 2.5–3 cm, just enough for it to be secured in the carotid artery. It does not need to be placed within the aorta proper.
When called about a patient who needs to be considered for ECMO, there are certain very specific steps that should be taken: assessment of the patient’s candidacy in the context of available institutional resources; activation of the “ECMO team”; STAT echocardiography and head ultrasound; type and crossmatch for blood needed to prime the ECMO circuit; and informed consent obtained from the parents after a frank discussion of the indications, alternatives, anticipated benefits (ECMO is a bridge, not a therapy in and of itself), and potential risks of the procedure (especially bleeding and intracranial hemorrhage). For every patient, a limit should be agreed upon regarding what length of time is reasonable before considering ECMO a failure.
Ultimately, the goal for any patient on ECMO is to get them off ECMO as soon as possible, and every maneuver every day should be with that goal in mind. It is important to avoid the situation in which the focus is on the minutia of everyday care while the bigger picture is being neglected: avoid fluid overload and start a forced diuresis early; gently increase ventilator settings to recruit native lung function as early as is practical; wean pressors and unnecessary medications; provide adequate intravenous nutrition throughout the ECMO run; and have frequent discussions with the family to review progress and discuss limitations. Finally, when the infant is not responding to therapeutic measures as expected, repeat the echocardiogram a second or third time as some congenital heart lesions can be very difficult to identify early on.