Extracorporeal Membrane Oxygenation for Pregnant and Postpartum Patients

Michael J. Wong, MD; Shobana Bharadwaj, MBBS; Jessica L. Galey, MD; Allison S. Lankford, MD; Samuel Galvagno, DO, PhD; Bhavani Shankar Kodali, MD


Anesth Analg. 2022;135(2):277-289. 

In This Article


Although pregnancy is a hypercoagulable state, characterized by increases in factors VII, VIII, X, XII, tissue factor, von Willebrand factor, and fibrinogen,[80] there is little evidence that pregnancy increases susceptibility to oxygenator or ECMO circuit thrombosis.[35] Nonetheless, vigilance for clot formation within the ECMO circuit is advisable for all patients. Unfractionated heparin is the systemic anticoagulant of choice for ECMO during pregnancy, since it is widely available, inexpensive, reversible, does not cross the placenta, and is not associated with congenital malformations.[73,74,81–83] As in the general population, optimal anticoagulation monitoring for obstetric patients on ECMO has not yet been determined.[84] Table 4 summarizes laboratory and point-of-care tests for monitoring heparin effect during ECMO. At our institution, heparin is given as a bolus during cannulation (50–100 units/kg) and then as a continuous infusion titrated according to laboratory testing and clinical evaluation. For VA-ECMO, our activated partial thromboplastin time (aPTT) goal is 60 to 80 seconds. During VV-ECMO, our aPTT goal is typically 45 to 55 seconds for both pregnant and nonpregnant patients, though a higher goal of 60 to 80 seconds is used in the presence of COVID-19.

Although the threat of circuit thrombosis usually necessitates systemic anticoagulation during general usage, heparin-bonded circuit tubing reduces this risk.[79] There is increasing experience with maintaining ECMO support while withholding systemic anticoagulation, in situations such as hemorrhage. In addition to heparin-bonded circuit tubing, polymethylpentene oxygenators and centrifugal pumps have also promised to reduce the risk of circuit thrombosis. Recent case series in nonpregnant patients indicate that circuit thrombosis risk is not increased in VV-ECMO without systemic anticoagulation, and the practice may decrease bleeding events and transfusion requirements.[85,86] Depending on the duration of ECMO support without systemic anticoagulation, multiple changes of oxygenators or circuit components may be reasonably expected. It is essential to maintain ongoing vigilance for visible circuit thrombosis or flow drops that may promote clotting, and the team must always be prepared to urgently change circuit components if necessary. Serial markers of hemolysis such as lactate dehydrogenase and plasma free hemoglobin further inform ongoing decisions about anticoagulation resumption. At our institution, we have had favorable outcomes in patients continued on prolonged VV-ECMO runs without systemic anticoagulation. Although our experience and comfort with this practice have improved during recent years, we still consider it preferable to maintain systematic anticoagulation during ECMO support, particularly for VA-ECMO that has a 9% incidence of circuit-related clot, which may be catastrophic.[87–89]

When there are indications for fetal delivery, the decision to reverse or hold systemic anticoagulation before delivery must be individualized, as data to inform best practices are few. At our institution, systemic heparin is stopped for at least 2 hours before a scheduled cesarean delivery on ECMO, and we consider restarting it after 24 to 48 hours if hemostasis is adequate. With heparin-bonded components, we have not had frequent thrombotic complications from pausing systemic anticoagulation periprocedurally.

Heparin-induced thrombocytopenia (HIT) during ECMO is an uncommon complication associated with significant thrombotic morbidity and mortality. The risk of HIT in pregnant women on ECMO support has not specifically been evaluated. However, pregnancy itself is potentially protective against HIT, as demonstrated by the relatively lower incidence of HIT among hospitalized pregnant women (below 0.1%) compared to nonpregnant hospitalized patients.[90] Due to the elevated risk of systemic thrombosis in patients with HIT, heparin infusions should be discontinued and alternative anticoagulation agents administered.[91] Argatroban, danaparoid, fondaparinux, and bivalirudin have been used for alternative systemic anticoagulation during ECMO in patients with HIT,[91] and they are also appropriate for pregnant patients.[92,93] If heparin-bonded circuit tubing is used, the heparin coating can potentially contribute to HIT and platelet consumption; therefore, it is preferable to promptly change these components when patient stability permits.

On discontinuation of VV-ECMO, 85% of patients demonstrate cannula site thrombosis, which is independent of their ECMO run time or anticoagulation regimen.[94] During VA-ECMO support, there is a 14% rate of arterial complications, including thrombosis and vessel stenosis, and up to 7% of patients develop critical limb ischemia requiring urgent thrombectomy.[95] While it is unclear whether pregnant patients are more susceptible to thromboembolic complications, given the thrombophilic tendency in pregnancy, it is prudent to perform duplex ultrasound of cannula sites after decannulation. Patients with evidence of cannula site thrombosis should receive therapeutic anticoagulation until sonographic resolution. Additionally, massive pulmonary embolism or severe maternal cardiomyopathy is a relatively common indication for VA-ECMO initiation in pregnant patients, and these conditions would also necessitate prolonged anticoagulation after decannulation.

Hemorrhagic complications are seen in 37% of obstetric patients on ECMO,[15] and this is comparable to the general population.[73] Bleeding may occur from numerous potential sources, including the pulmonary vasculature, thorax, cranium, gastrointestinal tract, cannulation sites, or tracheostomy.[68] If bleeding occurs after fetal delivery, obstetric causes of bleeding must also be considered, such as uterine atony, retained placenta, and traumatic delivery. For uncontrolled hemorrhage, temporary suspension of systemic anticoagulation may be balanced against the risk of circuit thrombosis. Viscoelastic tests help to target treatment of specific coagulopathies.[73]