Venous thrombosis after VV ECMO: What is the true prevalence?

Venous thromboembolism is considered one of the most preventable causes of in-hospital death. Venovenous extracorporeal membrane oxygenation (VV ECMO) utilization for severe respiratory failure has increased in the decade following the 2009 influenza A H1N1 pandemic and the publication of the CESAR trial.1 The interaction between a patient’s blood and the ECMO circuit produces an inflammatory response that can provoke both thrombotic and bleeding complications. VV ECMO is associated with a risk of thrombosis in both the circuit and the patient. To mitigate risk of thrombosis in extracorporeal life support (ECLS), the most widely used anticoagulant is unfractionated heparin, both as a systemic agent as well as heparin-bonded circuit components.3 The use of direct thrombin inhibitors in ECLS has also been expanding due to the reliable pharmacokinetics and greater reduction of thrombin as compared to unfractionated heparin. In a systematic review of patients with H1N1 published in 2013, the prevalence of deep venous thrombosis was estimated to be as low as 10 percent; however, more recent data suggests the true prevalence of venous thrombosis after decannulation from VV ECMO is much higher.2,4,6


A retrospective review in 2015 later examined the prevalence of DVT in the cannulated vessel of patients receiving VV ECMO. Utilizing empiric venous duplex at a median of 73 hours, cannula-associated DVT (CaDVT) was identified in 18 percent of survivors, with dual-lumen jugular cannulation having a higher observed rate of thrombosis compared to single lumen femoral cannulation.2 In 2017, Menaker et al. at Shock Trauma assessed the prevalence of CaDVT in 70 patients undergoing VV ECMO for severe ARDS and observed that 85 percent of patients demonstrated venous thrombosis on routine ultrasound 24 hours after decannulation.5  Neither the duration of ECMO nor the level of anticoagulation was associated with development of CaDVT in this study. 


Last year, a cohort of 172 patients weaned from VV ECMO were screened for CaDVT and venous thrombosis was identified in 62 percent of patients after decannulation. Only 6 of the 172 patients underwent CT, while the rest were diagnosed by venous duplex. This was the largest study to date in 2019 assessing cannula-related thrombosis in VV ECMO patients. The authors acknowledged that the modality of duplex sonography limited the identification of CaDVT due to the inability to evaluate the iliac and caval veins.4


This prompted the most recent observational study of CaDVT prevalence published this month in Critical Care Medicine. The authors screened 105 patients with severe ARDS treated with VV ECMO using a novel approach of empiric CT scan to identify venous thrombosis within 4 days of decannulation.6 They observed that CaDVT after VV ECMO is a common event, as CaDVT was identified in 71 percent of survivors. The other striking finding of the study was that 47 percent of clots were “distal thrombi,” located in the vena cava, which would be missed with a traditional bedside ultrasound diagnostic approach. Additionally, 16 percent of patients were diagnosed with pulmonary embolism. While the sample size was relatively small, this is the first study to examine the prevalence of CaDVT after VV ECMO using CT scan. Despite the small size of the French ICU (14 beds), it is an experienced ECMO center with approximately 50 to 60 VV ECMO runs performed per year. The median duration of ECMO was 10 days; however, the duration of cannulation was not associated with development of DVT, as seen in prior studies. The only protective factor for CaDVT observed in this study was thrombocytopenia, despite the thrombocytopenic patients having a greater time with a sub-therapeutic PTT (likely due holding anticoagulation out of concern for bleeding). The authors utilized a heparin protocol targeting PTT 1.2 to 1.5 times normal. They concluded that prior studies using ultrasound to assess the rate of CaDVT are likely underestimating the prevalence, as this modality does not identify clots in the pelvic veins or vena cava.6 


When considering a screening algorithm for DVT after VV ECMO decannulation, it is imperative to take into account a patient's renal function and overall stability to tolerate transport to radiology for CT scan.  Based on the results of these observational studies, the true prevalence of CaDVT is much higher than originally estimated during the H1N1 pandemic (>70 percent versus 10 percent). An algorithmic approach to the diagnosis of CaDVT should be developed at ECMO centers. One method may be to start with venous duplex and, if CaDVT is identified, simply treat and avoid contrasted CT scan. However, if the duplex ultrasound is negative, this data suggests that CT should be pursued to identify highly prevalent distal thrombi located in the vena cava.  



1. CESAR trial. Efficacy and economic assessment of conventional ECMO for severe adult respiratory failure (CESAR): a multicenter randomized controlled trial. Lancet 2009;374(9698):1351-1363.


2. Cooper E et al. Prevalence of venous thrombosis following venovenous extracorporeal membrane oxygenation in patient with severe respiratory failure. Crit Care Med 2015;43:e581-584. 


3. ELSO Anticoagulation Guideline. 2014 Available from:


4. Fisser C et al. Incidence and risk factors for cannula-related venous thrombosis after VV ECMO in adult patients with respiratory failure. Crit Care Med 2019;47(4):e332-339. 


5. Menaker et al. Incidence of cannula-associated DVT after VV ECMO. ASAIO Journal 2017;63:588-591. 


6. Parzy G et al. Prevalence and risk factors for thrombotic complications following VV ECMO: a CT scan study. Crit Care Med 2020;48(2):192-199. 


7. Zangrillo A, et al. ECMO in patients with H1N1 influenza infection: a systematic review and meta-analysis including 8 studies and 266 receiving ECMO. Critical Care 2013;17(1):R30.