Results
During the study period, there were 181 patients treated with ECMO, of which, 72 had cardiovascular failure and were treated with VA ECMO. Of these, 14 patients were placed on VA-V ECMO. At our institution, all patients on VA ECMO receive a right radial, brachial or axillary artery catheter for blood gas sampling (except for patients who have a right axillary side-graft for reinfusion). There were more males (11/14, 79%) than females in this cohort. Their median age was 54 (range 18 – 75) with median BMI of 30.3 (range 25 – 54). Patient characteristics are summarized in Table 1 .
Underlying diseases included myocardial infarction (4/14), cardiac arrest (3/14, two with concurrently diagnosed pneumonia [PNA]), pulmonary embolism (2/14), anti-hypertensive overdose (2/14, one with coexisting pneumonia), decompensated pulmonary arterial hypertension [PAH], penetrating chest injury, and postpericardiotomy syndrome. Seven patients (50%) were transferred to our institution for ECMO, while 50% (7/14) of patients suffered pre-ECMO cardiac arrest. Our study population had a median SOFA-0 score of 15 (predicted ICU mortality of 95.2%) and the last arterial blood gas prior to ECMO cannulation had a median pH of 7.20, PaO2/FiO2 ratio of 73, and PaCO2 of 50; one patient did not have a documented ABG prior to cannulation. Five patients had a left ventricular ejection fraction ≤ 30% on pre-ECMO echocardiogram, and five patients had moderately or severely reduced right ventricular ejection fraction (RVEF); two patients did not have a documented pre-ECMO transthoracic echocardiogram (TTE), and the RVEF of two patients was not assessed due to technical limitations.
Underlying disease, initial ECMO indication, and indication for VA-V conversion are displayed in Table 2 . Seven (50%) patients were initiated on VA ECMO with a median SAVE score of -12 (risk class V, predicted in-hospital survival of 18%), and the remaining seven patients were placed directly on VA-V ECMO due to present or imminent combined cardiopulmonary failure. The majority of patients initially cannulated on VA ECMO were quickly transitioned to a VA-V configuration, however three patients were on VA ECMO for a significant amount of time. None of the patients in our study population were initiated on a veno-venous configuration. Sequence of ECMO configurations with cannula size and locations are shown in Table 3 . Two patients were cannulated at outside hospitals prior to transfer, and the cannula sizes for one of these patients were not documented in our EMR. A superficial femoral artery distal perfusion cannula was placed in 9 of 13 (69%) patients who were cannulated via the femoral artery, and four patients (29%) had an Impella in place while on ECMO.
Arterial blood gas data is shown in Table 4 . On Day 1 of ECMO, our patients had a median PaO2 of 271 on a median ECMO flow of 4.22 L/min. Indications for VA to VA-V conversion included differential hypoxemia (5/7, 71%), persistent hypoxemia, and respiratory failure secondary to diffuse alveolar hemorrhage.
ECMO adjuvants, anticoagulation strategy, and blood product administration are shown in Table 5 . One patient was only anticoagulated with Impella purge solution, eight (57%) received a heparin infusion, and five (36%) were placed on Argatroban infusion (two of these patients were transitioned from heparin due to concern for HIT). Anticoagulation was held due to bleeding in 79% (11/14) of patients at some point in their ECMO run, and a median of 7.5 units of PRBCs and 2.5 units of platelets were transfused while on ECMO.
Of the 11 patients that demonstrated cardiac and pulmonary recovery allowing for weaning of ECMO support, eight (73%) were transitioned to VV prior to decannulation and three (27%) were decannulated directly from VA-V. The first arterial blood gas following ECMO had a median pH of 7.39, PaO2/FiO2 ratio of 262, and PaCO2 of 49. Six of eleven patients had documented post-ECMO transthoracic echocardiograms, which are shown in Table 6 .
Complications are listed by subcategory in Table 7 . In order of frequency, these included acute kidney injury (93%), clinically significant bleeding (50%), thrombotic non-circuit (36%), limb ischemia (36%), infection (29%), neurologic (29%), and mechanical/circuit (14%). Two patients with limb ischemia required intervention: one patient required bilateral below knee amputation and another with SFA occlusion required thrombectomy. All 4 infectious complications were due to pneumonia. With regards to neurologic complications, 2 patients had embolic strokes and 2 patients developed anoxic brain injury. Additionally, mechanical/circuit complications occurred in 2 patients: 1 patient had oxygenator fibrin accumulation requiring circuit exchange, and another had distal perfusion cannula clotting that required catheter exchange.
Patient outcomes are listed in Table 8 . VA-V support was utilized for 92.4 out of 148.2 total ECMO hours (median values with first and third quartiles displayed). Causes of death on ECMO included cor pulmonale, asystolic cardiac arrest, and withdrawal of care due to persistent cardiogenic shock as well as ARDS with severe hypoxic respiratory failure. Two patients did not survive from decannulation to discharge due to respiratory failure and withdrawal of care in the setting of anoxic brain injury. A total of 79% (11/14) of patients survived to ECMO decannulation and 64% (9/14) survived to hospital discharge, with 82% of decannulated patients surviving to discharge. One patient was transferred back to their original hospital and the remaining patients were discharged to an extended care facility (11%), physical rehabilitation facilities (33%), or home (44%). Cumulatively, these patients spent a median of 12.5 days on a ventilator, 14.5 days in the ICU, and 22 days in the hospital with a median of 12 ventilator free, 28.5 ICU free, and 1.5 hospital free days in 60.