Outcomes of Extubated COVID and Non-COVID Patients Receiving Awake Venovenous Extracorporeal Membrane Oxygenation

Extracorporeal membrane oxygenation (ECMO) is an efficacious treatment option for respiratory failure refractory to conventional therapy such as prone positioning and optimization of volume status.^1-4 Previous studies have focused on this modality as a potential means of improved survival for patients with severe acute respiratory distress syndrome (ARDS).^5-7 The use of ECMO varies by institution and available resources; however, as ECMO continues to evolve, it has increasingly been used to support patients with respiratory failure as a consequence of various infectious etiologies, including severe influenza and coronavirus-19 (COVID-19 or COVID). The standard of practice for most centers has been intubation with prolonged continuous analgesia and sedation before, during, and after ECMO. In recent years, ECMO has been initiated early to avoid the detrimental effects of prolonged mechanical ventilation and sedation. In addition, awake ECMO has gained interest as an alternative to mechanical ventilation in awake, nonintubated, and spontaneously breathing patients with respiratory failure.⁸

The Extracorporeal Life Support Organization (ELSO) has identified select patients with respiratory failure receiving ECMO support which may be safely managed without invasive mechanical ventilation. An ELSO guidance document for endotracheal extubation in patients with respiratory failure receiving VV ECMO describes considerations for readiness to extubate, including hemodynamic stability, patient’s ability to protect their airway, cooperativeness without requiring frequent analgosedation, and manageable secretions. In addition, patients should have an acceptable arterial blood gas with minimal ventilator settings (e.g., FiO₂ 0.4 and PEEP 5) before extubation.⁹

The potential benefits of an awake ECMO strategy include use of respiratory muscles and diaphragm, which maintains functional residual capacity, a decrease in exposure to continuous infusion analgesic and sedative medications, which in turn, helps to facilitate participation in physical rehabilitation, increases in communication and engagement with family and ICU staff, and decreases the risk of delirium, and subsequently, ICU and hospital LOS and mortality. There is also a decreased risk of ventilator-associated events such as ventilator-associated pneumonia, barotrauma, and increased oral nutrition.⁹ The benefits of this strategy, however, must be balanced with risks such as insufficient gas exchange support, increased work of breathing and energy expenditure, decreased ability to suction and provide secretion clearance, cannula displacement, and potential to experience pain, discomfort, and anxiety during the ICU stay.^8,9

There are limited studies analyzing awake ECMO as an alternative strategy to invasive mechanical ventilation. Yeo et al.¹⁰ retrospectively analyzed 10 patients who underwent awake ECMO due to postoperative ARDS and reported successful weaning in eight patients (80%), with a total ECMO duration of 9.13 ± 2.2 days (range, 6–12 days) and mean ventilator use duration of 6.8 ± 4.7 days (range, 2–16 days). Data regarding the use of an awake ECMO strategy remains limited, with most reports limited to descriptive data, minimal outcome data, and small sample sizes. Therefore, the purpose of this study was to evaluate our outcomes in patients with ARDS patients receiving awake venovenous (VV) ECMO.

Materials and Methods

This is a retrospective, observational analysis of patients diagnosed with respiratory failure and failing conventional therapies who underwent ECMO from November 2018 to April 2021. Before and during the COVID-19 pandemic, a consecutive cohort of patients requiring VV ECMO was included in this analysis. During this time frame, the practice caring for awake VV ECMO patients by members of the cardiovascular team at Rush University Medical Center remained consistent. Conventional therapies before ECMO cannulation included lung-protective ventilation, prone positioning, neuromuscular blockade, and volume optimization. Patients initially cannulated on ECMO for combined cardiorespiratory support and those with multiorgan failure were not included in the analysis. Data were collected through the use of electronic medical records. The study was approved by the local Institutional Review Board, and a waiver of consent was obtained. The primary outcome was survival to hospital discharge. Secondary outcomes included days requiring ECMO, time from cannulation to extubation, complications, patients requiring tracheostomy, hospital length of stay (LOS), intensive care unit (ICU) LOS, and discharge disposition. In a subgroup analysis, survival to hospital discharge, days requiring ECMO, time from cannulation to extubation, patients requiring tracheostomy, hospital LOS, ICU LOS, and discharge disposition were compared between non-COVID and COVID awake ECMO patients.

Cannulation Criteria and Strategies

Patients age <70, body mass index <50 kg/m², and those who failed to improve despite maximal supportive care were cannulated on ECMO. Indications for cannulation followed EOLIA study guidelines, including acidosis, hypoxia, or hypercarbia despite maximal ventilator support and medical therapy, including sedation, neuromuscular blockade, and inhaled nitric oxide.¹¹ Cannulation criteria and a decision-making tree for cannulation strategy are included in Table 1 and Figure 1.

Patients were cannulated using the internal jugular and femoral vein approach (Fem-IJ) or single-site dual lumen approach (ProtekDuo or Crescent). Under fluoroscopic and transesophageal echocardiographic guidance using the Fem-IJ approach, a super-stiff Amplatz wire was utilized, and in standard Seldinger fashion, the sites are dilated and cannulae placed. When using internal jugular and femoral cannulation, the cannulae were separated by at least 10 cm on fluoroscopy to minimize mixing of the inflow and outflow. A ProtekDuo or Crescent was placed under ultrasound-guided percutaneous access of the right internal jugular vein. For the ProtekDuo, a Swan was advanced through this access and a double-J Lunderquist wire placed under fluoroscopic and transesophageal echocardiographic guidance. The site was then dilated, and the cannula was placed. For the Crescent cannula, a super-stiff Amplatz was placed into the inferior vena cava under fluoroscopic guidance. The site was dilated, and the cannula was placed using both fluoroscopic and echocardiographic guidance. With the Crescent cannula, it is imperative that echocardiographic guidance is utilized to ensure the outflow jet is positioned to cross the tricuspid valve. This often requires both craniocaudal and rotational manipulation. For our cohort of patients with COVID-related ARDS, our institution favored the use of the ProtekDuo due to the stability of this cannulation strategy.

Considerations for two-site cannulation included stability at the time of cannulation (two-site cannulation is expeditious and requires less technical acumen), expected duration of ECMO, diagnosis prompting ECMO cannulation, and resources available at the time of cannulation.

Patients were supported with a Medtronic Cortiva-coated circuit with a Maquet Quadrox or Medtronic MC3 Nautilus oxygenator. These circuits were placed on the LivaNova Revolution or the Abbott CentriMag continuous flow pumps. The LivaNova Revolution is an open impeller design with a ferrous particle impregnated nylon magnet with no seals and low friction bearings. The CentriMag pump utilized MagLev flow technology with a free-floating magnetically levitated rotor with no seals, bearings, or valves for streamlined blood flow.

Our Awake Extracorporeal Membrane Oxygenation Approach

Our approach to ECMO includes a goal of extubation. Oxygen content is not determined by oxygenation saturation and is optimized to meet the metabolic demands of the patient as evidenced by normal lactate and mixed venous oxygen saturations; thereby ensuring optimal end-organ perfusion.¹² Adequate oxygen content was achieved by maximizing hemoglobin concentrations. The timing of extubation is dependent on several factors, including lung contribution, time on the ventilator, and dosing and duration of opioids and sedatives. To prepare for extubation, paralysis was discontinued upon ECMO cannulation. Subsequently, in the early phases of ECMO, continuous infusions such as propofol were weaned with the use of enteral agents such as quetiapine. Enteral benzodiazepines were added depending on the dose and duration of continuous benzodiazepines before ECMO cannulation. Continuous infusions of opioids were weaned with the use of enteral opioids. Dexmedetomidine was continued after extubation and was the last continuous infusion to be weaned off on a patient. Oral clonidine was used to help facilitate weaning of dexmedetomidine. Episodes of hypoxemia were tolerated if end-organ perfusion was not compromised; however, severe episodes of agitation leading to desaturation were treated with an increasing sweep or intravenous pushes of analgesics and sedatives. Extubation is considered if patients are awake and without the need for high doses of continuous infusion sedatives, able to cough, follow commands, tolerate minimal sedation and enteral medications, stable flows on the ECMO circuit without significant recirculation, no evidence of bleeding with optimization of hemoglobin (above 10 mg/dl), and hemodynamic stability without the need for frequent fluid boluses. In addition, lactate levels less than 2 mmol/L and arterial saturations over 70% while awake without maximal flow or sweep settings were targeted to allow for increases in ECMO support after extubation (Table 2). Once liberated from mechanical ventilation, patients participated in physical therapy, occupational therapy, and speech therapy. Lastly, once hemostasis was achieved, anticoagulation was considered. Historically, unfractionated heparin (UFH) has remained the standard anticoagulant of choice and is used in our non-COVID ECMO group.¹³ Although UFH has a favorable pharmacokinetic profile and antidote, concerns surrounding the use of UFH include its nonlinear effects, variable half-life, and dosing dependence on antithrombin concentrations, which has been shown to decrease with the institution of ECMO.^14,15 Given these concerns and the hypercoagulable state described in COVID-19 patients, anticoagulation with a direct thrombin inhibitor, bivalirudin, was started at our institution immediately after cannulation once hemostasis was achieved and continued until bleeding precluded further use. Despite the increasing use of direct thrombin inhibitors, the optimal anticoagulant and intensity of anticoagulation have remained an area of research interest in patients receiving ECMO.

During the maintenance phase of ECMO, episodic periods of hypoxemia were identified; however, laboratory parameters and hemodynamic goals were met to ensure these episodes were not associated with compromised end-organ perfusion. While secondary infections are anticipated given the prolonged LOS and need for invasive lines, including ECMO cannulas for recovery, removal of central access as soon as feasible is highly considered. During this phase, oral opioids and sedatives are weaned gradually to allow patient interaction and enhance participation in physical therapy, occupational therapy, and speech therapy. Dyspnea and tachypnea are managed by increasing sweep and hemoglobin optimization. While arterial blood gases are evaluated, the clinical condition of the patient is considered in the decision to wean from ECMO. In addition, improved lung recruitment occurred over time using incentive spirometry, ambulation, and physical therapy. Formal chest physiotherapy and routine bronchoscopy were not applied. Avoidance of positive pressure and repeated manipulation of the airway are both important. Instrumentation of the pleural spaces should be undertaken when needed; however, timing should be optimized due to a significant risk of bleeding. Diuretics are considered in this phase, given that patients are fluid avid in the early phase of ECMO. Caution must be exercised as aggressive diuresis in the setting of unstable flows, or line chattering can increase the risk of acute kidney injury (AKI). Activity is progressively increased from standing in a KREG bed (Kreg Therapeutics, Melrose Park, IL) to ambulation in the ICU. Lastly, patients are advanced from enteral feeds to oral diets as tolerated.

In the late phase of ECMO, weaning of sweep is performed if CO₂ is within the goal and there is no evidence of dyspnea or tachypnea at rest. Permissive hypercarbia may be considered if the patient is otherwise stable. Lung recovery is demonstrated by improved aeration on chest X-ray, appropriate blood gas parameters, return to normal saturations, activity tolerance, and limited need for supplemental oxygen (less than 6 L nasal cannula at rest and 15 L high flow nasal cannula with significant exertion). If lower sweep and flows are tolerated, FiO₂ is weaned to 40%. Before decannulation, patients are trialed off the sweep for at least 24 hours and must ambulate off the sweep.

Data Analysis

Analyses were performed using Statistical Package for the Social Sciences (SPSS, Inc, Armonk, NY), version 22.0. Continuous data were reported as the median and interquartile range and analyzed using the Mann–Whitney U test. Categorical data were reported as frequencies and percentages. The χ² or Fisher exact tests were used for nominal data. Days requiring ECMO, hospital LOS and ICU LOS, and discharge disposition were assessed in patients who survived. A p-value of less than 0.05 was considered statistically significant.

Results

Baseline Characteristics

A total of 62 patients were included in the analysis. A majority of patients were male (58.1%) and Hispanic (48.4%) with a median age of 46.5 (39.5–52.3) and median BMI of 34.2 kg/m² (28.3–40.0) (Table 3). Coronavirus-19, influenza, and bacterial pneumonia comprised the majority of indications for ECMO. The initial cannulation strategy for most patients was a single-site dual lumen approach (ProtekDuo) (64.5%) or Fem-IJ approach (25.8%). The median PaO₂/FiO₂ ratio before cannulation was 78.9 (62.8–102.0), median initial PCO₂ before cannulation was 56.0 mm Hg (47.5–69.0) and median PaO₂ was 74.5 mm Hg (61.2–97.5). Serum creatinine was similar between on the day of admission (0.95 mg/dl [0.74–1.31]) and on the day of cannulation (0.91 mg/dl [0.65–1.71]) (Table 4).

Baseline characteristics, including age, sex, BMI, sequential organ failure assessment scores before ECMO cannulation, and comorbidities were similar between COVID and non-COVID ECMO groups. Influenza combined with respiratory failure due to bacterial pneumonia comprised the majority of indications for ECMO in the non-COVID group (Table 3). The COVID ECMO group had a higher PaO₂ before cannulation (79.5 [64.5–105.3] vs. 67.5 [58.3–82.8], p = 0.021) whereas both groups had similar initial PCO₂ before cannulation (58.0 mm Hg [49.5–71.3] vs. 54.5 mm Hg [46.0–64.0], p = 0.270). Serum creatinine was similar between groups on the day of admission (0.96 mg/dl [0.75–1.31] vs. 0.88 mg/dl [0.72–1.33], p = 0.826) but higher in the non-COVID ECMO group on the day of cannulation (0.76 mg/dl [0.60–1.30] vs. 1.42 mg/dl [0.74–3.02], p = 0.002) (Table 4).

Complications were reported on all awake ECMO patients included in this analysis. Acute kidney injury was defined as a serum creatinine increase > 0.3 mg/dl within 48 hours as compared to baseline or requiring renal replacement therapy (RRT) and reported in 18 patients (29.0%) in the total cohort. AKI was more prevalent in the non-COVID ECMO group (11.8% vs. 50.0%, p = 0.002). Additionally, venous thromboembolism was reported in 14 patients in the total cohort (22.6%). Bleeding and infection were reported in 80.6% and 82.3% of patients, respectively, and were similar in both COVID and non-COVID ECMO groups (Table 5).

The primary outcome of survival to hospital discharge was 85.5% in the cohort. Days requiring ECMO was 33.0 (0.0–75.0), and the median days from ECMO cannulation to extubation was 6.0 (4.0–11.0). There were 23 (37%) patients requiring reintubation. Three patients in the cohort received a tracheostomy (4.8%). Intensive care unit length of stay was 46.0 (29.0–90.0) and hospital LOS was 51.0 (32.0–91.0). Over half of the patients (51.6%) were discharged to an acute rehabilitation facility, and 27.4% were discharged home (Table 6). There was similar survival to hospital discharge between the COVID and non-COVID awake ECMO patients (85% in both groups, p = 1.000). There was no difference in time from cannulation to extubation (5 [4–11] vs. 6 [3–11], p = 0.901); however, days requiring ECMO (p = 0.004), ICU (p = 0.016), and hospital LOS (p = 0.011) were significantly higher in the COVID group. Discharge disposition was similar to that reported in the total cohort (Table 6).

Discussion

To our knowledge, evidence describing outcomes in awake ECMO patients receiving VV ECMO is limited to case reports without survival outcomes; therefore, it is difficult to adequately compare the outcomes of this study to those reported in the literature.

The authors of this study attribute our survival rate to a number of factors. First, our patients were managed with the involvement of a dedicated and experienced multidisciplinary ECMO team consisting of surgeons, intensivists, nurses, perfusionists, pharmacists, physical and occupational therapists, respiratory therapists, and speech therapists. Second, patients were evaluated for ECMO and met criteria for cannulation if worsening respiratory failure was noted despite maximal medical therapy. Patients were considered for ECMO if they had AKI; however, patients were deemed ineligible if they were receiving RRT. Third, a majority of patients were cannulated with a ProtekDuo, which provides stable and reliable ECMO support with rare needs for cannula revision or repositions. Additional benefits of this unique configuration include right ventricular support, minimization of recirculation due to the distance between the two ports, the ability for patients to remain awake, and the option for patient mobility and rehabilitation while on support.¹⁶

Our cohort of patients included 34 patients with ARDS secondary to COVID-19. Our survival rate in this subset of patients is higher than those in previous reports, which concluded potentially withholding ECMO support in patients with COVID-19.¹⁷ Early data reported nearly half of the patients treated with ECMO died from septic shock and multiple organ failure with observed late complications of bleeding and infection.¹⁸ A more recent study by Schmidt et al.¹⁹ reported an estimated 31% probability of day-60 mortality for patients on ECMO. Data from the ELSO registry reported an in-hospital mortality 90 days after initiation of ECMO at 38%.²⁰ These rates are also comparable to that described in the EOLIA trial, which reported, at 60 days, 44 of 124 patients (35%) in the ECMO group had died.¹¹

Cessation of paralysis and weaning of continuous infusions of analgesics and sedatives were imperative in our approach to early extubation. At the time of cannulation, paralytics were discontinued, and analgesics and sedatives were weaned. During episodic periods of hypoxemia, bolus doses of analgesics and sedatives were administered rather than increasing doses of continuous infusions or adjusting ventilator settings. Dexmedetomidine, oral opioids, and antipsychotics were introduced to facilitate weaning of intravenous analgesics and sedatives and continued throughout extubation. These agents were gradually weaned in the maintenance phase of ECMO with lung recovery.

In our analysis, most reintubations were a result of uncontrolled nasopharyngeal bleeding, anxiety, airway protection, or the need for procedures. Once these factors were controlled and the patient was optimized, an attempt was made at extubation. Given the limited sample size, we were unable to identify specific factors that predicted the need for reintubation.

While tracheostomy to facilitate ECMO weaning was consistent with previous practices at our institution, our approach to ECMO has evolved to avoid tracheostomy. Indications for tracheostomy in our cohort included the inability to provide adequate support with ECMO alone and rehabilitate due to morbid obesity and progression of lung disease. The median time from cannulation to extubation supports this approach and facilitates a goal of early mobility. This differs from findings in studies by Boulos et al.²¹ and Swol et al.,²² which reported tracheostomy as a bridging option and favorable short-term outcomes associated with early compared to late tracheostomy for patients supported with VV ECMO for ARDS. Similar studies of COVID ECMO patients described early tracheostomy in most patients.^19,23 Liberation from mechanical ventilation and avoidance of tracheostomy in our cohort facilitated discharge home or an acute rehabilitation facility.

Complications seen in our cohort were consistent with those reported in the literature. Additionally, consistent with previous data, bleeding and infection were the most common complications noted in our patients receiving ECMO. A meta-analysis of 20 studies, encompassing 1,866 patients, reported ECMO complication rates of major or significant bleeding, 40.8% (26.8–56.6%) and significant infection, 30.4% (19.5–44.0%). Similarly, a meta-analysis of 12 studies encompassing 1,763 patients reported complications rates of bacterial pneumonia (33%) and any bleeding (33%).^24,25 Additionally, in recent data on ECMO use in COVID patients, bleeding and infection were the highest reported complications.¹⁹ Patients experienced bleeding episodes independent of anticoagulation. Patients also required circuit and oxygenator changes dependent on the presence of fibrin stranding, sepsis, and functionality of the circuit. Line pressure, changes in dP/dT, and trends in lactate dehydrogenase were also taken into consideration.

Our number of days receiving ECMO, ICU LOS, and hospital LOS provides insight into the labor and resources required to manage ECMO patients. A recent ELSO document outlined the need to consider ECMO on a case-by-case basis with consideration on overall patient load, staffing, and other resource constraints. Patient factors such as age, comorbidities, and organ failure should also be considered in the decision to cannulate.²⁶ While this document refers to COVID-19 patients, these elements should be considered in all ECMO candidates.

There are several limitations to this study. This was a retrospective review with data collection limited to chart review and outside hospital records. The sample size was limited with the lack of a control group and comparison to standard of practice. In addition, our study was not powered to detect a difference in outcome based upon cannulation strategy used and outcomes utilizing this strategy in patients with multiorgan failure.

Conclusions

This study highlights the impact of an awake ECMO approach on survival to hospital discharge. Future studies are needed to evaluate outcomes of this approach as compared to current practice to determine if this should become the standard.

Acknowledgments

Robert H. Bartlett, MD, Erica Bak, RN, and the ECMO team at Rush University Medical Center.