An Interprofessional Approach to Mobilizing Patients With COVID-19 Receiving Extracorporeal Membrane Oxygenation

This study was an 18-month preintervention and postintervention study to retrospectively compare 2 cohorts: adult patients with COVID-19 who were receiving VV-ECMO who were nonmobilized (May 2020 through December 2020; preintervention group) and adult patients with COVID-19 who were receiving VV-ECMO who were mobilized (January 2021 through November 2021; postintervention group). It was approved by our system’s institutional review board (IRB 014-179), which waived consent. In the preintervention group, patients were treated with sedation and NMBAs to keep them immobilized. In the postintervention group, patients were awakened and we executed an active rehabilitation care plan to mobilize them while they were receiving ECMO. We used an interprofessional approach to achieve this intervention (Table 1).

The setting of our study is a 137-bed, suburban, cardiac subspecialty hospital, with over 7 years of ECMO experience and approximately 350 ECMO cases managed to date. Historically, we have focused on patients with cardiac failure who require veno-arterial ECMO rather than those in respiratory failure requiring VV-ECMO. However, the pandemic rapidly shifted our practice from being a predominantly veno-arterial ECMO to a majority VV-ECMO center during the duration of this study, which spanned from May 2020 through November 2021.

All patients included in our study were those at our facility who required ECMO support, from May 2020 until November 2021, predominantly in respiratory failure, with a laboratory-confirmed COVID-19 infection. Patients with primarily cardiogenic shock were excluded. All participants were unvaccinated at the time of ECMO initiation.

All patients met established criteria for refractory respiratory failure requiring ECMO cannulation.^4,5  Because of infection control concerns, all patients were cannulated in the intensive care unit and then placed on standard lung-protective ventilation settings.²¹  All patients, except 1, were referred to us from diverse outside facilities; therefore, the patients’ initial COVID-19 therapeutics varied significantly. Each case was managed by an expert in infectious diseases and additional treatments during the patients’ course varied in accordance with the most current paradigm at the time.

To improve COVID-19 VV-ECMO survival outcomes, we implemented an interprofessional mobility strategy, with each service line contributing uniquely to this intervention (Table 1). Physicians revised isolation orders to be discontinued after 20 days from the onset of symptoms, and they performed early tracheostomies. Advanced practice providers adjusted intravenous (IV) infusion orders to promote the discontinuation of paralytics early and the modification of IV sedatives and paralytics. They also prescribed supplemental oral anxiolytics, antipsychotics, analgesics, and antitussives. Registered nurses and pharmacists optimized medicine administration within order parameters to wean patients off NMBAs and IV sedatives while supplementing treatment with an oral regimen to achieve an awake status. Respiratory therapists (RTs) managed ventilators to maintain rest settings but increased inspired oxygen delivery during rehabilitation sessions. Physical therapists and occupational therapists developed and fostered the progression of rehabilitation sessions from bed rest all the way up to full ambulation as the patient tolerated. And finally, the ECMO specialists (ESs) transiently increased ECMO settings during periods of anxiety and physical activity, while continuously monitoring the circuit’s integrity for complications during rehabilitation sessions.

Our sample included all adult patients with COVID-19 and receiving ECMO in our facility during the study period. The primary outcome was survival to discharge. Secondary outcomes included use, in days, of various sedatives, anxiolytics, and NMBAs; duration of ECMO support; and the need for cannula repositioning. The stratification of our 2 groups was based on our ECMO mobilization strategy (ie, nonmobile vs mobile).

Our initial nonmobile strategy included traditional lung-protective ventilation and complete paralysis with heavy sedation administration, strict bed rest, and no rehabilitation activities. In comparison, our interventions to achieve mobility in the second group included earlier insertion of tracheostomies and quicker removal of isolation restrictions, followed by an immediate weaning of NMBAs and sedatives. This was accomplished with a modification in continuous infusion administration, use of oral or transdermal medications, and an aggressive implementation of rehabilitation therapy sessions.

Patient variables included demographics, comorbidities, laboratory values, and measures of acuity. From the medical record, we extracted data on days of administration of IV paralytics, IV sedatives, oral anxiolytics, oral and transdermal analgesics, oral antipsychotics, oral β-blockers, IV-push β-blockers, IV-infused β-blockers, and antitussive regimens. We calculated survival to discharge, duration of support, and the number of occurrences that an echocardiographic-guided intervention was required for an ECMO cannula readjustment.

Continuous variables with a normal distribution were compared using the Student t test, and those with a nonnormal distribution were compared using the Kruskal-Wallis test. Categorical variables were assessed using the χ² test. For all statistical measures, a 2-tailed P < .05 was considered significant. Survival was compared using the χ² test. Throughout the article, normally distributed continuous variables are presented as mean values with standard deviations, and nonparametric variables are presented as medians with interquartile ranges (IQRs). Categorical variables are presented as whole numbers with percentages. Statistical analysis was performed using Stata/SE 17.0 (StataCorp).

Since the admission of our first patient with COVID-19 who required ECMO, in May 2020, until the completion of this study in November 2021, a total of 48 patients received ECMO support for respiratory failure secondary to a COVID-19 infection. Three patients remained on support at the end of this study, with indeterminant outcomes, and another 3 had inadequate medical records, so they were excluded from the analysis. Of the 42 patients included, 41 received VV- ECMO support, and 1 patient initially received veno-arterial ECMO but eventually was converted to VV-ECMO. All VV-ECMO configurations were cannulated as internal jugular vein, single-site, dual-lumen cannulas, which included the Avalon (Getinge) and Crescent (Medtronic), except in the case of 1 patient who was transported to our facility with bilateral, femoral venous cannulations; these were converted to a single-site cannulation after 15 days. For 2 patients, cannulas were changed during their course to a subclavian vein, single-site, dual-lumen Protek Duo (Liva-Nova) right ventricular assist device cannula with an oxygenator in the circuit. Fluoroscopy was only used for right ventricular assist device placement because of the complexity of the procedure. All other cannulations were performed with transesophageal echocardiography.

The majority of the patients were male (n = 32; 76.2%), the mean (SD) age was 47.2 (7.9) years, and a plurality was of Hispanic descent (n = 18; 42.9%). The majority were considered overweight (body mass index, mean [SD] 32.8 [5.4], calculated as kg/m²), had normal renal function (creatinine level, mean [SD] 0.79 [0.31] mg/dL), normal hepatic function (total bilirubin, mean [SD] 0.58 [0.23] mg/dL), and required medicine for cardiopulmonary support before cannulation in the form of inhaled epoprostenol (n = 33; 78.6%) and/or a vasopressor (n = 25; 59.5%), although there was a baseline higher amount of vasopressor needed for the mobilized group before ECMO initiation (n = 19 [73.1%]; P = .02) (Table 2). The patients’ overall survival to discharge was calculated to be 61.9%; the median duration of ECMO support was 23 (IQR, 13-56) days.

When stratified by mobility, 16 patients (38.1%) received traditional management and remained nonmobile, whereas 26 patients (61.9%) received an aggressive mobility therapy strategy. Analysis showed that the implementation of a mobility strategy resulted in statistically significant improved survival when compared with traditional management (73.1% vs 43.8%; P = .04). When the continuous IV-infusion regimens were analyzed, the secondary outcomes showed that the mobilized patients’ median days (IQR) of paralytics were reduced from 20 (10-26.5) to 3 (1-7) (P < .01). Administration of midazolam was reduced from a median of 22 (13-27) days to 6 (2-14) days (P < .01) and fentanyl from 20 (13-28) days to 5 (1-10) days (P < .01). The mobilized group of patients did show an increase in median days (IQR) of treatment with dexmedetomidine from 0 (0-0) to 26 (9-53) (P < .01), morphine from 0 (0-0) days to 2 (0-22) days (P < .01), and ketamine from 0 (0-0) days to 0 (0-7) days (P = .01), when compared with the nonmobilized group.

When compared, more patients in the mobilized cohort than the nonmobilized cohort had been administered oral anxiolytics (P < .01), oral or transdermal analgesics (P < .01), and oral antipsychotics (P < .01), and more required an antitussive regimen via oral forms (P < .01) and/or inhaled lidocaine nebulizers (P = .01). Surprisingly, there was no significant variance in the use of propofol (P = .66) or β-blockers (oral or IV push, P = .10; IV continuous infusion, P = .07) when data from both groups were analyzed.

Our findings demonstrate that mobilizing our patients was associated with a greater need for cannula repositioning (n = 16 times vs 1 time; P = .03). There was no substantial difference in the number of days receiving ECMO when the 2 groups were compared (P = .06) (Table 3).

After noting our outcome survival rates in caring for patients with COVID-19 who were receiving VV-ECMO were lower than those reported internationally, our facility chose to undergo a complete culture change and implement an unfamiliar strategy to improve outcomes. Instead of treating these patients with full sedation and NMBAs, we implemented an interprofessional approach to wake and mobilize them during their ECMO course. Comparing the 2 cohorts showed a significant improvement in the survival rate of the group receiving the early-mobility intervention.

Secondary analysis showed that our interprofessional intervention resulted in our mobilized cohort requiring less administration of NMBAs, midazolam, and fentanyl infusions but increased use of dexmedetomidine, morphine, and ketamine infusions. There was also a significant increase in administration of oral anxiolytics, analgesics, antipsychotics, and antitussives in the postintervention group. The differences in medication regimen between the 2 cohorts allowed our team to achieve a sufficiently awake status for patients with COVID-19 who were receiving ECMO so that they could participate in active rehabilitation during their ECMO course and, ultimately, increase their survival. Each discipline contributed uniquely with interventions to achieve these results. In the following section, we discuss the clinical practice implications for each service line.

A previous barrier to mobilization was the isolation restrictions that are applied to every patient with COVID-19. Physicians evaluated this protocol and revised their orders to best reflect current standards at that time. Early in the pandemic, 2 consecutive, negative COVID-19 polymerase chain reaction tests were required to move individuals out of isolation. This recommendation, however, lengthened isolation requirements, as evidenced by one of our initial patients with COVID-19 who was receiving ECMO and remained in full isolation for over 2 months, delaying our mobilization treatment attempts.

As more data became available, studies showed that a patient may still test positive on a polymerase chain reaction COVID-19 test but not truly be infectious, because most polymerase chain reaction tests cannot discern between a viable active virus and a non-infectious state.²²  Thus, physicians switched from mandating 2 negative COVID-19 tests to removing isolation precautions after 20 days of the onset of symptoms (for immunocompromised populations), congruent with the Centers for Disease Control and Prevention guidelines at the time of this intervention.²³  This change in isolation protocols allowed the nurse to be directly in the patient’s room at any given time and permitted the family to be present during the sedation-vacation trials, which may have reduced our patients’ anxiety.²⁴  Both of these modifications fostered an increase in the patients’ interactions with staff and family during the process of achieving a fully conscious status.

In addition, with the proper use of full personal protective equipment, if needed, physicians began performing early tracheostomies within 48 hours of ECMO cannulation, a treatment that had previously been discouraged for any isolated patient. The insertion of a tracheostomy made patients more comfortable while waking, provided a more secure advanced airway,²⁵  and gave them ventilatory protection while mobilizing.

Advanced practice providers were vital to helping mobilize patients. They assumed the responsibility of prescribing the appropriate medication regimen to encourage the safe weaning off paralytics. They also altered IV sedative or analgesic orders toward infusions that allowed a more awake status, and they added supplemental oral medications so the patient could eventually be transitioned off of IV forms of anxiolytics and analgesics.

The greatest change was executed by our intensive care unit nursing staff. Our naïveté about mobilizing patients receiving ECMO, in general, was intensified by adding the COVID-19–specific population to the dilemma. We battled the difficulty of treating anxiety and caring for the awake patient’s psychological needs, all while attempting to maintain stable hemodynamics.

The biggest challenge the intensive care unit nurses had to navigate was titrating medicines to achieve the patients’ alert state while also keeping them comfortable. Upon our retrospective review of secondary outcomes, we found the registered nurses had significantly reduced IV continuous infusion administration of paralytics, fentanyl, and midazolam. However, to help patients remain calm during this transition, the nurses had to increase their use of IV continuous infusions of dexmedetomidine, morphine, and ketamine. This change in opioid infusion administration from fentanyl to morphine was more consistent with literature about pharmacologic sequestration, showing that morphine is less absorbed by the ECMO circuit than fentanyl.^26,27  There was no noted difference in the nurses’ use of propofol for either cohort.

Continuous infusion titrations were not the only regimen altered to achieve an awake status; oral and transdermal medicines were manipulated as well. Patients in the postintervention group received significantly more oral or transdermal analgesics, oral anxiolytics, and oral antipsychotics. This goal of changing from IV infusions toward oral or transdermal maintenance to control pain and anxiety was evaluated daily by nurses to ensure they administered the optimal ordered regimen to achieve an awake status.

As expected, once paralytics were discontinued and sedation lifted, many patients displayed vigorous coughing propensities. To combat this, nurses had to administer significantly greater amounts of oral antitussive medicines and lidocaine nebulizer treatments to help quell the coughing paroxysms. The addition of such high amounts of extra medications administered as needed may have potentially increased the cost of patient care.

Once patients were awake, significant struggles we encountered were heart arrhythmias and hemodynamic challenges. Our experience reflected what the literature shows: that many patients with COVID-19 experienced tachycardia, especially while waking.¹⁷  We used a combination of ordered oral, IV push, or continuous IV infusions of β-blockers to combat the tachycardia; however, when compared with the nonmobilized group, we did not see a significant difference in β-blocker use. Nevertheless, battling the tachycardia and arrhythmias that affect patients with COVID-19 became a challenging aspect to conquer when waking them. This was difficult partially because both tachycardia and high cardiac-output states, such as anxiety, can contribute to hypoxia and decrease the effectiveness of VV-ECMO.^28,29 

Pharmacists’ interventions were vital during this study. They ensured there were minimal adverse drug interactions with the medication order changes, and they optimized the best regimen with minimal side effects to help achieve the most awake status. Education about sequestration of drugs by the ECMO circuit was also communicated by the pharmacists to our advanced practice providers when medication regimens were being developed.

The RT department played an important role in achieving patient mobilization. During each rehabilitation session, the RTs preemptively increased the ventilator fraction of inspired oxygen (Fio₂) to 1.0 to optimize the oxygen delivery during the highly consumptive state of physical activity. No change was made to the set tidal volumes or respiratory rates, because those changes in ventilation were compensated by the ES’s increase in the sweep (ie, the setting on ECMO that decreases carbon dioxide in the blood). These steps allowed for ventilator rest settings to be maintained and barotrauma to remain minimized. After each session, RTs weaned the ventilator’s Fio₂ slowly back to baseline, following standard COVID-19 ECMO guidelines of maintaining a ventilator’s Fio₂ at 0.50 or less.³⁰ 

One unique aspect noted about many of the patients with COVID-19 who were receiving ECMO was their predisposition for very low tidal volumes (often <100 mL). Lenience for low tidal volumes, tachypnea, and poor lung compliance were all needed to allow these patients to wake, and so the RTs would lower the tidal volume limit and increase the respiratory rate thresholds on the ventilator to prevent alarm fatigue. Serum lactate levels were checked frequently to determine the patient’s tolerance, and permissive tachypnea was allowed, as long as hemodynamic and cognitive stability were maintained.

The most inexperienced discipline relative to our ECMO population at the time was our rehabilitation personnel. Although we had been using ECMO for many years, the physical therapists and occupational therapists had never cared for our population of patients receiving ECMO, because we had not previously mobilized these patient; so change required the support of the registered nurses, ESs, and RTs to guide them. Collaboration of leadership between the Rehabilitation Services and the ECMO department helped distribute information on the fundamentals of how ECMO worked and the general safety concerns that accompany it. Basic reviews of the ECMO circuit settings and functions were offered by the ES for all the physical therapists and occupational therapists who were new to working with patients receiving ECMO.

Independent literature reviews^9,31  were also completed by our physical therapy and occupational therapy staff to help guide them through this new mobility strategy. Sessions were created with a variety of activities based on the patient’s alertness and ability to engage in therapy tasks. Examples of bedside activities included sensory motor tasks with basic auditory and visual tasks, upper and lower range-of-motion exercises, fine and gross motor coordination activities, and functional communication. Mobility then progressed to sitting at the edge of the bed, standing, marching in place, and, ultimately, ambulating. Early participation in activities of daily living included grooming and toileting training, resistance training, cycling using a stationary bike, and interacting with an electronic game device. Focusing on overall flexibility and upright posture became critical to advance mobility in these patients, because of their prolonged bed rest.

Last, although skilled in caring for patients receiving ECMO, the ESs were just as inexperienced as the rest of the staff in mobilizing awake patients receiving ECMO at the time of implementation. We consulted the most current international guidelines at the time for ECMO mobility³⁰  and also another facility in our health care system that did have experience with mobilizing their pretransplant VV-ECMO population. The ESs had to learn how to adjust the equipment into “transport mode,” which included the conversion of gas blenders to portable oxygen tanks, adaptation to battery mode on the pumps, and the detachment of the heater and cooler lines during mobile therapy sessions. Overall, the ES team had to maintain a new heightened awareness so the extracorporeal circuit would stay safe and intact during all physical activities.

Mobilizing patients receiving ECMO can increase the risk for circuit rupture, dislodgment, and occlusion with each rehabilitation session performed.⁷  There is also the potential for omitting vital steps when converting from rest to transport mode (or vice versa) on ECMO. Because of this risk, we strictly enforced a “High 5 Safety Checkoff” required to be performed by 2 ESs before and after each mobilization session. This checkoff involves 5 major safety equipment aspects to be double-checked by 2 ECMO-trained personnel, with a “high 5” hand gesture done once completed. This was to ensure that all equipment was converted safely from rest, to transport mode, and back to rest status (Table 4).

The ESs discovered that slightly liberalizing the sweep gas and the fraction of delivered oxygen setting on the ECMO circuit during rehabilitation sessions, but also during periods of transient high anxiety, promoted patients’ overall comfort and tolerance of activity. When awake or during moments of exertion, many patients with COVID-19 who were receiving ECMO complained of feeling dyspneic, so the ES would temporarily increase the sweep gas approximately 1 to 3 L higher than baseline to decrease blood carbon dioxide levels and also increase the fraction of delivered oxygen (the setting on ECMO that supplies oxygen to the blood) to 1.0 to increase oxygen delivery. The majority of the time, this action would ease the patient’s anxiety, and their noticeable dyspnea would resolve within minutes. Once the anxiety settled or after the physical activity was complete, the ES would slowly return the settings to baseline. Care was always taken not to overshoot, because a substantial decrease in serum carbon dioxide levels has been associated with neurologic complications.^32,33 

Finally, the ESs needed to be vigilant for the possibility of cannula malposition and the tendency toward increased recirculation on mobilized patients. Recirculation occurs when the VV-ECMO circuit siphons the delivery of the oxygenated blood (that has already been processed by the ECMO circuit) back up into the circuit itself, so that fraction of blood does not get delivered to the patient.^27,28  Recirculation renders the percentage that is recirculated essentially ineffective, and this decrease in effective flow can lead to hypoxia or underperfused states. One of the known major contributors to recirculation is cannula malposition.^27,28,34  Our facility used single-site, right internal jugular, dual-lumen cannulas on 41 of the patients (98%) included in this study, and we noted a much higher number of position readjustments were needed for those who were mobilized. We found an echocardiographic intervention to optimize the cannula position from recirculating was needed 16 times in the mobile group versus only once in the nonmobilized group.

Benefits of mobilization may include increased survival, increased engagement of staff, and the potential for increased patient and family satisfaction, because of the increased personal interactions while waking. One other inadvertent benefit noted was that the intensive care unit nurses were able to remove the urinary catheter from 4 of our awake patients during their ECMO runs, which minimized the risk of catheter-associated urinary tract infections.³⁵ 

Although literature shows that mobilizing patients receiving ECMO is safe,³⁶  there are inherent risks. The risks of implementing a mobilization strategy include the possibility of invasive line removal or dislodgment, the need for additional personnel to perform the rehabilitation services, an advanced safety-training requirement of the staff, the hindrance that rehabilitation therapies can be time consuming, and, ultimately, that mobility can affect vital sign stability and increase the risk of danger to the ECMO circuit integrity.

Significant limitations of this study include the small sample size as well as the fact that it was a retrospective review. The patients were not randomly assigned within this comparison study. We also lost 3 patients because of incomplete data documentation. Selective bias may have occurred, and we recommend prospective randomized trials be conducted to confirm our findings. Although no significant demographic differences were found between the 2 arms of the study, there was a greater amount of vasopressor use in the mobilized cohort before they underwent the intervention, suggesting the mobilized group potentially had a higher severity of illness at baseline. This inexplicable result could be an overt limitation of our study, because of its small sample size.

Data on the amount or distance of mobility, and any ECMO settings, such as flow, were not analyzed in this study. Throughout the study period, various COVID-19–specific therapies were implemented at referring facilities and our own facility, depending on the contemporary scientific paradigm. We do not believe these therapies made any significant impact on the ability to mobilize patients or on patient survival; however, we do not have sufficient data to assess the impact of these therapies, and this lack remains a potential limitation.

We did not test each patient to identify their specific COVID-19 variant, so we do not know with certainty if there was any impact of the variants in this study. However, in a recent multicenter study, researchers found no difference in survival of patients with COVID-19 who were receiving ECMO when a timeline of prevalent variants in the community was compared with survival outcomes.³⁷  And finally, although this was retrospective study of a small cohort, overall, we believe our findings are substantive and should lead to more complex, prospective studies.

Six months into the pandemic, our facility was not meeting international standards of survival among our population of patients with COVID-19 who were receiving ECMO support. Implementing an interprofessional approach of progressive mobility improved these outcomes. Changing the ECMO treatment paradigm with more rapid removal of isolation protocols, performing tracheostomies within 48 hours of ECMO initiation, and the faster weaning off paralytics and sedatives all allowed aggressive application of rehabilitation therapies.

Analysis of administration of continuously infused medication showed that the mobilized group had significantly fewer days of treatment with paralytics, fentanyl, and midazolam, but more days of dexmedetomidine, morphine, and ketamine. Concomitantly, these patients received more oral or transdermal analgesics, anxiolytics, and antipsychotics. They also had a greater use of both antitussives and lidocaine nebulizer treatments, and this group was associated with more single-site VV-ECMO cannula readjustments.

Ultimately, we found a significant increase in survival when we analyzed the preintervention and postintervention groups. This finding showed that an implementation of a progressive mobility strategy for the population of patients with COVID-19 receiving ECMO support was feasible and beneficial when using an interprofessional plan of care.

The authors thank Karen Stanzo, PhD, RN, IBCLC, NEA-BC, for manuscript review, and they thank the interprofessional critical care staff at Baylor Scott & White The Heart Hospital Plano.