Mobility Levels With Physical Rehabilitation Delivered During and After Extracorporeal Membrane Oxygenation: A Marker of Illness Severity or an Indication of Recovery?

Abstract

Objective

The aims of this study were to determine whether physical rehabilitation intervention for individuals who required extracorporeal membrane oxygenation (ECMO) is associated with clinical outcomes and to assess whether the patient mobility response over initial rehabilitation sessions early in the intensive care unit (ICU) course predicts or is associated with survival, lengths of stay, discharge disposition, and 30-day readmissions.

Methods

This study was a 10-year retrospective practice analysis of adults who were critically ill and required ECMO for >72 hours in the cardiothoracic ICU at an academic medical center. Physical rehabilitation implemented during or following the initiation of ECMO was quantified on the basis of timing, frequency, and change in mobility level in response to the intervention over the first 4 consecutive sessions. The primary dependent outcome was in-hospital mortality. Secondary outcomes included 30-day readmission and discharge disposition ranked on an ordinal scale.

Results

Three hundred fifteen individuals (mean age = 50 years [SD = 15 years]; 63% men; mean Sequential Organ Failure Assessment score = 11.6 [SD = 3.3]) met the inclusion criteria. Two hundred eighteen individuals (69%) received at least 1 physical rehabilitation session while requiring ECMO, 70 (22%) received rehabilitation after ECMO was discontinued, and 27 (9%) never received rehabilitation. Individuals discharged alive achieved higher mobility levels and had a steeper, more positive rate of change in mobility over the first 4 sessions than individuals who died in the hospital (2.8 vs 0.38; degrees of freedom = 199, t = 8.24). Those who received rehabilitation and achieved the milestones of sitting on the edge of the bed and walking for >45 m were more likely to survive (47% vs 13%; χ2 = 156) than those who did not (26% vs 3.5%; χ2 = 80).

Conclusion

A positive rate of change in mobility and the ability to achieve mobility milestones with rehabilitation were associated with improved clinical outcomes.

Impact

An individual’s mobility response to physical rehabilitation early in the ICU course is an important indicator of illness and should be used with clinical presentation to guide clinical decision-making and predict outcomes.

Introduction

Physical rehabilitation in the intensive care unit (ICU) is safe and feasible for a diverse range of critically ill individuals, including those requiring mechanical ventilation,^1,2 continuous renal replacement therapies,³ and extracorporeal membrane oxygenation (ECMO).^4,5 Although physical rehabilitation is purported to have beneficial effects on patient-centered outcomes such as improved muscle strength, quality of life, and greater probability of walking without assistance at hospital discharge in the general ICU population, the receipt of physical rehabilitation on the whole has not demonstrated consistent effects on clinical outcomes such as reducing length of stays (LOS), improving survival, and reducing readmission rates.^6–8 Overall, there is limited information on the association of physical rehabilitation with outcomes specifically for those who have required ECMO.^6,9

Physical rehabilitation in the ICU is a complex process. The provision of rehabilitation by physical and occupational therapists requires significant preparation, coordination, and communication with the interdisciplinary ICU team.^10,11 In addition, rehabilitation intervention needs to be individualized to optimize the physiological and functional response of the patient, which can fluctuate in the midst of critical illness. As such, implementation necessitates prescription of intervention with consideration for parameters, including the modality or type, dosage (eg, duration and frequency), and intensity,¹² as well as adapting parameters to changes in clinical variables such as hemodynamic instability.^13–15 Individuals may respond and progress differently to rehabilitation intervention based on the parameters chosen as well as potentially modifiable factors such as sedation reduction or sedation holidays⁹ and nonmodifiable factors such as age and prehospitalization physical functional status. Therefore, merely reporting whether physical rehabilitation was received as a binomial parameter, as is frequently done in critical care rehabilitation research including randomized controlled trials, may limit understanding of the efficacy.

In recent years, ECMO utilization for individuals with critical conditions such as acute respiratory distress syndrome, pulmonary embolism, cardiogenic shock, and cardiac arrest have increased.^16–18 Given the intense level of care, expertise, and resources required, there is an imminent need to optimize outcomes for this patient population. Patient case reports, case series, and observational studies provide preliminary evidence that physical rehabilitation-focused interventions delivered to individuals on ECMO may be safe, feasible, and potentially associated with improved outcomes.^4,19–24 The specific parameters that yield those benefits, however, are unknown.

The 2 aims of this study were to determine whether physical rehabilitation intervention for individuals who required ECMO is associated with clinical outcomes and to assess whether the patient mobility response over initial rehabilitation sessions early in the ICU course predicts or is associated with survival, lengths of stay, discharge disposition, and 30-day readmissions.

Methods

Ethical Considerations

The study was approved by the internal review board at the University of Kentucky with informed consent waived because of the retrospective study design.

Study Design

The study is a 10-year retrospective analysis of the University of Kentucky Healthcare’s electronic medical database for adult (>18 years old) individuals requiring ECMO for any indication with any modality (ie, type described below in ECMO parameters). Individuals were included in the analysis if they were admitted to the University of Kentucky from January 1, 2010, to December 31, 2019, and required ECMO for at least 72 hours. The 72-hour time period was selected a priori to exclude individuals who would not receive rehabilitation during a short ECMO duration. Pediatric patients, pregnant individuals, and prisoners as well as individuals requiring ECMO for <72 hours were excluded from the analysis.

Outcomes

The primary dependent outcome of interest was in-hospital mortality. Secondary outcomes included hospital LOS, discharge destination, and 30-day readmission rates. Discharge destination was ranked on an ordinal scale: 0 = mortality; 1 = discharge to inpatient hospice; 2 = discharge to long-term acute care facility equivalent to ventilator weaning and skilled nursing facilities; 3 = discharge to acute and subacute care rehabilitation facilities; 4 = discharge to home with home health rehabilitation services; and 5 = discharge to home with no services. Because of inconsistencies in the specific assessments and reporting over the study period, standardized muscle or physical functional outcome measures at ICU or hospital discharge were not used.

Institutional Standards for Physical Rehabilitation

Physical rehabilitation at our medical center includes the provision of physical and occupational therapist services to individuals during and following ECMO in the cardiothoracic ICU.²⁵ Intervention is delivered by physical and/or occupational therapists with nursing and respiratory clinicians assisting as needed per individual cases. Therapists receive at a minimum of 4 weeks of on-job training from a senior thnerapist on intervention delivery for mechanical circulatory support systems emphasizing patient safety, hemodynamic monitoring, presession screening via chart review, and triage with bedside nurse as well as rounding with the teams. In 2015, the inpatient rehabilitation department developed and implemented formalized annual competencies with training documents and manual skills assessment for mechanical circulatory support and ECMO (Suppl. File). Institutional internal guidelines state the goal is to provide rehabilitation sessions at a frequency of 4 to 7 times per week with the objective that individuals on ECMO progress towards out of bed mobilization if physiologically and physically applicable. Prior to 2016, physical rehabilitation did not commence until a consult was ordered by the attending physician. After 2016, an automated consult order was integrated into the electronic health record for every individual on ECMO, triggering an initial physical or occupational therapy chart review to determine level of appropriateness for intervention. Prior to 2015, patient appropriateness included a medical chart review, bedside triage with nursing, or rounding with medical team to examine hemodynamic stability defined as extubated state, mean arterial pressure (MAP) >60, oxygen saturation >85%, no heart rhythm in need of cardioversion, no code within past 24 hours, no increase in oxygen requirement in past 2 hours, or other specific parameters as determined by the team. After 2015, appropriateness for physical rehabilitation and mobility participation were based on guidelines following the Richmond Agitation-Sedation Scale (RASS) score (−2 to +2) and safety criteria as previously published in Hodgson et al.¹ In 2018, a formalized hemodynamic screening tool and ECMO mobility screening tool were developed (previously published)²⁵ based on Wells et al.⁴ Femoral cannulation is not a formal contraindication to physical rehabilitation at our medical center, but individual physicians and recommendations from Hodgson et al¹ may have restricted rehabilitation to only in bed activities.

Physical Rehabilitation Intervention Variables

Parameters were assessed as follows: received physical therapy (yes, no), received occupational therapy (yes, no); time to first rehabilitation session (hours from ECMO initiation to participation in the initial session), frequency of rehabilitation during ECMO (number of sessions completed per ECMO days as ratio), and frequency of rehabilitation during the entire hospital admission. Finally, mobility milestones were examined to understand if individuals achieving higher levels of mobility during their hospitalization associates with improved patient outcomes. Mobility milestones achieved during rehabilitation were examined as the time from ECMO initiation to the time achieving first sitting on the edge of the bed; first participating in standing; first walking for >1.5 m; and first walking for >45 m.

Response to Early Rehabilitation

The first 4 consecutive rehabilitation (physical or occupational therapy) sessions were analyzed for highest level of mobility achieved using the ICU Mobility Scale²⁶ (IMS). “Early” was defined a priori as the first 4 rehabilitation sessions following and within 10 days of ECMO initiation. Physical rehabilitation implemented prior to ECMO was excluded. The initial rehabilitation session was defined as the first session after the initiation of ECMO. The highest level of mobility achieved per session was examined to represent the patient functional response. Change in mobility from session 1 to session 4 was examined to represent the trajectory in response to early rehabilitation.

Demographic and Clinical Variables

Demographics extracted from the electronic health record included age, sex, body mass index, and race. Comorbid burden was quantified using the Charlson Comorbidity Index. Clinical data from the ICU and hospital stays extracted from the electronic health records included admitting diagnosis; ICU and hospital LOS (days); ECMO parameters (described below); receipt and duration of mechanical ventilation; receipt and duration of continuous renal replacement therapy; tracheostomy (yes, no); major surgery (yes, no), including organ transplantation, ventricular assist device implantation, sternotomy, and thoracotomy; receipt and time to initiation of enteral nutrition; receipt of continuous intravenous infusion of a neuromuscular blocker (cisatracurium and rocuronium); receipt of corticosteroid; receipt of vasopressor or inotrope; RASS score, a surrogate marker of sedative state, assessed from daily documented nursing notes and quantified by the worse score in the first 72 hours of ECMO initiation and the average during the ICU stay; and severity of illness, measured with the Sequential Organ Failure Assessment (SOFA) score, as the peak in the first 48 hours of admission to the cardiothoracic ICU.

ECMO parameters of interest included indication for ECMO; time from hospital admission to ECMO initiation (days); total duration (hours); modality (type), including venoarterial (VA), venovenous (VV) single lumen, VV double lumen, and hybrid (VA-VV); and cannulation location categorized as femoral vessel (artery or vein), internal jugular vein, aorta, subclavian artery, and atria. The most mobility-limiting access site was used to determine the category for individuals receiving 2 different cannulation locations (eg, femoral artery and internal jugular vein cannulation was categorized as femoral because of the mobility limitations imposed by the femoral artery cannulation site).

Data Extraction

Demographic, rehabilitation, and clinical data were extracted from the electronic health record by a computer science engineer with experience in medical database and data mining. To ensure accuracy of the extracted data, 5% of the sample was examined manually by an independent researcher. Before 2018, the IMS was not routinely reported. Thus, the IMS for the first 4 rehabilitation sessions was retrospectively assigned by a team of physical therapy students led by an expert. If multiple rehabilitation sessions (cotreatments or individual sessions) occurred on the same day, students were trained to record the highest IMS score achieved per day. The physical therapy students received 1-on-1 training and underwent a trial period scoring the 4 sessions on the same 10 individuals compared with the expert. The students were not informed of the design and focus of this study.

Statistical Analysis

Data were assessed using descriptive statistics including mean and SD or median and interquartile range, histograms, and Shapiro–Wilk test for normality. A 1-way analysis of variance was also performed to examine differences in independent and dependent variables based on the 4 ECMO modality types previously mentioned because of the known differences in severity of illness, physiology, and application with ECMO delivery. One-way analysis of variance was performed to assess different in-patient demographics, clinical variables, and participant outcomes for 3 groups of participants: receipt of rehabilitation treatment during ECMO; receipt of rehabilitation treatment after ECMO; and did not receive rehabilitation interventions at any time during admission. One-way analysis of variance was also utilized to understand if differences in rehabilitation delivery and patient outcomes existed based differences in practice per year of admission per standardized reporting above (2010–2014; 2015–2016; and 2017–2019). Independent t test and chi-square test were performed to assess whether the receipt of physical rehabilitation on ECMO improved outcomes compared with the group of individuals receiving rehabilitation after ECMO. Chi-square tests were performed to assess whether the achievement of mobility milestones was related to survival. Multivariate logistic regression was performed as an exploratory analysis to predict in-hospital mortality with independent variables defined a priori per the authors with support of a statistician to develop a directed acyclic graph (Suppl. Fig. 1). Individuals participating in at least 4 rehabilitation sessions were considered in the model predicting mortality. Data were analyzed and visualized using GraphPad Prism 8.2 (GraphPad Software, San Diego, CA, USA) and SAS (SAS Institute Inc, Cary, NC, USA).

Role of the Funding Source

The funder played no role in the design, conduct, or reporting of this study.

Results

A total of 546 individuals required ECMO from 2010 to 2019; 231 individuals were excluded from analysis (229 were on ECMO for <72 hours, 1 was a prisoner, and 1 was a pregnant individual). A total of 315 individuals with a mean age of 50 years (SD = 15 years), 63% of whom were men and predominantly White (89%) and had a mean Charlson Comorbidity Index of 3.3 (SD = 2.4), met the inclusion criteria (Tab. 1). On admission to the cardiothoracic ICU, patients had a mean SOFA score of 11.6 (SD = 3.5), and the average length of ICU stay was 29.3 (SD = 30.6) days. The primary indications for ECMO were acute respiratory failure and/or acute respiratory distress syndrome (n = 123; 39%), cardiogenic shock (n = 63; 20%), and pulmonary embolism (n = 23; 7%). Individuals required a mean of 12.8 (SD = 14.9) days on ECMO, with a maximum duration of 142 days (median = 7 days [interquartile range = 5 to 16]). The most common ECMO modality was VA, at 49%, followed by VV and VV double lumen, at 23% each, and hybrid at 6% (Tab. 1). Individuals receiving the VA ECMO modality were more likely to be men and older and to have a higher preexisting comorbidity burden than individuals receiving the other ECMO modalities but also had the shortest ECMO duration. Patients requiring the VA and hybrid (VA-VV) modalities had higher in-hospital mortality (Tab. 3). Demographic and clinical variables stratified by the 4 ECMO modalities are shown in Table 3.

Physical Rehabilitation Intervention

Two hundred eighteen individuals (69%) received at least 1 physical rehabilitation session while on ECMO. Seventy individuals (22.2%) received physical rehabilitation after ECMO was discontinued, and 27 individuals (8.6%) never received intervention during their hospitalization. Individuals who never received physical rehabilitation intervention had statistically higher SOFA scores, greater sedation levels (as defined by RASS), the VA modality, a shorter duration on ECMO, and a higher mortality (Tab. 1). Individuals who received physical rehabilitation while on ECMO were not more likely to survive than patients receiving rehabilitation therapy after ECMO treatment (52% vs 63%; χ² = 2.37; P = .161). Survivors who participated in rehabilitation during ECMO and those engaging after ECMO did not differ in 30-day hospital readmission rates (14% vs 12%; χ² = 0.04; P = .952) (Tab. 1).

For the 288 patients receiving physical rehabilitation, the mean time from ECMO initiation to the first session conducted by a physical therapist was 4.7 (SD = 4.8) days; for the first session conducted by an occupational therapist, the mean time was 6.2 (SD = 5.1) days. Patients participated on average in 7.4 (SD = 12.7) physical therapist–led and 2.2 (SD = 5.5) occupational therapist–led sessions while on ECMO. The combined mean frequency of physical therapist– and occupational therapist–led sessions during ECMO was 0.65, equating to 4.5 sessions of physical rehabilitation per week.

Patients receiving the VV double-lumen modality had the highest frequency of physical rehabilitation sessions (Tab. 2). ECMO modality influenced time to participation and the mobility status during rehabilitation (Tab. 2). In addition, patients admitted from 2017 to 2019 were more likely to receive physical rehabilitation, had a shorter wait from ECMO initiation to the first rehabilitation session, but had a lower frequency of physical rehabilitation sessions while requiring ECMO (Suppl. Tab. 1). There were no differences in patient outcomes when individuals were stratified into 3 groups on the basis of year of admission (Suppl. Tab. 1).

Mobility Milestones

A total of 188 individuals achieved the milestone of sitting on the edge of the bed during physical rehabilitation (IMS score = 3), with a mean time to the event of 13.2 (SD = 11.7) days after the ECMO start date (Tab. 3). Individuals who achieved sitting on the edge of the bed were more likely to survive than those who never achieved the milestone (47% vs 13%; χ² = 156; P < .0001). These trends remained statistically significant for individuals who achieved standing (44% vs 9.5%; χ² = 153; P < .0001), walking for >1.5 m (35% vs 6%; χ² = 109; P < .0001), and walking for >45 m (26% vs 3.5%; χ² = 80; P < .0001).

Response to Physical Rehabilitation

A total of 2333 physical rehabilitation sessions occurred for 216 individuals while on ECMO, with 597 being classified as early physical rehabilitation sessions. The early physical rehabilitation interventions were delivered as passive activity in bed for 38% of the early sessions (IMS score = 0), in-bed exercises for 23% (IMS score = 1), sitting and standing for 30% (IMS scores = 2 – 6), and walking for 9% (IMS score ≥7).

The average mobility level achieved on the first physical rehabilitation visit had an IMS score of 1.3 (SD = 1.8); the score increased, on average, to 2.0 (SD = 2.2) by the fourth session, which occurred, on average, within 6.9 (SD = 3.2) days of ECMO initiation (Tab. 3). Individuals discharged alive achieved higher mobility levels and had a steeper, more positive rate of change in mobility in the first 4 sessions than patients who died in the hospital (2.8 vs 0.38; t₁₉₉ = 8.24; P < .001) (Fig. 1A). Patients who were discharged to home had the highest rate of change in mobility (3.5 [SD = 2.3]) compared with patients discharged to acute care rehabilitation (2.3 [SD = 1.5]) and long-term care/nursing home (1.8 [SD = 1.5]; P < .001; Fig. 1B). Individuals with a higher initial severity of illness achieved lower mobility scores with less change over the first 4 sessions (χ² = 27.5; P < .0001) (Fig. 2A). Sedative status measured by mean RASS scores during the first 72 hours of ICU admission were significantly related to mobility status in the first 4 sessions (Fig. 2B).

Prediction Model

Less positive change in mobility from rehabilitation session 1 to rehabilitation session 4, not reaching the mobility milestones of sitting on the edge of the bed and walking >45 m, and having a higher Charlson Comorbidity Index were significantly associated with mortality (Suppl. Tab. 2).

Discussion

The findings of this study demonstrate that the ability to engage in physical rehabilitation and a positive response to early intervention are associated with survival for patients on ECMO regardless of modality. The data in this retrospective analysis reveal that positive change in mobility during the first 4 physical rehabilitation treatment sessions may be associated with survival and discharge to home. For example, individuals who were discharged to home were more likely to achieve standing by the fourth rehabilitation session. The findings of this study, however, should be interpreted with caution because of the heterogeneity of patient populations including differences noted when stratified by ECMO modality and admission year. The approach to assessing mobility levels and mobility milestones in rehabilitation sessions as indicators of functional recovery for patients who require ECMO is novel and signifies the importance of patient response during rehabilitation beyond merely the receipt of rehabilitation.

Early rehabilitation is purported to improve patient outcomes for diverse populations of critically ill patients.^6,9,27–29 ECMO therapies are life-sustaining for patients with high severity of illness typically in acute organ failure or end-stage lung or heart failure.³⁰ ECMO may have advantages over traditional invasive mechanical ventilation in the ability to wean sedatives and paralytics. “Awake” or “ambulatory” ECMO are terms describing the minimal sedative state that promotes the engagement in mobility and rehabilitation.³¹ In a recent systematic review by Ferreira et al, multiple studies have correlated rehabilitation during ECMO with improved outcomes; however, most of the included studies had low methodological quality with underlying risk of biases.^32,33 Data from our study support the observation that physical rehabilitation is associated with patient outcomes. As previously observed, individuals with single-site access ambulate more easily; however, our data confirm that a positive response to early rehabilitation is association with improved outcomes in patients supported on ECMO, regardless of modality and cannulation strategy.

Rehabilitation parameters and the patient response during the rehabilitation session provide objective data on the individual’s evolving medical state. In the general ICU population, the minimally important difference of the IMS ranges from 0.89 to 3.0,²⁶ and positive changes in mobility are associated with survival to 90 days.³⁴ Therefore, the response with rehabilitation is a vital indicator of severity of illness and may further enhance the ability to predict outcomes. Clinically, it is purported that physical rehabilitation may play a role in healing and recovery by addressing respiratory muscle weakness³⁵ and preventing the negative effects of bed-rest.³⁶ Physical rehabilitation in individuals with acute respiratory failure has been previously shown in randomized controlled trials to improve outcomes.⁹ The body of literature in multiple randomized controlled trials of physical rehabilitation in the ICU, however, has not demonstrated significant patient benefit.^7,8,37,38 Moreover, there has not been a physical rehabilitation randomized controlled trial assessing efficacy specifically in individuals on ECMO. A previous pilot randomized controlled trial demonstrated that individuals in an intensive rehabilitation group can achieve standing at the side of the hospital bed 15 days earlier than the standard of care.³⁹ The authors suggest that ICU rehabilitation for individuals requiring ECMO may be too conservative because delivery of interventions did not lead to differences in physiological variables between the 2 groups and safety events were very low.⁴⁰

Based on our data, achievement of mobility milestones may be an excellent surrogate or adjunct marker for levels of illness, indicators of recovery, and potentially predict patient outcomes. The individuals with lower initial mobility, combined with a poor rate of change in mobility, have a higher risk of in-hospital mortality that should be used in concert with the clinical presentation to inform the family and make medical decisions. We demonstrate that ECMO patients who achieved sitting on the edge of the bed, standing, and walking were more likely to survive. Perhaps these individuals are not as critically ill and thus would have favorable outcomes regardless of achievement of milestones. Survivorship bias may influence the strength of these arguments, because patients alive at discharge may have had better outcomes regardless of the rehabilitation intervention.

Individuals in deep sedation early in the ICU, measured with a surrogate marker in this study (RASS), had lower mobility, less positive change in mobility, and higher likelihood of death. Sedative use is an independent predictor or outcomes and certainly highly related to severity of illness.^41–43 Moreover, eliminating use of sedatives,⁴⁴ sedative interruption,^45,46 and sedation holidays specifically to increase physical rehabilitation⁹ leads to improved patient outcomes. Thus, we suggest a delicate interplay between severity of illness, sedation, and rehabilitation. Patients with the highest severity of illness may require deep sedation and thus have a high likelihood of death regardless of implementation of rehabilitation. Sedation is frequently recognized as a barrier influencing rehabilitation in the ICU.^47–49 Patient eligibility and prioritization for physical rehabilitation is a debated topic that is frequently influenced by patient disease, hemodynamic stability, and staffing availability in the ICU.^50–52 The findings from our study indicate that rehabilitation participation and level of mobility achieved are important barometers of illness that should be considered in the plan of care. Likewise, team discussions regarding the capacity for sedation holidays, patient actual or anticipated tolerance, and response to active mobility and rehabilitation are necessary.

The retrospective design of the study prevents the ability to demonstrate causality. Our findings are consistent with a previous analysis of 107 individuals on ECMO actively sitting on the edge of the bed as defined by the IMS score (≥3), which was associated with a decrease in mortality.³³ The findings inform that if an individual is not sitting upright at the edge of the bed in the first few rehabilitation sessions, then the patient is at risk of poor outcomes. The data support the notion that phenotypes of patients may exist based on response to rehabilitation: responders, both quick response and delayed response, and nonresponders. Phenotypes may provide an avenue for individualized medicine. For instance, individuals with the nonresponding phenotype may be a lower or higher priority for rehabilitation and elicit ethical questions as follows:

• Do all critically ill individuals who require ECMO need rehabilitation intervention?

• Should individuals with high probability of death and low potential to respond to intervention be deprioritized to focus resources on individuals with ability to respond?

• And, if we had a better understanding of intervention parameters (intensity, frequency, and timing) as well as addressed known contributors (sedation), would the concept of nonresponders be obsolete?

This study has limitations consistent with a retrospective review of a large administrative database. Retrospective studies are susceptible to selection and representation biases. This study occurred over a 10-year period, which would have included changes in medical management, technology, and care providers as well as protocols for rehabilitation treatment. The objectives did not include changes related to year of admission. We did stratify data based on important changes in rehabilitation practice, including a different screening approach before 2014 and the implementation of automatic rehabilitation orders after 2016, which may confound the findings. We were unable to control for potential practice changes related to different medical approaches and potential difference in nursing-driven mobility. We were unable to control for nurse-led mobility sessions because of infrequent reporting prior to 2016. Moreover, the definition of early as the first 4 rehabilitation sessions is relative. The rehabilitation and active mobilization literature defines early with significant heterogeneity,⁵³ including studies implementing within 48 hours of admission to ICU,⁹ and other studies demonstrating time to interventions of 8 days⁸ and even 14 days.⁵⁴ The timing of physical rehabilitation is a crucial parameter that may influence outcomes. The mean time to physical therapist–led session was 4.5 days after ECMO initiation; this value did decrease in the patients on ECMO after 2016 to a mean of 2.2 days. Thus, it is reasonable to suggest that individuals in more recent years may have had advantages of improvements in critical care management, particularly faster time to the first rehabilitation session.

Similarly, the methodological approach to assessing rehabilitation at the onset and following ECMO utilization may undervalue any rehabilitation that occurred prior to ECMO initiation. We specifically excluded rehabilitation prior to ECMO use with emphasis placed on early rehabilitation defined as the first 4 sessions once ECMO was initiated. The retrospective design combined with the clinical trajectories of critically ill patients, 48% of survivors discharged to long-term acute care facility or acute care rehabilitation facility with potential to remain institutionalized at the 30-day time point, may underrepresent readmission rates in this population. There were no differences in 30-day hospital readmissions between the group of individuals receiving rehabilitation during ECMO and the group receiving rehabilitation after ECMO. This should be explored further and raises potential questions about the optimal timing of rehabilitation (during ECMO or after ECMO). Individuals who received rehabilitation after ECMO had significantly lower ECMO durations; this finding may suggest a reduced time window for implementing rehabilitation while ECMO is required. Furthermore, individuals receiving rehabilitation after ECMO initiation had a higher initial SOFA score; this finding clinically may suggest a highly acute illness with a rapid response and quick discontinuation of ECMO.

The findings of this retrospective analysis of 315 critically ill individuals requiring ECMO demonstrates that the patient functional response during physical rehabilitation is an important indicator of illness and potential recovery. Additionally, the data support that a positive response to early rehabilitation is associated with improved outcomes. Prospective, randomized, controlled clinical trials to determine if rehabilitation can assist in healing and recovery of critically ill patients on ECMO are urgently needed; only then can the effectiveness be examined. Finally, future clinical investigations should focus on addressing the delivery of rehabilitation for individuals with critical illness (ie, how clinicians should prioritize implementation and, subsequently, what the optimal dose and intensity of interventions are).

Author Contributions

Concept/idea/research design: K.P. Mayer, A.M. Pastva, B. Maher, P.E. Morris, J.B. Zwischenberger

Writing: K.P. Mayer, A.M. Pastva, S.P. Hatchett, A.N. Henning, P.E. Morris, J.B. Zwischenberger

Data collection: K.P. Mayer, S.P. Hatchett, M. Chang

Data analysis: K.P. Mayer, G. Du, S.P. Hatchett, M. Chang, P.E. Morris, J.B. Zwischenberger

Project management: K.P. Mayer, J.B. Zwischenberger

Fund procurement: K.P. Mayer

Providing participants: B. Maher, P.E. Morris

Providing facilities/equipment: P.E. Morris

Providing institutional liaisons: P.E. Morris

Clerical/secretarial support: P.E. Morris

Consultation (including review of manuscript before submitting): A.M. Pastva, A.N. Henning, B. Maher, P.E. Morris, J.B. Zwischenberger

K.P. Mayer, P.E. Morris, and J.B. Zwischenberger contributed to all phases of the research study and manuscript writing. G. Du, S.P. Hatchett, and M. Chang performed and assisted with data management and data analyses. A.M. Pastva, A.N. Henning, B. Maher, and J.B. Zwischenberger provided scientific and clinical oversight and assisted with manuscript writing. All authors read and approved the final manuscript.

Acknowledgments

The authors acknowledge Thomas A. Tribble, BA, CST; Julia Jones-Akhtarekhavari, BSN, RN; and Erin L. Abner, PhD, as well as physical therapist students Chandi R. Wilson, Davoud A. Saghaian, Morgan W. Garrett, and Zachary L. Culver for their assistance with the study.

Funding

This project was supported by APTA Acute Care, an Academy of the American Physical Therapy Association. The content is solely the responsibility of the authors and does not necessarily represent the official views of APTA Acute Care.

Ethics Approval

This study was approved by the Institutional Review Board at the University of Kentucky with informed consent waiver due to the retrospective design and analysis of deidentified data.

Data Availability

The minimal data are included in this published article. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

References

Author notes