The impact of early rehabilitation in intensive care unit for COVID-19 patients

Abstract Purpose To evaluate the impact of early rehabilitation in intensive care unit (ICU) on the survival and functional outcomes of COVID-19 patients admitted to ICU at acute phase. Materials and methods We conducted a prospective quasi-experimental study including 346 eligible COVID-19 patients from all admitted cases in an ICU in Vietnam, divided into three groups: no rehabilitation (n = 32), late rehabilitation (n = 109), and early rehabilitation (n = 205). Baseline characteristics and survival information of patient were collected with BORG-CR10 scale and PFIT; the data were collected at different time points: before intervention, when switching to oxygen-therapy, and at discharge. Results The control group (patients not using rehabilitation therapy) has worse survival than both early rehabilitation group (hazard ratio [HR] 0.553; 95% confidence interval [CI] 0.380–0.806; p value < 0.001) and late rehabilitation group (HR 0.374; CI 0.235–0.594; p value < 0.001). Regarding functional improvement, during the first five days, rehabilitation did not make a significant impact on the patients (p value > 0.05), however if continued from day 5 to day 20, the early-rehabilitation patients obtained a statistically significant improvement for BORG-CR10 (p value < 0.01). No clear association was found for PFIT (p value > 0.05). Conclusion The research emphasises the benefits of the early rehabilitation in ICU for COVID-19 patients. IMPLICATION FOR REHABILITATION Rehabilitation for severe COVID-19 patients in the intensive care unit (ICU) can improve patient survival during the ICU stay. This study suggests the benefit of early rehabilitation in ICU for COVID-19 patients. Early rehabilitation shows statistically significant improvement for exertion in patients who underwent rehabilitation at least 5 d.


Introduction
Coronavirus disease 2019 (COVID-19) is a respiratory infection caused by a recent new type of Corona virus [1].According to the report of the World Health Organisation (WHO) as of 25 May 2023, there were 766 million confirmed cases and over 6.9 million deaths reported globally [2].The rate of COVID-19 infections increased in the past time, consequently burdening the global health system and causing many inadequacies in emergency, treatment and prevention of functional impairment for patients [1].Clinical manifestations of SARS-CoV-2 infection range from asymptomatic to severe including acute respiratory distress syndrome (ARDS) and multi-organ failure requiring hospitalisation in the intensive care unit (ICU).
A systematic review and meta-analysis by Li et al. [3] from 39 studies with 80,487 patients showed that 10.96% of SARS-CoV-2 patients required ICU admission.The use of ventilators, sedation or analgesics, and prolonged immobilisation in ICU lead to many patients complicating pressure ulcers, atelectasis, hospital-acquired pneumonia, thromboembolism, microvascular changes, joint contractures, and delirium.These result in prolongation of mechanical ventilation, increasement of length of ICU stay and mortality, and development of disability [4].Muscle weakness, a common complication in the ICU, is also present in COVID-19 patients with ARDS and multiple organ failure [5], which are associated with severe disability and long recovery time after discharge [6].Significant muscle atrophy during the first week in ICU is up to 20% [5].Many patients who have survived have a reduced quality of life after discharge from ICU, related to a multifactorial illness that predisposes them to functional, physical, cognitive, and/or psychological disabilities lasting even months to years [6].
Prior to the pandemic, early rehabilitation for both hospitalised and ICU patients [7,8] had been shown to be effective and well tolerated, helping to reduce mortality risk, mortality, length of hospital stays, and medical costs [9].During the COVID-19 pandemic, which had brought serious medical burdens and consequences, an early recovery was required [9].There are several studies on the impacts of early rehabilitation programmes for patients with severe COVID-19 in the ICU stage.For example, a recent study (n = 10) reported that early rehabilitation of COVID-19 patients could effectively reduce the time to extubation and duration of ICU stay [10].Another study [11] using a retrospective cohort (n = 159) concluded that rehabilitation in acute severe COVID-19 pneumonia was safe for patients as well as healthcare workers and could reduce the duration of hospitalisation stay.Similar studies focusing on functional outcomes in COVID-19 patients requiring invasive ventilation are also reported elsewhere [12,13].
This study aims to highlight the impacts of early rehabilitation on acute-phase COVID patients by investigating the association of early rehabilitation with their survival and functional outcomes using a cohort of 346 patients from an ICU in Vietnam.

Study design and setting
We conducted a prospective quasi-experimental study of non-randomised intervention with control group [14] to analyse the impacts of early rehabilitation on COVID-19 patients using the data collected from the Bach Mai Hospital

Patient selection and allocation
We selected all patients from 18 years old and above, with a confirmed diagnosis of COVID-19, and were admitted to the ICU of Bach Mai Hospital 16th Facility.SARS-CoV-2 was confirmed by a test using reverse transcriptase-polymerase chain reaction (RT-PCR) from a respiratory specimen, performed at standard testing facilities validated by the Ministry of Health, Vietnam, and WHO 2021 [15].All intensive COVID-19 patients were defined as severe following the living guidance and received general medical treatment following the WHO guidelines [15].Exclusive criteria consisted of all cases who acquired an orthopaedic injury (contraindication with mobilisation), pre-existing neuromuscular disorder required mechanical ventilation aid prior to COVID-19, or a progress neurological deficit (unable to return to Glasgow coma scale [16] greater than 14).Moreover, patients were excluded if they did not approve of the rehabilitation programme.
All ICU-admitted patients met the inclusion criteria (n = 346) were enrolled in the study, then allocated in three groups: early rehabilitation (EarlyRehab, n = 205), late rehabilitation (LateRehab, n = 109) and no rehabilitation (NoRehab, n = 32), see Figure 1.The classification was based on their admission time before or after 1 September 2021, when the rehabilitation group (sub-unit) was established (RehabU start-time).Thus, all patients who were discharged from the ICU before the RehabU start-time were not enrolled in the rehabilitation programme, then were allocated to the NoRehab group.Patients who were admitted to the ICU before the RehabU start-time and discharged after that day were allocated to the LateRehab group.Finally, all patients admitted after the RehabU start-time were assigned to the EarlyRehab group.

Rehabilitation programme
The intervention programme was considered for all patients admitted after the RehabU start-time and applied on the whole ICU.The interdisciplinary team includes ICU physicians and rehabilitation clinicians to monitor patients, detect unstable conditions or contraindications to rehabilitation exercise, and consider letting the patients return to exercise as soon as when they met safety criteria [16].The safety rehabilitation criteria in ICU were monitored by the experts who made recommendations on safety criteria for active mobilisation of critically ill adults [17].The rehabilitation programme for COVID-19 patients in the acute phase was determined in the guidelines of the Ministry of Health, Vietnam, issued in April 2020 [18].According to the planning rehabilitation programme of the guidelines, the exercise programme for a patient was planned two sessions per day, with a suggestion of a total exercise time of at least 35-45 min for one session, depending on patient condition.This includes Passive range of motion exercise in 15 min, Postural drainage technique in 10-15 min, Chest percussion and vibration for 10-15 min, and other supportive exercises, such as postural therapy, assisted-breathing technique (Supplementary Note A).However, in the actual performed rehabilitation, which was used in this study, each session for a patient was limited to 15 min, twice per day, plus 15 min of transportation, infection control procedures, and medical record review.Thus, the total time of exercises for a patient per day reduces by about a half in comparison to the planning rehabilitation programme.This was mainly due to the limitation of human resources and working condition while taking care too many patients at that moment, which required the programme adjustment.For example, the personal protective equipment of rehabilitation staff in the COVID-19 outbreak required more time to change and took longer to move to the ward.Furthermore, the poor tolerability of acute COVID-19 patients for the rehabilitation exercises usually did not allow them to endure the whole exercise time as in the planning programme.The training frequency is 2 sessions per day, and 7 d per week.The training programme was personalised, according to the Glasgow Consciousness score (GCS) [16] and muscle strength level of the Medical Research Council (MRC) [19].For the early rehabilitation group, when a patient was admitted to the ICU, the rehabilitation was started whenever he/she satisfied the safety rehabilitation criteria in ICU following the guidelines of the Ministry of Health, Vietnam, which was described in detail in the Supplementary Note A. For the late rehabilitation group, the same procedure was applied after the RehabU start-time.The details of the rehabilitation programme are provided in Supplementary Note A.

Outcome measure
The primary outcomes were survival rate and the length of stay in ICU.The secondary outcomes were the functional capacity by BORG CR10 scale [21] and Physical Function in Intensive Care Test (PFIT) [22].The functional capacity of all patients was assessed at baseline, discharge, and time points of change in oxygen support.The BORG CR10 is a widely accepted category-ratio scale used to rate the perception of exertion during a task.The PFIT was specifically developed for use with critically ill patients who may not be able to perform the Six-Minute Walk Test (6MWT) or other submaximal exercise tests [22].The details of BORG CR10 and PFIT are provided in Supplementary Note B.

Statistical analysis
The description of data was described using appropriate statistics for different patient groups.Qualitative variables were represented as a count (n) and a ratio (%), and mean (standard deviation [sd]) was used for quantitative variables.For univariate comparison, we used the Wilcoxon signed-rank and Chi-squared test for quantitative and qualitative variables, respectively.For statistical analysis, a subtraction from max value and then logarithmic transformation were applied for SpO 2 to better approximate a normal distribution.
The statistical analysis mainly focused on two groups: i) data of patients with early rehabilitation therapy and ii) a control group where the data were collected before the RehabU start-time.For survival analysis, the control group contained the patient data (survival status and length of stay in ICU) censored before the RehabU start-time.The Kaplan-Meier estimate was used to compare the survival of patients in the control group against early and late rehabilitation groups.The association of rehabilitation therapy with patient survival was performed using Cox regression models, adjusted for the baseline characteristics.Group IV was used as the reference for comparison with other groups using the NIV methods.Crude (without adjustment), adjusted values of hazard ratio (HR), confidence interval (CI) and p value were reported.
In the secondary analysis, the control group included the data of functional capacity measurements before the RehabU start-time.The baseline measurement of patients was recorded at the first day of ICU admission without any rehabilitation therapy.The ratio of the follow-up measurements collected after the admission day to the baseline measurement was used to evaluate the impacts of early rehabilitation on patient's functional capacity (BORG CR10 and PFIT).The analysis was performed using linear regression models.Due to the shortage of measurements in the group of patients who died in the ICU (only 6 measurements for both BORG-CR10 and PFIT from 5 patients), the analysis focused only on the patients who were alive during their ICU stay.
Data analyses were performed using R software version 4.1.2(The R Foundation, Vienna, Austria, https://www.r-project.org/).A p value < 0.05 was considered statistically significant.

Results
The demographic characteristics were surveyed in all groups and presented in Table 1.The mean age of the study subjects was 58.7 (14.6), and the female-to-male ratio was approximate to 1:1, both had no statistically significant difference between the EarlyRehab and control groups.The comorbidities with the highest frequency were hypertension (47.7%) followed by diabetes mellitus (35.3%).These two comorbidities also had statistically significant differences between the control and EarlyRehab groups.No significant results were found in other comorbidities.Regarding clinical baseline characteristics, the rate of COVID-19 patients admitted ICU with a low level of consciousness (GCS < 8 points) was only 13.9% and the rate of patients with muscle strength (MRC) inferior to level 3 was 19.7%.The mean respiratory rate was 25.3 (4.6) breaths per minute, and the mean of blood oxygen saturation (SPO 2 ) was of 91.3 (7.7) %, indicating that COVID-19 hospitalisation was related to respiratory problems.The mean blood pressure and pulse rate values were within physiological limits.When compared to the control group, the EarlyRehab group had a less severe condition at admission with a lower frequency of patients with GCS < 8 and MRC < 3, but higher mean values for systolic blood pressure, breathing rate, and SPO 2 (p value < 0.05, see Table 1).
The initial requirement for respiratory support at admission was seen in most COVID-19 cases (97.7%) admitted to the ICU, of which IV support was required for only 11.8% of the participants.Among NIV supports, more than half (55.8%) were admitted with a mask, followed by HFNC (18.5%), and then NC (10.4%).There is no significant difference in respiratory support between the EarlyRehab group and the control group.
The COVID-19 patients had an average of 13.57(7.2) d staying in the ICU, and 134 patients (38.73%) died during the period of the cohort study, see Table 1.The EarlyRehab group had a shorter length of stay in ICU than the control group (p value < 0.001).However, their survival rates are not statistically different (p value > 0.05).
The out of bed (OFB) and standing/walking (SW) exercises were assigned to the patients in the LateRehab and EarlyRehab groups who met the criteria for BORG score and oxygen therapy, see Table 1.Particularly, OFB exercises were applied for the HFNC and MASK patients with the BORG-CR10 score ≤ 6, and SW exercises were applied for the NC patients with BORG-CR10 score ≤ 3. LateRehab and EarlyRehab had n = 77 (70.6%) and n = 120 (58.5%) patients using OFB exercises, respectively.For the LateRehab group, the average onset time (calculated from the start of rehabilitation) of bed training is 3.3 (sd = 6.3) d, which is 1.5 times greater than the time of the EarlyRehab group with the average of 2.2 (sd = 4.6) d.However, this difference is not statistically different (p value > 0.05).Regarding walking exercise, a total of 165 patients were assigned to walking training, with 69 (63.3%) patients in the LateRehab group and 96 (47.1%) patients in the EarlyRehab group, p value < 0.001.The average onset time of walking training for all patients was 7.8 (sd = 7.5) d.In comparison with the LateRehab group (mean = 9.6, sd = 8.9 d), the EarlyRehab group was assigned 3 d earlier (mean= 6.5, sd = 6.0 d, and p value < 0.05).

Association of early rehabilitation and patient survival
To investigate the association of early rehabilitation therapy with patient survival, we divided the survival data of the LateRehab patients into two parts: before (LateRehab−) and after (LateRehab+) the RehabU start-time (see Figure 2(A)).LateRehab − and NoRehab groups were combined to construct a new control group which was used for the survival analysis against the rehabilitation therapy groups including EarlyRehab and LateRehab+.
The Kaplan-Meier curves in Figure 2(B) show that the control group had worse survival compared to both groups of patients with rehabilitation therapy: EarlyRehab (crude HR 0.463; CI 0.327-0.656;p value < 0.001) and LateRehab+ (crude HR 0.355; CI 0.226-0.558;p value < 0.001).After adjusted for the baseline characteristics using multivariate Cox regression models, the association is still strongly significant for both EarlyRehab group (adjusted HR 0.553; CI 0.380-0.806;p value = 0.002) and LateRehab + group (adjusted HR 0.374; CI 0.235-0.594;p value < 0.001), respectively (see Table 2).Other risk factors including age, MRC, pulse rate, and SPO 2 are also statistically associated with patient survival.Comparing to the serious patients in group IV, patients from the NIV group including HFNC, MASK, or NC had better survival (p values < 0.001, see Table 2).The association is not significant for the group of patients without a need of ventilation at admission (NoVent group).The small sample size (n = 8) of the group might account for this observation.
Figure 2(B) shows no obvious separation between the survival curves of the LateRehab + group and the EarlyRehab group.We observe the similar result with the multivariate Cox regression model (adjusted HR 1.393; CI 0.860-2.257;p value > 0.05, see Supplementary Figure S1).Thus, in this study, the patients achieved benefits from both early rehabilitation and late rehabilitation therapy, and the difference between those was not significant.

Association of early rehabilitation and functional improvement on the BORG and PFIT scales
In this analysis, we evaluate the impact of the rehabilitation on the functional capacity of the COVIV-19 patients who were alive during the ICU stay.Table 3 presents the baseline measurements of functional capacity assessments taken at the date of ICU admission (before rehabilitation).A total of 129 patients (37.28%) had baseline measurement that met the criteria, among those 15, 89, and 25 patients were admitted to ICU with HFNC, MASK, and NC, respectively.The mean of BORG-CR10 decreases in the order of HFNC, MASK, and NC, reflecting the severity of the patient groups from high to low, see Supplementary Figure S2.The separation between the NIV groups is significant for BORG-CR10 (F-test p value < 0.001) but not for PFIT (F-test p value > 0.05).However, there are no significant differences between the EarlyRehab and control groups across all types of ventilation at admission.It took a median of five days since the ICU admission to perform the follow-up measurements across all patients.We then used the median value to classify the data into two intervals: i) the first five days of ICU admission (days 0-4) and ii) the period of two weeks later (days 5-20).The ratio of the follow-up measurement to the baseline measurement was used to evaluate the progression of the patient.For BORG-CR10, if the ratio was less than 1 (the follow-up measurement is smaller than the baseline value), the patient achieved improvement; otherwise, the functional capacity of the patient was not improved.It is the opposite for PFIT that the patient expected improvement when the follow-up measurement was greater than the baseline value (the ratio is greater than 1).Table 4 presents the results of multiple linear regression to investigate the association between the rehabilitation therapy and the functional capacities in the two-time intervals.The p values were adjusted for the baseline characteristics of the patients.For the first 5 d, the rehabilitation therapy did not make a significant statistical difference between the control and EarlyRehab groups.In the second interval from day 5 to day 20, the EarlyRehab group had a significantly lower BORG-CR10 proportion than the control group (adjusted p value = 0.009), indicating that the patients obtained benefits from the rehabilitation therapy to improve BORG-CR10 in this period.However, there was no statistically significant difference between the two groups for the PFIT score (see Table 4).

Discussion
Care services related to severe COVID-19 patients force enormous pressure on the health system, especially in ICUs.Rehabilitation  specialists have an important role in assisting patients for their recovery during and after COVID-19.To investigate the impacts of early rehabilitation on COVID-19 patients at an acute phase in the ICU, we performed a prospective quasi-experimental study using data from 346 patients collected patients during the 4th wave of the COVID-19 pandemic in the South of Vietnam.The results showed that early rehabilitation might facilitate the functional improvement and survival of COVID-19 patients in the ICU.The survival analysis in this study shows that the rehabilitation programme helps to improve the patient survival during ICU stay, which is similar to results from other studies [10,11].More interestingly, both groups of early and late rehabilitation gained similar benefits for survival in comparison with patients not using the therapy.The association of breathing rate, SpO 2 , and IV group with the patient survival can be due to the more severity of these patients in comparison with other patients.
Since the IV patients who received sedation and muscle relaxants were unconscious, only NIV patients were investigated for their functional improvement.An attempted explanation for the ineffectiveness during the first 5 d is that the duration is a very short time for a rehabilitation course to get clearly effective.Moreover, the first 5 d from admission are usually a critical period that the rehabilitation interventions mainly focus on effective breathing exercises, therapeutic postures, and gentle functional activities.Therefore, functional improvement measured by BORG-CR10 focusing on perceived exertion might not be obvious.After 5 d constantly supported with the rehabilitation programme, the patients would improve breathing difficulty, maintain muscle strength, consequently, improve their motor function, physical condition, and psychology, which would help these patients achieve significant improvement for BORG-CR10 in comparison with the patients not using the rehabilitation therapy.
Different from the IV-admitted patients who were unconscious and most severely ill, the NIV-admitted patients usually could perform simple movements, making high PFIT scores at the first evaluation.Thus, it was possible that the little change due to rehabilitation therapy could not make statistically different between the comparing groups.
Regarding the baseline characteristics, the average age of the patients in this study was close to the reports of the epidemiological studies from other countries: 58.7 comparing to 61.0 in Spain [11] and 65.0 in Canada [23].Similar to a recent report [23], there was also no difference in the sex ratio (female: male) of the patients in this cohort in Vietnam.For comorbidities, there was a statistical difference between the control and EarlyRehab group for hypertension and diabetes.We found that the proportion of patients with these comorbidities is time dependent, see Supplementary Figure S3.There was no difference between the EarlyRehab and control groups across all ventilation types that patients used at ICU admission.For clinical characteristics, GCS < 8, MRC < 3, Systolic blood pressure, Breathing rate, and SpO 2 are statistically different between the two groups.

Limitation
Conducting in a single setting is the major limitation of this study.The study utilised the data collected in the complicated period of the COVID-19 epidemic in Vietnam when the treatment facilities were overloaded.The patients collected for each group were non-randomised and biased due to the chronological order of admission.However, the bias was not avoidable in this type of study.Moreover, due to a lack of logistics in the emergency context of the hardest time of the COVID-19 outbreak in Vietnam, the follow-up data of these patients after ICU discharge was not able to be tracked and collected.Therefore, there was no re-evaluation of the patients in the long term.Besides, rehabilitation belongs to step E (Early mobility and Exercise) of the ABCDEF bundle [24], thus its outcomes might be influenced by the consequences of four previous steps including (A): Assess, Prevent, and Manage Pain, (B): Both Spontaneous Awakening Trials and Spontaneous Breathing Trials, (C): Choice of analgesia and sedation, (D): Delirium: Assess, Prevent, and Manage.However, data collection and assessment of these steps are beyond the scope of this study.In addition, this study did not consider intensive care unit-acquired weakness (ICU-AW) which focuses on the skeletal muscle disorder that commonly occurs following sepsis, mobility restriction, hyperglycaemia, and glucocorticoids or neuromuscular blocking agents use.In this study, we focused more on the functional improvement of the patients rather than their muscle strength.We reasoned that generally there were not many patients staying in the ICU for a long time, which might have neuromuscular complications causing ICU-AW.The short stay of these severe and acute COVID-19 patients was due to 1) their rapid death and 2) quick transfer to other medical care units if the severe patients passed the critical stage, as following the stratification and guides of the Ministry of Health of Vietnam.Therefore, we excluded the patients with pre-existing ICU-AW in the step of patient selection.Finally, the analysis of functional outcomes focuses on only alive patients during the ICU stay due to a shortage of data for patients who died in the ICU.Future studies with additional data and a well-designed randomisation trial are expected to provide more valuable evidence on the effectiveness of early rehabilitation.

Figure 2 .
Figure 2. (a) line plot of the cohort for survival analysis.the x-axis presents the day of admission of patients since the iCU starting date.each horizontal line presents the duration of a patient from iCU admission until discharged (circle) or died (triangle).the vertical dashed line indicates the RehabU start-time which is used for data censoring.the data on the left and right of the RehabU start-time contain the information of patients with and without rehabilitation therapy, respectively.(b) Kaplan-Meier survival curves for rehabilitation groups.For the lateRehab + group, the length of stay of a patient is calculated from the RehabU start-time.
16 Facility, from 11 August 2021 to 30 October 2021.The Bach Mai Hospital 16 Facility was established in District 7, Ho Chi Minh City with a scale of 2300 beds, including 500 beds for ICU supporting the treatments of severe COVID-19 patients during the 4th wave of the COVID-19 pandemic in the South of Vietnam.Beside the Rehabilitation Group, other specialised groups (sub-units) of Infectious Diseases, Microbiology, Radiology, Clinical Pharmacy, and Clinical Nutrition were established in the ICU to support COVID-19 treatments.The study was approved by the Bach Mai hospital's ethics review committee (Grant number 2726].

Table 1 .
baseline characteristics and outcomes of CoViD-19 patients in iCU, and comparison between the rehab group and the control group.
Note: Values are mean (standard deviation) or n (%). the bold values in the last column highlight statistical significance at the p-value <0.05.

Table 2 .
hazard ratio (hR) estimate and 95% confidence intervals (Ci) from univariate analysis of the effect of rehabilitation and multivariate analysis including baseline characteristics of patients.

Table 3 .
baseline measurement of functional capacity collected at iCU admission.

Table 4 .
Comparison between the control group and earlyRehab group using the boRD-CR10 ratio and PFit ratio. of each group is presented by mean (standard deviation).the bold value in the last column highlight statistical significance at the p-value <0.05.