Effects of early, combined endurance and resistance training in mechanically ventilated, critically ill patients: A randomised controlled trial

Introduction Neuromuscular weakness resulting in severe functional impairment is common in critical care survivors. This study aimed to evaluate effects of an early progressive rehabilitation intervention in mechanically ventilated adults at risk. Methods This was a parallel, two-arm, assessor-blinded, randomised controlled trial with 6-months follow-up that was conducted in a mixed ICU of an academic centre in Switzerland. Previously independent, mechanically ventilated, critically ill adults with expected critical care stay ≥72 hours (n = 115) were randomised to a control group receiving standard physiotherapy including early mobilisation or to an experimental group with early endurance and resistance training combined with mobilisation. Primary endpoints were functional capacity (6-Minute Walk Distance) and functional independence (Functional Independence Measure) at hospital discharge. Secondary endpoints including muscle strength were assessed at critical care discharge. Safety was monitored closely by standard monitoring and predefined adverse events. Results Physiotherapy started within 48 hours of critical care admission while 97% of participants were still ventilated and 68% on inotropes. Compared to the control group (n = 57), the experimental group (n = 58) received significantly more physiotherapy (sessions: 407 vs 377, p<0.001; time/session: 25min vs 18min, p<0.001) and had less days with sedation (p<0.001). Adverse events were rare (0.6%) and without consequences. There were no significant between-group differences in 6-Minute Walk Distance (experimental 123m (IQR 25–280) vs control 100m (IQR 0–300); p = 0.542) or functional independence (98 (IQR 66–119) vs 98 (IQR 18–115); p = 0.308). Likewise, no differences were found for the secondary outcomes, except a trend towards improved mental health in the experimental group after 6 months (84 (IQR 68–88) vs 70 (IQR 64–76); p = 0.023). Conclusions Early endurance and resistance training in mechanically ventilated, intensive care patients does not improve functional capacity or independence at hospital discharge compared to early standard physiotherapy but may improve mental health 6-months after critical care discharge. Trial registration German Clinical Trials Register (DRKS): DRKS00004347, registered on 10 September 2012.


Introduction
Intensive care unit acquired weakness (ICUAW) is common in critical illness survivors.It increases health-care costs and impairs quality of life [1].ICUAW mainly affects skeletal muscles (critical illness myopathy) and peripheral nerves (critical illness polyneuropathy) [2], but its causes are poorly understood [3].Symptoms of ICUAW develop within the first week of critical illness [4].Its risk factors include sepsis, multiple organ failure, mechanical ventilation (MV), immobilisation and hyperglycaemia [3].ICUAW is associated with worse short-and long-term outcomes, including weaning failure [5], prolonged ICU and hospital stays [6], increased mortality [7], and poor functional status with persistent disability in activities of daily living [1,[8][9][10].
Early rehabilitation and mobilisation may help to reduce the functional disabilities of ICUAW patients.Some promising short-term outcomes have been reported [11,12], but long-term benefits from randomised controlled trials are warranted [13][14][15].Delayed intervention onset (�1 week) [12,13,15], low ICUAW-risk in targeted population [11] or unspecific exercise delivery [11,14] may have contributed to the inconclusive results.Despite these limitations in evidence, many ICUs, including ours, have adopted early mobilisation, also as a consequence of using less sedation.We aimed to investigate a specific rehabilitation intervention that could be implemented early in a highly ill population with a high ICUAW risk.
Resistance training can restore strength after bed-rest in older adults [16].Combined endurance and resistance training (ERT) enhances exercise capacity and health-related quality of life in patients with chronic obstructive pulmonary disease [17] and prevents cardiovascular and skeletal muscle deconditioning during bedrest [18].Whether early ERT is feasible, safe and beneficial in critically ill patients of whom many are still mechanically ventilated and treated with opioids or recovering from sedation is unclear.We hypothesised that early ERT combined with mobilisation in the ICU is feasible and might improve function at hospital discharge.This randomised controlled trial compared the effects and safety of early ERT combined with early mobilisation to standard care including early mobilisation in critically ill, mechanically ventilated adults for a period of up to 6-months after hospital discharge.

Design and setting
This single-centre, parallel, two-arm, assessor-blinded randomised controlled trial with 6-months follow-up was conducted in the tertiary, mixed ICU of the Department of Intensive Care Medicine at the Inselspital, Bern University Hospital, Switzerland.The trial was approved by the local Ethics Committee and prospectively registered in the German Clinical Trials Register (DRKS00004347) on September 10, 2012.It adhered to Good Clinical Practice Guidelines and the CONSORT statement (S1 Checklist).The study protocol has previously been published [19].

Participants
Adults (�18 years) expected to stay on MV for at least 72 hours and who had been independent before the onset of critical illness were eligible for participation.Independence was defined as living without assistance and qualitatively determined by the patient's family or medical-health records.Exclusion criteria were previous muscle weakness, contraindications to cycling, enrolment in another intervention study, palliative care, admission diagnosis that excluded the possibility of walking at hospital discharge, and patients who did not understand German or French [19].

Procedures
Patients were screened daily by a research nurse and a specialist physiotherapist who evaluated cycling eligibility.The expected MV duration was judged by the responsible physician.Subsequently and before study inclusion, an independent medical doctor confirmed eligibility criteria.Physicians and physiotherapists were unaware of the randomisation sequence.A deferred consent was obtained from the next of kin within 72 hours after randomisation and a written informed consent from each patient as early as possible.Unrestricted, computer-generated randomisation (https://www.random.org/lists/) was performed by a study nurse partly involved in screening eligible candidates.Participants were allocated at a 1:1 ratio to either group using sequentially numbered opaque sealed envelopes to ensure allocation concealment [20].Coded data was managed with REDcap [21].
Blinding the responsible ICU staff was impossible, but the physiotherapy assessors were blinded for group allocation and distinct from the therapists providing the intervention.However, logistical reasons prevented therapist separation between the two treatment groups.Discharge decisions were made by the primary care team without the involvement of physiotherapists.

Interventions
Both groups received standard ICU care with protocol-guided sedation, weaning [22] and nutrition [23].In cases of ICU readmissions, treatment according to initial group allocation was continued.
Standard care (control group).Patients in the control group received European standard physiotherapy [24] including early mobilisation, respiratory therapy and passive or active exercises.These were physiotherapy-initiated and individually tailored, but subject to a medical prescription.Treatments took place once daily on weekdays and on weekends if the responsible therapists deemed an interruption of therapy to be harmful to patient's previous progress (e.g.intensive rehabilitation and weaning period or retained airway secretion in extubated patients).

Study intervention (experimental group).
In patients in the experimental group an early, progressive ERT programme combined with early mobilisation (S1 Fig) was initiated and each component implemented as intensive as possible and tolerated in individual patients.To this end, sedation was reduced if medically permitted prior to physiotherapy [19].Physiotherapists were encouraged to split therapy into two or more sessions to prevent overexertion.Therapy visits occurred from Monday to Friday up to a maximum of three sessions per day and on weekends if deemed helpful to patient's progress.
Endurance training was conducted with a motor-assisted bed-cycle (MOTOmed letto2, Reck-Technik, Betzenweiler, Germany) that allows passive, motor-assisted or active cycling in bed.Patients were comfortably positioned in a supine position with an individual head-of-bed elevation to permit optimal leg movement.Maximal training intensity for unresponsive patients was 20 minutes with a pedalling rate of 20 cycles/min [12].During each session, patients were prompted verbally and with the MOTOmed ServoCycle feature to actively participate.When patients were able to cycle with motor-assistance for 20 minutes, assistance was gradually decreased.Only then was training duration raised to 30 minutes.Subsequently, resistance and then duration were increased with a defined maximum of 60 minutes with full resistance (S1 Table ).
Resistance training included standardised exercises for both upper and lower limbs using weights or manually administered resistance from the therapist.The targeted training intensity was 8-12 repetitions with 2-5 sets (2min rest) on 50-70% of the estimated one-repetition maximum.Once the exercises were performed properly, participants were given an illustrated handout for further training with family members or nurses.In patients unable to perform resistance exercises because of lack of strength or comprehension, physiotherapists used passive movement or tactile facilitation to excite movement.
Early mobilisation was started with in-bed exercises and, if no medical contraindications were present, continued progressively from sitting on the bedside to sitting in a chair to standing and, finally, to walking.The overall aim was patients' active participation in functional activities of daily living in order to promote early independence.
Usual physiotherapy was provided for patients after ICU discharge once to twice a day.Sessions commonly involved functional exercises, cycling, walking, strengthening and breathing techniques.

Outcomes
The two primary outcomes were functional capacity (6-Minute Walking Distance (6MWD) [25]) and the ability to perform activities of daily living (Functional Independence Measure (FIM) [26]) at hospital discharge.
Secondary outcomes were FIM and muscle strength at ICU discharge.Muscle strength was assessed if patients could follow at least 3 out of 5 standardised questions [27].Assessments included the Medical Research Council (MRC) sum-score for ICUAW diagnosis [2], handgrip strength using the JAMAR dynamometer (Sammons Preston Rolyan, Bolingbrook, IL, USA) [28] and quadriceps muscle strength measured with a handheld dynamometer (Microfet 2, Biometrics, Almere, Netherlands) [29].Additionally, limitations in range-of-motion were recorded for shoulder flexion, elbow flexion and extension, fist closure, hip flexion, knee flexion and extension and foot dorsiflexion [30].To assess functional mobility the Timed 'Up & Go' was performed at hospital discharge [31].Further endpoints were time on MV, ICU and hospital length of stay along with achieved ICU mobility.Quality of life was determined with the Short Form 36 (SF-36) six months after hospital discharge.The SF-36 is a valid and reliable measurement to assess long-term physical and mental health in ICU survivors [32].
To judge training intensity and stability, physiological trends and individually-set limits instead of strict target numbers were used to account for each patient's condition and variability.Safety was closely monitored using standard ICU monitoring and indirect calorimetry for 30min before, during and 15min after physiotherapy.Physiological data was recorded in a patient data management system that stored 2-minute median values to remove artefacts.Prospectively defined adverse events (AE) included new hemodynamically relevant arrhythmias or otherwise unstable hemodynamics, oxygenation desaturation under 85%, a fall or other injury and any accidental removal of a tube, catheter or similar device.Per definition, AEs occurred during or up to 15min after physiotherapy and persisted despite an intervention or therapy interruption.Other known complications of bedrest, e.g.atelectasis or pressure ulcers, were recorded along with any serious AE which included death, a life-threatening injury or any event that caused an extension of hospital stay.

Statistical analysis
Sample size determination was based on the 6MWD to show a difference of 54m [33] and a mean walking distance of 301m (SD 81) [34].A statistical power of 80% and an α-level of 0.05 required a sample size of totally 72 patients (36 per group).This was further adjusted by 28 patients, due to two primary outcomes that were expected to highly correlate.Lastly, the overall sample size was increased by 15% to 115 participants because of expected attrition, including mortality.It was decided a priori that refusals to participate (presumed consent not obtained within 72h of randomisation) were compensated for and were not to be included in the intention-to-treat analysis.The number of these additional participants was determined after the inclusion of 115 patients.As a consequence, 130 patients were randomised until the number of 115 participants with presumed consent was reached.
Analysis was by intention-to-treat and per protocol using SPSS (IBM SPSS Statistics Version Premium GradPack 24), and R (Version 3.4.4).In the per protocol analysis, patients who died were allocated the lowest possible outcome for both primary outcomes.The intention-to-treat analysis was conducted with multiple imputation using a linear regression model that imputed all missing values (including death).Sensitivity analyses for both primary outcomes were conducted to test the robustness of the main analysis.Secondary outcomes were analysed as measured.Between group comparisons for normally distributed data included Student's t-tests and for nonnormally distributed data Mann-Whitney-U tests.Dichotomous data was analysed with Fisher's exact test, repetitive data with repeated measures analysis of variance (ANOVA) and correlations with Spearman correlation coefficients.Continuous data are presented as means with standard deviation (SD) or median with interquartile range (IQR) dependent upon their distribution.Categorical data are given as numbers with percentages.Treatment effects are reported with 95% Confidence intervals (95% CI) as mean differences (MD) for normally distributed data, as estimator for difference in location for non-normal data (Wilcoxon-Mann-Whitney test implemented in R package "coin"), and as incidence rate ratios (IRR) or risk differences (RD) for percentages.IRRs and RDs were calculated with R using Exact Poisson and Proportion test, respectively.Effect sizes were calculated with Cohen's d for normally distributed data and with Cohen's r for Wilcoxon-Mann-Whitney test statistic in case of non-normal data (using appropriate z value as in [35]).The level of significance was set at p < 0.05.A preliminary safety analysis has been conducted after the enrolment of 35 participants [36].

Study participants
From October 8, 2012 to April 5, 2016, we randomised 130 patients for whom 15 (12%) patient's proxy declined initial study participation.These were excluded from all further analyses, as requested by the Ethics Committee.Consequently, 115 patients (experimental n = 58, control n = 57) were included in the intention-to-treat analysis.(Fig 1).Baseline demographics and clinical characteristics were similar between groups (Table 1).

Primary outcomes
No significant differences were found between groups in the intention-to-treat and the per protocol analysis (Table 2).There was a high positive correlation (p<0.001,r = 0.833) between the two primary outcomes.Results were robust in sensitivity analyses (S2 Table ).

Secondary outcomes
There were no significant differences between groups at ICU and hospital discharge (Table 3).The incidence of ICUAW at ICU discharge, determined by MRC sum-score in cooperative patients, was 58% (23/40) in the experimental and 61% (26/43) in the control group (n = 83; p = 0.826).Limitations in range-of-motion were observed in 22 (38%) patients of the experimental and in 16 (28%) of the control group (n = 115; p = 0.323) with a median of 2 (IQR 0-3) affected joints per patient.No joint was particularly vulnerable for contracture.There were no significant differences for ICU (p = 0.694) and hospital discharge destinations (p = 0.555) between groups.The most frequent discharge destinations for survivors were rehabilitation (40%; experimental n = 25, control n = 21), local hospital (26.1%; experimental n = 17, control n = 13) and home (15.7%; experimental n = 7, control n = 11).Except for mental health (p = 0.023), there were no significant differences in quality of life at 6-months after hospital discharge (Table 3).Because of substantial missing data for the SF-36, characteristics for patients with and without the 6-month follow-up were compared.There were no significant differences between patients that completed the SF-36 versus patients that did not return the questionnaire (S3 Table ).

Safety
Safety details are given in Table 4. Overall, 25 (3%) of all physiotherapy sessions had to be discontinued (between-groups difference p = 0.839).The most frequent reasons for therapy discontinuation were exceedance of individually set limits (n = 8, 32%), lack of patient cooperation (n = 5, 20%) and fatigue (n = 3, 12%).There were 4 (0.6%) AEs, 1 (0.2%) in the experimental group due to an oxygen desaturation while cycling and 3 (0.8%) in the control group during mobilisation (one oxygen desaturation, two unstable haemodynamics).All AEs resolved after therapy discontinuation and did not have further consequences.The incidence of ICU complications did not differ between the two groups.ICU, hospital-and 6-month mortality rates were similar between groups over the study period.

Discussion
This randomised controlled trial did not find an advantage of early ERT combined with mobilisation over standard early mobilisation and rehabilitation.While participants in the experimental group received more therapy and had less days with sedation, control group participants had more out-of-bed mobilisations.Functional capacity and independence did not differ between these two different exercise regimes at hospital discharge.No differences were found in the secondary outcomes, except for a trend towards improved mental health at 6-months after hospital discharge.AEs were rare and a stepwise, individually-supervised training approach starting within 48 hours of ICU admission appears safe.ERT interventions are feasible and highly beneficial for functional improvement in chronic obstructive pulmonary disease [17] and we hypothesised that they may be equally feasible and effective in the critically ill.However, the transfer of these interventions may not be as straightforward as anticipated.Pulmonary rehabilitation is usually delivered over 8 to 12 weeks [17], while our exercise exposure was limited to the time spent in ICU, when patients were recovering from critical illness, mostly still on mechanical ventilation, receiving opioids and sedatives or at least recovering from previously received sedatives.Since the patients in our study spent the greater part of  their hospital-stay on wards or step-down units (experimental 77%, control 70%) and the main reason for missed therapy visits were weekends (58%), extended protocol-exposure with more resistance training might have enhanced potential group differences.The median ICUstay of 6 days seemed to have been too short to properly introduce resistance training, which was defined as using weight or resistance.However, assistive exercise (experimental group: 25% of movement exercises) in the very weak patient may affect strength as well.Wright et al [37] reported a higher but still modest percentage of sessions (33%) with this type of strengthening.In their study, similar to ours, sedation and patient fatigue prevented implementation or targeted intensity.Weakness (58% with ICUAW in our study) and ongoing effects of sedative agents despite significant reduction in their use in this group may both have contributed.
The main difference between our study and previous ones reporting benefits from early mobilisation is that the control group was also mobilised early, as per our ICU standard.We started the intervention very early in both groups, while patients were still mechanically ventilated and on vasoactive support, whereas most [12,13,15,37,38], but not all other intervention groups [11,39] started substantially later.Nevertheless, our control group received physiotherapy as early as the intervention group in the study of Schweickert et al [11], and more often than treatment groups in other trials [37,38].In contrast to our study, Schaller et al [39] found that early mobilisation shortened ICU length of stay and improved functional mobility in surgical ICU patients at hospital discharge.Contribution of treatment of control groups to the different results cannot be evaluated due to insufficient details of care received in Schaller's study.Yet the case mix appears very different: our patients were sicker (mean APACHE II 23 vs 16 in [39]) and had less surgical patients.It is therefore conceivable that the early and intensive physiotherapy in the control group of our study explains the lack of effect  of additional endurance training.Early beginning of physiotherapy seems crucial due to early onset of ICUAW [4].Accordingly, Schweickert et al [11] combined early mobilisation (within 1.7d of intubation) with daily sedation-interruption leading to reduced delirium and to significant functional improvement at hospital discharge.Delirium was equally prevalent in our groups, though the incidence was comparable to the intervention group of [11] (39 vs 33%).However, despite the early start, patients in our study achieved the specific mobilisation milestones substantially later [11].This may also be related to the case mix-our patients were substantially older and sicker.These patients might not always be suitable for early mobilisation [40], but could profit from appropriate exercise interventions over the whole ICU stay [12].However, recent evidence also highlights the impact of pre-ICU health [41,42] and post-extubation dysphagia [43] on patients' functional recovery.Our heterogenous sample might therefore conceal responsive subgroups.More research is needed whether sicker patients need longer rehabilitation exposure [44] and for which subgroups recovery will be achievable.The conducted early endurance training seemed to have unfavourable effects on mobilisation levels leading to less out-of-bed mobilisations in this group.This self-funded, pragmatic study did  not have additional resources for therapy-delivery.Physiotherapists may therefore have used available resources for cycling instead of early mobilisation.Although more out-of-bed mobilisations did not positively impact function of the control group either.The large AVERT trial even found that early, high-dose out-of-bed mobilisation after stroke enlarges the probability for disability when compared to low-dose, early mobilisation [45].Dose-response analyses found the greatest benefit for short, frequent mobilisations [46].This would favour the higher frequency of physiotherapy sessions (11%) in the intervention group, although this percentage might have been insufficient to reach clinical significance.Similarly, the stepwise approach of our study might have been too ambitious for the allocated therapy time (7min extra) and cycling insufficiently dosed (15min versus 30min reported by [47,48]) to induce musculoskeletal adaptations.However, the small but significant changes in hemodynamic or oxygen-transport variables appeared to have been sufficient to stimulate patients to meet the metabolic demands of exercise by adequate cardiorespiratory adaptations.In contrast, neither passive cycling in deeply sedated patients [49] nor passive chair transfer [50] increased metabolic costs.Our results are in line with recent studies [13,15,37,38,51] that failed to show a benefit of early rehabilitation.Heterogenous populations, difficulties in group separation and exercise dose may all have contributed.The discrepancy between planned and actual therapy delivery in these trials implies that the critically ill's ability to participate in functional exercises may be overestimated.More attention should be paid to patient fatigue and on whom to target, as very early muscle preservation might not be feasible in a seriously ill cohort with active catabolic pathways.Resistance training may therefore be more feasible during recovery.
This study had several limitations.First, blinding of participants and physiotherapists was impossible.Second, loss of follow-up at 6 months was substantial and unequally distributed between groups (41% versus 58%) but without any differences in responders and nonresponders.Nevertheless, results from the SF-36 are not adequately powered, possibly attributable to type I error and need to be interpreted with caution.Third, we targeted patients at risk of ICUAW who were previously independent.Therefore, data from this trial might not be applicable to patients with previous functional dependency.Fourth, the observed difference of 23m in the 6MWD may be of clinical significance [52], however, to test a difference of 25m (SD 150) would require at least 1128 participants.Similarly, the two primary outcomes should be adjusted for multiplicity to avoid type I error.
This study also has several strengths: patients were included very early after ICU admission and received physiotherapy within 48 hours.Physiotherapy interventions were pragmatic and reflect real-world resources.The novel use of individually set limits instead of strict target numbers allows for more rehabilitation flexibility and appears to be a safe approach with a low incidence of AEs (0.6%).Nevertheless, physiotherapists were experienced in ICU rehabilitation, which is shown in the exceptionally high control group participation.With a total of 1368 edge-of-bed and 1292 out-of-bed mobilisations on 1130 ICU days this trial further underlines the importance of ICU rehabilitation as a team effort.Additionally, the blinded outcome measurements were patient-centred, valid, reliable and relevant to the devastating consequences of ICUAW.
In the light of these data future studies should assure sufficiently large sample sizes, aim to increase exercise-exposure and ensure a targeted, frequent exercise delivery to the right patient.The effect of early exercise on mental health needs further evaluation, considering that one in two survivors might experience psychiatric symptoms [53].

Conclusions
We did not find functional benefits of early exercise training over an active usual care rehabilitation in mechanically ventilated, critically ill adults.Successively-adapted, early exercise training seemed to be safe.Early endurance training might improve long-term mental health, but the most effective rehabilitation intervention to excite neuromuscular adaptations still needs to be determined.

The
Department of Intensive Care Medicine has, or has had in the past, research contracts with Orion Corporation, Abbott Nutrition International, B. Braun Medical AG, CSEM SA, Edwards Lifesciences Services GmbH, Kenta Biotech Ltd, Maquet Critical Care AB, Omnicare Clinical Research AG and research & development/ consulting contracts with Edwards Lifesciences SA, Maquet Critical Care AB, and Nestle ´.The money was paid into a departmental fund; JT and SMJ received no personal financial gain.The Department of Intensive Care Medicine has received unrestricted educational grants from the following organizations for organizing a quarterly postgraduate educational symposium, the Berner Forum for Intensive Care (until 2015): Fresenius Kabi, gsk, MSD, Lilly, Baxter, astellas, AstraZeneca, B | Braun, CSL Behring, Maquet, Novartis, Covidien, Nycomed, Pierre Fabre Pharma AG (formerly known as RobaPharm), Pfizer, Orion Pharma, Bard Medica S.A., Abbott AG, Anandic Medical Systems.The Department of Intensive Care Medicine has received unrestricted educational grants from the following organizations for organizing bi-annual postgraduate courses in the fields of critical care ultrasound, management of ECMO and mechanical ventilation: Pierre Fabre Pharma AG (formerly known as RobaPharm), Pfizer AG, Bard Medica S.A., Abbott AG, Anandic Medical Systems, PanGas AG Healthcare, Orion Pharma, Bracco, Edwards Lifesciences AG, Hamilton Medical AG, Fresenius Kabi (Schweiz) AG, Getinge Group Maquet AG, Dra ¨ger Schweiz AG, Teleflex Medical GmbH.No author was employed by any of these companies.None of the companies played a role in this study.Our relationship with the companies does not alter our adherence to PLOS ONE policies on sharing data and materials.

Table 2 . Results for the two primary outcomes per protocol and intention-to-treat analysis.
Multiple imputation from a linear regression model based on age, gender, BMI, weight, APACHE II score, TISS-28, Tiss-76, SOFA score, ICU days until randomisation, length of stay in ICU and hospital, duration of mechanical ventilation, FIM score and MRC sum-score at ICU discharge.All missing values (including death before hospital discharge) were imputed (6MWD: n = 33, FIM: n = 32). a

Table 5 . Treatment details for the control and experimental groups. Experimental group (n = 58, 478 study days) Control group (n = 57, 558 study days) p value Group difference (95% CI) Physiotherapy (PT)
a Median of all patients' median session duration