Cardiorespiratory response to early rehabilitation in critically ill adults: A secondary analysis of a randomised controlled trial.

INTRODUCTION: Early rehabilitation is indicated in critically ill adults to counter functional complications. However, the physiological response to rehabilitation is poorly understood. This study aimed to determine the cardiorespiratory response to rehabilitation and to investigate the effect of explanatory variables on physiological changes during rehabilitation and recovery.
METHODS: In a prospectively planned, secondary analysis of a randomised controlled trial conducted in a tertiary, mixed intensive care unit (ICU), we analysed the 716 physiotherapy-led, pragmatic rehabilitation sessions (including exercise, cycling and mobilisation). Participants were previously functionally independent, mechanically ventilated, critically ill adults (n = 108). Physiological data (2-minute medians) were collected with standard ICU monitoring and indirect calorimetry, and their medians calculated for baseline (30min before), training (during physiotherapy) and recovery (15min after). We visualised physiological trajectories and investigated explanatory variables on their estimated effect with mixed-effects models.
RESULTS: This study found a large range of variation within and across participants' sessions with clinically relevant variations (>10%) occurring in more than 1 out of 4 sessions in mean arterial pressure, minute ventilation (MV) and oxygen consumption (VO2), although early rehabilitation did not generally affect physiological values from baseline to training or recovery. Active patient participation increased MV (mean difference 0.7l/min [0.4-1.0, p<0.001]) and VO2 (23ml/min [95%CI: 13-34, p<0.001]) during training when compared to passive participation. Similarly, session type 'mobilisation' increased heart rate (6.6bpm [2.1-11.2, p = 0.006]) during recovery when compared to 'exercise'. Other modifiable explanatory variables included session duration, mobilisation level and daily medication, while non-modifiable variables were age, gender, body mass index and the daily Sequential Organ Failure Assessment.
CONCLUSIONS: A large range of variation during rehabilitation and recovery mirrors the heterogenous interventions and patient reactions. This warrants close monitoring and individual tailoring, whereby the best option to stimulate a cardiorespiratory response seems to be active patient participation, shorter session durations and mobilisation. TRIAL REGISTRATION: German Clinical Trials Register (DRKS) identification number: DRKS00004347, registered on 10 September 2012.

Introduction

Intensive care unit (ICU) survivors frequently experience poor recovery with multifactorial physical, cognitive and mental health impairments [1-5]. Early rehabilitation and mobilisation in critically ill patients are advocated to attenuate these complications [6]. Their safety has been confirmed in a large meta-analysis of >22,000 rehabilitation sessions [7]. Moreover, they increase functional mobility and muscle strength [8] and reduce days with delirium and mechanical ventilation [9]. However, insufficient intervention reporting of frequency, intensity, type and timing limit their implementation in clinical practice [10]. Additionally, the appropriate exercise type or intensity to induce appropriate physiological adaptations remains yet to be determined.

There is previous evidence that sitting on the edge of the bed may be associated with higher metabolic cost compared with a passive chair transfer [11] or passive in-bed cycling [12]. The hierarchy of these mobility intensities are mirrored in the ICU mobility scale [13] although the highest activity levels may not be the most physically intensive [14, 15]. Similarly, the cardiorespiratory response is likely to vary between different types of exercise, exercise-modalities or individuals [16, 17]. Individually tailored rehabilitation interventions therefore seem appropriate. However, there is little data to guide the physiotherapist’s clinical decision-making in delivering optimal training intensity.

The purpose of this planned secondary analysis was to determine the cardiorespiratory response to early physiotherapy-led rehabilitation in mechanically ventilated, critically ill adults in a mixed ICU. First, we aimed to analyse physiological variables from before, during and after rehabilitation to describe both the response to rehabilitation and to estimate recovery. Second, we aimed to characterise effects of a-priori selected explanatory variables on cardiorespiratory reactions from before to during rehabilitation and from before to after rehabilitation. We hypothesised that session type, mobilisation level, exercise modality and duration as well as individual patient characteristics would be the main drivers of the physiological response to rehabilitation. Finally, we re-evaluated safety by examining sessions with clinically relevant variations (>10%), an adverse event or therapy discontinuation to determine predictors and explored individual characteristics of patients with a strong physiological reaction.

Materials and methods

Study design

This response analysis was a planned secondary analysis of the physiological data collected during a pragmatic, randomised controlled trial comparing very early endurance and resistance training combined with mobilisation to usual care in mechanically ventilated, critically ill adults [18]. The trial was conducted in a mixed ICU of a Swiss academic centre (Department of Intensive Care Medicine, Inselspital, Bern University Hospital) between October 8, 2012 and April 5, 2016. No significant differences were found in the primary or secondary outcomes with the exception of improved mental health six months after hospital discharge for the experimental group [19]. Consequently, this secondary analysis considered the two randomised groups as one population using data from all trial participants with at least one rehabilitation session. A preliminary safety analysis of the physiological data of the first 35 subjects indicated a moderately increased workload with increased heart rate and oxygen consumption but stable oxygen saturation from before to during rehabilitation [20].

The local ethics committee approved the study that was registered in the German Clinical Trials Register (DRKS00004347) on September 10, 2012. Written informed consent was obtained from all participants or their representatives.

Population

Participants were ≥ 18 years old, functionally independent before ICU admission and expected to remain ventilated for ≥ 72 hours. They were ineligible in cases of suspected previous muscle weakness, contraindications to cycling, palliative care, admission diagnosis that precluded walking at hospital discharge or insufficient command of German or French [18].

Intervention delivery

Early rehabilitation interventions have been previously described [18, 19]. In brief, they started within 48 hours of ICU admission with common rehabilitation interventions provided by certified physiotherapists. Within the context of the RCT, exercise frequency, intensity, type and duration were individually tailored based on the clinical judgement of the treating physiotherapist in one group, while the other used a stepwise, standardised approach dependent upon tolerance and stability (S1 File). Patients in both groups were closely monitored and treatment interventions recorded. A pragmatic trial design was chosen to reflect real-world resources and to enhance clinical implementation. In consequence, any rehabilitation session may have included short, individually-set breaks between efforts. Physiotherapists could prematurely cease a session based on their clinical reasoning, but had to note the reason for therapy discontinuation. Adverse events were prospectively defined as ‘persisting despite an intervention or therapy interruption’ and included a new unstable hemodynamic, oxygen desaturation (<85%), accidental fall or device dislocation [18]. This flexible approach considered individually set limits instead of strict target numbers which might be too rigid for this heterogenous population. Finally, all participants had a resting period of 30 minutes before and 15 minutes after each physiotherapy session, where they should not be disturbed.

Data collection and measurements

Physiological data were collected by standard ICU monitoring for heart rate (HR [beats per minute: bpm]; electrocardiogram), mean arterial pressure (MAP [mmHg]; arterial line), minute ventilation (MV [l/min]; Serv0-i V3.0, Maquet Getinge Group, Gossau, Switzerland), oxygen saturation (SpO2 [%]; pulse oximetry) and by indirect calorimetry for oxygen consumption (VO2 [ml/min]; CARESCAPE patient monitor B850 with E-COVX-00 Module, GE Healthcare, Finland). Indirect calorimetry was installed for planned sessions in mechanically ventilated patients from Monday to Friday who did not have a contraindication such as high FiO2 (>60%), respiratory rate (>30 bpm) or intolerance for airway leak (e.g., PEEP >15 cmH2O). On days with more than one physiotherapy session (<10%), only one, nonspecific session was monitored by indirect calorimetry.

All values were delivered to our patient data management system (PDMS: Centricity Critical Care Clinisoft, GE, Barrington, IL, USA) that recorded 2-minute medians to remove artefacts. We further screened and deleted isolated values that were out of range (HR ≥ 220bpm; MAP negative or ≥ 200mmHg; VO2 ≤ 30ml/min). Afterwards we collected the median, maximum, minimum, mean and coefficient of variability (CV: defined as maximum minus minimum divided by mean multiplied by 100 –calculated to estimate fluctuations) for all values in the three prespecified timepoints: baseline (30min before), training (physiotherapy duration) and recovery (15min after). After each session, the treating physiotherapists recorded the time for beginning and end of physiotherapy, types of interventions, treatment modality (active, passive, mixed patient participation), mobilisation level (in-bed, edge-of-bed, out-of-bed), discontinuation or adverse events while a study nurse noted patient baseline characteristics, airway management, daily medications and Sequential Organ Failure Assessment (SOFA). Within the context of the original pragmatic randomised trial, we chose not to analyse the physiological response to isolated types of interventions, but investigated common treatment packages in the critically ill. Based on the treatment interventions performed we coded each session into seven predefined categories. These categories–termed ‘session type’—were cycling, mobilisation, respiratory management, exercise, exercise and respiratory management, complex exercise and mobilisation (meaning multifaceted rehabilitation that included mobilisation), and complex cycling and mobilisation (meaning multifaceted rehabilitation that included mobilisation and cycling) (S1 File).

Statistical analysis

Descriptive analysis

We describe the study population in terms of counts (n), percentages (%), means with standard deviation (SD) and medians with interquartile range (IQR). We visualized physiological trajectories over the three time-points and calculated correlations between physiological values.

Cardiorespiratory response

The impact of explanatory variables on physiological values during (training) and after (recovery) rehabilitation was investigated in respect to before the physiotherapy session and variability (CV) with Gaussian mixed-effects models which accounted for correlation of within-individual measurements (S1 File) based on Vickers et al. [21]. We excluded physiological values from HR and MAP with a corresponding zero CV to account for cardiac pacing. Explanatory variables were prospectively determined using extensive clinical reasoning and previous evidence to account for confounders. They included patient characteristics (age, gender, Body Mass Index (BMI), daily SOFA), attributes of the physiotherapy session (session type, mobilisation level, treatment modality, session duration, time since ICU admission), and ICU environment (airway, daily medication). We report p-values from a likelihood test to test the overall significance of categorical variables. The correlation between explanatory variables was examined to avoid collinearity (without taking into account the correlations among explanatory variables of the same subjects). The estimated coefficients may be interpreted as the impact of the explanatory variable on the change. Accordingly, these estimates characterise the impact of the explanatory variables on the average values of ‘during’ or ‘after’ rehabilitation in respect to the value ‘before’ rehabilitation.

Safety analysis

We used a mixed-effects logistic regression model which accounted for correlation of within-individual measurements to investigate explanatory variables related to clinically relevant variations and report odds ratios with 95% confidence intervals (CI). The safety cut-off for a clinically relevant variation in physiological measurements was defined as >10% variation–based on half of what is commonly reported as an adverse event (>20%) in the literature [22]. Exercise is structured and repetitive physical activity that results in energy expenditure [23]. Thus, a 10% threshold might be a safe cardiorespiratory training intensity in the critically ill adult.

Considering the exploratory, hypothesis-generating purpose of our secondary analysis, the significance threshold was set to 0.05 without adjustment for multiple testing. The statistical analysis was performed with R version 4.0.2 (2020-06-22).

Results

During the trial, a total of 784 physiotherapy-led rehabilitation sessions were conducted in 113 participants. The duration of physiotherapy was missing for 15 sessions, which were subsequently excluded. Thus, a total of 769 sessions for 113 participants were available. After the removal of sessions with a CV of zero for HR and MAP, we analysed physiological data from 716 sessions for 108 participants with a median of 3 [2-8] sessions per subject (S1 Fig). The characteristics of these sessions and participants are described in Table 1. Numerical physiological values and their correlations from before, during and after the sessions are described in S2 File (S1 Table, S2 Fig). The strongest correlations (r > 0.5) occurred within the same physiological value between the three timepoints and between MV and VO2 over all timepoints.

Table 1

Participant and physiotherapy session characteristics.

Participant variables	N	Median [25%, 75% quantile] or counts (%)
Age (years)	108	66.7 [55.1, 74.4]
Gender (male)	108	72 (67%)
Body Mass Index (kg/m2)	108	26.4 [23.7, 29.7]
APACHE II score (0–71)	108	22.0 [17.8, 27.2]
SOFA score at ICU admission (0–24)	108	8 [6–11]
ICU days until study inclusion	108	1.8 [0.9–2.6]
ICU length of stay (days)	108	7.0 [4.6–13.9]
Physiotherapy sessions per subject	108	3 [2–8]
Randomised to the experimental group	716	372 (52%)
Time from ICU admission to start of first session (days)	108	2.0 [1.4–3.1]
Time from ICU admission to start of individual session (days) a	716	8.6 [3.9, 19.6]
Physiotherapy session duration (min)	716	22.0 [16.8, 30.0]
Daily SOFA score during physiotherapy (0–24)	713	8 [5, 12]
Session types	716
exercise		193 (27%)
cycling		160 (22%)
mobilisation		178 (25%)
respiratory management		66 (9%)
exercise and respiratory management		54 (8%)
complex cycling and mobilisation		10 (1%)
complex exercise and mobilisation		55 (8%)
Treatment modality	610
passive		401 (66%)
active		187 (31%)
mixed		22 (4%)
Mobilisation level	716
in-bed		488 (68%)
edge-of-bed		150 (21%)
out-of-bed		78 (11%)
Physiotherapy session discontinuation b	716	23 (3%)
Adverse event during physiotherapy	716	4 (1%)
Airway support	716
endotracheal tube		327 (46%)
tracheostomy		271 (38%)
none		118 (16%)
Neuromuscular blocking agents on day of session	716	98 (14%)
Vasoactive support on day of session	716	371 (52%)
Opiates on day of session	716	662 (92%)
Sedatives on day of session	716	539 (75%)

a number of sessions varied between patients, this variable takes into account the time from ICU admission to the start of each, individual session in the individual patient.

b for all physiotherapy sessions (n = 784) there were 25 (3%) therapy discontinuations (two sessions with a therapy discontinuation from one subject were excluded in this analysis because of zero CV).

Abbreviations: SD = standard deviation, APACHE = Acute Physiology and Chronic Health Evaluation, SOFA = Sequential Organ Failure Assessment.

Descriptive analysis

The differences between the timepoints from before to during (training) and before to after (recovery) for medians of physiological values are illustrated in Fig 1. On average, these differences were not clinically relevant although wide 95% CI indicate fluctuations across sessions (S2 File: S1 Table). Clinically relevant variations (>10%) were highest for MV (training: 35.6% of sessions, recovery: 33.8%), VO2 (26.0%, 26.1%) and MAP (21.8%, 28.2%) (S2 File: S2 Table). These fluctuations become more apparent when plotting physiological trajectories across the three timepoints according to ‘session type’ (Fig 2). Differences for CV were equal across the three timepoints, but again varied in regards to ‘session type’ (S2 File: S3 and S4 Figs).

Fig 1

Differences in median physiological values from before to during/after physiotherapy.

Median differences were calculated as “during minus before” (training) and “after minus before” (recovery) physiotherapy for all physiological values. Measurement units: HR (bpm), MAP (mmHg), MV (l/min), SpO2 (%), VO2 (ml/min). Numerical differences (median [25%, 75%]): HR: training 1bpm [-1, 3.5], recovery 0.5bpm [−2, 3]; MAP: training 1.5mmHg [−2, 5], recovery -0.5 mmHg [−4, 4]; MV training 0.35l/min [-0.15, 1.2], recovery 0l/min [-0.6, 0.75]; SpO2: training 0% [−1, 1], recovery 0% [−1, 1]; VO2 training 8.8ml/min [-2.47, 25], recovery 1.65ml/min [-10.9, 18.3] (S2 File: S1 Table).

Fig 2

Trajectories of median physiological values according to session type.

A) Average trajectories with standard errors. M) median trajectories with IQR of physiological measurements. Standard errors are computed under the assumption of independences among all the observations. Measurement units: HR (bpm), MAP (mmHg), MV (l/min), SpO2 (%), VO2 (ml/min).

Cardiorespiratory responses

Explanatory variables had a low correlation with themselves and were all kept in the analysis (S2 File: S2 Fig). We report the estimated effect of explanatory variables for HR, MAP, MV and VO2 ‘during’ (Table 2) and ‘after’ physiotherapy-led rehabilitation (Table 3). Non-modifiable explanatory variables that mostly affected physiological responses during and after physiotherapy were age, gender, BMI and the daily SOFA score, whereas modifiable explanatory variables were session type, treatment modality, session duration, mobilisation level and daily medication. Shorter ‘session duration’ and ‘active treatment modality’ generally increased physiological parameters during rehabilitation, while cardiorespiratory parameters did not return back to baseline for ‘session type’ and ‘mobilisation level’ during the prespecified 15-min recovery-phase. Explanatory variables with an effect on SpO2 were ‘mobilisation level’ and ‘airway support’ (S2 File: S4 Table).

Table 2

Estimated fixed effect of explanatory variables on HR, MAP, MV and VO2 ‘during’ rehabilitation.

Explanatory variables	HR during (95%-CI)	MAP during (95%-CI)	MV during (95%-CI)	VO2 during (95%-CI)
Number of sessions	571	535	442	312
Age (years) a	0.02 (-0.02, 0.06)	0.01 (-0.03, 0.05)	-0.002 (-0.01, 0.01)	-0.54 (-0.87, -0.21)
Gender (male is reference)	0.24 (-0.84, 1.32)	1.03 (0.01, 2.05)	-0.42 (-0.72, -0.12)	-23.78 (-32.74, -15.39)
Body Mass Index (kg/m2) a	0.02 (-0.09, 0.12)	0.04 (-0.07, 0.14)	-0.01 (-0.04, 0.02)	0.88 (-0.02, 1.77)
Daily SOFA score (0–24) a	0.002 (-0.11, 0.12)	-0.02 (-0.15, 0.11)	0.01 (-0.02, 0.05)	0.29 (-0.73, 1.32)
Session duration (min) a	0.05 (0.01, 0.09)	-0.06 (-0.12, -0.01)	-0.02 (-0.03, -0.00)	-0.59 (-1.10, -0.13)
Time from ICU admission to start of session (days) a	0.02 (-0.01, 0.06)	0.03 (-0.02, 0.07)	0.001 (-0.01, 0.01)	-0.18 (-0.59, 0.20)
Session type (exercise is reference)
p-value b	0.136	0.627	0.434	0.615
cycling	-0.93 (-2.05, 0.24)	0.45 (-0.94, 1.84)	0.15 (-0.17, 0.48)	3.52 (-6.38, 14.18)
mobilisation	2.05 (-1.79, 5.92)	0.28 (-4.68, 5.25)	0.79 (-0.36, 1.93)	38.54 (-14.80, 89.51)
respiratory management	0.72 (-0.92, 2.40)	1.83 (-0.33, 4.00)	0.36 (-0.36, 1.14)	1.56 (-62.48, 65.85)
exercise and respiratory management	0.29 (-1.15, 1.70)	0.25 (-1.55, 2.06)	-0.02 (-0.46, 0.43)	-0.04 (-14.13, 14.72)
complex cycling and mobilisation	-2.59 (-5.74, 0.54)	1.98 (-1.99, 5.95)	0.19 (-0.70, 1.08)	-7.51 (-35.63, 20.44)
complex exercise and mobilisation	2.21 (-1.50, 5.98)	1.28 (-3.53, 6.08)	0.24 (-0.86, 1.33)	19.71 (-27.29, 66.16)
Treatment modality (passive is reference)
p-value b	0.064	0.016	<0.001	<0.001
mixed	-0.38 (-2.35, 1.59)	2.88 (0.26, 5.51)	0.53 (-0.06, 1.12)	1.11 (-17.97, 20.88)
active	1.16 (0.13, 2.19)	1.41 (0.13, 2.69)	0.72 (0.40, 1.04)	23.01 (12.51, 33.97)
Mobilisation level (in-bed is reference)
p-value b	0.610	0.900	0.294	0.681
edge-of-bed	-0.78 (-4.58, 3.03)	0.88 (-4.01, 5.78)	0.74 (-0.41, 1.88)	-2.85 (-52.04, 47.78)
out-of-bed	0.13 (-3.91, 4.08)	0.81 (-4.36, 5.99)	0.36 (-0.84, 1.55)	6.87 (-45.68, 63.19)
Airway support (none is reference)
p-value b	0.502	0.382	0.379	0.988
tracheostomy	-0.90 (-2.36, 0.59)	0.94 (-0.89, 2.77)	0.30 (-0.33, 0.96)	1.36 (-24.32, 27.77)
endotracheal tube	-0.49 (-1.77, 0.83)	1.14 (-0.52, 2.81)	0.05 (-0.54, 0.68)	0.85 (-23.42, 25.36)
Opiates on session day c	0.39 (-1.10, 1.81)	0.62 (-1.41, 2.64)	0.12 (-0.44, 0.67)	-7.44 (-24.94, 9.70)
Vasoactive on session day c	-0.11 (-1.04, 0.81)	-1.79 (-2.95, -0.64)	-0.13 (-0.41, 0.15)	-6.92 (-15.76, 2.16)
Sedatives on session day c	1.15 (0.06, 2.20)	1.32 (-0.05, 2.68)	0.04 (-0.34, 0.41)	4.88 (-8.80, 19.34)
Neuromuscular blocking agents on session day c	-0.19 (-1.25, 0.89)	-0.37 (-1.73, 0.98)	-0.21 (-0.51, 0.11)	-6.73 (-16.85, 3.20)

a per one-unit increase (for continuous variables).

b likelihood test for overall significance of categorical variables.

c none is reference.

Reported effects of explanatory variables are mean differences of median values and need to be considered under the assumption of ‘all other covariates being constant’. All models are adjusted for measured values and CV before physiotherapy (estimates not shown). Examples for interpretation: (1) Categorical data: Median VO2 significantly increased during training by 23.01ml/min with active patient participation when compared to passive patient participation. (2) Continuous data: Median VO2 decreased during training per one additional year of age by 0.54ml/min.

Table 3

Estimated fixed effect of explanatory variables on HR, MAP, MV and VO2 ‘after’ rehabilitation.

Explanatory variables	HR after (95%-CI)	MAP after (95%-CI)	MV after (95%-CI)	VO2 after (95%-CI)
Number of sessions	569	534	437	308
Age (years) a	0.05 (0.01, 0.08)	0.001 (-0.05, 0.05)	0.002 (-0.01, 0.01)	-0.30 (-0.68, 0.08)
Gender (male is reference)	0.47 (-0.39, 1.52)	0.50 (-0.75, 1.74)	-0.29 (-0.58, -0.01)	-18.23 (-27.91, -8.55)
Body Mass Index (kg/m2) a	-0.06 (-0.15, 0.04)	-0.13 (-0.26, -0.00)	-0.02 (-0.05, 0.01)	0.52 (-0.51, 1.55)
Daily SOFA score (0–24) a	-0.13 (-0.24, -0.01)	-0.20 (-0.36, -0.04)	-0.01 (-0.04, 0.03)	-0.32 (-1.50, 0.86)
Session duration (min) a	0.01 (-0.04, 0.06)	-0.01 (-0.07, 0.05)	-0.0002 (-0.02, 0.01)	-0.28 (-0.86, 0.30)
Time from ICU admission to start of session (days) a	0.03 (-0.01, 0.07)	0.01 (-0.04, 0.06)	-0.003 (-0.02, 0.01)	0.08 (-0.38, 0.53)
Session type (exercise is reference)
p-value b	0.016	0.916	0.048	0.982
cycling	-0.53 (-1.73, 0.77)	-0.37 (-2.06, 1.33)	-0.04 (-0.39, 0.31)	-2.38 (-14.13, 9.37)
mobilisation	6.56 (2.14, 11.24)	-1.78 (-7.82, 4.27)	1.16 (-0.13, 2.45)	-12.17 (-73.47, 49.13)
respiratory management	0.55 (-1.42, 2.40)	-0.69 (-3.33, 1.94)	-0.81 (-1.66, 0.05)	-15.98 (-90.71, 58.75)
exercise and respiratory management	0.12 (-1.64, 1.67)	0.32 (-1.87, 2.52)	0.07 (-0.43, 0.57)	-4.17 (-21.07, 12.73)
complex cycling and mobilisation	-2.79 (-6.37, 0.96)	0.71 (-4.12, 5.53)	0.18 (-0.81, 1.17)	-11.64 (-44.25, 20.96)
complex exercise and mobilisation	4.52 (0.22, 9.07)	-0.35 (-6.19, 5.50)	0.18 (-1.05, 1.41)	-12.63 (-67.08, 41.82)
Treatment modality (passive is reference)
p-value b	0.167	0.178	0.181	0.095
mixed	-0.95 (-3.33, 1.30)	0.18 (-3.02, 3.37)	-0.43 (-1.10, 0.24)	-6.55 (-29.10, 15.99)
active	0.97 (-0.27, 2.07)	1.44 (-0.13, 3.00)	0.19 (-0.16, 0.55)	11.98 (-0.52, 24.48)
Mobilisation level (in-bed is reference)
p-value b	0.027	0.578	0.050	0.001
edge-of-bed	-5.85 (-10.43, -1.42)	-1.35 (-7.30, 4.61)	-0.39 (-1.67, 0.89)	13.41 (-44.79, 71.60)
out-of-bed	-6.06 (-11.05, -1.67)	-2.62 (-8.92, 3.68)	-1.18 (-2.51, 0.16)	-40.92 (-104.37, 22.52)
Airway support (none is reference)
p-value b	0.416	0.471	0.678	0.886
tracheostomy	0.11 (-1.48, 1.76)	0.50 (-1.73, 2.74)	0.30 (-0.42, 1.02)	4.88 (-25.36, 35.11)
endotracheal tube	0.87 (-0.66, 2.33)	1.16 (-0.86, 3.19)	0.18 (-0.51, 0.87)	6.34 (-22.03, 34.70)
Opiates on session day c	0.43 (-1.25, 2.08)	-0.06 (-2.56, 2.43)	0.35 (-0.28, 0.97)	-8.27 (-28.25, 11.72)
Vasoactive on session day c	0.26 (-0.78, 1.33)	-2.42 (-3.82, -1.01)	-0.28 (-0.59, 0.03)	-11.78 (-22.17, -1.39)
Sedatives on session day c	1.32 (0.00, 2.44)	1.03 (-0.64, 2.69)	0.24 (-0.18, 0.66)	10.86 (-5.72, 27.43)
Neuromuscular blocking agents on session day c	0.18 (-1.04, 1.45)	-0.08 (-1.73, 1.57)	-0.01 (-0.35, 0.34)	-10.12 (-21.78, 1.54)

a per one-unit increase (for continuous variables).

b likelihood test for overall significance of categorical variables.

c none is reference.

Reported effects of explanatory variables are mean differences of median values and need to be considered under the assumption of ‘all other covariates being constant’. All models are adjusted for measured values and CV before physiotherapy (estimates not shown). Examples for interpretation: (1) Categorical data: Median HR ‘after’ (recovery) significantly increased for the ‘session type’ mobilisation with 6.56pbm when compared to exercise. Co-occurrence of ‘mobilisation levels’ and ‘session types’ are reported in S2 File (S3 Table E3). (2) Continuous data: A one-point increase in ‘SOFA score’ significantly reduced median MAP during recovery by 0.2mmHg.

Safety analysis

There were a total of 27 sessions where either an adverse event (4 [0.6%]) or a discontinuation of rehabilitation (23 [3%]) occurred (Table 1). Events and participants’ characteristics are described in S3 File. Overall, there were too few adverse and discontinuation events to investigate their patterns or predictors. We did, however, investigate factors for clinically relevant variations along with the two outliers in Fig 1. Thereby, ‘mobilisation level’, ‘treatment modality’ and ‘session type’ were most likely to induce clinically relevant variations. For example, patients with an out-of-bed/edge-of-bed mobilisation had a 2.3 times higher chance (odds ratio 95%CI 0.4–13.4, p = 0.36) to demonstrate a clinically relevant 10% variations in ‘VO2 during’ and a 4.7 times higher chance (95%CI 0.46–48.42, p = 0.19) in ‘VO2 after’ compared to an in-bed mobilisation (S3 File).

Discussion

This study provides new insight into the cardiorespiratory response to early rehabilitation in mechanically ventilated, critically ill adults in a mixed ICU. The main findings are:

Physiological values had a large range of variation during rehabilitation and recovery within and across participants’ sessions. Clinically relevant variations occurred in a substantial number of patients, despite overall changes not being statistically significant in the whole group of patients.

Key modifiable explanatory variables for physiological changes were session type, treatment modality, session duration, mobilisation level and daily medication, while key non-modifiable explanatory variables were age, gender, BMI and the SOFA score on the day of physiotherapy. Specifically, clinicians should be aware that during rehabilitation cardiorespiratory parameters (HR, MAP, MV and VO2) increase when using active participation versus passive therapy and decrease (with the exception of HR) with each additional minute of rehabilitation. In contrast, HR remains elevated in the recovery phase after (multimodal) mobilisation when compared to exercise and drops following higher mobilisation levels when compared to in-bed rehabilitation.

The findings from our study enhance clinical reasoning and decision-making about the beginning, type, duration and intensity of physiotherapy-led rehabilitation by informing clinicians about the estimated effects on physiological values based on individual patient characteristics. For example, we found no evidence that the ‘time from ICU admission to the start of physiotherapy’ particularly affected physiological values. Instead ‘session type’, ‘treatment modality’ and ‘session duration’ seem to be the major drivers of cardiorespiratory changes and should be used to plan appropriate rehabilitation interventions. Given the good correlation between VO2 and MV, we recommend to monitor routine data such as HR, MAP, MV and SpO2 and to tailor training intensity as well as to ensure sufficient recovery.

The goal of rehabilitation is to optimise physical functioning in order to enhance autonomy and participation [24]. Physiological instability is considered a barrier to the safe implementation of early rehabilitation in the ICU [25]. Our partly contradictory results–no general effect, despite large variations–might mirror the challenge of providing safe rehabilitation within critical care. Therapists need to balance safety against a sufficient training stimulation. We found that a clinically relevant variation was achieved in 1 out of 4 sessions. The physiological reaction of critically ill patients can vary from day to day within and across participants as indicated by the high variation and large confidence intervals of physiological values. Early rehabilitation should therefore be closely monitored and individually tailored, whereby changes in physiological parameters might necessitate adaptions within a session. Clearly, early rehabilitation is not a ‘one size fits all’ approach but rather requires continuous clinical reasoning and interprofessional collaboration.

The finding that per additional minute of rehabilitation cardiorespiratory parameters decreased is interesting. A potential explanation is that shorter sessions were more intensive with less breaks in-between. Alternatively, repeated muscle activation might be limited in the critically ill leading to early onset of muscle fatigue and patient inactivity [26]. From a training perspective, shorter and more frequent sessions might therefore be preferred to achieve an adequate training response. This strategy was associated with improved 3-month outcomes after stroke [27] and should be investigated in future randomised controlled trials.

Our results further reveal differences in the cardiorespiratory response of males versus females, elderly patients (>63 years), who may not achieve a sufficient cardiorespiratory response to training anymore and thus may require longer time for recovery, or of patients with a higher SOFA score possibly due to bioenergetic dysfunction [17]. There are few studies specifically investigating the impact of ‘illness severity’ on the cardiorespiratory response to physiotherapy. One retrospective study (n = 23) examining early mobilisation in the elderly (>75 years, APACHE II: 27) did not find any adverse reactions in haemodynamic parameters [28]. Another trial investigating early cycling in adults with septic shock (n = 18) suggests that this intervention is safe (0.4% adverse events) and might preserve muscle cross-sectional areas, though authors do not report haemodynamic data to support this [29]. Finally, a recent study testing early graded, passive cycling in septic patients (n = 10, SOFA = 7.5) similarly found a high variation that ranged from improved to worsened left ventricular function between patients [30].

In our study, physiotherapy in patients with vasoactive support seemed safe, though drops in MAP were likely during both training and recovery. Rebel et al. [31] similarly found that mobilisation with vasoactive support is feasible and safe, but associated with a higher risk of hypotension. The risk of hypotension in our study seems higher in obese (BMI >27), sicker (SOFA ≥8) patients, with ‘vasoactive support’ or longer ‘session duration’ (>24min), but not with ‘session type’. Previous findings about passive exercise or cycling with or without vasoactive support not affecting haemodynamic parameters [12, 32], could therefore be due to their case-mix. Our results furthermore indicate that HR might slightly increase during and after rehabilitation in patients receiving sedatives. Clinicians need to be aware of such associations and adjust exercise to remain within their therapeutic targets. Still, early rehabilitation could detect readiness to wean sedatives–shortening time on mechanical ventilation–and is therefore recommended by international guidelines [33].

Finally, clinicians need to be aware that heart rate recovery may be prolonged in elderly patients (>63 years) or after mobilisation when compared to just exercise. This is substantiated by the study of Black et al. [17] who found prolonged recovery times (defined as return to 10% of baseline VO2) for 1 in 4 rehabilitation sessions. We also found >25% of clinically relevant variations during recovery in MAP, MV and VO2 with VO2 remaining slightly but not significantly elevated. It is important to note that in our data rehabilitation included the whole spectrum of interventions while Black et al. [17] specifically investigated active, out-of-bed activities in long-stayers. Our descriptive data might therefore underestimate the cardiorespiratory response to these activities and patient group, but seem more generalisable to mixed ICU patients across their whole ICU stay. Additionally, it allowed us to specifically analyse explanatory variables and their estimated effect. In this regard, our results support previous research that active exercises and out-of-bed mobilisations lead to stronger physiological reactions than passive exercises or in-bed mobilisations [11, 14–17, 34, 35]. Nevertheless, the cardiorespiratory response to ‘session duration’, ‘session type’ and ‘mobilisation level’ does not seem as straightforward due to considerable overlap (S2 File: S3 Table). For example, while session type ‘mobilisation’ was generally associated with increased HR in the recovery period, HR dropped substantially following an ‘out-of-bed’ mobilisation. This phenomenon might signify increased blood flow after returning to the supine position or might be a sign of exhaustion. Still, the incidence of adverse events remained very low (0.6% [5/784 physiotherapy sessions]), indicating that early rehabilitation in the critically ill, mechanically ventilated adult is safe.

This study has limitations. First, this analysis relied solely on physiological data and did not consider fatigue or other subjective measures to evaluate patients’ training load or recovery. This might be important because exhaustion remains a major barrier to patient participation [36] and might not be reflected by physiological data. Second, we cannot exclude potential measurement error as we relied on standard ICU monitoring, whereby median filtering should have reduced the risk of artefacts substantially [37]. Third, the original trial primarily aimed to assess the efficacy of early rehabilitation while monitoring the safety of the randomised interventions. The population was therefore highly selective. Additionally, the estimated effect of explanatory variables should be interpreted as hypothesis-generating and in the context of safety. Yet, to the best of our knowledge, this is the first analysis that explored various explanatory variables on a large critically ill sample across the whole ICU stay. These results therefore provide important information for future trials, but need to be validated in prospective studies. Fourth, while our results inform clinical decision-making on the intensity and duration of early rehabilitation, they cannot establish the effect on functional outcomes. Finally, the safety cut-off for a clinically relevant variation was chosen in absence of previous evidence. However, a cut-off of 20% still led to variations in 1 out of 10 sessions, particularly for MV and VO2 (S2 File: S2 Table). Our interventions were safe with only few, transient adverse events. Future trials should therefore investigate the feasibility and efficacy of different physiological training intensities as well as their association with neuromuscular activation and patients’ perceived rate of exertion.

Conclusions

Based on the physiological data from ‘before’, ‘during’ and ‘after’ 716 early physiotherapy-led rehabilitation sessions from 108 participants, our findings indicate that rehabilitation in the ICU is safe and does not negatively influence physiological parameters. Nevertheless, clinically relevant variations are common during training and the recovery period. Physiological parameters should therefore be closely monitored and exercise individually tailored. The explanatory variables identified guide clinicians’ decision-making in delivering the optimal type and intensity by enabling clinicians to estimate the cardiorespiratory response prospectively. Shorter sessions and active treatment therefore seem to increase the cardiorespiratory response during therapy, while sufficient time for recovery seems particularly necessary after a mobilisation session.

Detailed methodology.

(PDF)

Click here for additional data file.

Additional results including S1–S4 Tables and S2–S5 Figs.

(PDF)

Click here for additional data file.

Adverse events, therapy discontinuation and factors for clinically relevant variations.

(DOCX)

Click here for additional data file.

Codebook.

(XLSX)

Click here for additional data file.

Dataset.

(CSV)

Click here for additional data file.

Analysis code and regression output.

(PDF)

Click here for additional data file.

STROBE checklist.

(DOCX)

Click here for additional data file.

Flow diagram of rehabilitation sessions.

(TIF)

Click here for additional data file.

25 Oct 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Brenda M. Morrow, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Competing Interests section:

“We have read the journal's policy and the authors of this manuscript have the following competing interests: SE, GL, MLV, CHGB declare that they have no competing interests. II and AM are affiliated with CTU Bern, University of Bern, which has a staff policy of not accepting honoraria or consultancy fees. However, CTU Bern is involved in design, conduct, or analysis of clinical studies funded by not-for-profit and for-profit organizations. In particular, pharmaceutical and medical device companies provide direct funding to some of these studies. For an up-to-date list of CTU Bern’s conflicts of interest see http://www.ctu.unibe.ch/research/declaration_of_interest/index_eng.html. SMJ reports the following potential conflicts of interest: 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 Nestlé. The money was paid into a departmental fund; 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, Dräger Schweiz AG, Teleflex Medical GmbH. None of these companies had a role in this study.”

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear authors,

thank you very much for the opportunity to review the submitted manuscript. The authors conducted a secondary analysis of 108 critical care patients and analyzed the cardiorespiratory responses to rehabilitation exercises.

I am appreciating this work and have got some minor concerns.

SERIOUS CONCERNS

MINOR CONCERNS

Page 2, Line 30-33: wording: you used the terms “early rehabilitation”, “exercise”, “rehabilitation”, “training” and “recovery” as synonyms, I guess to improve readability. With a closer look, all those terms are different concepts, requiring different scientific approaches. Similar, I found in section introduction (p4, L105-110), that you used the term “early physiotherapy-led rehabilitation” here, but also “early mobilisation”, “exercise”, and “sessions”. At all, I suggest using one (1) term stringently.

P2, L37: would you please add the information of the seven different session types, with increasing activities. Please use a stringent reporting.

P2, Abstract: I am fan of clear structures, e.g. using always the same order of reported outcome parameters: MV, VO2, MAP. Within the sentences, this order changes. Would you please re-arrange the order, and keep it? See also lines 153-156.

P2, Abstract: I guess all results are mean differences? If so, please add this information.

P3, keywords: to improve your future citation, please check for pubmed’s MeSH terms (https://www.ncbi.nlm.nih.gov/mesh) for the used keywords, and schedule key words in alphabetically order.

L123: this was an RCT, but you analyse intervention- and control group as one single group? If so, please add some information about the intervention and an explanation why the summary is reasonable (if I remember right, the intervention group had a little bit more minutes of rehab each day, right?)

L125 “physiological data (n=35)” You don’t mean that you analyzed 35 data, but data of 35 patients, right?

L126: please insert a line break between “…physiotherapy [20].” and “The local…”

L130: please add “Population”, and a line break. I guess you mean “equal or older than 18 years”?

Section results is fine. The tables are hard to read, but I have no idea how to improve readability. The “examples for interpretation” are really helpful.

L316-319: would you please summarize the main results in their plain meaning, eg. “complex exercise and mobilisation led to increased VO2”, or else

Section discussion is fine and interesting.

L339, 346, 358, 369, 388: any new hypothesis or research questions?

Not part of the review and just a suggestion: You started very early, at day 2, and the muscle loss might not be severe in early days. I wonder if there are any differences in cardiorespiratory responses to PT’s rehab when delivered within early <7 days or late ≥7 days?

Very fine work, thank you.

Reviewer #2: Thank you for the opportunity to review this paper. The authors present the cardiorespiratory response in critically ill adult patients in response to various early rehabilitation interventions. This paper is well written and contributes novel data to the field, highlighting that early rehabilitation is safe.

The authors adequately identify limitations to the study. The heterogeneity in both the population and interventions and the impact on the conclusions is mentioned.

I have detailed a few minor points below:

1. It may be valuable to clarify/better define the "explanatory" variables. How were these decided on or chosen?

2. If understood correctly, the clinically relevant cutoff of >10% is related to safety; however further justification as to why this was chosen and not 20% is required. This will perhaps be important when weighing up safety versus adequate intensity to bring about physiological change. In the discussion it seems as if this clinically relevant response is linked to response to exercise and not safety – please clarify.

3. Stronger motivation is needed for not adjusting for multiple testing is needed.

4. In Table 1, please could you clarify what is meant by "Time from ICU admission to start of each session (days)"?

5. While recommendations are made for clinical practice in terms of monitoring and recovery, what may be more helpful to clinicians and advancing the cause of early rehabilitation in the ICU is know how these physiological responses link to various outcomes. As highlighted by the authors, there is a gap in terms of adequate prescription (FITT) in the critically ill patients. Such data sets, could help shed light on factors such as the frequency and intensity of rehabilitation interventions to effect change on outcomes.

Authors highlight these results enhance clinical decision making around frequency and intensity, however, I think without looking at outcomes – this enhancement is limited – it may be safe, but there is no way to know it is effective with the data provided.

Reviewer #3: This manuscript is a straightforward secondary analysis of data generated from a randomized controlled trial, investigating the effect of explanatory variables on physiological changes during training and recovery. While the study (and the analysis) looks timely, relevant, and on target, I state my thoughts below, primarily on the statistical analysis presented.

1. A multi-level logistic regression was proposed. However, it was not clear from the writeup (Statistical Analysis section) on how exactly the dichotomization of the response variable was considered, and how the multi-level part was handled? Was it via a generalized estimating equations, or via some mixed-effects models? A clear writeup is expected.

2. Tables 2 and 3 summarizes the results (estimated fixed effects of explanatory variables); it was not clear whether the table entries (estimates) are the log(oods-ratios), or something else? This needs to be made clear in the Table captions.

3. How about assessing the goodness-of-fit (GOF) after these multi-level logistic regression fits? While the Hosmer-Lemeshow is ideal for assessing the standard logistic regression fit, there exists various proposals for the multi-level setup; see below:

The authors may consider producing some summary GOF statistics in this context.

https://digscholarship.unco.edu/cgi/viewcontent.cgi?article=1243&context=dissertations

4. Interpretation of covariates/explanatory variables in the Results section should be considered in terms of increase/decrease in odds, with associated 95% confidence intervals.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Dr. Peter Nydahl

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

24 Nov 2021

Responses to Reviewers

The reviewers’ comments are shown in bold type, our responses in plain type and changed passages of the manuscript are shown in cursive type. All pages and lines are in reference to the unmarked manuscript version: titled “Manuscript”.

Reviewer #1

Dear authors,

thank you very much for the opportunity to review the submitted manuscript. The authors conducted a secondary analysis of 108 critical care patients and analyzed the cardiorespiratory responses to rehabilitation exercises.

I am appreciating this work and have got some minor concerns.

We thank the reviewer for taking the time to review our manuscript and are grateful for the constructive and valuable feedback. All comments have been answered below.

SERIOUS CONCERNS

MINOR CONCERNS

Page 2, Line 30-33: wording: you used the terms “early rehabilitation”, “exercise”, “rehabilitation”, “training” and “recovery” as synonyms, I guess to improve readability. With a closer look, all those terms are different concepts, requiring different scientific approaches. Similar, I found in section introduction (p4, L105-110), that you used the term “early physiotherapy-led rehabilitation” here, but also “early mobilisation”, “exercise”, and “sessions”. At all, I suggest using one (1) term stringently.

We agree with the reviewer that these terms include different concepts. To improve clarity, we reworded the abstract and introduction section using the term ‘rehabilitation’ as suggested. We took this even further and reworded, where appropriate, throughout the manuscript.

We did, however, keep the term ‘recovery’ because it also includes the period after rehabilitation which was an important aspect of our analysis.

The sections mentioned by the reviewer now read:

Abstract (page 2, lines 30-33):

“Early rehabilitation is indicated in critically ill adults to counter functional complications. However, the physiological response to rehabilitation is poorly understood. This study aimed to determine the cardiorespiratory response to rehabilitation and to investigate the effect of explanatory variables on physiological changes during rehabilitation and recovery.”

Introduction (page 4, lines 104-107):

“First, we aimed to analyse physiological variables from before, during and after rehabilitation to describe both the response to rehabilitation and to estimate recovery. Second, we aimed to characterise effects of a-priori selected explanatory variables on cardiorespiratory reactions from before to during rehabilitation and from before to after rehabilitation.”

P2, L37: would you please add the information of the seven different session types, with increasing activities. Please use a stringent reporting.

We are unfortunately limited by the allowed word count within the abstract (maximal 300 words, initially 292 words). We therefore cannot add all explanatory variables. Moreover, to add only ‘session type’ would unduly elevate the variable above the others. Our hypotheses about the main drivers of the cardiorespiratory response to early rehabilitation also included mobilisation level, exercise modality and duration. These explanatory variables should therefore be considered as important as session type. It is also important to note, that the seven session types are not reflecting increasing activities. They are based on commonly used, combined interventions in clinical practice. For example, early mobilisation is often combined with respiratory management, because sitting on the edge of bed might help to clear secretions or because deep-breathing exercises might calm patients after an effort to sit up. More detailed information on the seven session types is provided within the supplement (S1 File).

However, we concede that the term rehabilitation is used differently across the world and that some clarity of our concept might be needed. While our manuscript includes a definition of rehabilitation (page 17, lines 348-349), we agree that the concept must be clear within the abstract. We therefore slightly changed the suggested sentence to include the pragmatic study design and some type of interventions within the abstract (page 2, lines 37-38):

“…, we analysed the 716 physiotherapy-led, pragmatic rehabilitation sessions (including exercise, cycling and mobilisation).”

P2, Abstract: I am fan of clear structures, e.g. using always the same order of reported outcome parameters: MV, VO2, MAP. Within the sentences, this order changes. Would you please re-arrange the order, and keep it? See also lines 153-156.

We have amended the order throughout the manuscript to ‘HR, MAP, MV, SpO2 and VO2’ when it made sense to do so. For example, we did not change the order, when the magnitude of the change was discussed (e.g., page 10, lines 246-247), but changed tables 2 and 3 to the same order. When one variable was not discussed we left this variable out, for example, we wrote ‘HR, MV and VO2’.

P2, Abstract: I guess all results are mean differences? If so, please add this information.

Indeed these are mean differences, we added this description to the results in the abstract (page 2, lines 49-51):

“Active patient participation increased MV (mean difference 0.7l/min [0.4-1.0, p<0.001]) and VO2 (23ml/min [95%CI: 13-34, p<0.001]) during training when compared to passive participation.”

P3, keywords: to improve your future citation, please check for pubmed’s MeSH terms (https://www.ncbi.nlm.nih.gov/mesh) for the used keywords, and schedule key words in alphabetically order.

We re-ordered keywords alphabetically. Five out of the nine keywords are MeSH-terms:

1. Critical Illness [C23.550.291.625]

2. Early ambulation (E02.760.169.063.500.335, E02.831.335)

3. Exercise [I03.350]

4. Oxygen Consumption [G03.680]

5. Physical Therapy Specialty [H02.010.625]

We further added the MeSH-term: Rehabilitation [E02.831], because this term contains the concept of recovery. We kept the terms physiotherapy; physiological reaction; recovery and training response because they are commonly used synonyms and thus complement the MeSH-terms. The manuscript has been changed as follows:

Page 3, lines 71-72:

“critical illness; early ambulation; exercise, oxygen consumption; physical therapy specialty; physiological reaction; physiotherapy; recovery; rehabilitation; training response”

L123: this was an RCT, but you analyse intervention- and control group as one single group? If so, please add some information about the intervention and an explanation why the summary is reasonable (if I remember right, the intervention group had a little bit more minutes of rehab each day, right?)

Yes, the reviewer’s assumption is correct. Our reasoning is as follows: This is a secondary analysis of an RCT that did not show any significant or clinically relevant differences in the prespecified primary and secondary outcomes. The two groups’ intervention did therefore not affect outcomes. In consequence the two treatments can be considered equal, which allowed us to combine both groups into one population to answer our research questions. The current dataset therefore includes the whole original population using data from all trial participants who had at least one rehabilitation session.

We slightly revised the manuscript to clarify this part (page 5, lines 121-125)

“No significant differences were found in the primary or secondary outcomes with the exception of improved mental health six months after hospital discharge for the experimental group [19]. Consequently, this secondary analysis considered the two randomised groups as one population using data from all trial participants with at least one rehabilitation session.

Nevertheless, we did not ignore the allocated intervention within our analysis. Again, the reviewer is correct that the intervention group had significantly longer rehabilitation sessions. Also, only the intervention group cycled. However, these factors are accounted for within our analysis by the explanatory variables ‘session duration’ and ‘session type’. We decided a-priori which explanatory variables to include within our model. These variables were based on clinical reasoning and earlier research to limit confounding (also see revised lines 196-197 based on comments of reviewer #2). During this process we considered a variable ‘randomisation group’. However, we discarded it to avoid multicollinearity within the model and because we were rather interested on how the variables themselves affected the cardiorespiratory response (e.g. ‘session duration’).

Nevertheless, based on the reviewer’s comments, we performed a sensititvity analysis that includes the variable ‘randomisation group’ in our model. Compared to our main analysis, the estimates barely changed for MAP, MV and VO2. While we found a significant effect for ‘randomisation group’ for HR (‘during’: mean difference 1.47bpm (0.17 to 2.78), p=0.029; ‘after’: 2.1bpm (0.79 to 3.41), p=0.002), the estimates for our chosen explanatory variables changed only slightly when compared to our main analyses. Most importantly, our results and conclusions do not change but rather reinforce our earlier reasoning to exclude this variable. We therefore limit the results of this sensitivity analysis to this letter, but did not add them to the supplementary material.

Outcome Explanatory variable Mean differences Lower 95% CI Upper 95% CI p-value

HR during

Age (years) 0.02 -0.02 0.06 0.329

Gender (male is reference) 0.21 -0.9 1.32 0.706

Body Mass Index (kg/m2) -0.01 -0.11 0.1 0.908

Randomisation group: intervention 1.47 0.17 2.78 0.029

Daily SOFA score (0-24) 0 -0.12 0.12 0.959

Session duration (min) 0.05 0.01 0.09 0.009

Time from ICU admission to start of session (days) 0.03 -0.01 0.07 0.179

Session type: cycling -1.66 -2.99 -0.33 0.015

Session type: mobilisation 2.04 -1.88 5.95 0.309

Session type: respiratory management 0.58 -1.11 2.27 0.501

Session type: exercise and respiratory management 0.39 -1.06 1.84 0.597

Session type: complex cycling and mobilisation -3.22 -6.45 0.02 0.052

Session type: complex exercise and mobilisation 2.35 -1.45 6.15 0.226

Treatment modality: mixed -0.45 -2.45 1.56 0.661

Treatment modality: active 1.18 0.13 2.23 0.028

Mobilisation level: edge-of-bed -0.98 -4.85 2.89 0.621

Mobilisation level: out-of-bed 0.08 -3.98 4.14 0.97

Airway support: tracheostomy -1.02 -2.52 0.49 0.186

Airway support: tube -0.53 -1.84 0.79 0.433

Opiates on session day 0.27 -1.21 1.75 0.719

Vasoactive on session day -0.2 -1.15 0.74 0.676

Sedatives on session day 1.28 0.19 2.37 0.021

Neuromuscular blocking agents on session day -0.21 -1.3 0.88 0.705

HR after

Age (years) 0.05 0.01 0.09 0.012

Gender (male is reference) 0.49 -0.54 1.52 0.36

Body Mass Index (kg/m2) -0.09 -0.19 0.01 0.084

Randomisation group: intervention 2.1 0.79 3.41 0.002

Daily SOFA score (0-24) -0.13 -0.26 -0.01 0.035

Session duration (min) 0.01 -0.04 0.06 0.686

Time from ICU admission to start of session (days) 0.04 0 0.08 0.046

Session type: cycling -1.88 -3.42 -0.34 0.017

Session type: mobilisation 6.66 2.04 11.27 0.005

Session type: respiratory management 0.27 -1.68 2.22 0.785

Session type: exercise and respiratory management 0.28 -1.41 1.97 0.745

Session type: complex cycling and mobilisation -3.99 -7.79 -0.19 0.04

Session type: complex exercise and mobilisation 4.82 0.34 9.31 0.035

Treatment modality: mixed -1.03 -3.38 1.32 0.391

Treatment modality: active 1.02 -0.18 2.22 0.097

Mobilisation level: edge-of-bed -6.34 -10.91 -1.77 0.007

Mobilisation level: out-of-bed -6.21 -10.98 -1.45 0.011

Airway support: tracheostomy -0.17 -1.85 1.51 0.843

Airway support: tube 0.81 -0.71 2.33 0.299

Opiates on session day 0.29 -1.41 1.99 0.737

Vasoactive on session day 0.07 -1.01 1.15 0.899

Sedatives on session day 1.63 0.37 2.89 0.012

Neuromuscular blocking agents on session day 0.09 -1.17 1.36 0.886

MAP during

Age (years) 0.01 -0.03 0.05 0.615

Gender (male is reference) 1.02 0 2.04 0.051

Body Mass Index (kg/m2) 0.03 -0.08 0.14 0.575

Randomisation group: intervention 0.28 -1.07 1.64 0.683

Daily SOFA score (0-24) -0.02 -0.15 0.11 0.737

Session duration (min) -0.06 -0.12 -0.01 0.017

Time from ICU admission to start of session (days) 0.03 -0.02 0.07 0.242

Session type: cycling 0.26 -1.4 1.92 0.761

Session type: mobilisation 0.29 -4.66 5.25 0.907

Session type: respiratory management 1.78 -0.4 3.96 0.109

Session type: exercise and respiratory management 0.27 -1.54 2.07 0.772

Session type: complex cycling and mobilisation 1.8 -2.26 5.86 0.386

Session type: complex exercise and mobilisation 1.31 -3.48 6.11 0.592

Treatment modality: mixed 2.87 0.25 5.49 0.032

Treatment modality: active 1.42 0.14 2.7 0.03

Mobilisation level: edge-of-bed 0.8 -4.1 5.7 0.749

Mobilisation level: out-of-bed 0.79 -4.38 5.96 0.764

Airway support: tracheostomy 0.9 -0.94 2.74 0.338

Airway support: tube 1.14 -0.53 2.8 0.181

Opiates on session day 0.6 -1.42 2.62 0.561

Vasoactive on session day -1.82 -2.98 -0.66 0.002

Sedatives on session day 1.38 -0.02 2.77 0.054

Neuromuscular blocking agents on session day -0.39 -1.75 0.96 0.568

MAP after

Age (years) 0 -0.05 0.05 0.977

Gender (male is reference) 0.5 -0.75 1.74 0.432

Body Mass Index (kg/m2) -0.13 -0.26 0 0.054

Randomisation group: intervention -0.01 -1.66 1.64 0.99

Daily SOFA score (0-24) -0.2 -0.36 -0.04 0.013

Session duration (min) -0.01 -0.07 0.05 0.735

Time from ICU admission to start of session (days) 0.01 -0.05 0.06 0.727

Session type: cycling -0.36 -2.38 1.67 0.728

Session type: mobilisation -1.78 -7.81 4.26 0.564

Session type: respiratory management -0.69 -3.34 1.96 0.61

Session type: exercise and respiratory management 0.32 -1.87 2.52 0.773

Session type: complex cycling and mobilisation 0.71 -4.23 5.65 0.778

Session type: complex exercise and mobilisation -0.35 -6.19 5.49 0.907

Treatment modality: mixed 0.18 -3.02 3.37 0.914

Treatment modality: active 1.44 -0.13 3 0.072

Mobilisation level: edge-of-bed -1.35 -7.31 4.62 0.658

Mobilisation level: out-of-bed -2.62 -8.91 3.68 0.415

Airway support: tracheostomy 0.51 -1.74 2.75 0.659

Airway support: tube 1.16 -0.86 3.19 0.261

Opiates on session day -0.06 -2.56 2.43 0.96

Vasoactive on session day -2.42 -3.83 -1 0.001

Sedatives on session day 1.02 -0.68 2.73 0.239

Neuromuscular blocking agents on session day -0.08 -1.73 1.57 0.926

MV during

Age (years) 0 -0.01 0.01 0.823

Gender (male is reference) -0.43 -0.74 -0.12 0.008

Body Mass Index (kg/m2) -0.01 -0.04 0.02 0.366

Randomisation group: intervention 0.31 -0.08 0.7 0.12

Daily SOFA score (0-24) 0.01 -0.02 0.05 0.429

Session duration (min) -0.02 -0.03 0 0.011

Time from ICU admission to start of session (days) 0 -0.01 0.02 0.745

Session type: cycling -0.02 -0.42 0.37 0.912

Session type: mobilisation 0.8 -0.38 1.97 0.184

Session type: respiratory management 0.31 -0.45 1.07 0.419

Session type: exercise and respiratory management 0.03 -0.43 0.49 0.892

Session type: complex cycling and mobilisation 0.01 -0.93 0.95 0.979

Session type: complex exercise and mobilisation 0.29 -0.84 1.41 0.618

Treatment modality: mixed 0.51 -0.1 1.12 0.104

Treatment modality: active 0.73 0.4 1.06 0

Mobilisation level: edge-of-bed 0.68 -0.49 1.86 0.256

Mobilisation level: out-of-bed 0.37 -0.86 1.59 0.556

Airway support: tracheostomy 0.29 -0.36 0.95 0.378

Airway support: tube 0.08 -0.54 0.7 0.802

Opiates on session day 0.09 -0.48 0.66 0.753

Vasoactive on session day -0.15 -0.44 0.13 0.3

Sedatives on session day 0.09 -0.3 0.47 0.668

Neuromuscular blocking agents on session day -0.22 -0.54 0.1 0.18

MV after

Age (years) 0 -0.01 0.01 0.653

Gender (male is reference) -0.3 -0.58 -0.01 0.043

Body Mass Index (kg/m2) -0.02 -0.05 0.01 0.2

Randomisation group: intervention 0.18 -0.2 0.55 0.355

Daily SOFA score (0-24) -0.01 -0.04 0.03 0.6

Session duration (min) 0 -0.02 0.01 0.988

Time from ICU admission to start of session (days) 0 -0.02 0.01 0.78

Session type: cycling -0.16 -0.58 0.27 0.475

Session type: mobilisation 1.18 -0.11 2.47 0.073

Session type: respiratory management -0.87 -1.73 -0.01 0.048

Session type: exercise and respiratory management 0.1 -0.4 0.6 0.7

Session type: complex cycling and mobilisation 0.06 -0.96 1.08 0.911

Session type: complex exercise and mobilisation 0.22 -1.01 1.45 0.723

Treatment modality: mixed -0.44 -1.1 0.23 0.199

Treatment modality: active 0.2 -0.16 0.56 0.273

Mobilisation level: edge-of-bed -0.44 -1.73 0.84 0.497

Mobilisation level: out-of-bed -1.18 -2.51 0.15 0.083

Airway support: tracheostomy 0.29 -0.43 1.01 0.427

Airway support: tube 0.21 -0.48 0.9 0.554

Opiates on session day 0.35 -0.27 0.97 0.27

Vasoactive on session day -0.3 -0.61 0.01 0.061

Sedatives on session day 0.28 -0.15 0.71 0.203

Neuromuscular blocking agents on session day -0.02 -0.37 0.33 0.909

VO2 during

Age (years) -0.54 -0.88 -0.2 0.002

Gender (male is reference) -23.7 -32.62 -14.78 0

Body Mass Index (kg/m2) 0.84 -0.1 1.78 0.081

Randomisation group: intervention 7.33 -6.46 21.12 0.299

Daily SOFA score (0-24) 0.25 -0.83 1.32 0.654

Session duration (min) -0.59 -1.08 -0.1 0.02

Time from ICU admission to start of session (days) -0.16 -0.57 0.25 0.451

Session type: cycling -1.97 -16.99 13.05 0.798

Session type: mobilisation 37.81 -16.1 91.72 0.17

Session type: respiratory management 2.6 -64.05 69.24 0.939

Session type: exercise and respiratory management 0.99 -14.03 16.01 0.897

Session type: complex cycling and mobilisation -12.78 -43.55 17.99 0.416

Session type: complex exercise and mobilisation 20.3 -28.18 68.79 0.412

Treatment modality: mixed 0.32 -19.89 20.53 0.975

Treatment modality: active 22.89 11.75 34.02 0

Mobilisation level: edge-of-bed -3.19 -54.93 48.55 0.904

Mobilisation level: out-of-bed 7.99 -47.98 63.97 0.78

Airway support: tracheostomy 2.54 -24.57 29.64 0.854

Airway support: tube 2.58 -22.98 28.14 0.843

Opiates on session day -7.66 -25.72 10.4 0.407

Vasoactive on session day -7.39 -16.76 1.99 0.124

Sedatives on session day 5.83 -8.8 20.47 0.435

Neuromuscular blocking agents on session day -7.15 -17.59 3.29 0.18

VO2 after

Age (years) -0.3 -0.68 0.08 0.121

Gender (male is reference) -18.26 -27.94 -8.59 0

Body Mass Index (kg/m2) 0.52 -0.5 1.55 0.318

Randomisation group: intervention -0.72 -15.78 14.33 0.925

Daily SOFA score (0-24) -0.32 -1.5 0.86 0.599

Session duration (min) -0.28 -0.86 0.3 0.342

Time from ICU admission to start of session (days) 0.07 -0.38 0.53 0.753

Session type: cycling -1.81 -18.55 14.94 0.833

Session type: mobilisation -12.14 -73.29 49 0.697

Session type: respiratory management -16.09 -90.67 58.48 0.673

Session type: exercise and respiratory management -4.26 -21.21 12.7 0.623

Session type: complex cycling and mobilisation -11.11 -45.52 23.31 0.528

Session type: complex exercise and mobilisation -12.72 -67.06 41.62 0.647

Treatment modality: mixed -6.46 -29.04 16.12 0.575

Treatment modality: active 12 -0.48 24.48 0.06

Mobilisation level: edge-of-bed 13.5 -44.58 71.58 0.649

Mobilisation level: out-of-bed -40.96 -104.24 22.33 0.206

Airway support: tracheostomy 4.77 -25.47 35.01 0.757

Airway support: tube 6.16 -22.37 34.69 0.672

Opiates on session day -8.27 -28.2 11.67 0.417

Vasoactive on session day -11.72 -22.17 -1.28 0.029

Sedatives on session day 10.77 -5.85 27.4 0.205

Neuromuscular blocking agents on session day -10.08 -21.74 1.58 0.091

L125 “physiological data (n=35)” You don’t mean that you analyzed 35 data, but data of 35 patients, right?

This is correct. We revised this to (page 5, lines 125-127):

“A preliminary safety analysis of the physiological data of the first 35 subjects indicated a moderately increased workload with increased heart rate and oxygen consumption but stable oxygen saturation from before to during rehabilitation [20].”

L126: please insert a line break between “…physiotherapy [20].” and “The local…”

This has been implemented as suggested.

L130: please add “Population”, and a line break. I guess you mean “equal or older than 18 years”?

We added a new heading ‘Population’ (line 132). Accordingly, we deleted ‘population’ from the heading ‘study design’. We further revised the sentence as follows (page 5, lines 133-134):

“Participants were ≥ 18 years old, functionally independent before ICU admission and expected to remain ventilated for ≥ 72 hours.”

Section results is fine. The tables are hard to read, but I have no idea how to improve readability. The “examples for interpretation” are really helpful.

As described previously, we changed the order of the column headings to HR, MAP, MV and VO2 to give a clear structure throughout the manuscript. We did not perform any other changes as readability should increase with article formatting.

L316-319: would you please summarize the main results in their plain meaning, eg. “complex exercise and mobilisation led to increased VO2”, or else

We concede that tables 2 and 3 are complex. We believe that the examples for interpretation and the highlighted relevant 95% confidence intervals will help readers to interpret results. To further increase understanding, we give a full summary here (based on 95% confidence intervals) along with a brief summary within the manuscript.

Full summary for ‘during rehabilitation’

• HR: increased by ‘session duration’ (per each additional minute of therapy), active participation (when compared to passive therapy), and by sedatives (when compared to none).

• MAP: increased by female gender (when compared to male), mixed or active participation (when compared to passive therapy), decreased by vasoactive drugs (when compared to none) and ‘session duration’ (per each additional minute of therapy)

• MV: increased by active participation (when compared to passive therapy), decreased by female gender (when compared to male) and by ‘session duration’ (per each additional minute of therapy)

• VO2: increased by active participation (when compared to passive therapy), decreased by age (per each additional year of age), female gender (when compared to male) and ‘session duration’ (per each additional minute of therapy)

Full summary for ‘after rehabilitation’

• HR: increased by age (per each additional year of age), session types ‘mobilisation’ and ‘complex exercise and mobilisation’ (when compared to ‘exercise) and by sedatives (when compared to none), decreased by mobilisation level ‘edge-of-bed’ and ‘out-of-bed’ (when compared to ‘in-bed) and by daily SOFA score (per each additional point)

• MAP: decreased by BMI (per each additional unit), daily SOFA score (per each additional point) and by vasoactive drugs (when compared to none)

• MV: decreased by female gender (when compared to male)

• VO2: increased by mobilisation level ‘edge-of-bed’ (when compared to ‘out-of-bed’), decreased by female gender (when compared to male) and by vasoactive drugs (when compared to none)

We newly summarised this in the manuscript as follows (page 16, lines 333-338):

“Specifically, clinicians should be aware that during rehabilitation cardiorespiratory parameters (HR, MAP, MV and VO2) increase when using active participation versus passive therapy and decrease (with the exception of HR) with each additional minute of rehabilitation. In contrast, HR remains elevated in the recovery phase after (multimodal) mobilisation when compared to exercise and drops following higher mobilisation levels when compared to in-bed rehabilitation.”

Additionally, a summary is available within the results section (page 11, lines 275-279):

“Shorter ‘session duration’ and ‘active treatment modality’ generally increased physiological parameters during rehabilitation, while cardiorespiratory parameters did not return back to baseline for ‘session type’ and ‘mobilisation level’ during the prespecified 15-min recovery-phase. Explanatory variables with an effect on SpO2 were ‘mobilisation level’ and ‘airway support’ (S2 File: S4 Table).”

Section discussion is fine and interesting.

L339, 346, 358, 369, 388: any new hypothesis or research questions?

Yes, there are several research questions arising from our analysis. Most importantly, we found that ‘treatment modality’ and ‘session duration’ were the major drivers of cardiorespiratory changes during early rehabilitation. However, a sufficient cardiorespiratory response does not automatically translate into an adequate neuromuscular response or functional benefits. Consequently, future trials should investigate if any of these factors can improve patient-centred, functional outcomes. For example, shorter sessions were associated with an increased cardiorespiratory response. We propose to investigate whether shorter, more frequent sessions improve functional outcomes, when compared to only one session per day.

We revised the manuscript to include this more clearly (page 17, lines 363-365):

“This strategy was associated with improved 3-month outcomes after stroke [27] and should be investigated in future randomised controlled trials.”

Patient participation (treatment modality) is dependent upon sedation but also on patients’ motivation or their perceived rate of exertion. There are a few studies who specifically investigate the impact of sedation on patient participation and functional outcomes [1], but patient fatigue is often a limiting factor [2]. A qualitative study that explores enabling factors and barriers to being active while critically ill might give insights into how future clinical trials should be shaped to achieve intensity-targets.

Additionally, we suggest to perform prospective studies that explore the optimal target for cardiorespiratory parameters and how this relates with patients’ perceived rate of exertion as well as neuromuscular activation (proof-of-concept study). For example, we chose a conservative cut-off of 10% for our analysis. However, this might not be sufficient to elicit a neuromuscular benefit. We would therefore propose to investigate the feasibility of different physiological targets (e.g., 10% versus 20% versus 30%) preferably using oxygen consumption. (Rationale: oxygen consumption seems the most important variable, when talking about exercise that is defined as a planned, structured, repetitive bodily movement produced by skeletal muscles that results in energy expenditure and aims to improve or maintain one or more components of physical fitness [3]). Research questions would include:

• Can patients achieve and sustain these intensities?

• Are higher intensities associated with improved muscle activation (for example when measured with electromyography)?

• If these physiological targets prove feasible (and remain safe as within our study), what is the effect of these different targets on functional outcomes?

We revised the manuscript to express some of these thoughts (page 19, lines 423-425):

“Future trials should therefore investigate the feasibility and efficacy of different physiological training intensities as well as their association with neuromuscular activation and patients’ perceived rate of exertion.”

Finally, our findings should be validated in a prospective study with prospectively planned hypotheses (also see revision based on comments of reviewer #2: lines 418-419).

Not part of the review and just a suggestion: You started very early, at day 2, and the muscle loss might not be severe in early days. I wonder if there are any differences in cardiorespiratory responses to PT’s rehab when delivered within early <7 days or late ≥7 days?

Our analysis accounted the time from ICU admission to each session (median of 9 days with an interquartile range from 4 to 20 days, which includes the 7 day difference). The estimated effect on cardiorespiratory parameters was very minimal, clinically irrelevant and not statistically significant [HR: 0.02 (-0.01, 0.06); MAP: 0.03 (-0.02, 0.07); MV: 0.001 (-0.01, 0.01), VO2: -0.18 (-0.59, 0.20)]. Accordingly, we concluded that an early or late session did not affect cardiorespiratory parameters (page 17, lines 341-343).

However, cardiorespiratory parameters are a poor surrogate measure for muscle mass, thus to infer about potential neuromuscular effects, a study using electromyography might be necessary.

Very fine work, thank you.

Reviewer #2

Thank you for the opportunity to review this paper. The authors present the cardiorespiratory response in critically ill adult patients in response to various early rehabilitation interventions. This paper is well written and contributes novel data to the field, highlighting that early rehabilitation is safe.

The authors adequately identify limitations to the study. The heterogeneity in both the population and interventions and the impact on the conclusions is mentioned.

We thank the reviewer for taking the time to review our manuscript and are grateful for the constructive and valuable feedback. All comments have been answered below.

I have detailed a few minor points below:

1. It may be valuable to clarify/better define the "explanatory" variables. How were these decided on or chosen?

We agree, that this is a highly relevant point. As mentioned within the comments of Reviewer #1, we extensively discussed parameters within the research team and weighted them against the previous evidence. We modified the manuscript accordingly (page 7, lines 196-197):

“Explanatory variables were prospectively determined using extensive clinical reasoning and previous evidence to account for confounders. They included…”

2. If understood correctly, the clinically relevant cutoff of >10% is related to safety; however further justification as to why this was chosen and not 20% is required. This will perhaps be important when weighing up safety versus adequate intensity to bring about physiological change. In the discussion it seems as if this clinically relevant response is linked to response to exercise and not safety – please clarify.

Indeed, we chose the 10% cut-off based on safety recommendations that generally consider an increase of >20% as an adverse event. To the best of our knowledge, there is currently no guidance on the dose and safety of any training intensity available in the literature. We therefore chose the more conservative cut-off that was considered safe in previous studies. This is clearly a limitation, that we acknowledge within the manuscript.

As the reviewer mentions, 20% might be a more adequate intensity to elicit a neuromuscular response potentially improving functional outcomes. To this end, we performed a sensitivity analysis using a 20% cut-off (S2 File: S2 Table), whereby 1 out of 10 sessions was affected. Adverse events within our trial were not based on fixed cut-offs, but rather on individually set limits by the treating physician. Overall, we reported only few, transient adverse events. Accordingly, we believe that a 20%-threshold might be safe and should be considered in future trials as a potential training intensity. We further included this in the manuscript which has been revised to:

Page 19, lines 421-425:

“Finally, the cut-off for a clinically relevant variation was chosen in absence of previous evidence. However, a cut-off of 20% still led to variations in 1 out of 10 sessions, particularly for MV and VO2 (S2 File: S2 Table). Our interventions were safe with only few, transient adverse events. Future trials should therefore investigate the feasibility and efficacy of different physiological training intensities as well as their association with neuromuscular activation and patients’ perceived rate of exertion.”

Additionally, we slightly revised the following passages to clarify that this cut-off was primarily about safety:

Page 8, lines 214-215:

“Thus, a 10% threshold might be a safe cardiorespiratory training intensity in the critically ill adult.”

Page 17, lines 352-353:

“We found that a clinically relevant variation was achieved in 1 out of 4 sessions.”

Additionally, it is important to note that the main aim of this study was to describe and explore the physiological response (changes) by early rehabilitation. The response to exercise is therefore an important topic within our discussion. To avoid any confusion between terms, we further substituted ‘clinically relevant change’ to ‘clinically relevant variation’ throughout the manuscript.

Finally, we incorporated subheadings that align with the results section to clarify that the 10% cut-off belongs to the safety analysis. The slightly revised section about the cut-off definition has been moved from the last paragraph of the heading ‘Data collection and measurements’ to the heading ‘statistical analysis’ (page 8, lines 208-215):

“Safety analysis

We used a mixed-effects logistic regression model which accounted for correlation of within-individual measurements to investigate explanatory variables related to clinically relevant variations and report odds ratios with 95% confidence intervals (CI). A clinically relevant variation in physiological measurements was defined as >10% variation– based on half of what is commonly reported as an adverse event (>20%) in the literature [22]. Exercise is structured and repetitive physical activity that results in energy expenditure [23]. Thus, a 10% threshold might be a safe cardiorespiratory training intensity in the critically ill adult.”

3. Stronger motivation is needed for not adjusting for multiple testing is needed.

We concede that the chance of false-positive findings is increased without adjustment for multiple testing. However, this is a secondary, hypothesis-generating analysis and not a primary hypothesis-testing study (null versus alternative hypothesis). Adjustments for multiple testing might therefore discard potentially useful observations [4, 5]. We therefore do not think that multiple testing is appropriate for our type of analysis. However, we concede that we need to clearly state this limitation to readers.

We therefore revised the manuscript as follows:

Page 8, lines 216-217:

“Considering the exploratory, hypothesis-generating purpose of our secondary analysis, the significance threshold was set to 0.05 without adjustment for multiple testing.”

Page 19, lines 415-416:

“Additionally, the estimated effect of explanatory variables should be interpreted as hypothesis-generating and in the context of safety.”

Page 19, lines 418-419:

“These results therefore provide important information for future trials, but need to be validated in prospective studies.”

4. In Table 1, please could you clarify what is meant by "Time from ICU admission to start of each session (days)"?

We investigated 716 rehabilitation sessions. The 108 participants had a median of 3 [2-8] sessions (line 226). Accordingly, the time to each session varied in respect to ICU admission. For example, the first session could have been on the first day after admission, the second on the second day and the third session on the fifth day. This variable accounts for these different time periods.

We renamed this variable to “Time from ICU admission to start of individual session (days)” and added a footnote to better explain the variable (Page 10, lines 234-235):

5. While recommendations are made for clinical practice in terms of monitoring and recovery, what may be more helpful to clinicians and advancing the cause of early rehabilitation in the ICU is know how these physiological responses link to various outcomes. As highlighted by the authors, there is a gap in terms of adequate prescription (FITT) in the critically ill patients. Such data sets, could help shed light on factors such as the frequency and intensity of rehabilitation interventions to effect change on outcomes.

Authors highlight these results enhance clinical decision making around frequency and intensity, however, I think without looking at outcomes – this enhancement is limited – it may be safe, but there is no way to know it is effective with the data provided.

We agree with the reviewer that studies linking physiological responses to functional outcomes are needed. However, before this research can be conducted safety should be established. Additionally, we would need proof-of-concept studies, for example, is a higher intensity associated with a better neuromuscular response? However, our data-set was primarily obtained to answer the question about the efficacy of two different interventions. This analysis would not be in line with our initial research questions and conclusions. We therefore advocate for future prospective studies. We integrated this within the manuscript (page 19, lines 423-425):

“Future trials should therefore investigate the feasibility and efficacy of different physiological training intensities as well as their association with neuromuscular activation and patients’ perceived rate of exertion.”

We further added this limitation to the manuscript (page 19, lines 419-420):

“Fourth, while our results inform clinical decision-making on the intensity and duration of early rehabilitation, they cannot establish the effect on functional outcomes.”

Reviewer #3

This manuscript is a straightforward secondary analysis of data generated from a randomized controlled trial, investigating the effect of explanatory variables on physiological changes during training and recovery. While the study (and the analysis) looks timely, relevant, and on target, I state my thoughts below, primarily on the statistical analysis presented.

We thank the reviewer for taking the time to review our manuscript and are grateful for the constructive and valuable feedback. All comments have been answered below.

1. A multi-level logistic regression was proposed. However, it was not clear from the writeup (Statistical Analysis section) on how exactly the dichotomization of the response variable was considered, and how the multi-level part was handled? Was it via a generalized estimating equations, or via some mixed-effects models? A clear writeup is expected.

We apologise for the imprecise formulation of our multiple analyses. To clarify, we introduced three sub-headings within the statistical analysis chapter (analogous to the sub-headings within the results section):

• Descriptive analysis (line 186)

• Cardiorespiratory response (line 191)

• Safety analysis (line 208)

Our primary analysis (cardiorespiratory response) used an ANCOVA approach with Gaussian mixed-effects models which accounted for the fact that the measurements within the same individual are correlated. We changed the statistical section to (page 7, lines 192-195):

“The impact of explanatory variables on physiological values during (training) and after (recovery) rehabilitation was investigated in respect to before the physiotherapy session and variability (CV) with Gaussian mixed-effects models which accounted for correlation of within-individual measurements (S1 File) based on Vickers et al. [21].”

Safety analyses were also based on a mixed-effects (we changed the wording multilevel to mixed-effects for consistency) regression model approach for a binary outcome. We corrected the statistical section to (page 8, lines 209-210):

“We used a mixed-effects logistic regression model which accounted for correlation of within-individual measurements to investigate…”

2. Tables 2 and 3 summarizes the results (estimated fixed effects of explanatory variables); it was not clear whether the table entries (estimates) are the log(oods-ratios), or something else? This needs to be made clear in the Table captions.

Again, we apologize for the unclear write-up of our statistical analysis section that we revised to include sub-headings.

Table 2 and 3 report the fixed effect estimates from Gaussian mixed-effect models (e.g. from the cardiorespiratory response). Accordingly, the estimates in Table 2 and 3 are mean differences of median values (i.e. median values of 2 min recorded values for each session). We changed the legend caption to clarify this (page 13, line 287 and page 15, line 301):

“Legend: Reported effects of explanatory variables are mean differences of median values and…”

The safety analys which used mixed-effect logistic regression models reports odds ratio (see S3 File and page 16, lines 317-320):

“For example, patients with an out-of-bed/edge-of-bed mobilisation had a 2.3 times higher chance (odds ratio 95%-CI 0.4-13.4, p=0.36) to demonstrate a clinically relevant 10% variations in ‘VO2 during’ and a 4.7 times higher chance (95%CI 0.46-48.42, p=0.19) in ‘VO2 after’ compared to an in-bed mobilisation (S3 File).”

3. How about assessing the goodness-of-fit (GOF) after these multi-level logistic regression fits? While the Hosmer-Lemeshow is ideal for assessing the standard logistic regression fit, there exists various proposals for the multi-level setup; see below:

The authors may consider producing some summary GOF statistics in this context.

https://digscholarship.unco.edu/cgi/viewcontent.cgi?article=1243&context=dissertations

Since our primary research question (cardiorespiratory response) was to investigate the association of a priori defined factors on study outcomes (and not predictions), we did not report GOF measures. In the table below, we provide the reviewer the GOF measures R-squared (R2), Akaike information criterion (AIC) and the root mean square error (RMSE). For all study outcomes we found R2 values ranging from 0.796 to 0.927. RMSE ranged from 1.11 to 35.9.

Outcome GOF measure Value

HR during R2 0.927

AIC 3376

RMSE 4.03

HR after R2 0.904

AIC 3541

RMSE 5.02

MAP during R2 0.826

AIC 3389

RMSE 5.39

MAP after R2 0.76

AIC 3582

RMSE 6.56

MV during R2 0.858

AIC 1528

RMSE 1.11

MV after R2 0.836

AIC 1583

RMSE 1.28

VO2 during R2 0.871

AIC 3013

RMSE 31.4

VO2 after R2 0.796

AIC 3035

RMSE 35.9

The reporting of GOF measures likely increases the complexity of the manuscript further (see comment from Reviewer #1). In consequence, we decided not to include this table in the manuscript.

4. Interpretation of covariates/explanatory variables in the Results section should be considered in terms of increase/decrease in odds, with associated 95% confidence intervals.

As highlighted in our comments above, Table 2 and 3 report mean differences of median values and not odds ratios as these results are based on our primary analysis (cardiorespiratory response). The results from the safety analysis are reported within S3 File using odds ratios as effect measures. We hope that we could clarify this by using different sub-headings within our statistical analysis chapter.

References

1. Schweickert WD, Pohlman MC, Pohlman AS, Nigos C, Pawlik AJ, Esbrook CL, et al. Early physical and occupational therapy in mechanically ventilated, critically ill patients: a randomised controlled trial. Lancet. 2009;373(9678):1874-82. Epub 2009/05/19. doi: 10.1016/S0140-6736(09)60658-9. PubMed PMID: 19446324.

2. Wright SE, Thomas K, Watson G, Baker C, Bryant A, Chadwick TJ, et al. Intensive versus standard physical rehabilitation therapy in the critically ill (EPICC): a multicentre, parallel-group, randomised controlled trial. Thorax. 2017. Epub 2017/08/07. doi: 10.1136/thoraxjnl-2016-209858. PubMed PMID: 28780504.

3. Caspersen CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions and distinctions for health-related research. Public health reports (Washington, DC : 1974). 1985;100(2):126-31. Epub 1985/03/01. PubMed PMID: 3920711; PubMed Central PMCID: PMCPMC1424733.

4. Streiner DL, Norman GR. Correction for multiple testing: is there a resolution? Chest. 2011;140(1):16-8. Epub 2011/07/07. doi: 10.1378/chest.11-0523. PubMed PMID: 21729890.

5. Althouse AD. Adjust for Multiple Comparisons? It's Not That Simple. Ann Thorac Surg. 2016;101(5):1644-5. Epub 2016/04/24. doi: 10.1016/j.athoracsur.2015.11.024. PubMed PMID: 27106412.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

16 Dec 2021

17 Dec 2021

Responses to Reviewers

The reviewers’ comments are shown in bold type, our responses in plain type and changed passages of the manuscript are shown in cursive type. All pages and lines are in reference to the revised manuscript in the provided PDF-proof.

Reviewer #2:

Thank you for your comprehensive response and changes to the manuscript.

I think it would be good to clarify that "clinically relevant" (if I understand correctly) is related to safety - not sure this is entirely clear.

Perhaps line 431 should read "negatively effect physiological parameters" linking it back to safety. I think this is important as the next step would then be navigating the line between safety and efficacy (intensity)

We appreciate the reviewer’s comment and changed the manuscript as follows:

Lines 211-213:

“The safety cut-off for a clinically relevant variation in physiological measurements was defined as >10% variation – based on half of what is commonly reported as an adverse event (>20%) in the literature [22].”

We would further like to mention, that above sentence is written under the subheading ‘safety analysis’ which should make it very clear to readers that this is about safety.

Lines 421-422:

“Finally, the safety cut-off for a clinically relevant variation was chosen in absence of previous evidence.”

Lines 431-432:

“[…], our findings indicate that rehabilitation in the ICU is safe and does not negatively influence physiological parameters.”

We would further like to highlight the following (unchanged) passage within our manuscript which elucidates that ‘clinically relevant variations’ are in relation to safety:

Lines 109-112:

“Finally, we re-evaluated safety by examining sessions with clinically relevant variations (>10%), an adverse event or therapy discontinuation to determine predictors and explored individual characteristics of patients with a strong physiological reaction.”

We thank the reviewer for their time and dedication to improve our manuscript. We hope to have fully satisfied all concerns in order to see this manuscript published.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

5 Jan 2022

Cardiorespiratory response to early rehabilitation in critically ill adults: a secondary analysis of a randomised controlled trial

PONE-D-21-27687R2

Dear Dr. Eggmann,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Brenda M. Morrow, PhD

Academic Editor

PLOS ONE

Reviewers' comments:

26 Jan 2022

PONE-D-21-27687R2

Cardiorespiratory response to early rehabilitation in critically ill adults: a secondary analysis of a randomised controlled trial

Dear Dr. Eggmann:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Professor Brenda M. Morrow

Academic Editor

PLOS ONE