Background: Few studies showed that early rehabilitation on the Intensive Care Unit (ICU) results in a better functional outcome at hospital discharge and reduced hospital stay. Effects of early rehabilitation at the general ward directly after transfer from the ICU on the length of stay (LOS) in hospital and on the functional outcome are missing.

Methods: In a prospective randomised controlled trial (RCT) 53 consecutive critical-illness-survivors were enrolled at nine ICUs. Early rehabilitation program (protocol) consisted of exercise therapy, active breathing techniques and electrical stimulation after discharge from the ICU to ward-based care. The usual care group received physical therapy as ordered by the primary care team after discharge from the ICU. LOS at the general ward after transfer from the ICU was recorded. Furthermore, Early-Rehabilitation-Barthel-Index (ERBI), Visual-Analogue-Scale for pain (VAS), 3-Minute-walk-test (3min), Beck-Depression-Inventory (BDI), State-Trait-Anxiety-Inventory (STAI), and Medical-Research-Council-scale (MRCS) were assessed.

Methods Patients All the procedures described in this study were approved by the local Ethics Committee (No. 375/2007), and informed consent was obtained from all patients or by proxy. The study was conducted prospectively between 2008 and 2011 in nine ICUs (three medical, six surgical; all together 72 beds) located in the General Hospital of Vienna/Medical University of Vienna according to the Helsinki Declaration of 1975, as revised in 1983.

Procedures ICU patients were recruited as soon as they were suitable for early ICU rehabilitation (see inclusion criteria). After baseline measurements patients were randomly assigned to the early rehabilitation group. A computer-generated, permuted block randomization scheme was used considering sex, age, LOS at the ICU and diagnostic category to allocate patients to the early rehabilitation group or usual care group. Each assignment was operated through a web-based randomization system (https://www.meduniwien.ac.at/randomizer) by an investigator with no further involvement. All included patients received the same usual care physical therapy, as ordered by the primary care team, during their stay at the ICUs. Unresponsive patients underwent passive range of motion exercise for all limbs. Once patient interaction was achieved, no matter if the patient was mechanically ventilated, a progressive early rehabilitation program was initiated starting with active assisted or active range of motion exercises and breathing techniques. If these exercises were tolerated, treatment was advanced to bed mobility exercises, transferring to upright sitting and sitting balance activities. The early rehabilitation program progressed to participation in activities of daily living, sit-to-stand transferring, pre-gait exercise and finally up to walking. Progression of activities was dependent on patient tolerance and stability (systolic blood pressure between 65 and 200 mm HG, heart rate between 40 and 130 beats per minute, respiratory rate between 5 and 40 breaths per minute, and pulse oximetry above 88%). Patients were discharged to a general ward as soon as their medical condition was sufficiently stable, regardless of their functional status.

Directly after transfer to a general ward the early rehabilitation group started with an early rehabilitation program, as ordered by an experienced physiatrist. The early rehabilitation program consisted of a coordinated exercise therapy and neuromuscular electrical stimulation, for two hours and five days a week until discharge from the acute hospital. Exercise therapy included strength training (three days/week, one to three sets of 12-15 repetitions, 8-10 exercises involving the major muscle groups: thighs, calves, buttocks, trunk, shoulder girdle), aerobic exercise (large muscle activities five days/week, 10-30 min/session, 50-80% maximal heart rate) and active breathing techniques. Exercise equipment being used: Resistance bands, free weights, weight machines, shoulder pulleys, stationary and bedside cycle ergometers, and wobble boards. Progression of activities was dependent on patient tolerance and stability. For neuromuscular electrical stimulation we applied Compex-P© devices producing biphasic symmetric impulses with a frequency of 50 Hz and a pulse width of 0.35 ms (stimulus regime: 8 s on/24 s off) five days/week, 30 min/session involving the anterior thighs and buttocks. The usual care group received single physical therapy sessions as ordered by the primary care team after transfer from the ICU to the general ward (exercise therapy, neuromuscular electrical stimulation).

Patients were assessed by two medical students who were unaware of randomization assignment. The medical students were trained by an experienced senior physician to perform the functional measures and questionnaires. To maintain blinding, the patients and any visitors were instructed not to discuss previous interventions with the assessors or the primary care team. Furthermore, assessments occurred after 4 pm at a time distant from the therapy intervention.

Statistics On the basis of ten pilot critical-illness-survivors who were admitted to intensive care, we noted a mean difference of seven days after transfer to a general ward between the early rehabilitation protocol and usual care revealing an effect size of 0.58 (G*power 3.1.3) [39]. We calculated that a total sample size of 40 critical-illness-survivors per group (for a total of 80) would be needed to reach a power of 80% using a one sided Wilcoxon-Mann-Whitney-Test at a significance level of 0.05. However the inclusion time period for new patients was limited to two and half years and only 53 patients met the inclusions criteria and were willing to participate.

The primary endpoint, days in hospital after transfer from the ICU to a general ward, was analyzed by intention to treat and per protocol analyses. The intention to treat sample included all patients, who could be transferred to a general ward in the General Hospital Vienna (early rehabilitation group: n = 23, usual care group: n = 27), whereby all patients were randomized at the ICU before transfer. Three patients already randomized had to be excluded as they were either unpredictable directly transferred to another hospital from the ICU (early rehabilitation group: n = 1, usual care group: n = 1), or died (usual care group: n = 1) at the ICU.

The per protocol sample included all patients who were randomized at the ICU, transferred to a general ward in the General Hospital Vienna, and for who the treatment was really started (early rehabilitation group: n = 19, usual care group: n = 27). For the per protocol analysis, in addition to the intention to treat sample, four patients had to be excluded in the early rehabilitation group because they did not receive additional treatment.

The difference between treatment and usual care group was analyzed univariately by log rank tests, and Kaplan-Meier curves were plotted. A Cox proportional hazard model was applied to analyze the difference between groups, also accounting for age, sex, diagnostic category at admission, and days at the ICU. We took the time at the general ward for patients who were treated at the General Hospital Vienna until discharge, and the total time at the general ward (days at the General Hospital Vienna + other hospital) for patients who were treated at the General Hospital Vienna and then transferred to another hospital. For some patients, the event (discharge from the general ward) did not occur within the observation period, as they were either lost to follow up (that means they were transferred to another hospital with unknown LOS) or died. For these patients we took the time at the general ward of the General Hospital Vienna until transfer or until death, respectively, and censored them. In the statistical analyses, the censoring considered that these patients did not experience the event (discharge from the general ward) until the time of censoring. No assumption was made when it may have happened afterwards. The intention to treat sample included five censored patients (3 early rehabilitation group, 2 usual care group) and the per protocol sample three censored patients (1 early rehabilitation group, 2 usual care group).

For analyses of the secondary endpoints ERBI, VAS, 3min, BDI, STAI, and MRCS, descriptive statistics (median, first quartile, third quartile, and range) were carried out at baseline and discharge from hospital as well as for the difference between discharge and baseline for each treatment group separately including data of all patients randomized in the study.

To compare the hospital costs after transfer from the ICU to the general ward between early rehabilitation and usual care group, descriptive statistics were assessed for the intention to treat sample as well as for the per protocol sample. All analyses were calculated with the statistics program SAS 9.2 and R 3.0.1. For the primary analysis, a p-value < 0.05 was considered as statistically significant.

Inclusion Criteria:

• > 16 years
• ICU stay for at least five days
• APACHE II Score ≥ 20
• ERBI ≥ -150
• ability of sitting on the edge of the bed for at least one minute

Exclusion Criteria:

• disease of the central motor nervous system (e.g. stroke)
• multiple trauma
• diagnosis of irreversible condition with little rehabilitation potential (e.g. cardiac insufficiency - New York Heart Association IV)
• language barrier
• if a transfer in another hospital was foreseeable