Effects of robotic rehabilitation using the Erigo® device on patients with neurological injury: a systematic review and meta-analysis of randomized clinical trials

Abstract Objective To systematically review the effects of robotic rehabilitation with the Erigo® device on patients with neurological injury on safety, spasticity, muscle strength, functionality, gait/balance, and changes in the level of consciousness. Methods MEDLINE, SciELO, EMBASE, The Cochrane Library - CENTRAL and PEDro databases were consulted without the restriction of date and language. Randomized controlled trials that evaluated the robotic rehabilitation and compared it to conventional or placebo therapy, isolated or in association with other therapy, were selected. Studies in which the treatment time was less than 10 sessions were excluded. The risk of bias was assessed with the use of the RoB 2.0 tool. Results Nine studies were included, totaling 347 patients. The robotic rehabilitation performed by the Erigo® device proved to be safe for neurological patients. The meta-analysis showed an improvement for spasticity (MD = 0.29; 95% CI = −0.49 to −0.08; I2 = 0%), but there was no significant increase in muscle strength in patients with stroke (MD = 0.25; CI 95% = −0.22 to −0,71; I2 = 0%). Erigo® showed inconclusive effects on functionality, gait/balance and level of consciousness in patients with severe acquired brain injury and vegetative or minimally conscious state. All studies present some concerns for the risk of bias. Conclusion Erigo® as a robotic rehabilitation strategy is safe for patients with acquired brain injury and appears to reduce spasticity in patients with stroke. The effects on muscle strength, functionality, gait and balance and level of consciousness remain uncertain and the methodological quality of the clinical trials included in this review is limited. IMPLICATIONS FOR REHABILITATION Erigo® device is safe to rehabilitation of patients with neurological injury. Rehabilitation with Erigo® device appears to reduce the spasticity of patients with stroke. New randomized clinical trials are needed to establish the effects on muscle strength, functionality, gait/balance and level of consciousness.


Background
Neurological injuries cover a group of vascular, neurodegenerative, or traumatic brain diseases and can lead to temporary or permanent disorders, depending on the affected brain region.They are one of the main causes of disabilities and deaths worldwide, impacting the quality of life and generating great socioeconomic burdens [1].
Among the neurological injuries, it is important to highlight, spinal cord injuries, stroke, persistent vegetative state or brain injury.These conditions cause a series of complications with physical and cognitive deficiencies, which have an impact on activities of daily living and on the quality of life of these patients.Among the physical changes are the reduction or severe impairment of functionality, decline of muscle strength, increased spasticity, gait and balance.Among the cognitive alterations, they range from a coma state, with complete functional dependence, to a confusional syndrome in which the patient manages to have some communication and has better functionality [2,3].
The early treatment in a neurological patient determines motor and sensory recovery, as well as the return to activities of daily living [4].Among the devices resources to optimize the rehabilitation process, there is the Erigo V R robotic board (Hocoma, Volketswil, Switzerland).This equipment presents all the advantages of lower limb movement and orthostatism.It allows an inclination of up to 90 � and the rhythmic passive movement of the knees and hips with a speed of up to 40 steps per minute, associated with functional electrical stimulation (FES).The combination of these resources helps to improve tolerance to orthostatism, alleviates complications related to immobility, thus complementing conventional physiotherapy in patients with spinal cord injury [5].
[6] demonstrated in a preliminary randomized controlled trial that robotic rehabilitation with the Erigo V R device can be a safe and viable therapy, especially for orthostatic reactions in patients with stroke.On the other hand, Krewer et al. [7] in a study with Vegetative State or Minimally Conscious State patients, observed that the most cause for an interruption in treatment with Erigo V R was the occurrence of hypotension, with a reduction in symptoms over the course of time.
Considering that studies provide conflicting information regarding the safety of this device, it is necessary to establish a consensus regarding this issue for considering the effective implementation of this therapy in clinical practice.For this reason, our study aimed to systematically review the effects of robotic rehabilitation with the Erigo V R device compared it to placebo or conventional physical therapy in patients diagnosed with spinal cord injuries, stroke, persistent vegetative state or brain injury.Safety was considered the primary investigational outcome.Spasticity, muscle strength, functionality, gait/balance, and changes in the level of consciousness were considered secondary outcomes.

Methods
This study is a systematic review of randomized clinical trials (RCTs) with meta-analysis, carried out according to the recommendations of Cochrane Handbook for Systematic Reviews of Interventions [8] and reported as specified by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [9].The protocol is registered on the PROSPERO platform (CRD42020173603).

Data sources and search strategy
The databases MEDLINE (via PubMed), Scientific Electronic Library Online (SciELO), EMBASE, The Cochrane Library -Cochrane Central Register of Controlled Trials (CENTRAL), Physiotherapy Evidence Database (PEDro) were searched from inception to January 6, 2022.In addition, the references from articles included in the review were examined to check other potential studies.The search terms included "Spinal Cord Injuries", "Stroke", "Persistent Vegetative State", "Brain Injury", "Chronic", "Robotics", "ERIGO".There were no language or publication date restrictions.The complete search strategy for PubMed is available in Table 1 (Supplementary material) and it has been adapted for the other databases.

Selection of studies
The RCTs that evaluated the effect of robotic rehabilitation using the Erigo V R device and compared it to placebo or conventional physical therapy, isolated or in association with other therapy, in patients diagnosed with spinal cord injuries, stroke, persistent vegetative state or brain injury, with impairment of functionality, were included.Studies in which the treatment time was less than 10 sessions and studies with incomplete data were excluded.
The safety of the therapy assessed by monitoring the cardiac parameters (blood pressure, heart rate, oxygen saturation) was defined as the primary outcome.Adverse response during therapy was identified as reduced of 20 mmHg for systolic blood pressure and 10 mmHg for diastolic blood pressure, peripheral oxygen saturation less than 90% and heart rate of �120 beats/minutes or an increase of �30 beats/minutes when a patient moves from a supine to a standing position [16,17].
The secondary outcomes were: functionality assessed by the Fugl Meyer or Functional Independence Measure; gait and a balance by Tinetti scale; muscle strength by the Medical Research Council (MRC) scale or dynamometry; spasticity by the Ashworth scale; coma state and level of consciousness by the Coma Recovery Scale-Revised (CRSr).
Initially, studies were selected by titles and abstracts by two independent reviewers.Abstracts that did not provide sufficient information for eligibility at this stage were selected for full-text evaluation, which was also performed independently by peers.Disagreements were discussed and if there was no consensus, a third reviewer was consulted.

Data extraction and risk of bias assessment
The reviewers extracted the data on a standardized form.Variables such as author and year of publication, type of study, population, sample size, gender and age, comparator group, type and protocol of intervention, outcomes of interest, and instruments for measuring outcomes were collected.
The evaluation of the methodological quality of the studies was carried out using the "Risk of Bias Toll for Randomized Trials" (RoB 2.0) scale of the Cochrane collaboration.The quality of each item was categorized into "low risk of bias", "high risk of bias" or "some concerns" and, a global score was given for each outcome assessed [18].

Data synthesis and analysis
Most of the results of the studies were presented descriptively due to the variety of populations as well as the measurement instruments and/or intervention protocols.
When possible, data were grouped with a meta-analytical approach.R statistical software, version 3.5.2,was used to perform the meta-analysis.Pooled effect estimates were obtained by the comparison of the change from baseline at the end of the study for each group.A random-effects model was used, with the DerSimonian and Laird variance estimator, and the results were presented as mean difference with a 95% confidence interval.A value of p < 0.05 was considered significant.Statistical heterogeneity between studies was assessed by the I 2 .

Description of the studies
The search strategy identified 3,064 studies, of which nine met the eligibility criteria, totaling 347 patients.Only two studies with stroke patients were included in the meta-analysis, since, the others did not present comparative values (pre and post-intervention), the data were incomplete or represented different populations.One study presented eligibility criteria [19], however, the data was very similar to one presented in another publication by the same author [20].We tried to contact the author to clarify if they are replicated data, but we were not successful.Furthermore, we did not identify the registration of these studies to draw conclusions.Because of that the study by Kumar et al. [19] was excluded.Figure 1 shows the flowchart of the selection of studies.
The most frequent neurological injury in the studies was acquired brain injury, represented by ischemic and hemorrhagic strokes in five studies.Also, all studies (347 patients) reported the monitoring of cardiorespiratory signs during the interventions.However, two studies [7,11] did not describe the absolute values in the manuscript when compared to conventional therapy.The comparison groups to the therapy with Erigo V R device performed conventional physiotherapy, associated or not with verticalization.One study [12] used it as a comparator for robotic rehabilitation with Erigo V R with conventional physiotherapy sessions associated with a device for mobilizing the legs in supine position.
Regarding the protocol with the Erigo V R device, there was a variation in the maximum angle of the board from 60 � to 90 � .Only three studies [10,11,13] reported the steps/minutes ratio, which varied between 18 to 40 steps/minute.The parameters used for the FES were reported in only one study [6], but it only mentions the intensity in which the electric current was applied (30 mA) (table 1).
In general, protocols included 10 to 24 sessions, lasting 15 to 60 min, and a weekly frequency of three to six times/ week.Table 1 shows in detail the main characteristics of the studies.

Risk of bias assessment
The overall for all included RCTs showed "some concerns".The details of the risk of bias assessment for each RoB 2.0 domain and outcome can be seen in table 2.

Stroke and severe brain trauma
There was no evidence of postural hypotension or other adverse events during the intervention with the Erigo V R device in five studies [6, 10,12,14,15].One study [20] did not describe whether or not there was an adverse event during the protocol.Due to lack of data or only pre-intervention values, it was not possible to perform the meta-analysis for the safety outcome.
The assessment of muscle strength is reported in three of the included studies (160 patients)[6, 14,20], and all used the Medical Research Council scale as a measuring instrument.The metaanalysis (figure 3) performed with two articles [14,20] did not show a significant increase in muscle strength in patients with stroke when compared to the control group (MD ¼ À 0.25; CI 95% ¼ À 0.22 to À 0.71; I 2 ¼ 0%).
The study by Calabr� o et al. [6] did not enter the quantitative analysis due to the presentation of the MRC scale scores in a different format.Their sample was composed of patients in the subacute phase of stroke, who underwent the intervention with Erigo V R and showed an increase in muscle strength of the lower limbs compared to the control group that received Bobath sessions and training for the vertebral position sitting and standing (Erigo V R : 2 ± 1 vs control: 1 ± 1; p ¼ 0.03).
Functionality after robotic rehabilitation with Erigo V R was verified in two [6,10] of the included studies (64 patients).The evaluation was carried out with two scales: in the study by Ancona et al. [10] Functional Independence Measure (FIM) scales were used, while the Fugl-Meyer scale for lower limbs was used in the study by Calabr� o et al. [6].
Ancona et al. [10] presented the results for functionality to patients with severe acquired brain injury (ischemic stroke, hemorrhagic stroke and traumatic brain injury) in relation to the baseline (T0) after the 10 th session with Erigo V R (T1) and at the end of the treatment cycle (after the 20 th session with Erigo V R -T2).An improvement was observed in the intervention group with the Erigo V R device in the FIM scale (T0: 52.82 ± 16.60; T1: 58.55 ± 18.71; T2: 62.59 ± 21.54; p < 0.001).In the control group, an analogous response was observed in the improvement from T0 to T2 and from T1 to T2, respectively, in the FIM scales (T0-T2: 52.55 ± 23.92 to 61.23 ± 26.93; T1-T2: 55.64 ± 24.80 to 61.23 ± 26.93; p < 0.001).The authors did not present a comparison between groups for this outcome.
Regarding the study by Calabr� o et al.
[6] performed with stroke patients, the group that received the Erigo V R the stroke group that received the Erigo V R intervention performed better in its score for the Fugl-Meyer scale, compared to the control group (Erigo V R :þ92 ± 10; control:þ58 ± 7; p ¼ 0.008).It was not possible to perform a meta-analysis for functionality outcome because the studies used different assessment instruments and they cover more than one population.
Gait and balance after rehabilitation with Erigo V R was verified only by the study using by Ancona et al. [10] used the Tinetti scale scores.An improvement was observed in the intervention group with the Erigo V R device in the Tinetti scale scores (T0: 2.73 ± 3.73; T1: 6.14 ± 4.06; T2: 8.50 ± 4.82; p < 0.001).In the control group, an analogous response was observed in the improvement from T0 to T2 and from T1 to T2, respectively, in the Tinetti (T0-T2: 4.00 ± 4.04 to 7.77 ± 5.04; T1-T2: 5.68 ± 4.39 to 7.77 ± 5.04; p < 0.05).The authors did not present a comparison between groups for this outcome.

Vegetative state or minimally conscious state
One study [11] did not evidence postural hypotension or other adverse events during the intervention with the Erigo V R .In the studies by Krewer et al. [7] and Taveggia et al. [13] the highest number of therapy interruptions occurred due to postural hypotension, with greater frequency in patients who performed only the verticalization of the table in comparison to those who performed the mobilization of the lower limbs in addition to the verticalization (Erigo V R ).
In    The level of consciousness was assessed in studies [6,7,10,11,13,15].However, the Coma Recovery Scale-Revised (CRSr) proposed as an evaluation method for this outcome in the protocol of this systematic review was only used by three studies included [7,11,13], which represents 83 patients.
In the rehabilitation protocol used by Frazzitta et al. [11] it is possible to identify those 11 patients in the intervention group with the Erigo V R device (73.3%) versus seven in the control group (43.7%) reached the maximum CRSr score (23 points).Comparing the groups, there was a significant improvement for the group that performed the verticalization with Erigo V R (p ¼ 0.03).
The study by Taveggia et al. [13] involving only eight patients, showed no change in the level of consciousness regarding the interventions made with the Erigo V R versus verticalization, after 24 sessions.
It was not possible to perform the meta-analysis for this outcome due to lack of pre-and post-intervention data and because a study provides only percentage values.

Discussion
This is the first systematic review with meta-analysis to demonstrate that the robotic rehabilitation with Erigo V R device is safe and appears to reduce the spasticity of patients with stroke.The effects on muscle strength, functionality, gait/balance and level of consciousness in patients with stroke, severe acquired brain injury and vegetative or minimally conscious state remain uncertain.
The safety of robotic rehabilitation with the Erigo V R device was assessed by analysis of the variation in vital signs and two studies only reported episodes of hypotension.The main adverse effect of immobilization and bed rest is the alteration of the cardiovascular system, leading to a sharp reduction in blood pressure when patients face orthostasis.When the individual transfers from the supine position to the standing position, in response to the baroreflex, there is an increase in the sympathetic activity, subsequent increase in peripheral vascular resistance, venous return, and cardiac output.This compensatory mechanism leads to a drop in systolic blood pressure (5 to 10 mmHg), an increase in diastolic blood pressure (5 to 10 mmHg), and an increase in heart rate (10 to 25 beats) [21,22].Also, the use of antihypertensive drugs and even the reduced sensitivity of baroreceptors, which is related to age, may justify the inability that some patients have to adapt [22].
Training with the Erigo V R device showed improvement in spasticity in stroke patients.However, the quantitative analysis included small sample size, wide confidence interval, and clinical trials that presented some concerns for methodological biases, indicating the need for further studies to strengthen this finding.Our results corroborate the systematic review carried out by Fang et al. [23] who investigated the effect of robot-assisted gait training on spasticity and pain in patients with spinal cord injury.The authors highlight that such results can be justified by the sensory information that the rhythmic movement provided by robotic devices generates, reducing, hence, reflexes and joint stiffness.
In addition to this sensory information provided by the device, the simultaneous combination with FES also helps to reduce spasticity.FES increases motor coordination and reduces spasticity while conventional electrical stimulation presents greater gains in strength and muscle mass.These findings are reinforced in the study by Sharif et al. [24] which compared the efficacy of FES versus conventional electrical stimulation in the gait rehabilitation of patients after stroke, as the results demonstrated a superior benefit of FES therapy in reducing of spasticity.Some of the clinical trials included in this systematic review that evaluated muscle strength in stroke patients did not present complete data (pre and post-intervention), so the meta-analysis was performed with only two studies and the results were not significant when comparing robotic rehabilitation to the control group.Similarly, to the spasticity outcome, the sample size of the analyzed studies was small, the confidence intervals wide, and there were concerns related to the risk of bias, indicating the need for studies with greater methodological rigor.Corroborating our findings, the systematic review by Ambrosini et al. [25] analyzed studies with sub-acute stroke, using cycling training versus cycling associated with FES, and showed no significant difference for the increase in muscle strength, change in the tone and static balance.Also, the quality of the evidence was limited (low to moderate), since the results were based on publications with small sample sizes and high heterogeneity.
On the other hand, the lack of positive effects on muscle strength can be attributed to the training volume of the studies included in the meta-analysis (only 24 sessions).Kuznetsov et al. [26] through a controlled, but non-randomized study, obtained an increase in lower limb muscle strength in stroke patients who had performed 30 sessions of robotic rehabilitation and FES, compared to the group that used only the tilt table training.
Regarding functionality, robotic rehabilitation using the Erigo V R device seems to have positive effects on this outcome in patients with stroke, as demonstrated by Calabr� o et al. [6] In contrast, Ancona et al. [10] showed positive results on functionality after robotic rehabilitation, but they did not present a comparison between the intervention and control groups.For this reason, we cannot affirm that Erigo V R is superior to conventional neurorehabilitation therapy for patients with ischemic stroke, hemorrhagic stroke and traumatic brain injury.The difference found in the findings of these two studies can be partly explained by the scales used to assess functionality and by the populations involved.Calabr� o et al. [6] evaluates only one population and it uses the Fugl-Meyer scale, which is more specific, finding positive results.On the other hand, Ancona et al. [10] evaluates the effect of robotic rehabilitation in individuals with different conditions and it uses a less sensitive instrument to assess functionality (FIM).This could explain the absence of expressive results.
Other systematic reviews [27,28] reiterate that robotic rehabilitation by means of different devices significantly improves functionality in patients with neurological injuries when compared to conventional treatment.This is supposedly due to the mechanism of neural plasticity since new neural connections are favored when high-intensity, repetitive training occurs.Therefore, we reinforce the importance of new and more robust clinical trials that evaluate the effect of the Erigo V R device in patients with neurological injury.
Regarding the findings for gait and balance, in the Ancona et al. [10] study included in this review shown that these outcomes presented similar improvement in the Erigo V R group and in the control group over time.However, the intervention group that performance the conventional neurorehabilitation plus the intervention with Erigo V R device showed increases in scores earlier than those observed in the control group (with 10 sessions versus 20 sessions).A constant and significative increase observed with 10 sessions with the equipment.Swinnen et al. [29] demonstrated in their systematic review with post-stroke patients that there is some evidence of balance improved after robot-assisted gait training.The authors also suggest that rehabilitation with robotic gait should be started as early as possible due to brain plasticity.However, it is still unclear which type of robot produces better results for this outcome.
The level of consciousness seems to improve for patients who perform robotic rehabilitation using Erigo V R plus conventional physical therapy, in comparison to conventional physical therapy alone, as demonstrated by Frazzita et al. [11] However, considering the studies by Krewer et al. [7] and Taveggia et al. [13] Erigo V R was not shown to be superior to tilt table verticalization therapy for this outcome.Nevertheless, it is highlighted that the Taveggia et al. [13] study evaluated only eight patients and despite the Krewer et al. [7] study having a larger sample size, the authors performed only ten Erigo V R sessions when compared to 15 sessions of the Frazzita et al. [11] study.
Multisensory stimulation helps to increase the consciousness of comatose patients, who can be stimulated by recruiting the ascending reticular activator system or collateral pathways if the former is injured.This system is responsible for the state of cortical alertness and to rehabilitate the verticalization is used.It can also help the contralateral sensorimotor cortical organization and reduces adverse actions to orthostatism [30,31].
Regarding the risk of bias, all studies included in this review had "some concerns" for the evaluated outcomes.The domain of the RoB 2.0 tool that presented the greatest weakness was "Deviations from intended intervention", as all studies present a very poor description of the interventions to infer any judgment.In addition, there is no information on blinding patients and therapists.The domain "Selection of the reported result" was the second most penalized, since, of the nine studies included in this review, seven had "some concerns".This occurs since the studies are not registered on platforms such as ClinicalTrials.gov,making the transparency of the evaluated outcomes compromised.Finally, the domains "Randomization process" and "Measurement of the outcome" also presented "some concerns" for more than 50% of the evaluated studies, because the randomization process is not described and it is not mentioned if the outcome assessor was blinded.
Among the limitations of this systematic review, highlights the low reproducibility of RCTs published so far on the effects of Erigo V R in patients with acquired brain injury.The lack of information on the applied protocols, heterogeneous studies and lack of robust evidence makes it impossible to make further inferences about this technology.Another important limitation is the lack of pre-and post-intervention assessments and standardization of evaluation scales, that made unfeasible to carry out a meta-analysis for some outcomes.
Robotic rehabilitation using the Erigo V R device can be an initial rehabilitation strategy during hospitalization and after hospital discharge for patients with motor and cognitive impairments after brain and spinal cord injuries.However, better-designed RCTs with larger sample sizes are needed to strengthen these findings and elucidate the real benefits of this therapy for implementation in clinical practice.

Conclusion
Robotic rehabilitation using Erigo V R device is safe for patients with neurological injury and appears to reduce spasticity in patients with stroke.The effects on muscle strength, functionality, gait/ balance, and level of consciousness remain uncertain.It is important to reiterate that the methodological quality of the clinical trials included in this review is limited and the results should be interpreted with caution.

Figure 2 .
Figure 2. MD and 95% CI on spasticity evaluated by Scale Modified Ashworth for ErigoV R vs control group.CI: confidence interval; MD: mean difference.

Figure 3 .
Figure 3. MD and 95% CI on muscle strength evaluated by Medical Research Council for ErigoV R vs control group.CI: confidence interval; MD: mean difference.

Table 1 .
Characteristics of included studies.
relation to spasticity, Krewer et al. [7] randomized 44 for treatment with the Erigo V R device versus tilt table.At the initial evaluation, 29 patients had mild spasticity (n ¼ 14 Erigo V R ; n ¼ 15 tilt table) and seven presented moderate spasticity (n ¼ 4 Erigo V R ; n ¼ 3 tilt table).After three weeks of interventions, there was no difference between the groups for this outcome (p ¼ 0.277).

Table 2 .
Evaluation of the risk of bias using the Rob 2 tool from Cochrane.