Content » Vol 51, Issue 2

Review article

Feasibility and effectiveness of repetitive gait training early after stroke: a systematic review and meta-analysis

Jonas Schröder, MSc1,2, Steven Truijen, MSc, PhD1,2, Tamaya Van Criekinge, MSc1,2 and Wim Saeys, MSc, PhD1–3

From the 1Research group MOVANT (MOVement ANTwerp), Department of Rehabilitation Sciences and Physiotherapy, University of Antwerp, Wilrijk, 2Multidisciplinary Motor Centre Antwerp (M2OCEAN), University of Antwerp, and 3Rehabilitation Hospital Revarte, Edegem, Belgium

Abstract

Background: Pre-clinical evidence suggests a period
early after stroke during which the brain is most receptive to rehabilitation, if it is provided as high-dose motor training.

Objective: To evaluate the feasibility of repetitive gait training within the first 3 months post-stroke and the effects on gait-specific outcomes.

Methods: PubMed, Web of Science, Cochrane Library, Rehab Data and PEDro databases were searched systematically. Randomized controlled trials were included to descriptively analyse the feasibility and quantitatively investigate the effectiveness of repetitive gait training compared with conventional
therapy.

Results: Fifteen randomized controlled trials were included. Repetitive training can safely be provided through body weight support and locomotor assistance from therapists or a robotic device. No difference in drop-out rates was reported despite the demanding nature of the intervention. The meta-analysis yielded significant, but small, effects on walking independence and endurance. Training with end-effector robots appears most effective.

Conclusion: Robots enable a substantial, yet feasible, increase in the quantity of walking practice early post-stroke, which might enhance functional recovery. However, the mechanisms underlying these effects remain poorly understood.

Key words: humans; stroke; stroke rehabilitation; exercise therapy; robotics; walking; paresis.

Accepted Oct 8, 2018; Epub ahead of print Dec 4, 2018

J Rehabil Med 2019: 51: 00–00

Correspondence address: Jonas Schröder, Research group MOVANT (MOVement ANTwerp), Department of Rehabilitation Sciences and Physiotherapy, University of Antwerp, Wilrijk, Belgium. Email: jonas.schroder@uantwerp.be

Lay abstract

Animal models suggest that rehabilitation provided as repetitive motor training is most effective early after stroke. To investigate whether such a rehabilitation approach can enhance long-term walking recovery in human patients with stroke, this review gathered clinical studies on the effects of repetitive gait training. We found that robots, in particular, provide a substantial, yet feasible, increase in the amount of walking practice in those stroke patients who are unable to walk. This increase in rehabilitation dose improves walking ability in the long-term. However, these effects are inconsistent, rather small, and in contrast with neutral effects on motor functions of the paretic leg. Therefore, the effects of repetitive training in the context of early stroke rehabilitation remain poorly understood and further research is required.

Introduction

Stroke is a leading cause of long-term disability worldwide and a dramatic increase in incidence is expected (1). The economic consequences are enormous (2), particularly regarding stroke survivors who remain dependent on continuous support (3). For social participation, regaining mobility is obligatory (4). However, more than 20% of stroke survivors do not reach independent walking (5, 6) and even if achieving independence, the great majority struggle to ambulate in the community (7). These persisting disabilities will aggravate physical inactivity, leading to deconditioning and poor long-term outcome (5, 8). If research fails to provide effective rehabilitation, the increasing incidence will inevitably lead to a growing dependent stroke population.

Considering that no therapeutic approach, to date, has proven superior (9) and effect sizes in clinical research are, in general, low (10), it seems reasonable to reflect on basic research. Interesting pre-clinical evidence on timing of stroke rehabilitation has been published. In rodents, motor training loses effectiveness if provided delayed, i.e. 7 (11) and 30 days (12) post-stroke, respectively, compared with earlier exposure. This activity-induced recovery pattern matches the temporal pattern of increased gene expression important for neuronal growth and plasticity in the post-stroke brain (13, 14). Thus, it appears that a limited period of heightened plasticity is induced early after stroke, in which the brain is most receptive to rehabilitation (14). Since this period is time-dependent, it is best described as a critical time-window for stroke rehabilitation (15).

In human stroke survivors, greatest gait recovery gains occur within the first 3 months post-stroke (5, 16) and rehabilitative interventions outside this period have rather modest effects (15, 17, 18). This time-dependent recovery profile corresponds highly to characteristics of a critical time-window, as mentioned above, which might be reflected in the association between earlier rehabilitation and improved outcome (19, 20). This emphasizes the need to develop a rehabilitative approach designed to take advantage of this time-window.

Such an approach should include high-dose training initiated within the first weeks aiming at the recovery of normal function (15). This is in great contrast with how rehabilitation is provided in current practice (21, 22). Therefore, this review aims to detect therapeutic strategies allowing such intensive therapy in the early phase when patients usually have severe weakness and are unable to walk. It is hypothesized that highly-repetitive gait training has the potential to improve long-term outcome when temporally matching the critical time-window.

However, there are concerns that application of rehabilitation too early might slow recovery (23, 24) or even induce infarct-expansion (25). In addition, clinicians might limit the patient’s effort to engage in exercise, since this can lead to short-term increases in spasticity (26) and an increased risk of falling (27). To clarify these aspects, all trials on early repetitive gait training will be collected to investigate feasibility as well as effectiveness.

Which strategies providing repetitive walking practice to non-ambulatory patients early post-stroke have already been investigated in the scientific literature?

Is early-initiated repetitive gait training feasible in terms of safety and patients’ acceptance?

Is repetitive gait training early after stroke more effective than conventional physiotherapy in terms of gait recovery and do these effects persist?

Methods

The current review was developed in adherence to the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) (28).

Definitions

According to the World Health Organization (WHO), stroke is defined as rapidly developing signs of disturbance of cerebral functions lasting > 24 h (unless interrupted by surgery/death), with no apparent non-vascular cause (29).

The focus here is on the early rehabilitation phase, defined as the first 3 months post-stroke, i.e. the period during which most gait recovery gains are observed (5, 16). Studies initiating gait training within a mean of at most 31 days post-stroke were included, to guarantee that the investigated population was exposed to the intervention within this time-window.

Furthermore, participants included in this review were non-ambulatory (Functional Ambulation Classification (FAC) ≤3, or equivalent) (30) as we aim to report interventions which can be provided to patients who are dependent in walking.

The intervention was considered repetitive gait training if an “active motor sequence was performed repetitively within a single training session, and the practice was aimed towards a clear functional goal” (31). In this case, the motor sequence was defined as whole, complex gait cycles and the functional goal as independent walking. The training might be provided with the assistance of therapists or with (electro-)mechanical devices. Trials were excluded if training is combined with another intervention (e.g. electrical stimulation) and the effects could not be isolated.

A study was identified as a randomized controlled trial (RCT) if the participants were assigned prospectively to 1 of 2 (or more) alternative forms of intervention using random allocation. In included trials all groups spend an equal amount of time on therapy.

Literature search

In October 2017, the following databases were searched for trials published between January 2000 and October 2017: PubMed, Web of Science, the Cochrane Library, PEDro and Rehab Data. Indexing terms and free-text words of the following key terms and synonyms were used: (Participants) “stroke” and “ (sub-)acute” or “inpatient”; (Intervention) “exercise therapy” or “task-specific training”; (Outcome) “gait” or “walking”; (Study design) “RCT”. A detailed search strategy used in PubMed can be found in supplemental material (see Table SI1). A search revision was scheduled while finalizing the manuscript to avoid missing recently published studies.

Search records were saved in EndNote X8. Duplicates were identified and removed. Afterwards, different screening phases based on abstracts and full-texts were conducted. Disagreement between 2 reviewers (JS, WS) performing study selection independently were discussed with a third reviewer (ST) to reach consensus.

Studies were included when: (i) patients had been diagnosed with stroke, (ii) the mean stroke interval (time between stroke onset and randomization) was at most 31 days, (iii) patients were non-ambulant (FAC ≤ 3), (iv) effects of repetitive gait training were investigated and (v) compared with conventional physiotherapy, (vi) the study used an RCT design, and (vii) the article was written in English, German or Dutch.

Methodological quality

The Physiotherapy Evidence Database Scale (PEDro), an 11-item scale, was used to assess methodological quality of included RCTs. All scores were obtained from the PEDro database. The first item, eligibility criteria, does not account for the total score and blinding of patients (item 5) and therapists (item 6) is impossible due to the nature of the intervention. Therefore, the maximum score is considered to be 8 and the following classification is used: a study with a PEDro score of 7–8 is considered good quality, while a score of 5–6 indicates moderate quality. To guarantee high-quality reporting, trials with a high risk of bias, i.e. a PEDro score of ≤ 4, were excluded.

Outcomes

The following data were extracted from selected studies: sample size, stroke interval, baseline impairment, type of experimental intervention and characteristics, between-group differences in the occurrence of adverse events and drop-outs, and effects on gait-specific outcomes. In case of missing data or inadequate documentation, the corresponding author was contacted.

Outcomes of this review were reported in correspondence with the proposed research questions. This firstly includes a description of therapeutic strategies allowing non-ambulant stroke patients to repetitively train walking. Secondly, the 2 feasibility items safety, measured by the incidence of adverse events, and adherence to therapy, defined as the number of drop-outs, were investigated. Thirdly, outcomes on effectiveness were investigated and classified according to the domains of the International Classification of Functioning Disability and Health (ICF) model (32). All included studies had to include the ability to walk independently (primary outcome) as an outcome measure. Secondary gait-related outcomes were included, such as motor impairments of the affected leg and different measures on walking performance.

Quantitative analysis

Review Manager software (RevMan 5.3) was used for the quantitative synthesis on the comparative effectiveness. Therefore, the number of participants in both groups together with the means of post-intervention and follow-up scores and its standard deviations were entered in RevMan 5.3 by one reviewer (JS) and cross-checked by another reviewer (WS or ST). If the scores were provided in medians and interquartile ranges, an algorithm developed by Wan et al. (33) was used to estimate means and standard deviations. Summary effect sizes (SES) were calculated with 95% confidence interval (95% CI) based on the effect sizes of individual studies. The mean differences (MD) were calculated since identical measures were used per comparison. When dichotomized outcome on walking independence was reported, an odds ratio was additionally calculated. The I2 statistic was used to determine between-study heterogeneity in results. If heterogeneity was high (I2 > 50%) a random-effects model was used. In each comparison, a sub-analysis on the intervention type was performed. If at least 3 RCTs could be included in a sub-group, the results were reported separately. In addition, if results of 2 or more subgroups were given, the subgroup difference was established using a χ2 test. Finally, the level of evidence drawn from the quantitative analysis were graded using a classification adapted from the Scottish Intercollegiate Guideline Network (SIGN) guidelines (34), where the methodological quality of included RCTs and consistency of results (based on the I2 test for heterogeneity) will be taken into account (see Table I).


Table I. Rating the level of evidence adapted from the Scottish Intercollegiate Guideline Network (SIGN) guidelines

Results
Literature search

In PubMed, the search strategy (see Table SI1) led to 330 hits on 24 October 2017 and a similar strategy was used in Web of Science. After identifying the 2 main interventions, body weight supported treadmill training (BWSTT) and robot-assisted gait training (RAGT), in other databases (Cochrane Library, Rehab Data, PEDro) the reviewers searched explicitly for those interventions. After de-duplication and a first phase screening on eligibility, 132 unique studies were included for detailed screening on abstract and afterwards on full-text. Finally, 15 studies were included (see Fig. 1). A revision in August 2018 did not reveal additional eligible studies.


Fig. 1. Flow diagram of study identification and selection process. P: participants; I: intervention; S: study design.

Methodological quality

In the final screening phase, 4 studies were excluded due to insufficient quality. Of the remaining 15 studies, 9 presented good (PEDro score 8 (35–37); 7 (38–41)) and 6 moderate quality (PEDro score 6 (42–47); 5 (48, 49)) (see Table II). A detailed scoring is shown in Table SII1.


Table II. Characteristics of included studies regarding the methodological quality, recruited population and applied intervention

Outcomes

In the 15 studies, a total of 915 participants were treated and evaluated: RAGT was provided to 286 participants and BWSTT to 152 participants, while 425 participants were allocated to the control groups.

Descriptive analysis of the intervention

All participants were provided with repetitive gait training or conventional therapy as an add-on to usual care, depending on the group to which they were allocated. Usual care included 25–60 min of physio-therapy daily.

To allow non-ambulant patients in the experimental group to repetitively practice gait, various forms of manual and (electro-) mechanical assistance were provided. In all but 1 (46) included studies, participants’ body weight and trunk was (partly) supported by an overhead harness system, while stationary practicing walking on a moving treadmill or footplates. In a single study, the trunk is not suspended in a harness, but a robot device supports the legs and trunk for stance stability to allow full weight-bearing by the patient (46). Body weight supported training can be divided into BWSTT and RAGT depending on the kind of assistance in locomotion provided. During BWSTT, patients train on a treadmill while therapists manually assist stance stability, swing initiation and forward progression of the paretic leg in a cyclical motion (36, 37, 41, 47). RAGT involves a similar stationary set-up while patients are not manually assisted by the therapist(s), but with an electromechanical device, i.e. a robot. Two different kinds of robots can be classified based on the motion they apply to the patient (see Fig. 2):


Fig. 2. Graphic illustration of identified interventions.

  • The Gait Trainer (35, 39, 40, 42–44, 48) is an end-effector device, meaning that motion is applied to the feet of the patient only by 2 footplates whose driven movement simulates stance- and swing-phase of a symmetrical gait pattern. This kind of assistance differs from treadmill-based training, as during the whole gait-cycle the feet are in contact with the footplates and therefore there is no foot clearance during the swing phase.
  • The exoskeleton Lokomat (38, 45, 49) is a robotic-driven orthosis consisting of actuators applying motion to the hip and knee joints of the patient to guide locomotion in a pre-programmed kinematic trajectory based on characteristics of a healthy symmetrical gait pattern.

There is an exception to this division. Ochi et al. (46) investigated a treadmill-based system including robotic arms, which guide the thighs and legs to reproduce a physiological gait pattern. As this system resembles the characteristics of before-mentioned exoskeletons (i.e. precise control of kinematics in the hip and knee joints), it will be accounted as such in the analysis.

These training modalities are compared with a control group, which is provided with conventional physiotherapy. This includes pre-gait exercises aiming at paretic leg strengthening and sitting balance. If possible, manual-assisted over-ground balance and gait training was provided. However, the exact content of the control intervention throughout the included studies is poorly described.

Few studies provided detailed information on the therapy dose. Ada et al. (36) documented that participants were able to walk 129 m during the first session of BWSTT compared with only 26 m in the control group. Tong et al. (39) documented that participants performed 500–1,000 steps during a session using an end-effector robot and during conventional therapy 50–100 steps only. Pohl et al. (35) found that participants walked with the same device 851 to 1,076 steps, similar to the results of Morone et al. (43). In addition, Peurala et al. (48) found that, with robot assistance, participants were able to initially walk 20 min without needing resting breaks, while none in the control group were able to. A similar documentation on exoskeletons is lacking, but authors declared that the exoskeleton allowed patients to practice walking at much higher doses compared with the control condition (45, 49). Overall therapy dose, as measured by the total augmented time spent walking in the experimental group, is found to vary between 300 and 960 min. Most studies provided additional 400 min of walking practice in 20 sessions over 4 weeks, meaning that 5 training sessions were provided weekly (see Table II).

Descriptive analysis of the feasibility

In total, 53 patients dropped out of the experimental group, while 55 dropped out of the control group (see Table III). The great majority of drop-outs were unrelated to the intervention (e.g. scheduling interference). In a single study, adverse events were reported without any difference between experimental and control group (43). In addition, few studies reported minor events caused by training, such as discomfort due to the harness (47), hypotension (43), pain (36, 43) or pressure sores (45), which led to a temporal discontinuity of the intervention. However, no study documented a significant difference between groups in the occurrence of such events (see Table III).


Table III. Extracted data from included studies on feasibility and effectiveness on gait-specific outcomes, as documented in the published article

Quantitative analysis of the effectiveness

The following outcome measures on the comparative effectiveness of repetitive gait training were detected and classified according to the ICF.

1. Categorization

Activity level: The measurements assessing the ability to walk are classified under the activity domain “walking” (ICF d450).

  • Walking independence: Independence, the primary outcome, is either measured with the FAC, a 5-item scale measuring the degree of assistance required to walk, or by dichotomized outcome where the number of patients achieving independence (e.g. FAC ≥ 4) is scored.
  • Walking speed: The time is measured while participants walk over a 5- or 10-m distance at a comfortable pace to calculate walking speed.
  • Walking endurance: Endurance is assessed by asking the participant to walk the greatest possible distance during a period of 6 min.

Body function level:

  • Motor control: The motor subscale for the lower extremity of the Fugl Meyer Assessment (FM-L) is classified under the domain “control of voluntary movements functions” (ICF b760).
  • Muscle strength: The (Medical Research Council) Motricity Index subscale for the lower limb (MI-L) measures the strength of major leg muscle groups and is classified as “muscles power functions” (ICF b730).

The results of the meta-analysis for each outcome, as defined above (see Table IV), are described below. Forest plots are derived from RevMan (see Figs S1–S71).


Table IV. Results of the quantitative analysis on the comparative effectiveness of repetitive gait training. This includes gait-specific outcome on both body function and activity level. Sub-analyses based on the intervention type are performed for each comparison and results are analysed if at least 3 RCTs could be included

2. Results

Activity level:

  • Walking independence: Post-intervention FAC scores were reported in 10 RCTs. Pooling yielded a nonsignificant heterogeneous SES (10 RCTs; n = 671 [exp 338; ctr 333]; MD = 0.38 [random]; 95% CI –0.03 to 0.78; p = 0.07; I2 = 78%). There is a significant subgroup difference (p = 0.002) and the sub-analysis revealed end-effector training to be effective only (5 RCTs, n = 385 [exp 187; ctr 195]; MD = 0.73 [random]; 95% CI 0.17–1.30; p = 0.01; I2 = 75%).
  • At follow-up (≥ 3 months), pooling of 7 RCTs resulted in a significant heterogeneous SES (7 RCTs; n = 538 [exp 266; ctr 272]; MD = 0.57 [random]; 95% CI 0.14–1.01; p = 0.01; I2 = 66%). If pooling outcome of end-effector studies only, a heterogeneous SES is identified (5 RCTs; n = 381 [exp 186; ctr 195]; MD = 0.72 [random]; 95% CI 0.16–1.28; p = 0.007; I2 = 69%).
  • In addition, dichotomous outcome was pooled to calculate an odds ratio. Follow-up data was entered if provided and otherwise post-intervention data was used. This yielded significant heterogeneous results (8 RCTs; n = 627 [exp 312; ctr 315]; OR = 1.99 [random]; 95% CI 1.13–3.53; p = 0.02; I2 = 60%). If pooling end-effector studies only, a greater, but non-significant, SES is identified (5 RCTs; n = 382 [exp 187; ctr 195]; OR = 2.15 [random]; 95% CI 0.88–5.28; p = 0.1; I2 = 75%). Taking the inconsistency of results into account, there is level B evidence for improved walking independence after repetitive gait training.
  • Walking speed: This was assessed in 9 RCTs and pooling yielded a non-significant homogenous SES (9 RCTs; n = 672 [exp 342; ctr 330]; MD = 0.05 [fixed]; 95% CI –0.00 to 0.11; p = 0.06; I2 = 42%). No significant subgroup difference was found (p = 0.41).
  • Walking endurance: In 4 RCTs, endurance was measured. When pooling data, a significant homogenous SES (4 RCTs; n = 406 [exp 206; ctr 200]; MD = 24.36 [fixed]; 95% CI 3.58–45.14; p = 0.02; I2 = 43%) is found. Taking the inconsistency of results into account, there is level B evidence for improved walking endurance after repetitive gait training. If pooling end-effector trials only, a homogeneous significant SES is calculated (3 RCTs; n = 309 [exp 154; ctr 155]; MD = 32.08 [fixed]; 95% CI 8.30–55.86; p = 0.008; I2 = 44%).

Body function level:

  • Motor control: Four RCTs assessed the FM-L and pooling results yielded a non-significant homogeneous SES (4 RCTs; n = 179 [exp 91; ctr 88]; MD = 0.52 [fixed]; 95% CI –1.52 to 2.59; p = 0.62; I2 = 28%). If pooling exoskeleton trials only, a nonsignificant heterogeneous SES is calculated (3 RCTs; n = 119 [exp 63; ctr 56]; MD = 0.76 [fixed]; 95% CI –1.83 to 3.36; p = 0.56; I2 = 51%).
  • Muscle strength: For a comparison on the MI-L, results of 5 RCTs could be pooled. This resulted in a non-significant heterogeneous SES (5 RCTs; n = 364 [exp 185; ctr 179]; MD = 3.64 [random]; 95% CI –2.88 to 10.17; p = 0.27; I2 = 56%). If isolating effects of end-effector trials a significant homogeneous SES (3 RCTs; n = 230 [exp 113; ctr 117]; MD = 8.00 [random]; 95% CI 2.08–13.93; p = 0.008; I2 = 8%) is identified.
Discussion

No between-group difference in the occurrence of adverse events and drop-outs was found despite the demanding nature of the intervention. This suggests it is feasible to provide repetitive training early after stroke by the use of an overhead harness system for support of body weight and manual or electromechanical assistance in forward progression of the paretic leg. Statistical significant effects on walking independence (at follow-up) and endurance is found in favour of repetitive training, according to level B evidence. Sub-analyses revealed that these effects are based mainly on studies investigating RAGT provided with an end-effector robot.

Dose-response relation in stroke rehabilitation

In the context of neurological rehabilitation, repetitive training leads to task-specific improvements (10, 31) and associated neuroplastic re-organization (50) if a sufficient dose of practice is provided. In animal models, synaptic changes in the motor cortex are observed after 400, but not after 60 reach-movements (51) and gait training is effective only if at least 1,000 steps are performed during a treadmill session (52). Corresponding findings in clinical research are in favour of a dose-response relation in stroke rehabilitation (17, 53). Despite this solid association between larger quantities of practice and greater gains, inpatient rehabilitation is described as a time of being physically inactive (54, 55) and the practice dose is far less than is provided in previously mentioned stroke models: patients walked for a mean of 250 (21) steps, while non-ambulatory patients walked for as little as 6–16 steps (56) during a therapeutic session aiming at gait recovery.

Technological advances can be of great value in providing more intensive rehabilitation, as robots let non-ambulatory patients train at much higher doses (57). For example, the Gait Trainer allows patients to execute approximately 1,000 steps in a session, while assistance of a single therapist is usually sufficient (35, 39, 43). In line with a dose-response relation, training with such a device appears effective in improving walking independence and endurance. This implies the importance of practice repetitions when designing effective interventions (58) and, in more general terms, the significance of motor learning in stroke rehabilitation (14, 50, 59).

However, the dose-response relationship is not linear, indicating that other factors have an influence (53). Morone et al. (43, 44) compared responsiveness to training between groups who differ in baseline scores on the MI-L (MI-L≈16 vs 52). Outcomes clearly demonstrate that only the more impaired patients benefit (43, 44), which is supported by Pohl et al. (35) as they found impressive effects in a more affected population (MI-L≈32; see Table II). Interestingly, the initial muscle strength of the paretic leg (e.g. assessed with the MI-L (6)) measured within the first days to weeks post-stroke is associated with walking ability at 6 months (5, 6, 60). Therefore, it appears that robot-assisted training was most effective in those patients with a poor prognosis. This might be related to a greater treatment contrast, since the more affected patients do not engage in intensive rehabilitation under conventional conditions (56). As suggested by Morone et al. (57) future research should not investigate if RAGT is effective, but rather who may benefit (43, 57).

What drives improved walking ability after repetitive training?

It is essential to consider that the task performance in the context of stroke rehabilitation can improve either via restitution of impairments or compensation strategies (15, 61–63). While the included participants potentially improved their ability to walk, we do not know how these changes are achieved, as the FAC does not reflect whether patients “returned towards more normal patterns of motor control” (i.e. restitution), or learned to “accomplish the goal through a new approach by the use of intact muscles, joints and effectors” (62, 64). Since participants improved their ability to walk without normalization of motor control and strength of the paretic leg (see Table IV), it seems that it is rather through compensation that the included patients improved. Indeed, compensation is frequently observed in the recovery of standing balance (65–67) and walking (68) as patients adopt an asymmetrical pattern to shift the kinetic control towards the unaffected side, while normalization of motor control of the paretic leg is almost lacking (69, 70). However, robots provide practice in a symmetrical pattern, which at first sight appears paradoxical. In future trials, analyses of the quality of the gait pattern, including inter-limb coordination in spatiotemporal and kinetic parameters, should be included to provide definitive evidence on mechanisms underlying effectiveness of early training (64, 71). This knowledge will have major implications for practice and designing robots for rehabilitation (i.e. trying to improve impairments or teaching compensation strategies) (15, 72).

Future directions for robots in rehabilitation

Despite evidence in support of repetitive training, small effect sizes are found. These are statistically significant, but the clinical importance is questionable, e.g. an improvement of 24 m on the 6-min walk test does not exceed the minimal clinically important difference (50 m (73)). We have to consider that interventions investigated to date are treadmill- or footplate-based. This means that massed practice of the same action is provided while the device controls balance via a supporting harness and gait via the pre-set belt speed (74, 75). Consequently, the patient is simply exposed to repetitive monotonous movement. However, animal models established that it is not such exposure to movement, but skill learning, that guides neuroplasticity (50). This suggests that stroke rehabilitation requires a whole different concept of RAGT, where the patient is constantly challenged in engaging environments and through variable practice (59). The introduction of mobile exoskeletons might enable the combination of high-dose practice through robotic assistance with the challenging nature of over-ground walking, since the patient has to actively initiate each step and control their balance, meaning that every step taken is treated as a novel problem to solve (59, 74). However, while this intervention may be promising, by exposing the patient to a learning environment, research on the usage of such devices is just beginning.

The need to change the current scientific approach

Only 15 studies, of which 3 are dependent follow-up studies, met the inclusion criteria. Those are mostly phase I or II trials with small sample sizes. Therefore, this review agrees with Stinear et al. (76), who found that less than 10% of clinical trials are initiated in the first 30 days post-stroke (76). A priority shift in research towards the first weeks is required (15, 62).

This research requires a new approach (62, 77, 78). Stratification seems important, since a growing body of evidence suggests that not all patients have the same potential to recover (6, 79). Using prognostic models will help to discriminate between these groups and to identify those patients who are most likely to benefit (80), e.g. by assessing muscle strength of the paretic leg when enrolling participants (6). In addition, our quantitative analysis is based on post-intervention data, which means that the process of recovery is measured as a single outcome score assessed on an arbitrary time-point. Considering that such trials are taking place in the background of spontaneous neurological recovery, a time-dependent process responsible for the majority of improvements on both body function and activity level (63, 81), recruitment and assessments of participants should be performed repetitively and at fixed time-points (77). This allows us to encapsulate the process rather than simply the outcome of recovery. Besides that, the majority of trials describe characteristics of the interventions poorly. A documentation on the dose in terms of repetitions is a far more accurate outcome compared with time scheduled for therapy (23, 53) and would allow us to quantify the treatment contrast between groups to analyse a dose-response relationship in more detail (53). This review highlights the great need to shift the selection of outcome measures from scales simply measuring task accomplishment to those measuring the quality of movement, to gather evidence on how patients improved when engaged in repetitive task practice (14, 15, 63). Taken together, not only a priority shift toward the first weeks is required in rehabilitation research, but also a corresponding shift in methodology with a need for more precision in our trials (77).

Conclusion

In total, 15 eligible studies were identified, which are in general pilot studies with small sample sizes. Consequently, well-designed motor rehabilitation trials starting in the first month post-stroke remain scarce. Repetitive gait training appears feasible and safe. Such training can lead to long-term functional improvements if provided early, but these effects are small. In sub-analyses, RAGT provided with an end-effector appears most effective and it seems that the more impaired patients benefit most. However, analyses on body function level yielded neutral effects and consequently the mechanisms underlying functional gains achieved after augmented gait training remain poorly understood. In the context of walking recovery after stroke, this review suggests that clinical research on early motor rehabilitation and robot-assisted training is still in its infancy.

Acknowledgement

The author would like to thank Professor Gert Kwakkel for scientific guidance and his support with the interpretation and documentation of results.

The authors have no conflicts of interest to declare.

References
  1. The author would like to thank Professor Gert Kwakkel for scientific guidance and his support with the interpretation and documentation of results.
    View article    Google Scholar
  2. The authors have no conflicts of interest to declare.
    View article    Google Scholar
  3. REFERENCES
    View article    Google Scholar
  4. Truelsen T, Piechowski-Jozwiak B, Bonita R, Mathers C, Bogousslavsky J, Boysen G. Stroke incidence and prevalence in Europe: a review of available data. Eur J Neurol 2006; 13: 581–598.
    View article    Google Scholar
  5. Gustavsson A, Svensson M, Jacobi F, Allgulander C, Alonso J, Beghi E, et al. Cost of disorders of the brain in Europe 2010. Eur Neuropsychopharmacol 2011; 21: 718–779.
    View article    Google Scholar
  6. Joo H, George MG, Fang J, Wang G. A literature review of indirect costs associated with stroke. J Stroke Cerebrovasc Dis 2014; 23: 1753–1763.
    View article    Google Scholar
  7. Lord SE, McPherson K, McNaughton HK, Rochester L, Weatherall M. Community ambulation after stroke: how important and obtainable is it and what measures appear predictive? Arch Phys Med Rehabil 2004; 85: 234–239.
    View article    Google Scholar
  8. Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS. Recovery of walking function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil 1995; 76: 27–32.
    View article    Google Scholar
  9. Veerbeek JM, Van Wegen EE, Harmeling-Van der Wel BC, Kwakkel G, Investigators E. Is accurate prediction of gait in nonambulatory stroke patients possible within 72 hours poststroke? The EPOS study. Neurorehabil Neural Repair 2011; 25: 268–274.
    View article    Google Scholar
  10. Hill K, Ellis P, Bernhardt J, Maggs P, Hull S. Balance and mobility outcomes for stroke patients: a comprehensive audit. Australian J Physiother 1997; 43: 173–180.
    View article    Google Scholar
  11. Mayo NE, Wood-Dauphinee S, Ahmed S, Gordon C, Higgins J, McEwen S, et al. Disablement following stroke. Disabil Rehabil 1999; 21: 258–268.
    View article    Google Scholar
  12. Pollock A, Baer G, Campbell P, Choo PL, Forster A, Morris J, et al. Physical rehabilitation approaches for the recovery of function and mobility following stroke. Cochrane Database Syst Rev 2014: CD001920.
    View article    Google Scholar
  13. Veerbeek JM, van Wegen E, van Peppen R, van der Wees PJ, Hendriks E, Rietberg M, et al. What is the evidence for physical therapy poststroke? A systematic review and meta-analysis. PLoS One 2014; 9: e87987.
    View article    Google Scholar
  14. Ng KL, Gibson EM, Hubbard R, Yang J, Caffo B, O’Brien RJ, et al. Fluoxetine maintains a state of heightened responsiveness to motor training early after stroke in a mouse model. Stroke 2015; 46: 2951–2960.
    View article    Google Scholar
  15. Biernaskie J, Chernenko G, Corbett D. Efficacy of rehabilitative experience declines with time after focal ischemic brain injury. J Neurosci 2004; 24: 1245–1254.
    View article    Google Scholar
  16. Murphy TH, Corbett D. Plasticity during stroke recovery: from synapse to behaviour. Nat Rev Neurosci 2009; 10: 861–872.
    View article    Google Scholar
  17. Zeiler SR, Krakauer JW. The interaction between training and plasticity in the poststroke brain. Curr Opin Neurol 2013; 26: 609–616.
    View article    Google Scholar
  18. Krakauer JW, Carmichael ST, Corbett D, Wittenberg GF. Getting neurorehabilitation right: what can be learned from animal models? Neurorehabil Neural Repair 2012; 26: 923–931.
    View article    Google Scholar
  19. Kollen B, van de Port I, Lindeman E, Twisk J, Kwakkel G. Predicting improvement in gait after stroke: a longitudinal prospective study. Stroke 2005; 36: 2676–2680.
    View article    Google Scholar
  20. Kwakkel G, van Peppen R, Wagenaar RC, Wood Dauphinee S, Richards C, Ashburn A, et al. Effects of augmented exercise therapy time after stroke: a meta-analysis. Stroke 2004; 35: 2529–2539.
    View article    Google Scholar
  21. Mehrholz J, Thomas S, Werner C, Kugler J, Pohl M, Elsner B. Electromechanical-assisted training for walking after stroke. Cochrane Database Syst Rev 2017; 5: Cd006185.
    View article    Google Scholar
  22. Horn SD, DeJong G, Smout RJ, Gassaway J, James R, Conroy B. Stroke rehabilitation patients, practice, and outcomes: is earlier and more aggressive therapy better? Arch Phys Med Rehabil 2005; 86: S101–S114.
    View article    Google Scholar
  23. Askim T, Bernhardt J, Salvesen O, Indredavik B. Physical activity early after stroke and its association to functional outcome 3 months later. J Stroke Cerebrovasc Dis 2014; 23: e305–312.
    View article    Google Scholar
  24. Lang CE, Macdonald JR, Reisman DS, Boyd L, Jacobson Kimberley T, Schindler-Ivens SM, et al. Observation of amounts of movement practice provided during stroke rehabilitation. Arch Phys Med Rehabil 2009; 90: 1692–1698.
    View article    Google Scholar
  25. Tyson SF, Woodward-Nutt K, Plant S. How are balance and mobility problems after stroke treated in England? An observational study of the content, dose and context of physiotherapy. Clin Rehabil 2018; 32: 1145–1152.
    View article    Google Scholar
  26. Lang CE, Lohse KR, Birkenmeier RL. Dose and timing in neurorehabilitation: prescribing motor therapy after stroke. Curr Opin Neurol 2015; 28: 549–555.
    View article    Google Scholar
  27. Langhorne P, Wu O, Rodgers H, Ashburn A, Bernhardt J. A Very Early Rehabilitation Trial after stroke (AVERT): a phase III, multicentre, randomised controlled trial. Health Technology Assessment (Winchester, England) 2017; 21: 1–120.
    View article    Google Scholar
  28. Kozlowski DA, James DC, Schallert T. Use-dependent exaggeration of neuronal injury after unilateral sensorimotor cortex lesions. J Neurosci 1996; 16: 4776–4786.
    View article    Google Scholar
  29. Kline TL, Schmit BD, Kamper DG. Exaggerated interlimb neural coupling following stroke. Brain 2007; 130: 159–169.
    View article    Google Scholar
  30. Czernuszenko A, Czlonkowska A. Risk factors for falls in stroke patients during inpatient rehabilitation. Clin Rehabil 2009; 23: 176–188.
    View article    Google Scholar
  31. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol 2009; 62: e1–e34.
    View article    Google Scholar
  32. Truelsen T, Mahonen M, Tolonen H, Asplund K, Bonita R, Vanuzzo D, et al. Trends in stroke and coronary heart disease in the WHO MONICA Project. Stroke 2003; 34: 1346–1352.
    View article    Google Scholar
  33. Holden MK, Gill KM, Magliozzi MR, Nathan J, Piehl-Baker L. Clinical gait assessment in the neurologically impaired. Reliability and meaningfulness. Phys Ther 1984; 64: 35–40.
    View article    Google Scholar
  34. French B, Thomas LH, Coupe J, McMahon NE, Connell L, Harrison J, et al. Repetitive task training for improving functional ability after stroke. Cochrane Database Syst Rev 2016; 11: CD006073.
    View article    Google Scholar
  35. World Health Organization. The International Classification of Functioning, Disability and Health. Geneva: WHO; 2015.
    View article    Google Scholar
  36. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol 2014; 14: 135.
    View article    Google Scholar
  37. Harbour R, Miller J. A new system for grading recommendations in evidence based guidelines. BMJ 2001; 323: 334–336.
    View article    Google Scholar
  38. Pohl M, Werner C, Holzgraefe M, Kroczek G, Mehrholz J, Wingendorf I, et al. Repetitive locomotor training and physiotherapy improve walking and basic activities of daily living after stroke: a single-blind, randomized multicentre trial (DEutsche GAngtrainerStudie, DEGAS). Clin Rehabil 2007; 21: 17–27.
    View article    Google Scholar
  39. Ada L, Dean CM, Morris ME, Simpson JM, Katrak P. Randomized trial of treadmill walking with body weight support to establish walking in subacute stroke: the MOBILISE trial. Stroke 2010; 41: 1237–1242.
    View article    Google Scholar
  40. Dean CM, Ada L, Bampton J, Morris ME, Katrak PH, Potts S. Treadmill walking with body weight support in subacute non-ambulatory stroke improves walking capacity more than overground walking: a randomised trial. J Physiother 2010; 56: 97–103.
    View article    Google Scholar
  41. Chang WH, Kim MS, Huh JP, Lee PK, Kim YH. Effects of robot-assisted gait training on cardiopulmonary fitness in subacute stroke patients: a randomized controlled study. Neurorehabil Neural Repair 2012; 26: 318–324.
    View article    Google Scholar
  42. Tong RK, Ng MF, Li LS. Effectiveness of gait training using an electromechanical gait trainer, with and without functional electric stimulation, in subacute stroke: a randomized controlled trial. Arch Phys Med Rehabil 2006; 87: 1298–1304.
    View article    Google Scholar
  43. Chua J, Culpan J, Menon E. Efficacy of an electromechanical gait trainer poststroke in Singapore: a randomized controlled trial. Arch Phys Med Rehabil 2016; 97: 683–690.
    View article    Google Scholar
  44. Nilsson L, Carlsson J, Danielsson A, Fugl-Meyer A, Hellstrom K, Kristensen L, et al. Walking training of patients with hemiparesis at an early stage after stroke: a comparison of walking training on a treadmill with body weight support and walking training on the ground. Clin Rehabil 2001; 15: 515–527.
    View article    Google Scholar
  45. Ng MF, Tong RK, Li LS. A pilot study of randomized clinical controlled trial of gait training in subacute stroke patients with partial body-weight support electromechanical gait trainer and functional electrical stimulation: six-month follow-up. Stroke 2008; 39: 154–160.
    View article    Google Scholar
  46. Morone G, Bragoni M, Iosa M, De Angelis D, Venturiero V, Coiro P, et al. Who may benefit from robotic-assisted gait training? A randomized clinical trial in patients with subacute stroke. Neurorehabil Neural Repair 2011; 25: 636–644.
    View article    Google Scholar
  47. Morone G, Iosa M, Bragoni M, De Angelis D, Venturiero V, Coiro P, et al. Who may have durable benefit from robotic gait training? A 2-year follow-up randomized controlled trial in patients with subacute stroke. Stroke 2012; 43: 1140–1142.
    View article    Google Scholar
  48. Schwartz I, Sajin A, Fisher I, Neeb M, Shochina M, Katz-Leurer M, et al. The effectiveness of locomotor therapy using robotic-assisted gait training in subacute stroke patients: a randomized controlled trial. PM R 2009; 1: 516–523.
    View article    Google Scholar
  49. Ochi M, Wada F, Saeki S, Hachisuka K. Gait training in subacute non-ambulatory stroke patients using a full weight-bearing gait-assistance robot: a prospective, randomized, open, blinded-endpoint trial. J Neurolog Sci 2015; 353: 130–136.
    View article    Google Scholar
  50. Franceschini M, Carda S, Agosti M, Antenucci R, Malgrati D, Cisari C. Walking after stroke: what does treadmill training with body weight support add to overground gait training in patients early after stroke? A single-blind, randomized, controlled trial. Stroke 2009; 40: 3079–3085.
    View article    Google Scholar
  51. Peurala SH, Airaksinen O, Huuskonen P, Jakala P, Juhakoski M, Sandell K, et al. Effects of intensive therapy using gait trainer or floor walking exercises early after stroke. J Rehabil Med 2009; 41: 166–173.
    View article    Google Scholar
  52. Han EY, Im SH, Kim BR, Seo MJ, Kim MO. Robot-assisted gait training improves brachial-ankle pulse wave velocity and peak aerobic capacity in subacute stroke patients with totally dependent ambulation: randomized controlled trial. Medicine 2016; 95: e5078.
    View article    Google Scholar
  53. Nudo RJ. Recovery after brain injury: mechanisms and principles. Front Hum Neurosci 2013; 7: 887.
    View article    Google Scholar
  54. Remple MS, Bruneau RM, VandenBerg PM, Goertzen C, Kleim JA. Sensitivity of cortical movement representations to motor experience: evidence that skill learning but not strength training induces cortical reorganization. Behav Brain Res 2001; 123: 133–141.
    View article    Google Scholar
  55. Cha J, Heng C, Reinkensmeyer DJ, Roy RR, Edgerton VR, De Leon RD. Locomotor ability in spinal rats is dependent on the amount of activity imposed on the hindlimbs during treadmill training. J Neurotrauma 2007; 24: 1000–1012.
    View article    Google Scholar
  56. Lohse KR, Lang CE, Boyd LA. Is more better? Using metadata to explore dose-response relationships in stroke rehabilitation. Stroke 2014; 45: 2053–2058.
    View article    Google Scholar
  57. Bernhardt J, Dewey H, Thrift A, Donnan G. Inactive and alone: physical activity within the first 14 days of acute stroke unit care. Stroke 2004; 35: 1005–1009.
    View article    Google Scholar
  58. Astrand A, Saxin C, Sjoholm A, Skarin M, Linden T, Stoker A, et al. Poststroke physical activity levels no higher in rehabilitation than in the acute hospital. J Stroke Cerebrovasc Dis 2016; 25: 938–945.
    View article    Google Scholar
  59. Rand D, Eng JJ. Disparity between functional recovery and daily use of the upper and lower extremities during subacute stroke rehabilitation. Neurorehabil Neural Repair 2012; 26: 76–84.
    View article    Google Scholar
  60. Morone G, Paolucci S, Cherubini A, De Angelis D, Venturiero V, Coiro P, et al. Robot-assisted gait training for stroke patients: current state of the art and perspectives of robotics. Neuropsychiatr Dis Treat 2017; 13: 1303–1311.
    View article    Google Scholar
  61. Moore JL, Roth EJ, Killian C, Hornby TG. Locomotor training improves daily stepping activity and gait efficiency in individuals poststroke who have reached a “plateau” in recovery. Stroke 2010; 41: 129–135.
    View article    Google Scholar
  62. Kitago T, Krakauer JW. Motor learning principles for neurorehabilitation. Handb Clin Neurol 2013; 110: 93–103.
    View article    Google Scholar
  63. Patel AT, Duncan PW, Lai SM, Studenski S. The relation between impairments and functional outcomes poststroke. Arch Phys Med Rehabil 2000; 81: 1357–1363.
    View article    Google Scholar
  64. Levin MF, Kleim JA, Wolf SL. What do motor “recovery” and “compensation” mean in patients following stroke? Neurorehabil Neural Repair 2009; 23: 313–319.
    View article    Google Scholar
  65. Bernhardt J, Hayward KS, Kwakkel G, Ward NS, Wolf SL, Borschmann K, et al. Agreed definitions and a shared vision for new standards in stroke recovery research: The Stroke Recovery and Rehabilitation Roundtable Taskforce. Neurorehabil Neural Repair 2017; 31: 793–799.
    View article    Google Scholar
  66. Buma F, Kwakkel G, Ramsey N. Understanding upper limb recovery after stroke. Restorative Neurol Neurosci 2013; 31: 707–722.
    View article    Google Scholar
  67. Kwakkel G, Lannin NA, Borschmann K, English C, Ali M, Churilov L, et al. Standardized measurement of sensorimotor recovery in stroke trials: consensus-based core recommendations from the Stroke Recovery and Rehabilitation Roundtable. Int J Stroke 2017; 12: 451–461.
    View article    Google Scholar
  68. van Asseldonk EH, Buurke JH, Bloem BR, Renzenbrink GJ, Nene AV, van der Helm FC, et al. Disentangling the contribution of the paretic and non-paretic ankle to balance control in stroke patients. Exp Neurol 2006; 201: 441–451.
    View article    Google Scholar
  69. de Haart M, Geurts AC, Huidekoper SC, Fasotti L, van Limbeek J. Recovery of standing balance in postacute stroke patients: a rehabilitation cohort study. Arch Phys Med Rehabil 2004; 85: 886–895.
    View article    Google Scholar
  70. Roerdink M, Geurts AC, de Haart M, Beek PJ. On the relative contribution of the paretic leg to the control of posture after stroke. Neurorehabil Neural Repair 2009; 23: 267–274.
    View article    Google Scholar
  71. Bowden MG, Balasubramanian CK, Neptune RR, Kautz SA. Anterior-posterior ground reaction forces as a measure of paretic leg contribution in hemiparetic walking. Stroke 2006; 37: 872–876.
    View article    Google Scholar
  72. Den Otter AR, Geurts AC, Mulder T, Duysens J. Gait recovery is not associated with changes in the temporal patterning of muscle activity during treadmill walking in patients with post-stroke hemiparesis. Clin Neurophysiol 2006; 117: 4–15.
    View article    Google Scholar
  73. Buurke JH, Nene AV, Kwakkel G, Erren-Wolters V, Ijzerman MJ, Hermens HJ. Recovery of gait after stroke: what changes? Neurorehabil Neural Repair 2008; 22: 676–683.
    View article    Google Scholar
  74. Bowden MG, Behrman AL, Woodbury M, Gregory CM, Velozo CA, Kautz SA. Advancing measurement of locomotor rehabilitation outcomes to optimize interventions and differentiate between recovery versus compensation. J Neurolog Phys Ther 2012; 36: 38–44.
    View article    Google Scholar
  75. Kwakkel G, Kollen B, Lindeman E. Understanding the pattern of functional recovery after stroke: facts and theories. Restorative Neurol Neurosci 2004; 22: 281–299.
    View article    Google Scholar
  76. Perera S, Mody SH, Woodman RC, Studenski SA. Meaningful change and responsiveness in common physical performance measures in older adults. J Am Geriatr Soc 2006; 54: 743–749.
    View article    Google Scholar
  77. Louie DR, Eng JJ. Powered robotic exoskeletons in post-stroke rehabilitation of gait: a scoping review. J Neuroeng Rehabil 2016; 13: 53.
    View article    Google Scholar
  78. Hidler J, Nichols D, Pelliccio M, Brady K, Campbell DD, Kahn JH, et al. Multicenter randomized clinical trial evaluating the effectiveness of the Lokomat in subacute stroke. Neurorehabil Neural Repair 2009; 23: 5–13.
    View article    Google Scholar
  79. Stinear C, Ackerley S, Byblow W. Rehabilitation is initiated early after stroke, but most motor rehabilitation trials are not: a systematic review. Stroke 2013; 44: 2039–2045.
    View article    Google Scholar
  80. Winters C, Kwakkel G, van Wegen EEH, Nijland RHM, Veerbeek JM, Meskers CGM. Moving stroke rehabilitation forward: the need to change research. NeuroRehabilitation 2018; 43: 19–30.
    View article    Google Scholar
  81. Stinear CM. Stroke rehabilitation research needs to be different to make a difference. F1000Res 2016; 5.
    View article    Google Scholar
  82. Veerbeek JM, Winters C, van Wegen EEH, Kwakkel G. Is the proportional recovery rule applicable to the lower limb after a first-ever ischemic stroke? PLoS One 2018; 13: e0189279.
    View article    Google Scholar
  83. Stinear CM. Prediction of motor recovery after stroke: advances in biomarkers. Lancet Neurol 2017; 16: 826–836.
    View article    Google Scholar
  84. Kwakkel G, Kollen B, Twisk J. Impact of time on improvement of outcome after stroke. Stroke 2006; 37: 2348–2353.
    View article    Google Scholar
Supplementary content
Figure S1-S7
Table SI-II

Comments

Do you want to comment on this paper? The comments will show up here and if appropriate the comments will also separately be forwarded to the authors. You need to login/create an account to comment on articles. Click here to login/create an account.