Authors: Joel Stein, MD Lauri Bishop, DPT Daniel J. Stein, MD, MPH Christopher Kevin Wong, PT, PhD
Gait
Affiliations: From the Department of Rehabilitation and Regenerative Medicine, Columbia University College of Physicians and Surgeons (JS, LB, CKW); Division of Rehabilitation Medicine, Weill Cornell Medical College (JS); Department of Rehabilitation Medicine, New York-Presbyterian Hospital, New York, New York (JS); and Department of Medicine, University of Virginia Medical Center, Charlottesville, Virginia (DJS).
Correspondence: All correspondence and requests for reprints should be addressed to: Joel Stein, MD, Department of Rehabilitation and Regenerative Medicine, Columbia University College of Physicians and Surgeons, Harkness Pavilion Room 1-165, New York, NY 10032.
ORIGINAL RESEARCH ARTICLE
Gait Training with a Robotic Leg Brace After Stroke A Randomized Controlled Pilot Study ABSTRACT Stein J, Bishop L, Stein DJ, Wong CK: Gait training with a robotic leg brace after stroke: a randomized controlled pilot study. Am J Phys Med Rehabil 2014;93:987Y994.
Objective: Robot-aided exercise therapy is a promising approach to enhance walking ability in stroke survivors. This study was designed to test a new robotic knee brace for restoring mobility in stroke survivors. Design: Twenty-four ambulatory individuals with chronic hemiparesis after
Disclosures: Supported by Tibion, Inc (now part of AlterG, Fremont, CA), which provided feedback on study design, but did not participate in the conduct of the study, interpretation, or reporting of the results. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
0894-9115/14/9311-0987 American Journal of Physical Medicine & Rehabilitation Copyright * 2014 by Lippincott Williams & Wilkins
stroke were enrolled in this pilot study. The participants were randomly assigned in equal numbers to either treatment with the experimental device or to a group exercise program and received a total of 18 hrs of their assigned therapy during a 6-wk training period. The primary outcome was gait velocity, as measured with the 10-m walk test. Secondary measures included 6-min walk test, Timed Up and Go test, Five-Times-Sit-to-Stand test, Romberg test, Emory Functional Ambulation Profile, Berg Balance scale, and the California Functional Evaluation 40.
Results: Twenty subjects completed the entire protocol and all follow-up visits. No significant differences between the two groups were found for the primary outcome measure at either the completion of training (week 6) or at the 3-mo follow-up (week 19), with inconsistent findings for secondary measures. No withingroup changes were seen in the primary outcome measure (10-m walk test) in either group. Within-group improvements were seen in several of the secondary measures for both groups. No complications of robotic therapy were observed.
Conclusions: Robotic therapy for ambulatory stroke patients with chronic DOI: 10.1097/PHM.0000000000000119
hemiparesis using a robotic knee brace resulted in only modest functional benefits that were comparable with a group exercise intervention. Key Words:
www.ajpmr.com
Robotics, Stroke, Hemiparesis, Gait, Exercise Therapy
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
987
S
troke affects approximately 795,000 individuals annually in the United States, with considerable resulting disability.1 Hemiparesis after stroke is a major contributor to disability, with more than 60% of stroke survivors experiencing limitations in mobility immediately after stroke. Of these, approximately 36% have continued mobility limitations on completion of their rehabilitation course.2 Mobility disturbances from hemiparesis include difficulty with sit-to-stand transfers, reduced balance, slowed and inefficient gait, falls, and damage to joints from altered biomechanics. Physical therapy has been shown to provide significant benefit, but substantial disability often persists.2 Moreover, many ambulatory stroke survivors continue to require leg braces (typically ankle foot orthoses) and/or ambulatory aids such as canes.3 Robotic devices to provide gait training have been developed, including the Lokomat (Hocoma, Inc, Volketswil, Switzerland) and the G-EO (Reha Technologies, Inc, Olten, Switzerland). These workstation-type devices provide gait training using a powered exoskeleton with treadmill (Lokomat) or powered footplates with programmable trajectories, analogous to a robotic elliptical trainer (G-EO). Results with the Lokomat have lagged those achieved with therapist provided partial body weight supported treadmill training.4 Moreover, these workstation robotic devices are large as well as costly and can only provide treatment in a rehabilitation gym setting. Wearable powered orthoses provide another approach to remediate motor deficits after stroke. Upper limb devices include electromyography-controlled powered elbow braces (Myomo, Inc, Cambridge, MA) to provide assistance to the elbow for flexion or extension.5 Surface electromyography does not perfectly indicate intended effort/force, however, and the optimal control mechanism and algorithm for providing assistance are not yet clear. A tethered lowerlimb exoskeletal device, the Anklebot (Interactive Motion Technologies, Inc, Watertown, MA), seems to improve gait parameters, even when training occurs in the seated position,6 but is not a fully wearable, autonomous device usable for community ambulation. Bilateral wearable powered exoskeletal robotic systems have been developed, including the ReWalk (Argo Medical Technologies Ltd, Yokneam Ilit, Israel) and eLegs (Ekso Bionics, Inc, Richmond, CA), but are primarily intended for use by individuals with paraplegia rather than stroke survivors with hemiparesis. AlterG (Fremont, CA; formerly Tibion, Inc) has developed the Bionic Leg, a powered knee orthosis intended to provide assistance and training for patients with unilateral neurologic or orthopedic
988
Stein et al.
conditions (Fig. 1). The device relies on multiple sensors, including pressure sensors placed in the user’s shoes, accelerometers, and joint angle detectors to create a model of the user’s activities, including sit-to-stand transfers, gait on level surfaces, and stair climbing/descent, and provides a mechanized assist with these activities through actuators at the knee joint.7 The device is fully wearable (i.e., untethered during use) and contains its own power source sufficient for several hours of use. The model used in this study weighs approximately 8 lbs (3.6 kg) and attaches with Velcro straps, requiring only 2Y3 mins to don/doff.8 The device is programmable, and a trained physical therapist establishes initial parameters and updates the device over the course of therapy.8,9 Donning and doffing the device requires some assistance from another individual, but once the device is in place, a patient could potentially use it without direct therapist supervision for home exercise between physical therapy sessions. This device has particular appeal for certain populations of hemiparetic stroke survivors with mobility impairments. Potential applications are the exercise training for the remediation of stiff-legged gait, treatment of knee hyperextension (recurvatum), difficulty with ambulation, sit-to-stand transfers, and/ or difficulty with stair climbing. The feedback provided through force feedback to the user is similar conceptually to the feedback provided during conventional gait training and is intended to help the user develop biomechanically more efficient gait through the adoption of a more normal gait pattern.
FIGURE 1 Bionic leg device in use.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
The device was well tolerated and did not cause any adverse effects in a small case series of several stroke patients with hemiparesis.8 The authors hypothesized that ambulation training with the use of the device would result in improved gait patterns more closely approximating normal gait because of the feedback and assistance provided by the device and thus result in greater gait speed and endurance. The authors therefore chose to conduct a pilot randomized controlled trial of this device in stroke survivors who were already independent in household ambulation but who had reduced gait velocity to determine the effect of physical therapy using this powered knee orthosis on gait speed and endurance. This study was intended to confirm the ability of ambulatory stroke patients with hemiparesis to successfully use the device and obtain preliminary data on efficacy to plan future clinical trials.
METHODS Subjects A total of 24 subjects were recruited from community-dwelling outpatients receiving care at the authors’ institution and/or enrolled in a registry of stroke survivors wishing to be considered for participation in clinical research. Inclusion criteria included a history of a single stroke (ischemic or hemorrhagic) causing significant leg weakness and gait alterations at least 6 mos before study entry. The stroke must have been confirmed through computed tomography or magnetic resonance imaging. Subjects with a single stroke based on clinical history but with evidence of previous asymptomatic strokes on imaging studies were considered eligible for participation. Subjects were required to be independent in household ambulation (with or without the use of a unilateral assistive device [e.g., standard cane, four-pronged cane, or hemiwalker] and with or without the use of an ankle foot orthoses). Exclusion criteria included subjects receiving ongoing physical therapy for the leg and/or gait and mobility training at the time of study entry. Subjects were permitted to have received previous botulinum toxin injections as long as such injections were received at least 3 mos before study entry and provided that no further injections were planned during the study period. Other exclusion criteria included the presence of other neurologic disorders, such as Parkinson disease, a history of more than one symptomatic stroke, excessive spasticity (defined as Ashworth scale of greater than 3 [of 4] at any lower limb joint), uncontrolled hypertension, unstable coronary artery disease, conwww.ajpmr.com
tractures of either lower limb, impaired cognition defined by a Folstein MiniYMental State Examination score below 24, or other medical conditions that might interfere with the subject’s ability to complete the study. Subject enrollment and flow is shown in Figure 2. No formal power analysis was performed because of insufficient information about anticipated effect size. The subjects were randomized in blocks of six using a 50:50 concealed allocation to either the robotic treatment group or the exercise control group.
Training Regimen The subjects in the robotic treatment group received 1 hr of individualized physical therapy with the device 3 days per week for six consecutive weeks (a total of 18 sessions). Each 1-hr session included brief rest periods, with a goal of administering approximately 50 mins of active therapy. Therapy sessions were conducted by an experienced physical therapist trained in the use of the device. Sessions consisted of a set of standardized over-ground functional tasks including sit-to-stand transfers from variable height with and without chair arms; standing balance while turning, reaching, stepping, and functional activities; gait training on level ground and surfaces; and mobility on stairs and curbs. All activities were performed to the tolerance and abilities of the subjects and supervised and assisted by a physical therapist present throughout the training session. The subjects assigned to the exercise control group received 1 hr of group exercise without the use of the robotic device 3 days/wks for six consecutive weeks (a total of 18 sessions). The exercise intervention was designed to control for the effects of interacting with a physical therapist by providing an equivalent number of formal therapy sessions while providing minimal gait training. Group exercise sessions included 3 subjects per treatment session. Therapy sessions were conducted by the same unblinded physical therapist treating the robotic group. Activities included relaxation/meditation, self-stretching, and gentle upper and lower limb active range of motion exercises. The subjects spent approximately 5 mins of the session walking at a comfortable walking pace (two 2.5-min walks during the 1-hr session).
Outcome Measures The subjects underwent functional gait and balance assessments on 4 separate occasions: at study entry, after completion of the 6-wk training program (week 6), as well as at 1 and 3 mos after completion of the training program (weeks 10 and 19). All Gait Training with Robotics After Stroke
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
989
FIGURE 2 Subject enrollment and flow.
assessments were performed by one physical therapist blinded to group assignment. Baseline and follow-up assessments of functional gait and balance performance included the Timed Up and Go (TUG),10 10-m walk test,11 6-min walk test,11 Five-Times-Sit-to-Stand test,12 the Berg Balance scale,13 California Functional Evaluation 40,14 and the Emory Functional Ambulation Profile (EFAP).15,16 Gait velocity was classified as household ambulation (G0.4 m/sec), limited community ambulation (0.4Y0.8 m/sec), and full community ambulation (90.8 m/sec) to determine how many subjects improved from one class to the next.17 The 10-m walk test measures gait velocity and was designated as the primary outcome measure. It was performed on a 14-m walkway, and measurement was performed for the middle 10 m on this
990
Stein et al.
walkway. Three trials were timed by stopwatch and averaged. The 6-min walk test measures a combination of gait endurance and velocity. The TUG and Five-Times-Sit-to-Stand test are measures of transfer ability, with the TUG including a component of gait velocity as well. The Berg Balance scale measures balance. The California Functional Evaluation 40 and EFAP are assessments of overall gait functional ability. Unpaired t tests were used to compare baseline characteristics of the two groups, except in cases where tests of normality failed, when Mann-Whitney rank sum tests were performed. Paired t tests were used for an intention-to-treat analysis of change scores at the conclusion of treatment. A linear mixed model was used to analyze change over time in perprotocol subjects to account for the correlation between the multiple time points inherent in repeated
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
measures. This allowed consideration of all available time points when determining the effect of the intervention. A constant correlation (exchangeable) model was found to provide the best fit and used for all analyses. The authors used SAS version 9.3 for statistical analyses (SAS Institute, Cary, NC). The study was approved by the institutional review board. All subjects provided informed consent.
RESULTS Thirty-four subjects underwent in-person screening for participation, of whom 24 subjects were enrolled, with 12 subjects allocated to each treatment group. The most frequent reasons prospective subjects were excluded from the study were ongoing physical therapy, recent botulinum toxin injection in the lower limb, excessive spasticity, and low score on the Folstein MiniYMental State Examination. Baseline subject characteristics were similar between the two treatment groups (see Table 1). All subjects completed the training regimen, and there were no adverse effects of either therapy. One subject in the exercise control group sustained a fall during his 1-mo follow-up assessment, resulting in a nonYdisplaced fracture of the fifth metatarsal bone. One subject in the robotic group had a second stroke between the 1-mo (week 10) and 3-mo (week 19) follow-up visits that was deemed unrelated to the study treatment. One additional subject in the control exercise group failed to complete the 1- and 3-mo follow-up, and one additional subject in the robotic treatment group was lost to follow-up before the 3-mo follow-up assessment.
No significant differences were found between the two groups for the primary outcome measure at the completion of training (week 6) using paired t test analyses (n = 24). Results for the Berg Balance scale at completion of training (week 6) favored the robotic treatment group; conversely, results for the EFAP at completion of training favored the control exercise group. The remainder of secondary measures showed no difference between the two groups at completion of training. A per-protocol analysis of the 20 subjects who completed all of the assessments found no significant differences between the two groups using a linear mixed model through the 3-mo follow-up period for either primary or secondary measures (see Table 2 for a summary of outcome measurements). The authors also examined the number of subjects who advanced from one gait velocity classification (e.g., household ambulation, G0.4 m/sec) to the next and found no statistically significant change from either intervention (see Table 3). Within-group analyses revealed no change in the primary outcome measure for either group from baseline (week 0) to the completion of treatment (week 6) or the 3-mo follow-up (week 19). Several secondary measures demonstrated within-group improvements at these two time points, as shown in Table 2.
DISCUSSION No differences in the primary outcome measure of gait velocity were demonstrated between the two groups, nor were there any within-group
TABLE 1 Baseline subject characteristics
Subjects Mean age Percentage (male) Duration after stroke, mean (SD), mos Folstein MiniYMental State Examination score, mean (SD) Baseline 10-m walk test, mean (SD) 6-min walk test, mean (SD) 5-times-sit-to-stand test, secs TUG, secs Berg Balance scale CAFE´ 40 EFAP Romberg, eyes open Romberg, eyes closed
Robotic Training Group (SD)
Exercise Group (SD)
P
12 57.6 (10.7) 83% 49.1 (38.9) 28.8 (1.2)
12 56.6 (15.1) 58% 88.5 (153.0) 29.2 (0.9)
0.85 0.37 1.0a 0.46
22.9 (20.0) 185.9 (95.9) 31.6 (26.4) 22.7 (9.7) 46.1 (11.7) 135.2 (38.8) 91.7 (42.5) 14.3 (14.6) 2.5 (3.5)
27.5 (21.2) 169.3 (89.6) 24.1 (9.8) 33.7 (26.4) 49.4 (5.8) 137.8 (34.6) 131.8 (93.2) 20.4 (14.2) 7.4 (9.7)
0.40a 0.67 0.39 0.21 0.39 0.87 0.21 0.28a 0.26a
a Mann-Whitney rank sum test. CAFE´ 40, California Functional Evaluation 40.
www.ajpmr.com
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
991
TABLE 2 Outcome measures
Baseline (Week 0)
10-m walk test, secs
Completion of Training (Week 6)
1-mo Follow-up (Week 10)
3-mo Follow-up (Week 19)
Robotic Therapy (n = 12)
Exercise Control (n = 12)
Robotic Therapy (n = 12)
Exercise Control (n = 12)
Robotic Therapy (n = 12)
Exercise Control (n = 10)
Robotic Therapy (n = 10)
Exercise Control (n = 10)
22.9 (20.0)
27.5 (21.2)
20.3 (15.0)
19.2 (9.1)
20.5 (16.3)
21.0 (12.4)
15.1 (7.0)
20.2 (11.2)
6-min walk test, m 185.9 (95.9) 169.3 (89.7) 213. 4 (108.2) 194.8 (83.2) 217.8 (121.1) 203.8 (87.1) 252.7 (108.4)
214 (94.7)
5-times-sit-to-stand test, secs TUG, secs
31.6 (26.4)
24.1 (9.8)
25.3 (16.7)
18.8 (6.1)
19.4 (8.3)
16.5 (6.8)
26.3 (32.1)
15.8 (4.3)
22.7 (9.7)
33.7 (26.4)
26.7 (16.9)
28.0 (17.1)
24.8 (16.8)
26.8 (18.4)
18.4 (7.9)
25.5 (15.5)
Berg Balance scale
46.1 (11.7)
49.4 (5.8)
48.6 (10.2)
49.2 (5.8)
50.4 (8.2)
51.6 (5.4)
52.4 (5.4)
52.5 (4.7)
CAFE´ 40
135.2 (38.8) 137.8 (34.6) 148.8 (31.7)
EFAP
91.7 (42.5) 131.8 (93.2) 115.1 (84.7)
Romberg, eyes open, secs Romberg, eyes closed, secs
14.3 (14.6)
20.4 (14.2)
2.5 (3.5)
7.4 (9.7)
137.3 (34.5) 149.8 (30.5)
148.9 (40.3) 161.1 (37.2)
153.8 (37.3)
101.7 (55.4)
82.7 (31.4)
98.4 (53.3)
76.2 (30.8)
97.8 (51.1)
23.9 (10.3)
22.2 (11.0)
23.1 (12.5)
23.8 (10.6)
25.7 (9.9)
22.9 (11.7)
7.4 (9.4)
11.7 (12.6)
9.8 (11.3)
12.3 (12.9)
14.6 (13.5)
13.9 (12.1)
a
Linear mixed model. Mann-Whitney rank sum test. c Statistically significant. CI, confidence interval. b
changes for this measure. Secondary measures provided conflicting results, with one measure (Berg Balance score) favoring robotic therapy, another (EFAP) favoring the exercise control group, and the remainder of the secondary measures without any difference between the two groups. The authors did, however, find within-group improvements in walking endurance and in several other secondary measures. In the aggregate, these findings support the use of exercise therapy to improve mobility in individuals with chronic hemiparesis but do not suggest an advantage to robotic leg brace for this purpose. There are several potential explanations for the lack of any incremental benefit of robot-aided training over group therapy. One potential explanation for the authors’ findings is that the robotic device used and/or the training regimen implemented does not provide the optimal training experience for this population. It is possible that the dose of gait training provided was inadequate and that training at a faster cadence, more intensive training (more hours per week), or longer-duration training (more total weeks of train-
992
Stein et al.
ing) would have provided greater benefit. Although the authors did not measure the number of steps taken during each therapy session, these were likely fewer than achievable with a treadmill-based system because the device functions at the self-selected cadence of the user rather than at a faster rate used in some treadmill training protocols. The intensity and duration of the training in the authors’ study were selected arbitrarily on the basis of logistic considerations rather than on the evidence that this protocol is optimal. Another possible explanation of the authors’ findings might be that walking itself, rather than therapy involving treadmills, robots, or other devices, is the most effective therapy for improving gait in ambulatory stroke survivors. Other studies showing better results18 or comparable outcomes19 for conventional gait training compared with robotic or partial body weightYsupported treadmill training may support this argument. The authors used a control intervention that was intended to provide a comparable amount of time in a therapeutic environment interacting with
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Difference Within Groups Mean Change from Baseline (Week 0) to Completion of Training (Week 6) Exercise Control (95% CI)
Robotic Therapy (95% CI) 2.6 (j1.6 to 6.8) P = 0.20 27.5 (j5.4 to 60.4) P = 0.09 6.3 (j3.2 to 15.8) P = 0.17 0.4 (j3.7 to 4.5) P = 0.85 2.5 (0.9Y4.1) P = 0.005c 13.7 (j3.7 to 31.0) P = 0.11 1.2 (j10.2 to 12.5) P = 0.83 9.6 (j0.5 to 19.7) P = 0.06 4.9, P = 0.049bc
Difference Between Groups
Mean Change from Baseline (Week 0) to 3-mo (Week 19) Follow-up Robotic Therapy
Baseline (Week 0) to Baseline (Week 0) Completion (Week 6) to 3-mo (Week 19) of Training Follow-upa
Exercise Control
8.3 (j0.6 to 17.3) P = 0.06 25.5 (6.6Y44.4) P = 0.01c 5.3 (2.7Y8.9) P = 0.08 5.7 (j1.0 to 12.4) P = 0.09 0.3 (j2.0 to 2.5) P = 0.81 4.6 (j4.1 to 13.4) P = 0.27 29.7, P = 0.005bc
0.5 (j3.7 to 4.6) P = 0.81 42.1 (j2.1 to 86.2) P = 0.06 3.7 (j6.1 to 13.5) P = 0.41 2.7 (j0.2 to 5.6) P = 0.07 2.1 (j0.6 to 4.8) P = 0.11 18.0 (j2.2 to 38.2) P = 0.08 7.9 (0.8Y15.0) P = 0.03c 10.2 (j3.9 to 24.3) P = 1.0b P = 0.14 3.6 (j3.4 to 10.6) 11.6 (j1.8 to 21.5) P = 0.27 P = 0.03c
5.7 (j3.9 to 15.2) 5.8, P = 0.13b P = 0.22 28 (11.9Y37.5) 2.0 (j33.7 to 37.7), P = 0.038c P = 0.91 6.5, P = 0.04bc 0.9 (j9.3 to 11.2) P = 0.85 5.3 (j2.3 to 12.9) 6.3, P = 0.28b P = 0.16 2.3, P = 0.19b 2.8 (0.13Y5.4) P = 0.04c 9.0 (j9.8 to 27.9) 18.1 (5.0Y31.2) P = 0.33 P = 0.01c 30.6, P = 0.049bc 30.8, P = 0.009 bc
P = 0.18
1.4, P = 0.88b
8.0, P = 0.25b
P = 0.65
5.2 (0.03Y10.4) P = 0.049c
1.7, P = 0.60b
P = 0.67
P = 0.38 P = 0.24 P = 0.22 P = 0.64 P = 0.88 P = 0.25
of the study do not allow the authors to determine what portion of the improvement in gait endurance resulted from treatment sessions vs. a nonYspecific effect of study participation, a further examination of this hypothesis would be worthwhile. The improvements seen in the study’s control group emphasize the impact of participation in a clinical trial and contact time with a physical therapist on performance even in the absence of goal-directed gait training and should inform design of control interventions in future studies of rehabilitation therapies. One strategy to provide greater amounts of therapy would be to provide a home-based program
a physical therapist while minimizing the amount of formal gait training provided. The use of a more conventional physical therapy program focusing on gait training might have provided different results, perhaps even superior to the robotic leg brace intervention used in the study’ protocol. The authors speculate that some portion of the improvement in walking endurance seen may have resulted from a nonYspecific effect of participating in a clinical trial. The frequent visits to the authors’ facility for treatment sessions represent an opportunity to increase activity levels simply by the act of traveling to and from the treatment sessions. Although the data TABLE 3 Gait velocity classification
Gait Velocity G0.4 m/sec
0.4Y0.8 m/sec
90.8 m/sec
0 0 2 2
5 7 5 5
5 3 3 3
Robotic therapy group (baseline) Robotic therapy group (after therapy) Exercise control group (baseline) Exercise control group (after therapy) Number of subjects falling within each gait velocity class.
www.ajpmr.com
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
993
to supplement therapist-supervised training. Planned enhancements in the device that will allow retrieval of activity data would permit a more rigorous quantification of therapy dose in the future and allow the determination of a dose-response curve. All of subjects of this study were ambulatory for at least household distances independently at study entry and may not have been the ideal target population for robot-aided therapy. It is possible that this form of robotic therapy may prove more effective for individuals who lack the ability to walk independently and who would therefore benefit from the physical assistance provided by the robot. Moreover, there may be a window of opportunity earlier after stroke during which the recovery curve may be altered favorably through the incorporation of robotaided rehabilitation. Because this study focused exclusively on ambulatory stroke patients with chronic stable deficits, further study with nonYambulatory stroke patients during the early phase of gait recovery after stroke would be required to test this hypothesis. In conclusion, no overall difference in outcome was found between the robot-assisted and control group exercise therapies. The subjects within each group showed no change in gait velocity but did experience improvements in gait endurance and some other secondary measures. REFERENCES
6. Forrester LW, Roy A, Krebs HI, et al: Ankle training with a robotic device improves hemiparetic gait after a stroke. Neurorehabil Neural Repair 2011;25: 369Y77 7. Horst RW: A bio-robotic leg orthosis for rehabilitation and mobility enhancement. Conf Proc IEEE Eng Med Biol Soc 2009;2009:5030Y3 8. Wong CW, Bishop L, Stein J: A wearable robotic knee orthosis for gait training: A case series of hemiparetic stroke survivors. Prosthet Orthot Int 2012;36:119Y26 9. Bishop L, Stein J, Wong CK: Robot-aided gait training after spinal cord injury: A case report. J Neurol Phys Therapy 2012;6:138Y43 10. Mathias S, Nayak US, Isaacs B: Balance in elderly patients: the Bget-up and go[ test. Arch Phys Med Rehabil 1986;67:387Y9 11. Flansbjer UB, Holmback AM, Downham D, et al: Reliability of gait performance tests in men and women with hemiparesis after stroke. J Rehabil Med 2005;37:75Y82 12. Mong Y, Teo TW, Ng SS: 5-repetition sit-to-stand test in subjects with chronic stroke: reliability and validity. Arch Phys Med Rehabil 2010;91:407Y13 13. Berg KO, Wood-Dauphinee SL, Williams JI, et al: Measuring balance in the elderly: Validation of an instrument. Can J Public Health 1992;83:S7Y11 14. Fung S, Byl N, Melnick M, et al: Functional outcomes: The development of a new instrument to monitor the effectiveness of physical therapy. Eur J Phys Med Rehabil 1997;7:31Y41
1. Roger VL, Go AS, Lloyd-Jones DM, et al: on behalf of the American Heart Association Statistics Committee and Stroke Statistics Subcommittee: AHA Statistical Update: Heart disease and stroke statisticsV2012 update: A report from the American Heart Association. Circulation 2012;125:e2Y220
15. Baer HR, Wolf SL: Modified Emory Functional Ambulation Profile: An outcome measure for the rehabilitation of poststroke gait dysfunction. Stroke 2001; 32:973Y9
2. Jorgensen HS, Nakayama H, Raaschou HO, et al: Recovery of walking function in stroke patients: The Copenhagen stroke study. Arch Phys Med Rehabil 1995;76:27Y32
16. Wolf SL, Catlin PA, Gage K, et al: Establishing the reliability and validity of measurements of walking time using the Emory Functional Ambulation Profile. Phys Ther 1999;79:1122Y33
3. Teasell RW, McRae MP, Foley N, et al: Physical and functional correlations of ankle-foot orthosis use in the rehabilitation of stroke patients. Arch Phys Med Rehabil 2001;82:1047Y9
17. Schmid A, Duncan PW, Studenski S, et al: Improvements in speed-based gait classifications are meaningful. Stroke 2007;38:2096Y100
4. Hornby TG, Campbell DD, Kahn JH, et al: Enhanced gait-related improvements following therapist- vs. robotic-assisted locomotor training in subjects with chronic stroke: A randomized controlled study. Stroke 2008;39:1786Y92 5. Stein J, Narendran K, McBean J, et al: EMG-controlled exoskeletal upper limb powered orthosis for exercise
994
training post-stroke. Am J Phys Med Rehabil 2007; 85:255Y61
Stein et al.
18. Hidler J, Nichols D, Pelliccio M, et al: Multicenter randomized clinical trial evaluating the effectiveness of the Lokomat in subacute stroke. Neurorehabil Neural Repair 2009;23:5Y13 19. Duncan PW, Sullivan KJ, Behrman AL, et al: LEAPS Investigative Team: Body-weight-supported treadmill rehabilitation after stroke. N Engl J Med 2011;364: 2026Y36
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Gait
Affiliations: From the Department of Rehabilitation and Regenerative Medicine, Columbia University College of Physicians and Surgeons (JS, LB, CKW); Division of Rehabilitation Medicine, Weill Cornell Medical College (JS); Department of Rehabilitation Medicine, New York-Presbyterian Hospital, New York, New York (JS); and Department of Medicine, University of Virginia Medical Center, Charlottesville, Virginia (DJS).
Correspondence: All correspondence and requests for reprints should be addressed to: Joel Stein, MD, Department of Rehabilitation and Regenerative Medicine, Columbia University College of Physicians and Surgeons, Harkness Pavilion Room 1-165, New York, NY 10032.
ORIGINAL RESEARCH ARTICLE
Gait Training with a Robotic Leg Brace After Stroke A Randomized Controlled Pilot Study ABSTRACT Stein J, Bishop L, Stein DJ, Wong CK: Gait training with a robotic leg brace after stroke: a randomized controlled pilot study. Am J Phys Med Rehabil 2014;93:987Y994.
Objective: Robot-aided exercise therapy is a promising approach to enhance walking ability in stroke survivors. This study was designed to test a new robotic knee brace for restoring mobility in stroke survivors. Design: Twenty-four ambulatory individuals with chronic hemiparesis after
Disclosures: Supported by Tibion, Inc (now part of AlterG, Fremont, CA), which provided feedback on study design, but did not participate in the conduct of the study, interpretation, or reporting of the results. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
0894-9115/14/9311-0987 American Journal of Physical Medicine & Rehabilitation Copyright * 2014 by Lippincott Williams & Wilkins
stroke were enrolled in this pilot study. The participants were randomly assigned in equal numbers to either treatment with the experimental device or to a group exercise program and received a total of 18 hrs of their assigned therapy during a 6-wk training period. The primary outcome was gait velocity, as measured with the 10-m walk test. Secondary measures included 6-min walk test, Timed Up and Go test, Five-Times-Sit-to-Stand test, Romberg test, Emory Functional Ambulation Profile, Berg Balance scale, and the California Functional Evaluation 40.
Results: Twenty subjects completed the entire protocol and all follow-up visits. No significant differences between the two groups were found for the primary outcome measure at either the completion of training (week 6) or at the 3-mo follow-up (week 19), with inconsistent findings for secondary measures. No withingroup changes were seen in the primary outcome measure (10-m walk test) in either group. Within-group improvements were seen in several of the secondary measures for both groups. No complications of robotic therapy were observed.
Conclusions: Robotic therapy for ambulatory stroke patients with chronic DOI: 10.1097/PHM.0000000000000119
hemiparesis using a robotic knee brace resulted in only modest functional benefits that were comparable with a group exercise intervention. Key Words:
www.ajpmr.com
Robotics, Stroke, Hemiparesis, Gait, Exercise Therapy
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
987
S
troke affects approximately 795,000 individuals annually in the United States, with considerable resulting disability.1 Hemiparesis after stroke is a major contributor to disability, with more than 60% of stroke survivors experiencing limitations in mobility immediately after stroke. Of these, approximately 36% have continued mobility limitations on completion of their rehabilitation course.2 Mobility disturbances from hemiparesis include difficulty with sit-to-stand transfers, reduced balance, slowed and inefficient gait, falls, and damage to joints from altered biomechanics. Physical therapy has been shown to provide significant benefit, but substantial disability often persists.2 Moreover, many ambulatory stroke survivors continue to require leg braces (typically ankle foot orthoses) and/or ambulatory aids such as canes.3 Robotic devices to provide gait training have been developed, including the Lokomat (Hocoma, Inc, Volketswil, Switzerland) and the G-EO (Reha Technologies, Inc, Olten, Switzerland). These workstation-type devices provide gait training using a powered exoskeleton with treadmill (Lokomat) or powered footplates with programmable trajectories, analogous to a robotic elliptical trainer (G-EO). Results with the Lokomat have lagged those achieved with therapist provided partial body weight supported treadmill training.4 Moreover, these workstation robotic devices are large as well as costly and can only provide treatment in a rehabilitation gym setting. Wearable powered orthoses provide another approach to remediate motor deficits after stroke. Upper limb devices include electromyography-controlled powered elbow braces (Myomo, Inc, Cambridge, MA) to provide assistance to the elbow for flexion or extension.5 Surface electromyography does not perfectly indicate intended effort/force, however, and the optimal control mechanism and algorithm for providing assistance are not yet clear. A tethered lowerlimb exoskeletal device, the Anklebot (Interactive Motion Technologies, Inc, Watertown, MA), seems to improve gait parameters, even when training occurs in the seated position,6 but is not a fully wearable, autonomous device usable for community ambulation. Bilateral wearable powered exoskeletal robotic systems have been developed, including the ReWalk (Argo Medical Technologies Ltd, Yokneam Ilit, Israel) and eLegs (Ekso Bionics, Inc, Richmond, CA), but are primarily intended for use by individuals with paraplegia rather than stroke survivors with hemiparesis. AlterG (Fremont, CA; formerly Tibion, Inc) has developed the Bionic Leg, a powered knee orthosis intended to provide assistance and training for patients with unilateral neurologic or orthopedic
988
Stein et al.
conditions (Fig. 1). The device relies on multiple sensors, including pressure sensors placed in the user’s shoes, accelerometers, and joint angle detectors to create a model of the user’s activities, including sit-to-stand transfers, gait on level surfaces, and stair climbing/descent, and provides a mechanized assist with these activities through actuators at the knee joint.7 The device is fully wearable (i.e., untethered during use) and contains its own power source sufficient for several hours of use. The model used in this study weighs approximately 8 lbs (3.6 kg) and attaches with Velcro straps, requiring only 2Y3 mins to don/doff.8 The device is programmable, and a trained physical therapist establishes initial parameters and updates the device over the course of therapy.8,9 Donning and doffing the device requires some assistance from another individual, but once the device is in place, a patient could potentially use it without direct therapist supervision for home exercise between physical therapy sessions. This device has particular appeal for certain populations of hemiparetic stroke survivors with mobility impairments. Potential applications are the exercise training for the remediation of stiff-legged gait, treatment of knee hyperextension (recurvatum), difficulty with ambulation, sit-to-stand transfers, and/ or difficulty with stair climbing. The feedback provided through force feedback to the user is similar conceptually to the feedback provided during conventional gait training and is intended to help the user develop biomechanically more efficient gait through the adoption of a more normal gait pattern.
FIGURE 1 Bionic leg device in use.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
The device was well tolerated and did not cause any adverse effects in a small case series of several stroke patients with hemiparesis.8 The authors hypothesized that ambulation training with the use of the device would result in improved gait patterns more closely approximating normal gait because of the feedback and assistance provided by the device and thus result in greater gait speed and endurance. The authors therefore chose to conduct a pilot randomized controlled trial of this device in stroke survivors who were already independent in household ambulation but who had reduced gait velocity to determine the effect of physical therapy using this powered knee orthosis on gait speed and endurance. This study was intended to confirm the ability of ambulatory stroke patients with hemiparesis to successfully use the device and obtain preliminary data on efficacy to plan future clinical trials.
METHODS Subjects A total of 24 subjects were recruited from community-dwelling outpatients receiving care at the authors’ institution and/or enrolled in a registry of stroke survivors wishing to be considered for participation in clinical research. Inclusion criteria included a history of a single stroke (ischemic or hemorrhagic) causing significant leg weakness and gait alterations at least 6 mos before study entry. The stroke must have been confirmed through computed tomography or magnetic resonance imaging. Subjects with a single stroke based on clinical history but with evidence of previous asymptomatic strokes on imaging studies were considered eligible for participation. Subjects were required to be independent in household ambulation (with or without the use of a unilateral assistive device [e.g., standard cane, four-pronged cane, or hemiwalker] and with or without the use of an ankle foot orthoses). Exclusion criteria included subjects receiving ongoing physical therapy for the leg and/or gait and mobility training at the time of study entry. Subjects were permitted to have received previous botulinum toxin injections as long as such injections were received at least 3 mos before study entry and provided that no further injections were planned during the study period. Other exclusion criteria included the presence of other neurologic disorders, such as Parkinson disease, a history of more than one symptomatic stroke, excessive spasticity (defined as Ashworth scale of greater than 3 [of 4] at any lower limb joint), uncontrolled hypertension, unstable coronary artery disease, conwww.ajpmr.com
tractures of either lower limb, impaired cognition defined by a Folstein MiniYMental State Examination score below 24, or other medical conditions that might interfere with the subject’s ability to complete the study. Subject enrollment and flow is shown in Figure 2. No formal power analysis was performed because of insufficient information about anticipated effect size. The subjects were randomized in blocks of six using a 50:50 concealed allocation to either the robotic treatment group or the exercise control group.
Training Regimen The subjects in the robotic treatment group received 1 hr of individualized physical therapy with the device 3 days per week for six consecutive weeks (a total of 18 sessions). Each 1-hr session included brief rest periods, with a goal of administering approximately 50 mins of active therapy. Therapy sessions were conducted by an experienced physical therapist trained in the use of the device. Sessions consisted of a set of standardized over-ground functional tasks including sit-to-stand transfers from variable height with and without chair arms; standing balance while turning, reaching, stepping, and functional activities; gait training on level ground and surfaces; and mobility on stairs and curbs. All activities were performed to the tolerance and abilities of the subjects and supervised and assisted by a physical therapist present throughout the training session. The subjects assigned to the exercise control group received 1 hr of group exercise without the use of the robotic device 3 days/wks for six consecutive weeks (a total of 18 sessions). The exercise intervention was designed to control for the effects of interacting with a physical therapist by providing an equivalent number of formal therapy sessions while providing minimal gait training. Group exercise sessions included 3 subjects per treatment session. Therapy sessions were conducted by the same unblinded physical therapist treating the robotic group. Activities included relaxation/meditation, self-stretching, and gentle upper and lower limb active range of motion exercises. The subjects spent approximately 5 mins of the session walking at a comfortable walking pace (two 2.5-min walks during the 1-hr session).
Outcome Measures The subjects underwent functional gait and balance assessments on 4 separate occasions: at study entry, after completion of the 6-wk training program (week 6), as well as at 1 and 3 mos after completion of the training program (weeks 10 and 19). All Gait Training with Robotics After Stroke
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
989
FIGURE 2 Subject enrollment and flow.
assessments were performed by one physical therapist blinded to group assignment. Baseline and follow-up assessments of functional gait and balance performance included the Timed Up and Go (TUG),10 10-m walk test,11 6-min walk test,11 Five-Times-Sit-to-Stand test,12 the Berg Balance scale,13 California Functional Evaluation 40,14 and the Emory Functional Ambulation Profile (EFAP).15,16 Gait velocity was classified as household ambulation (G0.4 m/sec), limited community ambulation (0.4Y0.8 m/sec), and full community ambulation (90.8 m/sec) to determine how many subjects improved from one class to the next.17 The 10-m walk test measures gait velocity and was designated as the primary outcome measure. It was performed on a 14-m walkway, and measurement was performed for the middle 10 m on this
990
Stein et al.
walkway. Three trials were timed by stopwatch and averaged. The 6-min walk test measures a combination of gait endurance and velocity. The TUG and Five-Times-Sit-to-Stand test are measures of transfer ability, with the TUG including a component of gait velocity as well. The Berg Balance scale measures balance. The California Functional Evaluation 40 and EFAP are assessments of overall gait functional ability. Unpaired t tests were used to compare baseline characteristics of the two groups, except in cases where tests of normality failed, when Mann-Whitney rank sum tests were performed. Paired t tests were used for an intention-to-treat analysis of change scores at the conclusion of treatment. A linear mixed model was used to analyze change over time in perprotocol subjects to account for the correlation between the multiple time points inherent in repeated
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
measures. This allowed consideration of all available time points when determining the effect of the intervention. A constant correlation (exchangeable) model was found to provide the best fit and used for all analyses. The authors used SAS version 9.3 for statistical analyses (SAS Institute, Cary, NC). The study was approved by the institutional review board. All subjects provided informed consent.
RESULTS Thirty-four subjects underwent in-person screening for participation, of whom 24 subjects were enrolled, with 12 subjects allocated to each treatment group. The most frequent reasons prospective subjects were excluded from the study were ongoing physical therapy, recent botulinum toxin injection in the lower limb, excessive spasticity, and low score on the Folstein MiniYMental State Examination. Baseline subject characteristics were similar between the two treatment groups (see Table 1). All subjects completed the training regimen, and there were no adverse effects of either therapy. One subject in the exercise control group sustained a fall during his 1-mo follow-up assessment, resulting in a nonYdisplaced fracture of the fifth metatarsal bone. One subject in the robotic group had a second stroke between the 1-mo (week 10) and 3-mo (week 19) follow-up visits that was deemed unrelated to the study treatment. One additional subject in the control exercise group failed to complete the 1- and 3-mo follow-up, and one additional subject in the robotic treatment group was lost to follow-up before the 3-mo follow-up assessment.
No significant differences were found between the two groups for the primary outcome measure at the completion of training (week 6) using paired t test analyses (n = 24). Results for the Berg Balance scale at completion of training (week 6) favored the robotic treatment group; conversely, results for the EFAP at completion of training favored the control exercise group. The remainder of secondary measures showed no difference between the two groups at completion of training. A per-protocol analysis of the 20 subjects who completed all of the assessments found no significant differences between the two groups using a linear mixed model through the 3-mo follow-up period for either primary or secondary measures (see Table 2 for a summary of outcome measurements). The authors also examined the number of subjects who advanced from one gait velocity classification (e.g., household ambulation, G0.4 m/sec) to the next and found no statistically significant change from either intervention (see Table 3). Within-group analyses revealed no change in the primary outcome measure for either group from baseline (week 0) to the completion of treatment (week 6) or the 3-mo follow-up (week 19). Several secondary measures demonstrated within-group improvements at these two time points, as shown in Table 2.
DISCUSSION No differences in the primary outcome measure of gait velocity were demonstrated between the two groups, nor were there any within-group
TABLE 1 Baseline subject characteristics
Subjects Mean age Percentage (male) Duration after stroke, mean (SD), mos Folstein MiniYMental State Examination score, mean (SD) Baseline 10-m walk test, mean (SD) 6-min walk test, mean (SD) 5-times-sit-to-stand test, secs TUG, secs Berg Balance scale CAFE´ 40 EFAP Romberg, eyes open Romberg, eyes closed
Robotic Training Group (SD)
Exercise Group (SD)
P
12 57.6 (10.7) 83% 49.1 (38.9) 28.8 (1.2)
12 56.6 (15.1) 58% 88.5 (153.0) 29.2 (0.9)
0.85 0.37 1.0a 0.46
22.9 (20.0) 185.9 (95.9) 31.6 (26.4) 22.7 (9.7) 46.1 (11.7) 135.2 (38.8) 91.7 (42.5) 14.3 (14.6) 2.5 (3.5)
27.5 (21.2) 169.3 (89.6) 24.1 (9.8) 33.7 (26.4) 49.4 (5.8) 137.8 (34.6) 131.8 (93.2) 20.4 (14.2) 7.4 (9.7)
0.40a 0.67 0.39 0.21 0.39 0.87 0.21 0.28a 0.26a
a Mann-Whitney rank sum test. CAFE´ 40, California Functional Evaluation 40.
www.ajpmr.com
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
991
TABLE 2 Outcome measures
Baseline (Week 0)
10-m walk test, secs
Completion of Training (Week 6)
1-mo Follow-up (Week 10)
3-mo Follow-up (Week 19)
Robotic Therapy (n = 12)
Exercise Control (n = 12)
Robotic Therapy (n = 12)
Exercise Control (n = 12)
Robotic Therapy (n = 12)
Exercise Control (n = 10)
Robotic Therapy (n = 10)
Exercise Control (n = 10)
22.9 (20.0)
27.5 (21.2)
20.3 (15.0)
19.2 (9.1)
20.5 (16.3)
21.0 (12.4)
15.1 (7.0)
20.2 (11.2)
6-min walk test, m 185.9 (95.9) 169.3 (89.7) 213. 4 (108.2) 194.8 (83.2) 217.8 (121.1) 203.8 (87.1) 252.7 (108.4)
214 (94.7)
5-times-sit-to-stand test, secs TUG, secs
31.6 (26.4)
24.1 (9.8)
25.3 (16.7)
18.8 (6.1)
19.4 (8.3)
16.5 (6.8)
26.3 (32.1)
15.8 (4.3)
22.7 (9.7)
33.7 (26.4)
26.7 (16.9)
28.0 (17.1)
24.8 (16.8)
26.8 (18.4)
18.4 (7.9)
25.5 (15.5)
Berg Balance scale
46.1 (11.7)
49.4 (5.8)
48.6 (10.2)
49.2 (5.8)
50.4 (8.2)
51.6 (5.4)
52.4 (5.4)
52.5 (4.7)
CAFE´ 40
135.2 (38.8) 137.8 (34.6) 148.8 (31.7)
EFAP
91.7 (42.5) 131.8 (93.2) 115.1 (84.7)
Romberg, eyes open, secs Romberg, eyes closed, secs
14.3 (14.6)
20.4 (14.2)
2.5 (3.5)
7.4 (9.7)
137.3 (34.5) 149.8 (30.5)
148.9 (40.3) 161.1 (37.2)
153.8 (37.3)
101.7 (55.4)
82.7 (31.4)
98.4 (53.3)
76.2 (30.8)
97.8 (51.1)
23.9 (10.3)
22.2 (11.0)
23.1 (12.5)
23.8 (10.6)
25.7 (9.9)
22.9 (11.7)
7.4 (9.4)
11.7 (12.6)
9.8 (11.3)
12.3 (12.9)
14.6 (13.5)
13.9 (12.1)
a
Linear mixed model. Mann-Whitney rank sum test. c Statistically significant. CI, confidence interval. b
changes for this measure. Secondary measures provided conflicting results, with one measure (Berg Balance score) favoring robotic therapy, another (EFAP) favoring the exercise control group, and the remainder of the secondary measures without any difference between the two groups. The authors did, however, find within-group improvements in walking endurance and in several other secondary measures. In the aggregate, these findings support the use of exercise therapy to improve mobility in individuals with chronic hemiparesis but do not suggest an advantage to robotic leg brace for this purpose. There are several potential explanations for the lack of any incremental benefit of robot-aided training over group therapy. One potential explanation for the authors’ findings is that the robotic device used and/or the training regimen implemented does not provide the optimal training experience for this population. It is possible that the dose of gait training provided was inadequate and that training at a faster cadence, more intensive training (more hours per week), or longer-duration training (more total weeks of train-
992
Stein et al.
ing) would have provided greater benefit. Although the authors did not measure the number of steps taken during each therapy session, these were likely fewer than achievable with a treadmill-based system because the device functions at the self-selected cadence of the user rather than at a faster rate used in some treadmill training protocols. The intensity and duration of the training in the authors’ study were selected arbitrarily on the basis of logistic considerations rather than on the evidence that this protocol is optimal. Another possible explanation of the authors’ findings might be that walking itself, rather than therapy involving treadmills, robots, or other devices, is the most effective therapy for improving gait in ambulatory stroke survivors. Other studies showing better results18 or comparable outcomes19 for conventional gait training compared with robotic or partial body weightYsupported treadmill training may support this argument. The authors used a control intervention that was intended to provide a comparable amount of time in a therapeutic environment interacting with
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Difference Within Groups Mean Change from Baseline (Week 0) to Completion of Training (Week 6) Exercise Control (95% CI)
Robotic Therapy (95% CI) 2.6 (j1.6 to 6.8) P = 0.20 27.5 (j5.4 to 60.4) P = 0.09 6.3 (j3.2 to 15.8) P = 0.17 0.4 (j3.7 to 4.5) P = 0.85 2.5 (0.9Y4.1) P = 0.005c 13.7 (j3.7 to 31.0) P = 0.11 1.2 (j10.2 to 12.5) P = 0.83 9.6 (j0.5 to 19.7) P = 0.06 4.9, P = 0.049bc
Difference Between Groups
Mean Change from Baseline (Week 0) to 3-mo (Week 19) Follow-up Robotic Therapy
Baseline (Week 0) to Baseline (Week 0) Completion (Week 6) to 3-mo (Week 19) of Training Follow-upa
Exercise Control
8.3 (j0.6 to 17.3) P = 0.06 25.5 (6.6Y44.4) P = 0.01c 5.3 (2.7Y8.9) P = 0.08 5.7 (j1.0 to 12.4) P = 0.09 0.3 (j2.0 to 2.5) P = 0.81 4.6 (j4.1 to 13.4) P = 0.27 29.7, P = 0.005bc
0.5 (j3.7 to 4.6) P = 0.81 42.1 (j2.1 to 86.2) P = 0.06 3.7 (j6.1 to 13.5) P = 0.41 2.7 (j0.2 to 5.6) P = 0.07 2.1 (j0.6 to 4.8) P = 0.11 18.0 (j2.2 to 38.2) P = 0.08 7.9 (0.8Y15.0) P = 0.03c 10.2 (j3.9 to 24.3) P = 1.0b P = 0.14 3.6 (j3.4 to 10.6) 11.6 (j1.8 to 21.5) P = 0.27 P = 0.03c
5.7 (j3.9 to 15.2) 5.8, P = 0.13b P = 0.22 28 (11.9Y37.5) 2.0 (j33.7 to 37.7), P = 0.038c P = 0.91 6.5, P = 0.04bc 0.9 (j9.3 to 11.2) P = 0.85 5.3 (j2.3 to 12.9) 6.3, P = 0.28b P = 0.16 2.3, P = 0.19b 2.8 (0.13Y5.4) P = 0.04c 9.0 (j9.8 to 27.9) 18.1 (5.0Y31.2) P = 0.33 P = 0.01c 30.6, P = 0.049bc 30.8, P = 0.009 bc
P = 0.18
1.4, P = 0.88b
8.0, P = 0.25b
P = 0.65
5.2 (0.03Y10.4) P = 0.049c
1.7, P = 0.60b
P = 0.67
P = 0.38 P = 0.24 P = 0.22 P = 0.64 P = 0.88 P = 0.25
of the study do not allow the authors to determine what portion of the improvement in gait endurance resulted from treatment sessions vs. a nonYspecific effect of study participation, a further examination of this hypothesis would be worthwhile. The improvements seen in the study’s control group emphasize the impact of participation in a clinical trial and contact time with a physical therapist on performance even in the absence of goal-directed gait training and should inform design of control interventions in future studies of rehabilitation therapies. One strategy to provide greater amounts of therapy would be to provide a home-based program
a physical therapist while minimizing the amount of formal gait training provided. The use of a more conventional physical therapy program focusing on gait training might have provided different results, perhaps even superior to the robotic leg brace intervention used in the study’ protocol. The authors speculate that some portion of the improvement in walking endurance seen may have resulted from a nonYspecific effect of participating in a clinical trial. The frequent visits to the authors’ facility for treatment sessions represent an opportunity to increase activity levels simply by the act of traveling to and from the treatment sessions. Although the data TABLE 3 Gait velocity classification
Gait Velocity G0.4 m/sec
0.4Y0.8 m/sec
90.8 m/sec
0 0 2 2
5 7 5 5
5 3 3 3
Robotic therapy group (baseline) Robotic therapy group (after therapy) Exercise control group (baseline) Exercise control group (after therapy) Number of subjects falling within each gait velocity class.
www.ajpmr.com
Gait Training with Robotics After Stroke Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
993
to supplement therapist-supervised training. Planned enhancements in the device that will allow retrieval of activity data would permit a more rigorous quantification of therapy dose in the future and allow the determination of a dose-response curve. All of subjects of this study were ambulatory for at least household distances independently at study entry and may not have been the ideal target population for robot-aided therapy. It is possible that this form of robotic therapy may prove more effective for individuals who lack the ability to walk independently and who would therefore benefit from the physical assistance provided by the robot. Moreover, there may be a window of opportunity earlier after stroke during which the recovery curve may be altered favorably through the incorporation of robotaided rehabilitation. Because this study focused exclusively on ambulatory stroke patients with chronic stable deficits, further study with nonYambulatory stroke patients during the early phase of gait recovery after stroke would be required to test this hypothesis. In conclusion, no overall difference in outcome was found between the robot-assisted and control group exercise therapies. The subjects within each group showed no change in gait velocity but did experience improvements in gait endurance and some other secondary measures. REFERENCES
6. Forrester LW, Roy A, Krebs HI, et al: Ankle training with a robotic device improves hemiparetic gait after a stroke. Neurorehabil Neural Repair 2011;25: 369Y77 7. Horst RW: A bio-robotic leg orthosis for rehabilitation and mobility enhancement. Conf Proc IEEE Eng Med Biol Soc 2009;2009:5030Y3 8. Wong CW, Bishop L, Stein J: A wearable robotic knee orthosis for gait training: A case series of hemiparetic stroke survivors. Prosthet Orthot Int 2012;36:119Y26 9. Bishop L, Stein J, Wong CK: Robot-aided gait training after spinal cord injury: A case report. J Neurol Phys Therapy 2012;6:138Y43 10. Mathias S, Nayak US, Isaacs B: Balance in elderly patients: the Bget-up and go[ test. Arch Phys Med Rehabil 1986;67:387Y9 11. Flansbjer UB, Holmback AM, Downham D, et al: Reliability of gait performance tests in men and women with hemiparesis after stroke. J Rehabil Med 2005;37:75Y82 12. Mong Y, Teo TW, Ng SS: 5-repetition sit-to-stand test in subjects with chronic stroke: reliability and validity. Arch Phys Med Rehabil 2010;91:407Y13 13. Berg KO, Wood-Dauphinee SL, Williams JI, et al: Measuring balance in the elderly: Validation of an instrument. Can J Public Health 1992;83:S7Y11 14. Fung S, Byl N, Melnick M, et al: Functional outcomes: The development of a new instrument to monitor the effectiveness of physical therapy. Eur J Phys Med Rehabil 1997;7:31Y41
1. Roger VL, Go AS, Lloyd-Jones DM, et al: on behalf of the American Heart Association Statistics Committee and Stroke Statistics Subcommittee: AHA Statistical Update: Heart disease and stroke statisticsV2012 update: A report from the American Heart Association. Circulation 2012;125:e2Y220
15. Baer HR, Wolf SL: Modified Emory Functional Ambulation Profile: An outcome measure for the rehabilitation of poststroke gait dysfunction. Stroke 2001; 32:973Y9
2. Jorgensen HS, Nakayama H, Raaschou HO, et al: Recovery of walking function in stroke patients: The Copenhagen stroke study. Arch Phys Med Rehabil 1995;76:27Y32
16. Wolf SL, Catlin PA, Gage K, et al: Establishing the reliability and validity of measurements of walking time using the Emory Functional Ambulation Profile. Phys Ther 1999;79:1122Y33
3. Teasell RW, McRae MP, Foley N, et al: Physical and functional correlations of ankle-foot orthosis use in the rehabilitation of stroke patients. Arch Phys Med Rehabil 2001;82:1047Y9
17. Schmid A, Duncan PW, Studenski S, et al: Improvements in speed-based gait classifications are meaningful. Stroke 2007;38:2096Y100
4. Hornby TG, Campbell DD, Kahn JH, et al: Enhanced gait-related improvements following therapist- vs. robotic-assisted locomotor training in subjects with chronic stroke: A randomized controlled study. Stroke 2008;39:1786Y92 5. Stein J, Narendran K, McBean J, et al: EMG-controlled exoskeletal upper limb powered orthosis for exercise
994
training post-stroke. Am J Phys Med Rehabil 2007; 85:255Y61
Stein et al.
18. Hidler J, Nichols D, Pelliccio M, et al: Multicenter randomized clinical trial evaluating the effectiveness of the Lokomat in subacute stroke. Neurorehabil Neural Repair 2009;23:5Y13 19. Duncan PW, Sullivan KJ, Behrman AL, et al: LEAPS Investigative Team: Body-weight-supported treadmill rehabilitation after stroke. N Engl J Med 2011;364: 2026Y36
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 11, November 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
