Despite its impact and pervasiveness, severe arm hemiparesis is often not impacted by rehabilitative therapies.1 Newer rehabilitative regimens have incorporated repetitive task-specific training (RTP) and brought about increased affected arm use and function.2-5 Yet, despite their promise, the strategies inherent in some RTP approaches require exceptional time and/or human resources2 and/or are not reimbursed by insurance.2,5 Moreover, patients must typically exhibit active, distal movement to be eligible for the previously mentioned therapies: an ability not exhibited by many individuals with stroke. Sophisticated mechanized approaches have been proposed as an alternative;6 however, many of these treatments incorporate devices offering equivocal efficacy7 or no greater efficacy than that achieved with conventional therapy.8 Moreover, the cost, complexity, and size of existing devices make their purchase and/or use prohibitive in most clinical and home environments.
The Myomo e100 (Myomo, Inc, Boston, Massachusetts) is a lightweight, neurorobotic brace that is worn while the individual performs RTP. The neurorobotic device obtains surface electromyography (EMG) signals that are transmitted to a powered orthosis, which assists affected elbow flexion and extension, as needed. The intended users of the device is Individuals with stroke who exhibit some active movement of the shoulder, but limited movement at the affected elbow, such that active assistance may be needed. The proximal-to-distal recovery pattern often exhibited after stroke means that many of these individuals exhibit little or no movement in the joints distal to the elbow. A case series9 conducted by members of our research team suggested that the use of the neurorobotic device as part of a program of RTP may reduce affected arm impairment. However, the study did not apply rigorous study criteria and did not examine outcomes in individuals with severe impairment.
In preparation for a larger trial, we performed a case study to examine the outcomes associated with neurorobotic device training in a subject having severe arm impairment, as characterized by the absence of active elbow extension and absence of active movement at any joint distal to the elbow. To our knowledge, this was the first study examining a successful robotic-based RTP strategy in an individual exhibiting this severity of stroke-induced arm hemiparesis.
The subject of the case study was a 53-year-old African American man who responded to a study advertisement placed in local outpatient therapy clinics in the Midwestern United States. The first volunteer who responded to the advertisement was screened to determine whether he met the inclusion criteria (see later). Before screening and participation, he signed an informed consent form to participate in the study, which had been approved by the Medical Institutional Review Board at the University of Cincinnati. The subject had experienced a right hemorrhagic stroke 30 months before study enrollment. He was right-hand dominant. He received inpatient speech, occupational, and physical therapy immediately poststroke, occurring every weekday for 3 weeks. On discharge, he was administered outpatient physical and occupational therapy, occurring 2 days/week for approximately 8 weeks. At the time of study enrollment, he was not participating in any formal rehabilitative program but performed affected leg and arm raises daily and reported being “very, very interested” in further rehabilitative therapies targeting his arm. At the time of study enrollment, the subject was taking phenytoin (100 mg [4× per day]), topiramate (50 mg [2× per day]), digoxin (0.25 mg [4× per day]), and aspirin (81 mg [1× per day]). Aside from the stroke, the subject's medical history was significant for diabetes and heart disease (he had a pacemaker). He lived at home with his wife who worked full-time, and he performed some household chores. He walked independently using an ankle foot orthosis.
The subject met the following study inclusion criteria: (1) upper extremity Fugl Meyer score ranging between 10 and 25; (2) stroke experienced greater than 12 months before study enrollment; (3) a score of greater than or equal to 25 on the Folstein Mini-Mental Status Examination; (4) age ranging between 21 years and 75 years; (5) experienced a single stroke; (6) discharged from all forms of physical rehabilitation; (7) presence of volitionally activated EMG signal from the paretic biceps brachii of at least 5 μV in amplitude; (8) neurorobotic device able to fit on affected arm properly and without discomfort (ie, no red marks or discomfort observed in 10 minutes of use during fitting. Exclusion criteria were as follows (1) excessive pain in the affected hand, arm, or shoulder, defined as a score greater than or equal to 5 on a 10-point visual analog scale; (2) excessive spasticity in the affected elbow, defined as a score of greater than or equal to 3 on the Modified Ashworth Spasticity scale (MAS); (3) prior botulinum toxin injection to any portion of the affected arm within the past 4 months or phenol injections within the past 12 months before study participation; (4) stroke below the level of the midbrain; (5) substantial contracture of elbow, defined as greater than 20° of elbow flexion at the baseline evaluation (the e100 device cannot function in the presence of reduced range of motion because of contractures).
During screening, it was observed that the subject was able to produce some active movement of the shoulder and elbow, including elbow flexion, but no active elbow extension, and no active movement in any joint below the elbow. Values for active and passive range of motion of the affected arm obtained during screening are given in Table 1. The subject exhibited MAS scores of 2 at the affected elbow (during both elbow flexion and extension) and 3 at the affected wrist, digits, and thumb.
Given that motor return usually occurs in a proximal-to-distal fashion after stroke, we reasoned that an intervention encouraging movement of the affected elbow within the context of activities of daily living (ADL) might increase elbow extension, as well as possibly having a positive influence on affected arm use, and quality of movement during ADL.
After screening, a battery of outcome measures was administered in our laboratory by a research team member who was blinded to the intervention to be administered. It took approximately an hour and a half to administer all measures to the participant. The primary outcome measure was the upper extremity section of the Fugl-Meyer Impairment scale (FM),10 which assessed changes in arm impairment. Specific FM items were derived from Brunnstrom's stages of poststroke motor recovery. Data arise from a 3-point ordinal scale (0 = cannot perform, 2 = can perform fully) applied to each item for a total score of 66 on the upper extremity scale. The FM exhibits high test-retest reliability (total = 0.98-0.99, subtests = 0.87-1.00), interrater reliability, and construct validity.11,12
In addition to the subject's ability to actively perform isolated movements, we administered the Canadian Occupational Performance Measure (COPM), a standardized interview used to identify occupational performance problems and to measure satisfaction and importance of tasks according to the subject. Specifically, after the client identifies activities that he or she has difficulty performing, he or she rates each activity on a scale from 0 to 10 according to its importance to the subject, perception of performance with the skill, and his or her satisfaction with the performance. The therapist then confirms the most important activities with the client and asks the subject to rate each task on his or her performance and his or her satisfaction with the performance. The performance score and the satisfaction score are then each calculated by adding together the performance or satisfaction scores for all problems and dividing by the number of problems. Once the top 5 activities are determined, these tasks are used to guide the treatment. The test is usually administered at baseline and discharge to determine change in performance and satisfaction. If the number is positive, then there is change for the better. If the number is negative, then the person's perception of his or her satisfaction and/or performance of the skill has lessened. In 2004, the COPM was reported to be a valid, reliable, clinically useful, and responsive outcome measure,13 including after stroke.14 In the current study, the activities chosen by the subject using the COPM included using a floor scrubber; preparing a simple meal, such as tacos or chili; opening containers; using an upright vacuum; and managing a checkbook. He also noted the desire to stand and shoot a basketball and cast and spin cast a fishing rod as leisure goals.
Finally, to determine whether the previously mentioned changes conveyed changes in quality of life, we administered the Stroke Impact Scale 3.0 (SIS).15 The instrument is a 59-item self-report measure, with items further broken down into 8 domains (strength, hand function, ADL, mobility, communication, emotion, memory, and social participation). Scores for each domain range from 0 to 100, and higher scores indicate a better health-related quality of life. In previous studies, SIS domains were examined by comparing the SIS to existing stroke measures and by comparing differences in SIS scores across Rankin scale levels. Using these techniques, each domain met or approached the standard of α coefficient of 0.9 for comparing the same subjects across time. The intraclass correlation coefficients for test-retest reliability ranged from 0.70 to 0.92.15
All 3 measures are responsive in stroke, with the FM and SIS being stroke specific and the COPM being used in a number of neurorehabilitative and stroke studies.
Five days after pretesting, the participant returned to the laboratory and met with the treating therapist. During a 2-hour session, the neurorobotic device was fitted and the subject was educated on device use by a research team member. This included proper donning and doffing and the use of the device. At this time, we also confirmed that there was sufficient EMG signal to use the device. The Myomo e100 (Figure 1) is a Food and Drug Administration–approved, noninvasive, lightweight (≈5 lb), and wearable system that continuously monitors the surface EMG signals from either the flexor or extensor muscles of the affected arm. These EMG signals control the powered neurorobotic orthosis and assist the active muscle with movement of the affected arm, and it can also provide passive assist in the opposite direction when the muscle being sensed is turned off or relaxes (eg, when the sensor is on the biceps, it can provide assistance with elbow flexion when the biceps is active or assistance with passive elbow extension when the biceps is relaxed; the opposite is true when the sensor is moved to the triceps).
The EMG signals are filtered and processed to infer a desired joint torque. The signal processing of the measured surface EMG is accomplished through a system that comprises off-the-shelf EMG sensors, analog signal–processing components, and digital signal–processing components.
The signal-processing algorithm enables bidirectional control at the elbow into flexion or extension. The system gain parameter (ie, amount of assistance in the active assist direction) varies during the course of a session as the subject fatigues. The base unit for software gain is 12 V of motor voltage per volt of surface EMG voltage. The passive opposing force parameter is generally constant throughout a therapy session, usually changing slightly from session to session to account for changes in muscle tone.
During use, changing the location of the EMG sensor (which can be accomplished without removing the brace) allows the user to alternate between active assistance with elbow flexion or with extension, without the need to change any device settings. The user's central nervous system is incorporated into the control loop through a combination of kinesthetic, proprioceptive, tactile, and visual sensory feedback. Specifically, the user's intention to move is detected via his or her EMG. This allows the user to accomplish position control, with the neurorobotic device acting as a forward-loop strength amplifier. The treating therapist can adjust the system parameters to alter the amount of assistance that the device provides.
Beginning 1 week after the education session, the subject received individualized, 1-hour, RTP sessions, which occurred while wearing the neurorobotic device. The sessions were held 3 days/week for 8 weeks. All therapy was administered by the same therapist and in the same environment.
On the basis of pilot work,9 the research team had compiled a battery of tasks that could be performed by using the neurorobotic device and that would engage the hemiparetic arm. The activities involved using the affected arm for either actively performing the task unilaterally, or actively assisting in the task, in a meaningful way. The particular activities in which the subject engaged (listed in Table 2) were chosen by the treating therapist on the basis of the activities that the subject identified via the COPM. Components of strength, flexibility, endurance, and proprioception were included to allow the muscles to coordinate in the performance of a purposeful final movement. Given the bimanual nature of most ADL, we asked each subject to work on only 1 bilateral task and 1 unilateral reaching task incorporating the affected arm during each session. The unilateral reaching task was one of the subject's chosen ADL. Finally, to target elbow extension and flexion (which were component parts of the reaching task and several other ADL that the subject identified), we chose 1 task that emphasized biceps flexion and 1 that emphasized triceps extension. Each task was further graded as having a difficulty level of 1 (less difficult) or 2 (more difficult); this allowed the therapist ability to progress to more difficult tasks when the subject had attained proficiency in 8 of 10 attempts in a less-difficult task.
It was decided that allowing practice of all of the tasks over the 8 weeks would permit maximum time for the subject to identify muscles that needed to be used during each task and would provide the opportunity to learn how to optimize performance of each task through trial-and-error practice. Thus, each activity listed in Table 2 was practiced at least twice during the 8-week intervention. While other modalities (eg, electrical stimulation) were not permitted, the treating therapist was given the option to use 5 to 10 minutes as needed during each session to perform range of motion or other exercises to help with performance of the movements in Table 2. This decision was made given the moderate MAS levels that the subject exhibited at some joints during screening; these levels indicated to us that stretching would sometimes be necessary to transiently reduce muscle tone. The therapist maintained a treatment record, so that researchers could monitor compliance with the protocol.
The previous RTP therapy program, sometimes called “standard task training of the affected limb,” was consistent with the methods of previous affected arm studies.2-4 Moreover, RTP, as structured herein, did not use shaping, as was the case in 2 of the previously mentioned studies.2,4 Therefore, our approach had the advantage of repeating tasks with continuous feedback from the therapist, with little formal time necessary for review of immediate-past performance. The RTP used herein was also an accurate analog for actual therapy provided in most clinics, as it involved motor learning of functional activities performed continuously for a period of up to 30 minutes.16
During the course of the intervention, the participant complained of minor fatigue on 3 occasions and shoulder pain on 2 occasions; he had no persistent complaints or limitations. Adherence was 100%, and he attended all clinical sessions. After 10 weeks, the subject returned to the laboratory, where all outcome measures were again administered by an examiner. The examiner was blinded in that he routinely administered outcome measures to patients and was not aware whether this particular subject had been administered any intervention.
Before intervention, the subject exhibited minimal active movement in the paretic limb, indicated with FM score of 10 at initial testing. This reflected normal reflexes, ability to partially or fully complete all FM shoulder items, and ability to actively flex the affected elbow. Because of his minimal levels of active movement, he was rather unsatisfied with his ability to perform valued activities, reflected both by his low score on the ADL scale of the SIS (Table 3) and by the low satisfaction and performance ratings that were given by the subject on the COPM (2.2 points on each scale).
After intervention, the subject maintained his preexisting active movements in the affected arm. Moreover, he gained the ability to actively extend his elbow and to flex the affected wrist with the elbow at his side at 90°. Consequently, his FM score increased to 12 points (an increase of 2 points). He also exhibited 1-point reductions in MAS scores at the affected elbow (extension and flexion) and affected thumb, but spasticity levels in the affected wrist and digits remained unchanged. These changes were accompanied by increased satisfaction and performance on the COPM (+ 2 and +1.8, respectively), and with the exception of the mobility scale, there were also increases in all scales of the SIS, including a 30-point increase in the subject's rating of his recovery (Table 3).
The subject's wife reported that she was pleased with the subject's new movement abilities and that he was attempting to use the affected arm more at home, particularly for cleaning and cooking ADL. After the completion of the study, the subject was also able to return to outpatient physical therapy for electrical stimulation to the affected arm and additional therapy to improve skill with ADL.
Despite their increasing pervasiveness and disabling impact, few treatments reduce stroke-induced hemiparesis in the individuals with the most severe impairments. Moreover, the most promising regimens require some active movement at the affected wrists and fingers: a prerequisite that excludes many individuals with stroke. The current intervention reports positive outcomes associated with the use of a portable neurorobotic brace on active affected arm movement in a subject with stroke exhibiting severe hemiparesis.
After 8 weeks of neurorobotic device training incorporating supervised, task-specific practice, the subject exhibited a 2-point FM increase with the score increase, resulting from additional active movement at the affected elbow and new movement in the affected wrist. The magnitude of total FM change was somewhat small (+2 points). Nonetheless, these new movements were clinically significant, conveying the ability to perform a limited number of ADL that he could not previously perform, including helping around the house (eg, using a dust rag, using a mop). In turn, performance of these new ADL caused the subject to feel better about his contribution to the household, which had been diminished since the time of his stroke. These impressions likely increased his satisfaction and quality of life, as reflected in the increased COPM scores.
In light of the relatively small FM changes, a more surprising finding was the change in SIS scores that was observed. For example, it was striking that, although there were no changes in active hand movement on the FM, there was a 13-point change in the SIS hand domain. We believe that at least part of the reason for the larger and more widespread SIS changes is the psychological effect of intervention participation. Specifically, after intervention, the subject reported using the affected arm for a variety of household chores. Although his active arm movement might not have changed appreciably, we believe that his subjective perception of his arm's movement and utility changed, which caused the self-reported SIS changes. Future research may want to use qualitative methods in concert with behavioral measures to better assess the role that subjects’ perception of affected extremity change plays in outcomes of self-reported and objectively measured assessment.
An additional related factor that might have contributed to the notable SIS and/or COPM changes was the subject's apparent increase in affected arm use, which caused him to reintegrate the affected arm into the performance of several household chores and to feel better about his contribution to the household. Given the known relationships between affected arm use, neuroplasticity, and movement, we would argue that any change in perceived utility and subsequent use is welcome, even if actual movement (as measured by the FM) is not markedly changed. Yet, although increased affected arm use was reported by his wife, we did not formally measure this construct. In future studies, it may be of value to examine affected arm use, perhaps by using activity monitors that provide quantified, objective ascertainments of affected and unaffected arm use patterns. Kinematic measurements will also enable additional, objective assessment of subtle movement changes that rater-based measures may not discern.
The subject exhibited new elbow and wrist movement and increases on the previously mentioned outcome measures. Given the chronicity of his stroke and the fact that the improvements were exhibited in a period of time in which no other rehabilitation was being provided, the changes were likely because of the intervention herein described. Consequently, the subject's physician was able to use these data to justify a prescription for the subject to return to rehabilitative therapy and to have payment for the therapy reimbursed by the patient's managed care provider. Thus, for the field of rehabilitation, the significance of this intervention may be not only in its comparatively smaller cost and size to other robotic devices but also in its ability to act as a “bridge” to other interventions for patients with severe impairment of the upper extremity.
Several, other robotic devices target the stroke-affected arm. However, most existing robotic devices use large platform systems, with some incorporating force transducers and visual feedback via therapy-centric video games to provide the patient with upper extremity assistance, ranging from full assistance to partial assistance. Moreover, most existing devices do not provide the patient with the ability to practice ADLs when working with a therapist: a prerequisite for some types of reimbursement. While the case study approach we used cannot suggest causality, the changes in our single subject described herein were comparable in magnitude to outcomes reported in a recent trial with the T-WREX device.17 Conversely, a study with another device indicated that outcomes were not superior to guided practice with a therapist.8 The device tested herein may thus potentially constitute a superior value compared with other robotic devices on the market, in terms of its portability, lower price,7 comparable magnitude of change on outcome measures, and ability to enable RTP within the context of a meaningful functional goal. These facets, the outcomes of this case report, and the outcomes of the previous study of the neurorobotic device in less-impaired subjects9 justify further exploration of this promising and innovative technique in patients exhibiting greater impairment of the affected arm.
In an individual with severe arm hemiparesis due to chronic stroke, participation in a regimen of RTP incorporating a portable, EMG-triggered, neurorobotic device was associated with reduced impairment of the affected arm. The subject also experienced increased satisfaction with his performance of a limited set of valued tasks, functional tasks, and quality of life. Importantly, the training-related changes conveyed ability to perform some activities that the subject could not previously perform. The movement changes were sufficiently robust to enable the subject to be readmitted to outpatient rehabilitative therapy, where he experienced additional gains.
1. Duncan PW. Synthesis of intervention trials to improve motor recovery following stroke. Top Stroke Rehabil. 1997;3:1–20.
2. Wolf SL, Winstein CJ, Miller JP, et al.. EXCITE Investigators. Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: the EXCITE randomized clinical trial. JAMA. 2006;1,296(17):2095–2104.
3. Page SJ, Levine P, Leonard A. Mental practice in chronic stroke: results of a randomized, placebo-controlled trial. Stroke. 2007; 38(4):1293–1297.
4. Page SJ, Levine P, Leonard AC, Szaflarski J, Kissela B. Modified constraint-induced therapy in stroke: results of a single blinded, randomized controlled trial. Phys Ther. 2008; 88(3):333–340.
5. Kimberley TJ, Lewis SM, Auerbach EJ, Dorsey LL, Lojovich JM, Carey JR. Electrical stimulation driving functional improvements and cortical changes in subjects with stroke. Exp Brain Res. 2004; 154(4):450–460.
6. Reinkensmeyer D, Kahn LE, Averbuch M, McKenna-Cole A, Schmit BD, Rymer WZ. Understanding and treating arm movement impairment after chronic brain injury: progress with the ARM guide. J Rehab Res Dev. 2000; 37(6):653–662.
7. Richards LG, Senesac CR, Davis SB, Woodbury ML, Nadeau SE. Bilateral arm training with rhythmic auditory cueing in chronic stroke: not always efficacious. Neurorehabil Neural Repair. 2008; 22(2):180–184.
8. Volpe BT, Lynch D, Rykman-Berland A, et al.. Intensive sensorimotor arm training mediated by therapist or robot improves hemiparesis in patients with chronic stroke. Neurorehabil Neural Repair. 2008; 22(3):305–310.
9. Stein J, Narendran K, McBean J, Krebs K, Hughes R. Electromyography-controlled exoskeletal upper-limb–powered orthosis for exercise training after stroke. Am J Phys Med Rehabil. 2007;86:255–261.
10. Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S. The post-stroke hemiplegic patient. I. A method for evaluation of physical performance. Scand J Rehabil Med. 1975;7:13–31.
11. Duncan PW, Propst M, Nelson SG. Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cerebrovascular accident. Phys Ther. 1983;63:1606–1610.
12. DiFabio RP, Badke RB. Relationship of sensory organization to balance function in patients with hemiplegia. Phys Ther. 1990; 70(9):542–548.
13. Carsweel A, McColl M, Baptiste S, Law M, Polatajko H, Pollock N. The Canadian Occupational Performance Measure: a research and clinical literature review. Can J Occ Ther. 2004; 71(4):210–222.
14. Cup EHC, Scholte op Reimer WJM, Thijssen MCE, van Kuyk-Minis MAH. Reliability and validity of the Canadian Occupational Performance Measure in stroke patients. Clin Rehabil. 2003;17:402–409.
15. Duncan PW, Bode RK, Min Lai S, Perera S, Glycine Antagonist in Neuroprotection Americans Investigators. Rasch analysis of a new stroke-specific outcome scale: the Stroke Impact Scale. Arch Phys Med Rehabil. 2003;84(7):950–963.
16. Latham NK, Jette DU, Coster W, et al.. Occupational therapy activities and intervention techniques for clients with stroke in six rehabilitation hospitals. Am J Occup Ther. 2006; 60(4):369–378.
17. Housman SJ, Scott KM, Reinkensmeyer DJ. A randomized controlled trial of gravity-supported, computer-enhanced arm exercise for individuals with severe hemiparesis. Neurorehabil Neural Repair. 2009; 23(5):505–514.
hemiplegia; occupational therapy; robotics; stroke