Each year, nearly 2 million people experience a stroke in the United States and Europe combined,1,2 leaving roughly half of the survivors severely impaired, with a chronically nonfunctional hand or arm.3–5 There are few treatment options for persons with severe sensorimotor impairment of the hand.6 Conventional treatment for severely impaired individuals focuses primarily on compensation, rather than restoring normal motor function.7–10
Impairment-oriented therapy may be more efficacious than functional retraining in persons with severe hemiparesis.11–13 Conceptually, impairment-oriented therapy involves the diagnosis and treatment of specific somatic sensorimotor impairments associated with stroke and other central nervous system injuries. Typical impairments are weakness, hyperactive tone, contracture, co-contraction, and somatosensory loss. Specificity in impairment is defined not only by the type of impairment but also by the specific joint and direction of joint motion at the impaired joint. Different approaches may be required to treat different types of impairment, but accurate diagnostic and effective therapeutic approaches remain to be developed for those with severe impairment following stroke.4,6,11
The most commonly observed motor impairment associated with stroke is weakness in finger extension,6,14 which severely restricts hand-opening and thereby limits many activities of daily living.15,16 Finger extension may also be limited by “oppositional impairments” such as hyperactive flexor tone, contracture, and flexor co-contraction, although flexor co-contraction is thought to contribute more significantly to paresis than hyperactive tone or contracture.17–20 In addition to motor impairments, somatosensory deficits likely also contribute to paresis following stroke.6,11 Recovery of somatosensation may contribute to the restoration of movement following stroke.1,21 As oppositional impairments can limit finger extension, functional recovery of the hand may require some remediation of oppositional impairments in addition to strengthening the finger extensor.22,23
In people with severe upper extremity impairment, a clinical focus on retraining function rather than reversing impairment might be related to a lack of adequate tools to treat different types of impairments.7 We developed an intervention that combines robotic-assisted movement with muscle vibration (assisted movement with enhanced sensation [AMES]) intended to remediate impairments in the limbs by combining assisted movement, biofeedback of active torque or electromyographic (EMG) activity, and muscle vibration to remediate weakness, hyperactive tone, co-contraction, and somatosensory loss. The robotic device moves the user's distal arm or leg while vibrating the antagonist (ie, lengthening) muscle(s). The user participates by simultaneously attempting to move the limb in the same direction as the applied motion.24 A real-time visual display of either active torque or EMG informs the user if he or she is assisting the motion adequately, that is, at 25% strength. The robotic device cyclically rotates the joint (5°/second, ±15°) in the flexion-extension direction while muscle vibration is applied to the lengthening (antagonist) muscles.
In a previous study, we showed that a combination of assisted movement, muscle vibration, and torque biofeedback applied over 6 months (90-100 thirty-minute sessions) resulted in significant and sustained increases in strength and joint position control in the arm or leg of stroke survivors with chronic, moderately severe to severe impairment.24 This reduction in impairment was associated with a return of limb function, but the 6-month intervention period may be prohibitively lengthy for many people with stroke. In the present study, our goal was to assess whether a shorter intervention period (30 sessions over 8-10 weeks) would mitigate chronic impairment and restore some functional finger extension in hands rendered plegic or severely paretic hand by stroke. In addition, we compared EMG biofeedback25–28 with torque biofeedback in people with impairment of finger extension following stroke. We hypothesized that EMG biofeedback would be more efficacious than torque biofeedback, because the plegic hand is unable to produce any net extension torque with the fingers.
Participants with stroke were consented, screened, and enrolled at 3 study sites: Emory University, Northwestern University, and Oregon Health & Science University. The baseline impairment level of enrolled participants was defined as an absence of all or most finger extension in the impaired hand, but with measurable volitional EMG in the finger extensor. Inclusion criteria consisted of (1) hemispheric (cortical/subcortical) ischemic or hemorrhagic stroke; (2) 1 year or more since the most recent stroke; (3) ages 18 to 80 years; (4) most active finger extension less than 5 cm, (5) absence of substantially elevated tone (≤3 Ashworth scale), (6) ability to produce volitionally 20 μV or more EMG in extensor digitorum communis, (7) physically/cognitively capable of consenting and complying with the protocol, and (8) capable of communicating with investigators. Exclusion criteria included (1) upper-limb flaccidity; (2) significant proprioceptive deficit (<14/20 correct on the Joint Position Test), (3) impaired upper limb use or upper limb pain not related to stroke, (4) cardiopulmonary compromise, (5) major psychiatric disorder, (6) severe cognitive or communication deficit, (7) upper limb too large/small for the device, (8) severe upper limb contracture, (9) progressive neurodegenerative disorders, (10) uncontrolled seizures, (11) drug/alcohol abuse, (12) terminal illness with anticipated survival less than 12 months, (13) concurrent participation in another upper limb study, (14) change in or initiation of ongoing prescribed therapy, including antispasmodic medication, (15) NIH Stroke Scale29 scores greater than 1 on the Sensory Item and greater than 1 on the Neglect Item. All enrolled participants completed the Montreal Cognitive Assessment (MOCA),30 with an average score (±SD) of 24.2 (±4.4) (median = 25, minimum = 15, maximum = 29).
The study coordinators at each of the 3 study sites used a computer-generated randomization table to allocate participants to 1 of 2 treatment groups: (1) robot-assisted movement and muscle vibration combined with EMG biofeedback (EMG Group) and (2) robot-assisted movement and muscle vibration combined with torque biofeedback (TOR Group). There was no control group. The evaluators were blinded to treatment group assignments.
To determine eligibility, the principal investigator or study physician at each study site screened candidates by reviewing their medical history and testing them with the Sensory and Neglect items of the NIH Stroke Scale. Hand proprioception was tested with the Joint Position Test31 by manually flexing or extending the candidate's hand at the wrist. Ten flexion movements and 10 extension movements were performed in a pseudorandom order, and the candidate, with eyes closed, reported the perceived direction of motion. Active finger extension range of motion was measured by having the candidate volitionally extend the fingers with the hand supported and the forearm positioned in neutral. The evaluator measured the linear distance of fingertip extension of the most mobile finger (if any) 3 times using a thin ruler, dusted with talcum powder and inserted between the fingers. Maximal volitional EMG in extensor digitorum communis was measured by having each candidate exert 3 maximal finger/thumb extensions and flexions with the hand-held isometric (see the “EMG Recording” section). Finger and wrist flexor tone was assessed with the modified Ashworth Scale.32
Standardized EMG recording interfaces were used to screen participants and to treat those assigned to the EMG Group. Before commencing the study, the principal investigator trained the treatment therapists at each site in the application of EMG electrodes using an illustrated manual. Pairs of 1-cm diameter, Ag-AgCl surface electrodes (Tyco, Hydro-Snap ES40036, Mansfield, Massachusetts) were applied 2.5 cm apart, in a proximal-distal orientation, over extensor digitorum communis and flexor digitorum profundus of the affected arm. A reference electrode was applied to the dorsum of the affected hand. For each participant, the locations of the 2 muscle recording sites were referenced to existing landmarks (eg, moles, freckles) or marked with indelible ink during each session to ensure consistent intersession placement. Any residual variability in electrode placement was assumed to be inconsequential due to the large number of recordings obtained from each participant (ie, 30). Normalization of EMG amplitude with respect to maximal levels was impractical due to very low levels of EMG, increases in volitional EMG levels due to the intervention, and incomplete recruitment inherent to maximal efforts.33,34
Raw EMG signals generated from the 2 forearm muscles were processed by an EMG amplifier (Noraxon 1200, Scottsdale, Arizona), which included band-pass filtering (10-500 Hz), amplification (×1000), and full-wave rectification. The analog outputs of the EMG amplifier were digitized at 1 kHz and smoothed with a 1-second moving average. In the absence of any EMG signal, the typical noise level was 10 μV or less. The EMG recorded from the agonist and antagonist muscles was used to move graphical objects (ie, 2 horizontal bars as biofeedback on a computer monitor).
The thumb-and-finger manipulator of the AMES investigational device was used in this study (Figure 1). The participant sat to the right or left of the device and placed the affected forearm in the device postured in a precision grip. The trainer secured the participant's arm in the device and positioned the plastic probes of 3 vibrators over the tendons of (1) flexors digitorum profundus and pollicis longus, (2) extensor digitorum communis, and (3) extensor pollicis longus.
Evaluation of Impairment and Activity
The experimental procedures are illustrated in the flowchart in Figure 2. Shortly after screening, enrollment, and randomization of a participant to 1 of the 2 treatment groups, a therapist evaluated the participant's affected hand. These evaluations were standardized by an independent therapist experienced in the evaluation tools used in this study, and all evaluators were trained in this standardized approach. To quantify hand impairment, the evaluators used the Upper Extremity Fugl-Meyer Assessment (UE-FMA), with a maximum score of 66.35 The Box-and-Block Test (BBT) was used to assess ability to perform functional hand activities. Participants grasped, transported, and released as many blocks as possible in 60 seconds.36 The BBT was designated as the primary outcome measure for the study. Finally, participants completed the Stroke Impact Scale (SIS), a self-evaluation questionnaire about perceived stroke recovery.37 After participants completed all 30 treatments, they were reevaluated with the UE-FMA, BBT, and SIS (Figure 2). The evaluating therapist was blinded as to each participant's treatment group assignment.
Each participant participated in 30 treatment sessions (Figure 2). At the beginning of each treatment session, a calibration procedure was performed to differentiate volitional components from nonvolitional components of torque or EMG. Participants held the hand as relaxed as possible while the device passively moved the thumb and fingers (5°/second, ±15°) 3 full cycles while recording the resistive torque (for the TOR Group) or EMG (for the EMG Group). Subsequently, during treatment, the AMES device subtracted the averaged calibration torque (or EMG) from the overall torque (or EMG) recording in order to feed back only the participants’ volitional effort.
The participant's peak volitional finger/thumb strength at the beginning of each session was measured using the AMES device. The participant performed 3 maximal hand-openings and hand-closings with the hand-held isometric, closed for extension strength and open (60° thumb-finger separation) for flexion strength. Each effort lasted 5 or fewer seconds, during which time the participant attempted, often in a discontinuous manner, to exceed the previous maximum. During the Strength Test, the participant viewed real-time visual feedback of active torque (TOR Group) or EMG (EMG Group). If the participant's flexor co-contraction was substantial, a net negative (ie, flexor) torque or EMG was sometimes produced while testing extension strength. Because participants’ efforts often fluctuated during this test, we limited our quantification of strength to peak torque in all participants, and to peak flexor and extensor EMG in the EMG Group.
Prior to the study, therapists at each site were trained for standardized use of the AMES device. During treatment, participants assisted the device as it cyclically moved the joints of the affected hand through their range of motion at 5°/second between hand closure (ie, precision grip) and 30° of thumb and finger extension (ie, 60° of separation) over a 30-minute period (ie, 150 total cycles/session). Figure 3 illustrates the recorded signals during 1 full cycle of treatment from an EMG-group participant.
Real-time visual biofeedback provided participants with a graphic display of the amount of assistance being provided with 1 (TOR Group) or 2 (EMG Group) horizontal bars, the bar-width being proportional to the participant's volitional torque or EMG. A target line on each bar corresponded to approximately 25% of the participant's maximum flexion or extension strength. During each half cycle of motion, the computer screen indicated each time the participant achieved this target torque or EMG.
The TOR Group participants viewed the same volitional torque graphic during every treatment. However, with the EMG Group, the computer screen displayed a single bar depicting agonist EMG amplitude in sessions 1 to 3, antagonist EMG amplitude in sessions 4 to 5, and 2 bars depicting agonist and antagonist EMG amplitudes in sessions 6 to 30. Participants were instructed to reach the agonist EMG target while minimizing antagonist EMG.
To determine whether the treatment affected the participants’ levels of impairment (UE-FMA, Strength Test), functional activity (BBT), and quality of life (SIS), the distribution of scores from each test was first assessed for normality with the Kolmogorov-Smirnov Test. The UE-FMA and SIS scores were normally distributed and, therefore, were individually subjected to analysis of covariance with SCORE as the dependent variable and TREATMENT group as the independent variable (SPSS v19; IBM Inc, Armonk, NY). The pretreatment scores from the UE-FMA and SIS were used as covariates to adjust for each participant's baseline level of impairment. Change scores were tested with a paired t test. SIS items Strength, Activities of Daily Living, Mobility, and Hand were averaged to provide a Physical Performance subscore.
The BBT and strength test torque scores were not normally distributed; therefore, the Wilcoxon signed rank test was used to compare the pretreatment and posttreatment scores using medians and 25th and 75th percentiles. For those participants showing improvement more than 0 on the BBT, we regressed the BBT change scores on the UE-FMA baseline scores.
Strength Test data consisted of torque values for TOR Group participants and both torque and EMG values for EMG Group participants. Baseline flexion and extension strength in all participants were defined as the average of the first 9 peak torque scores (TOR and EMG groups) from the first 3 sessions; posttreatment strength was defined as the last 9 scores from the last 3 sessions.24 For the analysis of EMG data (EMG Group), co-contraction was defined as the amplitude of antagonist EMG divided by that of agonist EMG. Decreases in co-contraction were defined as positive. A value of P ≤ 0.05 was used as the threshold for statistical significance for all outcome measures.
Forty-nine adult participants (aged 20-82 years; median = 55 years) were enrolled in the study. Time since stroke ranged from 1 to 39 years (median = 4 years, mean = 9 years). There were no significant differences in demographics and medical history between treatment groups (Table 1). Three participants dropped out without completing treatments or evaluations, in 2 due to the travel distance, and in the other due to a protocol deviation (excessive alcohol use). Three other enrolled participants were unable to participate because of study closure. The data presented here were obtained from 43 participants (22 women and 21 men) from whom both pre- and posttreatment data were available. There was a significant difference in baseline UE-FMA scores (Table 2), which indicated that the EMG Group was more impaired than the TOR Group at baseline. The 43 participants completing the study were severely impaired, and as shown in Figure 4, 28 of 43 participants had no active range of motion at baseline.
The UE-FMA scores within both treatment groups showed a significant increase (average change score = 4.1; paired t = 7.2, P < 0.001). After adjusting for baseline differences in the 2 groups (F1,38 = 125, P < 0.001), there was no significant between-group difference in UE-FMA change score over the course of treatment (F1,38 = 0.07, P = 0.79). The average UE-FMA results for the pooled treatment groups showed a significant increase in scores for items II/III (Dynamic Movement, P < 0.001), item IV (Mixing Synergies, P < 0.001), item V (Little Synergy, P < 0.01), and item VIII (Hand, P < 0.001), and a trend was observed for item VII (Wrist, P = 0.057).
The Strength Test results showed an increase in flexion torque strength, but not in extension torque strength (Table 2). Flexion torque strength scores increased from pretreatment to posttreatment when the treatment groups were combined (P < 0.001). The increase for the TOR Group was larger than that in the EMG Group (P < 0.002). Extension strength scores did not change from pretreatment to posttreatment (P = 0.11).
Activity and Quality-of-Life Outcomes
While the BBT scores indicated no between-group differences (P = 0.61), these scores increased across-groups from pre- to posttreatment (P < 0.005). The average score for the TOR Group increased from 3.9 to 5.8 blocks (P ≤ 0.05) and, for the EMG Group, from 0.4 to 1.0 blocks (P = 0.079). At baseline, 28 of 43 participants could not pick up a single block, but by the end of treatment, 4 of the 28 could pick up at least 1 block. The median scores (Table 3) did not change because of a disproportionate number of 0 scores.
Our analysis indicated that whether a particular participant improved in the BBT depended on that person's baseline impairment level. As shown in Figure 5, participants whose baseline UE-FMA score was 17 or more might improve on this test with 30 treatments, whereas no participant with a baseline UE-FMA score of less than 17 showed an improvement in the BBT. For those participants whose BBT score did improve, the amount of improvement was roughly proportional to their baseline UE-FMA scores. Of the 4 participants with no baseline finger extension whose BBT score increased, 3 were from the EMG Group, with final scores of 4, 3, and 1, whereas only 1 was from the TOR Group, with a final score of 1.
The Physical Performance subscore of the SIS (Table 3), within and across treatment groups, increased significantly from pretreatment to posttreatment (P ≤ 0.001). After adjusting for baseline (F1,40 = 25.6, P ≤ 0.001), there was no difference between treatment groups (F1,40 = 3.317, P = 0.076). Analysis of individual SIS items showed a significant increase in strength (P ≤ 0.0001), but there was no significant difference in activities of daily living, mobility, or hand. Item 9 (stroke recovery) also showed a significant increase from pretreatment to posttreatment (P < 0.001), without a significant difference between treatment groups. The study participants’ perceptions were that their quality of life improved over the course of treatment.
Change in Agonist and Antagonist EMG
During Strength Tests, both EMG and torque amplitudes were measured in the EMG Group (Table 4). During hand-opening Strength Tests with the EMG Group, mean finger extensor EMG increased from 23 μV to 51 μV (P ≤ 0.001) between pre- and posttreatment but did not change during hand-closing (P = 0.66). In contrast, finger flexor EMG did not change for either hand-opening (t test; P = 0.38) or hand-closing (P = 0.58). During hand-opening Strength Tests, the average level of co-contraction decreased (P ≤ 0.05), but co-contraction did not change for hand-closing Strength Tests (P = 0.83).
The outcomes of this study indicate that 30 treatments of assisted movement with muscle vibration reduce impairment in the severely paretic or plegic hand and, thereby create the possibility of functional recovery, from either continued use of the study device or other, more functionally oriented interventions. The UE-FMA, our primary measure of impairment, showed an overall average change score = 4.1, which is slightly less than a minimum detectable change of 5.2 previously reported for people with chronic spastic hemiplegia.38 However, the participants in that study were less impaired at baseline than those in the present study, and the UE-FMA is apparently more sensitive with severely, compared to moderately, impaired people.38 To date, the minimal clinically important difference (MCID) for the UE-FMA in people at the chronic stage of stroke recovery has not been determined, although Lin et al39 assumed an MCID of 6 to 10 (10%-15%) for this group. For people at the subacute stage of stroke recovery, Arya et al40 determined an MCID of 9 (19%). Based on these previous results, the UE-FMA change score observed in the present study reached the level of MCID on a percentage basis, but not on absolute score.
A fundamental premise of impairment-oriented therapy is that severe impairments not only limit function but also interfere with the recovery of function, and need to be remediated before functional recovery can take place.11,13 Accordingly, AMES was developed to address sensorimotor impairments.24 The inclusion of muscle vibration was intended to reduce hyperactive reflexes41–43 and to enhance somatosensory processing in the cortex.44,45 Furthermore, the combination of assisted movement and antagonist muscle vibration was intended to strengthen motor connections between the brain and spinal cord and, thereby, to increase strength by repetitively and synchronously pairing functionally related motor and proprioceptive activity in the brain. Somatosensory stimulation was intended to support such Hebbian-style plasticity. In this study, EMG biofeedback was combined with AMES with the intent to reduce flexor co-contraction in the hand.17
In this study, we extended earlier findings24 by assessing the value of assisted movement combined with vibration in a more impaired cohort of stroke survivors and involving fewer treatment sessions than had been studied previously, and by adding EMG biofeedback. In the earlier study, most participants had some volitional finger extension at baseline, while in the present study at enrollment most participants were unable to volitionally extend their fingers (Figure 4). We hypothesized that the plegic hand would respond better to EMG biofeedback, since with EMG biofeedback, participants with plegic hands could determine when they were activating the correct muscle. Our results, however, did not support this hypothesis, since no between-group differences were identified. One possible reason for this discrepancy is that, despite their randomized allocation, participants allocated to the EMG Group were significantly more impaired (UE-FMA) and less functional (BBT) than the TOR Group at baseline. Another possible explanation is that, after plegic TOR Group participants received several minutes of AMES treatment, vibration may have relaxed the spastic flexor muscles sufficiently that weak extensor contractions could then produce some net extension torque, which would be visible in the torque biofeedback. Extension torque and/or EMG typically increased during each session, and at the end of each session, participants could often extend more digits further than just prior to beginning treatment that day.
One limitation of the study was that the treatment and the evaluation of impairment were performed with the hand supported against gravity. Previous studies of people with spastic hemiplegia46,47 have shown that finger and wrist flexors involuntarily co-contract when the unsupported arm is actively moved, as during reaching and pointing. While this involuntary activation of the flexor synergy may limit volitional movement outside of the clinic, supporting the arm during treatment may not impact the efficacy of the intervention, since several of our initially plegic participants regained the ability to transfer blocks, and the half of participants who were able to move at least 1 block at baseline improved their scores on the BBT.
Further research is needed to determine whether AMES, with either EMG or torque biofeedback, is sufficiently efficacious to warrant its clinical use for severe hand paresis or plegia. To be effective, impairment-oriented therapy must reduce impairments enough to regain some functional use of the hand, or else gains achieved during therapy could be lost when therapy is discontinued. While this study did not involve a follow-up evaluation, our previous study24 on a slightly less impaired group of stroke survivors showed retention of gains 6 months posttreatment. Future studies should be designed to ensure that treatment groups are balanced at baseline through the use of a fixed or adaptive randomization procedure and, possibly, by treating only extension motion so that finger flexors are not strengthened, since the latter might increase the resistance to finger extension in hands with flexor co-contraction. If these changes provide a more robust demonstration of AMES efficacy in those with severe hand impairment, a path can then be established to sequence best practice approaches from the most impaired to the least impaired, since at present, little benefit is obtained in severely impaired stroke survivors from existing interventions.
Assisted movement and muscle vibration, combined with either EMG or torque biofeedback, appear to reduce upper limb impairment, improve volitional activation of the hand muscles, and restore a modicum of hand function in some persons with severe hand impairment due to chronic stroke. Balanced treatment groups will be required in a future study of the severely impaired upper limb, to determine whether EMG biofeedback is more efficacious than torque biofeedback in restoring hand-opening when this biofeedback is combined with assisted movement and muscle vibration.
1. Doyle S, Bennett S, Fasoli SE, McKenna KT. Interventions for sensory impairment in the upper limb after stroke (Review). Cochrane Collaboration. 2010;6:1–56.
2. Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics—2023 update: a report from the American Heart Association. Circulation. 2012;125:e2–e220.
3. Kwakkel G, Kollen BJ, Krebs HI. Effects of robot-assisted therapy on upper limb recovery after stroke: a systematic review. Neurorehabil Neural Repair. 2008;22:111–121.
4. Platz T, Eickhof C, van Kaick S, et al. Impairment-oriented training or Bobath therapy for severe arm paresis after stroke: a single-blind multicentre randomized controlled trial. Clin Rehabil. 2005;19:714–724.
5. Wade DT. Measuring arm impairment and disability after stroke. Int Disabil Stud. 1989;11:89–92.
6. Raghavan P. The nature of hand motor impairment after stroke and its treatment. Curr Treat Options Cardiovasc Med. 2007;9:221–228.
7. Barreca S, Wolf SL, Fasoli S, Bohannon R. Treatment interventions for the paretic upper limb of stroke survivors: a critical review. Neurorehabil Neural Repair. 2003;17:220–226.
8. Teasell RW, Foley NC, Bhogal SK, Chakravertty R, Bluvol A. A rehabilitation program for patients recovering from severe stroke. Can J Neurosci. 2005;32:512–517.
9. Pereira S, Teasell R, Graham R. Efficacy of severe stroke rehabilitation. Stroke. 2010;41:P166.
10. Finley MA, Fasoli SE, Dipeitro L, et al. Short-duration robotic therapy in stroke patients with severe upper-limb motor impairment. J Rehabil Res Dev. 2005;42:683–692.
11. Platz T. Impairment-oriented training (IOT)—scientific concept and evidence-based treatment strategies. Restor Neurol Neurosci. 2004;22:301–315.
12. Krebs HI, Mernoff S, Fasoli SE, Hughes R, Stein J, Hogan N. A comparison of functional and impairment-based robotic training in severe to moderate chronic stroke: a pilot study. Neurorehabilitation. 2008;23:81–87.
13. Platz T, van Kaick S, Mehrholz J, Leidner O, Eickhof C, Pohl M. Best conventional therapy versus modular impairment-oriented training for arm paresis after stroke: a single-blind, multicenter randomized controlled trial. Neurorehabil Neural Repair. 2009;23:706–716.
14. Trombly CA, Thayer-Nason L, Bliss G, Girard CA, Lyrist LA, Brexa-Hooson A. The effectiveness of therapy in improving finger extension in stroke patients. Am J Occup Ther. 1986;40:612–617.
15. Lang CE, DeJong SL, Beebe JA. Recovery of thumb and finger extension and its relation to grasp performance after stroke. J Neurophysiol. 2009;102:451–459.
16. Fritz SL, Light KE, Patterson TS, Behrman AL, Davis SB. Active finger extension predicts outcomes after constraint-induced movement therapy for individuals with hemiparesis after stroke. Stroke. 2005;36:1172–1177.
17. Kamper DG, Rymer WZ: Impairment of voluntary control of finger motion following stroke: role of inappropriate muscle coactivation. Muscle Nerve. 2001;24:673–681.
18. Kamper DG, Fischer HC, Cruz EG, Rymer WZ: Weakness
is the primary contributor to finger impairment in chronic stroke. Arch Phys Med Rehabil. 2006;87:1262–1269.
19. Wagner JM, Lang CE, Sahrmann SA, et al. Relationships between sensorimotor impairments and reaching deficits in acute hemiparesis. Neurorehabil Neural Repair. 2006;20:406–416.
20. Leonard CT, Gardipee KA, Koontz JR, Anderson J, Wilkins SA. Correlation between impairment and motor performance during reaching tasks in subjects with spastic hemiparesis. J Rehabil Med. 2006;38:243–249.
21. Casadio M, Morasso P, Sanguineti V, Giannoni P. Minimally assistive robot training for proprioceptive enhancement. Exp Brain Res. 2009;194:219–231.
22. Wolf SL, Baker MP, Kelly JL. EMG
biofeedback in stroke: effect of patient characteristics. Arch Phys Med Rehabil. 1979;60:96–102.
23. Wolf SL, Baker MP, Kelly JL. EMG
biofeedback in stroke: a one year follow-up on the effect of patient characteristics. Arch Phys Med Rehabil. 1980;61:351–355.
24. Cordo P, Lutsep H, Cordo L, Wright W, Cacciatore T, Skoss R. Assisted movement with enhanced sensation (AMES): coupling motor and sensory to remediate motor deficits in chronic stroke patients. Neurorehabil Neural Repair. 2009;23:67–77.
25. Dogan-Aslan M, Nakipoglu-Yüzer GF, Dogan A, Karabay I, Ozgirgin N. The effect of electromyographic biofeedback treatment in improving upper extremity functioning of patients with hemiplegic stroke. J Stroke Cerebrovasc Dis. 2012;21:187–192.
26. Lourenção MI, Battistella LR, de Brito CM, Tsukimoto GR, Miyazaki MH. Effect of biofeedback accompanying occupational therapy and functional electrical stimulation in hemiplegic patients. Int J Rehabil Res. 2008;31:33–41.
27. Wolf SL. Electromyographic biofeedback applications to stroke patients. Phys Ther. 1983;63:1448–1459.
28. Wolf SL, Binder-MacLeod SA. Electromyographic biofeedback applications to the hemiplegic patient. Changes in upper extremity neuromuscular and functional status. Phys Ther. 1983;63:1393–1403.
29. Goldstein LB, Bertels C, Davis JN. Interrater reliability of the NIH Stroke Scale. Arch Neurol. 1989;46:660–662.
30. Nasreddine AS, Phillips NA, Bédirian V, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53:695–699.
31. Gilman S. Joint position sense and vibration sense: anatomical organization and assessment. J Neurol Neurosurg Psychiatry. 2002;73:473–477.
32. Bohannon RW, Smith MB. Interrater reliability of a modified Ashworth scale of muscle spasticity
. Phys Ther. 1987;67:206–207.
33. Berlanger A, McComas A. Extent of motor unit activation during effort. J Appl Physiol. 1981;51:1131–1135.
34. Herbert R, Gandevia S. Twitch interpolation in human muscles: mechanisms and implications for measurement of voluntary activation. J Neurophysiol. 1999;82:2271–2283.
35. Fugl-Meyer AR. Post-stroke hemiplegia assessment of physical properties. Scand J Rehabil Med Suppl. 1980;7:85–93.
36. Duncan P, Wallace D, Lai S, Johnson D, Embretson S, Laster LJ. The Stroke Impact Scale Version 2.0. Evaluation of reliability, validity, and sensitivity to change. Stroke. 1999;30:2131–2140.
37. Derosiers J, Bravo G, Hébert R, Butil E, Mercier L. Validation of the Box and Block Test as a measure of dexterity of elderly people: reliability, validity, and norms studies. Arch Phys Med Rehabil. 1994;75:751–755.
38. Lin J, Hsu M, Sheu C, et al. Psychometric comparisons of 4 measures for assessing upper-extremity function in people with stroke. Phys Ther. 2009;89:840–850.
39. Lin K, Hsieh Y, Wu C, Chen C, Jang Y, Liu J. Minimal detectable change and clinically important difference of the Wolf Motor Function Test in stroke patients. Neurorehabil Neural Repair. 2009;23:429–434.
40. Arya KN, Verma R, Garg RK. Estimating the minimal clinically important difference of an upper extremity recovery measure in subacute stroke patients. Top Stroke Rehabil. 2011;18:599–610.
41. Hagbarth K-E, Eklund G. The effects of muscle vibration in spasticity
, rigidity, and cerebellar disorders. J Neurol Neurosurg Psychiat. 1968;31:207–213.
42. Ageranioti S, Hayes KC. Effect of vibration on hypertonia and hyperreflexia in the wrist joint of patients with spastic hemiparesis. Physiother Can. 1990;42:24–33.
43. Noma T, Matsumoto S, Etoh S, Shimodozono M, Kawahira K. Anti-spastic effects of the direct application of vibratory stimuli to the spastic muscle of hemiplegic limbs in post-stroke patients. Brain Injury. 2009;23:623–631.
44. Marconi B, Filippi GM, Koch G, et al. Long-term effects on cortical excitability and motor recovery induced by repeated muscle vibration in chronic stroke patients. Neurorehabil Neural Repair. 2011;25:48–60.
45. Forner-Cordero A, Steyvers M, Levin O, Alaerts K, Swinnen SP. Changes in corticomotor excitability following proplonged muscle tendon vibration. Behav Brain Res. 2008;190:41–49.
46. Wolf SL, Binder-Macleod SA. EMG
biofeedback applications to the hemiplegic patient: changes in upper extremity neuromuscular and functional status. Phys Ther. 1983;63:1393–1403.
47. Miller L, Dewald J. Involuntary paretic wrist/finger flexion forces and EMG
increase with shoulder abduction load in individuals with chronic stroke. Clin Neurophysiol. 2012;123:1216–1225.