Secondary Logo

Journal Logo

Original Research Articles

Nerve Stimulation Enhances Task-Oriented Training for Moderate-to-Severe Hemiparesis 3–12 Months After Stroke

A Randomized Trial

Carrico, Cheryl MS, OT/L; Westgate, Philip M. PhD; Salmon Powell, Elizabeth MS; Chelette, Kenneth C. MS; Nichols, Laurie BS, OT/L; Pettigrew, L. Creed MD, MPH; Sawaki, Lumy MD, PhD

Author Information
American Journal of Physical Medicine & Rehabilitation: November 2018 - Volume 97 - Issue 11 - p 808-815
doi: 10.1097/PHM.0000000000000971


Neuroplasticity (i.e., the capacity for reorganization of the central nervous system) is a basis for functional recovery after stroke.1 Repetitive sensory activation has been shown to modulate neuroplasticity (i.e., enlarge motor cortical representation) and improve motor function in animal models.2 This evidence indicated that sensory stimulation can modulate motor cortical plasticity, thus establishing a mechanism by which sensory input plays a role in motor skill acquisition.2–4 A therapeutic intervention based on this mechanism in humans is called somatosensory stimulation (SS). Somatosensory stimulation applies repetitive, noninvasive, low-level electrical currents to activate large cutaneous and proprioceptive sensory fibers without producing muscle contractions.5 There is some evidence that SS modulates neuroplasticity and can enhance upper limb (UL) motor function in humans after stroke.6–8 For example, Sawaki and colleagues conducted a sham-controlled study of the effects of 2 hours of SS on voluntary thumb movement in seven subjects with mild hemiparesis (i.e., able to perform isolated movement of paretic thumb) at more than 6 mos after stroke. Results demonstrated that SS increased encoding of kinematic details of movements evoked with transcranial magnetic stimulation.8 In addition, Laufer and colleagues' systematic review of randomized and quasi-randomized trials concluded that SS may enhance motor recovery after stroke.7 However, most of the 15 reviewed studies were conducted during the chronic stage of recovery, and the review did not aim to establish how severity of deficit may determine intervention responsiveness. In sum, it has not yet been established how SS may impact moderately-to-severely impaired UL function in the 3- to 12-mo period after stroke.

Intensive, task-oriented motor training has been associated with extensive neuroplastic change, as well as more significant functional gains than usual and customary care, in cases of mild-to-moderately impaired UL motor function after stroke.9 Similarly, studies showing that SS can enhance outcomes of intensive task-oriented motor training have primarily targeted mild-to-moderate impairment.10–13,14 The purpose of the present study was to investigate whether active SS would enhance outcomes of intensive task-oriented motor training significantly more than sham SS in subjects with moderately-to-severely impaired UL motor function in the 3- to 12-mo period after stroke.


In accordance with the Declaration of the World Medical Association (, this study was approved by the authorized institutional human research review board governing the research (i.e., the University of Kentucky and Cardinal Hill Hospital in Lexington, Kentucky). All subjects provided written informed consent after receiving a verbal and written explanation of the purposes, procedures, and potential hazards of this study. All study procedures were followed in accordance with institutional guidelines. Subjects were recruited from communities, hospitals, and clinics in the local and regional areas surrounding the study site. The study was conducted at a neurorehabilitation research laboratory located at HealthSouth Cardinal Hill Rehabilitation Hospital in Lexington, Kentucky. The first tier of inclusion criteria included the following: (a) having sustained a single ischemic or hemorrhagic stroke during the 3- to 12-mo period preceding enrollment; (b) inability at the time of screening to demonstrate active extension of the paretic metacarpophalangeal and interphalangeal joints at least 10 degrees and the wrist 20 degrees; and (c) 18 years or older. The second tier of inclusion criteria included having a baseline score of 47 or lower on the modified 30-item Fugl-Meyer Assessment (FMA) that measures UL motor function.15 Subjects with impaired sensation at baseline were not excluded, and subjects with various degrees of impaired sensation were enrolled. Exclusion criteria were selected to minimize potential confounding variables. These criteria included the following: (a) history of carpal tunnel syndrome and/or documented peripheral neuropathy; (b) addition or change in the dosage of drugs known to exert detrimental effects on motor recovery within 3 mos of recruitment16; and (c) aphasia or cognitive deficit severe enough to preclude informed consent.

This study was a randomized controlled parallel group superiority trial. Standardized evaluations of UL motor function were administered by an occupational therapist at baseline, after completion of the intervention period, and at 1- and 4-mo follow-up time points. After baseline evaluation, the principal investigator used a computer program to generate a simple random allocation sequence for assigning subjects into two groups (i.e., either active SS paired with intensive task-oriented training; or sham (control) SS paired with intensive task-oriented training). The only independent variable was SS. To maximize enrollment and completion of the study, eligible participants were randomized into a 1.5 active/1 sham ratio. One of the members of the research study staff (LS) enrolled all participants and made all assignments to interventions according to the randomizer program. Each intervention session took place 3 times per week for 6 wks (18 total sessions) and consisted of either active or sham SS (2 hours) immediately preceding intensive task-oriented motor training (4 hours). Subjects, care providers, and evaluators of motor function were blinded to group assignment (i.e., they were not provided with information about which SS condition any subject received). Somatosensory stimulation was delivered by one of the members of the research study staff (LS) or a biomedical engineer. Therapists who evaluated motor function or who administered intensive task-oriented training did not administer SS.

Sample Size

Sample size was calculated based on preliminary data in seven subjects with chronic, UL hemiparesis after stroke. In the preliminary study, 4 subjects received 2-hour daily hand SS preceding 4-hour motor training for 10 consecutive weekdays. The remaining 3 patients received sham SS preceding 4-hour motor training for 10 consecutive weekdays. The difference of changes in scores on Wolf Motor Function Test (WMFT) (logarithm scaled) after intervention for experimental and control groups was 0.07, and the common standard deviation was approximately 0.09. To detect the same effect size as that observed in the preliminary study with 80% power, assuming a similar standard deviation in the change, a total of 55 subjects were enrolled in the present study.

Evaluation and Outcome Measures

The WMFT served as the primary outcome measure. The time-based portion of the WMFT encompasses a battery of 15 tasks that simulate functional tasks and that are ordered according to complexity.9 The WMFT scoring calculates the mean of the 15 time-based tasks. Because of the skewed distribution of the WMFT, a logarithmic transformation is performed on the mean of these 15 time-based WMFT measures. There are no units, because it is a logarithm. Values are denoted as log (mean WMFT). The WMFT has established reliability and validity and has been extensively applied in research to evaluate UL motor capacity after stroke.9,17 The minimal clinically important difference (MCID) for WMFT has been cited as 1.5 to 4 seconds18 (0.18–0.6 log scaled); however, there are no available data on MCID for WMFT as measured in stroke populations with severely impaired motor function. Secondary outcome measures included the FMA UL motor score, the Action Research Arm Test (ARAT), and the Stroke Impact Scale (SIS). The FMA is a quantitative measure of motor recovery, sensation, coordination, and speed and is based on the principle that motor recovery occurs in a predictable progression.19 The FMA has an extensive history of clinical and research application in stroke populations19 as well as high interrater reliability (=0.89–0.98 according to the subset for lower or UL) and test-retest reliability (=0.99).20 The modified 30-item FMA15 does not calculate reflex scores and is a unidimensional measure of volitional movement in which the highest possible motor score for a tested UL is 60.21 The MCID for FMA is 9 to 10 points18; however, the MCID for the modified 30-item FMA15 has not been established relative to any level of motor impairment after stroke. The ARAT measures grasp, grip, pinch, and other indices of rehabilitation-related change in UL motor capacity.17,22 The highest possible ARAT score for a tested UL is 57. The MCID for ARAT has been cited as 5.7 points18; however, there are no available data on MCID for ARAT as measured in stroke populations with severely impaired motor function. The SIS is a 59-item measure that assesses the following 8 domains: strength; hand function; activities of daily living/instrumental activities of daily living; mobility; communication; emotion; memory and thinking; and participation/role function. Each item is rated on a five-point Likert scale. Scores range from 0 to 100. Several studies have proven its reliable psychometric attributes, including reliability, validity, and sensitivity to change.23,24

Intervention Component 1: SS (Currently Still an Investigational Device per Food and Drug Administration)

Somatosensory stimulation was delivered for 120 mins at the beginning of each session. Optimal positions to stimulate the Erb's point, radial, and median nerves were determined by applying a surface bar electrode with the cathode placed distally on the paretic UL. During identification of the optimal site for stimulation, the irradiation of the stimulation was recorded in cases where subjects could feel the SS. Where subjects could not feel the SS, the investigators elicited movements of the muscle that corresponded to the nerve that investigators wanted to stimulate. Once the location was set, the intervention stimulation was set to evoke compound muscle action potentials of 50–100 μV in the deltoid, triceps, and opponens pollicis brevis, respectively, which in most cases fell below sensory threshold (i.e., nonperceived afferent stimulation) (Fig. 1). When subjects had substantially impaired sensation, optimal stimulation sites were identified by visualizing muscle contractions of the deltoid, triceps, and opponens pollicis brevis muscles, respectively. To stimulate each nerve trunk, 10-mm gold-plated stimulating electrodes were placed on the skin to stimulate sites concurrently with the cathode positioned proximally over each of the optimal positions identified by the bar electrode.5 The stimulation consisted of 10-Hz trains of five pulses, with each pulse lasting 1 millisecond. Trains of pulses were delivered at 1 Hz. Disposable surface electromyographic electrodes were placed over the belly of the deltoid, triceps, and opponens pollicis brevis muscles. Electromyographic activity was amplified and filtered (bandpass, 10–3000 Hz) and recorded using a data collection program written in LabVIEW (National Instruments, Austin, TX). For active SS, the stimulus intensity was adjusted to elicit small compound muscle action potentials of approximately 50 to 100 μV without visible muscle movements.5 This low stimulus intensity and the stimulus duration of 1 millisecond has been shown to preferentially activate large cutaneous and proprioceptive sensory fibers without producing muscle contractions.5 While efferent activation can occur at a small level with this intensity, the PNS electrode montage was set to direct the majority of current from distal to proximal. For sham SS, an identical protocol was implemented except that the stimulator amplitude was set to 0 V after compound muscle action potentials were identified. Each subject was informed in each session that he or she might or might not feel sensations associated with SS and that sensations did not indicate whether SS was being delivered. Because attention seems to play an important role in neuroplastic change,25 subjects were required to stay awake during SS. During stimulation, subjects sat with the paretic arm over a pillow in a comfortable position per subject report during TV watching or reading.

Electrode position and pulse train. This figure illustrates the placement of electrodes on the paretic UL for delivery of SS to the targeted nerves. The image in the lower left corner illustrates the timing of the pulse train.

Intervention Component 2: Intensive Task-Oriented UL Motor Training

In each session, each subject participated in 4 hours of intensive task-oriented UL motor training immediately after SS. An occupational therapist blinded to SS condition administered the training. Training focused on highly repetitive use of the paretic UL in unilateral and bilateral task-oriented motor activities. Tasks had progressive difficulty to elicit movement skill just beyond the level already achieved (i.e., shaping).26 Each session incorporated tasks eliciting progressive increase in endurance. These tasks occupied a minimum of 2 hours of training and a maximum of 3.5 hours of training by the end of the second week of training. The remainder of the 4-hour window in each session was used for rest or bathroom breaks. The total time for rest or bathroom breaks varied across subjects and sessions according to each subject's fatigue or personal needs. Tasks were repeatable and targeted functional goals (such as activities of daily living) or prerequisites to function (e.g., releasing, grasping, reaching, supination). Training included rest breaks and grading of activities according to subject fatigue and impairment. There was no movement monitoring system, but therapists provided each subject with verbal or visual (graph) feedback on the quality of task performance during training. Rest breaks lasted no longer than the practice segment. The target range for repetitions of any given task was 10 to 50 according to the demands of the task as well as reported levels of fatigue and engagement of each subject with a given task. Therapy took place in a 1:1 therapist-to-subject ratio. Missed sessions were rescheduled so that each participant completed all 18 sessions.


For each outcome of interest, a longitudinal repeated measures model that accounts for categorical time, trial arm, and their interaction was fit. Each model incorporates an unstructured working covariance matrix, and the Kenward and Roger27 degrees of freedom method was used for inference. These analyses correspond to the use of repeated measures multivariate analysis of variance, but with the allowance of missing data. Primary interest was in the comparison of mean changes in outcomes from baseline to immediately postintervention and to 1- and 4-mo follow-up for the 2 trial arms. To ensure the appropriateness of statistical assumptions for inference, the square root was applied to ARAT and log(mean WMFT) values. With the WMFT, log(mean WMFT) values were subtracted from the maximum sample value of 2.08 before applying the square root. To convey the clinical relevance of results, estimated means and standard errors are presented on the original scale for ARAT and WMFT. All available data were used for analyses. All tests were two-sided, with statistical significance prespecified as P < 0.05. Analyses were conducted in SAS Version 9.4 (SAS Institute, Cary, NC).


Supplemental Digital Content 1 ( shows how this study met the CONSORT checklist for randomized controlled trials. Table 1 summarizes baseline scores and demographics of the sample. Fig. 2 depicts the study flow. The date range defining the periods of data collection was July 2008 to February 2015.

Baseline scores and demographics of the sample
Study flow.

The trial ended because funding was completed. There were no complications or serious adverse events related to the study. No evidence of unintended effects in each group was found. Analysis was by original assigned groups. Fig. 3 depicts between-group differences as well as within-group mean change associated with each intervention condition. Results of all analyses are presented in Tables 2 to 4. For more details from the models, the separate impacts of each trial arm on the mean change of each outcome are presented.

Graphs of change on all outcomes. Intervention-related changes on WMFT, ARAT, FMA, and SIS are shown. All graphs indicate mean change in score relative to baseline for each group. Change associated with active SS is shown in the darker shade. To ensure the appropriateness of statistical assumptions, the square root was applied to ARAT and log(mean WMFT) values. The WMFT and ARAT values were back-transformed to raw values to calculate the means and standard errors shown to facilitate interpretation of the results. However, the statistics shown are based on the transformed values. For WMFT, improvement is represented as a negative change along the y-axis. For ARAT, FMA, and SIS, improvement is depicted as positive change along the y-axis. Significant within-group change is denoted with a single asterisk. Significant between-groups difference is denoted with a double asterisk. Error bars denote 1 standard error of the mean.
Estimated means, standard errors, and P values corresponding to mean change on all outcomes from baseline to postintervention
Estimated means, standard errors, and P values corresponding to mean change on all outcome measures from baseline to 1-mo follow-up
Estimated means, standard errors, and P values corresponding to mean change on all outcomes from baseline to 4-mo follow-up

Statistically significant between-groups differences favored active SS on the following evaluations at the following time points: WMFT at post; and ARAT at post, 1 mo, and 4 mos. No significant between-groups differences on FMA were present at any time. Statistically significant between-groups differences favoring sham SS were evident on SIS at 1 mo.

Statistically significant improvement was associated with active SS as indicated by within-group mean change on the following evaluations at the following time points: WMFT at post, 1 mo, and 4 mos; ARAT at post, 1 mo, and 4 mos; FMA at post, 1 mo, and 4 mos; and SIS at post and 4 mos.

Statistically significant improvement was associated with sham SS as indicated by within-group mean change on the following evaluations at the following time points: FMA at post, 1 mo, and 4 mos; and SIS at post, 1 mo, and 4 mos. No statistically significant within-group mean change on WMFT or ARAT was associated with sham SS at any time point.


With regard to the main hypothesis, between-groups analyses revealed that active SS enhanced outcomes of intensive task-oriented training significantly more than sham SS. In addition, there was statistically significant within-group improvement associated with active SS on all time points for all outcome measures except SIS at 1-mo follow-up. In contrast, sham SS was associated with statistically significant within-group improvement on only two of the four outcome measures. Thus, SS seems to have potential to serve as a clinical strategy for significantly improving outcomes of intensive task-oriented motor training in cases of moderate-to-severe UL hemiparesis during the 3- to12-mo period after stroke. However, more studies are needed in this regard, in part because the ARAT was the only outcome measure that reflected statistically significant differences between groups at every time point after intervention. In addition, as the MCID for each outcome measure has not been established with regard to severely impaired motor function after stroke, it is unclear what conclusions are warranted about the clinical significance of this study's findings. Even though some findings reflected statistical significance, the clinical effects may be small.

The FMA predominantly measures body structure and function,28 whereas WMFT, ARAT, and SIS predominantly measure activity.17,28 Specifically, WMFT and ARAT measure what an individual can do in a standardized environment29 and questionnaires such as the SIS measure perceived performance.17 Thus, based on the present study's statistical findings, active SS paired with motor training can be considered to have improved not only body structure and function (i.e., FMA) but also activity (i.e., WMFT, ARAT, SIS). Because sham SS did not significantly improve any outcomes measured by ARAT and WMFT, motor training alone may be insufficient to improve what can be performed in a standardized environment. Future studies incorporating measures that are sensitive and specific to effects at home or outside the laboratory (e.g., movement monitoring with accelerometers, actual performance, activity limitations, participation restrictions), as well as how these measures relate to disability and clinical relevance, are recommended. In addition, there were no obvious indications that impaired sensation affected outcomes, but future studies should investigate this possibility.

Findings from this study should be considered in light of possible limitations. First, WMFT was prespecified as the primary outcome in this study because the WMFT was used in the Extremity Constraint Induced Therapy Evaluation (EXCITE) trial, which showed that constraint-induced movement therapy led to more significant motor improvements than usual care in 222 subjects with mild-to-moderate hemiparesis at 3 to 9 mos after stroke.9 However, other studies have indicated that the WMFT may not be optimal in cases of severely impaired motor function17,30 because the instrument becomes susceptible to floor effects, as well as nonnormal distribution of data, in such cases.30 A second limitation in the present study is that the FMA's possible lack of sensitivity to change in cases of chronically impaired motor function31 may have obscured finding significant differences between groups on what the FMA purports to measure. Additional limitations included unequal allocation ratio as well as losses to follow-up at the 1- and 4-mo time points that reduced the statistical power and value of the outcomes at those times. Although unequal allocation has been cited as an acceptable approach in double-blind randomized trials,32 use of equal allocation ratio should be investigated in future studies. Finally, to aid in interpretation of this study's statistically significant findings, further research is needed to establish the MCID on each outcome measure relative to stroke populations with severely impaired motor function.


Active SS seems to have potential for translation to a clinical strategy that may optimize outcomes of intensive task-oriented motor training for people with moderate-to-severe UL hemiparesis in the 3- to 12-mo period after stroke. Future studies should measure actual performance, activity limitations, and participation restrictions to substantiate how SS paired with motor training can impact disability and functional recovery for this population. In addition, to elucidate neurophysiological mechanisms underlying the effects of intervention, future studies should incorporate outcome measures that quantify neuroplastic change.


1. Nudo RJ, Plautz EJ, Frost SB: Role of adaptive plasticity in recovery of function after damage to motor cortex. Muscle Nerve 2001;24:1000–19
2. Recanzone GH, Allard TT, Jenkins WM, et al.: Receptive-field changes induced by peripheral nerve stimulation in si of adult cats. J Neurophysiol 1990;63:1213–25
3. Pavlides C, Miyashita E, Asanuma H: Projection from the sensory to the motor cortex is important in learning motor skills in the monkey. J Neurophysiol 1993;70:733–41
4. Sakamoto T, Arissian K, Asanuma H: Functional role of the sensory cortex in learning motor skills in cats. Brain Res 1989;503:258–64
5. Kaelin-Lang A, Luft AR, Sawaki L, et al.: Modulation of human corticomotor excitability by somatosensory input. J Physiol 2002;540:623–33
6. Ikuno K, Kawaguchi S, Kitabeppu S, et al.: Effects of peripheral sensory nerve stimulation plus task-oriented training on upper extremity function in patients with subacute stroke: a pilot randomized crossover trial. Clin Rehabil 2012;26:999–1009
7. Laufer Y, Elboim-Gabyzon M: Does sensory transcutaneous electrical stimulation enhance motor recovery following a stroke? A systematic review. Neurorehabil Neural Repair 2011;25:799–809
8. Sawaki L, Wu CW, Kaelin-Lang A, et al.: Effects of somatosensory stimulation on use-dependent plasticity in chronic stroke. Stroke 2006;37:246–7
9. Wolf SL, Winstein CJ, Miller JP, et al.: Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: The EXCITE randomized clinical trial. JAMA 2006;296:2095–104
10. McDonnell MN, Hillier SL, Miles TS, et al.: Influence of combined afferent stimulation and task-specific training following stroke: a pilot randomized controlled trial. Neurorehabil Neural Repair 2007;21:435–43
11. Kim TH, In TS, Cho HY: Task-related training combined with transcutaneous electrical nerve stimulation promotes upper limb functions in patients with chronic stroke. Tohoku J Exp Med 2013;231:93–100
12. Fleming MK, Sorinola IO, Roberts-Lewis SF, et al.: The effect of combined somatosensory stimulation and task-specific training on upper limb function in chronic stroke: a double-blind randomized controlled trial. Neurorehabil Neural Repair 2015;29:143–52
13. Dos Santos-Fontes RL, Ferreiro de Andrade KN, Sterr A, et al.: Home-based nerve stimulation to enhance effects of motor training in patients in the chronic phase after stroke: a proof-of-principle study. Neurorehabil Neural Repair 2013;27:483–90
14. Carrico C, Chelette KC 2nd, Westgate PM, et al.: Randomized trial of peripheral nerve stimulation to enhance modified constraint-induced therapy after stroke. Am J Phys Med Rehabil 2016;95:397–406
15. Woodbury ML, Velozo CA, Richards LG, et al.: Rasch analysis staging methodology to classify upper extremity movement impairment after stroke. Arch Phys Med Rehabil 2013;94:1527–33
16. Goldstein LB, Davis JN: Restorative neurology. Drugs and recovery following stroke. Stroke 1990;21:1636–40
17. Lemmens RJ, Timmermans AA, Janssen-Potten YJ, et al.: Valid and reliable instruments for arm-hand assessment at ICF activity level in persons with hemiplegia: a systematic review. BMC Neurol 2012;12:21
18. Pandian S, Arya KN: Stroke-related motor outcome measures: do they quantify the neurophysiological aspects of upper extremity recovery? J Bodyw Mov Ther 2014;18:412–23
19. Gladstone DJ, Danells CJ, Black SE: The Fugl-Meyer Assessment of motor recovery after stroke: a critical review of its measurement properties. Neurorehabil Neural Repair 2002;16:232–40
20. Duncan PW, Propst M, Nelson SG: Reliability of the Fugl-Meyer Assessment of sensorimotor recovery following cerebrovascular accident. Phys Ther 1983;63:1606–10
21. Woodbury ML, Velozo CA, Richards LG, et al.: Dimensionality and construct validity of the fugl-meyer assessment of the upper extremity. Arch Phys Med Rehabil 2007;88:715–23
22. Lyle RC: A performance test for assessment of upper limb function in physical rehabilitation treatment and research. Int J Rehabil Res 1981;4:483–92
23. Duncan PW, Lai SM, Bode RK, et al.: Stroke impact scale-16: a brief assessment of physical function. Neurology 2003;60:291–6
24. Lo AC, Guarino PD, Richards LG, et al.: Robot-assisted therapy for long-term upper-limb impairment after stroke. N Engl J Med 2010;362:1772–83
25. Ridding MC, McKay DR, Thompson PD, et al.: Changes in corticomotor representations induced by prolonged peripheral nerve stimulation in humans. Clin Neurophysiol 2001;112:1461–9
26. Wolf SL, Thompson PA, Winstein CJ, et al.: The EXCITE stroke trial: comparing early and delayed constraint-induced movement therapy. Stroke 2010;41:2309–15
27. Kenward MG, Roger JH: Small sample inference for fixed effects from restricted maximum likelihood. Biometrics 1997;53:983–97
28. Santisteban L, Teremetz M, Bleton JP, et al.: Upper limb outcome measures used in stroke rehabilitation studies: a systematic literature review. PLoS One 2016;11:e0154792
29. World Health Organization: Towards a common language for functioning, disability and health: ICF the International Classification of Functioning, Disability and Health. Published 2002. Accessed January 24, 2017
30. Hodics TM, Nakatsuka K, Upreti B, et al.: Wolf Motor Function Test for characterizing moderate to severe hemiparesis in stroke patients. Arch Phys Med Rehabil 2012;93:1963–7
31. van der Lee JH, Beckerman H, Lankhorst GJ, et al.: The responsiveness of the Action Research Arm Test and the Fugl-Meyer Assessment scale in chronic stroke patients. J Rehabil Med 2001;33:110–3
32. Doostfatemeh M, Taghi Ayatollah SM, Jafari P: Power and sample size calculations in clinical trials with patient-reported outcomes under equal and unequal group sizes based on graded response model: a simulation study. Value Health 2016;19:639–47

Upper Limb; Occupational Therapy; Humans; Transcutaneous Electric Nerve Stimulation; Neuronal Plasticity

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.