Abstract
Introduction
This scoping review aimed to identify and chart rehabilitation methods for severe upper limb motor impairments in patients with chronic stroke, incorporating both non-randomized controlled trials and randomized controlled trials (RCTs) to provide a broader perspective.
Methods
A search was conducted using PubMed and Web of Science. Data were managed using the Rayyan software. Two independent reviewers screened titles, abstracts, and full-text articles. Inclusion criteria included severe upper limb motor impairment, chronic stroke, RCTs, quasi-RCTs, and crossover studies.
Results
From 802 extracted articles, 16 met the inclusion criteria: 87.5% were RCTs, 12.5% were crossover trials, whereas 0% were quasi-RCTs. Major interventions included robot-assisted training (25%), noninvasive brain stimulation (25%), and electrical stimulation (31.3%). Although motor function improvements were frequently reported, few studies showed statistically significant differences compared with control groups.
Discussion
This review systematically mapped the range of study designs and rehabilitation methods identified in the literature, providing a structured overview of current evidence for severe upper limb motor impairments in patients with chronic stroke.
Introduction
Stroke is often accompanied by significant sequelae that substantially affect activities of daily living (ADL). Upper limb motor impairment is common after stroke, affecting up to 80% of survivors early after onset. 1 In acute stroke cohorts, upper limb weakness has been observed in approximately one-third of patients within the first days poststroke. 2 Recovery remains limited for many individuals, with fewer than half regaining upper limb function by 6 months. 3 Motor impairment severity has been shown to correlate with the use of the paretic upper limb in daily activities, with more severe impairments leading to reduced use of the affected limb. 4 This phenomenon of acquired nonuse may further exacerbate functional decline. 5 For patients with chronic stroke, recovery of upper limb function remains a critical rehabilitation priority. 6 Severe motor impairment is also associated with substantial psychological burden, including feelings of loss and frustration.7,8 Therefore, improving severe motor impairments remains essential for enhancing quality of life in this population.
Various upper limb rehabilitation strategies, including Constraint-Induced Movement Therapy (CIMT), task-oriented training, mirror therapy, electrical stimulation, and motor imagery therapy, have been effectively used in clinical practice. Emerging interventions such as repetitive transcranial magnetic stimulation are under investigation for their potential clinical application. However, the diversity of these methods has complicated the formulation of an effective rehabilitation strategy. Recently, systematic reviews using network meta-analyses (NMAs) have been conducted to address this issue.7,9 Saikaley et al. 7 collected comprehensive evidence without limiting the disease stage or poststroke severity and reported the effectiveness of CIMT, mirror therapy, and motor imagery therapy. Mirror therapy has been widely investigated in stroke rehabilitation. A Cochrane review published in 2018 10 reported beneficial effects of mirror therapy on motor function and activities of daily living in people after stroke. However, the included studies encompassed a broad range of stroke stages and severity levels. The evidence specifically targeting individuals with severe upper limb motor impairment in the chronic phase appears to be comparatively limited. In addition, intensive upper limb rehabilitation programs have demonstrated meaningful improvements even in individuals in the chronic phase of stroke. For example, Ward et al. 11 reported clinically relevant gains following a high-intensity upper limb neurorehabilitation program (Queen Square programme), highlighting the potential of structured, high-dose interventions in chronic stroke populations. However, in clinical practice, developing optimal rehabilitation strategies based on the disease stage and severity of each patient poststroke is crucial. For instance, although CIMT has generally been effective in patients with mild motor impairments,12,13 evidence regarding its effectiveness in patients with severe motor impairments is limited. 14 Although mirror therapy has shown broad efficacy in the chronic phase,15–17 studies have reported its ineffectiveness in the acute phase.18,19 To date, most systematic reviews have focused on interventions for patients with mild-to-moderate impairment or have included mixed severity populations without stratification. Consequently, a significant gap remains in the literature regarding evidence-based rehabilitation options specifically for patients with severe upper limb motor impairment in the chronic phase of stroke, who often experience persistent disability and for whom treatment options are limited. This lack of consolidated evidence leaves clinicians with insufficient guidance when selecting appropriate interventions for this subgroup, despite the high clinical demand and unmet patient needs.
Furthermore, existing systematic reviews have primarily relied on randomized controlled trials (RCTs), which may not fully capture the range of available interventions or reflect real-world clinical practice. Including a broader range of study designs (e.g., quasi-RCTs, crossover trials) can offer a more comprehensive understanding of feasible and emerging rehabilitation approaches for this challenging population. Therefore, this study aimed to comprehensively explore and chart the available evidence for rehabilitation strategies targeting patients with severe upper limb motor impairment in the chronic phase of stroke, using a scoping review methodology that incorporates multiple study designs. The results of this review are intended to help guide future research and clinical decision-making for this understudied and clinically important population.
Materials and methods
This scoping review was conducted in accordance with the PRISMA extension for Scoping Reviews (PRISMA-ScR) guidelines and followed the methodological framework proposed by Arksey and O’Malley, 20 further refined by Levac et al. 21 The study protocol was registered on the Open Science Framework (registration doi:10.17605/OSF.IO/AVKYC). Two databases, PubMed and Web of Science, were searched for articles published from 1975 to May 31, 2025. These databases were selected because they comprehensively index peer-reviewed biomedical and rehabilitation research literature. The search terms used in PubMed were (“Randomized Controlled Trial”) OR (“CrossOver Studies” [Mesh]) AND (“Stroke” [Mesh]) AND (“Upper Extremity” [Mesh]) AND (“chronic” [Title/Abstract]) AND (“rehabilitation” [MeSH Terms]). The search terms used in the Web of Science were ((MeSH HEADING:exp: (Stroke) AND MeSH HEADING:exp: (Rehabilitation)) AND MeSH HEADING:exp: (Upper Extremity)) AND TOPIC: (chronic) AND ((Randomized Controlled Trial) OR (Cross-Over Studies)). All articles were imported via Rayyan (Qatar Computing Research Institute, Qatar), and duplicate articles were removed. Two independent reviewers (Ryo Harigai and Shohei Shimizu) used Rayyan to screen the titles, abstracts, and full texts to determine the eligibility of each article. In cases of disagreement, the reviewers discussed the issue, and if it remained unresolved, a third reviewer (Katsuya Sakai) was consulted.
The inclusion criteria for the RCTs, quasi-RCTs, and crossover trials were as follows: (1) patients aged ≥18 years; (2) patients diagnosed with stroke (ischemic or hemorrhagic) with upper limb motor impairment; and (3) patients with >6 months poststroke history. Additionally, the studies were required to (4) have an average Fugl-Meyer Assessment Upper Extremity (FMA-UE) score of <22 at the start of the intervention; (5) involve upper limb rehabilitation interventions; (6) include a control group receiving no intervention (sham) or usual care; (7) evaluate upper limb motor function (FMA-UE); (8) describe pre- and postintervention assessment results; (9) be written in English; and (10) have the full text available. The FMA-UE criterion was selected because it is a widely used and well-established measure of upper limb motor impairment after stroke. The cut-off value of <22 was adopted in accordance with Doumen et al., 22 who defined severe upper limb motor impairment using this threshold in a review of rehabilitation interventions. By applying the same predefined criterion, we aimed to ensure methodological consistency and alignment with existing literature when identifying a clearly defined severe population. As this study is a scoping review aimed at comprehensively collecting literature, we also included studies with small sample sizes. The following individuals and studies were excluded: (1) patients with traumatic brain injury, transient ischemic attack, subarachnoid hemorrhage, or other neurological disorders; (2) studies involving nonhuman subjects; (3) those not specifying the duration since stroke onset; (4) those not reporting the average FMA-UE score at the start of the intervention; (5) those lacking rehabilitation intervention details; (6) those using pharmacological therapy in the intervention group; and (7) those not clearly describing the pre- and postintervention assessment results. Furthermore, (8) review articles, (9) conference proceedings, and (10) protocol papers were excluded.
The following information was collected for data extraction: (1) author names; (2) year of publication; (3) country; (4) study design; (5) number of participants; (6) average age; (7) time since stroke onset; (8) baseline assessments, including FMA-UE, Action Research Arm Test (ARAT), Wolf Motor Function Test (WMFT), Box and Block Test (BBT), Motor Assessment Scale (MAS), Stroke Impairment Assessment Set (SIAS), and National Institutes of Health Stroke Scale (NIHSS); (9) intervention details for each group (method, duration, frequency, and intensity); and (10) outcome measures, including pre- and postintervention results. If the necessary information was not available in the articles, the authors were contacted via email. If no response was received, it was assumed that the authors did not provide any relevant information.
The risk of bias in the included studies was assessed using the PEDro scale. Two independent reviewers (Ryo Harigai and Shohei Shimizu) performed the assessments; in cases of disagreement, a third reviewer (Katsuya Sakai) was consulted to reach a consensus. The third reviewer decided whether to accept or reject the paper based on the inclusion and exclusion criteria.
Although methodological quality was assessed using the PEDro scale, a formal reporting checklist evaluation (e.g., CONSORT adherence) was not conducted, consistent with the scoping review framework, which aims to map the breadth of available evidence rather than perform an exhaustive critical appraisal of reporting quality.
Results
Figure 1 illustrates the PRISMA diagram of the review process. A total of 802 articles were identified from the two databases. After removing duplicates, 660 records were subjected to primary screening, of which 421 were excluded. The remaining 239 records underwent secondary screening, of which 223 were excluded. Ultimately, 16 studies met the inclusion criteria and were included in this review. The reasons for exclusion included being <6 months poststroke, not having an RCT or randomized crossover trial design, differing outcomes, lack of a specified FMA-UE score, or unavailability of the full text. Finally, 16 studies were included in this review. The results of these 16 studies are summarized in Table 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram for scoping review identification. Results of mapping. ARAT: Action Research Arm Test, BBT: Box and Block Test, CG: Control Group, CT: Conventional Therapy, cTBS: Continuous Theta Burst Stimulation, ES: Electrical Stimulation, FES: Functional Electrical Stimulation, FMA-UE: Fugl-Meyer Assessment Upper Extremity, IG: Intervention Group, MAS: Motor Assessment Scale, MT: Mirror Therapy, n.d.: not described, RAT: Robotic Assisted Therapy, RCT: Randomized Controlled Trial, ROM: Range of Motion, rTMS: Repetitive Transcranial Magnetic Stimulation, tDCS: Transcranial Direct Current Stimulation, VR: Virtual Reality, WMFT: Wolf Motor Function Test.
Study design, sample size, and stroke phase
Among the 16 included articles, 14 (87.5%) were RCTs, two (12.5%) were randomized crossover trials, whereas 0% were quasi-RCTs. The sample sizes ranged from 9 to 62 participants (mean 25.8 ± 12.8), with a total of 412 participants. The duration since stroke onset ranged from 214.20 ± 69.52 days to 112.4 ± 128.5 months. The preintervention FMA-UE scores ranged from 8.5 ± 1.2 to 21.33 ± 0.8 points.
Intervention
The breakdown of intervention methods was as follows: Four studies (25%) involved robot-assisted training; four studies (25%) involved noninvasive brain stimulation; five studies (31.3%) involved electrical stimulation; one study (6.3%) involved virtual reality; one study (6.3%) involved mirror therapy; and one study (6.3%) involved exercise therapy with assistive devices. The intervention frequency varied from at least two to five times per week, and the intervention duration ranged from 2 to 12 weeks, with no consistent pattern. Robot-assisted training utilizes devices such as InMotion,28,30 T-WREX, 27 and HEXO-UR30A, 29 all of which support the movement of the paretic upper limb and facilitate exercise therapy. Noninvasive brain stimulation includes low-frequency repetitive transcranial magnetic stimulation (LF-rTMS),23,24 continuous theta burst stimulation (cTBS), 24 and transcranial direct current stimulation (tDCS).25,26 Among the 16 studies, 14 investigated pre- and postintervention changes or group differences using statistical analysis. Among these, 11 studies reported improvements postintervention, with six studies reporting significant differences between groups.23,24,26,32–34 The breakdown of studies reporting group differences was as follows: three studies involved noninvasive brain stimulation,23,24,26 and three studies involved electrical stimulation.32–34
Although six types of rehabilitation methods were identified in the included studies, three methods (VR, mirror therapy, and exercise therapy with assistive devices) were represented individually by only one study. While four studies involved noninvasive brain stimulation, differences in the stimulation methods used existed, including LF-rTMS, cTBS, and tDCS. Four studies involved robotic devices; however, the devices used varied across studies.
Risk of bias
Risk of bias.
Discussion
In this scoping review, we conducted a comprehensive analysis of 16 studies to examine rehabilitation strategies for severe upper limb motor impairment in patients with chronic stroke. The primary findings revealed that various interventions, including robot-assisted training, noninvasive brain stimulation, and electrical stimulation, were utilized, with many reporting improvements in motor function postintervention in RCT and crossover study designs. However, few studies demonstrated statistically significant differences when compared with control groups, suggesting that although rehabilitation for severe motor impairment poststroke can improve outcomes, superior intervention options are limited. Previous research has validated the effectiveness of rehabilitation strategies for severe motor impairment during the acute to subacute phases7,9; this study provides an overview of evidence for rehabilitation strategies for severe upper limb motor impairment during the chronic phase.
Our findings provide an overview of current evidence on rehabilitation strategies for severe upper limb motor impairment in patients with chronic stroke, including RCT and crossover study designs. Robot-assisted training and noninvasive brain stimulation have been extensively studied and are frequently reported in the literature.39–44 These interventions have been applied across various stages of stroke recovery, including the chronic phase, indicating their potential applicability for severe motor impairment.
Regarding noninvasive brain stimulation, different stimulation methods such as LF-rTMS, cTBS, and tDCS have been employed, which poses a challenge for direct comparison of their effects.23–26 The variability in stimulation parameters and settings across studies highlights the need for research to establish effective conditions and propose standardized protocols. Similarly, variations in the devices used across robot-assisted training studies exist, affecting consistency and reproducibility in clinical applications.
Additionally, rehabilitation strategies for upper limb motor impairment seem applicable across different disease stages, provided that the severity is similar. Doumen et al. 22 reported effective strategies for severe motor impairment during the acute and recovery phases, such as robot-assisted training and electrical stimulation, which were also found to be effective in patients with chronic stroke in this study. However, recent neuroscientific research has indicated differing pathophysiologies between mild-to-moderate and severe motor impairments,45–48 suggesting the need for tailored rehabilitation strategies. Future research should focus on developing strategies tailored to the specific pathophysiology of severe motor impairments.49–52
A significant challenge identified was the limited number of studies on interventions such as VR and mirror therapy. High-quality studies are required to establish the efficacy of these methods. Additionally, despite the relatively high number of studies on noninvasive brain stimulation and robot-assisted training, the diversity of devices and stimulation methods used means that only a limited number of individual studies exists for each method. Therefore, the reported effectiveness may change as more studies are conducted.
The findings of this review highlight that a majority of the included studies (11 out of 16) were of high quality, as indicated by their PEDro scores of ≥6. Importantly, five out of six studies showing statistically significant differences between the intervention and control groups also met this quality threshold, suggesting that higher methodological rigor may be associated with positive outcomes in poststroke rehabilitation. The presence of these bias risks highlights the need for future research to improve the rigor of study designs, particularly in terms of methodological transparency and adherence to high-quality standards. More robust study designs, with clear reporting on allocation concealment, comprehensive blinding strategies, and proper implementation of ITT analysis, would strengthen the evidence base for poststroke rehabilitation interventions and facilitate the translation of findings into clinical practice.
Strengths and limitations
This is the first scoping review to summarize rehabilitation strategies for severe upper limb motor impairment in patients with chronic stroke. By analyzing the study design, participant characteristics, intervention types, and measured outcomes in detail, we clarified the current state of rehabilitation strategies. However, this study had several limitations. First, the restrictions on certain languages and databases may have led to the exclusion of relevant studies. Second, the limited number of included studies warrants caution when generalizing the results. Addressing these limitations requires systematic reviews and NMAs to provide detailed insights. Third, restricting inclusion to studies reporting FMA-UE scores to define severe motor impairment may have excluded studies using alternative outcome measures and contributed to the limited number of eligible studies.
Conclusion
In summary, rehabilitation strategies for severe upper limb motor impairment in patients with chronic stroke are diverse, with many showing significant efficacy. However, the evidence for several interventions requires further elucidation. Although this review demonstrated that high-quality studies tend to report favorable outcomes, addressing common sources of bias will be critical to advancing the field. Future research should focus on filling these gaps, particularly through high-quality RCTs, to verify the efficacy of new interventions. By improving study design and addressing methodological concerns, future research can provide a more definitive understanding of effective rehabilitation strategies for patients with stroke.
Footnotes
Author contributions
Conceptualization: Ryo Harigai (RH), Methodology: RH, Katsuya Sakai (KS), Software: RH, Formal analysis: RH, Shohei Shimizu (SS), Data curation: RH, Resources: RH, Writing-Original draft preparation: RH, Visualization: RH, Investigation: RH, SS, Supervision: KS, Validation: KS, Writing-Reviewing and Editing: RH, KS, Project administration: RH. All authors agree to be accountable for all aspects of the work.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Ryo HARIGAI is employed by BRAIN Co., Ltd., which provides rehabilitation services. This employment is unrelated to the conduct and content of this study. The other authors declare no conflicts of interest.
Data Availability Statement
Data are available on request from the authors.