Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2014;95:968-85

REVIEW ARTICLE

Assistive Technologies: Can They Contribute to Rehabilitation of the Upper Limb After Stroke? Sybil Eleanor Farmer, PhD,a Venugopal Durairaj, MSc,a Ian Swain, PhD,b,c Anand David Pandyan, PhDa,d From the aInstitute for Science and Technology in Medicine, Keele University, Keele; bSchool of Design, Computing and Engineering, Bournemouth University, Bournemouth; cDepartment of Medical Physics and Clinical Engineering, Salisbury NHS Foundation Trust, Salisbury; and dSchool of Health and Rehabilitation, Keele University, Keele, United Kingdom.

Abstract Objective: To systematically identify, review, and explore the evidence for use of assistive technologies (ATs) in poststroke upper limb rehabilitation. Data Sources: AMED, CINAHL, Cochrane Library, Compendex, CSA Illumina, EMBASE, MEDLINE, PEDro, PyscINFO, and Web of Science were last searched in September 2011. Study Selection: Two independent researchers screened for inclusion criteria (adult poststroke subjects, upper limb rehabilitation with an AT). The risk of bias was assessed. Randomized controlled trials of poststroke subjects with baseline equivalence as assessed by blinded assessors were selected for data extraction. Data Extraction: Details of subjects, experimental and control treatments, and all outcomes were recorded in a spreadsheet. Data Synthesis: These data were used to calculate effect sizes for all outcome measures. Impairment measures ranged from
.39 (95% confidence interval [CI],
1.14 to .62) to 1.46 (95% CI, .72e2.20). Measures of activity effect sizes were from .04 (95% CI,
.35 to .44) to .93 (95% CI,
.39 to 2.25); for Motor Activity Log, from .07 (95% CI,
.66 to .80) to 1.24 (95% CI, .47e2.01); and for participation, from
3.32 (95% CI,
4.52 to 2.11) to 1.78 (95% CI, 0e3.56). Conclusions: AT treatments appear to give modest additional benefit when compared with usual care or in addition to usual care. This is most apparent for subjects early poststroke with 2 caveats: high-intensity constraint-induced movement therapy and electrical stimulation exclusively to the shoulder appear detrimental. The heterogeneity of treatment parameters and population characteristics precludes specific recommendations. Research would benefit from modeling studies to explicitly define criteria of population, intervention, comparator, and outcomes for effective treatments before the development of efficiently integrated care pathways. Archives of Physical Medicine and Rehabilitation 2014;95:968-85 ª 2014 by the American Congress of Rehabilitation Medicine

After a stroke, people present with a variety of neurologic symptoms. Disordered motor control is common1 and can present as spasticity (defined variably from stretch-induced muscle activity to loss in fractionated control of voluntary movement), paresis (complete or incomplete), or both.2,3 Disordered motor control leads to a reduction in functional use of the limbs and

Supported by the National Institute for Health Research (NIHR) under its Program Grants for Applied Research Program (grant no. RP-PG-0707-10012). The views expressed in this article are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health. No commercial party having a direct financial interest in the results of the research supporting this article has conferred or will confer a benefit on the authors or on any organization with which the authors are associated.

consequent atrophy,4 diminished limb range of movement, and increased joint and intramuscular stiffness,5 and at the shoulder there can be pain and subluxation.6 Loss of sensory awareness can also contribute to poor coordination of movement, neglect, and learned nonuse.7 Dexterity can also be compromised by this impaired control. These combined impairments limit active arm use and inhibit participation of many aspects of life.8 Changed cognition and mental status are also common and can affect the ability to carry out activities essential for daily living.9,10 It is estimated that less than 50% of stroke survivors will recover arm function.11 Rehabilitation therapies are the mainstay of treatment to facilitate recovery of function and integration into society after a stroke.

0003-9993/14/$36 - see front matter ª 2014 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2013.12.020

Assistive technologies for poststroke arm rehabilitation Approximately 10% of the National Health Service budget for stroke is likely to be spent on stroke rehabilitation services.12 Traditionally, rehabilitation has primarily depended on manual therapies and passive devices (eg, splints and positioning devices). Since the 1950s, there has been an interest in identifying technological solutions to enhance the recovery potential of stroke survivors.13,14 The aim of this study was to systematically review the efficacy of these technologies currently used to rehabilitate the arm after stroke (Assistive Technologies for Rehabilitation of the Arm following Stroke [ATRAS] was a National Institute for Health Researchefunded research program). For the purposes of this review, an assistive technology (AT) was defined as a mechanical or electrical device used in a functional taskeoriented training process that will have a systematic or rehabilitative effect on a person. With the use of this definition, the following technologies are considered in this review: biofeedback, brain stimulation, neuromuscular electrical stimulation (NMES), robotics, virtual reality, and constraint-induced movement therapy (CIMT). (CIMT was considered to be an assistive device as the mitt it is used in to induce active learning, thereby having a rehabilitative effect.) The objective of this study was to consider the published evidence for the effects of ATs in remediating the impairments of the upper limb after stroke and the consequent effects on activity and participation.15

Methods Data sources The databases accessed were AMED, CINAHL, Cochrane Library, COMPENDEX, CSA Illumina, EMBASE, MEDLINE, PEDro, PsycINFO, and Web of Science. The duration of search was from the origins of the databases to September 2011. The search terms and strategy for MEDLINE are listed in appendix 1.

969 bias of the potentially relevant articles is shown by the results in this appendix, presented as the number of studies (percentage of total studies) that reviewers agreed should be recorded negatively. We determined to use these results to identify studies that reported randomized controlled trials with independent assessors, and subjects with a similar prognosis at baseline, since such studies are thought to have a low risk of bias and may be used to identify patients who might benefit from treatments studied.

Data extraction The number of subjects, AT and treatment parameters, comparator, time poststroke of initial treatment, and data of all outcome measures were extracted from these studies by a researcher and then independently checked by another researcher. Outcomes were classified as “improved” if the score increased from baseline, “no change” if the score remained the same, and “deteriorated” if the score decreased from baseline.

Data synthesis We considered that the heterogeneity in the subject and treatment characteristics was too great to combine data in a meta-analysis, so we treated each article as an individual data point. These above extracted data were used to calculate effect sizes17 for all outcome measures of each study separately (including measures that were ordinal). Appendix 3 shows the equations used to calculate effect size (equation 1) and the SE of the effect size (used to calculate confidence intervals [CIs]; equation 2). Odds ratios18 were calculated for the 1 article that showed the number of subjects benefiting from treatment.19 Effect sizes were considered with respect to the International Classification of Functioning, Disability and Health and time poststroke.

Results Study selection Study details from searches were then imported into a reference manager (RefWorksa). After duplicate removal, the titles and abstracts of all studies were screened by 2 independent researchers to ensure that articles met the inclusion criteria (peer-reviewed articles in English of poststroke adult human studies for treatment of the upper limb with ATs). Full copies of these articles were then assessed for risk of bias, external validity, and availability of extractible data by 2 independent reviewers using a modified version of the quality assessment tool developed by van Tulder et al.16 The questions used are shown in appendix 2. The risk of

List of abbreviations: ARAT Action Research Arm Test AT assistive technology ATRAS Assistive Technology for Rehabilitation of the Arm following Stroke CI confidence interval CIMT constraint-induced movement therapy FM Fugl-Meyer MAL Motor Activity Log MCID minimally clinically important difference NMES neuromuscular electrical stimulation SIS Stroke Impact Scale WMFT Wolf Motor Function Test

www.archives-pmr.org

After search, a total of 4633 articles were found; this reduced to 2871 articles after duplicates were removed. A total of 467 experimental studies met the inclusion criteria. Forty-one studies were identified as randomized controlled trials of subjects with baseline equivalence, with outcomes measured by independent assessors. Most of these 41 studies were either NMES (17 articles19-35) or CIMT (12 articles36-47). Seven studies48-54 considered robotics, 255,56 used biofeedback, and 1 study each used brain stimulation,57 virtual reality,58 and stochastic resonance.59 The number of articles from each database and for each AT is shown in figure 1. A wide range of outcome measures were used in these articles, so we consider the effects of these treatments by grouping outcome measures with respect to impairment (range of motion, grip strength, subjective assessment of strength using the Medical Research Council Scale for muscle strength, Fugl Myer [FM]), activity (Action Research Arm Test [ARAT], Wolf Motor Function Test [WMFT]), and participation (Motor Activity Log [MAL] Amount of Use, MAL Quality of Movement, FIM, Barthel index, Rankin score, Stroke Impact Scale [SIS]). In tables 1 through 3, results are grouped into ATs in the following order: biofeedback, CIMT, robotics, and electrical stimulation (including stochastic resonance). Comparators are listed. Joints treated are shown in the next column, ordered by number of joints treated (most to least joints). Sample sizes are shown as number of control and experimental treatment subjects and

970

S.E. Farmer et al

Fig 1 Flowchart of AT article selection. Abbreviations: RCTs, randomized controlled trials; Stim, stimulation; TMS, transcranial magnetic stimulation.

expressed as a percentage of the total number of subjects. For follow-up, sample size and length of follow-up are reported. Effect sizes and 95% CIs for immediate effects and long-term effects are shown for measures of impairment in table 1, measures of activity in table 2, and measures of participation in table 3. Alongside effect sizes, the change in outcome measure score is shown (improvement, [; no change, 4; deterioration, Y), with most studies reporting improvement in both control and experimental subjects. There is time equivalence between the control and AT in 2421-23,25-27,29,31-33,37,39,40-45,48,49,51,53,54,56,59 of the 41 studies. Seven studies21,25,29,33,53,55,56 used an AT in conjunction with routine therapy, with the comparator being an extended period of routine therapy. Seven studies22,26,27,31,55,56,59 used a placebo, but 431,55,56,59 of these also provided either routine treatment or exercises to the control arm. The main results for the single-blind studies comparing AT with a control treatment or no treatment are summarized in tables 1 through 3. These 31 studies had a total of 1267 participants; numbers of control and experimental subjects are reported and expressed as a percentage of the total number of participants in these studies. It was not possible to extract data from a number of articles that had presented their

data graphically.28,34,46,47 Other studies14,20,24,30,52,57,58 are described at the end of this section.

Impairment Effect sizes for impairment varied from
.39 (95% CI,
1.14 to .62), which was associated with a loss of range of motion,49 to 1.46 (95% CI, .72e2.20), which was associated with an increase in deltoid strength.53 Grip strength and extensor strength improved with electrical stimulation27,29 but did not significantly change in 3 individual studies: 1 CIMT,36 1 robotics,49 and 1 NMES21 study. The FM score is more frequently reported, with some studies reporting total FM and others reporting proximal and distal subsection scores (see table 1). Chae et al31 provided stimulation to the extensor digitorum communis and the extensor pollicis, followed by functional training with the experimental group, achieving lower FM scores after the NMES phase of treatment but showing greater improvements after functional training than the control group (that received sensory stimulation). FM total effect size after functional www.archives-pmr.org

Measures of impairment

Impairment Measures

AT

Comparator

Measures of Biofeedback Routine treatment and ROM, grip, placebo treatment vs and Routine treatment and strength experimental treatment Time equivalence56 CIMT Usual care vs Experimental treatment36

Robotics

Routine treatment vs Experimental treatment Time equivalence49 Routine treatment þ control treatment vs Routine treatment þ experimental treatment53

ES

Routine therapy þ control treatment vs Routine therapy þ experimental treatment Time equivalence21 Routine therapy þ experimental treatment vs Routine therapy30 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence29 Experimental treatment vs Placebo treatment27

Joints Treated W/F

Sample Size (% of Total Sample Size) CZ13 TZ14 (2.13%)

Sh/E/W/F CZ116 TZ106 (17.52%) Sh/E/W/F CZ116 TZ10 (17.52%) Sh/E/W/F CZ10 TZ8 (1.42%) Sh/E/W/F CZ16 TZ15 (2.44%) Sh/E CZ18 TZ17 (2.76%) Sh/E CZ18 TZ17 (2.76%) Sh/E CZ18 TZ17 (2.76%) Sh/W/F CZ10 TZ10 (1.57%) Sh/W/F CZ10 TZ10 (1.57%)

Outcome Measure AROM

Immediate Effects .49 (
.57 to 1.54)

NR

NR

WMFT weight

NR

NR

ROM


.39 (
1.14 to .62)

C[; T[

Grip strength MRC biceps* MRC deltoid* MRC wrist flex* AROM Grip power

0 (
.70 to .70) .37 1.46
.14 .90

0 (
.88 to .88)

C[; T[

W/F

CZ28 TZ29 (4.50%) CZ28 TZ29 (4.50%) CZ28 TZ29 (4.50%) CZ8 TZ8 (1.26%)

Active extension

.11 (
.59 to .82)

Wrist moment at 0 Grip strength

.99 (.44 to 1.53) .38 (
.16 to .93)

Strength

.31 (
.85 to 1.47)

.11 (
.64 to .86)

C[; T[

.52 (
.17 to 1.20)

C[; T[

(6m) (8m) 1.70 (.92 to 2.47)

C[; T[

.12 (
.54 to .79)

C[; T[

(8m) (8mo)

C[; T[


.13 (
1.87 to 1.61) C[; T4 CZ6 TZ6 (3mo)

Wrist position

W/F

CZ116
.07 (
4.86 to 1.52) C[; T[ TZ106 (12mo) CZ116 .05 (
1.1 to 2.88) C[; T[ TZ106 (12mo)

CZ14 TZ14 (
.30 to 1.04) CY; T[ CZ18 TZ17 (.72 to 2.20) CY; T[ CZ18 TZ17 (
.81 to .52) C4; T[ CZ18 TZ17 (
3.37 to 5.18) C[; T[

CZ6 TZ6 (.95%)

W/F

C[;T[

WMFT grip

E/W/F

W/F

Sample Size at FU (Length of FU) Long-Term Effects

Assistive technologies for poststroke arm rehabilitation

www.archives-pmr.org

Table 1

C[; T[

CZ28 TZ29 (6mo) C4; T[ CZ28 TZ29 (6mo) C[; T[ CZ28 TZ29 (6mo) C[; T[


.10 (
1.61 to 1.41) C[; T[

.15 (
.71 to 1.00)

C[; T[

.68 (.12 to 1.23)

C4; T[


.30 (
.93 to .33)

C[; TY

(continued on next page)

971

972

Table 1 (continued )

AT

Comparator

Total FM

BF (LF)

Routine therapy þ experimental treatment vs Routine therapy þ placebo treatment55 Experimental treatment vs Routine therapy Time equivalence37 Experimental treatment vs Routine treatment Time equivalence39 Experimental treatment vs Routine treatment Time equivalence41 Experimental treatment vs Control treatment Time equivalence54 Experimental treatment vs Control treatment Time equivalence54 Routine treatment þ experimental treatment vs Routine treatment33 Routine treatment þ experimental treatment vs Routine treatment33 High-intensity experimental treatment vs Control treatment35 Routine treatment þ Experimental treatment vs Routine treatment32 Routine therapy þ control treatment vs Routine therapy þ experimental treatment Time equivalence21 Experimental treatment vs Placebo treatment followed by functional training

BF(HF) CIMT

Robotics (LI) Robotics (HI) ES (LI)

ES (HI)

ES

Joints Treated

Sample Size (% of Total Sample Size)

W/F

Outcome Measure

Sample Size at FU (Length of FU) Long-Term Effects

Immediate Effects

www.archives-pmr.org

CZ12 TZ9 (1.66%) W/F CZ8 TZ11 (1.50%) Sh/E/W/F CZ7 TZ9 (1.26%)

Total FM

.78 (
1.45 to 3.01) C[; T[

Total FM

.03 (
.89 to .95)

Total FM*

.65 (
2.44 to 3.74) C[; T[

Sh/E/W/F CZ16 TZ16 (2.53%)

Total FM*

.34 (
.82 to 1.51)

C[; T[

Sh/E/W/F CZ12 TZ13 (1.97%)

Total FM*

.27 (
.82 to 1.36)

C[; T[

W

CZ6 TZ6 (.97%)

Total FM*

.24 (
1.27 to 1.75) C[; T[

W

CZ6 TZ6 (.95%)

Total FM*


.04 (
1.19 to 1.10) C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Total FM*

.75 (
1.87 to 3.36) C[; T[

CZ22 TZ22 (2mo)

.80 (
2.19 to 3.78) C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Total FM*

.58 (
1.63 to 2.80) C[; T[

CZ22 TZ22 (2mo)

.75 (
2.42 to 3.91) C[; T[

W/F

CZ10 TZ9 (1.50%)

Total FM

1.37 (
.11 to 2.64)

C[; T[

CZ10 TZ9 (6mo)

.45 (
.71 to 2.3)

Sh/E/W/F CZ18 TZ19 (2.13%)

Total FM*

2.43 (
.74 to 5.59)

C[; T[

CZ18 TZ19 (6m)

Sh/W/F

CZ10 TZ10 (1.57%)

Total FM*

1.06 (
.89 to 3.00)

C[; T[

CZ10 TZ10

W/F

CZ12 TZ11 (1.81%)

Total FM*

.01 (
.80 to .81)

C[; T[

CZ11 TZ9 (6mo)

C[; T[

CZ12 .68 (
1.59 to 2.96) C[; T[ TZ9 (1.5mo) CZ8
.30 (
1.99 to 1.39) C[; T[ TZ11 (1.5mo) CZ7 .74 (
2.73 to 4.21) C[; T[ TZ9 (3mo)

C[; T[

1.04 (
1.18 to 3.25) C[; T[

.15 (
.96 to 1.27)

C[; T[

(continued on next page)

S.E. Farmer et al

Impairment Measures

Impairment Measures

AT

Joints Treated

Comparator

Sample Size (% of Total Sample Size)

Outcome Measure

Immediate Effects

Sample Size at FU (Length of FU) Long-Term Effects

31

Distal FM

Time equivalence Experimental treatment vs Placebo treatment Time equivalence26 Stochastic Experimental treatment þ resonance exercises vs Placebo treatment þ exercises Time equivalence59 Robotics Experimental treatment vs Control treatment Time equivalence51 Experimental treatment vs Control treatment Time equivalence48 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence53 ES (LI) Routine treatment þ experimental treatment vs Routine treatment33 ES (HI) Routine treatment þ experimental treatment vs Routine treatment33 ES Routine treatment þ experimental treatment vs Routine treatment32 Routine treatment þ experimental treatment vs Routine treatment32 Experimental treatment vs Placebo treatment followed by functional training

W/F

CZ14 TZ14 (2.21%)

Total FM*

1.05 (
.83 to 2.93)

C[; T[

Sh/E

CZ15 TZ15 (2.37%)

Total FM*

.04 (
.69 to .76)

C[; T[

CZ15 TZ15 (1mo)

.05 (
.68 to .77)

C[; T[

Sh/E

CZ10 TZ11 (1.66%)

Distal FM*

.05 (
.81 to .91)

C[; T[

CZ10 TZ11 (3mo)


.03 (
.89 to .83)

C[; T[

Sh/E

CZ14 TZ13 (2.13%)

Distal FM*


.06 (
.81 to .70)

C[; T[

CZ14 TZ13 (6mo)


.38 (
1.15 to .40)

C[; T[

Sh/E

CZ18 TZ17 (2.76%)

Distal FM*

.08 (
.59 to .74)

C[; T[

CZ18 TZ17 (8mo)

.06 (
.61 to .72)

C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Distal FM*

.54 (
.60 to 1.69)

C[; T[

CZ22 TZ22 (2mo)

.82 (
.97 to 2.61)

C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Distal FM*

.41 (
.53 to 1.35)

C[; T[

CZ22 TZ22 (2mo)

.61 (
.89 to 2.11)

C[; T[

Sh/E/W/F CZ18 TZ19 (2.92%)

Distal FM (hand)*

.03 (
.62 to .67)

C[; T[

CZ18 TZ19 (6mo)

.17 (
.49 to .84)

C[; T[

Sh/E/W/F CZ18 TZ19 (2.92%)

Distal FM (wrist)*

.33 (
.33 to .99)

C[; T[

CZ18 TZ19 (6mo)

.56 (
.15 to 1.27)

C[; T[

W/F

Distal FM (hand)*
.37 (
1.25 to .52)

C[; TY

CZ11 TZ9 (6mo)

CZ13 TZ11 (1.89%)


.08 (
.96 to .81)

Assistive technologies for poststroke arm rehabilitation

www.archives-pmr.org

Table 1 (continued )

C[; T[

(continued on next page)

973

974

Table 1 (continued ) Impairment Measures

AT

Joints Treated

Comparator

Sample Size (% of Total Sample Size)

Outcome Measure

Sample Size at FU (Length of FU) Long-Term Effects

Immediate Effects

31

Proximal FM Robotics

ES (LI)

ES (HI)

Time equivalence Experimental treatment vs Placebo treatment followed by functional training Time equivalence31 Routine treatment þ experimental treatment vs Routine treatment32 Experimental treatment vs Control treatment Time equivalence51 Experimental treatment vs Control treatment Time equivalence48 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence53 Routine treatment þ experimental treatment vs Routine treatment33 Routine treatment þ experimental treatment vs Routine treatment33

Distal FM (wrist)*

.21 (
.61 to 1.03)

C[; T4 CZ11 TZ9 (6mo)

.27 (
.65 to 1.20)

CY; T[

Sh/E/W/F CZ18 TZ19 (2.92%)

Proximal FM*

.58 (
.25 to 1.41)

C[; T[

CZ18 TZ19 (6mo)

.65 (
.15 to 1.27)

C[; T[

Sh/E

CZ10 TZ11 (1.66%)

Proximal FM*


.12 (
1.00 to .76)

C[; T[

CZ10 TZ11 (3mo)

Sh/E

CZ14 TZ13 (2.13%)

Proximal FM*

.88 (0 to 1.76)

C[; T[

CZ14 TZ13 (6mo)

.24 (
.54 to 1.03)

Sh/E

CZ18 TZ17 (2.76%)

Proximal FM*

.68 (
.73 to 2.09)

C[; T[

CZ18 TZ17 (8mo)

.87 (
1.45 to 3.20) C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Proximal FM*

.79 (
.83 to 2.41)

C[; T[

CZ22 TZ22 (2mo)

1.07 (
1.08 to 3.22) C[; T[

Sh/W/F

CZ22 TZ22 (3.47%)

Proximal FM*

.68 (
.80 to 2.17)

C[; T[

CZ22 TZ22 (2mo)

W/F

CZ13 TZ11 (1.89%)


.07 (
.93 to .80)

C[; T[

C[; T[

.81 (
1.00 to 2.62) C[; T[

Abbreviations: AROM, active range of motion; BF, biofeedback; C, control; E, elbow; ES, electrical stimulation; F, fingers; FU, follow-up; HF, high functioning; HI, high intensity; LF, low functioning; LI, low intensity; MRC, Medical Research Council; NR, not reported; ROM, range of motion; Sh, shoulder; T, experimental treatment; W, wrist; [, increase in mean score; 4, no change in mean score; Y, decrease in mean score. * Scale-based measurement.

S.E. Farmer et al

www.archives-pmr.org

Measures of activity Joints Treated

AT

Comparator

BF to LF BF to HF

Routine therapy þ experimental treatment vs Routine therapy þ placebo treatment55

W/F

CIMT

Usual care vs Experimental treatment36

Sh/E/W/F

W/F

Sh/E/W/F

Robotics

ES

ES LI ES HI SR

Routine therapy þ usual aftercare vs Experimental therapy þ usual aftercare38 Routine therapy vs Experimental therapy Time equivalence40 Experimental treatment vs Routine therapy41 Experimental treatment vs No treatment41 Routine therapy vs Experimental therapy Time equivalence44 Experimental treatment vs Control treatment Time equivalence51 Experimental treatment vs Placebo treatment Time equivalence22 Routine therapy and experimental therapy vs Routine therapy þ increased contact time Time equivalence29 Routine treatment þ experimental treatment vs Routine treatment33 Routine treatment þ experimental treatment vs Routine treatment33 Experimental treatment þ exercises vs Placebo treatment þ exercises Time equivalence59

Sh/E/W/F

Sh/E/W/F Sh/E/W/F Sh/E/W/F Sh/E/W/F Sh/E

Sample Size (% of Total Sample Size) CZ12 TZ9 (1.66%) CZ8 TZ11 (1.50%) CZ116 TZ106 (17.52%) CZ116 TZ106 (17.52%) CZ12 TZ18 (2.37%) CZ20 TZ23 CZ12 TZ13 CZ10 TZ13 CZ9 TZ11 CZ10 TZ11

Outcome Measure

Sample Size at FU (Length of FU)

Immediate Effects

Long-Term Effects

ARAT*

.93 (
.39 to 2.25)

C4; T[

ARAT*

.08 (
1.01 to 2.06)

C[; T[

WMFT FAS* WMFT PT

NR

NR

NR

NR

WMFT FAS

.18 (
.55 to .92)

C[; T[

ARAT*

.79 (
1.51 to 3.09)

C[; T[

ARAT*

.86 (
1.44 to 3.16)

C[; T[

NR

ARAT*

.79 (
2.19 to 3.76)

C[; T[

NR

ARAT*

.74 (
2.21 to 2.25)

C[; T[

NR

ARAT*

.15 (
.75 to 1.05)

C[; T[

NR

(3.39%)

CZ12 TZ9 (1.5mo) CZ8 TZ11 (1.5mo) CZ116 TZ106 (12mo) CZ116 TZ106 (12mo) CZ12 TZ18 (6mo) CZ19 TZ18 (3mo)

.77 (
.97 to 2.50)

C4; T[


.36 (
2.98 to 2.25)

C[; T[

.08 (
.06 to .27)

C[; T[

.11 (
2.79 to 3.01)

C[; T[

0 (
.73 to .73)

C[; T[

.48 (
1.16 to 2.12)

C[; T[

(1.97%) (1.82%) (1.58%) (1.66%)
.04 (
.40 to .32)

Sh

CZ86 TZ90 (13.89%)

ARAT*

.04 (
.35 to .44)

C[; T[

CZ75 TZ80 (3mo)

W/F

CZ28 TZ27 (4.34%)

ARAT*

.47 (
1.08 to 2.02)

C[; T[

CZ23 TZ25 (6mo)

.27 (
.78 to 1.32)

C[; T4

Sh/W/F

CZ22 TZ22 CZ22 TZ22 CZ15 TZ15 CZ15 TZ15

ARAT*

.50 (
.69 to 1.68)

C[; T[

.86 (
1.92 to 3.64)

C[; T[

ARAT*

.46 (
.75 to 1.66)

C[; T[

CZ22 TZ22 (2mo) CZ22 TZ22 (2mo)

.81 (
1.74 to 3.37)

C[; T[

ARAT*

.15 (
.66 to .96)

C[; T[

NR

WMFT


.15 (
6.85 to 6.54)

C[; T[

NR

Sh/W/F E/W/F E/W/F

Assistive technologies for poststroke arm rehabilitation

www.archives-pmr.org

Table 2

(3.47%) (3.47%)

C[; T[

(2.37%) (2.37%)

975

Abbreviations: BF, biofeedback; C, control; E, elbow; ES, electrical stimulation; F, fingers; FAS, Functional Activity Scale; FU, follow-up; HF, high functioning; LF, low functioning; NR, not reported; PT, performance time; Sh, shoulder; SR, stochastic resonance; T, experimental treatment; W, wrist; [, increase in mean score; 4, no change in mean score; Y, decrease in mean score. * Scale-based measurement.

Measures of participation

Participation Measures AT MAL (AoU)

CIMT

Comparator

Usual care vs Experimental treatment36 Routine therapy þ usual aftercare vs Experimental therapy þ usual aftercare38 Experimental treatment vs Routine treatment Time equivalence39 mCIMT Experimental treatment vs Routine treatment Time equivalence42 Experimental treatment vs Routine treatment Time equivalence43 Robotics Experimental treatment vs Control treatment Time equivalence49 Robotics (LI) Experimental treatment vs Control treatment Time equivalence54 Robotics (HI) Experimental treatment vs Control treatment Time equivalence54 ES (HF) Experimental treatment vs Control treatment Time equivalence23 ES (LF) Experimental treatment vs Control treatment Time equivalence23 SR Experimental treatment þ exercises vs Placebo treatment þ exercises Time equivalence59 CIMT Usual care vs Experimental treatment36 Routine therapy þ usual aftercare vs Experimental therapyþ usual aftercare38

Joint Treated

Sample Size (% of Total Sample Size)

Outcome Measure

Sh/E/W/F CZ116 MAL (AoU) TZ106 (17.52%) Sh/E/W/F CZ12 MAL (AoU) TZ18 (2.37%)

Sample Size at FU (Length of FU) Long-Term Effects

Immediate Effects

.24 (
.50 to .97)

C[; T[

CZ116 TZ106 (12mo) CZ12 TZ18 (6mo)

.15 (
.02 to .68)

C[; T[


.10 (
.83 to .63)

C[; T[

CZ14 TZ14 (6mo)

.18 (
.57 to .92)

C[; T[

C[; T[

Sh/E/W/F CZ16 TZ16 (2.52%)

MAL (AoU)

.12 (
.58 to .81)

C[; T[

Sh/E/W/F CZ15 TZ17 (2.52%)

MAL (AoU)

1.12 (.37 to 1.87)

C[; T[

Sh/E/W/F CZ15 TZ15 (2.37%)

MAL (AoU)

.72 (
.04 to 1.48)

C[; T[

Sh

CZ16 TZ15 (2.45%)

MAL (AoU)

.28 (
.42 to .99)

C[; T[

W

CZ6 TZ6 (.95%)

MAL (AoU)

.14 (
.99 to 1.27)

C[; T[

W

CZ6 TZ6 (.95%)

MAL (AoU)

.29 (
.85 to 1.42)

C[; T[

F

CZ6 TZ6 (.95%)

MAL (AoU)

CZ6 TZ6 (6mo)

2.24 (
3.24 to 7.72)

F

CZ8 TZ8 (1.26%)

MAL (AoU)

CZ8 TZ8 (6mo)

2.52 (
8.09 to 13.13) C[; T[

E/W/F

CZ15 TZ15 (2.37%)

MAL (AoU)

Sh/E/W/F CZ116 MAL (QoM) TZ106 (17.52%) Sh/E/W/F CZ12 MAL (QoM) TZ18 (2.37%)

.16 (
.55 to .88)

.07 (
.66 to .80)

C[; T[

C[; T[

CZ116 TZ106 (12mo) CZ12 TZ18 (6mo)

.16 (.06 to .72)

C[; T[


.02 (
.75 to .71)

C[; T[

(continued on next page)

S.E. Farmer et al

www.archives-pmr.org

MAL(QoM)

976

Table 3

Participation Measures AT

FIM

Comparator

Experimental treatment vs Routine treatment Time equivalence39 mCIMT Experimental treatment vs Routine treatment Time equivalence42 Experimental treatment vs Routine treatment Time equivalence43 Robotics Experimental treatment vs Control treatment Time equivalence49 Robotics (LI) Experimental treatment vs Control treatment Time equivalence54 Robotics (HI) Experimental treatment vs Control treatment Time equivalence54 ES (HF) Experimental treatment vs Control treatment Time equivalence23 ES (LF) Experimental treatment vs Control treatment Time equivalence23 CIMT Routine therapy þ usual aftercare vs Experimental therapy þ usual aftercare38 Experimental treatment vs Routine treatment Time equivalence39 mCIMT Experimental treatment vs Routine treatment Time equivalence42 Experimental treatment vs Routine treatment Time equivalence43

Joint Treated

Sample Size (% of Total Sample Size)

Outcome Measure

Sample Size at FU (Length of FU) Long-Term Effects

Immediate Effects

Sh/E/W/F CZ16 TZ16 (2.52%)

MAL (QoM)

.30 (
.40 to 1.00)

Sh/E/W/F CZ15 TZ17 (2.52%)

MAL (QoM)

Sh/E/W/F CZ15 TZ15 (2.37%)

MAL (QoM)

.57 (
.18 to 1.31)

C[; T[

Sh

CZ16 TZ15 (2.45%)

MAL (QoM)

.00 (
.70 to .80)

C[; T[

W

CZ6 TZ6 (.95%)

MAL (QoM)

.11 (
1.03 to 1.24)

C[; T[

W

CZ6 TZ6 (.95%)

MAL (QoM)

.45 (
.69 to 1.60)

C[; T[

F

CZ6 TZ6 (.95%)

MAL (QoM)

CZ6 TZ6 (6mo)

2.48 (
7.03 to 11.99) C[; T[

F

CZ8 TZ8 (1.26%)

MAL (QoM)

CZ8 TZ8 (6mo)

2.09 (
1.76 to 5.94)

C[; T[

CZ12 TZ18 (6mo)

.18 (
.63 to .99)

C[; T[

1.24 (.47 to 2.01)

C[; T[

C[; T[

Sh/E/W/F CZ12 TZ18 (2.37%)

FIM

.05 (
.69 to .79)

C[; T[

Sh/E/W/F CZ16 TZ16 (2.52%)

FIM

.34 (
.78 to 1.47)

C[; T[

Sh/E/W/F CZ15 TZ17 (2.52%)

FIM

.34 (
1.15 to 1.84)

C[; T[

Sh/E/W/F CZ15 TZ15 (2.37%)

FIM

.27 (
1.20 to 1.74)

C[; T[

CZ14 TZ14 (6mo)

.22 (
.52 to .96)

C[; T[

Assistive technologies for poststroke arm rehabilitation

www.archives-pmr.org

Table 3 (continued )

(continued on next page)

977

978

Table 3 (continued ) Participation Measures AT CIMT (HI)

CIMT (LI)

Robotics

ES

Barthel

Robotics

ES

Experimental treatment vs Control treatment Time equivalence45 Experimental treatment vs Control treatment Time equivalence45 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence53 Experimental treatment vs Control treatment Time equivalence48 Routine therapy þ control treatment vs Routine therapy þ experimental treatment Time equivalence21 Routine therapy þ experimental therapy vs Routine therapy þ extra therapy Time equivalence25 Experimental treatment vs Placebo treatment Time equivalence26 Experimental treatment vs Control treatment Time equivalence48 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence29 Routine treatment þ experimental treatment vs Routine treatment þ control treatment Time equivalence29

Sample Size (% of Total Sample Size)

Sample Size at FU (Length of FU) Long-Term Effects

Outcome Measure

Immediate Effects

Sh/E/W/F CZ17 TZ16 (2.60%)

FIM UE


3.32 (
4.52 to
2.11) C[; T[

Sh/E/W/F CZ17 TZ19 (2.84%)

FIM UE

.11 (
.55 to .76)

Sh/E

FIM

Sh/E

Sh/E

CZ18 TZ17 (2.76%) CZ18 TZ17 (2.76%)

FIM motor

CZ17 TZ16 (3mo)


3.44 (
4.68 to 2.21)

C[; T[

C[; T[

CZ17 TZ19 (3mo)


.86 (
1.56 to 5.94)

C[; T[

.78 (
1.01 to 2.57)

CY; TY

1.13 (
2.31 to 4.57)

CY; TY

1.75 (
1.83 to 5.32)

C[; T[

CZ15 TZ15 (8mo) CZ15 TZ15 (8mo)

1.27 (
3.09 to 5.63)

C[; T[

.80 (
.15 to 1.79)

C[; T[

CZ14 TZ13 (2.13%)

FIM

.41 (
.34 to 1.17)

C4; T4 CZ14 TZ13 (6mo)

Sh/E/W/F CZ10 TZ10 (1.58%)

FIM

.52 (
.78 to 1.82)

C[; T[

W

CZ5 TZ4 (.71%)

FIM

W/F

CZ14 TZ14 (2.21%)

FIM self-care

.15 (
.61 to .91)

C[; T[

Sh/E

CZ14 TZ13 (2.13%)

Barthel Index

.40 (
.42 to 1.22)

C4; T[

CZ14 TZ13 (6mo)

.50 (
.38 to 1.38)

C[; T[

W/F

CZ28 TZ27 (4.34%)

Barthel Index

.30 (
.26 to .86)

CY; TY

CZ23 TZ25 (6mo)

.67 (
.15 to 1.50)

CY; TY

W/F

CZ28 TZ27 (4.34%)

Rankin

0 (
.53 to .53)

CY; TY

CZ23 TZ25 (6mo)

0 (
.57 to .57)

CY; TY

1.78 (0 to 3.56)

C[; T[

(continued on next page)

S.E. Farmer et al

www.archives-pmr.org

ES

Comparator

Joint Treated

CY; T[ .29 (
.82 to 1.40)

C[; TY
.03 (
.77 to .71) SR

CIMT SIS

Comparator

Usual care vs Experimental Sh/E/W/F CZ116 SIS (hand) treatment36 TZ106 (17.52%) Sh/E/W/F CZ12 SIS (hand) Routine therapy þ usual TZ18 (2.37%) aftercare vs Experimental therapy þ usual aftercare38 CZ15 SIS Experimental treatment þ E/W/F TZ15 (2.37%) exercises vs Placebo treatment þ exercises Time equivalence59

Immediate Effects Outcome Measure Sample Size (% of Total Sample Size) Joint Treated Participation Measures AT

Table 3 (continued )

www.archives-pmr.org

Abbreviations: AoU, amount of use; C, control; E, elbow; ES, electrical stimulation; F, fingers; FU, follow-up; HF, high functioning; HI, high intensity; LF, low functioning; LI, low intensity; mCIMT, modified CIMT; QoM, quality of movement; Sh, shoulder; SR, stochastic resonance; T, experimental treatment; UE, upper extremity; W, wrist; [, increase in mean score; 4, no change in mean score; Y, decrease in mean score.

C[; T[ .12 (
.60 to 14.66)

CZ116 TZ106 (12mo) CZ12 TZ18 (6mo)

Sample Size at FU (Length of FU) Long-Term Effects

Assistive technologies for poststroke arm rehabilitation

979 training was small (.19; 95% CI,
.88 to 1.27). There was some retention of these benefits at 6 months’ follow-up that were greater than those of the control group. Stein et al59 reported no significant differences in sensory awareness after stochastic resonance, with the effect size for proprioception of .08 (95% CI,
.51 to .67); for vibration, 0 (95% CI,
.72 to .72); for pinprick,
.18 (95% CI,
.56 to 1.28); and for SemmesWeinstein monofilament testing,
.46 (95% CI,
1.22 to .30).

Activity Functional activity was measured by most researchers using ARAT, or WMFT performance time and Functional Activity Scale, with results summarized in table 2. Nine studies reported ARAT scores. Effect sizes estimated immediately at the end of the treatment period were variable and are summarized below:  Biofeedback55: .08 (95% CI,
1.01 to 2.06) to .93 (95% CI,
.39 to 2.25)  CIMT36,38,40,43,44: .18 (95% CI,
.55 to .92) to .86 (95% CI,
1.44 to 3.16)  Robotics51: .15 (95% CI,
.75 to 1.05)  Electrical stimulation22,29,33: .04 (95% CI,
.35 to .44) to 0.5 (95% CI,
.69 to 1.68)  Stochastic resonance59: .15 (95% CI,
.66 to .96) Little benefit is derived posttreatment from electrical stimulation to the shoulder only, from robotics to the shoulder and elbow,51 or from stochastic resonance59 or from biofeedback for the higher-functioning subject.55 Some benefit is gained posttreatment from CIMT,40,41,44 electrical stimulation to the arm (shoulder and/or wrist and fingers),29,33 and from biofeedback to low-functioning subjects.55 Further improvements are shown after electrical stimulation to the shoulder and/or wrist and fingers at follow-up29,33; however, in 1 study,22 there appears to be a negative effect of electrical stimulation to the shoulder at follow-up when shoulder muscles were stimulated exclusively. Four studies used the WMFT: CIMT,36,38 electrical stimulation,33 and stochastic resonance.59 Effect sizes ranged from
.15 to .18 (see table 2). Wolf et al36 reported posttreatment results without SDs, so it has not been possible to calculate the effect size posttreatment as was done for the follow-up results by using the reported CIs to estimate pooled SD. Stochastic resonance59 did not improve WMFT effect sizes. Both low- and high-dose electrical stimulation gave improvements after treatment that showed further improvement at follow-up.33

Participation Results from measures of participation are shown in table 3. Studies variably report using a range of outcomes: Barthel Index,29,48 with effect sizes ranging from 0.3 (95% CI,
.26 to .86) to 0.4 (95% CI,
.42 to 1.22); Rankin score,29 with an effect size of 0.0 (95% CI,
.53 to .53); and FIM21,25,29,38,39,42,43,45,48,53 giving the overall level of disability. Dromerick et al45 compared low-intensity and high-intensity CIMT with control subjects, showing that a high intensity of CIMT in acute subjects had a negative effect after treatment (effect size,
3.32; 95% CI,
4.52 to 2.11), with further deterioration at follow-up (effect size, 3.44;

980 95% CI,
4.68 to 2.21). This contrasts with other regimens and modalities that show some improvement, with effect sizes ranging from .05 (95% CI,
.69 to .79) to 1.78 (95% CI, 0e3.56). Studies report MAL Amount of Use23,36,38,39,42,43,49,54,59 indicating some increase in the reported amount of use, with effect sizes from .12 (95% CI,
.58 to .81) to 1.12 (95% CI, .37e 1.97). Although Popovic et al23 showed higher levels of use at follow-up, this result should be viewed with caution because of wide CIs and a small sample size. MAL Quality of Use is reported,23,36,38,39,43,54 with effect sizes from .07 (95% CI,
.66 to .80) to 1.24 (95% CI, .47e2.01). The immediate effect size for SIS was
.03 (95% CI,
.77 to .71).59 Long term SIS (hand components) was reported with effect sizes of .29 (95% CI,
.82 to 1.40)38 and .12 (95% CI,
.60 to 14.66).36 However, it would appear that in these studies, neither CIMT nor stochastic resonance has greatly changed the impact of the stroke on the subjects studied. Some studies could not be easily summarized in tables, and they are described below. The single study57 of brain stimulation compared repetitive transcranial stimulation with sham stimulation. This study had 10 experimental subjects and 5 control subjects. The effect size for the Jebsen-Taylor Test immediately after treatment was .46 (95% CI,
12.19 to 13.10), and at 2 weeks after treatment the effect size (comparing pretreatment with follow-up) was 0.2 (95% CI,
5.28 to 5.67). Saposnik et al58 examined the feasibility and safety of the Nintendo Wii compared with recreational therapy. The effect size calculated immediately after treatment for WMFT performance time was .16 (95% CI,
1.31 to 1.63). Hemmen and Seelen20 compared 2 treatments that use electrical stimulation: conventional repetitive electrical stimulation and electromyography-triggered stimulation with audiovisual feedback. Although neither regimen was more effective as shown by effect sizes immediately after treatment (0; 95% CI,
.84 to .84) and at follow-up (.08; 95% CI,
.93 to .77), both regimens were reported to result in an improvement in ARAT scores of 19.414.5, indicating that both regimens are beneficial. De Kroon et al24 also showed improvements in both a cyclical stimulation group and an electromyography-triggered group, with ARAT scores increasing by 2.32.9 and by 4.26.7, respectively. Effect size was in favor of the latter group immediately after treatment (.35; 95% CI,
.67 to 1.39) but regressed at follow-up to be in favor of the cyclical stimulation group (.42; 95% CI,
.54 to 1.37). In the former study,20 subjects were within 6 weeks of stroke, whereas in the latter study,24 subjects were more than a year poststroke. Hesse et al30 compared the effect of computerized arm training with electrical stimulation, with the 22 arm-training subjects performing 12,000 wrist and forearm movement cycles and the 22 NMES subjects just 1800 to 2400 wrist movements in total. The primary outcome measure, FM score, gave an effect size of 1.15 (95% CI, e1.75 to 4.05) posttreatment in favor of the more intensive arm training. The proximal component of FM gave an effect size of .99 (95% CI,
.47 to 2.45) and the distal component an effect size of 2.38 (95% CI, .57e4.19), indicating that where the forearm was the focus of treatment that was most beneficial effect. Daly et al52 compared 6 robotic trainer subjects with 6 electrical stimulation subjects, with both groups also undertaking motor learning. This gave a modest benefit in favor of the robotic

S.E. Farmer et al trainer, with an effect size on the FM score of .23 (95% CI,
1.05 to 1.52).

Time poststroke Figure 2 shows that the greatest effects are achieved by treatment in the acute phase. There are 2 exceptionsdelectrical stimulation to the shoulder22 only and CIMT45 (see table 1)dwhere early treatment appears deleterious in a subsample of stroke patients. When treatment starts at 6 weeks poststroke,21,26,29,32,33,37,53 the range of effect sizes is wide, from
.14 to 2.43; this contrasts with effect sizes in chronic stroke27,30,31,39,41,48,49,51,52,56 that range from
.39 to .88.

Discussion ATs can contribute to improving outcomes after stroke, and the risk of harm is rare. The additional benefit of using ATs as an adjunct to routine therapy is often relatively small in terms of improved function, and this can only suggest that the benefit from rehabilitation provided by an AT is similar to that being obtained with routinely used rehabilitation therapies. (Note: If P values suggest that for a given set of data, there is only a 5% probability that similarities exist, then one can infer a statistically significant difference between the data sets. The reason for a lack of a statistical difference in most of the studies reviewed was an improvement that occurred in both the treatment and control groups. In these circumstances, the most likely inference is that the 2 arms have similar effectiveness. However, since the effect size estimates were predominantly positive, one can infer that the additional benefit was relatively small.) There are beneficial changes at the impairment level, and this is consistent with effects associated with task-specific training. However, the improvements at the level of the impairment rarely translate into increased activity or give benefit at the level of participation. It is uncertain whether minimally clinically importance differences (MCIDs) have since been achieved for many outcomes. The following discussion explores the possible reasons for such findings.

Insufficient difference between control and experimental treatments or subjects The relatively small benefits seen may be due to the lack of difference between experimental and control treatments. This can be due to the design of studies that use time equivalence between the groups studied. Differences in the treatments may be further reduced when experimental treatment is combined with routine treatment in the experimental arm and this is compared with routine treatment of extended duration to give time equivalence. There can be a lack of clarity related to the selection of treatment parameters when routine therapy is used as the control. This is usually described in rather general terms (eg, “conventional treatment that targeted proximal upper-limb function that was based on neurodevelopmental therapy”),48(p954) thus each subject is likely to receive an individualized treatment. Although the shaping associated with CIMT was described as this technique was initially developed,60 explicit detail that would permit replication is less apparent for more recent studies.42,43,45 Routine treatment may include practicing functional tasks and therefore may be similar in content to CIMT. www.archives-pmr.org

Assistive technologies for poststroke arm rehabilitation

Fig 2

Effect size (ES) of functional measures with time poststroke.

A further factor reducing the difference between subjects can be competitive bias, which was reported by Rodgers et al,61 and should be considered in rehabilitation studies where single blinding is the norm because of the difficulties of blinding subjects to equipment-based treatments. The heterogeneity of the subjects within individual studies may also affect outcomes. In some studies, the subjects time post has a wide range, eg, Time from stroke 4.35.3 years for experimental subjects and 2.81.99 years for control subjects.31 Although there may be little statistical difference between these groups, some subjects are still at a relatively early point in their recovery and may benefit from the intervention, while others are in the chronic period where the confounding factors, secondary complications, and comorbidities reduce any beneficial effects. Thus, the improvement gained when acute and chronic patients are mixed is likely to be less than if subjects are in the acute phase only, therefore leading to an underestimation of treatment effects, whereas in studies where all subjects are in the chronic phase, outcomes are generally lower than those where all subjects are in the acute phase (see fig 2).

Population, intervention, comparator, and outcome factors The original brief for the ATRAS was to inform a clinical trial, developing the integration of ATs into the care pathway. Although there has been research into some treatment parameters, many aspects that would define the population, intervention, comparators, and outcomes for such a trial are lacking. The consequences of loss of motor control need to be addressed on a number of levels. Key to this would be an assessment that identifies residual deficits that required remediation. Most studies, however, do not identify specific deficits in their inclusion criteria, thereby reducing the chance of matching the deficit with appropriate treatment. Furthermore, each deficit may benefit from a different treatment strategy so that in any given www.archives-pmr.org

981

research, only one of the many problems is effectively treated. The remaining untreated deficits diminish any beneficial effects; this alone can explain the small effect sizes reported in the literature. At best, studies hint at the way forward. Subjects with low function appear to have a better response to biofeedback than highfunctioning subjects55; conversely, high-functioning subjects have a greater improvement from NMES.23 Figure 2 demonstrates the tendency to greater improvement with early intervention. Two studies, however, indicate detrimental effects with early treatment. First, Dromerick45 considered that their higher dose of CIMT in acute stroke was detrimental, suggesting that overtraining or the “blocked schedule” interfered with motor relearning when compared with “distributed practice.” Second, Church et al22 found some risk of harm with purely shoulder treatment in the severely disabled stroke patient, since overall functional recovery is impeded. There is some evidence to guide the duration of treatment sessions for electrical stimulation and CIMT. Treatment sessions of 30 minutes of electrical stimulation compared with 60 minutes of stimulation showed little additional benefit for increased treatment.33 Electrical stimulation triggered by voluntary movement appeared more beneficial than other NMES modes.62 In CIMT, a shorter duration of treatment sessions (2h shaping with up to 6h of restraint of the nonaffected limb)42,43 may be more beneficial than the originally proposed treatment of longer duration38,39 (see table 2). The higher intensity of treatment with robotics54 indicates that sufficient practice of an action is required; however, most robotic studies use 1 hour of treatment daily over a 3- to 6-week period. There is evidence that greater effect sizes are associated with movements that stimulate functionally useful movements, as seen in studies that stimulate more than 1 joint.28,32,33 Lin et al32 showed the greatest effect size posttreatment (2.43; 95% CI,
.74 to 5.59) with a treatment regimen stimulating shoulder abduction and wrist extension. In the more severely disabled patient, there is evidence that therapeutic stimulation protocols (ie, stimulation that is cyclical) have

982

S.E. Farmer et al

some benefit at the level of the impairment and may facilitate recovery of function if applied distally.29 Sensory deficits can contribute to loss of function, but few studies (2 studies on biofeedback,55,56 1 feasibility study using virtual reality58) were identified that specifically considered the effectiveness of modalities that substitute for these losses. Some studies show further gains at follow-up, perhaps allowing the subject to use the limb in activities of daily living; however, the precise reasons were not distilled. Other studies that report follow-up outcomes show a tendency for subjects to regress. This is likely to be a consequence of designs that cease treatment after a predetermined time interval rather than when the subject has achieved a predetermined goal, or if progress has ceased. It would seem that small treatment effects may be due to some of these treatment inadequacies.

result of this limitation it is possible, particularly in studies with small samples, that we have overestimated the effect size. The MCID varies with the measure being used and the sample within which the MCID is estimated. There are no clear guidelines in the literature on how the MCID can be individualized to the samples within the selected articles. We have discussed, where appropriate from first scientific principles, the relationship between effect size and clinically important difference, hence we have not specifically and systematically discussed the relationship between effect size and MCID. Although we have considered ATs that can be used in rehabilitation, we have not included studies or devices that may provide ongoing compensation for residual deficits.

Theoretical considerations

Conclusions

AT can be used to achieve a number of treatment objectives. Muscle atrophy can be limited or muscle strength improved when the stimulated muscle works as per principles of overload. Cyclic stimulation may be used to mobilize joints and intramuscular connective tissue, thereby reducing the rate of contracture development. Motor relearning could be prompted by electrical stimulation used to assist movement at a number of joints simulating a functional movement (eg, reach to grasp). Similarly, robots can be used to provide weight support to weak muscles or resistance to stronger muscles so that repeated actions strengthen muscle. Joint mobility may be increased by guided motion that may also contribute to motor learning. Such actions require knowledge of performance, possibly provided by biofeedback to reinforce learning. The selection of treatment parameters to achieve such objectives, the length of treatment required to maximize recovery and the complexities of these treatments are rarely addressed in the literature. Current AT approaches seem to be more effective in improving impairments, and the effects are not carrying over into activity and participation. This suggests that new strategies are required to bring changes at a functional level. Current treatment tends to focus on grasp and release or simultaneous bilateral training. Although arms and hands can be used singly, they often work in collaboration (as in throwing a large ball), but there are many activities that require counteractive collaboration between the hands (eg, cutting food with a knife and holding the food with the other hand). Therapists may attempt to reeducate the counteractive utilization of the hands so essential to many functional activities in dressing, personal hygiene, and eating, but few if any AT systems support practice of such activities. As more advanced multichannel programmable stimulators are developed, it may be possible to facilitate more complete functional movements during rehabilitation. There have been studies on robotic systems that provide actuation and guidance for multiple degrees of freedom, including individual interphalangeal joints. After the system design and testing, treatment parameters for such systems should be explored systematically before clinical evaluation. In summary, there appears to be a need for significant phase I/ II research aimed at identifying optimal treatment methods, parameters, and durations before the phase III studies are carried out.

This review highlights the challenges in designing and executing rehabilitation research. Many studies combine the AT with “routine therapy” (exercises, stretching, and sensory stimulation) or an extended period of therapy, or both, diminishing the contrast between the experimental AT treatment and the control treatment. Routine therapy, with its individualized treatment, will thus tend to confound outcomes. AT can assist in producing benefit at the level of impairment with modest increases in effect size compared with routine therapy alone, with 2 possible exceptions; negative effects were seen with isolated shoulder stimulation in severely impaired subjects or with high-intensity CIMT in the acute phase of stroke. AT-based therapy appears to be able to deliver benefit at least equivalent to that provided by routine therapy. At present, the optimal treatment parameters are unknown, but there is evidence that treatment of the arm early after stroke appears most beneficial. Contrary to the general opinion that recovery and therefore effects of treatment plateau at 6 months, subjects showed benefit from late rehabilitation treatment in chronic stroke. There is still much work to be done that builds on basic biophysical principles to address and integrate treatments for all aspects of impairment. Informed with effective treatments for impairments, researchers would then be able to address multiple deficits and integrate a number of treatment modalities to reeducate the many complex movements that require coordination between joints within the upper limb and between both upper limbs that are used in activities of daily living. Such approaches would inform integrated rehabilitation pathways that restore arm function to the level that improves the quality of independent lifestyle for stroke survivors beyond that achieved by current methods. Perhaps the greatest benefit of technological solutions is in their potential to deliver therapy in ways that are less personnel intensive and more convenient to patients, but as yet no studies appear to have investigated such issues. There will, of course, be issues to resolve to develop and support such self-care pathways.

Supplier a. RefWorks-COS, a business unit of ProQuest LLC, 7200 Wisconsin Ave, Ste 601, Bethesda, MD 20814.

Study limitations

Keywords This study is limited by including only studies in English. In this review, we have estimated effect sizes, assuming normality of sample distribution within the selected studies. As a

Assistive technology; Rehabilitation; Review; systematic; Upper extremity www.archives-pmr.org

Assistive technologies for poststroke arm rehabilitation

Corresponding author Anand David Pandyan, PhD, Professor of Rehabilitation Technology for Health, MacKay Bldg, Keele University, Keele, Staffordshire, England ST5 5BG. E-mail address: [email protected].

Acknowledgments We thank the ATRAS team for their assessment of articles.

Appendix 1 MEDLINE Search Strategy Search Terms 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

MeSH : Stroke or stroke (text) MeSH : Upper extremity or ‘upper extremity’(Text) Upper limb* Arm* Hand* Shoulder* Elbow* Wrist* Finger* Thumb* “Virtual Reality” or VR or “Wii” Biofeedback or Bio feedback Cast* Orthot* Orthos* Splint* Dynamic devices “Contracture Correction Device” OR CCD Electric* Stimulat* FES or ‘Functional Electrical Stimulation’ ES ENS or ‘electrical neuromuscular stimulation’ IFT or “Inferential Therapy” “Cortical Stim*” “Transcranial Magnetic Stimulation” or TMS Robo* Electro* Mech* CPM or ‘continuous passive movement’ or ‘continuous passive motion’ 29. CIMT mitt 30. “Constraint induced” OR “Forced Use” Searches AZ1 B Z 2 OR 3 OR 4 OR 5 OR 6 OR 7 OR 8 OR 9 OR 10 CZA&B D Z C & 11 E Z C & 12 D Z C & (13 OR 14 OR 15 OR 16 OR 17 OR 18) E Z C & (19 OR 20 OR 21 OR 22 OR 23) F Z C & (24 OR 25) G Z C & (26 OR 27) H Z C & (28 OR 29 OR 30) I Z C & (11 OR 12 OR 13 OR 14 OR 15 OR 16 OR 17 OR 18 OR 19 OR 20 OR 21 OR 22 OR 23 OR 24 OR 25 OR 26 OR 27 OR 28 OR 29 OR 30) www.archives-pmr.org

983 Appendix 2

Quality Assessment Questionnaire

Assessment Questions 1. Were the eligibility criteria specified? 2. Was a method of randomization performed? 3. Was treatment allocation concealed? 4. Were prognostic indicators similar for groups at baseline? 5. Were the index and control interventions explicitly described? 6. Was the care provider blinded to the intervention? 7. Were co-interventions avoided or comparable? 8. Was the compliance reported in all groups? 9. Was the patient blinded to the intervention? 10. Was the outcome assessor blinded to the intervention? 11. Were the outcome measures relevant? 12. Were adverse effects described? 13. Was the withdrawal/dropout rate described? 14. Was a short-term follow-up measurement performed? 15. Was a long-term follow-up measurement performed? 16. Was timing comparable for outcome assessment in both groups? 17. Was the sample size for each group described? 18. Did the analysis include an intentionto-treat analysis? 19. Was the variability given for primary outcome measures?

Answers Percentage Negative Negative or Unclear or Unclear 108 299

23.13 64.03

417 346

89.29 74.09

172

36.83

451

96.57

357

76.45

309

66.17

421

90.15

307

65.74

53 336 272

11.35 71.95 58.24

258

55.25

405

86.72

230

49.25

161

34.48

437

93.58

167

35.76

Appendix 3 Effect Size Equations Xe
Xc Effect size[sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðNe
1Þd2e þ ðNc
1Þd2c Ne þ Nc
2 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2 u u Xe
Xc tNe þ Nc þ SE of Effect sizeZ Ne Nc 2ðNe þ Nc Þ

Equation 1

Equation 2

Abbreviations: Xe , experimental outcome; Xc , mean control outcome; Ne, number of experimental subjects; Nc, number of control subjects; de, standard error of experimental subjects; dc, standard error of control subjects.

References 1. Carr JH, Shepherd RB. A motor relearning programme for stroke. 2nd ed. London: Heinemann Medical; 1987.

984 2. Malhotra S, Cousins E, Ward A, et al. An investigation into the agreement between clinical, biomechanical and neurophysiological measures of spasticity. Clin Rehabil 2008;22:1105-15. 3. O’Dwyer NJ, Ada L, Neilson PD. Spasticity and muscle contracture following stroke. Brain 1996;119:1737-49. 4. Lieber RL. Skeletal muscle structure, function and plasticity; the physiological basis of rehabilitation. 3rd ed. Baltimore: Lippincott, Williams & Wilkins; 2010. 5. Malhotra S, Pandyan A, Rosewilliam S, Roffe C, Hermens H. Spasticity and contractures at the wrist after stroke: time course of development and their association with functional recovery of the upper limb. Clin Rehabil 2011;25:184-91. 6. Price CIM, Pandyan AD. Electrical stimulation for preventing and treating post-stroke shoulder pain: a systematic Cochrane review. Clin Rehabil 2001;15:5-19. 7. Doyle S, Bennett S, Fasoli SE, McKenna KT. Interventions for sensory impairment in the upper limb after stroke. Cochrane Database Syst Rev 2010;(6):CD006331. 8. Desrosiers J, Rochette A, Noreau L, Bourbonnais D, Bravo G, Bourget A. Long-term changes in participation after stroke. Top Stroke Rehabil 2006;13:86-96. 9. Douiri A, Rudd AG, Wolfe CD. Prevalence of poststroke cognitive impairment: South London Stroke Register 1995-2010. Stroke 2013; 44:138-45. 10. Kouwenhoven SE, Gay CL, Bakken LN, Lerdal A. Depressive symptoms in acute stroke: a cross-sectional study of their association with sociodemographics and clinical factors. Neuropsychol Rehabil 2013;23:658-77. 11. Nakayama H, Jørgensen H, Raaschou H, Olsen TS. Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil 1994;75:394. 12. Stroke rehabilitation guide: supporting London commissioners to commission quality services in 2010/11. Available at: http://www. londonhp.nhs.uk/wp-content/uploads/2011/03/Stroke-RehabilitationGuide.pdf. Accessed February 21, 2014. 13. Long C, Masciarelli V. An electrophysiologic splint for the hand. Arch Phys Med Rehabil 1963;44:499-503. 14. Juvinall RC, Smith EM, Timm RF. Analysis of design bases for upper-extremity ortheses. Ann Arbor: University of Michigan Medical School; 1961. 15. World Health Organization. International Classification of Functioning, Disability and Health. Available at: http://www.who.int/ classifications/icf/en/index.html. Accessed May 9, 2013. 16. Van Tulder MW, Koes BW, Bouter LM. Conservative treatment of acute and chronic nonspecific low back pain: a systematic review of randomized controlled trials of the most common interventions. Spine (Phila Pa 1976) 1997;22:2128-56. 17. Hedge L, Olkin I. Statistical methods for meta-analysis. Orlando: Academic Pr; 1985. 18. Morris J, Gardner M. Calculating confidence intervals for relative risks (odds ratios) and standardised ratios and rates. Br Med J (Clin Res Ed) 1988;296:1313-6. 19. Yu DT, Chae J, Walker ME, et al. Intramuscular neuromuscular electric stimulation for poststroke shoulder pain: a multicenter randomized clinical trial. Arch Phys Med Rehabil 2004;85:695-704. 20. Hemmen B, Seelen HA. Effects of movement imagery and electromyography-triggered feedback on arm hand function in stroke patients in the subacute phase. Clin Rehabil 2007;21:587-94. 21. Chan MK, Tong RK, Chung KY. Bilateral upper limb training with functional electric stimulation in patients with chronic stroke. Neurorehabil Neural Repair 2009;23:357-65. 22. Church C, Price C, Pandyan AD, Huntley S, Curless R, Rodgers H. Randomized controlled trial to evaluate the effect of surface neuromuscular electrical stimulation to the shoulder after acute stroke. Stroke 2006;37:2995-3001. 23. Popovic MB, Popovic DB, Sinkjær T, Stefanovic A, Schwirtlich L. Clinical evaluation of functional electrical therapy in acute hemiplegic subjects. J Rehabil Res Dev 2003;40:443-53.

S.E. Farmer et al 24. De Kroon J, IJzerman MJ. Electrical stimulation of the upper extremity in stroke: cyclic versus EMG-triggered stimulation. Clin Rehabil 2008;22:690-7. 25. Francisco G, Chae J, Chawla H, et al. Electromyogram-triggered neuromuscular stimulation for improving the arm function of acute stroke survivors: a randomized pilot study. Arch Phys Med Rehabil 1998;79:570-5. 26. Chae J, Bethoux F, Bohinc T, Dobos L, Davis T, Friedl A. Neuromuscular stimulation for upper extremity motor and functional recovery in acute hemiplegia. Stroke 1998;29:975-9. 27. 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: 450-60. 28. Thrasher TA, Zivanovic V, McIlroy W, Popovic MR. Rehabilitation of reaching and grasping function in severe hemiplegic patients using functional electrical stimulation therapy. Neurorehabil Neural Repair 2008;22:706-14. 29. Powell J, Pandyan AD, Granat M, Cameron M, Stott DJ. Electrical stimulation of wrist extensors in poststroke hemiplegia. Stroke 1999; 30:1384-9. 30. Hesse S, Reiter F, Konrad M, Jahnke MT. Botulinum toxin type A and short-term electrical stimulation in the treatment of upper limb flexor spasticity after stroke: a randomized, double-blind, placebocontrolled trial. Clin Rehabil 1998;12:381-8. 31. Chae J, Harley MY, Hisel TZ, et al. Intramuscular electrical stimulation for upper limb recovery in chronic hemiparesis: an exploratory randomized clinical trial. Neurorehabil Neural Repair 2009;23:569-78. 32. Lin ZL, Yan TB. Long-term effectiveness of neuromuscular electrical stimulation for promoting motor recovery of the upper extremity after stroke. J Rehabil Med 2011;43:506-10. 33. Hsu SS, Hu MH, Wang YH, Yip PK, Chiu JW, Hsieh CL. Dose-response relation between neuromuscular electrical stimulation and upperextremity function in patients with stroke. Stroke 2010;41:821-4. 34. McDonnell MN, Hillier SL, Miles TS, Thompson PD, Ridding MC. Influence of combined afferent stimulation and task-specific training following stroke: a pilot randomized controlled trial. Neurorehabil Neural Repair 2007;21:435-43. 35. Kowalczewski J, Gritsenko V, Ashworth N, Ellaway P, Prochazka A. Upper-extremity functional electric stimulation-assisted exercises on a workstation in the subacute phase of stroke recovery. Arch Phys Med Rehabil 2007;88:833-9. 36. 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. 37. Boake C, Noser EA, Ro T, et al. Constraint-induced movement therapy during early stroke rehabilitation. Neurorehabil Neural Repair 2007;21:14-24. 38. Dahl AE, Askim T, Stock R, Langørgen E, Lydersen S, Indredavik B. Short- and long-term outcome of constraint-induced movement therapy after stroke: a randomized controlled feasibility trial. Clin Rehabil 2008;22:436-47. 39. Lin KC, Wu CY, Liu JS, Chen YT, Hsu CJ. Constraint-induced therapy versus dose-matched control intervention to improve motor ability, basic/extended daily functions, and quality of life in stroke. Neurorehabil Neural Repair 2009;23:160-5. 40. Myint JM, Yuen GF, Yu TK, et al. A study of constraint-induced movement therapy in subacute stroke patients in Hong Kong. Clin Rehabil 2008;22:112-24. 41. Page SJ, Levine P, Leonard A, Szaflarski JP, Kissela BM. Modified constraint-induced therapy in chronic stroke: results of a singleblinded randomized controlled trial. Phys Ther 2008;88:333-40. 42. Lin KC, Wu CY, Wei TH, Lee CY, Liu JS. Effects of modified constraint-induced movement therapy on reach-to-grasp movements and functional performance after chronic stroke: a randomized controlled study. Clin Rehabil 2007;21:1075-86. 43. Wu CY, Lin KC, Chen HC, Chen IH, Hong WH. Effects of modified constraint-induced movement therapy on movement kinematics and

www.archives-pmr.org

Assistive technologies for poststroke arm rehabilitation

44.

45.

46.

47.

48.

49.

50.

51.

52.

daily function in patients with stroke: a kinematic study of motor control mechanisms. Neurorehabil Neural Repair 2007;21:460-6. Dromerick AW, Edwards DF, Hahn M. Does the application of constraint-induced movement therapy during acute rehabilitation reduce arm impairment after ischemic stroke? Stroke 2000;31:2984-8. Dromerick AW, Lang CE, Birkenmeier RL, et al. Very Early Constraint-Induced Movement during Stroke Rehabilitation (VECTORS): a single-center RCT. Neurology 2009;73:195-201. Broga˚rdh C, Vestling M, Sjo¨lund BH. Shortened constraint-induced movement therapy in subacute strokedno effect of using a restraint: a randomized controlled study with independent observers. J Rehabil Med 2009;41:231-6. Broga˚rdh C, Lexell J. A 1-year follow-up after shortened constraintinduced movement therapy with and without mitt poststroke. Arch Phys Med Rehabil 2010;91:460-4. Lum PS, Burgar CG, Shor PC, Majmundar M, Van der Loos M. Robot-assisted movement training compared with conventional therapy techniques for the rehabilitation of upper-limb motor function after stroke. Arch Phys Med Rehabil 2002;83:952-9. 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;23:505-14. Hesse S, Werner C, Pohl M, Rueckriem S, Mehrholz J, Lingnau ML. Computerized arm training improves the motor control of the severely affected arm after stroke: a single-blinded randomized trial in two centers. Stroke 2005;36:1960-6. 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: 305-10. Daly JJ, Hogan N, Perepezko EM, et al. Response to upper-limb robotics and functional neuromuscular stimulation following stroke. J Rehabil Res Dev 2005;42:723-36.

www.archives-pmr.org

985 53. Masiero S, Celia A, Rosati G, Armani M. Robotic-assisted rehabilitation of the upper limb after acute stroke. Arch Phys Med Rehabil 2007;88:142-9. 54. Hsieh YW, Wu CY, Liao WW, Lin KC, Wu KY, Lee CY. Effects of treatment intensity in upper limb robot-assisted therapy for chronic stroke: a pilot randomized controlled trial. Neurorehabil Neural Repair 2011;25:503-11. 55. Crow JL, Lincoln NB, De Weerdt N, Nouri FM. The effectiveness of EMG biofeedback in the treatment of arm function after stroke. Int Disabil Stud 1989;11:155-60. 56. Armagan O, Tascioglu F, Oner C. Electromyographic biofeedback in the treatment of the hemiplegic handda placebo-controlled study. Am J Phys Med Rehabil 2003;82:856-61. 57. Fregni F, Boggio PS, Valle AC, et al. A sham-controlled trial of a 5-day course of repetitive transcranial magnetic stimulation of the unaffected hemisphere in stroke patients. Stroke 2006;37: 2115-22. 58. Saposnik G, Teasell R, Mamdani M, et al. Effectiveness of virtual reality using Wii gaming technology in stroke rehabilitation: a pilot randomized clinical trial and proof of principle. Stroke 2010;41:1477-84. 59. Stein J, Hughes R, D’Andrea S, et al. Stochastic resonance stimulation for upper limb rehabilitation poststroke. Am J Phys Med Rehabil 2010;89:697-705. 60. Uswatte G, Taub E, Morris D, Barman J, Crago J. Contribution of the shaping and restraint components of constraint-induced movement therapy to treatment outcome. NeuroRehabilitation 2006;21: 147-56. 61. Rodgers H, Macintosh J, Price C, et al. Does an early increasedintensity interdisciplinary upper limb therapy programme following stroke improve outcome? Clin Rehabil 2003;17:579-89. 62. de Kroon JR, Ijzerman MJ, Chae J, Lankhorst GJ, Zilvold G. Relation between stimulation characteristics and clinical outcome in studies using electrical stimulation to improve motor control of the upper extremity in stroke. J Rehabil Med 2005;37:65-74.