G Model

REHAB 1194 1–6

Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

Available online at

ScienceDirect
www.sciencedirect.com

1
2 Review

3 Adding electrical stimulation during standard rehabilitation

4 after stroke to improve motor function. A systematic review
5 and meta-analysis
6 Q1 Sharareh Sharififar a,*, Jonathan J. Shuster b, Mark D. Bishop c
7 a
Physiotherapy Department, University of Florida, Gainesville, Florida, USA
8 b
Department of Health Outcomes and Policy, Biostatistics Epidemiology and Research Design, Clinical and Translational Science Institute, Biostatistician, UF
9 Clinical Research Center, College of Medicine, University of Florida, Gainesville, Florida, USA
10 c
Department of Physical Therapy, College of Public Health and Health Professions, University of Florida, Gainesville, Florida, USA

A R T I C L E I N F O A B S T R A C T

Article history: Background: Clinical studies have shown that sensory input improves motor function when added to
Received 18 September 2017 active training after neurological injuries in the spinal cord.
Accepted 10 June 2018 Objective: We aimed to determine the effect on motor function of extremities of adding an electrical
sensory modality without motor recruitment before or with routine rehabilitation for hemiparesis after
Keywords: stroke by a comprehensive systematic review and meta-analysis.
Sensory Methods: We searched databases including MEDLINE via PubMed and the Cochrane Central Register of
Stimulation
Controlled Trials from 1978 to the end of November 2017 for reports of randomized controlled trials or
Stroke
Function
controlled studies of patients with a clinical diagnosis of stroke who underwent 1) transcutaneous
electrical nerve stimulation (TENS) or peripheral electromyography-triggered sensory stimulation over a
peripheral nerve and associated muscles or 2) acupuncture to areas that produced sensory effects,
without motor recruitment, along with routine rehabilitation. Outcome measures were motor
impairment, activity, and participation outcomes defined by the International Classification of
Functioning, Disability and Health.
Results: The search yielded 11 studies with data that could be included in a meta-analysis. Electrical
sensory inputs, when paired with routine therapy, improved peak torque dorsiflexion (mean difference
[MD] 2.44 Nm, 95% confidence interval [CI] 0.26–4.63). On subgroup analysis, the combined therapy
yielded a significant difference in terms of sensory stimulation without motor recruitment only on the
Timed Up and Go test in the chronic phase of stroke (MD 3.51 sec, 95% CI 3.05–3.98). The spasticity score
was reduced but not significantly (MD 1.11 points, 95% CI -2.35 0.13).
Conclusion: Electrical sensory input can contribute to routine rehabilitation to improve early post-stroke
lower-extremity impairment and late motor function, with no change in spasticity. Prolonged periods of
sensory stimulation such as TENS combined with activity can have beneficial effects on impairment and
function after stroke.
C 2018 Published by Elsevier Masson SAS.

11
12 1. Introduction few months after brain insult reduces the individual’s potential to 16
participate in therapy [2,3]. The initial damage to the neuronal 17
13 Hemiparesis is one of the prominent impairments caused by pathways followed by functional reorganization halt the motor 18
14 stroke that affects activities of daily living and quality of life recovery; however, many stimuli have shown potential therapeu- 19
15 [1]. Clinically, the failure to recover motor deficits rapidly within a tic benefit to improve functional recovery after stroke. Tradition- 20
ally, in stroke research, potential therapeutic strategies targeting 21
motor recovery emphasize motor stimulation associated with 22
* Corresponding author. Department of Physical Therapy, College of Public Health massed motor practice for the recovery of motor skills [4]. 23
and Health Professions, University of Florida, UFHSC, Box, 100154 Gainesville, Afferent motor stimulation leads to neurological improve- 24
Florida, USA. ments, but the recovery of function varies. The effects of increased 25
E-mail addresses: shararehsharififar@gmail.com (S. Sharififar), shusterj@ufl.edu
(J.J. Shuster), bish@phhp.ufl.edu (M.D. Bishop).
sensory input on motor outcomes have been relatively neglected in 26

https://doi.org/10.1016/j.rehab.2018.06.005
1877-0657/ C 2018 Published by Elsevier Masson SAS.

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005
 G Model
REHAB 1194 1–6

2 S. Sharififar et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

27 the rehabilitation literature as compared with those of other SPORTDiscus, and Cumulative Index to Nursing and Allied Health 87
28 interventions, even though a number of studies suggested clinical Literature (CINAHL), Google, and WorldCat from 1978 to the end of 88
29 benefits. In 1915, Franz et al. were the first to report success in November 2017. We also hand-searched the reference lists of 89
30 recovering upper extremity motor function by using simple published reviews and narrative review articles. In addition, 90
31 rehabilitative techniques such as massage and vibration. [5] In abstracts published in relevant conference proceedings were 91
32 particular, cutaneous and proprioceptive afferent information polled. 92
33 facilitates the improvement of motor performance and promotes The search terms included ‘‘cerebrovascular accident’’ OR 93
34 effective motor learning because it increases the cortico-motor ‘‘stroke’’ OR ‘‘hemiplegia’’ OR ‘‘hemiparesis’’ OR ‘‘hemiparetic,’’ 94
35 excitability in areas representing the stimulated body parts [6– AND ‘‘vibration,’’ ‘‘pressure’’ OR ‘‘haptics,’’ ‘‘kinesthetic*,’’ ‘‘stretch,’’ 95
36 10]. Subsequent research has shown that electrical sensory input ‘‘weight,’’ ‘‘joint angle,’’ OR ‘‘tactile,’’ ‘‘touch,’’ OR ‘‘texture,’’ OR 96
37 alters the sensory and motor cortical maps [3,11–13] and clinical ‘‘transcutaneous electrical nerve stimulation,’’ ‘‘TENS,’’ ‘‘neuro- 97
38 studies have shown that sensory input improved motor function muscular training,’’ AND ‘‘motor,’’ ‘‘movement,’’ OR ‘‘motion,’’ OR 98
39 when added to active training after neurological injuries in the ‘‘mobility,’’ OR ‘‘function,’’ OR ‘‘performance’’. Although each 99
40 spinal cord [14]. database used differing syntax, this set of search terms was the 100
41 A recent systematic review and a meta-analysis of the effects of basis for the searches in each of those databases. 101
42 augmenting rehabilitation with sensory stimulation were publis- We included reports of randomized controlled trials or 102
43 hed in the past 10 years [15,16]. The review by Laufer et al. did not controlled trials with parallel or crossover designs. In studies 103
44 lead to a meta-analysis because of too few studies that met with more than 2 arms, the values related to e-stim or 104
45 inclusion criteria and the meta-analysis by Veerbeek et al. did not placebo interventions combined with the routine therapy were 105
46 separate sensory stimulations that lead to motor recruitment from recorded. 106
47 pure sensory-level stimulation. All reports of sensory modalities investigated such as vibration, 107
48 In the current analysis we expanded on these studies. In pressure, haptics, stretch, weight, proprioception, tactile, touch, 108
49 particular, we used a newer method for the meta-analysis. The texture, thermal, and transcutaneous electrical nerve stimulation 109
50 random-effects method is commonly used, but it forces the (TENS) as adjunct therapy to routine rehabilitation to improve 110
51 distribution of effect sizes to be independent of the study design, post-stroke motor function were included. However, the primarily 111
52 for no relation between a study’s effect size and the size of study. sensory modality of interest was e-stim. Different e-stim inter- 112
53 This is a strong assumption, which is not necessarily true for all ventions such as repetitive electrical nerve stimulation, acupunc- 113
54 studies, so the estimate of the between-studies variance will have ture or muscle stimulation have common sensory nerve axon 114
55 poor precision, limiting investigators’ and clinicians’ ability to stimulation at specific peripheral sites. The comparison groups 115
56 apply the commonly used random-effects method appropriately. included active treatment (active or passive exercise, e-stim over 116
57 Because of major issues with this method [17,18], we used the different sites), placebo (sham, sub-sensory threshold stimulus 117
58 approach proposed by Shuster et al. to overcome the fundamental intensity), and no treatment. 118
59 limitations of the empirical weighting method. In this approach, The outcome measures were any measurable activity-based 119
60 ‘‘studies-at-random’’ assumes that studies are drawn from a motor function. To increase the generalizability of the results to the 120
61 population of studies that are independent and implies that the population of interest, we included studies of participants with all 121
62 weight given to each study’s effect size is a random variable [19]. durations of stroke-related sequelae (acute, sub-acute, and 122
63 Thus, this current study extended the previous work by: chronic) and severity of stroke-induced hemiparesis. The Interna- 123
tional Classification of Functioning, Disability and Health (ICF) 124
65  potentially broadening the available literature that could be allowed us to document functioning and disability outcomes 125
66 reviewed; reported by the included studies across the levels of function (the 126
67  using a method suitable for pooling a small number of studies body, the person, the society). 127
68 with a small number of participants;
69  examining the effects of electrical sensory stimulation without 2.2. Assessment of quality of studies 128
70 motor recruitment across levels of function.
Two independent raters (SSh and MDB) assessed the methodo- 129
71 The primary questions driving this current analysis were as logical quality of the included articles by using the standardized 130
72 follows: validated PEDro scale for the quality of controlled clinical trials 131
[20], an 11-item scale [21] previously used in systematic reviews 132
74  Is there evidence for the efficacy of augmenting rehabilitation [22], with total scores ranging from 0 to 10. Agreement in quality 133
75 with sensory input using electrical stimulation (e-stim) to assessment on PEDro rating between evaluators was measured by 134
76 improve motor function in a person after stroke? the Kappa statistic, with 95% confidence intervals (CIs). A kappa of 135
77  What are the specific effects of adding an electrical sensory 1 ndicates perfect agreement, and a kappa of 0 indicates agree- 136
78 modality before or along with routine rehabilitation on motor ment equivalent to chance. 137
79 impairment, activity, and participation outcomes after stroke?
2.3. Data analysis 138

The initial screening step involved examining the article title 139
80 2. Methods and major key words, then abstracts and full texts. 140
One author (SSh) extracted the relevant data from each article 141
81 2.1. Identification and selection of studies and recorded them on standardized Microsoft Excel spreadsheets. 142
To ensure accurate copying of the data, a second reviewer (MDB) 143
82 We extracted key words associated from relevant articles and independently checked the information in the forms with the 144
83 used Medical Subject Heading (MeSH terms) and Major Headings related articles. The data recorded were the study design, 145
84 to search for human studies in any language in the databases participant characteristics, type of intervention and co-interven- 146
85 MEDLINE via PubMed, Cochrane Central Register of Controlled tion, region of the body to which the stimulus was applied, and the 147
86 Trials, Cochrane Database of Systematic Reviews, Web of Science, mean pre-and post-intervention values for each measure. 148

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005
 G Model
REHAB 1194 1–6

S. Sharififar et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 3

Fig. 1. Flow of studies in the review. *Papers may have been excluded for failing to meet more than one inclusion criteria. rTMS, repetitive transcranial magnetic stimulation.

149 2.4. Statistical analysis Characteristics of the included studies are in Table 1. The meta- 174
analysis included 11 studies that investigated the effects of 175
150 The meta-analysis involved use of SAS v9.4 (SAS Institute, Inc. electrical sensory input without motor recruitment across 176
151 Cary, NC). We used a studies at random approach, which estimated levels of function. These studies included 450 participants with 177
152 the patient-level effect size by assigning the weight of each point stroke (the ratio of males to females 5:1). The mean (SD) age of 178
153 estimate based on the total sample size [23]. We chose this method participants was 61.28 years (range 49–72), with mean post-stroke 179
154 given limitations to the common used random-effects model. period 37.29 months (SD 36.21). Each study had a small sample 180
155 The mean effect sizes of the outcome measures were calculated size: < 50 individuals in each group (range 14–67 total partici- 181
156 by using the information provided in the articles. When the number of pants). The single controlled trial examined the effectiveness of 182
157 trials was > 5, the difference in means (MD; post-intervention mean e-stim using sock or glove electrodes with an intensity that was 183
158 values minus pre-intervention mean values) in the experimental and sufficient for sensory level but insufficient to produce a tetanic 184
159 comparison groups was considered the effect size of each of the trials muscle contraction [25]. The remaining studies used randomized 185
160 [23]. When the number of studies was small (< 5), we used a controlled trial designs, repeated measures, or crossover designs. 186
161 conservativet-approximation method [19,23]. Both periods of crossover Most of the studies usually validated and justified their measure- 187
162 trials were included in the analyses [24]. Secondary analyses stratified ments and outcomes and described the intervention effectively. In 188
163 data by chronicity, comparing participants < 6 months post-stroke with most studies, the description of the baseline characteristics was 189
164 those > 6 months post-stroke for all outcomes with sufficient data. clear, and the groups did not differ significantly in terms of certain 190
165 Two-sided P < 0.05 was considered statistically significant. characteristics. However, the studies failed to screen for specific 191
deficits such as type, location, chronicity, and severity of stroke. The 192
associated dysfunction in the studies’ inclusion criteria resulted in 193
166 3. Results
non-tailored treatments and the treatment of only one of the many 194
deficits manifested in the population under the study. 195
167 The results of the systematic search are summarized in
The methods, amount of exposure to the e-stim, and associated 196
168 Fig. 1. Among the 676 articles identified, 189 were considered
exercise protocols (passive versus active) varied among studies. 197
169 for potential inclusion; reports of 11 were finally included.
Three of the included studies investigated TENS, and the remaining 198
170 3.1. Characteristics of studies studies investigated peripheral, sensory level stimulation modali- 199
ties over a peripheral nerve and its distributed muscle supply, or 200
171 The mean (SD) PEDro quality score for the 11 selected reports acupuncture areas. 201
172 was 7.18/10 (1.32, range 5–). The Kappa score was 0.36 (95% CI Ten studies investigated electrical sensory input with routine 202
173 0.05–0.67) (P < 0.00). therapy or active treatment of specific standardized isotonic 203

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005
 G Model
REHAB 1194 1–6

4 S. Sharififar et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

Table 1
Q3 Summary of included studies.

Study Design Participants Intervention Outcome measures

a
Chen, Chen et al., RCT n = 24 All subjects received standard physical, Impairment = Fmax/Max (%) H-
2005 [35] Sex = 10F occupational, and speech therapy interventions reflex latencyb (ms)
Age (yr) = 57 as per routine of the inpatient stroke Activity = 10 m walking time (s)
Mean (SD) chronicity rehabilitation program. Exp = ES of paretic
(mo) = 23.5 (1.6) gastrocnemius
Max constant intensity without inducing
muscle contraction, 0.2 ms pulse width at 20 Hz
20 min (*)/day  6/wk  4 wk
Con = stimulation intensity at zero
Hui-Chan, Ng, and RCT home-based n = 52 Exp = TENS over paretic leg on 4 acupuncture Impairment = composite spasticity
Mak, 2009 [33] Sex = NR points scale, maximum voluntary
Age (yr) = 56 2-3 times sensory threshold, 0.2 ms square contraction for plantar and
Mean (SD) chronicity (yr) = 4.7 pulses at 100 Hz dorsiflexion
(3.4) 60 min (*)/day  5/wk  4 wk Activity = Timed Up and Go test,
Con = placebo 6 min walk test, gait velocity (m/s)
Both = FREc was followed by TENS/placebo Follow up = 0, 4 weeks
Mukherjee, McPeak RCT cross-over n = 14 Exp = Electro-acupuncture of paretic arm with Impairment = modified Ashworth
et al., 2007 [34] Mean (SD) age (yr) = 63.1 (1.7) unpleasant but tolerable intensity at 2 Hz for the wrist and the index finger
Sex = 6F 40 min (*)/day  2/wk  6 wk flexor and extensors, average speed
Mean (SD) chronicity (yr) = 9.4 Both = active-assisted, isokinetic, and isometric dependent reflex torque at the
(2.5) strengthening 4 different angular velocities:
30 min (*)/day  2/wk 20,45,60,75
Ng and Hui-Chan RCT home-based n = 41 Exp = TENS over paretic leg on 4 acupuncture Impairment = composite spasticity
2007 [26] Mean (SD) age (yr) = 55.9 (1.1) points scale, maximum voluntary
Sex = 8F 2-3 times sensory threshold, 0.2 ms square contraction for plantar and
Mean (SD) chronicity (yr) = 4.8 pulses at 100 Hz dorsiflexion
(0.5) 60 min (*)/day  5/wk  4 wk Activity = gait velocity (m/s)
Con = placebo Follow up = 0, 4 weeks
Both =TENS/placebo followed by FREd
Ng and Hui-Chan RCT home-based n = 52 Exp = TENS over paretic leg on 4 acupuncture Activity = Timed Up and Go test,
2009 [27] Mean (SD) age (yr) = 57.3 (1.1) points 6 min walk test, gait velocity (m/s)
Sex = 11F 2-3 times sensory threshold, 0.2 ms square Follow up = 0, 4 weeks
Mean (SD) chronicity (yr) = 4.5 pulses at 100 Hz
(0.4) 60 min (*)/day  5/wk  4 wk
Con = placebo
Both = FREd was followed by TENS/placebo
Park, Seo et al., RCT n = 29 Exp = ES of paretic leg over lateral and medial Impairment = modified-Ashworth
2014 [28] Mean (SD) age (yr) = 71.2 (0.6) quadriceps and gastrocnemius 0.01 mA up to for calf muscles, balance; anterior-
Sex = 9F 90% amplitude using sub sensory threshold, posterior and mediolateral sway,
Mean (SD) chronicity 200 ms at 100 Hz gait analysis using Opto-gait,
(mo) = 18.6 (0.4) 30 min (*)/day  5/wk  6 wk cadence, paretic step and stride
Con = placebo lengths
Both = general exercise program (ROM training, Activity = Timed Up and Go test
functional mat training, gait training each
10 min)
Peurala, Pitkanen CT n = 67 Exp = ES of paretic hand or foot Impairment = modified Ashworth,
et al., 2002 [25] Mean (SD) age (yr) = 54.4 (1.1) glove or sock electrode monophasic constant limb skin sensation somatosensory
Sex = 18F cutaneous twin pulses just below sensory evokes potentials
Mean (SD) chronicity (yr) = 3.3 threshold at 50 Hz Activity = walking speed (m/s)
(1.1) 20 min 2(*)/day  3 wk
Con = placebo
Both = individual PT without any specific
approach
Tekeoglu, Adak RCT n = 60 Exp = ES of the paretic hand over triceps brachii Impairment = Ashworth for the
et al., 1998 [29] Mean (SD) age (yr) = 54.1 (0.8) and the paretic leg over common peroneal elbow flexors, knee and ankle
Sex = 29F nerve Activity = Barthel index
Mean (SD) chronicity Sensory threshold at the bearable pain level
(mo) = 42.5 (1.5) with square pulses of 0.2 ms at 100 Hz
30 min (*)/day  5 day/wk  8 wk
Con = placebo
Both = Todd-Davis exc for 40 min in the
morning and afternoon
Yan 2002 [30] RCT n = 33 Exp = ES of the paretic lower extremity Impairment = max voluntary
Mean (SD) age (yr) = 70.8 (0.9) 20-30 mA at 2 to 3 times sensory threshold contraction and torque of ankle
Sex = 17F 0.3 ms at 30 Hz dorsi/plantar flexion,
Chronicity (wk) = 5.6 60 min (*)/day  5/wk  3 wk proprioception threshold of the
Con = placebo or no treatment knee joint
Both = standard rehabilitation including both Activity = Timed Up and Go test
PT and OT each lasting 60 min Follow up = 0, 5 weeks
Yan and Hui-Chan RCT n = 38 Exp = ES of the paretic lower extremity Impairment = max voluntary
2009 [31] Mean (SD) age (yr) = 70.6 (1.3) 0.2 ms at 100 Hz within the subject’s tolerance contraction torque (Nm),
Sex = 19F level on 3 lower-extremity acupuncture points Composite spasticity score
Mean (SD) chronicity 60 mins (*)/day  5/wk  3 wk Activity = Timed Up and Go Test
(day) = 9.5 (0.6) Con = placebo or no treatment Follow up = 0, 5 weeks
Both = standard rehabilitation including both
PT and OT each lasting 60 min

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005
 G Model
REHAB 1194 1–6

S. Sharififar et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx 5

Table 1 (Continued )

Study Design Participants Intervention Outcome measures

Yavuzer, Oken RCT n = 30 Exp = ES of the paretic lower extremity Impairment = modified-Ashworth,
et al., 2007 [32] Mean (SD) age (yr) = 63.1 (1.7) 10 mA below palpable contraction, duty cycle gait kinematics
Sex = 14F 10 s on/10 s off 240 ms pulse width 240 ms pulse Activity = lower extremity motor
Mean (SD) chronicity (mo) = 3.4 width asymmetric biphasic rectangular at recovery, walking velocity (m/s)
(0.4) 35 Hz Participation = FIM
30 min(*)/day  5/wk  4 wk
Con = placebo
Both = Conventional stroke rehabilitation
program 2 to 5 hr (*)/day  5/wk  4 wk

CT: controlled trial; BL: baseline; Con: control group; ES: electrical stimulation; Exp: experiment group; F: female; FIM: functional independence measure; FRE: functional
resistance exercise; NA: not applicable; max: maximum; mo: months; ROM: range of motion; OT: occupational therapy; PT: physiotherapy; wk: week; yr: year; TENS:
transcutaneous electrical nerve stimulation.
a
Standard stroke rehabilitation interventions such as neuromuscular re-education, RCT, randomized controlled trial; ROM, strengthening, neuromuscular facilitation,
functional training, and use of physical modalities and adaptive equipment as indicated.
b
Amount of the inhibition of the H reflex during vibration, excitability of soleus stretch reflex in terms of latency, onset angles and magnitude.
c
FRE included (1) loading exercise on the affected leg; (2) stepping-up exercise with the affected leg; (3) stepping-down exercise with the unaffected leg; (4) heel lifts from
a dorsiflexed position in standing and 2 functional training; (5) standing up from a chair, walking a short distance, and returning to the chair; and 6) walking with rhythmic
auditory cues generated by a metronome.
Q4 d The physiotherapist provided standardized progression by using higher wooden blocks when subjects could perform the weight-bearing exercises 20 times without
compensatory movement and by increasing the number of repetitions completed within 10 min. Walking was progressed by increasing its speed.

Table 2
Summary effect sizes of combined sensory and motor inputs.

ICF level Outcomes N/n MD 95% CI P-values

Activity Timed Up and Go test (sec) 4/171 7.87 7.08 to 22.82 0.19
Timed Up and Go test (sec), chronic 3/133 3.51* 3.05 to 3.98 0.00
Walking ability (m/s) 7/266 0.04 0.02 to 0.11 0.14
Walking ability (m/s), chronic 5/184 0.03 0.05 to 0.11 0.28
Body impairment Ashworth/modified Ashworth/CSS 8/487 0.83 1.77 to 0.10 0.07
Ashworth/modified Ashworth/CSS, chronic 6/182 0.17 0.85 to 0.51 0.53
Dorsrflexion torque (N.m) 4/158 2.44* 0.26 to 4.63 0.03
Peak plantarflexion torque (N.m) 3/126 1.05 4.52 to 6.62 0.50

ICF: International classification of functioning; disability and health; CSS: Composite Spasticity Scale; MD: mean difference; N: number of studies; n: number of participants.
*
P < 0.05

204 training with routine therapy. The participants received e-stim for mobility (t = 3.51, P = 0.0009). We found significant effects 234
205 a median of 20 sessions (range 12–61; interquartile range 6.75) associated with increased dorsiflexion torque and decreased time 235
206 over 3 to 8 weeks. The comparator group was placebo with routine to complete the TUG in a chronic subgroup of patients post-stroke. 236
207 therapy or active treatment of specific standardized isotonic
208 training [26–33]. One study used routine therapy alone in the 4. Discussion 237
209 comparison group [34]. One study used routine therapy before e-
210 stim compared to placebo [35]. Another study compared the Our findings from this systematic review and meta-analysis 238
211 effects of adding electrical sensory input to routine therapy before show that electrical sensory stimulation without muscle contrac- 239
212 active treatment of specific standardized isotonic training [27]. tion can help improve post-stroke lower-extremity muscle 240
213 None of the included studies performed blinded or concealed function and functional mobility before or with routine rehabili- 241
214 allocations. Blinding participants to the exercise intervention is tation for hemiparesis after stroke. An insufficient number of 242
215 difficult in rehabilitation studies. Almost all studies used sham or studies reported common outcome measures. We used the 243
216 sub-sensory threshold stimuli in controls. In half of the studies, t-approximation method to calculate the effect sizes for 5 of 244
217 evaluators who were blinded to the groups’ assignment evaluated 13 of the reported outcomes, which were pooled from more than 245
218 the baseline and post-intervention outcomes. Only a few studies 3 studies per outcome measure. The variability in methodology, in 246
219 included a therapist blinded to the groups’ assignment to deliver addition to the low power to find real effects in many of the trials, 247
220 treatment. A common problem in all these studies was the small may have contributed to these limited findings. 248
221 sample size: < 30 participants in each group. Thus, the studies Two studies showed enhanced dorsiflexion torque production 249
222 lacked the power necessary to assess the statistical significance of with e-stim applied to the peripheral nerve and acupuncture 250
223 the sensory inputs. points as compared with placebo during the acute and chronic 251
224 A wide range of outcome measures was used, so we considered stages after stroke. Patients in both studies received 60 min of e- 252
225 the effects of sensory input by grouping 5 outcome measures in stim with active exercise 4 to 6 hr/day every weekday for 3 to 253
226 terms of impairment on the ICF: the activity (Timed Up and Go test 4 weeks. Increases in primary somatosensory cortical area by 254
227 [TUG] and walking ability) and the body impairment (Ashworth/ sensory stimulation may have been caused by sensory cortical 255
228 modified Ashworth scale or Composite Spasticity Score, dorsi- long-term potentiation processes that led to skill development 256
229 flexion and peak plantar flexion torque) (Table 2). An insufficient [36,37]. This is a relatively short duration of e-stim applied to the 257
230 number of studies reported common outcomes during the peripheral nerve and acupuncture points that can be easily used in 258
231 participation. The additional benefit of using sensory input as an patients post-stroke in the therapy setting or at their homes. 259
232 adjunct to routine therapy appeared seemed to be relatively high Two studies showed beneficial effects of the TUG of sensory 260
233 for lower- extremity impairment (t = 2.44, P = 0.03) and functional input combined with passive and active exercises or routine 261

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005
 G Model
REHAB 1194 1–6

6 S. Sharififar et al. / Annals of Physical and Rehabilitation Medicine xxx (2018) xxx–xxx

262 therapy as compared with placebo. The TUG includes a series of [11] Albe-Fessard D, Liebeskind J. Origin of somato-sensitive messages 326
activating the cells of the motor cortex in monkeys. Exp Brain Res 1966;1: 327
263 motor tasks that also demand muscle strength and movement 127–46. 328
264 coordination in addition to balance control. Improvements in TUG [12] Kreisman NR, Zimmerman ID. Cortical unit responses to temprature stimula- 329
265 time have practical significance. For people in the chronic phase tion of the skin. Brain Res 1971;25:184–7. 330
[13] Marconi B, et al. Long-term effects on cortical excitability and motor recovery 331
266 after stroke who received TENS in addition to active exercises, the induced by repeated muscle vibration in chronic stroke patients. Neurorehabil 332
267 mean time to get up from a chair and walking speed changed from Neural Repair 2011;25:48–60. 333
268 25 to 19 sec. These participants were more likely independent in [14] Field-Fote JG-OAEC. Cortical vs. afferent stimulation as an adjunct to func- Q2 334
tional task practice training: a randomized, comparative pilot study in people 335
269 the transfer tasks needed for activities of daily living [38]. Because with cervical spinal cord injury. Clin Rehabil 2014;1–12. 336
270 TENS could be taught easily, it represents a low-cost treatment [15] Veerbeek JM, et al. What is the evidence for physical therapy poststroke? A 337
271 modality that could be readily applied to patients after stroke and systematic review and meta-analysis. PLoS One 2014;9:e87987. 338
272 help in promoting this treatment. [16] Laufer Y, Elboim-Gabyzon M. Does sensory transcutaneous electrical stimu- 339
lation enhance motor recovery following a stroke? A systematic review. 340
273 The placebo protocols in the included studies involved the use Neurorehabil Neural Repair 2011;25:799–809. 341
274 of electrodes placed over acupuncture points or non-motor areas [17] Emerson JD, Mosteller HDF. Simple robust procedures for combining risk 342
275 and delivered no or subliminal stimuli. Although this is encourag- differences in sets of 22 tables. Stat Med 1996;15:1465–88. 343
[18] Emerson JD, Hoaglin DC, Mosteller F. A comparison of procedures for combin- 344
276 ing evidence to support the benefits of sensory input, we cannot ing risk differences in sets of 22 tables from clinical trials. J Ital Stat Soc 345
277 definitively conclude that the addition of e-stim to rehabilitation 1993;2:269–90. 346
278 was a superior intervention given that no study evaluated the [19] Shuster JJ, Guo JD, Skyler JS. Meta-analysis of safety for low event-rate 347
binomial trials. Res Synth Methods 2012;3. 348
279 believability of the placebo or determined the expectations of the [20] Sherrington C, et al. PEDro. A database of randomized trials and systematic 349
280 effectiveness of the intervention. Studies of other rehabilitation reviews in physiotherapy. Man Ther 2000;5:223–36. 350
281 interventions indicate that baseline differences in the believability [21] Verhagen AP, et al. The Delphi list: a criteria list for quality assessment of 351
randomized clinical trials for conducting systematic reviews developed by 352
282 of a placebo or expectations for treatment effectiveness have Delphi consensus. J Clin Epidemiol 1998;51:1235–41. 353
283 potential to confound the results [39,40]. [22] Maher CG, et al. Reliability of the PEDro scale for rating quality of randomized 354
284 Several problems were identified, such as variability in how controlled trials. Phys Ther 2003;83:713–21. 355
285 researchers delivered somatosensory stimulations with the [23] Shuster JJ. Empirical vs natural weighting in random effects meta-analysis. 356
Stat Med 2010;29:1259–65. 357
286 associated exercise protocols (passive versus active) and variabili- [24] Curtin F, Elbourne D, Altman DG. Meta-analysis combining parallel and cross- 358
287 ty in the reported outcomes. The optimal dosage needs to be over clinical trials. II: Binary outcomes. Stat Med 2002;21:2145–59. 359
288 addressed in future studies. Also, researchers and clinicians should [25] Peurala SH, et al. Cutaneous electrical stimulation may enhance sensorimotor 360
recovery in chronic stroke. Clin Rehabil 2002;16:709–16. 361
289 use measurements that are specific to both the residual deficits of [26] Ng SS, Hui-Chan CW. Transcutaneous electrical nerve stimulation combined 362
290 the post-stroke patient and adjust treatment parameters based on with task-related training improves lower limb functions in subjects with 363
291 augmenting motor learning and motor function. chronic stroke. Stroke 2007;38:2953–9 [00392499]. 364
[27] Ng SSM, Hui-Chan CWY. Does the use of TENS increase the effectiveness of 365
292 In summary, our systematic review and meta-analysis of exercise for improving walking after stroke? A randomized controlled clinical 366
293 11 studies that augmented rehabilitation with electrical sensory trial. Clin Rehab 2009;23:1093–103. 367
294 stimulation without motor recruitment showed benefits for [28] Park J, et al. The effects of exercise with TENS on spasticity, balance, and gait in 368
patients with chronic stroke: a randomized controlled trial. Med Sci Monit 369
295 improving lower-extremity impairment and function. Future 2014;20:18901–21896. 370
296 studies could consider measures of impairment and function with [29] Tekeoglu Y, Adak B, Goksoy T. Effect of transcutaneous electrical nerve 371
297 stronger, more direct associations between the two. stimulation (TENS) on Barthel Activities of Daily Living (ADL) index score 372
following stroke. Clin Rehabil 1998;12:277–80. 373
[30] Yan T. The effectiveness of early neuromuscular electrical stimulation on 374
298 References lower extremity functions of stroke patient. Hong Kong Polytechnic 375
2002;289 [People’s Republic of China]. 376
299 [1] Urban PP, et al. Occurence and clinical predictors of spasticity after ischemic [31] Yan T, Hui-Chan CW. Transcutaneous electrical stimulation on acupuncture 377
300 stroke. Stroke 2010;41:2016–20. points improves muscle function in subjects after acute stroke: a randomized 378
301 [2] Han L, Law-Gibson D, Reding M. Key neurological impairments influence controlled trial. J Rehabil Med 2009;41:312–6. 379
302 function-related group outcomes after stroke. Stroke 2002;33:1920–4. [32] Yavuzer G, et al. Effect of sensory-amplitude electric stimulation on motor 380
303 [3] Nudo RJ, Friel KM, Delia SW. Role of sensory deficits in motor impairments recovery and gait kinematics after stroke: a randomized controlled study. Arch 381
304 after injury to primary motor cortex. Neuropharmacology 2000;39:733–42. Phys Med Rehab 2007;88:710–4. 382
305 [4] de Kroon JR, et al. Relation between stimulation characteristics and clinical [33] Hui-Chan CW, Ng SS, Mak MK. Effectiveness of a home-based rehabilitation 383
306 outcome in studies using electrical stimulation to improve motor control of programme on lower limb functions after stroke. Hong Kong Med J 2009;15: 384
307 the upper extremity in stroke. J Rehab Med 2005;37:65–74 [Taylor & Francis 42–6. 385
308 Ltd]. [34] Mukherjee M, et al. The effect of electro-acupuncture on spasticity of 386
309 [5] Franz SI, Sheets M, Wilson AA. Possibility of recovery of motor function in long- the wrist joint in chronic stroke survivors. Arch Phys Med Rehabil 2007;88: 387
310 standing hemiplegia;preliminary report. JAMA 1915;65:2150–4. 159–66. 388
311 [6] Cuypers K, et al. Long-term T.E.N.S. treatment improves tactile sensitivity in [35] Chen SC, et al. Effects of surface electrical stimulation on the muscle-tendon 389
312 MS patients. Neurorehabil Neural Repair 2010;24:420–7. junction of spastic gastrocnemius in stroke patients. Disabil Rehabil 390
313 [7] Floel A, et al. Influence of somatosensory input on interhemispheric inter- 2005;27:105–10. 391
314 actions in patients with chronic stroke. Neurorehabil Neural Repair 2008;22: [36] Buonomano DV, Merzenich MM. Cortical plasticity: from synapses to maps. 392
315 477–85. Annu Rev Neurosci 1998;21:149–86. 393
316 [8] Ridding MC, et al. Changes in muscle responses to stimulation of the motor [37] Dobkin BH. Do electrically stimulated sensory inputs and movements lead 394
317 cortex induced by peripheral nerve stimulation in human subjects. Exp Brain to long-term plasticity and rehabilitation gains? Curr Opin Neurol 2003;16: 395
318 Res 2000;131:135–43. 685–91. 396
319 [9] Tinazzi M, et al. Long-lasting modulation of human motor cortex following [38] Podsiadlo D, Richardson S. The timed ‘‘Up & Go’’: a test of basic functional 397
320 prolonged transcutaneous electrical nerve stimulation (TENS) of forearm mobility for frail elderly persons. J Am Geriatr Soc 1991;39:142–8. 398
321 muscles: evidence of reciprocal inhibition and facilitation. Exp Brain Res [39] Bialosky JE, et al. The influence of expectation on spinal manipulation induced 399
322 2005;161:457–64. hypoalgesia: an experimental study in normal subjects. BMC Musculoskelet 400
323 [10] Wu CW, Seo H, Cohen LG. Influence of electric somatosensory stimulation on Disord 2008;9:9. 401
324 paretic-hand function in chronic stroke. Arch Phys Med Rehabil 2006;87: [40] Bialosky JE, et al. Placebo response to manual therapy: something out of 402
325 351–7. nothing? J Man Manip Ther 2011;19:11–9. 403

Please cite this article in press as: Sharififar S, et al. Adding electrical stimulation during standard rehabilitation after stroke to improve
motor function. A systematic review and meta-analysis. Ann Phys Rehabil Med (2018), https://doi.org/10.1016/j.rehab.2018.06.005