Transcranial Direct Current Stimulation (tDCS): Does it Have Merit in Stroke Rehabilitation? A Systematic Review

Abstract

Transcranial direct current stimulation has been gaining increasing interest as a potential therapeutic treatment in stroke recovery. We performed a systematic review with meta-analysis of randomized controlled trials to collate the available evidence in adults with residual motor impairments as a result of stroke. The primary outcome was change in motor function or impairment as a result of transcranial direct current stimulation, using any reported electrode montage, with or without adjunct physical therapy. The search yielded 15 relevant studies comprising 315 subjects. Compared with sham, cortical stimulation did not produce statistically significant improvements in motor performance when measured immediately after the intervention (anodal stimulation: facilitation of the affected cortex: standardized mean difference = 0·05, P = 0·71; cathodal stimulation: inhibition of the nonaffected cortex: standardized mean difference = 0·39, P = 0·08; bihemispheric stimulation: standardized mean difference = 0·24, P = 0·39). When the data were analyzed according to stroke characteristics, statistically significant improvements were evident for those with chronic stroke (standardized mean difference = 0·45, P = 0·01) and subjects with mild-to-moderate stroke impairments (standardized mean difference = 0·37, P = 0·02). Transcranial direct current stimulation is likely to be effective in enhancing motor performance in the short term when applied selectively to patients with stroke. Given the range of stimulation variables and heterogeneous nature of stroke, this modality is still experimental and further research is required to determine its clinical merit in stroke rehabilitation.

Keywords

brain electric stimulation intervention rehabilitation stroke therapy

Introduction

Recovery from stroke remains sub-optimal and drives the compelling search for effective methods of stroke rehabilitation that are accessible, safe, and easy to administer. This has led to increasing interest in noninvasive cortical stimulation. Cortical stimulation may have a role in promoting both contralesional and ipsilesional plastic changes after stroke. This is based on the hypothesis that a focal lesion leads to reduced output from the lesioned hemisphere and disrupts the balance of interhemispheric communication. Electrical stimulation may be able to facilitate a shift of this imbalance toward the prestroke equilibrium by downregulating excitability via application to the nonlesioned hemisphere, thus releasing the lesioned hemisphere from excessive transcallosal inhibition (1). Conversely, it may be applied to the lesioned hemisphere to increase the excitability of the perilesional regions (2).

Transcranial direct current stimulation (tDCS) has emerged as one of the primary techniques under investigation. The tDCS stimulating device is a 13 cm × 21 cm portable box, with two rubber electrodes applied with conductive gel or water-soaked pads. Typically, the protocol for tDCS utilizes 1–2 mA of continuous current for a duration of 10–20 mins, with one electrode placed in the region of the motor cortex and the other on the contralateral supraorbital region. At a cost of $8000, the price, ease of application, and small size of the unit render it a practical concurrent therapy option for rehabilitation clinicians. The studies performed in healthy adults to date have consistently shown that cortical activity, including motor function, can be temporarily altered by tDCS, and the effects depend on the polarity and position of the electrodes, whereby brain activity is increased by anodal stimulation and conversely decreased by cathodal stimulation. The tDCS technique has been in existence since the 1960s, with a body of evidence in psychological conditions where reports have indicated it to be a safe technique that is well tolerated by patients (3).

Several narrative reviews have been conducted describing the effects of tDCS in a range of conditions including stroke (4 –10). The purpose of this review was to systematically review the potential of tDCS to enhance the motor recovery of stroke survivors. The research questions were

What are the effects of tDCS on body function or activity limitation in patients with stroke compared with no treatment or standard physical therapy?

Are the effects of tDCS in patients with stroke dependant on stimulation parameters (i.e. anodal vs. cathodal) or patient characteristics (i.e. chronic vs. acute stroke)? and

Is tDCS a safe modality for use in the stroke population?

Method

Identification of studies

A literature search was undertaken to locate all eligible published studies. Electronic searches of MEDLINE, PubMed, CINAHL, and ProQuest were performed using key words for the health condition, stroke, and terms for the intervention, tDCS, in various combinations to locate the relevant papers. There was no methodological filter used for the study design and no time constriction applied to the literature search that was completed in September 2012. In addition, the reference lists of all relevant articles were hand searched for further studies. Duplicates were removed manually. Based on titles and abstracts, the principal author (J. M.) retrieved relevant studies after which two authors (J. M. and P. v. V.) independently evaluated the studies for inclusion in this review. Any uncertainties regarding inclusion were clarified through discussion (see Box 1 for inclusion criteria).

Box. 1

Inclusion criteria.

Design

Randomised or quasi-randomised controlled trials

Participants

Adults >18years

Diagnosis of stroke (haemorrhagic or infarct, any location, acute or chronic, any level of disability)

Intervention

Transcranial direct current stimulation (either polarity, any configuration, single or multiple sessions, alone or as an adjunct to other interventions)

Outcome measures

Impairment or Functional measures (any validated tool of physical function or impairment eg Fugyl-Meyer assessment, Jebsen-Taylor test of hand function, grip strength, reaction time)

Comparisons

tDCS versus sham stimulation

tDCS in addition to other therapy versus other therapy

Anodal tDCS versus cathodal tDCS

Selection of studies

Studies involving adults with a diagnosis of stroke, as defined by the original authors irrespective of lesion location, severity, or classification, were eligible. Included literature was limited to full-text publications in English that utilized a controlled experimental design on human subjects and reported original data. Because the aim of our study was to assess benefits of tDCS in terms of movement performance, studies were excluded if they did not include at least one measure of impairment, body function, or activity limitation.

Two reviewers (J. M. and P. v. V.) worked collaboratively to assess the methodological quality of the studies using the criteria from the Physiotherapy Evidence Database scale (PEDro) (see Table 1 for results). Studies were only included if they used a randomized and controlled design. This included studies with random allocation to treatment groups as well as crossover research design with randomization to treatment order.

Table 1

Included studies: summary of research design

Study	Random allocation	Concealed allocation	Groups similar at baseline	Participant blinding	Therapist blinding	Assessor blinding	<15% dropouts	Intention to treat analysis	Between-group difference reported	Point estimate and variability reported	Total 0–10 (PEDro score)
Fregni et al. (11) 2005	✓	x	✓	✓	x	✓	✓	✓	✓	✓	8
Boggio et al. (1) 2007	✓	x	x	✓	x	✓	✓	✓	✓	✓	7
Kim et al. (12) 2009	✓	x	✓	✓	x	x	✓	✓	✓	✓	7
Celnik et al. (13) 2009	✓	x	✓	✓	x	✓	✓	✓	✓	✓	8
Lindenberg et al. (14) 2010	✓	x	✓	✓	✓	✓	✓	✓	✓	✓	9
Mahmoudi et al. (15) 2011	✓	x	x	✓	x	✓	✓	✓	✓	✓	7
Kim et al. (16) 2010	✓	✓	✓	✓	✓	✓	✓	x	✓	✓	9
Madhavan et al. (17) 2011	✓	x	✓	✓	x	✓	✓	✓	✓	✓	8
Nair et al. (18) 2011	✓	x	✓	✓	✓	✓	✓	✓	✓	✓	9
Geroin et al. (19) 2011	✓	x	✓	x	x	✓	✓	✓	✓	✓	7
Bolognini et al. (20) 2011	✓	x	✓	✓	✓	✓	x	x	✓	✓	7
Hesse et al. (21) 2011	✓	x	✓	✓	✓	✓	✓	✓	✓	✓	9
Zimerman et al. (22) 2012	✓	x	✓	✓	✓	✓	✓	✓	✓	✓	8
Rossi et al. (23) 2012	✓	x	✓	✓	✓	✓	✓	✓	✓	✓	9
Stagg et al. (24) 2012	✓	x	✓	✓	x	x	✓	✓	✓	✓	7

All included studies specified eligibility criteria.

Interventions

Several different applications of tDCS to modulate cortical excitability are described in the literature and we did not restrict our inclusion of studies based on this factor. For the purposes of this review, the following descriptions apply:

Anodal tDCS: The active electrode (anode) is positioned over the lesioned (stroke) motor cortex and the cathode (reference) over the contralesional supraorbital region, with the intent of upregulating the excitability of the perilesional regions.

Cathodal tDCS: The cathode is positioned over the contralesional motor cortex and the anode over the ipsilesional supraorbital region, with the intent of downregulating the excitability of the contralesional cortex.

Bihemispheric tDCS: The anode is positioned over the lesioned motor cortex and the cathode over the contralesional cortex, with the intent of simultaneously upregulating the excitability of the perilesional cortex and downregulating the excitability of the contralesional cortex.

Extracephalic tDCS: The anode is positioned over the lesioned motor cortex and the cathode over the contralesional deltoid muscle, with the intent of upregulating the excitability of the perilesional regions.

Sham tDCS: The electrodes are positioned as per anodal stimulation.

In each of these montages, the intensity of the stimulation is steadily increased to the selected threshold over a period of 30 s at the commencement of the intervention and similarly decreased over 30 s at the conclusion of the intervention. In the situation of sham tDCS, the stimulation is ramped up to give the patient the initial tingling sensation and then turned off without the subject being aware that the stimulation has ceased (25).

Outcome measures

The primary outcome measures were motor impairment, body function, and activity limitation. Typical measures of these outcomes include simple reaction time (SRT), Motor Assessment Scale, Jebsen-Taylor Test of Hand Function (JTT), Fugl-Meyer Assessment (FM), and Wolf Motor Function Test. Included studies were also examined for reports of adverse events such as discomfort, fatigue, and headache.

Data analysis

Details from each study including sample size, participant characteristics, tDCS parameters, and reported findings were extracted using a standardized form. The results extracted relate to outcomes measured immediately following the application of tDCS and not longer term effects due to either lack of follow-up data in the majority of studies or large variation in follow-up periods post intervention. Our results therefore indicate the presence or absence of immediate responses to tDCS and do not address change in long-term performance. Authors were contacted where there was difficulty extracting the data from the published paper.

Where means and standard deviation values were provided for pre-post intervention conditions, the standardized mean difference (SMD) was calculated. This allowed us to convert all outcomes to a common scale to compare studies that used different tools to measure the same outcome. We followed general practice to interpret a value of 0·2 to indicate a small effect, and 0·8 a large effect (26). Changes from baseline were used as the primary outcome. In the case where a decreased score on the assessment tool used in the original research represented an improvement, the positive form of the difference score was used to allow for comparison across the different scales. A meta-analysis was then conducted to obtain the average effect of the tDCS interventions and to compare the effects against sham intervention. Intertrial heterogeneity was quantified using I² (27). Trials in the meta-analysis were considered to have low statistical heterogeneity if I² was equal or less than 25% (28), in which case a fixed-effect model was used. If I² was greater than 25%, a random effects model was used to incorporate intertrial heterogeneity (27).

We calculated seven separate meta-analyses. First, we analyzed the effects of different types of tDCS: anodal, cathodal, and bihemispheric stimulation compared with sham stimulation. Then, we analyzed the effect of tDCS on different patient subgroups: those with chronic stroke and those with sub-acute stroke. According to customary convention, we defined acute stroke as within the first three-days of symptom onset, sub-acute stroke as less than three-months, and chronic stroke as greater than three-months since the initial symptoms. Finally, we pooled the data from studies that included subjects with mild-to-moderate impairments and those with moderate-to-severe impairments. As no standard definition of these categories exists, we used the definitions and criteria provided by the original authors to determine groupings. We were unable to include the results of two studies in any of the meta-analyses due to insufficient data (17,24).

Results

Flow of studies through the review

The search resulted in the identification of 289 articles after the removal of duplicates. Of these 35 were deemed potentially relevant. On closer scrutiny, a further 20 were excluded due to research design (11 studies failed to meet the randomization criteria); no inclusion of an impairment, function, or activity limitation measure (5 studies); insufficient information (1 study); and Russian or Chinese text only (3 studies). This left 15 studies for inclusion in the review (see Fig. 1 for the flow of studies through the review).

Fig. 1

Flow of studies through the review. *Papers may have been excluded for failing to meet more than one inclusion criteria. RCT, randomized controlled trial.

Characteristics of studies

Quality

The methodological quality of the 15 studies meeting all inclusion criteria was consistently high with a mean PEDro score of 7·9 out of 10 [standard deviation (SD) = 0·9, range 7–9]. All studies used randomization; however, only one reported concealed allocation (16). Each study, with the exception of one (19), used participant blinding; all but one (12) blinded the assessors, but only 6 of the 15 studies reported therapist blinding. All studies had excellent retention rates with only two studies (16,20) reporting dropouts, but this remained less than 15% of each sample.

Participants

The mean age of the 315 participants across the studies was 59·3 years, with a range from 28 to 87 years and a preponderance of men (61%). Time since stroke varied with 10 of 15 studies recruiting participants with chronic stroke, 4 studies recruiting sub-acute subjects, and 1 study recruiting a mixed sample of chronic and sub-acute subjects. However, the majority of participants were less than 12 months post stroke (58%). Both cortical and sub-cortical strokes were included in the samples, but cortical strokes predominated (cortical n = 137, sub-cortical n = 110, both n = 68).

Several different scales were used to classify stroke severity. This included grip strength, upper limb score of the Fugl- Meyer, and the ability to perform all items on the JTT. According to the classifications provided by the original authors, 11 studies included participants with mild/moderately affected participants, and only 4 studies recruited subjects with moderate/severe impairments.

Intervention

The majority of studies investigated the effects of anodal stimulation compared with the sham condition (11 studies). In addition, many studies analyzed the effect of several different applications of tDCS including cathodal stimulation (eight studies), bihemispheric stimulation (three studies), and extracephalic stimulation (one study). All studies used a sham control condition. Stimulation intensity was most typically 1 mA (10 studies), with a range of 0·5–2·0 mA and stimulation duration ranged from 7 to 40 mins. Seven studies reported the effects of tDCS following a single session of stimulation, others trialed weekly sessions or consecutive daily sessions. Concurrent physical therapy or training during stimulation was administered in nine of the studies.

Outcome measures

Several standardized tools were used to assess body function and activity limitation. This included the JTT (four studies), the FM (five studies), SRT tasks (four studies), six-minute walk test (one study), and the Box and Block Test (one study). Time-points for data collection were not consistent across the trials. Most studies provide outcomes immediately following the stimulation period, with three exceptions where there was a delay of up to seven-days after the cessation of the intervention prior to reassessment (14,16,18). Four studies collected short-term follow-up data out to at least seven-days post intervention, and only three studies collected long-term data out to at least three-months (16,21,23). Patient tolerance of the intervention was reported in all studies, and discomfort, fatigue, and attention data were formally collected via visual analog scales or questionnaires in five studies (11 –13,22,24). See Table 2 for a summary of the included studies.

Table 2

Included studies: Summary

Study	Design	Participants	Stimulation protocol	Intervention description	Outcome measures
Fregni et al. (11) 2005	Crossover, sham controlled, double blind Randomized, counterbalanced	n = 6 Mean age: 537 Range: 28–75 Chronic Mixed cortical and sub-cortical Mild-to-moderate deficits	Cathodal and anodal 1 mA/35 cm² 20 mins	Single session of anodal tDCS, cathodal tDCS, and sham separated by ≈ 48 h	JTT
Boggio et al. (1) 2007	Study 1: Crossover, sham controlled, double blind Study 2: Open-label study	n = 4 Mean age: 57·4 Range: 38–75 Chronic Sub-cortical lesions Mild deficits n = 5 Chronic Sub-cortical lesions Mild deficits	Anodal and cathodal 1 mA/35 cm² 20 mins Cathodal 1 mA/35 cm² 20 mins	One session weekly for four-weeks Five consecutive daily sessions	JTT
Kim et al. (12) 2009	Single (patient) blinded, sham controlled, counterbalanced, randomized	n = 10 Mean age: 62·8 Range: 28–75 Sub-acute Sub-cortical (one cortical) Mild deficits	Anodal 1 mA/25 cm² 20 mins	Single sessions of tDCS and sham separated by ≈ 24 h	BBT Finger acceleration
Celnik et al. (13) 2009	Crossover, sham controlled, double blind Randomized	n = 9 Mean age: 55·3 Range: 40–73 Chronic Mixed cortical and sub-cortical lesions Mild deficits	Anodal 1 mA/57 cm² 20 mins PNS median and ulnar nerve of paretic hand (1 hz for two-hours)	PNS + tDCS PNS + tDCS_sham tDCS + PNS_sham PNS_sham + tDCS_sham Sessions separated by ≈ six-days 28 mins of finger task practice followed by stimulation	Finger sequence task
Kim et al. (16) 2010	Double blind, sham controlled RCT with stratified randomization and long-term follow-up	n = 18 Mean age: 57·2 Range: 34–77 Sub-acute Mixed cortical and sub-cortical lesions Mild-to- moderate deficits	Cathodal and anodal 1 mA/25 cm² 20 mins	Ten sessions five times per week over two-weeks combined with concurrent 30 mins OT	FM (UE) BI
Lindenberg et al. (14) 2010	Crossover, sham controlled, double blind Block randomization and stratification for impairment	n = 20 Mean age: 56·2 Range: 34–77 Chronic MCA territory infarct Mild-to-moderate deficits	Bihemispheric 1·5 mA/16·3 cm² 30 mins	Five consecutive daily sessions of either bihemispheric tDCS or sham tDCS combined with 60 mins concurrent OT	FM (UE) WMFT
Mahmoudi et al. (15) 2011	Crossover, sham controlled, double blind Counterbalanced and randomized to order	n = 10 Mean age: 60·8 Range: 40–87 Mixed chronic and sub-acute Mixed cortical and sub-cortical lesions Mild-to-moderate deficits	1. Bihemispheric anodal and cathodal 2. Anodal 3. Cathodal 4. Extracephalic tDCS 5. Sham 1 mA/35 cm² 20 mins	Single sessions separated by >96 h	JTT
Madhavan et al. (17) 2011	Crossover, sham controlled, double blind Randomized	n = 9 Mean age: 65·4 Range: 50–87 Chronic Sub-cortical except one cortical Mild-to-moderate deficits	Anodal tDCS and anodal over nonlesioned cortex 0·5 mA/8 cm²	3 × 15 mins sessions separated by 48 h Concurrent practice of ankle-tracking task using paretic ankle	Tracking error in visuomotor ankle-tracking task FM (LE)
Nair et al. (18) 2011	Sham controlled, double blind, randomized	n = 14 Mean age: 55·8 Range: 40–76 Chronic Mixed: sub-cortical/cortical Moderate-to-severe deficits	Cathodal tDCS 1 mA 30 mins	Five-days/consecutive sessions Concurrent OT 60 mins	ROM FM (UE)
Geroin et al. (19) 2011	Sham controlled, single blind, randomized	n = 30 Mean age: 62·7 Range: 51–73 Chronic Mixed: sub-cortical/cortical Mild-to-moderate deficits	Anodal tDCS 1 mA/35 cm² Seven-minutes	Ten sessions Five times per week for two-weeks Concurrent robotic gait training	6MWT 10MWT
Bolognini et al. (20) 2011	Sham controlled, double blind, randomized	n = 14 Mean age: 46·7 Range: 26–75 Chronic Mixed: sub-cortical/cortical Moderate-to-severe deficits	Bihemispheric 2 mA/35 cm² 40 mins	Ten sessions Five times per week for two-weeks Concurrent CIMT: 90% of waking hours + four-hours shaping	JTT Grip strength FM (UE)
Hesse et al. (21) 2011	Sham controlled, double blind, randomized multicenter	n = 96 Mean age: 64·9 Range: 39–79 Sub-acute Mixed: sub-cortical/cortical Severe deficits	Cathodal and anodal 2 mA/35 cm² 20 mins	Thirty sessions Five times per week for six-weeks Concurrent robot training and regular therapy	FM (UE) Grip strength BI
Zimerman et al. (22) 2012	Sham controlled, double blind, crossover pseudo-randomized order	n = 12 Mean age: 58·3 Range: 31–73 Chronic Sub-cortical/cortical Mild deficits	Cathodal tDCS 1 mA/25 cm² 20 mins	Concurrent bilateral motor sequence training 15–20 mins Separated by one-week	Finger sequence task
Rossi et al. (23) 2012	Sham controlled, double blind, randomized	n = 50 Mean age: 68·2 Range: 30–80 Acute Mixed sub-cortical/cortical Moderate-to-severe deficits	Anodal tDCS 2 mA/35 cm² 20 mins	Five sessions Five times per week for one-week	FM (UE) BI
Stagg et al. (24) 2012	Sham controlled, single blind, randomized	n = 13 Mean age: 64 Range: 30–80 Chronic Sub-cortical: seven; cortical: six Mild-to-moderate deficits	Cathodal and anodal 1 mA/35 cm² 20 mins	Single sessions separated by one-week Concurrent blocks of visually cued response time task and grip strength task	Grip strength Response time

6MWT, six-minute walk test; 10MWT, 10–meter walk test; BBT, Box and Block Test; BI, Barthel Index; CIMT, constraint induced movement therapy; FM, Fugl-Meyer Test; JTT, Jebsen-Taylor Test of Hand Function; LE, lower extremity component; MCA, middle cerebral artery; OT, Occupational Therapy; PNS, peripheral nerve stimulation; PNSsham, PNS delivered to the deep peroneal and posterior peroneal nerves for two-hours; RCT, randomized controlled trial; ROM, range of joint motion; tDCS, transcranial direct current stimulation; UE, upper extremity component; WMFT, Wolf Motor Function Test.

Effect of tDCS

Physical function

The effect of anodal tDCS on motor performance was examined by pooling the data from nine studies involving 224 subjects. When compared with sham controls, anodal tDCS did not significantly alter motor performance [SMD = 0·05, confidence interval (CI) =–0·25–0·31, P = 0·71, see Fig. 2]. This finding was similar for cathodal stimulation when the data from seven studies involving 154 subjects were pooled (SMD = 0·39, CI =–0·05–0·82, P = 0·08, see Fig. 3) and bihemispheric stimulation using the data from three studies involving 54 subjects (SMD = 0·24, CI =–0·3–0·77, P = 0·39, see Fig. 4). Only one author investigated the use of an extracephalic reference electrode and report a nonsignificant effect relative to sham (P = 0·82) (15). Two studies conducted follow-up assessments three-months after the intervention and both report no between-group differences (P> 0·05) (21,23). Kim et al. (16) were the only authors to report long-term outcomes that were measured six-months after the intervention. Functional performance was significantly better six-months post cathodal stimulation compared with sham (P < 0·05), whereas anodal stimulation showed a trend toward improvement relative to sham but this did not reach statistical significance at six-months follow-up.

Fig. 2

SMD (95% CI) of the effect of anodal stimulation on motor performance compared with sham by pooling data from nine studies (n = 224). CI, confidence interval; SMD, standardized mean difference.

Fig. 3

SMD (95% CI) of the effect of cathodal stimulation on motor performance compared with sham by pooling data from seven studies (n = 154). CI, confidence interval; SMD, standardized mean difference.

Fig. 4

SMD (95% CI) of the effect of bihemispheric stimulation on motor performance compared with sham by pooling data from three studies (n = 54). CI, confidence interval; SMD, standardized mean difference.

Further analysis was conducted by comparing acute, sub-acute, and chronic stroke samples. Only one study investigated subjects with acute stroke (23). These authors report a nonsignificant effect associated with tDCS. The effect of tDCS on motor performance in people with chronic stroke was evaluated by pooling the data from eight studies involving 130 subjects. When compared with sham controls, tDCS significantly improved performance (SMD = 0·45, CI = 0·09–0·80, P = 0·001, see figure 5). This positive finding was not replicated when we pooled the data from the three studies (n = 49) that used sub-acute stroke samples (SMD = 0·01, CI = −0·39–0·42, P = 0·94, see figure 6). The final analyses revealed a statistically significant benefit of tDCS in nine studies involving 155 subjects who demonstrated mild/moderate impairment (SMD = 0·37, CI = 0·05–0·70, P = 0·02, see figure 7), but not those classified with moderate/severe impairment in four studies with 141 subjects (SMD = −0·05, CI = −0·38–0·28, P = 0·78), see figure 8.

Fig. 5

SMD (95% CI) of the effects of tDCS on motor performance of people with chronic stroke compared with sham by pooling data from eight studies (n = 130). CI, confidence interval; SMD, standardized mean difference; tDCS, transcranial direct current stimulation.

Fig. 6

SMD (95% CI) of effects of tDCS on motor performance of people with sub-acute stroke compared with sham by pooling data from three studies (n = 49). CI, confidence interval; SMD, standardized mean difference; tDCS, transcranial direct current stimulation.

Fig. 7

SMD (95% CI) of the effects of tDCS on motor performance of people with mild/moderate stroke impairments compared with sham by pooling data from nine studies (n = 155). CI, confidence interval; SMD, standardized mean difference; tDCS, transcranial direct current stimulation.

Fig. 8

SMD (95% CI) of the effects of tDCS on motor performance of people with moderate/severe stroke impairments compared with sham by pooling data from four studies (n = 141). CI, confidence interval; SMD, standardized mean difference; tDCS, transcranial direct current stimulation.

Safety

The absence of adverse events was recorded in 14 of the 15 studies; the remaining study (16) recorded two adverse episodes, whereby one participant experienced a headache (during anodal stimulation) and another dizziness (during cathodal stimulation). These symptoms lead the participants to discontinue their involvement in the study. There was no significant effect of stimulation on attention or fatigue. (See Table 3 for a summary of the findings of individual studies).

Table 3

Included studies: Summary of findings

Study	Outcome measures	Findings (mean % improvement relative to sham)
Fregni et al. (11) 2005	JTT	Cathodal 15·3% (P < 0·05) No significant difference between anodal and cathodal	Anodal 10·7% (P < 0·05)
Boggio et al. (1) 2007	JTT	Cathodal 9·5% (P= 0·016) No significant difference between anodal and cathodal (P = 0·56)	Anodal 7·3% (P= 0·046)
Kim et al. (12) 2009	BBT	Anodal: 17·8% (P < 0·05)
Celnik et al. (13) 2009	Finger sequence task (accuracy + speed)	Anodal: 18·6% (P < 0·05)	Anodal + peripheral nerve stimulation = 41·3% (P < 0·05)
Kim et al. (16) 2010	FM (UE)	Anodal: 27·3%No significant difference between groups at one-day post interventionCathodal better than sham at six-months in FM measure (P < 0·05)	Cathodal: 16·3%
Lindenberg et al. (14) 2010	FM (UE)	Bihemispheric: 11·7% (P < 0·001)
Mahmoudi et al. (15) 2011	JTT	Bihemispheric 13·9% (P = 0·011)No significant difference between these three groupsExtracephalic 2·4% (P= 0·82)	Anodal 9·3% (P = 0·016)Cathodal 6·8% (P = 0·010)
Madhavan et al. (17) 2011	Tracking error in visuomotor ankle-tracking task	Anodal = 10·8% (P= 0·0001)
Nair et al. (18) 2011	FM (UE)	Cathodal:FM: 7·5% (P= 0·048)ROM: 13·3% (P= 0·002)
Geroin et al. (19) 2011	6MWT10MWT	Anodal:6MWT: 9% (P= 0·14)10 min walk test: 3·4% (P = 0·32)
Bolognini et al. (20) 2011	JTTFM (UE)HS	Bihemispheric:JTT: 19% (P < 0·01)FM: 16·2% (P < 0·03)HS: 41·6% (P < 0·01)
Hesse et al. (21) 2011	FM (UE)HS	Anodal:FM: 10·9%HS: −1%No significant difference between these two groups or sham	Cathodal3·9%1·4%
Zimerman et al. (22) 2012	Correct finger sequences	Anodal:35·7% (P = 0·04)
Rossi et al. (23) 2012	FM	Anodal:FM: −25·8% (P= 0·82)
Stagg et al. (24) 2012	Response time	Anodal:12·6% (P = 0·002)	Cathodal:7·9%(P=0·04)

6MWT, six-minute walk test; 10MWT, ten-meter walk test; BBT, Box and Block Test; FM, Fugl-Meyer Test; JTT, Jebsen-Taylor Test of Hand Function; HS, handgrip strength; ROM, range of joint motion; UE, upper extremity component.

Discussion

The results of this systematic review provide evidence from 15 studies with relatively high methodological quality in support of tDCS when applied to selected patients with stroke. These positive findings are not consistent across all the included studies, possibly due to the heterogeneity of the participant characteristics and stimulation paradigms. Those most likely to benefit are patients with chronic stroke and/or mild-to-moderate motor impairments. Likewise, the size of the treatment effect is variable and at best modest with a maximum effect size of 35·7% improvement relative to sham.

Based on the current findings, we are unable to decipher if one form of tDCS is superior to another. Pooled data demonstrate a lesser treatment effect for the anodal stimulation montage (anodal SMD = 0·05; cathodal SMD = 0·39; bihemispheric SMD = 0·39), and while several individual studies demonstrate greater treatment effect following cathodal stimulation compared with anodal (1,11), these differences fail to reach statistical significance and others report the effects of cathodal stimulation to be significantly weaker than anodal (15). Although cathodal, anodal, and bihemispheric stimulation appear to have merit, there is currently no evidence to support the use of extracephalic stimulation paradigms in patients with stroke (15). Furthermore, the optimal number of tDCS sessions and session duration has not yet been well defined with conflicting reports in the literature. It has been proposed that repetition of tDCS in consecutive sessions can enhance the efficacy of the stimulation by cumulating or stabilizing the effects. Several authors have demonstrated the positive effects of tDCS enduring beyond the intervention period by one-week (14,18,20); however, this is disputed by other authors who utilized daily tDCS stimulation ranging from one- to six-weeks with no reported benefit (19,21,23). The ideal number and timing of sessions and the sustainability of the effects remain undetermined and requires long-term prospective investigation.

There is consensus that for motor improvements to be lasting tDCS must occur in conjunction with training (29). This may enhance skill acquisition by increasing afferent inputs to the cortex while its intrinsic excitability is being enhanced by tDCS. Transcranial direct current stimulation has been shown to beneficially enhance the effects of peripheral nerve stimulation (13), constraint-induced movement therapy (20), and occupational therapy training (14,16,18). In contrast, Rossi et al. (23) were unable to show any additional benefit of tDCS when supplementing routine therapy in acute stroke patients. Likewise, there is no evidence to support tDCS as an adjuvant to bilateral robot assisted limb training in either the upper limb (21) or the lower limb (19). These conflicting findings suggest that several factors may influence the outcome of the combined treatment approach and may include the temporal delivery of tDCS in relation to the training as well as the type of training.

Stroke is a heterogeneous disease affecting a diverse population. The establishment of participant selection criteria based on lesion location, time after stroke, and/or integrity of the corticospinal pathway may assist in determining which patients are most likely to benefit from tDCS. Corticospinal excitability in the sub-acute stages of stroke recovery may be different from that in the chronic stages, and therefore neuromodulatory agent such as tDCS may have differential effects. This is supported by our meta-analysis whereby there was a difference in the size of the treatment effect when the sample was composed of people with chronic stroke (SMD = 0·45) compared with sub-acute stroke (SMD = 0·01). However, Mahmoudi et al. (15) dispute this premise and found no correlation between time post stroke and outcome in their randomized controlled trial (P > 0·02). Furthermore, cautious interpretation is required as the positive findings in chronic stroke have predominantly been reported in mild to moderately affected patients who were able to grasp and manipulate objects with the affected hand. In contrast, the acute/sub-acute samples often included participants with severe deficits. Similarly, cortical excitability and current flow is likely to differ according to stroke location and lesion volume. Several authors support this theory reporting that patients with subcortical lesions had significantly greater improvements than those with cortical lesions (15,21). Indeed, there could be merit in defining damaged brain areas and networks more specifically through imaging in future studies to determine the presence of differential responses to tDCS according to lesion characteristics. It has been proposed that tDCS may be effective for a wide range of stroke types and motor deficits as long as part of the corticospinal output is preserved (16). Future research may need to consider matching controls for lesion characteristics as at this stage we are unable to conclusively determine if the effects of tDCS are independent of these factors.

This review indicates that tDCS is well tolerated by patients with only two documented dropouts in all the reported studies. These dropouts occurred following adverse events in the study with the highest dosage of tDCS (16). Although headache and dizziness are relatively minor symptoms, it may suggest that in sub-acute stroke 10 sessions of 2 mA direct current is reaching the threshold of patient tolerance. This highlights the need for the determination of stimulation parameters that maximize benefit but limit the associated risks.

The strength of this review is that it included functionally relevant outcome measures related to motor performance, which is often the major clinical focus of stroke rehabilitation. It presents an impartial synthesis of all the high-quality English publications in the field that provides a more robust estimate of the likely effect of tDCS than individual studies alone. Although 15 studies were identified by this review, the total number of participants was small (n = 315) and further caution with interpretation of these findings is warranted given the likelihood of publication bias toward positive findings. While this data confirm that tDCS has the potential to improve motor performance in patients with stroke, long-term effects on motor function remain unclear as only one study evaluated effects after six-months.

It is becoming increasingly apparent that the best intervention for stroke recovery will incorporate a combination of techniques to maximize neuronal plasticity (29). Research investigating a possible role for tDCS is mounting but remains inconclusive and hampered by both the heterogeneity of the patient and stimulation characteristics. It appears that in order to maximize the potential of this modality, prudent selection of patients and stimulation parameters will be required. Factors such as lesion patterns, severity of paresis, time course post stroke, and the type of adjuvant therapy are methodological issues that require further attention.

References

Boggio

Nunes

Rigonatti

. Repeated sessions of non-invasive brain DC stimulation is associated with motor function improvement in stroke patients. Restor Neurol Neurosci 2007; 25:123–9.

Ward

Cohen

. Mechanisms underlying recovery of motor function after stroke. Arch Neurol 2004; 61:1844–8.

Redfearn

JWT

Lippold

Constain

. A preliminary account of the clinical effects of polarizing the brain in certain psychiatric disorders. Br J Psychiatry 1964; 110:773–85.

Schlaug

Renga

. Transcranial direct current stimulation: a noninvasive tool to facilitate stroke recovery. Expert Rev Med Devices 2008; 5:759–68.

Schlaug

Renga

Nair

. Transcranial direct current stimulation in stroke recovery. Arch Neurol 2008; 65:1571–6.

Hummel

Cohen

. Non-invasive brain stimulation: a new strategy to improve neurorehabilitation after stroke? Lancet 2006; 5:708–12.

Harvey

Nudo

. Cortical brain stimulation: a potential therapeutic agent for upper limb motor recovery following stroke. Top Stroke Rehabil 2007; 14:54–67.

Bolognini

Pascual-Leone

Fregni

. Using non-invasive brain stimulation to augment motor training-induced plasticity. J Neuro Eng Rehabil 2009; 6:8, doi: 10.1186/1743-0003-6-8.

Alonso-Alonso

Fregni

Pascual-Leone

. Brain stimulation in poststroke rehabilitation. Cerebrovasc Dis 2007; 24:157–66.

10.

Williams

Imamura

Fregni

. Updates on the use of non-invasive brain stimulation in physical and rehabilitation medicine. J Rehabilit Med 2009; 41:305–11.

11.

Fregni

Boggio

Mansur

. Transcranial direct current stimulation of the unaffected hemisphere in stroke patients. Neuroreport 2005; 16:1551–5.

12.

Kim

Ohn

Yang

Park

C-I

Jung

. Enhancing motor performance by anodal transcranial direct current stimulation in sub-acute stroke patients. Am J Phys Med Rehabil 2009; 88:829–36.

13.

Celnik

Paik

Vandermeeren

Dimyan

Cohen

. Effects of combined peripheral nerve stimulation and brain polarization on performance of a motor sequence task after chronic stroke. Stroke 2009; 40:1764–71.

14.

Lindenberg

Renga

Zhu

Nair

Schlaug

. Bihemispheric brain stimulation facilitates motor recovery in chronic stroke patients. Neurology 2010; 75:2176–84.

15.

Mahmoudi

Haghighi

Petremfar

Jahanshani

Salehi

Fregni

. Transcranial direct current stimulation: electrode montage in stroke. Disabil Rehabil 2011; 33:1383–8.

16.

Kim

Lim

Kang

. Effect of transcranial direct current stimulation on motor recovery in patients with subacute stroke. Am J Phys Med Rehabil 2010; 89:879–86.

17.

Madhavan

Weber

Stinear

. Non-invasive brain stimulation enhances fine motor control of the hemiparetic ankle: implications for rehabilitation. Exp Brain Res 2011; 209:9–17.

18.

Nair

Renga

Lindenberg

Zhu

Schlaug

. Optimizing recovery potential through simultaneous occupational therapy and non-invasive brain stimulation using tDCS. Restor Neurol Neurosci 2011; 29:411–20.

19.

Geroin

Picelli

Munari

Waldner

Tomelleri

Smania

. Combined transcranial direct current stimulation and robot-assisted gair training in patients with chronic stroke: a preliminary comparison. Clin Rehabil 2011; 25:537–48.

20.

Bolognini

Vallar

Casati

. Neurophysiological and behavioural effects of tDCS combined with constraint-induced movement therapy in poststroke patients. Neurorehabil Neural Repair 2011; 25:819–29.

21.

Hesse

Waldner

Merholz

Tomelleri

Pohl

Werner

. Combined transcranial direct current stimulation and robot-assisted arm training in subacute stroke patients: an exploratory, randomized multicenter trial. Neurorehabil Neural Repair 2011; 25:838–46.

22.

Zimerman

Heise

Hoppe

Cohen

Gerloff

Hummel

. Modulation of training by single-session transcranial direct current stimulation to the intact motor cortex enhances motor skill acquisition of the paretic hand. Stroke 2012; 43:2185–91.

23.

Rossi

Sallustio

Di Legge

Stanzione

Koch

. Transcranial direct current stimulation of the affected hemisphere does not accelerate recovery of acute stroke patients. Eur J Neurol 2012; 20:202–4.

24.

Stagg

Bachtiar

O'shea

. Cortical activation changes underlying stimulation-induced behavioural gains in chronic stroke. Brain 2012; 135:276–84.

25.

Gandiga

Hummel

Cohen

. Transcranial DC stimulation (tDCS): a tool for double-blind sham controlled clinical studies in brain stimulation. Clin Neurophysiol 2006; 117:845–50.

26.

Valentine

Cooper

. Effect size substantive interpretation guidelines: issues in the interpretation of effect sizes. 2003. 24th April 2012.

27.

Higgins

JPT

Green

. Cochrane Handbook for Systematic Reviews of Interventions. Chichester, UK, John Wiley & Sons Ltd, 2006.

28.

Higgins

JPT

Thompson

Deeks

Altman

. Measuring inconsistency in meta-analysis. BMJ 2003; 327:557–60.

29.

Hummel

Celnik

Pascual-Leone

. Controversy: noninvasive and invasive cortical stimulation show efficacy in treating stroke patients. Brain Stimul 2008; 1:370–82.