Transcranial direct current stimulation (tDCS) for improving activities of daily living, and physical and cognitive functioning, in people after stroke

Bernhard Elsner; Joachim Kugler; Marcus Pohl; Jan Mehrholz

doi:10.1002/14651858.CD009645.pub4

. 2020 Nov 11;2020(11):CD009645. doi: 10.1002/14651858.CD009645.pub4

Transcranial direct current stimulation (tDCS) for improving activities of daily living, and physical and cognitive functioning, in people after stroke

Bernhard Elsner ^1,^4,^✉, Joachim Kugler ², Marcus Pohl ³, Jan Mehrholz ²

Editor: Cochrane Stroke Group

PMCID: PMC8095012 PMID: 33175411

Abstract

Background

Stroke is one of the leading causes of disability worldwide. Functional impairment, resulting in poor performance in activities of daily living (ADL) among stroke survivors is common. Current rehabilitation approaches have limited effectiveness in improving ADL performance, function, muscle strength, and cognitive abilities (including spatial neglect) after stroke, with improving cognition being the number one research priority in this field. A possible adjunct to stroke rehabilitation might be non‐invasive brain stimulation by transcranial direct current stimulation (tDCS) to modulate cortical excitability, and hence to improve these outcomes in people after stroke.

Objectives

To assess the effects of tDCS on ADL, arm and leg function, muscle strength and cognitive abilities (including spatial neglect), dropouts and adverse events in people after stroke.

Search methods

We searched the Cochrane Stroke Group Trials Register, CENTRAL, MEDLINE, Embase and seven other databases in January 2019. In an effort to identify further published, unpublished, and ongoing trials, we also searched trials registers and reference lists, handsearched conference proceedings, and contacted authors and equipment manufacturers.

Selection criteria

This is the update of an existing review. In the previous version of this review, we focused on the effects of tDCS on ADL and function. In this update, we broadened our inclusion criteria to compare any kind of active tDCS for improving ADL, function, muscle strength and cognitive abilities (including spatial neglect) versus any kind of placebo or control intervention.

Data collection and analysis

Two review authors independently assessed trial quality and risk of bias, extracted data, and applied GRADE criteria. If necessary, we contacted study authors to ask for additional information. We collected information on dropouts and adverse events from the trial reports.

Main results

We included 67 studies involving a total of 1729 patients after stroke. We also identified 116 ongoing studies. The risk of bias did not differ substantially for different comparisons and outcomes. The majority of participants had ischaemic stroke, with mean age between 43 and 75 years, in the acute, postacute, and chronic phase after stroke, and level of impairment ranged from severe to less severe. Included studies differed in terms of type, location and duration of stimulation, amount of current delivered, electrode size and positioning, as well as type and location of stroke.

We found 23 studies with 781 participants examining the effects of tDCS versus sham tDCS (or any other passive intervention) on our primary outcome measure, ADL after stroke. Nineteen studies with 686 participants reported absolute values and showed evidence of effect regarding ADL performance at the end of the intervention period (standardised mean difference (SMD) 0.28, 95% confidence interval (CI) 0.13 to 0.44; random‐effects model; moderate‐quality evidence). Four studies with 95 participants reported change scores, and showed an effect (SMD 0.48, 95% CI 0.02 to 0.95; moderate‐quality evidence). Six studies with 269 participants assessed the effects of tDCS on ADL at the end of follow‐up and provided absolute values, and found improved ADL (SMD 0.31, 95% CI 0.01 to 0.62; moderate‐quality evidence). One study with 16 participants provided change scores and found no effect (SMD ‐0.64, 95% CI ‐1.66 to 0.37; low‐quality evidence). However, the results did not persist in a sensitivity analysis that included only trials with proper allocation concealment.

Thirty‐four trials with a total of 985 participants measured upper extremity function at the end of the intervention period. Twenty‐four studies with 792 participants that presented absolute values found no effect in favour of tDCS (SMD 0.17, 95% CI ‐0.05 to 0.38; moderate‐quality evidence). Ten studies with 193 participants that presented change values also found no effect (SMD 0.33, 95% CI ‐0.12 to 0.79; low‐quality evidence). Regarding the effects of tDCS on upper extremity function at the end of follow‐up, we identified five studies with a total of 211 participants (absolute values) without an effect (SMD ‐0.00, 95% CI ‐0.39 to 0.39; moderate‐quality evidence). Three studies with 72 participants presenting change scores found an effect (SMD 1.07; 95% CI 0.04 to 2.11; low‐quality evidence). Twelve studies with 258 participants reported outcome data for lower extremity function and 18 studies with 553 participants reported outcome data on muscle strength at the end of the intervention period, but there was no effect (high‐quality evidence). Three studies with 156 participants reported outcome data on muscle strength at follow‐up, but there was no evidence of an effect (moderate‐quality evidence). Two studies with 56 participants found no evidence of effect of tDCS on cognitive abilities (low‐quality evidence), but one study with 30 participants found evidence of effect of tDCS for improving spatial neglect (very low‐quality evidence). In 47 studies with 1330 participants, the proportions of dropouts and adverse events were comparable between groups (risk ratio (RR) 1.25, 95% CI 0.74 to 2.13; random‐effects model; moderate‐quality evidence).

Authors' conclusions

There is evidence of very low to moderate quality on the effectiveness of tDCS versus control (sham intervention or any other intervention) for improving ADL outcomes after stroke. However, the results did not persist in a sensitivity analyses including only trials with proper allocation concealment. Evidence of low to high quality suggests that there is no effect of tDCS on arm function and leg function, muscle strength, and cognitive abilities in people after stroke. Evidence of very low quality suggests that there is an effect on hemispatial neglect. There was moderate‐quality evidence that adverse events and numbers of people discontinuing the treatment are not increased. Future studies should particularly engage with patients who may benefit the most from tDCS after stroke, but also should investigate the effects in routine application. Therefore, further large‐scale randomised controlled trials with a parallel‐group design and sample size estimation for tDCS are needed.

Keywords: Adult, Aged, Female, Humans, Male, Middle Aged, Activities of Daily Living, Bias, Cognition Disorders, Cognition Disorders/rehabilitation, Confidence Intervals, Lower Extremity, Lower Extremity/physiology, Motor Activity, Motor Activity/physiology, Muscle Strength, Patient Dropouts, Patient Dropouts/statistics & numerical data, Perceptual Disorders, Perceptual Disorders/rehabilitation, Randomized Controlled Trials as Topic, Recovery of Function, Stroke Rehabilitation, Transcranial Direct Current Stimulation, Upper Extremity, Upper Extremity/physiology

Plain language summary

Direct electrical current to the brain to improve rehabilitation outcomes

Review question

We reviewed the evidence about the effect of direct electrical current to the brain (transcranial direct current stimulation, tDCS) to reduce impairment in activities of daily living (ADL), arm and leg function, muscle strength and cognitive abilities (including spatial neglect), dropouts and adverse events in people after stroke.

Background

Stroke is one of the leading causes of disability worldwide. Most strokes take place when a blood clot blocks a blood vessel leading to the brain. Without a proper blood supply, the brain quickly suffers damage, which can be permanent. This damage often causes impairment of ADL, motor and cognitive function among stroke survivors. According to people with stroke, carers and health professionals, improving cognitive abilities after stroke is the number one research priority in this field of medicine. Therefore, neurological rehabilitation, including effective training strategies, is needed to facilitate recovery and to reduce the burden of stroke. Therapies tailored to patients' and carers' needs are especially important. Current rehabilitation strategies have limited effectiveness in improving these impairments. One possibility for enhancing the effects of rehabilitation might be the addition of brain stimulation without breaking the skin, by means of tDCS. This technique can alter how the brain works and may be used to reduce impairment of ADL and function. However, the effectiveness of this intervention for improving rehabilitation outcomes is still unknown.

Search date

The review is current to January 2019.

Study characteristics

We included 67 studies involving a total of 1729 adult participants with acute, postacute or chronic ischaemic or haemorrhagic stroke. The mean age in the experimental groups ranged from 43 years up to 70 years, and from 45 years up to 75 years in the control groups. The level of participants' impairment ranged from severe to moderate. The majority of studies were conducted in an inpatient setting. Several different stimulation types with different stimulation durations and dosages were administered and compared with sham tDCS or an active control intervention. Sham tDCS means that the stimulation is switched off covertly in the first minute of the intervention.

Key results

This review found that tDCS might enhance ADL, but does not improve arm and leg function, muscle strength and cognitive abilities. Proportions of adverse events and people discontinuing the treatment were comparable between groups. Included studies differed in terms of type, location and duration of stimulation, the amount of current delivered, electrode size and positioning,as well as type and location of stroke. Future research is needed in this area to foster the evidence base of these findings, especially regarding arm and leg function, muscle strength and cognitive abilities (including spatial neglect).

Quality of the evidence

The quality of evidence for tDCS for improving ADL ranged from very low to high. It was low to moderate for upper extremity function, and moderate for adverse events and people discontinuing the treatment.

Summary of findings

Summary of findings 1. tDCS versus any type of placebo or passive control intervention for improving activities of daily living, and physical and cognitive functioning at the end of intervention period, in people after stroke.

tDCS versus any type of placebo or passive control intervention for improving activities of daily living, and physical and cognitive functioning at the end of intervention period, in people after stroke
Patient or population: people with stroke Settings: inpatient and outpatient setting Intervention: tDCS versus any type of placebo or passive control intervention
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	TDCS versus any type of placebo or passive control intervention
Primary outcome measure: mean ADL at the end of the intervention period Measures of activities of daily living. Scale from: 0 to infinity.		Absolute values in the intervention groups were 0.28 standard deviations higher (absolute values) (0.13 to 0.44 higher)		686 (19 studies)	⊕⊕⊕⊝ moderate^a	SMD 0.28 (0.13 to 0.44); however, this effect was not sustained when including only studies with adequate allocation concealment (Table 2)
		Change scores in the intervention groups were 0.48 standard deviations higher (change scores) (0.02 to 0.95 higher)		95 (4 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.48 (0.02 to 0.95); however, this effect was not sustained when including only studies with adequate allocation concealment (Table 2)
Secondary outcome measure: mean upper extremity function at the end of the intervention period Clinical measures of upper extremity function. Scale from: 0 to infinity.		Absolute values in the intervention groups were 0.17 standard deviations higher (absolute values) (0.05 lower to 0.38 higher)		792 (24 studies)	⊕⊕⊕⊝ moderate^d	SMD 0.17 (‐0.05 to 0.38)
		Change scores in the intervention groups was 0.33 standard deviations higher (change scores) (0.12 lower to 0.79 higher)		193 (10 studies)	⊕⊕⊝⊝ low^b,e	SMD 0.33 (‐0.12 to 0.79)
Secondary outcome measure: mean lower extremity function at the end of the intervention period Clinical measures of lower extremity function. Scale from: 0 to infinity.		Absolute values in the intervention groups were 0.28 standard deviations higher (absolute values) (0.12 lower to 0.69 higher)		204 (8 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.28 (‐0.12 to 0.69)
		Change scores in the intervention groups was 0.46 standard deviations higher (change scores) (0.09 lower to 1.01 higher)		54 (4 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.46 (‐0.09 to 1.01)
Secondary outcome measure: mean muscle strength at the end of the intervention period Clinical measures of muscle strength. Scale from: 0 to infinity.		Absolute values in the intervention groups were 0.19 standard deviations higher (absolute values) (‐0.01 lower to 0.38 higher)		437 (13 studies)	⊕⊕⊕⊕ high	SMD 0.19 (‐0.01 to 0.38)
		Change scores in the intervention groups were 0.19 standard deviations higher (change scores) (‐0.01 lower to 0.38 higher)		116 (5 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.07 (‐0.66 to 0.8)
Secondary outcome measure: mean cognitive abilities at the end of the intervention period Clinical measures of cognitive abilities. Scale from: 0 to infinity.		Mean in the intervention groups was 0.46 standard deviations higher (0.1 lower to 1.02 higher)		56 (2 studies)	⊕⊕⊝⊝ low^b,e	SMD 0.46 (‐0.1 to 1.02)
Secondary outcome measure: mean hemispatial neglect at the end of intervention period		Mean in the intervention groups was 4.8 higher (0.13 to 9.47 higher)		30 (1 study)	⊕⊝⊝⊝ very low^b,c,e	No statistical pooling possible
Secondary outcome measure: dropouts, adverse events and deaths during the intervention period Number of adverse events, dropouts and deaths during the intervention period	Study population		RR 1.25  (0.74 to 2.13)	1330 (47 studies)	⊕⊕⊕⊝ moderate^d
	34 per 1000	42 per 1000 (25 to 72)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADL: Activities of daily life; CI: Confidence interval; RR: Risk ratio; SMD: Standardised mean difference; tDCS: transcranial direct current stimulation
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

Sensitivity analysis	Studies included in analysis	Effect estimate
All studies with proper allocation concealment presenting absolute values	Hesse 2011; Khedr 2013; Kim 2010; Rocha 2016; Tedesco Triccas 2015b; Wu 2013a	(SMD 0.25, 95% CI ‐0.03 to 0.53; participants = 304; studies = 6; I² = 22%; inverse variance method with random‐effects model)
All studies with proper allocation concealment presenting change scores	Andrade 2017; Rabadi 2017	(SMD 0.31, 95% CI ‐0.49 to 1.11; participants = 76; studies = 2; I² = 53%; inverse variance method with random‐effects model)
All studies with proper blinding of outcome assessor for primary outcome absolute values	Allman 2016; Andrade 2017; Ang 2012; Bang 2015; Boggio 2007a; Bolognini 2011; Cha 2014; Chang 2015; Chelette 2014; Cho 2017; Cunningham 2015; D'Agata 2016; Danzl 2012; Di Lazzaro 2014a; Di Lazzaro 2014b; Fusco 2013a; Fusco 2014; Geroin 2011; Hamoudi 2018; Hathaiareerug 2019; Hesse 2011; Ilić 2016; Khedr 2013; Kim 2010; Koo 2018; Lee 2014; Lindenberg 2010; Manji 2018; Mazzoleni 2019; Mortensen 2016; Nair 2011; Nicolo 2017; Park 2013; Park 2015; Picelli 2015; Qu 2009; Rabadi 2017; Rocha 2016; Rossi 2013; Saeys 2015; Salazar 2019; Sattler 2015; Seo 2017; Shaheiwola 2018; Sik 2015; Straudi 2016; Tahtis 2012; Tedesco Triccas 2015b; Utarapichat 2018; Viana 2014; Wang 2014; Wong 2015; Wu 2013a	(SMD 0.23, 95% CI 0.05 to 0.41; participants = 536; studies = 15; I² = 0%; inverse variance method with random‐effects model)
All studies with proper blinding of outcome assessor for primary outcome change values	Danzl 2012; Fusco 2014	(SMD 0.77, 95% CI ‐0.21 to 1.75; participants = 19; studies = 2; I² = 0%; inverse variance method with random‐effects model)
All studies with intention‐to‐treat analysis for primary outcome absolute values	Allman 2016; Andrade 2017; Ang 2012; Bang 2015; Boggio 2007a; Bolognini 2011; Cha 2014; Chang 2015; Chelette 2014; Cho 2017; Cunningham 2015; D'Agata 2016; Danzl 2012; Di Lazzaro 2014a; Di Lazzaro 2014b; Fusco 2013a; Fusco 2014; Geroin 2011; Hamoudi 2018; Hathaiareerug 2019; Hesse 2011; Ilić 2016; Khedr 2013; Koo 2018; Lindenberg 2010; Manji 2018; Mazzoleni 2019; Mortensen 2016; Nair 2011; Nicolo 2017; Park 2013; Park 2015; Picelli 2015; Qu 2009; Rabadi 2017; Rocha 2016; Rossi 2013; Saeys 2015; Salazar 2019; Sattler 2015; Seo 2017; Shaheiwola 2018; Sik 2015; Straudi 2016; Tahtis 2012; Utarapichat 2018; Viana 2014; Wang 2014; Wong 2015; Wu 2013a	(SMD 0.27, 95% CI 0.06 to 0.47; participants = 387; studies = 11; I² = 0%; inverse variance method with random‐effects model)
All studies with intention‐to‐treat analysis for primary outcome change scores	Danzl 2012	(SMD 1.36, 95% CI ‐0.31 to 3.03; participants = 8; studies = 1; I² = 0%; inverse variance method with random‐effects model)

tDCS versus any type of active control intervention for improving activities of daily living, and physical and cognitive functioning at the end of intervention phase, in people after stroke
Patient or population: people with stroke Settings: inpatient and outpatient setting Intervention: tDCS versus any type of active control intervention
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	TDCS versus any type of active control intervention
Primary outcome measure: mean ADL at the end of the intervention period Barthel Index. Scale from: 0 to 100.	Absolute values in the control groups was 69.2 Barthel Index Score	Absolute values in the intervention groups was 6.59 higher (1.26 to 11.91 higher)		121 (3 studies)	⊕⊕⊝⊝ low^a,b
Secondary outcome measure: mean upper extremity function at the end of the intervention period Clinical measures of upper extremity function. Scale from: 0 to infinity.		Absolute values in the intervention groups was 0.84 standard deviations higher (absolute values) (0.2 to 1.48 higher)		124 (5 studies)	⊕⊕⊝⊝ low^a,b	SMD 0.84 (0.2 to 1.48)
		Change scores in the intervention groups was 0.51 standard deviations higher (change scores) (0.2 to 1.22 higher)		32 (1 study)	⊕⊕⊝⊝ low^a,b	SMD 0.51 (0.20 to 1.22)
Secondary outcome measure: mean lower extremity function at the end of the intervention period		Mean in the intervention groups was 0.23 standard deviations higher (0.66 lower to 1.13 higher)		66 (3 studies)	⊕⊕⊕⊝ moderate^a	SMD 0.23 (‐0.66 to 1.13)
Secondary outcome measure: mean muscle strength at the end of the intervention period		Mean in the intervention groups was 0.08 standard deviations higher (0.44 lower to 0.6 higher)		57 (2 studies)	⊕⊕⊝⊝ low^a,b	SMD 0.08 (‐0.44 to 0.6)
Secondary outcome measure: cognitive abilities at the end of the intervention period	No evidence available
Secondary outcome measure: spatial neglect at the end of the intervention period	See comment	See comment	Not estimable	12 (1 study)	⊕⊕⊕⊝ moderate^a
Secondary outcome measure: dropouts, adverse events and deaths during the intervention period Adverse events, dropouts and deaths during the intervention period	Study population		RR 1.76  (0.43 to 7.17)	209 (7 studies)	⊕⊕⊕⊝ moderate^a
	19 per 1000	34 per 1000 (8 to 139)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADL: Activities of daily life; CI: Confidence interval; RR: Risk ratio; SMD: Standardised mean difference; tDCS: transcranial direct current stimulation
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

tDCS versus any type of placebo or passive control intervention for improving activities of daily living, and physical and cognitive functioning at the end of follow‐up, in people after stroke
Patient or population: patients with improving activities of daily living, and physical and cognitive functioning at the end of follow‐up, in people after stroke Settings: inpatient and outpatient Intervention: tDCS
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	tDCS
Primary outcome measure: mean ADL until the end of follow‐up Measures of activities of daily living. Scale from: 0 to infinity.		Absolute values in the intervention groups was 0.31 standard deviations higher (absolute values) (0.01 to 0.62 higher)		269 (6 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.31 (0.01 to 0.62)
		Change scores in the intervention groups was 0.64 standard deviations lower (change scores) (1.66 lower to 0.37 higher)		16 (1 study)	⊕⊕⊝⊝ low^a,b	SMD ‐0.64 (‐1.66 to 0.37)
Secondary outcome measure: mean upper extremity function to the end of follow‐up Clinical measures of upper extremity function. Scale from: 0 to infinity.		Absolute values in the intervention groups was 0 standard deviations higher (absolute values) (0.39 lower to 0.39 higher)		211 (5 studies)	⊕⊕⊕⊝ moderate^b	SMD 0 (‐0.39 to 0.39)
		Change scores in the intervention groups was 0.51 standard deviations higher (change scores) (‐0.20 to 1.22 higher)		32 (1 study)	⊕⊕⊝⊝ low^b,c	SMD 0.51 (‐0.20, 1.22)
Secondary outcome measure: lower extremity function to the end of follow‐up	No evidence available
Secondary outcome measure: mean muscle strength at the end of follow‐up		Mean in the intervention groups was 0.07 standard deviations higher (0.26 lower to 0.41 higher)		156 (3 studies)	⊕⊕⊕⊝ moderate^b	SMD 0.07 (‐0.26 to 0.41)
Secondary outcome measure: cognitive abilities at the end of follow‐up	No evidence available
Secondary outcome measure: hemispatial neglect at the end of follow‐up	No evidence available
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADL: Activities of daily living; CI: Confidence interval; SMD: Standardised mean difference; tDCS: transcranial direct current stimulation
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

Date	Event	Description
14 January 2020	New search has been performed	We have rerun and expanded the searches to January 2019 and revised the text as appropriate. We have included 67 trials involving 1729 participants in this update compared with 32 trials with 748 participants in the last version of this review from 2015.
14 January 2020	New citation required but conclusions have not changed	The conclusions have not changed: there is evidence of an effect of transcranial direct current stimulation for improving activities of daily living, but not for arm function.

Date	Event	Description
28 September 2015	New search has been performed	The scope of the updated review has broadened since the previous version. This was in response to a request from the Cochrane Stroke Group to incorporate evidence relating to cognitive function (including neglect) into this update. We have rerun and expanded the searches to February 2015 and revised the text as appropriate. We have included 32 trials involving 748 participants in this update compared with 15 trials with 455 participants in the last version of this review from 2013.
28 September 2015	New citation required and conclusions have changed	The conclusions have changed: there is evidence of an effect of transcranial direct current stimulation for improving activities of daily living, but not for arm function.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Primary outcome measure: ADL at the end of the intervention period	23		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 Absolute values	19	686	Std. Mean Difference (IV, Random, 95% CI)	0.28 [0.13, 0.44]
1.1.2 Change scores	4	95	Std. Mean Difference (IV, Random, 95% CI)	0.48 [0.02, 0.95]
1.2 Primary outcome measure: ADL until the end of follow‐up	7		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.2.1 Absolute values	6	269	Std. Mean Difference (IV, Random, 95% CI)	0.31 [0.01, 0.62]
1.2.2 Change scores	1	16	Std. Mean Difference (IV, Random, 95% CI)	‐0.64 [‐1.66, 0.37]
1.3 Secondary outcome measure: upper extremity function at the end of the intervention period	34		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.3.1 Absolute values	24	792	Std. Mean Difference (IV, Random, 95% CI)	0.17 [‐0.05, 0.38]
1.3.2 Change scores	10	193	Std. Mean Difference (IV, Random, 95% CI)	0.33 [‐0.12, 0.79]
1.4 Secondary outcome measure: upper extremity function to the end of follow‐up	8		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 Absolute values	5	211	Std. Mean Difference (IV, Random, 95% CI)	‐0.00 [‐0.39, 0.39]
1.4.2 Change scores	3	72	Std. Mean Difference (IV, Random, 95% CI)	1.07 [0.04, 2.11]
1.5 Secondary outcome measure: lower extremity function at the end of the intervention period	12		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.5.1 Absolute values	8	204	Std. Mean Difference (IV, Random, 95% CI)	0.28 [‐0.12, 0.69]
1.5.2 Change scores	4	54	Std. Mean Difference (IV, Random, 95% CI)	0.46 [‐0.09, 1.01]
1.6 Secondary outcome measure: muscle strength at the end of the intervention period	18		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.6.1 Absolute values	13	437	Std. Mean Difference (IV, Random, 95% CI)	0.19 [‐0.01, 0.38]
1.6.2 Change values	5	116	Std. Mean Difference (IV, Random, 95% CI)	0.07 [‐0.66, 0.80]
1.7 Secondary outcome measure: muscle strength at the end of follow‐up	3	156	Std. Mean Difference (IV, Random, 95% CI)	0.07 [‐0.26, 0.41]
1.8 Secondary outcome measure: cognitive abilities at the end of the intervention period	2	56	Std. Mean Difference (IV, Random, 95% CI)	0.46 [‐0.10, 1.02]
1.9 Secondary outcome measure: hemispatial neglect at the end of intervention period	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
1.10 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period	47	1330	Risk Ratio (M‐H, Random, 95% CI)	1.25 [0.74, 2.13]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Primary outcome measure: ADL at the end of the intervention period, absolute values	3	121	Mean Difference (IV, Random, 95% CI)	6.59 [1.26, 11.91]
2.2 Secondary outcome measure: upper extremity function at the end of the intervention period	6		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
2.2.1 Absolute values	5	124	Std. Mean Difference (IV, Random, 95% CI)	0.84 [0.20, 1.48]
2.2.2 Change scores	1	32	Std. Mean Difference (IV, Random, 95% CI)	0.51 [‐0.20, 1.22]
2.3 Secondary outcome measure: upper extremity function to the end of follow‐up	1	32	Mean Difference (IV, Random, 95% CI)	10.00 [‐0.07, 20.07]
2.4 Secondary outcome measure: lower extremity function at the end of the intervention period	3	66	Std. Mean Difference (IV, Random, 95% CI)	0.23 [‐0.66, 1.13]
2.5 Secondary outcome measure: muscle strength at the end of the intervention period	2	57	Std. Mean Difference (IV, Random, 95% CI)	0.08 [‐0.44, 0.60]
2.6 Secondary outcome measure: spatial neglect at the end of the intervention period	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.7 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period	7	209	Risk Ratio (M‐H, Random, 95% CI)	1.76 [0.43, 7.17]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Planned analysis: duration of illness ‐ acute/subacute phase versus postacute phase for ADL at the end of the intervention period	19		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.1.1 Acute/subacute phase (the first week after stroke and the second to the fourth week after stroke	5	237	Std. Mean Difference (IV, Random, 95% CI)	0.26 [‐0.01, 0.53]
3.1.2 Postacute phase (from the first to the sixth month after stroke)	5	271	Std. Mean Difference (IV, Random, 95% CI)	0.34 [0.09, 0.59]
3.1.3 Chronic phase (from the sixth month after stroke)	9	198	Std. Mean Difference (IV, Random, 95% CI)	0.14 [‐0.15, 0.42]
3.2 Planned analysis: effects of type of stimulation (A‐tDCS/C‐tDCS/dual‐tDCS) and location of stimulation (lesioned/non‐lesioned hemisphere) on ADL at the end of the intervention period (study groups collapsed)	18		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.2.1 A‐tDCS over the lesioned hemisphere	12	300	Std. Mean Difference (IV, Random, 95% CI)	0.08 [‐0.15, 0.31]
3.2.2 C‐tDCS over the lesioned hemisphere	10	388	Std. Mean Difference (IV, Random, 95% CI)	0.30 [0.09, 0.50]
3.2.3 Dual‐tDCS (A‐tDCS over the lesioned and C‐tDCS over the non‐lesioned hemisphere)	3	46	Std. Mean Difference (IV, Random, 95% CI)	0.33 [‐0.25, 0.92]
3.3 Planned analysis: type of control intervention (sham tDCS, conventional therapy or nothing)	21		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.3.1 Sham tDCS	19	686	Std. Mean Difference (IV, Random, 95% CI)	0.28 [0.13, 0.44]
3.3.2 Active control intervention	3	121	Std. Mean Difference (IV, Random, 95% CI)	0.57 [‐0.31, 1.45]

*Study characteristics*
Methods	Study design: RCT Number of dropouts: 2 (in experimental group) Number of adverse events: not stated Deaths: none ITT: no
Participants	Country: UK Sample size: 24 patients (11 in experimental and 13 in control group) Inclusion criteria: at least 6 months after a single unilateral ischaemic or haemorrhagic stroke affecting motor function in the contralesional hand, informed consent Exclusion criteria: previous stroke or stroke affecting the primary motor cortex, inability to provide informed consent due to severe language or cognitive impairment, and contraindications for tDCS
Interventions	2 arms: anodal tDCS (1 mA over the lesioned M1) during the first 20 minutes of daily self‐administered Graded Repetitive Arm Supplementary Program (GRASP) training for 60 minutes over 9 days sham tDCS (1 mA over the lesioned M1) during the first 10 seconds of daily self‐administered GRASP training for 60 minutes over 9 days
Outcomes	Outcomes were measured at baseline and at 10, 17, 30 and 90 days after study start: WMFT ARAT UE‐FM
Funding source	Supported by the Dunhill Medical Trust, Oxford NIHR (National Institute for Health Research) Biomedical Research Centre, Wellcome Trust, Medical Research Council, and The People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme FP7/2007‐2013/ under Research Executive Agency (REA) grant agreement no. PITN‐GA‐2011‐290011
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A random number generator was used to assign conditions in blocks of four, stratified by starting level on the motor training program"
Allocation concealment (selection bias)	Unclear risk	Quote: "Randomization was performed by a researcher (H.J.‐B.) who was not involved in any baseline assessments, and allocation was communicated to one other researcher (U.A.)."
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants were blinded. Personnel performing the baseline treatment were blinded, whereas personnel performing tDCS was not. Quote: "Motor training was carried out by a researcher blind to stimulation conditions [...] tDCS was delivered by a researcher who was aware of the stimulation conditions"
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded. Personnel performing the baseline treatment were blinded, whereas personnel performing tDCS was not. Quote: "Motor training was carried out by a researcher blind to stimulation conditions [...] tDCS was delivered by a researcher who was aware of the stimulation conditions"
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "All clinical assessments were scored by a researcher blind to stimulation conditions"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "All clinical assessments were scored by a researcher blind to stimulation conditions"
Incomplete outcome data (attrition bias) Subjective outcome measures	Unclear risk	There were two dropouts in the tDCS group. The reason for this was not described by the authors. There was no intention‐to‐treat analysis. Quote: "Of 1191 patients assessed for eligibility, 26 were randomized to receive either anodal tDCS or sham treatment, and 24 completed the intervention"
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	There were two dropouts in the tDCS group. The reason for this was not described by the authors. There was no intention‐to‐treat analysis. Quote: "Of 1191 patients assessed for eligibility, 26 were randomized to receive either anodal tDCS or sham treatment, and 24 completed the intervention"
Selective reporting (reporting bias)	Unclear risk	There is a published protocol for this study (NCT01414582). All outcomes listed in the protocol have been reported, except the reaction time task and the SIS.

*Study characteristics*
Methods	Study design: RCT Number of dropouts: none Number of adverse events: 16 out of 60 patients (27%) experienced mild adverse events Deaths: none ITT: yes
Participants	Country: Brazil Sample size: 60 (20 in 2 experimental groups and 20 in control group) Inclusion criteria: age between 18 and 65 years; diagnosis of unilateral, nonrecurring, subacute stroke, as defined by the International Classification of Diseases (ICD10) through CT or MRI conducted by neurologists, one to three months after vascular injury. Participants also had to be able, by using any method of pinch, to grasp a washcloth from a table top, lift it up a few inches, and release it. Exclusion criteria: actively extension of the wrist more than 10°, extend ≥ 2 digits more than 10° and abduct the thumb more than 10°, difficulties in understanding the instructions, cognitive deficits, tDCS contraindications
Interventions	3 arms: A‐tDCS (0.7 mA over M1, duration not described) on 5 consecutive days for 2 weeks prior to CIMT (3‐hour daily protocol) A‐tDCS (0.7 mA over PMC, duration not described) on 5 consecutive days for 2 weeks prior to CIMT (3‐hour daily protocol) S‐tDCS (0.7 mA over M1 for 30 seconds) on 5 consecutive days for 2 weeks prior to CIMT (3‐hour daily protocol)
Outcomes	Outcomes were measured at baseline and at the end of intervention: BI UE‐FM MAS BBT MRC adverse events questionnaire
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The method of randomization was a 1 : 1: 1 permuted block randomization generated by a web based randomization tool (https://www.random.org)."
Allocation concealment (selection bias)	Low risk	Quote: "This was employed by concealed allocation of sequentially numbered, opaque sealed envelopes, so that the person responsible for allocation had no contact with patients or with the work of others."
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Participants were blinded Personnel were not blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded Personnel were not blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Unclear risk	Not described by the study authors
Blinding of outcome assessment (detection bias) Objective outcome measures	Unclear risk	Not described by the study authors
Incomplete outcome data (attrition bias) Subjective outcome measures	Unclear risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	In comparison with the published trial protocol, the following outcomes are not reported: motor activity log, Biodex Balance electronic platform, Postural Assessment Scale for Stroke (PASS), Falls Efficacy Study, Short Physical Performance Battery (SPPB), Wolf Motor Function Test (WMFT)

*Study characteristics*
Methods	Study design: randomised controlled cross‐over trial Number of dropouts: none Number of adverse events: not described Deaths: none ITT: yes
Participants	Country: China Sample size: 10 participants; mean age (SD) 63 (6) years; mean UE‐FM (SD) 58 (8) Inclusion criteria: not explicitly stated; participants were recruited from a convenience sample from two patient self help groups for stroke; participants were < 80 years of age; had a single stroke more than a year prior to enrolment and had weakness in the affected upper limb and could perform a pincer grip with the index finger Exclusion criteria: not explicitly stated, but people excluded were either illiterate in Chinese, had a history of other neurologic disorders, metal in the head, musculoskeletal pathologies affecting movements in the upper limbs, had aphasia or < 18 points on the MMSE
Interventions	Each participant underwent all of the following conditions: A‐tDCS over M1 of the affected hemisphere (1 mA for 20 minutes) plus sham tDCS over M1 of the unaffected hemisphere (1 mA for 10 seconds) once C‐tDCS over M1 of the unaffected hemisphere (1 mA for 20 minutes) plus sham tDCS over M1 of the affected hemisphere (1 mA for 10 seconds) once Sham tDCS over M1 of the unaffected hemisphere plus sham tDCS over M1 of the affected hemisphere (1 mA for 10 seconds) once
Outcomes	Outcomes were measured at baseline and at the end of intervention period Primary outcome measures: Purdue Pegboard Test (hand dexterity) Color‐word Stroop Test (selective attention) Secondary outcome measures: pinch grip strength (handheld digital dynamometer) fatigue (NRS)
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The sequence was determined in advance for each subject by drawing lots from an envelope"
Allocation concealment (selection bias)	Unclear risk	Quote: "The sequence was determined in advance for each subject by drawing lots from an envelope"
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Participants were blinded, but personnel were not Quote: "It was the third investigator (C.Y.) who set the tDCS parameters for both channels and operated the machine behind the subject throughout the experimental procedure"
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded, but personnel were not Quote: "It was the third investigator (C.Y.) who set the tDCS parameters for both channels and operated the machine behind the subject throughout the experimental procedure"
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Outcome assessors were blinded Quote: "Two other investigators (J.W. and E.C.) who were blinded to the allocated tDCS conditions then assessed the baseline motor status of the subjects’ paretic upper limb"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Outcome assessors were blinded Quote: "Two other investigators (J.W. and E.C.) who were blinded to the allocated tDCS conditions then assessed the baseline motor status of the subjects’ paretic upper limb"
Incomplete outcome data (attrition bias) Subjective outcome measures	Unclear risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: randomised sham‐controlled cross‐over trial Dropouts: none Adverse events: none Deaths: none ITT: yes Duration: 16 weeks
Participants	Country: Brazil Number of participants: 4 Age: (mean ± SD) 60.75 ± 13.15 years Gender: 0 female Type of stroke: not described, most likely ischaemic stroke Time poststroke: (mean ± SD) 34.5 ± 27.74 months Severity: mean muscle strength of the finger flexors (MRC) 3.8; mean ASS 0.5 Inclusion criteria: not clearly stated, but all participants had chronic, subcortical stroke, were right‐handed and had their stroke at least 12 months before study enrolment Exclusion criteria: not stated
Interventions	Characteristics: 4 weekly sessions of A‐tDCS (1 mA) over the hand area of M1 of the lesioned hemisphere, or C‐tDCS (1 mA) over the hand area of M1 of the non‐lesioned hemisphere or sham tDCS over the hand area of M1 of the lesioned hemisphere for 20 minutes with at least 2 weeks of rest between stimulation conditions
Outcomes	Outcomes used: duration of JTT in seconds Time point(s) of measurement: at baseline, after the first and after the fourth session of each treatment condition
Funding source	This work was supported by a grant within the Harvard Medical School Scholars in Clinical Science Program (NIH K30 HL04095‐03) to F.F. and by K24 RR018875, RO1‐NS 47754, RO1‐NS 20068 to A.P.‐L.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Procedure not described Quote: "The order of these conditions was counterbalanced and randomised across subjects"
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded; blinding of personnel was not described
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "A blinded rater evaluated motor function using the Jebsen‐Taylor Hand Function Test"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: randomised controlled multicentre trial Dropouts: 7 Adverse events: none Deaths: not stated ITT: no
Participants	Country: not stated Number of participants: 14 participants from the outpatient population of 3 neurological research units Age: (mean ± SD) 46.71 ± 14.08 years Gender: 9 women (64%) Type of stroke: 2 haemorrhagic (14%) Time poststroke: (mean ± SD) 35.21 ± 26.45 months Severity: moderate to severe hemiparesis, as indexed by UE‐FM (mean score 26, range 8 to 50) Inclusion criteria: ischaemic or haemorrhagic first‐ever stroke, stroke onset > 6 months before the study, functional inclusion criteria as defined by the EXCITE trial Exclusion criteria: pre stroke motor impairment affecting the upper limbs, moderate to severe major depression, previous CIMT and/or tDCS and contraindications regarding CIMT and/or tDCS
Interventions	Number of arms: 2 14‐day CIMT with shaping techniques + A‐tDCS (2 mA, 40 minutes) over the lesioned primary motor cortex (M1) 14‐day CIMT with shaping techniques + sham tDCS (40 minutes) over the lesioned primary motor cortex (M1)
Outcomes	Outcomes used: motor assessments: duration of JTT in seconds, handgrip strength, MAL, UE‐FM visual analogue scales for anxiety and pain/discomfort, questionnaire for adverse effects time point of measurement: day 1, day 5, day 10 (end of treatment) and at 2 and 4 weeks of follow‐up
Funding source	This work was supported by the American Heart Association (0735535T) (FF), University of Milano‐Bicocca (NB, GV), IRCCS Istituto Auxologico Italiano (NB, GV, LT), Regione Lombardia‐Ricerca Finalizzata 2009 (LT, CC)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation list (Bolognini 2013 [pers comm])
Allocation concealment (selection bias)	Unclear risk	The principal investigator, who took no part in data collection, nor in participants' evaluations, nor in treatment, knew the randomisation list and performed allocation (Bolognini 2013 [pers comm])
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Participants were blinded; blinding of personnel was not described
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded; blinding of personnel was not described
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "The assessment of motor functions and the administration of the functional scales and questionnaires were performed by a trained staff, blinded to group assignment"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "The assessment of motor functions and the administration of the functional scales and questionnaires were performed by a trained staff, blinded to group assignment"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	Dropouts due to frustration and tiredness during assessment Quote: "Five patients (2 in the active group and 3 in the sham group) did not complete the JHFT. Two patients (1 in the active group and 1 in the sham group) did not complete the HS task." These participants have been excluded from analysis and presentation of results"
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	Dropouts due to frustration and tiredness during assessment Quote: "Five patients (2 in the active group and 3 in the sham group) did not complete the JHFT. Two patients (1 in the active group and 1 in the sham group) did not complete the HS task." These participants have been excluded from analysis and presentation of results"
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: randomised sham controlled cross‐over trial in a factorial design Number of dropouts: not clearly stated Number of adverse events: not clearly stated Deaths: not stated ITT: yes
Participants	Country: Italy Sample size: 34 (8 in experimental group tDCS+rTMS, 16 in experimental group rTMS+tDCS, and 10 in control group sham tDCS) Inclusion criteria: chronic ischaemic or haemorrhagic stroke (> 6 months), aged between 18 and 70 years Exclusion criteria: global cognitive impairment (MMSE<25), severe functional disability (BI < 45), severe psychiatric disorders, degenerative neurological disorders, epilepsy, and severe medical conditions, implanted drug infusion systems, spinal/brain stimulators, endovascular coils
Interventions	2 arms (1a and 1b were conducted in a cross‐over design and arm 2 in a parallel design): 1a. rTMS+Dual tDCS group received 10 daily sessions of rTMS for 2 weeks and after a washout period (at least 6 months) 10 daily sessions of dual tDCS + mirror therapy for 2 weeks. 1b. dual tDCS + mirror therapy group received 10 daily sessions of dual tDCS + mirror therapy for 2 weeks and after a washout period (at least 6 months) they received 10 daily sessions of rTMS for 2 weeks 2. sham tDCS + mirror therapy group received 10 daily sessions of dual tDCS + mirror therapy for 2 weeks
Outcomes	Outcomes were measured at baseline and at the end of each intervention period prior to cross‐over: ARAT MMSE Auditory evoked potential Forward and backward digit span Attentional Matrices Short Story Test Copy of Figure Visual Search and Cancellation Tasks Nelson MCST Hamilton Depression Rating Scale (HDRS)
Funding source	None reported
Notes	All analyses are based on the comparison of arm 1b and arm 2, since arm 1a did not contain the same base therapy
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Participants and personnel were blinded Quote: "The trial was randomized double blind (Subject, Caregiver, Outcomes Assessor)"
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and personnel were blinded Quote: "The trial was randomized double blind (Subject, Caregiver, Outcomes Assessor)"
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Participants and personnel were blinded Quote: "The trial was randomized double blind (Subject, Caregiver, Outcomes Assessor)"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Participants and personnel were blinded Quote: "The trial was randomized double blind (Subject, Caregiver, Outcomes Assessor)"
Incomplete outcome data (attrition bias) Subjective outcome measures	Unclear risk	There were 6 dropouts in the treatment groups, but it was not clear, whether they occurred during the first treatment period prior to crossing over or afterwards
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	There were 6 dropouts in the treatment groups, but it was not clear, whether they occurred during the first treatment period prior to crossing over or afterwards
Selective reporting (reporting bias)	Low risk	Outcome measures have been reported for the outcomes listed in the published trial protocol

*Study characteristics*
Methods	Study design: randomised double‐blind sham‐controlled cross‐over trial Dropouts: none Adverse events: none Deaths: none ITT: yes
Participants	Country: not clearly stated Number of participants: 6 participants with chronic stroke neuroimaging‐proofed diagnosis; all were right‐handed and all had their strokes at least 12 months before the study Age: (mean ± SD) 53.7 ± 16.6 years Gender: 4 women (66%) Type of stroke: not stated Time poststroke: 27.1 months (range 12 to 72 months) Severity: motor strength (mean ± SD) 4.18 ± 0.37; ASS (mean ± SD) 0.83 ± 0.75 Inclusion criteria: not clearly stated Exclusion criteria: not clearly stated, but the authors referred to Hummel 2005, where the exclusion criteria were as follows: severe depression, history of severe alcohol or drug abuse, severe language disturbances, particularly of a receptive nature, or serious cognitive deficits (MMSE < 23/30 points)
Interventions	Characteristics: each participant underwent 3 different conditions for 20 minutes, separated by at least 48 hours of rest: A‐tDCS of the lesioned hemisphere's M1 (1 mA). C‐tDCS of the non‐lesioned hemisphere's M1 (1 mA). sham tDCS (electrode montage not stated by the authors).
Outcomes	Outcomes used: duration of JTT in seconds Time point of measurement: at baseline after familiarisation session, during stimulation and directly after stimulation
Funding source	This work was supported by a grant within the Harvard Medical School Scholars in Clinical Science Program (NIHK30 HL04095‐03) to F.F. and by K24 RR018875, RO1‐NS 47754, RO1‐NS 20068 to A.P.‐L.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded; blinding of personnel was not described
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Outcome assessor was blinded Quote: "A blinded neuropsychologist—instructed not to communicate with the patient during the task—evaluated patients’ performance"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: double‐blinded, sham‐controlled, randomised cross‐over study Dropouts: none Adverse events: none Deaths: none ITT: yes
Participants	Country: Italy Number of participants: 9 Age (mean ± SD): 53.5 ± 20.7 years Gender: 4 (57%) women Type of stroke: 8 (89%) ischaemic, 1 (11%) haemorrhagic Time post‐stroke (mean ± SD): 28.3 ± 10.4 days Severity (mean ± SD): grip strength 17.83 ± 7.45 kg Inclusion criteria: cortical or subcortical first‐ever stroke (radiologically confirmed), possibility to perform pinch/grip test Exclusion criteria: history of chronic disabling pathologies of the upper limb; spasticity; presence of pacemaker or severe cardiovascular conditions; a history of tumour, prior neurosurgical brain intervention, severe cardiovascular conditions (including the presence of a pacemaker), a diagnosis of epilepsy or major psychiatric disorders
Interventions	Each participant underwent 1 of the following different stimulation conditions in 2 consecutive sessions on 2 consecutive days in random order (sham tDCS was obligatory) A‐tDCS for 15 minutes at 1.5 mA over M1 of the lesioned hemisphere C‐tDCS for 15 minutes at 1.5 mA over M1 of the non‐lesioned hemisphere dual‐tDCS for 15 minutes at 1.5 mA, with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere sham tDCS (dosage and application not clearly stated, probably as in the other groups)
Outcomes	Outcomes were measured at baseline and at the end of intervention period: Nine‐Hole Peg Test‐index (quote: "9HPT‐index=velocity LS/velocity HS∗100") maximum pinch and grasp force in kg (measured by specific dynamometers according to the Jamar method, with a higher value indicating greater pinch and grasp force) patient satisfaction as measured by the Quebec User Evaluation of Satisfaction with Assistive Technology
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "For the random sequence generation, we used the RAND function in Matlab"
Allocation concealment (selection bias)	Low risk	Quote: "Specifically, patients were asked to take a sealed envelope from a box, containing a piece of paper with the assignment, which was concealed until the envelope was opened"
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Quote: "Patients were blinded while physicians and assessors knew the treatment (real or sham)"
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Quote: "Patients were blinded while physicians and assessors knew the treatment (real or sham)"
Blinding of outcome assessment (detection bias) Subjective outcome measures	High risk	Quote: "Patients were blinded while physicians and assessors knew the treatment (real or sham)"
Blinding of outcome assessment (detection bias) Objective outcome measures	Unclear risk	Quote: "Patients were blinded while physicians and assessors knew the treatment (real or sham)"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Method: RCT Number of dropouts: 3 (2 (14%) in the experimental group, 1 (7%) in the control group) Number of adverse events: not reported Deaths: not described ITT: no
Participants	Country: Italy Sample size: 11 participants (5 in the experimental and 6 in the control group) Inclusion criteria: admission to stroke unit; age between 18 and 83 years; ischaemic stroke in the MCA area confirmed by MRI or CT; time since stroke less than 30 days; no history of severe cognitive impairment; written informed consent Exclusion criteria: inability to perform a motor rehabilitation training; haemorrhagic stroke or multiple foci of ischaemia; previous stroke; diagnosis of major psychiatric disorders; epilepsy; history of tumour; pacemaker; uncontrolled arrhythmias; non‐stabilised heart diseases; dementia or severe aphasia
Interventions	2 arms C‐tDCS (1.5 mA for 10 minutes) over M1 of the unaffected hemisphere on 5 consecutive days each week for 2 weeks prior to a rehabilitative session sham tDCS (not described) over M1 of the unaffected hemisphere on 5 consecutive days each week for 2 weeks prior to a rehabilitative session
Outcomes	Outcomes were measured at baseline, after the end of intervention period, 1 month after the intervention period and at the end of inpatient rehabilitation (75 to 110 days): Canadian Neurological Scale Barthel Index 9‐hole peg test grasp and pinch force Upper extremity Fugl‐Meyer Assessment Timed Up and Go Test 6‐Minute Walking Test 10‐Meter Walking Test Rivermead Mobility Index Functional Ambulation Categories
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The randomization was created in accordance with a binary sequence previously generated using MATLAB R2007b Software (TheMatworks Inc., USA)"
Allocation concealment (selection bias)	Unclear risk	Not described by the study authors
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Quote: "The patient was blind to the type of stimulation. An unblinded investigator administered the stimulation"
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Quote: "The patient was blind to the type of stimulation. An unblinded investigator administered the stimulation"
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "The patient was blind to the type of stimulation, as well as the physician performing the assessments"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "The patient was blind to the type of stimulation, as well as the physician performing the assessments"
Incomplete outcome data (attrition bias) Subjective outcome measures	High risk	Quote: "Two patients of EG dropped out from the study (one at the first and the other one at the second session). Also one patient of control group dropped out for an emergency transfer to another hospital." These participants have not been analysed
Incomplete outcome data (attrition bias) Objective outcome measures	High risk	Quote: "Two patients of EG dropped out from the study (one at the first and the other one at the second session). Also one patient of control group dropped out for an emergency transfer to another hospital." These participants have not been analysed
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: pilot RCT Dropouts: none Adverse effects: none Deaths: none ITT: yes
Participants	Country: Italy Number of participants: 30 outpatients Age: (mean ± SD) 62.7 ± 6.4 years Gender: 7 women (23%) Type of stroke: unilateral ischaemic stroke Time post‐stroke: (mean ± SD) 26.4 ± 5.5 months Severity: mean ESS score 79.93 (minimum score: 0, maximum score: 100; a completely healthy person would have a score of 100) Inclusion criteria: at least 12 months from first unilateral ischaemic stroke, age < 75 years, ESS score ≥ 75 and ≤ 85, MMSE‐score ≥ 24, ability to maintain standing position without aid for at least 5 minutes, ability to walk independently for at least 15 minutes with the use of walking aids Exclusion criteria: history of seizures, EEG suspect of elevated cortical excitability, metallic implants within the brain and previous brain neurosurgery, medications altering cortical excitability or with a presumed effect of brain plasticity, posterior circulation stroke, deficits of somatic sensations involving the paretic lower limb, presence of vestibular disorders/paroxysmal vertigo, severe cognitive or communicative disorders, cardiovascular comorbidity, rehabilitation treatment 3 months before study enrolment
Interventions	Number of arms: 3; all participants underwent 50‐minute training sessions 5 times a week for 2 consecutive weeks and 1 of the following interventions: robot‐assisted gait training + A‐tDCS of the lesioned hemisphere over the presumed leg area (1.5 mA for 7 minutes) robot‐assisted gait training + sham tDCS of the lesioned hemisphere over the presumed leg area (for 7 minutes) overground walking exercises according to the Bobath approach
Outcomes	Primary outcomes: 6‐Minute Walking Test, 10‐Metre Walking Test Secondary outcomes: GAITRite system, FAC, RMI, MI leg subscore and MAS Time point of measurement: at baseline, after treatment and at two weeks follow‐up
Funding source	This research received no specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "After baseline evaluation, patients were allocated to one of three treatment groups according to a simple software‐generated randomisation scheme"
Allocation concealment (selection bias)	Low risk	Quote: "We allocated patients to one of the three treatment arms according to a restricted randomisation scheme. One of the investigators checked correct patient allocation according to the randomisation list. After unmasking at the end of the study, we checked that no errors had been made in allocation" (Smania 2013 [pers comm])
Blinding of participants and personnel (performance bias) Subjective outcome measures	Unclear risk	Quote: "Asking the assessor to make an educated guess tested the success of blinding. The therapists were aware of the type of treatment received by the patients. Patients were aware of the type of treatment who underwent but they were not aware about the type of stimulation (Group 1 stimulation vs Group 2 sham stimulation)" (Smania 2013 [pers comm])
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Quote: "Asking the assessor to make an educated guess tested the success of blinding. The therapists were aware of the type of treatment received by the patients. Patients were aware of the type of treatment who underwent but they were not aware about the type of stimulation (Group 1 stimulation vs Group 2 sham stimulation)" (Smania 2013 [pers comm])
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "All patients were evaluated by the same examiner (an experienced internal coworker) who was not aware of the treatment received by the patients"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "All patients were evaluated by the same examiner (an experienced internal coworker) who was not aware of the treatment received by the patients"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes stated in the methods section were reported, except muscle tone as measured by MAS

*Study characteristics*
Methods	Study design: RCT Number of dropouts: 5 (1 in sham tDCS group and 4 in no‐training/no‐tDCS group) Number of adverse events: 8 (6 in sham tDCS group and 2 in A‐tDCS group) Deaths: none ITT: no
Participants	Country: Germany Sample size: 56 (19 in A‐tDCS group, 19 in sham tDCS group and 18 in passive control group) Inclusion criteria: age 18 to 80 years, unilateral, first ever ischemic stroke ≥ 3 months before study enrolment, mild to moderate hemiparesis with residual hand function sufficient for task performance, clear hand preference as assessed by the Edinburgh Handedness Inventory and sufficient cognitive function to comply with study requirements Exclusion criteria: not stated
Interventions	3 arms: A‐tDCS (1.2 mA for 20 min) over the ipsilesional M1 during computerised grip strength training for 45 min per day for 5 days sham tDCS (1.2 mA for 30 sec) over the ipsilesional M1 during computerised grip strength training for 45 min per day for 5 days passive control group (did not receive training or tDCS)
Outcomes	Outcomes were measured at baseline, at day 8, 29, 57, 85 and 113 after study start: overall training effect (time and group differences in skill) total learning (the sum of skill changes at the end of training) learning stages (online vs. offline learning) generalization (JTT overall time) online learning on the first day cumulative learning probability relation Between Online and Offline Learning changes in movement time, target error rate, and movement smoothness correlation of total learning with patient characteristics
Funding source	HMS was supported by a National Institutes of Health NINDS Intramural Competitive Postdoctoral Fellowship and K23NS078052; financial support for the work at the Freiburg site is provided by the German Research Foundation (MH, AS‐M, BF, JR; DFG grant number RE 2740/3‐1). LGC was supported by the Intramural Research Program of the National Institutes of Health, NINDS
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Allocation to a tDCS condition followed a balanced randomization list prepared prior to the experiment."
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants were blinded in both study centres and personnel was blinded at the Freiburg site, whereas it was not at the NIH site. 80% of participants felt stimulated by active tDCS in the sham tDCS group and 70% in the A‐tDCS group
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded in both study centres and personnel was blinded at the Freiburg site, whereas it was not at the NIH site. 80% of participants felt stimulated by active tDCS in the sham tDCS group and 70% in the A‐tDCS group
Blinding of outcome assessment (detection bias) Subjective outcome measures	High risk	Blinding of outcome assessors not described
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Blinding of outcome assessors not described
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	5 participants dropped out during follow‐up due to reasons unrelated with the intervention (1 in sham tDCS group due to finger infection and 4 in no‐training/no‐tDCS group due to medical interventions which altered task performance)
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	5 participants dropped out during follow‐up due to reasons unrelated with the intervention (1 in sham tDCS group due to finger infection and 4 in no‐training/no‐tDCS group due to medical interventions which altered task performance)
Selective reporting (reporting bias)	Unclear risk	There is an a priori published trial protocol, which, however did not state certain outcome measures

*Study characteristics*
Methods	Study design: RCT Number of dropouts: 1 (in the tDCS group until follow‐up) Number of adverse effects: not reported Deaths: unclear ITT: yes
Participants	Country: Thailand Sample size: 18 (9 in experimental and 9 in control group) Inclusion criteria: haemorrhagic or ischaemic stroke more than 1 month prior validated by cerebral imaging, age between 20 to 80 years, presence of moderate to severe UE functional impairment (UE‐FM score 0 to 47), being able to speak Thai Exclusion criteria: unstable medical condition, upper limb contracture, tDCS contraindication, or electro‐acupuncture, epilepsy, malignant cardiac arrhythmia, pregnancy, traumatic conditions of the affected hand or peripheral nerve injury, presence of cognitive impairment as evaluated with the Thai MMSE with a score of less than 24, or psychiatric disorder and being unable to perform the given task or understand instructions
Interventions	2 arms: dual tDCS (2 mA for 20 minutes) once a week for 3 weeks during intensive physical therapy and occupational therapy performed in hourly sessions 3 times per week for 3 weeks electro‐acupuncture once a week for 3 weeks during intensive physical therapy and occupational therapy performed in hourly sessions 3 times per week for 3 weeks
Outcomes	Outcomes were measured at baseline, at the end of intervention period and at 1 month follow‐up: UE‐FM PPT grip strength (handheld dynamometer)
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participants were allocated to either TDCS or EA groups by sealed envelope with computer‐generated blocks of four randomizations."
Allocation concealment (selection bias)	Low risk	Quote: "Participants were allocated to either TDCS or EA groups by sealed envelope with computer‐generated blocks of four randomizations."
Blinding of participants and personnel (performance bias) Subjective outcome measures	High risk	Participants and personnel apparently were not blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	High risk	Participants and personnel apparently were not blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "The assessments were achieved by a physician who was blinded to group allocation."
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "The assessments were achieved by a physician who was blinded to group allocation."
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There was 1 dropout (11%) in the dual tDCS group (reason was not provided)
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	There was 1 dropout (11%) in the dual tDCS group (reason was not provided)
Selective reporting (reporting bias)	Unclear risk	There is an a priori published protocol available on the Internet. The secondary outcome "Modified Ashworth Scale" was not reported in the publication

PERMALINK

Transcranial direct current stimulation (tDCS) for improving activities of daily living, and physical and cognitive functioning, in people after stroke

Bernhard Elsner

Joachim Kugler

Marcus Pohl

Jan Mehrholz

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Summary of findings 1. tDCS versus any type of placebo or passive control intervention for improving activities of daily living, and physical and cognitive functioning at the end of intervention period, in people after stroke.

1. Sensitivity analyses for comparison 1.1: primary outcome of ADL performance at the end of the intervention period.

Summary of findings 2. tDCS versus any type of active control intervention for improving activities of daily living, and physical and cognitive functioning at the end of intervention period, in people after stroke.

Summary of findings 3. tDCS versus any type of placebo or passive control intervention for improving activities of daily living, and physical and cognitive functioning at the end of follow‐up, in people after stroke.

Summary of findings 4. tDCS versus any type of active control intervention for improving activities of daily living, and physical and cognitive functioning at the end of follow‐up, in people after stroke.

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

2013 version

2016 version

2020 version

1.

Included studies

Design

Sample sizes

Setting

Participants

2. Patient characteristics.

Interventions

3. Demographics of studies, including dropouts and adverse events.

Outcomes

Excluded studies

Risk of bias in included studies

2.

3.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Comparison 1. tDCS versus any type of placebo or passive control intervention

Comparison 1.1 Primary outcome measure: ADL at the end of the intervention period

1.1.1 Studies presenting absolute values

*Study characteristics*
Methods	Study design: double‐blind randomised sham‐controlled multicentre trial Dropouts: 11 (11%) Adverse effects: none Deaths: 2 (2%) due to heart infarction and during stent surgery ITT: yes, 85 participants completed the study (89%)
Participants	Country: Germany/Italy Number of participants: 96 stroke participants from 3 study centres Mean age: 65.0, range 39 to 79 years Gender: 37 women (39%) Type of stroke: all ischaemic, 45 of 96 (47%) right‐hemispheric stroke Time poststroke: (mean ± SD) A‐tDCS group: 3.4 ± 1.8 weeks; C‐tDCS group: 3.8 ± 1.4 weeks; sham tDCS group: 3.8 ± 1.5 weeks Severity: at least wheelchair‐mobile participants, who had severe flaccid upper limb paresis with no (MRC 0) or minimal (MRC 1) volitional hand and finger extensor activity. 24 had an upper limb UE‐FM (range 0 to 66) < 18 and were unable to transfer 3 wooden blocks from 1 compartment to the other in the Box and Block test Inclusion criteria: age 18 to 79 years, first supratentorial ischaemic stroke with a stroke interval of 3 to 8 weeks' duration, and with participation in a comprehensive inpatient rehabilitation programme Exclusion criteria: history of epileptic seizures, EEG suspect of elevated cortical excitability, metallic implants in the brain, preceding brain surgery, sensitive scalp skin, anticonvulsant or neuroleptic medications
Interventions	Number of arms: 3; each participant practised for 6 weeks every working day for 20 minutes with the arm robot (AT) and simultaneously received one of the following interventions: (1) A‐tDCS (2 mA) with the anode positioned over the presumed hand area of the lesioned hemisphere (2) C‐tDCS (2 mA) with the cathode positioned over the presumed hand area of the non‐lesioned hemisphere sham tDCS (0 mA) with consecutive changing of the positions of arms (1) and (2)
Outcomes	Primary outcome: sensory and motor integrity, degree of synergy as assessed by UE‐FM assessment score (0 to 66, 0 = no movement, 66 = full motion) Secondary outcomes: upper limb muscle strength (MRC; 0 to 5, 0 = plegic, 5 = full power), muscle tone (MAS; 0 to 5, 0 = no increase, 5 = affected part rigid in flexion or extension), BI, upper limb function (as assessed by Box and Block test, the transfer of as many wooden blocks as possible with the affected hand from 1 compartment to the other within 1 minute, with a high value indicating good function) Time point of measurement: study onset, end of the 6‐week intervention and 3 months of follow‐up
Funding source	The Verein zur Förderung der Hirnforschung und Rehabilitation e.V. supported the TRAGAT study (NCT 00407667)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Following a telephone call, an independent person randomly allocated eligible patients to 1 of the 3 groups by drawing a lot out of an envelope containing 96 lots, indicating A, B, and C"
Allocation concealment (selection bias)	Low risk	Quote: "Following a telephone call, an independent person randomly allocated eligible patients to 1 of the 3 groups by drawing a lot out of an envelope containing 96 lots, indicating A, B, and C. He then informed the locally responsible person about the group assignment and the study started the next day"
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants and personnel were blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and personnel were blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "To ensure blinded evaluation of the FMS, videos of the assessment, where the patients sat on a chair and a mirror was placed 45° behind them, were sent to an experienced therapist off site" and "Two experienced physiotherapists, blinded with respect to group assignment, assessed the secondary parameters together" and "The blinding was maintained at all measurement points"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "To ensure blinded evaluation of the FMS, videos of the assessment, where the patients sat on a chair and a mirror was placed 45° behind them, were sent to an experienced therapist off site" and "Two experienced physiotherapists, blinded with respect to group assignment, assessed the secondary parameters together" and "The blinding was maintained at all measurement points"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	1 dropout occurred during the study period as the result of pneumonia, and 10 after the end of the intervention period until follow‐up (6 were caused by being unavailable, 2 resulted from refusal to further participate and 2 were caused by cardiac conditions). ITT analysis was performed
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	1 dropout occurred during the study period as the result of pneumonia, and 10 after the end of the intervention period until follow‐up (6 were caused by being unavailable, 2 resulted from refusal to further participate and 2 were caused by cardiac conditions). ITT analysis was performed
Selective reporting (reporting bias)	Low risk	All outcomes reported in the methods section and in the published trial protocol reported

*Study characteristics*
Methods	Method: Randomised cross‐over trial Number of dropouts: none Number of adverse effects: 6 Deaths: none ITT: yes
Participants	Country: Republic of Korea Sample size: 10 participants Inclusion criteria: unilateral right hemispheric stroke, younger than 70 years; noticeable cognitive disorder after stroke; written informed consent Exclusion criteria: seizures; metal implants in the head, cardiac pacemaker; history of neuropsychiatric diseases
Interventions	Each participant underwent one of the following treatments: single session of A‐tDCS over the DLPFC of the non‐lesioned hemisphere (2 mA for 30 minutes) followed by a single session sham tDCS over the DLPFC of the non‐lesioned hemisphere (2 mA for 10 seconds), separated by at least 48 hours wash‐out period single session sham tDCS over the DLPFC of the non‐lesioned hemisphere (2 mA for 10 seconds) followed by single A‐tDCS over the DLPFC of the non‐lesioned hemisphere (2 mA for 30 minutes), separated by at least 48 hours wash‐out period
Outcomes	Outcomes were measured at baseline and at the end of intervention period: response accuracy recognition accuracy response time of a two‐back verbal working memory task
Funding source	Supported by the Korea Research Foundation Grant funded by the Korean Government (KRF‐2008‐1093‐000) and by a KOSEF grant funded by the Korean government (M10644000022‐06N4400‐02210)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The order of stimulation was randomly assigned for all participants"
Allocation concealment (selection bias)	Unclear risk	Not described by the authors
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants have been blinded by sham tDCS; blinding of personnel not stated
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Not described by the authors, however all outcome data were acquired by a computerised assessment during cognitive tasks
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Method: randomised cross‐over trial Dropouts: none Adverse effects: none (Paik 2015 [pers comm]) Deaths: none ITT: yes, all participants completed the study
Participants	Country: Republic of Korea Sample size: 10 people with stroke aged 48 to 84 years Inclusion criteria: not explicitly stated; written informed consent Exclusion criteria: cerebellar or brainstem lesion; metallic body implant; pacemaker; cochlear implant; history of seizure; unstable medical condition; inability to perform outcome tasks; Na+ or Ca++ channel blockers
Interventions	Each participant underwent one of the following treatments: A‐tDCS over the left DLPFC (2 mA for 20 minutes) followed by sham tDCS over the left DLPFC (2 mA for 1 minute), separated by at least 48 hours wash‐out period S=sham tDCS over the left DLPFC (2 mA for 1 minute) followed by A‐tDCS over the left DLPFC (2 mA for 20 minutes), separated by at least 48 hours wash‐out period
Outcomes	Outcomes were measured at baseline, at the end of intervention period and at 3 hours postintervention: attention (Go/No‐Go test)
Funding source	This research was supported by a grant from Seoul National University College of Medicine (Grant No. 800‐20060236) to NJ Paik, and by a grant from the Korean Geriatric Society to EK Kang
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "We applied random order using computerized program. Randominzation program is freely available in the Internet." (Paik 2015 [pers comm]) Comment: However, patient‐ID and first session stimulation type were continuously alternated, as can be seen in Table 1
Allocation concealment (selection bias)	Unclear risk	Not described by the study authors
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants were blinded; blinding of personnel not stated
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Outcome assessor was blinded Quote: "Both patients and the investigator that carried out the behavioral measurements were unaware of the type of intervention, because tDCS and sham were administered by another investigator who did not participate in the behavioral task or data analysis"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported. There was no published a priori trial protocol (Paik 2015 [pers comm])

*Study characteristics*
Methods	Study design: RCT (parallel assignment) Dropouts: none Adverse effects: none Deaths: none ITT: yes, all participants completed the study
Participants	Country: Egypt Number of participants: 40 outpatients Age: (mean ± SD) years Gender: 14 women (35%) Type of stroke: acute single thromboembolic non‐haemorrhagic infarction, documented by MRI Time poststroke: (mean ± SD) 17.1 ± 3.6 days Severity: (range) 7 to 13 on NIHSS Exclusion criteria: extensive infarction (all territories of MCA), severe flaccid hemiplegia, head injury, neurological disease other than stroke, renal or hepatic impairment, previous administration of tranquilliser, inability to give informed consent, no MEP recorded from FDI muscle of the affected hand
Interventions	3 arms: A‐tDCS, 25 minutes at 2 mA daily for 6 consecutive days on M1 of the lesioned hemisphere, delivered by saline‐soaked pads (5 x 7 cm) C‐tDCS, 25 minutes at 2 mA daily for 6 consecutive days on M1 of the non‐lesioned hemisphere, delivered by saline‐soaked pads (5 x 7 cm) sham tDCS, 25 minutes daily (with a short ramp‐up and ramp‐down of the current at the beginning and at the end of each session) for 6 consecutive days on M1 of the lesioned hemisphere
Outcomes	NIHSS at baseline, at the end of the intervention period and at 1, 2 and 3‐month follow‐up (0 to 42, with higher scores indicating a more severe stroke) OMCASS at baseline, at the end of the intervention period and at 1, 2 and 3‐month follow‐up (0 to 100, with higher scores indicating no clinical impairment due to stroke) BI at baseline, at the end of the intervention period and at 1, 2 and 3‐month follow‐up (0 to 100, with higher scores indicating better global function) Muscle strength according to MRC at the end of the intervention period, at 1, 2 and 3‐month follow‐up (0 to 5, with higher scores indicating higher muscle strength) Cortical excitability (as measured by RMT and AMT) at the end of the intervention period, at 1, 2 and 3‐month follow‐up (with greater intensity indicating a higher threshold)
Funding source	The review author(s) received no financial support for the research, authorship, and/or publication of this article
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Each patient was given a serial number from a computer‐generated randomisation table"
Allocation concealment (selection bias)	Low risk	Quote: "Group allocations (Anodal, Cathodal, or Sham) were placed in serially numbered, opaque closed envelopes ... and each patient was placed in the appropriate group after opening the corresponding sealed envelope"
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants and therapists were blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and therapists were blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Outcome assessor was blinded
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Outcome assessor was blinded
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Low risk	All outcomes stated in the study protocol and listed in the methods section of the publication have been reported

*Study characteristics*
Methods	Study design: single‐blinded, sham‐controlled, randomised cross‐over study Dropouts: none Adverse effects: none Deaths: none ITT: yes
Participants	Country: Republic of Korea Number of participants: 10 subacute participants Age: (mean) 62.8 years Gender: seven women (70%) Type of stroke: first‐ever stroke, as confirmed by MRI; 2 had haemorrhagic stroke (20%) Time poststroke: (mean) 6.4 weeks, range 3 to 12 weeks Severity: participants could grasp and release independently; degree of strength according to MRC was ≥ 3 but < 5 for all paretic finger flexors and extensors. Participants did not have a family history of seizure, could understand the purpose of the study and did not have any deformities or contractures of the fingers, hands, elbows and shoulders Inclusion criteria: not explicitly stated Exclusion criteria: not explicitly stated
Interventions	Each participant underwent 2 different stimulation conditions, each for 20 minutes, separated by at least 24 hours of rest: A‐tDCS (1 mA) over the primary motor cortex of the first dorsal interossei muscle of the lesioned hemisphere sham tDCS over the primary motor cortex of the first dorsal interossei muscle of the lesioned hemisphere
Outcomes	Box and Block test (the transfer of as many wooden blocks as possible with the lesioned hand from 1 compartment to the other within 1 minute, with a high value indicating good function) and finger acceleration measurement (in g, with a higher value indicating higher acceleration) at baseline, at 5 minutes of stimulation, immediately and at 30 and 60 minutes after stimulation Visual analogue scales to assess attention and fatigue (score 1 to 7; 1 = no attention/fatigue; 7 = highest level of attention/fatigue) at baseline, immediately and at 30 and 60 minutes after stimulation
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "A doctor who works in tDCS's room, he randomised patients on his own sequence" (Kim 2013 [pers comm])
Allocation concealment (selection bias)	Unclear risk	Quote: "A doctor who works in tDCS's room, he randomised patients on his own sequence" (Kim 2013 [pers comm])
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Both participants and personnel were blinded (Kim 2013 [pers comm])
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Both participants and personnel were blinded (Kim 2013 [pers comm])
Blinding of outcome assessment (detection bias) Subjective outcome measures	High risk	No blinding of outcome assessors Quote: "An examiner who was aware of the stimulation method used was instructed not to communicate with patients during the task and evaluated patients’ performances"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	No blinding of outcome assessment, but the review authors judged that the outcome measurement is not likely to be influenced by lack of blinding Quote: "An examiner who was aware of the stimulation method used was instructed not to communicate with patients during the task and evaluated patients’ performances"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: double‐blind sham‐controlled multicentre randomised trial Dropouts: 1 participant discontinued treatment because of dizziness and another because of headache (2 out of 20) during follow‐up Adverse effects: none Deaths: none ITT: no
Participants	Country: Republic of Korea Number of participants: 20 participants from neurorehabilitation units at 2 tertiary university hospitals Age: (mean ± SD) 57.27 ± 4.95 Gender: 7 women (35%) Type of stroke: first‐ever cortical or subcortical ischaemic stroke Time poststroke: (mean ± SD) A‐tDCS group: 34 ± 27.1 days; C‐tDCS: 19.4 ± 9.3 days; sham tDCS: 22.9 ± 7.5 days Severity: mild to moderate motor deficits (MRC score ≥ 2) Inclusion criteria: first‐ever ischaemic strokes in the cortical or subcortical area within the previous 2 months and mild to moderate motor deficits (MRC score ≥ 2) Exclusion criteria: cerebellar or brainstem lesions; presence of a metallic foreign body implant, such as a pacemaker or an artificial cochlea; history of seizure or another unstable medical condition; severe language disturbance; neglect, depression or cognitive deficits (based on the MMSE, 10 of 30 points) that would limit participation; history of severe alcohol or drug abuse; previous stroke that resulted in residual disability; premorbid arm impairment; and hemiplegic shoulder pain; use Na⁺ or Ca²⁺channel blockers or NMDA receptor antagonists
Interventions	Number of arms: 3 Each participant received 10 sessions (5 times per week for 2 weeks during conventional occupational therapy aiming at improving the co‐ordination and strength of the paretic hand) of 1 of the following interventions: A‐tDCS over the primary motor cortex (M1) of the contralateral FDI muscle of the lesioned hemisphere (2 mA for 20 minutes) C‐tDCS over the M1 of the ipsilateral FDI of the non‐lesioned hemisphere (2 mA for 20 minutes) sham tDCS over the M1 of the contralateral FDI (for 20 minutes)
Outcomes	Outcomes used: FMA 0 to 66 (with higher scores indicating better function) for assessing upper limb motor function and MBI 0 to 100 (with higher scores indicating better global function) Time point of measurement: at baseline, 1 day and 6 months after intervention
Funding source	Supported by a grant from Helping Water Foundation (to NJP)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly assigned to one of the three groups (atDCS, ctDCS or Sham treatment) using a stratified randomisation procedure with permuted block size of 3 and an algorithm that balanced Brunnstrom stages"
Allocation concealment (selection bias)	Low risk	Quote: "Sealed opaque envelopes were used for randomisation"
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants and personnel were blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and personnel were blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "Two independent raters blinded to the type of intervention performed outcome measurements"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "Two independent raters blinded to the type of intervention performed outcome measurements"
Incomplete outcome data (attrition bias) Subjective outcome measures	Unclear risk	1 participant of each interventional arm (14% each) discontinued intervention; we excluded these participants from analysis
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	1 participant of each interventional arm (14% each) discontinued intervention; we excluded these participants from analysis
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in the methods section reported

*Study characteristics*
Methods	Study design: randomised cross‐over study Number of dropouts: none Number of adverse effects: none Deaths: none ITT: yes
Participants	Country: Thailand Sample size: 38 Inclusion criteria: first‐ever ischaemic stroke, confirmed by CT or MRI, with an onset ≤ 6 months, lower‐limb weakness but able to perform sit‐to‐stand independently and walk without physical assistance for at least 3 metres Exclusion criteria: intracranial metal implants, cochlear implants, cardiac pacemaker, history of seizures, no clear neurological history, psychiatric disorders, excessive pain in any joint of the lower limb
Interventions	Each participant underwent the following conditions, separated by a wash‐out period of ≥ 1 week: dual tDCS (2 mA for 20 minutes) once prior to 60 minutes of physical therapy sham tDCS (2 mA for 20 minutes) once prior to 60 minutes of physical therapy
Outcomes	Outcomes were measured at baseline and at the end of each intervention phase: knee extensor strength TUG FTSST
Funding source	This work was supported by a research grant from the Faculty of Physical Therapy, Mahidol University (2016/018.2901)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "participants were randomly allocated to receive the real or sham experiment for their first experiment."
Allocation concealment (selection bias)	Unclear risk	Quote: "participants were randomly allocated to receive the real or sham experiment for their first experiment."
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Both were blinded Quote: "tDCS was applied by a researcher who was blinded to the outcome assessment and data analysis."
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Both were blinded Quote: "tDCS was applied by a researcher who was blinded to the outcome assessment and data analysis."
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "These outcome measures were evaluated before and after the intervention by a researcher who was blinded to the intervention."
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "These outcome measures were evaluated before and after the intervention by a researcher who was blinded to the intervention."
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Incomplete outcome data (attrition bias) Objective outcome measures	Low risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Low risk	All outcomes listed in the prospectively registered trial protocol have been reported

*Study characteristics*
Methods	Method: RCT Number of dropouts: 5 (3 out of 42 in the experimental groups (7%) and 2 out of 22 in the control group (9%)) Number of adverse effects: no major adverse events Deaths: none ITT: no
Participants	Country: Republic of Korea Sample size: 59 people with stroke (39 in the experimental groups and 20 in the control group) Inclusion criteria: unilateral hemiparesis caused by stroke; first stroke within 1 month prior to enrolment; shoulder motor strength Medical Research Council grade ≤ 2 Exclusion criteria: contraindications to brain stimulation; previous history of brain neurosurgery or epilepsy; metallic implants in the brain; severe cognitive impairment; aphasia interfering with understanding study instructions; poor sitting balance; impaired vision; hemispatial neglect
Interventions	3 arms: C‐tDCS over the hand area of M1 over the non‐lesioned hemisphere (2 mA for 20 minutes) during occupational therapy aiming at functional improvement of the affected arm for 30 minutes per day, 5 times a week for 3 weeks; virtual reality training aiming at functional improvement of the affected arm for 30 minutes per day, 5 times a week for 3 weeks; C‐tDCS plus virtual reality training aiming at functional improvement of the affected arm for 30 minutes per day, 5 times a week for 3 weeks
Outcomes	Outcomes were measured at baseline and at the end of intervention period Modified Ashworth Scale Manual Muscle Testing Manual Function Test Fugl‐Meyer assessment, upper extremity subscale Korean‐Modified Barthel Index
Funding source	None reported
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "All of the enrolled patients were randomly assigned to 1 of 3 groups using a table of random numbers"
Allocation concealment (selection bias)	Unclear risk	Not described by the study authors
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	Participants and personnel providing the base treatment were blinded
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and personnel providing the base treatment were blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	Quote: "All evaluations were performed before and immediately after treatment by a single experienced occupational therapist who was not aware of the treatment allocation"
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "All evaluations were performed before and immediately after treatment by a single experienced occupational therapist who was not aware of the treatment allocation"
Incomplete outcome data (attrition bias) Subjective outcome measures	High risk	3 participants out of 42 (7%) in the experimental groups and 2 out of 22 (9%) were lost to follow‐up and excluded from the analysis. 2 out of the 3 losses to follow‐up in the experimental group dropped out due to "medical problem(s)"
Incomplete outcome data (attrition bias) Objective outcome measures	High risk	3 participants out of 42 (7%) in the experimental groups and 2 out of 22 (9%) were lost to follow‐up and excluded from the analysis. 2 out of the 3 losses to follow‐up in the experimental group dropped out due to "medical problem(s)"
Selective reporting (reporting bias)	Unclear risk	All outcomes listed in the methods section reported

*Study characteristics*
Methods	Study design: sham‐controlled double‐blinded randomised trial Dropouts: not stated Adverse effects: none Deaths: not stated, likely none ITT: not stated
Participants	Country: USA Number of participants: 20 chronic stroke participants Age: (mean ± SD) 55.8 ± 12.9 years Gender: 5 women (25%) Type of stroke: first and only ischaemic stroke Time poststroke: (mean ± SD) 40.3 ± 23.4 months Severity: UE‐FM Score (mean ± SD) 39.8 ± 11.5 Inclusion criteria: ischaemic stroke in the territory of the medial cerebral artery at least 5 months before enrolment; no previous or subsequent strokes; MRC strength grade of 3/5 in extensor muscles of the lesioned upper extremity in the acute phase with at least 15 degrees of active wrist dorsiflexion at enrolment Exclusion criteria: additional neurological or psychiatric disorders; concurrent use of CNS‐affecting drugs
Interventions	Number of arms: 2, each participant underwent 5 consecutive sessions of physical therapy/occupational therapy and 1 of the following interventions Dual‐tDCS: A‐tDCS over M1 of the lesioned hemisphere + C‐tDCS over M1 of the non‐lesioned hemisphere (1.5 mA each, for 30 minutes) Sham tDCS (for 30 minutes)
Outcomes	Primary outcome measure: UE‐FM scores (0 to 66, with higher scores reflecting better motor performance) Secondary outcome measure: WMFT (with lower scores indicating better motor performance) Time point of measurement: at baseline and at 3 and 7 days after the last intervention session
Funding source	Supported by the NIH/NINDS (NS045049)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly assigned to one of two groups ... using a block randomisation with 3 strata of impairment"
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants and personnel were blinded
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "Each patient underwent motor impairment assessments and MRI at baseline and after the intervention, conducted by trained individuals who were blinded to the type of intervention the patients received"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	All randomised participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes stated in the methods section were reported

*Study characteristics*
Methods	Study design: sham‐controlled cross‐over randomised trial Dropouts: not stated, most likely none Adverse effects: none Deaths: not stated, most likely none ITT: not stated
Participants	Country: Iran Number of participants: 10 right‐handed stroke participants with no sensory deficits Age: (mean ± SD) 60.8 ± 14.1 years Gender: 3 women (30%) Type of stroke: ischaemic Time poststroke: (mean ± SD) 8.3 ± 5.45, range 1 to 16 months Severity: median Brunnstrom stage 6 Inclusion criteria: single ischaemic stroke with more than 1 month's duration of mild to moderate motor deficit (to ensure that all participants could perform all items on the JTT Exclusion criteria: clinically significant or unstable medical or psychiatric disorder with history of substance abuse, any neuropsychiatric comorbidity other than stroke and contraindications to tDCS
Interventions	Each participant underwent 5 different treatments with at least 4 days of each of the following: A‐tDCS of lesioned M1 (with the cathodal electrode positioned at the contralateral supraorbital area, 1 mA for 20 minutes) A‐tDCS of lesioned M1 (with the cathodal electrode positioned at the contralateral deltoid muscle, 1 mA for 20 minutes) C‐tDCS of lesioned M1 (with the anodal electrode positioned at the contralateral supraorbital area, 1 mA for 20 minutes) dual‐tDCS: A‐tDCS of lesioned M1 + C‐tDCS of non‐lesioned M1 sham tDCS (20 minutes)
Outcomes	Outcomes used: JTT (with familiarisation sessions) Time points of measurement: at baseline and after stimulation
Funding source	This study was partially supported by an American Heart Association (AHA) grant (grant number 0735535T)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The order of these conditions was counterbalanced and randomised across patients"
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of participants and personnel (performance bias) Objective outcome measures	Low risk	Participants probably were blinded; blinding of personnel was not described. Quote: "Patients were then randomised to the double‐blinded, sham‐controlled cross over part of the experiment"
Blinding of outcome assessment (detection bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Blinding of outcome assessment (detection bias) Objective outcome measures	Low risk	Quote: "A blinded physiatrist—instructed not to communicate with the patients during the task—evaluated patients' performance"
Incomplete outcome data (attrition bias) Subjective outcome measures	Low risk	There were no subjective outcome measures
Incomplete outcome data (attrition bias) Objective outcome measures	Unclear risk	All participants apparently completed the study. No treatment withdrawals, no losses to follow‐up, no trial group changes and no major adverse events were stated
Selective reporting (reporting bias)	Unclear risk	All outcomes stated in the methods section were reported