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

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

Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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 1)
		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 1)
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.
^aDowngraded one level because 95% CI contains effect size of the minimal important difference. ^bDowngraded one level because the total sample size is less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^cDowngraded one level due to study ratings with 'high' risk of bias ^dDowngraded one level because 95% CI contains effect size of no difference and the minimal important difference. ^ePublication bias strongly suspected by visual inspection of funnel plot. Downgraded one level.

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

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)
CI: confidence interval SMD: standardised mean difference

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

Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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.
^aDowngraded one level due to total sample size being less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^bDowngraded one level due to several study ratings with 'high' risk of bias.

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

Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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.
^aDowngraded one level due to study ratings with 'high' risk of bias. ^bDowngraded one level because the total sample size is less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^cDowngraded one level because publication bias strongly suspected.

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

Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)
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
Patient or population: people with stroke Settings: inpatient and outpatient Intervention: tDCS
	Assumed risk	Corresponding risk
	Control	tDCS
Primary outcome measure: mean ADL at the end of follow‐up Scale from: 0 to 100.	No evidence available
Secondary outcome measure: mean upper extremity function to the end of follow‐up per cent change in Jebsen‐Taylor‐Test		Mean in the intervention groups was 10 higher (0.07 lower to 20.07 higher)	32 (1 study)	⊕⊕⊕⊝ moderate^a
Secondary outcome measure: lower extremity function at the end of follow‐up	No evidence available
Secondary outcome measure: muscle strength at the end of follow‐up	No evidence available
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 life; 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.
^aDowngraded one level due to total sample size being less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria).

Background

Description of the condition

Every year, 15 million people worldwide suffer from stroke (WHO 2011). Of these, nearly six million die (Mathers 2011). Another five million people are left permanently disabled by stroke every year (WHO 2011). Hence, stroke is one of the leading causes of death worldwide and has a considerable impact on disease burden (WHO 2011). Stroke affects function and many activities of daily living (ADL). Three out of four stroke patients have an impairment in performing ADL at hospital admission, and only about one‐third of patients who have completed rehabilitation have achieved normal neurological function (Jørgensen 1999). Around half of patients do not regain function of the affected arm six months after stroke (Kwakkel 2003). Three out of four people with stroke suffer from working memory impairment and may thus experience executive dysfunction (Riepe 2004). Based on ratings by people with stroke, carers and health professionals, improving cognition after stroke is the number one research priority in stroke medicine (Pollock 2012). Therefore, neurological rehabilitation (including effective training strategies) is needed to facilitate recovery and to reduce the burden of stroke (Barker 2005). Therapies tailored to patients' and carers' needs are especially important (Barker 2005).

Description of the intervention

Transcranial direct current stimulation (tDCS) is a non‐invasive method used to modulate cortical excitability by applying a direct current to the brain (Bindman 1964; Nowak 2009; Purpura 1965). Stimulation of the central nervous system by tDCS is inexpensive when compared with repetitive transcranial magnetic stimulation (rTMS) and epidural stimulation (Hesse 2011).

How the intervention might work

Transcranial direct current stimulation (tDCS) usually is delivered via saline‐soaked surface sponge electrodes, which are connected to a direct current stimulator of low intensity (Lang 2005). Three different applications might be used: 1) the anodal electrode may be placed over the presumed area of interest of the brain with the cathodal electrode placed above the contralateral orbit (anodal stimulation, A‐tDCS); 2) the cathodal electrode may be placed over the presumed area of interest of the brain with the anodal electrode placed above the contralateral orbit (cathodal stimulation, C‐tDCS) (Hesse 2011); or 3) anodal stimulation and cathodal stimulation may be applied simultaneously (dual‐tDCS) (Lindenberg 2010). Primarily resulting from a shift of the resting potential of the brain's neurons, tDCS using anodal stimulation might lead to increased cortical excitability, whereas cathodal stimulation might lead to decreased excitability (Bindman 1964; Floel 2010; Purpura 1965). Stimulation lasting for longer than five minutes might induce significant after‐effects (which probably are mainly due to changes in synaptic mechanisms), which could last up to several hours (Nitsche 2001; Nitsche 2003). These effects probably are 1) anatomically specific (referring to how the electrodes are positioned and which way the current takes to reach the targeted brain areas); 2) activity‐selective and task‐specific (meaning that neuronal networks active during a certain activity are preferentially stimulated by tDCS); and 3) input‐selective (meaning that tDCS would alter the neuronal system's input and thereby enhance information processing) (Bikson 2013). The facilitating effect of tDCS could be used to facilitate motor learning in healthy people (Boggio 2006; Jeffery 2007; Nitsche 2001; Nitsche 2003; Reis 2009), and appears to be a promising option in rehabilitation after stroke.

Why it is important to do this review

Previous versions of this review suggested that tDCS, with or without simultaneous upper extremity training, in people with stroke, results in greater improvement in arm motor function when compared with sham tDCS alone (Elsner 2013; Elsner 2016). Some pilot studies have even reported improvement in ADL, such as turning over playing cards, picking up beans with a spoon, and manipulating light and heavy objects with the arm (Fregni 2005; Hummel 2005; Kim 2009). However, these findings were not supported by a large‐scale multicentre randomised controlled trial (RCT), which did not find any effects on measures of ADL (Hesse 2011). There is contradictory evidence on the additional effect of tDCS on lower extremity function and gait (Cha 2014; Fusco 2014; Geroin 2011; Tahtis 2012). There are indications that tDCS might also improve working memory or neglect by modulating excitability of the corresponding brain areas (Au‐Yeung 2014; Jo 2008a; Kang 2008b; Ko 2008a; Park 2013; Sunwoo 2013a). However, in a systematic review of RCTs about the effects of tDCS on aphasia, no evidence of an effect was found (Elsner 2015). Despite the fact that adverse effects associated with the application of tDCS have been reported rarely so far, concerns about the safety of tDCS regarding its impact on cerebral autoregulation have recently emerged (List 2015; Nitsche 2015).

To date, studies of tDCS have tended to include small sample sizes. Currently, no systematic review has comprehensively synthesised the findings of available RCTs. Therefore, a systematic review of RCTs investigating the effectiveness and acceptability of tDCS for improving ADL, motor function and cognitive abilities (including spatial neglect) in people with stroke is required.

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.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs and randomised controlled cross‐over trials, from which we analysed only the first period as a parallel‐group design. We did not include quasi‐RCTs.

Types of participants

We included adult participants (18 years of age and older) who had experienced a stroke. We used the World Health Organization (WHO) definition of stroke (Hatano 1976), or a clinical definition, if not specifically stated (i.e. signs and symptoms persisting longer than 24 hours). We included participants regardless of initial level of impairment, duration of illness, or gender.

Types of interventions

This is the update of an existing review. In the previous versions 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 (i.e. sham tDCS, no intervention or conventional motor rehabilitation). We defined active tDCS as the longer‐lasting (lasting longer than two minutes) application of a direct current to the brain to stimulate the affected hemisphere, or to inhibit the healthy hemisphere (NItsche 2000). We defined sham tDCS as short‐term direct current stimulation (lasting less than two minutes; this is approximately the time it usually takes to fade in and fade out the current in sham‐controlled tDCS trials in order to produce perceivable sensations on the skin similar to active tDCS (Gandiga 2006), or placement of electrodes with no direct current applied.

Types of outcome measures

Below, we describe the primary and secondary outcomes.

Primary outcomes

The primary outcome was ADL, regardless of their outcome measurement. However, we prioritised generally accepted outcome measures in the following order to facilitate quantitative pooling.

Frenchay Activities Index (FAI) (Schuling 1993)
Barthel ADL Index (BI) (Mahoney 1965)
Rivermead ADL Assessment (Whiting 1980)
Modified Rankin Scale (mRS) (Bonita 1988)
Functional Independence Measure (FIM) (Hamilton 1994)

We analysed primary outcomes according to their time point of measurement as follows: 1) at the end of the study period; and 2) at follow‐up: from three to 12 months after the study end. In cases where included studies reported ADL in other measures than those mentioned above, all review authors discussed and reached consensus about the outcome measures to be included in the primary outcome analysis.

Secondary outcomes

In this update we defined secondary outcomes as upper limb function, lower limb function, muscle strength, cognitive abilities (including spatial neglect), safety, with appropriate measures as reported in the studies. We preferred interval‐scaled outcome measures rather than ordinal‐scaled or nominal‐scaled ones. We prioritised secondary outcome measures as follows.

For upper limb function:

Action Research Arm Test (ARAT) (Lyle 1981);
Fugl‐Meyer Score (Fugl‐Meyer 1975);
Nine‐Hole Peg Test (NHPT) (Sharpless 1982); and
Jebsen Taylor Hand Function Test (JTT) (Jebsen 1969).

For lower limb function:

walking velocity (in metres per second);
walking capacity (metres walked in six minutes); and
Functional Ambulation Categories (FAC) (Holden 1984).

For muscle strength:

grip force (measured by handheld dynamometer) (Boissy 1999); and
Motricity Index Score (Demeurisse 1980).

For cognitive abilities, such as working memory, attention and spatial neglect:

Montreal Cognitive Assessment (Nasreddine 2005);
Clock Drawing Test (Goodglass 1983);
Executive Function (assessments have been described in Chung 2013);
target cancellation (Molenberghs 2011);
line bisection (Molenberghs 2011);
other measures of cognitive abilities; and
other measures of spatial neglect.

For safety:

measured by the number of dropouts and adverse events (including death from all causes).

Depending on the measurements provided in the included trials, all review authors discussed and reached consensus about which outcome measures should be included in the analysis of secondary outcomes.

Search methods for identification of studies

See the methods for the Cochrane Stroke Group Specialised register. We searched for relevant trials in all languages and arranged translation of trial reports where necessary.

Electronic searches

According to the increased scope of this update we re‐ran our searches with updated search strategies of the Cochrane Stroke Group Trials Register (January 2019) and the following electronic bibliographic databases.

Cochrane Central Register of Controlled Trials (CENTRAL; the Cochrane Library; 2019, Issue 1) (Appendix 1)
MEDLINE Ovid (1948 to January 2019) (Appendix 2)
Embase Ovid (1980 to January 2019) (Appendix 3)
CINAHL Ebsco (Cumulative Index to Nursing and Allied Health Literature; 1982 to January 2019) (Appendix 4)
AMED Ovid (1985 to January 2019) (Appendix 5)
Science Citation Index (Web of Science) (1899 to February 2015) (Appendix 6)
Physiotherapy Evidence Database (PEDro) at www.pedro.org.au/ (January 2019) (Appendix 7)
Rehabdata at www.naric.com/?q=REHABDATA (1956 to January 2019) (Appendix 8)
Compendex (Engineering Village by Elsevier; 1969 to January 2019) (Appendix 9)
Inspec (Engineering Village by Elsevier; 1969 to January 2019) (Appendix 9)

We developed the MEDLINE search strategy with the help of the Cochrane Stroke Group Information Specialist and adapted it for the other databases.

We also searched the following ongoing trials and research registers (January 2019).

WHO International Clinical Trials Registry Platform (apps.who.int/trialsearch/)
ClinicalTrials.gov (clinicaltrials.gov)

Searching other resources

We carried out the following additional searches to identify further published, unpublished and ongoing trials not available in the aforementioned databases.

We handsearched the following relevant conference proceedings, which had not already been searched by the Cochrane Stroke Group.
1. 3rd, 4th, 5th, 6th and 7th World Congress of NeuroRehabilitation (2002, 2006, 2008, 2010, 2012, 2014, 2016 and 2018).
2. 1st, 2nd, 3rd, 4th, 5th and 6th World Congress of Physical and Rehabilitation Medicine (2001, 2003, 2005, 2007, 2009, 2011, 2013, 2015, 2017 and 2019).
3. Deutsche Gesellschaft für Neurotraumatologie und Klinische Neurorehabilitation (2001 to 2019).
4. Deutsche Gesellschaft für Neurologie (2000 to 2019).
5. Deutsche Gesellschaft für Neurorehabilitation (1999 to 2019).
6. Asian Oceania Conference of Physical and Rehabilitation Medicine (2008, 2010, 2012, 2014, 2017 and 2019).
We screened reference lists from relevant reviews, articles and textbooks.
We contacted authors of identified trials and other researchers in the field.
We used Science Citation Index Cited Reference Search for forward tracking of important articles.
We contacted the following equipment manufacturers (June 2015).
1. Activatek, Salt Lake City, USA (www.activatekinc.com)
2. Changsha Zhineng Electronics, Changsha City, Hunan, China (www.cszhineng.diytrade.com)
3. DJO Global, Vista, USA (www.djoglobal.com)
4. Grindhouse (www.grindhousewetware.com)
5. Magstim, Spring Gardens, UK (www.magstim.com)
6. Neuroconn, Ilmenau, Germany (www.neuroconn.de)
7. Neuroelectrics, Barcelona, Spain (www.neuroelectrics.com)
8. Newronika, Milano, Italy (www.newronika.it)
9. Soterix Medical, New York City, USA (www.soterixmedical.com)
10. Trans Cranial Technologies, Hong Kong (www.trans-cranial.com)

Data collection and analysis

Selection of studies

One review author (BE) read the titles and abstracts of records identified by the electronic searches and eliminated obviously irrelevant studies. We retrieved the full text articles of the remaining studies, and two review authors (JK and BE) independently ranked the studies as relevant, possibly relevant or irrelevant according to our inclusion criteria (types of studies, participants and aims of interventions). Two review authors (JM and MP) then examined whether the possibly relevant publications fit the population, intervention, comparison, outcome (PICO) strategy of our study question. We included all trials rated as relevant, or possibly relevant, and excluded all trials ranked as irrelevant. We resolved disagreements by discussion with all review authors. If we needed further information to resolve disagreements concerning including or excluding a study, we contacted the trial authors and requested the required information. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009), and listed in the Characteristics of excluded studies table all studies that did not match our inclusion criteria regarding types of studies, participants and aims of interventions.

Data extraction and management

Two review authors (BE and JM) independently extracted trial and outcome data from the selected trials. If one of the review authors was involved in an included trial, another review author extracted trial and outcome data from that trial. In accordance with the 'Risk of bias' tool implemented in Review Manager 5.3 (RevMan 2014) and Review Manager Web, we used a standard data extraction sheet to extract data on:

methods of random sequence generation;
methods of allocation concealment;
blinding of assessors;
use of an intention‐to‐treat (ITT) analysis;
adverse effects and dropouts;
important differences in prognostic factors;
participants (country, number of participants, age, gender, type of stroke, time from stroke onset to study entry and inclusion and exclusion criteria);
comparison (details of interventions in treatment and control groups, duration of treatment and details of cointerventions in the groups);
outcomes; and
investigators' time point of measurement.

Further, we extracted data on initial ADL ability or initial functional ability, or both.

BE and JM checked the extracted data for agreement. If necessary, we contacted trialists to obtain more information.

Assessment of risk of bias in included studies

Two review authors (JM and MP) independently assessed the risk of bias in the included trials, according to Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). We assessed the risk of bias according to the following domains.

Random sequence generation
Allocation concealment
Blinding of participants and personnel
Blinding of outcome assessment
Incomplete outcome data
Selective outcome reporting
Other bias

Two other review authors (JK and MP) checked the extracted data for agreement. All review authors discussed disagreements and, if necessary, sought arbitration by another review author. We judged each domain to be at high, low or unclear risk of bias. We provide a quote from the study report, together with a justification for our judgement, in the 'Risk of bias' table. We summarised the risk of bias judgements across different studies for each of the domains listed.

Measures of treatment effect

For all outcomes that were continuous data, we entered means and standard deviations (SDs). We calculated a pooled estimate of the mean difference (MD) with 95% confidence intervals (CIs). If studies did not use the same outcomes, we calculated standardised mean differences (SMDs) instead of MDs. For all binary outcomes, we calculated risk ratios (RRs) with 95% CIs. Where different scales measured the same outcome, with some using higher values to indicate better performance, and others using lower values, we ensured a consistent direction of the effect across all outcome measurements by multiplying the values of the corresponding scales by ‐1.

For all statistical comparisons, we used the current version of Review Manager 5 (RevMan 2014) and Review Manager Web.

Unit of analysis issues

There were no unit of analysis issues. If studies did not used parallel group designs, e.g. cross‐over RCTs, we only considered the outcomes between groups at the pre‐crossover period.

Dealing with missing data

In case of missing data we extracted data from diagrams or contacted study authors to acquire missing data. If median values and interquartile ranges (IQR) were provided, we estimated their corresponding mean and standard deviation following the approach of Wan 2014.

Assessment of heterogeneity

We used the I² statistic to assess heterogeneity. We used a random‐effects model, regardless of the level of heterogeneity. Thus, when heterogeneity occurred, we could not violate the preconditions of a fixed‐effect model approach.

We considered I² > 50% as representing substantial heterogeneity. If I² > 50%, we explored the individual trial characteristics to identify potential sources of heterogeneity.

Assessment of reporting biases

We tried to minimise reporting bias by using a sensitive search strategy, and by searching for studies in all languages, and by handsearching. Furthermore, we created funnel plots and examined them by visual inspection.

Data synthesis

We undertook meta‐analysis only if we judged participants, interventions, comparisons and outcomes to be sufficiently similar to ensure an answer that is clinically meaningful. If more than one active or sham or control group investigated the same content, we combined these into one group each (e.g. if two sham control groups were included, we combined them into a single sham group for comparison with the active group).

Subgroup analysis and investigation of heterogeneity

If at least two studies were available for each group (tDCS/sham), we conducted planned analyses of the following subgroups for our primary outcome of ADL.

Duration of illness: acute/subacute phase (the first week after stroke and the second to the fourth week after stroke, respectively) versus the postacute phase (from the first to the sixth month after stroke) versus the chronic phase (more than six months after stroke).
Type of stimulation: cathodal versus anodal and position of electrodes/location of stimulation.
Type of control intervention: active (e.g. conventional therapy) versus passive (sham tDCS or no intervention).

All stratified (subgroup) analyses were accompanied by appropriate tests for interaction (statistical tests for subgroup differences as described in the Cochrane Handbook (Higgins 2011b), as implemented in Review Manager 5 (RevMan 2014).

Sensitivity analysis

We incorporated a post hoc sensitivity analysis for methodological quality to test the robustness of our results for our primary outcome ADL. We analysed concealed allocation, blinding of assessors, and ITT.

Summary of findings and assessment of the certainty of the evidence

We created four 'Summary of findings' tables using the following outcomes (two comparisons (tDCS versus sham and tDCS versus active control) at the end of intervention period and at the end of follow‐up (i.e. three months or longer), respectively).

Primary outcome measure: ADL. Measures of activities of daily living. Scale from: 0 to infinity
Secondary outcome measure: upper extremity function. Clinical measures of upper extremity function. Scale from: 0 to infinity
Secondary outcome measure: lower extremity function. Clinical measures of lower extremity function. Scale from: 0 to infinity
Secondary outcome measure: muscle strength. Clinical measures of muscle strength. Scale from: 0 to infinity
Secondary outcome measure: cognitive abilities. Clinical measures of cognitive abilities. Scale from: 0 to infinity
Secondary outcome measure: hemispatial neglect. Clinical measures of hemispatial neglect. Scale from: 0 to infinity
Secondary outcome measure: dropouts, adverse events and deaths (during the intervention period only). Number of adverse events, dropouts and deaths during the intervention period

We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes (Atkins 2004). We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook (Higgins 2011c; Schünemann 2013) using GRADEproGDT software (GRADEproGDT). We justified all decisions to downgrade the quality of studies using footnotes, and we made comments to aid the reader's understanding of the review where necessary.

Results

Description of studies

We describe the included studies as follows.

Results of the search

2013 version

For the 2013 version of this review, we identified 6226 potentially relevant trials through electronic searching; we considered 92 full papers and included 15 trials with 455 participants.

2016 version

For the 2016 version, we identified a total of 2295 new records through the searches. After screening titles and abstracts, we obtained the full‐text of 52 new articles. After further assessment, we determined that 17 new studies met the review's inclusion criteria.

2020 version

In this update, we identified a total of 3407 new records through the searches. After screening titles and abstracts, we obtained the full‐text of 198 new articles. After further assessment, we determined that 35 new studies met the review inclusion criteria, and four studies are awaiting classification, as more information is required. We identified 61 ongoing pilot and large‐scale randomised trials.

The flow of references is shown in Figure 1.

Figure 1

Study flow diagram. Please note that the number of full‐texts is not necessarily equal to the number of studies (e.g. the studies Di Lazzaro 2014a and Di Lazzaro 2014b have been presented in a single full‐text. Moreover there often are several full‐texts of a single trial (e.g. as is the case for Hesse 2011 or Nair 2011).

Included studies

Design

We included 67 studies involving a total of 1729 participants in our qualitative analysis (see Characteristics of included studies). All studies investigated the effects of tDCS versus sham tDCS, except Bang 2015; Cha 2014; Cho 2017; Hamoudi 2018; Hathaiareerug 2019; Lee 2014; Park 2015 and Qu 2009, which compared tDCS with an active comparator. Fifteen trials with 183 participants were randomly assigned cross‐over trials (Au‐Yeung 2014; Boggio 2007a; D'Agata 2016; Fregni 2005a; Fusco 2013a; Klomjai 2018; Jo 2008a; Kang 2008b; Kim 2009; Ko 2008a; Mahmoudi 2011; Manji 2018; Sohn 2013; Sunwoo 2013a; Utarapichat 2018), whereas the remaining 52, with 1546 participants, were RCTs.

Sample sizes

The sample sizes of included studies ranged from four in Boggio 2007a to 96 in Hesse 2011, with a mean (SD) sample size of 26 (18). The median sample size was 20.

Setting

Seventeen of the included studies were conducted in the Republic of Korea, 10 in Italy, seven in the USA, six in China, four in Brazil, three in Thailand, two in Japan, two in Germany, one in Iran, one in Egypt, one in the UK, one in Singapore, one in Belgium, one in Switzerland, and one in Serbia. In three studies, the country was not stated clearly.

Participants

The proportion of participants with ischaemic stroke ranged from 36% in Sohn 2013 to 100% in Fusco 2014. The mean age in the experimental groups ranged from 43 years in Bolognini 2011 to 70 years in Kang 2008b, and from 45 years in Qu 2009 to 75 years in the control groups (Boggio 2007a). The proportion of women participating in the included studies ranged from 0% in Au‐Yeung 2014 and Boggio 2007a to 75% in Danzl 2012. See Table 2 for a comprehensive summary of participant characteristics.

Table 2. Patient characteristics

Study ID	Experimental: age, mean (SD)	Control: age, mean (SD)	Experimental: time post stroke, mean (SD)	Control: time post stroke, mean (SD)	Experimental: sex, n (%)	Control: sex, n (%)	Experimental: lesioned hemisphere, n (%)	Control: lesioned hemisphere, n (%)	Experimental: severity, mean (SD)	Control: severity, mean (SD)	Experimental: lesion cause/ location, n (%)	Control: lesion cause/ location, n (%)	Handedness, n (%)
Allman 2016	60 (12) years	67 (10) years	51 (33) months	57 (40) months	3 (27) female	4 (31) female	3 (27) left	4 (31) left	UE‐FM 39 (16)	UE‐FM 36 (17)	2 (18) cortical	4 (31) cortical	Not stated
Andrade 2017	54 (4) years	55 (4) years	2 (2) months	2 (1) months	18 (45) female	8 (40) female	20 (50) left	10 (50) left	NIHSS 17 (1)	NIHSS17 (1)	15 (38) haemorrhagic, 17 (43) cortical	5 (25) haemorrhagic, 8 (40) cortical	Not stated
Ang 2012	52 (12) years	56 (10) years	3 (2) years	3 (1) years	4 (40) female	1 (11) female	5 (50) left	6 (67) left	UE‐FM 35 (8)	UE‐FM 33 (8)	6 (60) ischaemic; 1 (10) cortical, 9 (90) subcortical	7 (78) ischaemic; 9 (100) subcortical	Not stated
Au‐Yeung 2014	63 (6) years		8 (3) years		0 female		5 (50) left		UE‐FM 58 (8); MMSE 29 (2)		8 (80) ischaemic		10 (100) right‐handed
Bang 2015	66 (4) years	66 (5) years	7 (2) weeks	7 (1) weeks	2 (50) female	2 (50) female	6 (100) right	6 (100) right	MBI 51 (5)	MBI 50 (6)	Not described		Not stated
Boggio 2007a	56 (11) years	75 (NA) years	33 (34) months	39 months	3 (100) male	1 (100) male	2 (67) left	1 (100) left	MRC 4.2 (0.53)	MRC 4.7 (NA)	3 (100) ischaemic and subcortical	1 (100) ischaemic and subcortical	12 (100) right‐handed
Bolognini 2011	43 (13) years	51 (15) years	44 (31) months	26 (18) months	4 (57) female	5 (71) female	4 (57) left	4 (57) left	BI 18.13 (2.42)	BI 14.33 (5.46)	2 (29) haemorrhagic, 5 (71) ischaemic	7 (100) ischaemic	14 (100) right‐handed
Cha 2014	60 (11) years	58 (10) years	14 (5) months	15 (4) months	Not stated	Not stated	4 (40) left	5 (50) left	Brunnstrom 5 (1)	Brunnstrom 5 (1)	Not stated	Not stated	Not stated
Chang 2015	60 (10) years	66 (11) years	16 (6) days	17 (5) days	9 (38) female		6 (50) left	5 (42) left	NIHSS 7 (4)	NIHSS 9 (5)	24 (100) ischaemic/11 (46) corona radiata, 7 (29) MCA, 4 (17) MCA border zone, 2 (8) internal capsule		Not stated
Chelette 2014	58(7) years	62 (5) years	5 (2) years	5 (1) years	9 (45) female	1 (17) female	14 (70) left	1 (17) left	SIS 62 (13)	SIS 57 (18)	16 (80) ischaemic/17 (81) cortical	6 (100) ischaemic/4 (67) cortical	16 (80) right‐handed
Cho 2017	61 (9) years	58 (13) years	14 (6) days	14 (5) days	6 (40) female	7 (47) female	7 (47) left	7 (47) left	UE‐FM 50 (19)	UE‐FM 41 (13)	12 (75) ischaemic/5 (33) cortical	13 (87) ischaemic/4 (27) cortical	Not stated
Cunningham 2015	64 (8) years	59 (10) years	63 (81) months	37 (27) months	2 (33) female	2 (33) female	2 (33) left	4 (67) left	UE‐FM 41 (14)	UE‐FM 47 (11)	2 (33) haemorrhagic	2 (33) haemorrhagic	Not stated
D'Agata 2016	57 (12) years	65 (12) years	41 (39) months	37 (32) months	8 (33) female	3 (39) female	12 (50) left	6 (60) left	Not described clearly		17 (71) ischaemic/6 (25) cortical, 15 (62) subcortical, 3 (13) corticosubcortical	7 (70) ischaemic/1 (10) cortical, 8 (80) subcortical, 1 (10) corticosubcortical	Not stated
Danzl 2012	65 (15) years	71 (11) years	57 /55) months	39 (33) months	1 (25) female	3 (75) female	4 (100) left	4 (100) left			2 (50) ischaemic/not described	4 (100) ischaemic/not described	Not stated
Di Lazzaro 2014a	66 (16) years	71 (14) years	3 (1) days	3 (1) days	2 (29) female	3 (43) female	3 (43) left	3 (43) left	NIHSS 7 (5)	NIHSS 7 (4)	7 (100) ischaemic; 3 (43) subcortical; 4 (57) corticosubcortical	7 (100) ischaemic; 2 (29) subcortical, 5 (71) corticosubcortical	Not stated
Di Lazzaro 2014b	61 (16) years	69 (12) years	3 (2) days	3 (1) days	4 (40) female	6 (60) male	2 (20) left	6 (60) left	NIHSS 6 (3)	NIHSS 6 (2)	10 (100) ischaemic; 4 (40) subcortical, 6 (60) corticosubcortical	10 (100) ischaemic; 4 (40) subcortical, 6 (60) corticosubcortical	Not stated
Fregni 2005a	54 (17) years		27 (24) months		2 (33) female		3 (50) left		MRC 4.18 (0.37)		Cause not clearly stated by the authors		6 (100) right‐handed
Fusco 2013a	44 (16) years	65 (22) years	31 (13) days	25 (5) days	3 (60) female	1 (25) female	3 (60) left	2 (50) left	Grasp force 17.83 (7.45) kg		5 (100) ischaemic	3 (75) ischaemic, 1 (25) haemorrhagic	9 (100) right‐handed
Fusco 2014	56 (15) years	60 (12) years	19 (8) days		3 (60) female	3 (50) female	2 (40) left	2 (33) left	BI 33 (22)	BI 51 (34)	5 (100) ischaemic	6 (100) ischaemic	9 (73) right‐handed
Geroin 2011	64 (7) years	63 (6) years	26 (6) months	27 (5) months	2 (20) female	4 (40) female	Not stated by the authors	Not stated by the authors	ESS 79.6 (4.1)	ESS 79.6 (2.7)	10 (100) ischaemic; 4 (40) cortical, 3 (30) corticosubcortical, 3 (30) subcortical	10 (100) ischaemic; 5 (50) cortical, 3 (30) corticosubcortical, 2 (20) subcortical	Not stated by the authors
Hamoudi 2018	62 (13) years	62 (13) years for sham tDCS and 65 (2) for passive control group	48 (80) months	44 (51) months for sham tDCS and 23 (4) months for passive control group	6 (33) female	3 (17) and 6 (43) female	9 (50) left	8 (44) and 7 (50) left	UE‐FM 59 (4)	UE‐FM 59 (4) and 59 (4)	18 (100) ischaemic/9 (50) subcortical	18 (100) ischaemic/9 (50) subcortical and 14 (100) ischaemic/7 (50) subcortical	EHI 78 in the Exp group, EHI 84 in the Sham group and EHI 90 in the Ctl group
Hathaiareerug 2019	56 (8) years	59 (10) years	6 (4) months	5 (3) months	1 (11) female	2 (22) female	4 (44) left	2 (22) left	UE‐FM 38 (17)	UE‐FM 32 (14)	6 (67) ischaemic/1 (11) cortical, 4 (44) subcortical, 4 (44) corticosubcortical	7 (77) ischaemic/1 (11) cortical, 3 (33) subcortical, 5 (55) corticosubcortical	89% right‐handed
Hesse 2011	65 (10) years	66 (10) years	4 (2) weeks	4 (2) weeks	26 (41) female	11 (34) female	35 (55) left	16 (50) left	BI 34.15 (6.97); UE‐FM 7.85 (3.58)	BI 35.0 (7.8); UE‐FM 8.2 (4.4)	64 (100) ischaemic; 29 (45) TACI, 20 (31) PACI, 15 (23) LACI	32 (100) ischaemic; 13 (41) TACI, 13 (41) PACI, 6 (18) LACI	Not stated by the authors
Ilić 2016	58 (8) years	62 (4) years	41 (24) months	37 (21) months	10 (71) female	7 (58) female	13 (50) left		UE‐FM 47 (8)	UE‐FM 51 (6)	26 (100) ischaemic/26 (100) subcortical		24 (92) right‐handed
Jo 2008a	48 (9) years		2 (1) months		3 (30) female		10 (100) right		Not reported		4 (40) ischaemic		Not stated by the authors
Kang 2008b	70 (3) years		544 (388) days		4 (40) female		7 (70) right		21 (1) MMSE		7 (70) ischaemic		Not stated by the authors
Khedr 2013	59 (9) years	57 (8) years	13 (5) days	13 (5) days	9 (33) female	5 (38) female	12 (44) left	6 (46) left	BI 32.76 (10.75)	BI 31.1 (12.6)	27 (100) ischaemic; 12 (44) cortical, 5 (19) corticosubcortical, 10 (37) subcortical	13 (100) ischaemic; 6 (42) cortical, 3 (23) corticosubcortical, 4 (31) subcortical	Not stated by the authors
Kim 2009	63 (13) years		6 (3) weeks		7 (70) female		8 (80) left		MRC between 3 and 5 for the all paretic finger flexors and extensors		8 (80) infarction, 2 (20) haemorrhage		Not stated by the authors
Kim 2010	54 (15) years	63 (9) years	27 (21) days	23 (8) days	2 (18) female	3 (43) female	7 (64) left	2 (29) left	BI 71.77 (23.86) UE‐FM 34.7 (15.0)	BI 67.9 (22.4) UE‐FM 41.0 (13.0)	11 (100) ischaemic; 3 (27) cortical, 3 (27) corticosubcortical, 5 (71) subcortical	7 (100) ischaemic; 2 (29) cortical, 1 (14) corticosubcortical, 4 (57) subcortical	Not stated by the authors
Kim 2016	59 (13) years	52 (11) years	15 (6) months	15 (7) months	5 (33) female	6 (40) female	8 (53) left	7 (47) left	FIM 67 (10)	FIM 80 (11)	4 (27) ischaemic/not stated	10 (67) ischaemic/not stated	Not stated by the authors
Ko 2008a	62 (9) years		29‐99 days		5 (33) female		15 (100) right		19 per cent deviation (11)		10 (66) ischaemic		15 (100) right‐handed
Koo 2018	52 (3) years	59 (3) years	19 (8) months	20 (8) months	7 (58) female	6 (50) female	6 (50) left	8 (75) left	MBI 35 (16)	MBI 38 (20)	4 (33) ischaemic; 3 (25) cortical, 9 (75) subcortical	7 (58) ischaemic;2 (17) cortical, 8 (67) subcortical, 2 (17) brain stem	24 (100) right handed
Klomjai 2018	57 (12) years		3 (2) months		5 (26) female		12 (63) right		TUG 21 (13) s	TUG 20 (13) s	19 (100) ischaemic		16 (84) right‐handed
Lee 2014	62 (11) years	61 (14) years	18 (8) days	17 (6) days	17 (44) female	9 (45) female	19 (49) left	13 (65)	UE‐FM 37 (23)	UE‐FM 35 (22)	21 (54) ischaemic; 21 (54) cortical; 18 (46) subcortical	14 (70) ischaemic; 10 (50) cortical; 10 (50) subcortical	Not stated by the authors
Lindenberg 2010	62 (15) years	56 (13) years	31 (21) months	40 (23) months	2 (20) female	3 (30) female	6 (60) left	7 (70) left	UE‐FM 38.2 (13.3)	UE‐FM 39.8 (11.5)	10 (100) ischaemic	10 (100) ischaemic	19 (95) right‐handed, 1 (5) both‐handed
Mahmoudi 2011	61 (14) years		8 (5) months		3 (33) female		6 (60) left, 3 (30) right, 1 (10) brainstem		JTT (without handwriting): 12.3 (7.3) s		10 (100) ischaemic		Not stated by the authors
Manji 2018	62 (10) years	64 (11) years	4 (2) months	5 (1) months	5 (33) female	4 (27) female	Not reported		FIM 107 (10)	FIM 104 (10)	9 (60) ischaemic	8 (16) ischaemic	Not stated by the authors
Mazzoleni 2019	68 (16) years	69 (16) years	Not reported		12 (60) female	12 (63) female	11 (55) left	11 (58) left	CMMSA 4.3 (1.4)	CMMSA 5.1 (1.1)	13 (65) ischaemic	16 (84) ischaemic	38 (97) right‐handed
Mortensen 2016	66 (11) years	61 (10) years	32 (16) months	29 (15) months	4 (50) female	2 (29) female	4 (50) left	4 (57) left	JTT 69 (29) s	JTT 55 (18) s	0 ischaemic	0 ischaemic	Not stated by the authors
Nair 2011	61 (12) years	56 (15) years	33 (20) months	28 (28) months	2 (29) female	3 (43) female	3 (43) left	5 (71) left	UE‐FM 30 (11)	UE‐FM 31 (10)	7 (100) ischaemic; 5 (71) cortical and corticosubcortical, 2 (29) subcortical	7 (100) ischaemic; 4 (56) cortical and corticosubcortical, 3 (43) subcortical	14 (100) right‐handed
Nicolo 2017	65 (12) years	64 (17) years	1 (0.4) months	1 (0.3) months	13 (46) female	5 (38) female	4 (29) left	5 (36) left	NIHSS 13 (6)	NIHSS 12 (5)	13 (46) ischaemic; 4 (14) cortical, 16 (67) corticosubcortical, 8 (29) subcortical	10 (71) ischaemic; 1 (8) cortical, 6 (46) corticosubcortical, 6 (46) subcortical	39 (95) right‐handed
Park 2013	65 (14) years	66 (11) years	29 (19) days	25 (17) days	6 (67) female	2 (40) female	2 (33) left	2 (40) left	NIHSS 8 (3)	NIHSS 10 (3)	4 (67) ischaemic	3 (60) ischaemic	Not stated by the authors
Park 2015	59 (6) years	60 (13) years	19 (12) months	24 (16) months	Not reported		9 (56) left	3 (19) left	Gait speed 0.7 (0.3) m/s	Gait speed 0.6 (0.3) m/s	4 (25) ischaemic	4 (50) ischaemic	Not stated by the authors
Picelli 2015	64 (9) years	61 (7) years	57 (35) months	55 (33) months	6 (30) female	2 (20) female	Not reported		6MWT 181 (79) m	6MWT 183 (51) m	7 (35) cortical; 7 (35) corticosubcortical; 6 (30) subcortical	4 (40) cortical; 4 (40) corticosubcortical; 2 (20) subcortical	Not stated by the authors
Qu 2009	45 (11) years	45 (14) years	6 (range 3 to 36) months	4 (range 3 to 12) months	4 (16) female	3 (12) female	14 (56) left	13 (52) left	BI 64 (17)	BI 72 (22)	10 (40) haemorrhagic, 15 (60) infarction	10 (40) haemorrhagic, 15 (60) infarction	Not stated by the authors
Qu 2017	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described
Rabadi 2017	62 (11) years	63 (6) years	7 (4) days	6 (3) days	0 female	0 female	4 (50) left	2 (25) left	FIM 61 (17)	FIM 59 (12)	8 (100) ischaemic	8 (100) ischaemic	15 (94) right‐handed
Rocha 2016	58 (range 41‐71) years	59 (range 46‐70) years	31 months (range 9‐67)	27 months (6‐46)	3 (21) female	3 (43) female	8 (57) left	3 (43) left	UE‐FM 48 (6)	UE‐FM 51 (9)	Not stated by the authors		21 (100) right‐handed
Rossi 2013	66 (14) years	70 (14) years	2 days	2 days	13 (52) female	11 (44) female	18 (72) left	16 (64) left	UE‐FM 4.1 (6.4)	FM 4.6 (7.8)	25 (100) ischaemic; 1 (4) cortical, 17 (68) corticosubcortical, 7 (28) subcortical	25 (100) ischaemic; 2 (8) cortical, 18 (72) corticosubcortical, 5 (20) subcortical	Not stated by the authors
Saeys 2015	62 (10) years	65 (7) years	46 (22) days	38 (15) days	7 (44) female	7 (47) female	11 (92) left	6 (55) left	Tinetti 8 (7)	Tinetti 9 (6)	15 (94) ischaemic	11 (73) ischaemic	Not stated by the authors
Salazar 2019	60 (10) years	56 (16) years	21 months (range 6‐59)	23 months (range 8‐59)	5 (33) female	5 (33) female	8 (53) left	8 (53) left	median UE‐FM 25 points (range 9‐46)	median UE‐FM 29 (range 16‐46)	14 (93) ischaemic	11 (73) ischaemic	27 (90) right handed
Sattler 2015	68 (10) years	63 (12) years	5 (3) days	6 (4) days	3 (30) female	3 (30) female	Not exactly described		NIHSS 3 (1); UE‐FM 47 (3)	NIHSS 3 (2), UE‐FM 49 (3)	Not exactly described		All patients were right handed
Seo 2017	61 (9) years	63 (9) years	76 (83) months	153 (123) months	2 (18) female	3 (30) female	6 (55) left	2 (20) left	MRS 3 (0.5)	MRS 3 (0.4)	9 (82) ischaemic	7 (70) ischaemic	Not stated by the authors
Shaheiwola 2018	49 (9) years	52 (11) years	18 (15) months (median(IQR))	16(13) months (median(IQR))	1 (7) female	2 (13) female	7 (47) left	9 (60) left	UE‐FM 16 (9)	UE‐FM 18 (13)	Not exactly described
Sik 2015	60 (IQR 54‐68) years	60 (IQR 55‐67) years	22 (32) months (median(IQR))	18 (19) months (median(IQR))	10 (50) female	3 (27) female	10 (50) left	5 (45) left	Not exactly described		19 (95) ischaemic	10 (91) ischaemic	Not stated by the authors
Sohn 2013	58 (15) years		63 (17) days		2 (18) female		6 (55) left		Not stated by the authors		4 (36) ischaemic		Not stated by the authors
Straudi 2016	53 (16) years	64 (10) years	41 (35) weeks	78 (62) weeks	7 (58) female	4 (36) female	9 (75) left	6 (55) left	UE‐FM 28 (19)	UE‐FM 37 (14)	7 (83) ischaemic; 9 (75) cortical, 3 (25) subcortical	9 (82) ischaemic; 5 (45) cortical, 6 (55) subcortical	Not stated by the authors
Sunwoo 2013a	63 (13) years		28 (60) months		6 (60) female		10 (100) left		MMSE 28 (2)		7 (70) ischaemic		10 (100) right‐handed
Tahtis 2012	67 (12) years	56 (12) years	20 (5) days	25 (11) days	2 (29) female	1 (14) female	3 (43) left	3 (43) left	MRS 2 (1)	MRS 3 (1)	7 (100) ischaemic; 4 (57) cortical, 3 (43) subcortical	7 (100) ischaemic; 3 (43) cortical; 4 (57) subcortical	Not stated by the authors
Tedesco Triccas 2015b	64 (10) years	63 (14) years	25 (31) months	13 (16) months	5 (42) female	4 (33) female	6 (50) left	5 (45) left	UE‐FM 28 (19)	UE‐FM 37 (14)	3 (25) ischaemic; 3 (25) cortical, 9 (75) subcortical	9 (81) ischaemic; 4 (36) cortical; 7 (64) subcortical	22 (96) right‐handed
Utarapichat 2018	57 (12) years		34 (19) months		4 (40) female		5 (50) left		MRC knee extensor 4		10 (100) ischaemic		Not stated by the authors
Viana 2014	56 (10) years	55 (12) years	32 (18) months	35 (20) months	1 (10) female	3 (30) female	5 (50) left	3 (30) left	UE‐FM 41 (16)	UE‐FM 39 (17)	9 (90) ischaemic	10 (100) ischaemic	19 (95) right‐handed
Wang 2014	54 (14) years	52 (9) years	Not explicitly stated, but all participants were enrolled between 1 and 4 weeks post stroke		1 (16) female	1 (33) female	2 (33) left	0 left	FIM 59 (18)	FIM 74 (8)	6 (100) ischaemic	3 (100) ischaemic	Not stated by the authors
Wong 2015	69 (10) years		11 (5) days		11 (65) female		Not explicitly stated		Not explicitly stated		Not stated by the authors		Not stated by the authors
Wu 2013a	46 (11) years	49 (13) years	5 (3) months	5 (3) months	11 (24) female	10 (22) female	24 (53) left	23 (51) left	BI 55 (range 0 to 85) UE‐FM 12.3 (5.5)	BI 55 (range 25 to 95) UE‐FM 11.8 (8.2)	27 (60) ischaemic, 18 (40) haemorrhagic	26 (58) ischaemic, 19 haemorrhagic (42)	Not stated by the authors
Yi 2016	62 (11) years	62 (10) years	Not stated		5 (25) female	4 (40) female	None	None	Not stated	Not stated	Not explicitly stated	Not explicitly stated	Not stated by the authors
Yun 2015	60 (14) years	69 (15) years	1.5 (1) months	1.5 (1) months	17 (57) female	8 (53) female	11 (37) left	4 (27) left	Not explicitly stated	Not explicitly stated	Not explicitly stated	Not explicitly stated	Not stated by the authors

BBT: Box and Block Test
BI: Barthel Index
CMMSA: Chedoke McMaster Stroke Assessment
ESS: European Stroke Scale
IQR: Interquartile Range
JTT: Jebsen Taylor Hand Function Test
LACI: lacunar stroke
MRC: Medical Research Council
NA: not applicable
NIHSS: National Institute of Health Stroke Scale
PACI: partial anterior circulation stroke
SD: standard deviation
TACI: total anterior circulation stroke
UE‐FM: Upper Extremity Fugl‐Meyer Score

Interventions

The experimental groups received anodal stimulation (A‐tDCS) (Allman 2016; Andrade 2017; Au‐Yeung 2014; Boggio 2007a; Bolognini 2011; Cha 2014; Chang 2015; Chelette 2014; Cunningham 2015; Danzl 2012; Fregni 2005a; Fusco 2013a; Geroin 2011; Hamoudi 2018; Hesse 2011; Ilić 2016; Jo 2008a; Kang 2008b; Khedr 2013; Kim 2009; Kim 2010; Kim 2016; Ko 2008a; Koo 2018; Mahmoudi 2011; Manji 2018; Mazzoleni 2019; Mortensen 2016; Park 2013; Park 2015; Picelli 2015; Rossi 2013; Seo 2017; Shaheiwola 2018; Sik 2015; Sohn 2013; Sunwoo 2013a; Tedesco Triccas 2015b; Utarapichat 2018; Viana 2014; Wang 2014; Wong 2015; Yi 2016); cathodal stimulation (C‐tDCS) (Au‐Yeung 2014; Boggio 2007a; Chelette 2014; Cho 2017; Fregni 2005a; Fusco 2013a; Fusco 2014; Hesse 2011; Khedr 2013; Kim 2010; Lee 2014; Mahmoudi 2011; Nair 2011; Nicolo 2017; Qu 2009; Qu 2017; Rabadi 2017; Wu 2013a; Yi 2016); or dual‐tDCS (anodal plus cathodal stimulation simultaneously) (Ang 2012; Bang 2015; Chelette 2014; D'Agata 2016; Di Lazzaro 2014a; Di Lazzaro 2014b; Fusco 2013a; Hathaiareerug 2019; Klomjai 2018; Lindenberg 2010; Mahmoudi 2011; Salazar 2019; Sik 2015; Straudi 2016; Sunwoo 2013a; Tahtis 2012). The control groups of all but eight included studies received sham tDCS. The remaining eight studies received physical therapy, occupational therapy, mirror therapy or virtual reality as a control intervention (Bang 2015; Cha 2014; Cho 2017; Hamoudi 2018; Hathaiareerug 2019; Lee 2014; Park 2015 Qu 2009). See Table 3 for a comprehensive summary of intervention characteristics, dropouts and adverse events.

See: 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; 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

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

Study ID	Type of intervention/ stimulation (polarity)	Electrode position and size	Reference electrode position	Treatment intensity		Base treatment	Dropouts	Adverse events	Source of information
Allman 2016	A‐tDCS	5 x 7–cm electrodes, encased in saline‐soaked sponges with the anode placed over ipsilesional primary motor cortex (5 cm lateral to Cz: C3) and the cathode over the contralateral supraorbital ridge		1 mA for 20 minutes	Base treatment plus 20 minutes of A‐tDCS or sham tDCS	Daily self‐administered Graded Repetitive Arm Supplementary Program (GRASP) training for 60 minutes over 9 days	2 (15%) in the EXP group due to organizational issues	Not described explicitely	Published
Allman 2016	Sham tDCS			1 mA for 10 seconds	Base treatment plus 20 minutes of A‐tDCS or sham tDCS		2 (15%) in the EXP group due to organizational issues	Not described explicitely	Published
Andrade 2017	A‐tDCS	6.4 x 2.5 cm anode over premotor cortex	On the supraorbital region in the contralateral hemisphere	0.7 mA (duration not described)	Base treatment plus unknown duration of A‐tDCS over PMC or M1 or sham tDCS	CIMT on a 3‐hour daily protocol of motor skills training for two weeks, supervised by a blinded physiotherapist (restriction of 90% of waking hours)	None	16 out of 60 patients reported mild side effects after stimulation (7 in the M1 group, 6 in PMC group, and 3 in the sham group): skin redness under the site of stimulation (5 in M1 group, 4 in PMC group, and 3 in sham group), mild headache (3 in M1 group and 2 in PMC group), and sleepiness (1 in PMC group). In all groups some subjects experienced multiple adverse effects.	Published
	A‐tDCS	6.4 x 2.5 cm anode over M1
	Sham tDCS	Not described
Ang 2012	Dual‐tDCS	Saline‐soaked sponge electrodes with the anode placed over M1 of the affected hemisphere and the cathode placed over M1 the unaffected hemisphere (size not stated)		1 mA for 20 minutes	20 minutes of dual‐tDCS or sham tDCS followed by 8 minutes of evaluation prior to base treatment	60 minutes of therapy using EEG‐based MI‐BCI with robotic feedback with the MIT‐Manus 5 times a week for 2 weeks	None	Unclear	Published
Ang 2012	Sham tDCS			1 mA for 30 seconds			None	Unclear	Published
Au‐Yeung 2014	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS, C‐tDCS and sham tDCS once in random order with at least 5 days wash‐out period	None	None	Unclear	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of both hemispheres		1 mA for 10 seconds
Bang 2015	Dual tDCS	Anodal sponge electrode of 35cm² was attached to the right posterior parietal cortex (P4) and accompanied by cathode tDCS of the second circuit was positioned over the left posterior parietal cortex (P3). Therefore, in the first tDCS circuit, the anode was placed over P4 and the cathode was placed over the left supraorbital area		1 mA for 20 minutes	Base treatment either with or without Dual tDCS	Mirror‐based feedback training	Not described	Unclear	Published
Bang 2015	Feedback training			NA	Base treatment either with or without Dual tDCS	Mirror‐based feedback training	Not described	Unclear	Published
Boggio 2007a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS, C‐tDCS or sham tDCS 4 days once a day	None	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Bolognini 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes; with the anode placed over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Base treatment + A‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks	CIMT up to 4 hours/day for 5 days a week for 2 consecutive weeks	7 (33%) due to frustration and tiredness during assessments (Bolognini 2013 [pers comm]); these participants have been excluded from analysis and presentation of results	None	Published and unpublished
Bolognini 2011	Sham tDCS			2 mA for 30 seconds				None	Published and unpublished
Cha 2014	A‐tDCS	Water‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base treatment + A‐tDCS for 20 minutes	Basic training for improving function of upper and lower extremities for 30 minutes per day, 5 times a week for four weeks	None	Unclear	Published
Cha 2014	PT	NA		NA	NA		None	Unclear	Published
Chang 2015	A‐tDCS	Saline‐soaked sponge surface electrodes with the 7 cm² anode over the tibialis anterior area of precentral gyrus of affected hemisphere	Saline‐soaked sponge surface electrodes with the 28 cm² cathode over the contralateral supraorbital area	2 mA for 10 minutes	Base treatment + either A‐tDCS or sham tDCS for 20 minutes	Conventional physical therapy	Not reported	Unclear	Published
Chang 2015	Sham tDCS			2 mA for 15 seconds	Base treatment + either A‐tDCS or sham tDCS for 20 minutes	Conventional physical therapy	Not reported	Unclear	Published
Chelette 2014	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1.4 mA for 20 minutes	Either A‐tDCS, C‐tDCS, dual tDCS or sham tDCS prior to base treatment	3 hours of intensive, task‐oriented UE motor training (a modified constraint‐based protocol)	Not reported	Unclear	Published
	C‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode contralesional supraorbital	35 cm² saline‐soaked sponge electrodes with the cathode over contralesional M
	Dual tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode over contralesional M1
	Sham tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1.4 mA for 30 seconds
Cho 2017	C‐tDCS	35 cm² wet sponge electrodes with the cathode over contralesional M1	35 cm² wet sponge electrodes with the anode contralesional supraorbital	2 mA for 20 minutes	Either base treatment plus C‐tDCS or base treatment only daily for 2 weeks	10 Hz and 90% rMT for 5 seconds with a 55‐second inter‐train interval, 90% of rMT intensity	None	No serious adverse events occured	Published
Cho 2017	rTMS	rTMS over ipsilesiona lM1 of the hand		1000 pulses over 20 min			None	No serious adverse events occured	Published
Cunningham 2015	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional PMC and SMA, identified with neuronavigation	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1 mA for 30 minutes	A‐tDCS or sham tDCS during each rehabilitation session	CIMT for 30 minutes, 3 times per week for 5 weeks with supervision from a physical therapist. Intensive functional exercises were performed via a graded, regimented, feedback‐driven approach. Patient‐specific goals were emphasized. Patients were asked to restrain the non‐paretic upper limb by placing it in a mitt for 2 hours every weekday while performing home exercises. Exercise log was monitored at each session	None	Unclear	Published
Cunningham 2015	Sham tDCS			1 mA for 30 seconds	A‐tDCS or sham tDCS during each rehabilitation session		None	Unclear	Published
D'Agata 2016	rTMS + dual tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 20 minutes	1a. 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	rTMS@1Hz at 80% rMT for 15 min (900 stimuli) over the non‐lesioned M1 of the hand area	Not clearly stated	Unclear	Published
	Dual tDCS + mirror therapy			1.5 mA for 20 minutes		Mirror box training with the plegic hand (3 series of 25 repetitions of 6 different movements)
	Sham tDCS + mirror therapy			Not described
Danzl 2012	A‐tDCS	25 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1 of the leg and the anode over the contralateral supraorbital forehead		2 mA for 20 minutes	A‐tDCS or sham tDCS prior to base treatment	Robot‐assisted walking training (20 to 40 minutes) 3 times per week for 4 weeks	2 (20%): 1 in the A‐tDCS and 1 in the sham group due to knee pain and contractures	None	Published
Danzl 2012	Sham tDCS			2 mA for 75 seconds	A‐tDCS or sham tDCS prior to base treatment			None	Published
Di Lazzaro 2014a	Dual‐tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Dual‐tDCS or sham tDCS on 5 continuous days	None	None	Unclear	Published
Di Lazzaro 2014a	Sham tDCS			2 mA for 30 seconds	Dual‐tDCS or sham tDCS on 5 continuous days	None	None	Unclear	Published
Di Lazzaro 2014b	Dual‐tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Base treatment + dual‐tDCS or sham tDCS on 5 continuous days	CIMT for at least 90% of waking hours, including 1.5 hours per day arm training	None	Unclear	Published
Di Lazzaro 2014b	Sham tDCS			2 mA for 30 seconds			None	Unclear	Published
Fusco 2013a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 15 minutes	1 active tDCS (A‐tDCS, C‐tDCS, dual‐tDCS) and 1 sham tDCS session in 2 consecutive days	None	None	None	Published and unpublished
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 15 minutes
	Dual‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 15 minutes
	Sham tDCS	Not described by the authors
Fusco 2014	C‐tDCS	Saline‐soaked 35 cm² gel‐sponge electrodes with the cathode over M1 of the non‐lesioned hemisphere	Above the right shoulder	1.5 mA for 10 minutes	Each participant underwent C‐tDCS and sham tDCS on 5 consecutive days each week for 2 weeks prior to a rehabilitative session in randomised order	Patient‐tailored motor rehabilitation focusing on recovery of upper limb for 45 minutes twice a day	2 (14%); reasons not described by the authors	Unclear	Published
Fusco 2014	Sham tDCS		Above the right shoulder	Not described			1 (7%); emergency transfer to another hospital	Unclear	Published
Fregni 2005a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent A‐tDCS, C‐tDCS and sham tDCS once, separated by at least 48 hours of rest	None	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Geroin 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 7 minutes	Base treatment + A‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks	50‐minute training sessions 5 days a week for 2 consecutive weeks, consisting of 20 minutes of robot‐assisted gait training and 30 minutes of lower limb strength and joint mobilisation training	None	None	Published
Geroin 2011	Sham tDCS		Over the contralateral supraorbital forehead	0 mA for 7 minutes			None	None	Published
Hamoudi 2018	A‐tDCS	25 cm² anode over ipsilesional M1 hotspot	25 cm² cathode over the contralateral supraorbital forehead	1.2 mA for 20 minutes	Either base treatment + A‐tDCS or sham tDCS or passive control group	Computerised grip strength training for 45 minutesper day for 5 days	No dropouts during intervention phase	1 (6) migraine, 1 (6) tingling sensation of the unaffected hand	Published
	Sham tDCS	25 cm² anode over ipsilesional M1 hotspot	25 cm² cathode over the contralateral supraorbital forehead	1.2 mA for 30 seconds				3 (17) mild headache, 1 (6) phosphene, 1 (6) abdominal pain, 1 (6) retching
	Passive control group	NA				No base treatment		None
Hathaiareerug 2019	Dual tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere	Saline‐soaked 35 cm² sponge electrodes with the cathode over M1 of the non‐lesioned hemisphere	2 mA for 20 minutes	Base treatment + either dual tDCS or electro‐acupuncture once a week for 3 weeks	Intensive physical therapy and occupational therapy performed in hourly sessions 3 times per week for 3 weeks	1 (11) dropped out during follow‐up	Unclear	Published
Hathaiareerug 2019	Electro‐acupuncture	NA					None	Unclear	Published
Hesse 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS 5 days a week for 6 consecutive weeks	20 minutes of robot‐assisted arm training 5 days a week for 6 consecutive weeks	11 (11%); 7 dropouts in the EXP‐groups: 1 (14%) during intervention period due to pneumonia and 6 (86%) until 3 months of follow‐up (2 deaths due to myocardial infarction and stent surgery, 3 due to being unavailable and 1 due to refusal of further enrolment); 4 dropouts in the CTL group: 3 (75%) due to being not available and 1 (25%) due to refusal of further enrolment	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the non‐lesioned hemisphere		2 mA for 20 minutes
	Sham tDCS	As in the A‐tDCS or the C‐tDCS group (changing consecutively)		0 mA for 20 minutes
Ilić 2016	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 hand area of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + either A‐tDCS or sham tDCS prior	Intensive task oriented training, delivered by OT and consisting of strength training, ROM exercises, manipulation exercises, pinch grip, grasp, release and simulating ADL	1 dropout in the sham group (reason not stated)	None	Published
Ilić 2016	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 60 seconds	Base treatment + either A‐tDCS or sham tDCS prior		1 dropout in the sham group (reason not stated)	None	Published
Jo 2008a	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over DLPFC of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 30 minutes	A‐tDCS once and sham tDCS once or vice versa, separated by at least 48 hours of resting period	None	None	6 Quote: "Transient aching or burning sensations were reported in six cases, and transient skin redness at the electrode contact site was reported in three cases."	Published
Jo 2008a	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 10 seconds		None	None		Published
Kang 2008b	A‐tDCS	25 cm² electrodes over the left DLPFC	Over the contralateral supraorbital forehead	2 mA for 20 minutes	A‐tDCS and sham tDCS or vice versa, separated by at least 48 hours of resting period	None	Not described	Unclear	Published
Kang 2008b	Sham tDCS	25 cm² electrodes over the left DLPFC	Over the contralateral supraorbital forehead	2 mA for 1 minute		None	Not described	Unclear	Published
Khedr 2013	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 25 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS for 6 consecutive days after	Rehabilitation program within 1 hour after each tDCS session, starting with passive movement and range of motion exercise up to active resistive exercise	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes, cathode over M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 25 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 2 minutes
Kim 2009	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent A‐tDCS and sham tDCS, separated by at least 24 hours of rest	None	None	None	Published and unpublished
Kim 2009	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds		None	None	None	Published and unpublished
Kim 2010	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the lesioned hemisphere (as confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks at the beginning of each therapy session	Occupational therapy according to a standardised protocol aimed at improving paretic hand function for 30 minutes 5 days a week for 2 consecutive weeks	2 of 20; 1 participant discontinued treatment because of dizziness and another because of headache (authors did not state corresponding groups)	Two	Published
	C‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the non‐lesioned hemisphere (confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 20 minutes
	Sham tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the lesioned hemisphere (confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 1 minutes
Kim 2016	A‐tDCS	Saline‐soaked 24 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base treatment + either A‐tDCS or sham tDCS	Traditional occupational therapy treatment	Not described	Unclear	Published
Kim 2016	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds	Base treatment + either A‐tDCS or sham tDCS	Traditional occupational therapy treatment	Not described	Unclear	Published
Ko 2008a	A‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes over right (lesioned) PPC	Over the contralateral supraorbital forehead	2 mA for 20 minutes	A‐tDCS once and sham tDCS once or vice versa, separated by at least 48 hours of resting period	None	Not described	None	Published
Ko 2008a	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 10 seconds		None	Not described	None	Published
Koo 2018	A‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes with the anode over S1 of the affected hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS or sham tDCS during 10 stimulation sessions over 10 days	None	Not described	None	Published
Koo 2018	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 20 seconds		None	Not described	None	Published
Klomjai 2018	Dual tDCS	Saline‐soaked sponge‐pad electrodes with 35cm² surface and electroconductive gel	Anodal tDCS over the M1 of the affected hemisphere and cathodal tDCS over the M1 of the unaffected hemisphere	2 mA for 20 minutes	Dual tDCS once prior to base treatment and sham tDCS once prior to base treatment or vice versa, separated by at least 7 days of resting period	Dose‐matched physical therapy for 60 minutes under expert supervision, aiming at improving strength in the lower extrimity	Not described	Unclear	Published
Klomjai 2018	Sham tDCS			2 mA for 120 seconds			Not described	Unclear	Published
Lee 2014	C‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes over hand area of M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	20 minutes per day, 5 times per week for 3 weeks	Occupational therapy for 30 minutes per day, 5 times per week for 3 weeks	3 of 42 (7%); 2 medical problems; 1 refused to participate	No major adverse events	Published
	C‐tDCS		Over the contralateral supraorbital forehead	2 mA for 20 minutes	20 minutes per day, 5 times per week for 3 weeks	Virtual reality therapy for 30 minutes per day, 5 times per week for 3 weeks	3 of 42 (7%); 2 medical problems; 1 refused to participate
	Virtual reality	NA		NA	NA	Virtual reality only for 30 minutes per day, 5 times per week for 3 weeks	2 of 22 (9%); 1 refused to participate; 1 early discharge
Lindenberg 2010	Dual‐tDCS	Saline‐soaked 16.3 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 30 minutes	Base treatment + dual‐tDCS or sham tDCS at 5 consecutive sessions on 5 consecutive days	Physical and occupational therapy sessions at 5 consecutive sessions on 5 consecutive days for 60 minutes, including functional motor tasks	None	None	Published
Lindenberg 2010	Sham tDCS			1.5 mA for 30 seconds			None	None	Published
Mahmoudi 2011	A‐tDCS1	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral orbit	1 mA for 20 minutes	Each participant underwent A‐tDCS1, A‐tDCS2, C‐tDCS, dual‐tDCS and sham tDCS once with a wash‐out period of at least 96 hours	None	None	Unclear	Published
	A‐tDCS2	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	On the contralateral deltoid muscle	1 mA for 20 minutes
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes, cathode over M1 of the non‐lesioned hemisphere	Over M1 of the lesioned hemisphere	1 mA for 20 minutes
	Dual‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Manji 2018	A‐tDCS	25 cm² saline‐soaked sponge electrodes with the anode over the SMA of the lesioned hemisphere	Over the inion	1 mA for 20 minutes	Each participant underwent A‐tDCS + base treatment or sham tDCS + base treatment in a random order, each once a day for a week	Body‐weight‐supported treadmill training (BWSTT) with 20% of body weight support for 20 minutes once a day for a week	None	Unclear	Published
Manji 2018	Sham tDCS		Over the inion	1 mA for 30 seconds			None	Unclear	Published
Mazzoleni 2019	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + 20 minutes either A‐tDCS or sham tDCS 5 times a week for 6 weeks	Robotic wrist‐training with appr. 1000 repetitions per session. The robot provided assistance, if necessary	1 out of 20 (5) in the CTL group dropped out due to robot failure	None	Published
Mazzoleni 2019	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 5 seconds				None	Published
Mortensen 2016	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 20 minutes	Base treatment + 20 minutes either A‐tDCS or sham tDCS on 5 consecutive days	30 minutes of home‐based occupational therapy, aiming at activities and functional tasks	1 out of 8 (13) in the CTL group dropped out during worsening of hand function	There were 6 moderate or severe adverse events (3 in the EXP group and 3 in the CTL group, respectively)	Published
Mortensen 2016	Sham tDCS		Over the contralateral supraorbital forehead	1.5 mA for 30 seconds					Published
Nair 2011	C‐tDCS	Saline‐soaked sponge electrodes with the cathode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 30 minutes	Base‐treatment + C‐tDCS or sham tDCS for 5 consecutive daily sessions, each at the beginning of the base treatment sessions	Occupational therapy (PNF; shoulder abduction, external rotation, elbow extension, forearm pronation) for 5 consecutive daily sessions (60 minutes each)	None	None	Published
Nair 2011	Sham tDCS	Not described by the authors		For 30 minutes			None	None	Published
Nicolo 2017	C‐tDCS	35 cm² saline‐soaked sponge electrodes with the cathode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 25 minutes	Base therapy + brain stimulation 3 times per week for 3 weeks during upper extremity functional motor training sessions	30 minutes of active functional motor practice, consisting of patient‐tailored exercises	None	None	Published
	Sham (tDCS, cTBS)		Over the contralateral supraorbital forehead	1 mA for 30 seconds
	cTBS	Over non‐lesioned M1	N/A	267 bursts, each consisting of 3 pulses at 30 Hz, repeated at inter‐burst intervals of 167 ms); 2 stimulation trains of 30 seconds (separated by 15 minutes)
Park 2013	A‐tDCS	Sponge electrodes with the anode positioned over the bilateral prefrontal cortex	At the non‐dominant arm	2 mA for 30 minutes	Base‐treatment + A‐tDCS or sham tDCS for 5 days a week for approximately 18 days	Computer‐assisted cognitive rehabilitation (CACR) with the ComCog program (15 minute attention and 15 minute memory training)	Unclear	None	Published
Park 2013	Sham tDCS		At the non‐dominant arm	2 mA for 30 seconds			Unclear	None	Published
Park 2015	A‐tDCS	Anode over Cz area of the left parietal lobe [sic]	Over the contralateral supraorbital forehead	2 mA for 15 minutes	Physiotherapy + either A‐tDCS or sham tDCS for 3 days a week during 4 weeks	Task related training for weight support ability improvement and stepping strategy Quote: "(1) lifting and maintaining the lower extremity; (2) lifting the heels; (3) lifting the lower extremity over the footstool followed by lowering; (4) lifting the lower extremity and lowering in onto a footstool; (5) walking back and forth over a 3‐m distance to a chair; and (6) going back and forth at a constant pace over 10‐m distance. The tasks were conducted one‐on‐one with a physical therapist."	None	None	Published
	Sham tDCS	Anode over Cz area of the left parietal lobe [sic]	Over the contralateral supraorbital forehead	Not described
	PT	N/A
Picelli 2015	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS with either cathodal transcutaneous spinal direct current stimulation (tsDCS) or with sham tsDCS	Robot‐assisted gait training on a G‐EO for 20 minutes, 5 times per week for 2 weeks	None	None	Published
Picelli 2015	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 2 minutes	Base treatment + sham tDCS and cathodal tsDCS		None	None	Published
Qu 2009	C‐tDCS	Saline‐soaked 18 cm² sponge electrodes over primary sensorimotor cortex of the lesioned hemisphere	Unclear	0.5 mA for 20 minutes, once a day for 5 consecutive days for 4 weeks		NA	None	None	Published
Qu 2009	PT	NA		Physical therapy according to the Bobath, Brunnstrom and Rood approaches for 40 minutes twice a day for 5 consecutive days for 4 weeks		NA	None	None	Published
Qu 2017	C‐tDCS	Not described	Not described	1.0 mA cathodal tDCS for 2 weeks, once a day, once for 20 minutes, 5 days a week	Not described	Not described	Not described	Unclear	Published
	C‐tDCS	Not described	Not described	2.0 mA cathodal DCS for two weeks, once a day, once for 20 minutes, 5 days a week
	Sham tDCS	Not described	Not described	Sham tDCS for 2 weeks, once a day, once for 20 minutes, 5 days a week
Rabadi 2017	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over PMC of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 30 minutes	Base therapy + C‐tDCS or sham tDCS 30 minutes a day on 5 consecutive days for 2 weeks	4 hours of standard occupational and physical therapy	There were no drop‐outs during intervention phase. Until 3 months follow‐up 3 dropouts (38) occured in the EXP group and 1 (13) in the CTL group. Reasons were not stated by the authors.	None	Published
Rabadi 2017	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds		4 hours of standard occupational and physical therapy		None	Published
Rocha 2016	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 13 minutes	A‐tDCS, C‐tDCS or sham tDCS 3 times a week for 4 consecutive weeks prior to base therapy	mCIMT (total immobilisation of the non‐paretic upper limb and intensive training of the paretic upper limb) for 6 continuous hours each day over 4 weeks plus 1 hour gross and fine motor activities training per day	There were 2 drop‐outs in each group (28%) due to unknown reasons	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the non‐lesioned hemisphere		1 mA for 9 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere		1 mA for 30 seconds
Rossi 2013	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Once a day for 5 consecutive days	Not described by the authors	None	None	Published
Rossi 2013	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds	Once a day for 5 consecutive days	Not described by the authors	None	None	Published
Saeys 2015	A‐tDCS	Over the motor cortex (on C4 or C3 of the 10–20 EEG system)	over the intact hemisphere	1.5 mA for 20 minutes	16 x 20‐minute sessions (4 times a week for 4 weeks)	Both groups received multidisciplinary regular physical and occupational therapy mainly focused on the neurodevelopmental treatment concept (1 hour daily)	None	None	Published
Saeys 2015	Sham tDCS	Over the motor cortex (on C4 or C3 of the 10–20 EEG system)	over the intact hemisphere	Stimulation turned off after 30 seconds	16 x 20‐minute sessions (4 times a week for 4 weeks)		None	None	Published
Salazar 2019	Dual‐tDCS	Over the the M1 area (C3 and C4 of the EEG system) Anode electrodes were positioned over the ipsilesional M1 and cathodes over the contralesional M1		Both groups received 10 sessions of concurrent tDCS and FES or sham tDCS and FES during 30 minutes, 5 times a week for 2 weeks		Before each stimulation session, participants had scapular, shoulder, elbow, wrist and finger passive mobilization for approximately 10 min	None	None	Published
Salazar 2019	Dual sham tDCS						None	None	Published
Sattler 2015	A‐tDCS	Over the M1 area (at the hotspot of the extensor carpi radialis muscle	Cathode over the contralesional supraorbital region	1.2 mA anodal tDCS	5 consecutive daily sessions for 13 minutes each	rPNS (5 Hz) was delivered to the radial nerve through bipolar round brass electrodes placed in the spiral grove of the paretic side and was applied at the same time as the real or sham tDCS stimulation. It was applied similarly in both active and sham conditions for 13 minutes. The intensity of was adjusted to be below the threshold for direct M response (0.7 x MT).	None	None	Published
Sattler 2015	Sham tDCS		Cathode over the contralesional supraorbital region	Stimulation (same site and same parameters) was turned off after 60 seconds of stimulation	5 consecutive daily sessions for 13 minutes each		None	None	Published
Seo 2017	A‐tDCS	Over the presumed leg area of the lesioned hemisphere, just lateral to the Cz position according to the 10–20 system	Cathode on the forehead above the contralateral orbit	2 mA for 20 minutes	20 minutes of tDCS for every weekday during 2 weeks (total 10 sessions)	RAGT for 45 minutes after tDCS	None at first follow‐up	None	Published
Seo 2017	Sham tDCS		Cathode on the forehead above the contralateral orbit	Stimulation intensity was slowly tapered down from 2 to 0 mA over several seconds after initial minute		RAGT for 45 minutes after tDCS	None at first follow‐up	None	Published
Shaheiwola 2018	A‐tDCS	Primary motor cortex using (abductor pollicis brevis) hot spot)	Cathode on the contralateral symmetrical area of non‐lesioned hemisphere	2.0 mA, time of ramp‐up: 10 seconds, time of ramp‐down: 10 seconds, 20 minutes	5 sessions per week on workdays and a total of 20 sessions during the 4 weeks	60 minutes FES each day	None	None	Published
Shaheiwola 2018	Sham tDCS					60 minutes FES each day	None	None	Published
Sik 2015	A‐tDCS g	Anodal tDCS over C3‐C4 area of the affected hemisphere	Opposite supraorbital region	2 mA, 20 mintes in patients with anodal stimulation ‐‐‐ 2 mA, 40 minutes in the bihemispheric‐treated patients (20 minutes anodal tDCS to the lesional hemisphere/20 minutes cathodal tDCS to the non‐lesional hemisphere)		tDCS application was started simultaneously with occupational therapy (15 sessions for 3 weeks) Physiotherapy and occupational therapy, (2 hours, including range of motion exercises, strengthening exercises, outreach activities)	5 (2 in A‐tDCS group and 2 in bihemispheric group and 1 in sham group)	None	Published
	Dual‐tDCS	Dual‐TDCs active electrode to the C3‐C4 area of the unaffected hemisphere in addition to its anodal application
	Sham tDCS	Sham: electrodes were placed as in the anodal group
Sohn 2013	A‐tDCS	25 cm² sponge electrodes over M1 of the affected hemisphere	Not described	2 mA for 10 minutes	A‐tDCS or sham tDCS once	None	Unclear	Unclear	Published
Sohn 2013	Sham tDCS	25 cm² sponge electrodes over M1 of the affected hemisphere	Not described	2 mA for 20 seconds	A‐tDCS or sham tDCS once	None	Unclear	Unclear	Published
Straudi 2016	Dual‐tDCS	Anode was placed on the M1 of the affected hemisphere. Electrodes were located at C3 and C4 according to the 10/20 international EEG system	Cathode on the contralateral M1 area	1 mA for 30 minutes, during RAT		Upper Extremity Robot‐Assisted Training	None	No severe adverse events (10 out of 23 reported mild adverse events)	Published
Straudi 2016	Sham tDCS		Cathode on the contralateral M1 area	Current was delivered for only 30 seconds and then the current was discontinued, but the tDCS apparatus was left in place for the same time as active tDCS (30 minutes)		Upper Extremity Robot‐Assisted Training	None		Published
Sunwoo 2013a	Dual‐tDCS	Saline‐soaked 25 cm² sponge electrodes over the right posterior parietal cortex (PPC) plus cathodal tDCS over the left PPC	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent dual‐tDCS, A‐tDCS and sham tDCS once with a wash‐out period of at least 24 hours	None	None	3 (30%) suffered from mild headache after dual‐tDCS, which disappeared spontaneously	Published
	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over the right PPC plus sham tDCS over the left PPC		1 mA for 20 minutes
	Sham tDCS	Saline‐soaked 25 cm² sponge electrodes over the right PPC plus sham tDCS over the left PPC		1 mA for 10 seconds
Tahtis 2012	Dual‐tDCS	Saline‐soaked 25 cm² electrodes with the anode placed over the leg area of the lesioned hemisphere and the cathode placed over leg area of the non‐lesioned hemisphere	Not described	2 mA for 15 minutes	Dual‐tDCS or sham tDCS once	None	Unclear	None	Published
Tahtis 2012	Sham tDCS		Not described	2 mA for < 30 seconds	Dual‐tDCS or sham tDCS once	None	Unclear	None	Published
Tedesco Triccas 2015b	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base therapy plus tDCS or sham tDCS for 18 sessions during 8 weeks (approximately 2 to 3 sessions per week)	Robotic arm training with the ArmeoSpring device (60 minutes per session) for 18 sessions during 8 weeks (approximately 2 to 3 sessions per week)	1 out of 12 (8%) in the A‐tDCS group due to a skin reaction after receiving four sessions of A‐tDCS	6 out of 12 (50%) in the A‐tDCS group reported adverse events such as pain, burning or headache after receiving A‐tDCS	Published/unpublished
Tedesco Triccas 2015b	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 20 seconds					Published/unpublished
Utarapichat 2018	A‐tDCS	Saline‐soaked 10 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	2 mA for 10 minutes	Not described	Not described	None	Unclear	Published
Utarapichat 2018	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds	Not described	Not described	None	Unclear	Published
Viana 2014	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	2 mA for 13 minutes	Base therapy + A‐tDCS or sham tDCS 3 times a week for 5 weeks	Virtual reality training using Nintendo Wii (Games used: Wii Sports resort, Wii Play Motion, Let's Tap) aiming at movements of shoulder, elbow, wrist, hand and fingers; each game was played for 15 minutes (total time per training session: 60 minutes); passive stretching exercises were performed before and after each training session	None	None	Published
Viana 2014	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds			None	None	Published
Wang 2014	Dual‐tDCS	35 cm² electrodes with the anode placed over M1 of the affected hemisphere	Over contralateral M1	1 mA for 20 minutes	Dual‐tDCS or sham‐tDCS once	Placebo methylphenidate 1 hour prior to stimulation	Unclear	No major adverse events; 3 participants (50%) from the dual‐tDCS group reported mild tingling sensation with tDCS stimulation	Published
	Dual‐tDCS			1 mA for 20 minutes		20 mg MP 1 hour prior to stimulation
	Sham‐tDCS			1 mA for 10 seconds		20 mg MP 1 hour prior to stimulation
Wong 2015	A‐tDCS	Over the hand area of primary motor cortex of the affected hemisphere	Cathodal electrode was placed over the contralateral supraorbital area	1 mA tDCS for 20 minutes	Not described	Not described	Not described	Unclear	Published
Wong 2015	5 consecutive sessions of intensive physiotherapy upper limb training			1 mA tDCS for 20 minutes	Not described	Not described	Not described	Unclear	Published
Wu 2013a	C‐tDCS	Saline‐soaked 24.75 cm² sponge electrodes over primary sensorimotor cortex of the lesioned hemisphere	Over the shoulder on the unaffected side	1.2 mA for 20 minutes	Once daily 5 days a week for 4 weeks	Quote: "Both groups received a conventional physical therapy program for 30 minutes twice daily, including maintaining good limb position, chronic stretching via casting or splinting, physical modalities and techniques, and movement training"	None	None	Published
Wu 2013a	Sham tDCS		Over the shoulder on the unaffected side	1.2 mA for 30 seconds	Once daily 5 days a week for 4 weeks		None	None	Published
Yi 2016	A‐tDCS	Over the right PPC (5 cm x 5 cm)	Over Cz	2 mA for 30 minutes	5 sessions per week for 3 weeks	Conventional physical therapy throughout the duration of the 3 weeks	2 out 32 (6%)	None	Published
	C‐tDCS	Over the left PPC		2 mA for 30 minutes
	Sham tDCS	Sham tDCS was performed in the same way as for anodal group		2 mA for 30 minutes
	Sham tDCS	Sham tDCS was performed in the same way as for anodal group		Stimulator was turned off after 30 seconds
Yun 2015	A‐tDCS left	At T3 for the left‐group and	Unclear	2 mA for 30 minutes	5 times a week for 3 weeks	Not described	None	None	Published
	A‐tDCS right	At T3 for the left‐group and	Unclear
	Sham tDCS	At T4 for the right‐group	Unclear
	Sham tDCS	Using the same method as for the left‐ group,	Unclear

A‐tDCS: anodal direct current stimulation
C‐tDCS: cathodal direct current stimulation
CIMT: constraint‐induced movement therapy
cTBS: Continuous Theta Burst Stimulation
Dual‐tDCS: A‐tDCS and C‐tDCS simultaneously
EEG: electroencephalography
FES: Functional electrical stimulation
M1: primary Motor Cortex
MEP: motor‐evoked potentials
MI‐BCI: motor imagery brain‐computer interface
MP: methylphenidate
NA: not applicable
PNF: proprioceptive neuromuscular facilitation
PPC: posterior parietal cortex
PT: physical therapy
RAGT: robotic‐assisted gait training
rPNS: Repetitive electrical stimulation
SD: standard deviation
tDCS: transcranial direct current stimulation
tsDCS: transcutaneous spinal direct current stimulation

Outcomes

Widely used outcomes for activities were the Barthel Index (BI, 13 of 67 studies, 20%) and the Motor Activity Log (MAL, seven of 67 studies, 11%). Widely used outcomes for upper extremity function were the Upper Extremity Fugl‐Meyer Score (UE‐FM, 30 of 67 studies, 45%), the Jebsen‐Taylor Test (JTT, nine of 67 studies, 13 %) and the Action Research Arm Test (ARAT, eight of 67 studies, 12%). Fifty‐six studies (84%) reported data on adverse events or drop‐outs.

We excluded 10 of the included trials from quantitative syntheses (meta‐analyses) because of missing information regarding the first intervention period of the cross‐over trial (Au‐Yeung 2014; Fregni 2005a; Jo 2008a; Kang 2008b; Kim 2009; Klomjai 2018; Ko 2008a; Mahmoudi 2011; Sohn 2013; Sunwoo 2013a).

Excluded studies

We excluded 49 trials from qualitative assessment, mainly because they were not RCTs, or because their outcomes did not measure function, ADL or cognition (see Characteristics of excluded studies).

Risk of bias in included studies

We provided information about the risk of bias in Characteristics of included studies. To complete the rating of methodological quality, we contacted all principal investigators of the included trials and of trials awaiting classification to request further information about methodological issues, if necessary. We made contact via letter and email, including email reminders once a month if we received no response. Some trialists provided all requested information, and some did not answer our requests. We used the 'Risk of bias' tool, as implemented in Review Manager 5.3, to assess risk of bias according to the aspects listed under Methods. A detailed description of risk of bias can be found in Characteristics of included studies. Information on risk of bias on study level and outcome level is provided in Figure 2 and in Figure 3.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Thirty‐two of the 67 included studies (48%) were at low risk of bias for sequence generation, whereas sixteen studies (24%) were at low risk of bias for allocation concealment.

Blinding

We deemed 34 of the 67 included studies (50%) to be at low risk of bias for blinding of participants and personnel for subjective outcomes and 52 studies (76%) for objective outcomes, respectively; three studies were at high risk of bias in this domain (Bang 2015; Cho 2017; Hathaiareerug 2019). Forty‐five studies (66%) were at low risk of bias for blinding of outcome assessment for subjective and objective outcomes, whereas we determined 10 studies to have high risk of bias in this domain (Chang 2015; Cho 2017; Fusco 2013a; Hamoudi 2018; Kim 2009; Kim 2016; Mazzoleni 2019; Rabadi 2017; Utarapichat 2018; Yi 2016).

Incomplete outcome data

Thirty‐eight of the 67 included studies (56%) were at low risk of bias for incomplete outcome data for objective and subjective outcomes, and three studies were at high risk of bias (Fusco 2014; Lee 2014; Yun 2015).

Selective reporting

Thirteen of the 67 included studies (38%) were at low risk of bias for selective outcome reporting, and three studies (5%) were at high risk of bias (Di Lazzaro 2014a; Di Lazzaro 2014b; Nair 2011).

Other potential sources of bias

We are not aware of other potential sources of bias.

Effects of interventions

Fifty‐seven of the 67 included studies (85%) were included in the meta‐analysis (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; Kim 2016; Koo 2018; Lee 2014; Lindenberg 2010; Manji 2018; Mazzoleni 2019; Mortensen 2016; Nair 2011; Nicolo 2017; Park 2013; Park 2015; Picelli 2015; Qu 2009; Qu 2017; 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; Yi 2016; Yun 2015).

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

We found 19 studies with 686 participants examining the effects of tDCS on ADL (Bolognini 2011; Chelette 2014; Cunningham 2015; Di Lazzaro 2014a; Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Kim 2010; Kim 2016; Koo 2018; Lee 2014; Nicolo 2017; Qu 2017; Rocha 2016; Straudi 2016; Tedesco Triccas 2015b; Wu 2013a; Yi 2016; Yun 2015). We found evidence of effect regarding ADL performance when we analysed the data with combined intervention groups, as stated in Methods (i.e. A‐tDCS and/or C‐tDCS versus sham tDCS; SMD 0.28, 95% CI 0.13 to 0.44; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.1; summary of findings Table 1).

1.1.2 Studies presenting change scores

Four studies with 95 participants reported the effects of tDCS on ADL as change values relative to baseline (Andrade 2017; Danzl 2012; Fusco 2014; Rabadi 2017). Moderate‐quality evidence suggests that there is evidence of an effect (SMD 0.48, 95% CI 0.02 to 0.95; inverse variance method with random‐effects model; Analysis 1.1; summary of findings Table 1).

The funnel plot of Analysis 1.1 can be found in Figure 4. By visual inspection, we concluded that there were no indications of substantial funnel plot asymmetry that would suggest the presence of publication bias.

Figure 4

Funnel plot of comparison: 1 Primary outcome measure: tDCS for improvement of ADL versus any type of placebo or control intervention, outcome: 1.1 ADL at the end of the intervention period, absolute values (BI points).

Comparison 1.2 Primary outcome measure: ADL until the end of follow‐up, absolute values (at least three months after the end of the intervention period)

1.2.1 Studies presenting absolute values

We included six studies with 269 participants (Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Kim 2010; Rossi 2013; Tedesco Triccas 2015b); investigators measured the effects of tDCS on ADL at the end of follow‐up. We found evidence of effect regarding ADL performance when we analysed the data with combined intervention groups (SMD 0.31, 95% CI 0.01 to 0.62; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.2; summary of findings Table 3).

1.2.2 Studies presenting change scores

One study with 16 participants reported the effects of tDCS on ADL as change values relative to baseline (Rabadi 2017). There is low‐quality evidence that there is no evidence of an effect (SMD ‐0.64, 95% CI ‐1.66 to 0.37; inverse variance method with random‐effects model; Analysis 1.2; summary of findings Table 3).

By visual inspection of the funnel plot of Analysis 1.2, we concluded that there were no indications of substantial asymmetry that would suggest the presence of publication bias (Figure 5).

Figure 5

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.2 Primary outcome measure: ADL until the end of follow‐up.

Comparison 1.3 Secondary outcome measure: upper extremity function at the end of the intervention period

1.3.1 Studies presenting absolute values

Twenty‐four trials with a total of 792 participants examined upper limb function at the end of the intervention period and provided absolute values for the outcome (Allman 2016; Andrade 2017; Bolognini 2011; Chelette 2014; Cunningham 2015; Di Lazzaro 2014a; Di Lazzaro 2014b; Fusco 2013a; Hesse 2011; Ilić 2016; Kim 2010; Koo 2018; Lee 2014; Lindenberg 2010; Nicolo 2017; Qu 2017; Rocha 2016; Rossi 2013; Salazar 2019; Shaheiwola 2018; Straudi 2016; Tedesco Triccas 2015b; Viana 2014; Wu 2013a). There was no evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 0.17, 95% CI ‐0.05 to 0.38; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.3; summary of findings Table 1).

1.3.2 Studies presenting change scores

We included 10 studies with 193 participants (Ang 2012; D'Agata 2016; Fusco 2014; Hamoudi 2018; Mazzoleni 2019; Mortensen 2016; Nair 2011; Rabadi 2017; Sattler 2015; Wang 2014). Investigators measured the effects of tDCS on upper limb function at the end of the intervention period and provided absolute values for the outcome. There was no evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 0.33, 95% CI ‐0.12 to 0.79; inverse variance method with random‐effects model; low‐quality evidence; Analysis 1.3; summary of findings Table 1).

By visual inspection of the funnel plot of Analysis 1.3, we concluded that there were some indications of asymmetry in the studies presenting change scores, suggesting that publication bias may be present (Figure 6).

Figure 6

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.3 Secondary outcome measure: upper extremity function at the end of the intervention period.

Comparison 1.4 Secondary outcome measure: upper extremity function to the end of follow‐up (at least three months after the end of the intervention period)

1.4.1 Studies presenting absolute values

Five studies with a total of 211 participants examined upper extremity function at the end of follow‐up and reported absolute values for this outcome (Allman 2016; Di Lazzaro 2014b; Hesse 2011; Rossi 2013; Tedesco Triccas 2015b). We found no evidence of effect regarding upper extremity function when we analysed the data with combined intervention groups (i.e. A‐tDCS and/or C‐tDCS versus sham tDCS; SMD ‐0.00, 95% CI ‐0.39 to 0.39; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.4; summary of findings Table 3).

1.4.2 Studies presenting change scores

We included three studies with 72 participants (D'Agata 2016; Hamoudi 2018; Kim 2010); the investigators measured the effects of tDCS on upper limb function at the end of follow‐up and provided change values for the outcome. There was evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 1.07, 95% CI 0.04 to 2.11; inverse variance method with random‐effects model; low‐quality evidence; Analysis 1.4; summary of findings Table 3).

By visual inspection of the funnel plot of Analysis 1.4, we concluded that there were some indications of asymmetry in the studies presenting change scores, suggesting that publication bias may be present (Figure 7).

Figure 7

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.4 Secondary outcome measure: upper extremity function to the end of follow‐up.

Comparison 1.5 Secondary outcome measure: lower extremity function at the end of the intervention period

1.5.1 Studies presenting absolute values

Eight studies with a total of 204 participants examined lower extremity function at the end of the intervention period and reported absolute values for this outcome (Cha 2014; Chang 2015; Geroin 2011; Koo 2018; Manji 2018; Park 2015; Picelli 2015; Yi 2016). We found no evidence of effect regarding lower extremity function when we analysed the data with combined intervention groups (i.e. A‐tDCS and/or C‐tDCS versus sham tDCS; SMD 0.28, 95% CI ‐0.12 to 0.69; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.5; summary of findings Table 1).

1.5.2 Studies presenting change scores

Four studies with a total of 54 participants examined lower extremity function at the end of the intervention period and reported change values for this outcome (Danzl 2012; Fusco 2014; Seo 2017; Tahtis 2012). We found no evidence of effect regarding lower extremity function when we analysed the data with combined intervention groups (i.e. A‐tDCS and/or C‐tDCS versus sham tDCS; SMD 0.46, 95% CI ‐0.09 to 1.01; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.5; summary of findings Table 1).

By visual inspection of the funnel plot of Analysis 1.5, we concluded that there were no indications for publication bias (Figure 8).

Figure 8

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.5 Secondary outcome measure: lower extremity function at the end of the intervention period.

There were no studies which examined the effects of tDCS on lower extremity function at follow‐up (i.e. after at least three months).

Comparison 1.6 Secondary outcome measure: muscle strength at the end of the intervention period

1.6.1 Studies presenting absolute values

We included 13 studies with 437 participants (Andrade 2017; Bolognini 2011; Di Lazzaro 2014a; Di Lazzaro 2014b; Fusco 2013a; Hesse 2011; Khedr 2013; Koo 2018; Lee 2014; Picelli 2015; Rocha 2016; Salazar 2019; Viana 2014); investigators measured the effects of tDCS on muscle strength at the end of the intervention period and provided absolute values for the outcome. There was no evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 0.19, 95% CI ‐0.01 to 0.38; inverse variance method with random‐effects model; high‐quality evidence; Analysis 1.6; summary of findings Table 1).

1.6.2 Studies presenting change scores

Five studies with a total of 116 participants examined muscle strength at the end of the intervention period and reported change values for this outcome (Fusco 2014; Geroin 2011; Mazzoleni 2019; Mortensen 2016; Seo 2017). We found no evidence of effect regarding muscle strength when we analysed the data with combined intervention groups (i.e. A‐tDCS and/or C‐tDCS versus sham tDCS; SMD 0.07, 95% CI ‐0.66 to 0.80; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.6; summary of findings Table 1).

By visual inspection, the authors concluded that there were no indications of funnel plot asymmetry that would suggest the presence of publication bias in Analysis 1.6 (Figure 9).

Figure 9

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.6 Secondary outcome measure: muscle strength at the end of the intervention period.

Comparison 1.7 Secondary outcome measure: muscle strength at the end of follow‐up (at least three months after the end of the intervention period), absolute values

We included three studies with 156 participants (Di Lazzaro 2014b; Hesse 2011; Khedr 2013). Investigators measured the effects of tDCS on muscle strength at the end of follow‐up and provided absolute values for the outcome. There was no evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 0.07, 95% CI ‐0.26 to 0.41; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 1.7; summary of findings Table 3).

By visual inspection, the authors concluded that there were no indications of funnel plot asymmetry that would suggest the presence of publication bias in Analysis 1.7 (Figure 10).

Figure 10

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.7 Secondary outcome measure: muscle strength at the end of follow‐up.

Comparison 1.8 Secondary outcome measure: cognitive abilities at the end of the intervention period

There were two studies with 56 participants that examined the effects of tDCS on cognitive abilities (Park 2013; Yun 2015); investigators measured the effects of tDCS on cognitive impairment at the end of intervention and provided absolute values for the outcome. There was no evidence of effect of tDCS when we analysed the data with combined intervention groups (SMD 0.46, 95% CI ‐0.10 to 1.02; inverse variance method with random‐effects model; low‐quality evidence; Analysis 1.8; summary of findings Table 1). We furthermore identified three randomised cross‐over trials that examined the effects of tDCS on cognitive abilities, but data extraction was not possible due to missing information regarding the first intervention period (Au‐Yeung 2014; Jo 2008a; Kang 2008b). However, each of the studies reported evidence of an effect in favour of tDCS regarding measures of attention. We did not identify any studies examining the effects of tDCS on cognitive abilities at follow‐up.

Comparison 1.9: Secondary outcome measure: spatial neglect

We identified one trial with 15 participants that examined the effects of tDCS on neglect, but data extraction was not possible due to missing information regarding the first intervention period (Ko 2008a). We included one study with 30 participants examining the effects of tDCS on spatial neglect (Yi 2016). This study reported improvement in neglect tests (MD 4.80, 95% CI 0.13 to 9.47; inverse variance method with random‐effects model; very low‐quality evidence). We did not identify any randomised studies examining the effects of tDCS on spatial neglect at follow‐up (Analysis 1.9).

Comparison 1.10 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period

Forty‐eight out of 67 studies (74%) reported data on dropouts, and 36 out of 67 studies (55%) reported data on adverse events. In 27 of 67 studies (40%), dropouts, adverse events or deaths that occurred during the intervention period were reported (Andrade 2017; Cho 2017; Danzl 2012; Fusco 2013a; Hamoudi 2018; Jo 2008a; Lee 2014; Mazzoleni 2019; Mortensen 2016; Nair 2011; Nicolo 2017; Park 2015; Picelli 2015; Rabadi 2017; Rocha 2016; Saeys 2015; Salazar 2019; Sattler 2015; Seo 2017; Shaheiwola 2018; Sik 2015; Straudi 2016; Tedesco Triccas 2015b; Utarapichat 2018; Viana 2014; Yi 2016; Yun 2015), whereas the remaining studies reported no dropouts, adverse events or deaths. When analysing 47 studies with 1330 participants, we found no evidence of effect regarding differences in dropouts, adverse effects and deaths between intervention and control groups (RR 1.25, 95% CI 0.74 to 2.13; Mantel‐Haenszel method with random‐effects model; analysis based only on studies that reported either on dropouts or on adverse events or on both; moderate‐quality evidence; Analysis 1.10; summary of findings Table 1). A detailed description of dropouts, adverse events and deaths during the intervention period can be found in Table 3.

By visual inspection, the authors concluded that there were no indications of funnel plot asymmetry that would suggest the presence of publication bias in Analysis 1.10 (Figure 11).

Figure 11

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.10 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period.

Comparison 2. tDCS versus any type of active control intervention

Comparison 2.1 Primary outcome measure: ADL at the end of the intervention period, absolute values

There were three studies with 121 participants that examined the effects of tDCS on ADL at the end of the intervention period and provided absolute values on this outcome (Bang 2015; Lee 2014; Qu 2009). There was evidence of effect of tDCS on ADL at the end of the intervention period (MD 6.59 BI points, 95% CI 1.26 to 11.91; inverse variance method with random‐effects model; low‐quality evidence; Analysis 2.1; summary of findings Table 2). We did not identify any study examining the effects of tDCS versus any type of active control intervention on ADL at follow‐up.

Comparison 2.2 Secondary outcome measure: upper extremity function at the end of the intervention period

2.2.1 Studies presenting absolute values

Five studies with a total of 124 participants which examined upper extremity function at the end of the intervention period and reported absolute values for this outcome (Cha 2014; Cho 2017; Hathaiareerug 2019; Lee 2014; Wong 2015). We found evidence of an effect regarding upper extremity function at the end of the intervention period (SMD 0.84, 95% CI 0.20 to 1.48; inverse variance method with random‐effects model; low‐quality evidence; Analysis 2.2; summary of findings Table 2). We did not identify any study examining the effects of tDCS versus any type of active control intervention on upper extremity function at follow‐up.

2.2.2 Studies presenting change scores

There was one study with 32 participants that examined the effects of tDCS on upper extremity function at the end of the intervention period and reported change values for this outcome (Hamoudi 2018). This study reported no evidence of effect of tDCS on upper extremity function at the end of the intervention period (SMD 0.51, 95% CI ‐0.20 to 1.22; inverse variance method with random‐effects model; low‐quality evidence; Analysis 2.2; summary of findings Table 2). We could not identify any study examining the effects of tDCS versus any type of active control intervention on upper extremity function at follow‐up.

Comparison 2.3 Secondary outcome measure: upper extremity function at the end of follow up

One study with 32 participants examined the effects of tDCS on upper extremity function at the end of the follow‐up (Hamoudi 2018). This study reported no evidence of effect of tDCS on upper extremity function (MD 10.00% in change of the time to complete the JTT, 95% CI ‐0.07 to 20.07; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 2.3; summary of findings Table 4).

Comparison 2.4 Secondary outcome measure: lower extremity function at the end of the intervention period

Three studies with a total of 66 participants which examined lower extremity function at the end of the intervention period (Cha 2014; Cho 2017; Park 2015). We found no evidence of an effect regarding lower extremity function at the end of the intervention period (SMD 0.23, 95% CI ‐0.66 to 1.13; inverse variance method with random‐effects model; moderate‐quality evidence; Analysis 2.4). We did not identify any study examining the effects of tDCS versus any type of active control intervention on lower extremity function at follow‐up.

Comparison 2.5 Secondary outcome measure: muscle strength at the end of the intervention period

There were two studies with 57 participants that examined the effects of tDCS on muscle strength at the end of the intervention period (Hathaiareerug 2019; Lee 2014). These studies reported no evidence of effect of tDCS on muscle strength at the end of the intervention period (SMD 0.08, 95% CI ‐0.44 to 0.60; inverse variance method with random‐effects model; low‐quality evidence; Analysis 2.5). We could not identify any study examining the effects of tDCS versus any type of active control intervention on muscle strength at follow‐up.

Comparison 2.6 Secondary outcome measure: spatial neglect at the end of the intervention period

There was one study with 12 participants that examined the effects of tDCS on upper extremity function at the end of the intervention period and reported change values for this outcome (Bang 2015). This study reported no evidence of effect of tDCS on lower extremity function at the end of the intervention period (MD ‐0.53 points in the line bisection test, 95% CI ‐0.93 to ‐0.13; moderate‐quality evidence; Analysis 2.6). We could not identify any study examining the effects of tDCS versus any type of active control intervention spatial neglect at follow‐up.

Comparison 2.7 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period

Seven studies with 209 participants reported dropouts, adverse events, or deaths that occurred during the intervention period (Hamoudi 2018; Hathaiareerug 2019; Lee 2014). We found no evidence of effect regarding differences in dropouts, adverse effects and deaths between intervention and control groups (RR 1.76, 95% CI 0.43 to 7.17; Mantel‐Haenszel method with random‐effects model; analysis based only on studies which reported either on dropouts or on adverse events or on both; moderate‐quality evidence; Analysis 2.7; summary of findings Table 2).

Comparison 3. Subgroup analyses

Outcome 3.1. Planned analysis: duration of illness ‐ acute/subacute versus postacute versus chronic phase for ADL at the end of the intervention period

In a planned subgroup analysis, we analysed the effects of tDCS on the primary outcome of ADL in the acute/subacute and postacute phases (Analysis 3.1). We found no evidence for different effects of tDCS between subgroups (P = 0.58).

Outcome 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

We performed a planned subgroup analysis regarding the location of electrode positioning and hence of stimulation (Analysis 3.2). No studies investigated the effects of A‐tDCS over the non‐lesioned hemisphere. We found no evidence of differences in effects of location and type of stimulation regarding ADL performance between subgroups (P = 0.34).

Outcome 3.3. Planned sensitivity analysis regarding types of control interventions (sham tDCS/conventional therapy/no intervention)

We performed a planned subgroup analysis regarding the type of control interventions (Analysis 3.3). We found no evidence of differences in effects of location and type of stimulation regarding ADL performance between subgroups (P = 0.53).

Sensitivity analyses

We conducted a sensitivity analysis of methodological quality to test the robustness of our results. We repeated the analysis of our primary outcome, ADL performance at the end of the intervention period and at the end of follow‐up, and considered only studies with correctly concealed allocation, blinding of assessors and ITT. The evidence of an effect of tDCS disappeared when we analysed only those studies with correct allocation concealment. See Table 1 and Table 4.

Table 4. Sensitivity analyses for comparison 1.2: primary outcome of ADL performance at the end of follow‐up at least 3 months after the end of the intervention period

Sensitivity analysis	Studies included in analysis	Effect estimate
All studies with proper allocation concealment for primary outcome absolute values	Hesse 2011; Khedr 2013; Kim 2010; Tedesco Triccas 2015b	(SMD 0.30, 95% CI ‐0.15 to 0.75; participants = 199; studies = 4; I² = 51%; inverse variance method with random‐effects model)
All studies with proper allocation concealment for primary outcome change scores	Rabadi 2017	(SMD 0.19, 95% CI ‐0.27 to 0.64; participants = 16; studies = 1; I² = 0%; inverse variance method with random‐effects model)
All studies with proper blinding of outcome assessor for primary outcome	Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Kim 2010; Rossi 2013; Tedesco Triccas 2015b	(SMD 0.31, 95% CI 0.01 to 0.62; participants = 269; studies = 6; I² = 27%; inverse variance method with random‐effects model)
All studies with intention‐to‐treat analysis	Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Rossi 2013	(SMD 0.38, 95% CI 0.05 to 0.70; participants = 205; studies = 4; I² = 16%; inverse variance method with random‐effects model)

CI: confidence interval
SMD: standardised mean difference

Discussion

Summary of main results

This review focused on evaluating the effectiveness of transcranial direct current stimulation (tDCS) (anodal stimulation (A‐tDCS)/cathodal stimulation (C‐tDCS)/(anodal plus cathodal stimulation simultaneously (dual‐tDCS)) versus any passive control intervention (sham tDCS or no intervention) and tDCS versus any active control intervention (any other approach) for improving ADL, arm and leg function, muscle strength and cognitive abilities (including spatial neglect), dropouts and adverse events in people after stroke. We included 67 studies involving a total of 1729 participants.

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

We found 19 studies with 686 participants examining the effects of tDCS on our primary outcome measure, ADL, after stroke. In addition to these studies presenting absolute values of the outcome, we found four studies with 95 participants, presenting change values for the outcome. We found moderate‐quality evidence of effect regarding ADL performance at the end of the intervention period for the studies presenting absolute values (SMD 0.28, 95% CI 0.13 to 0.44) and also moderate‐quality evidence for the studies presenting change scores (SMD 0.48, 95% CI 0.02 to 0.95). The funnel plot shows no evidence of a small‐study effect. Six studies with 269 participants reporting absolute values assessed the effects of tDCS on ADL at the end of follow‐up and one study with 16 participants reported change scores. Moderate‐quality evidence and low‐quality evidence suggested an effect regarding ADL performance (SMD 0.31, 95% CI 0.01 to 0.62 and SMD ‐0.64, 95% CI ‐1.66 to 0.37, respectively). However, this effect was not sustained when including only studies with adequate allocation concealment (Table 1; Table 4). Also, one could argue that the effect is not clinically important when using SMD 0.5 as a surrogate threshold for clinical relevance, as suggested by the GRADE working group (Schünemann 2013).

One of our secondary outcome measures was upper extremity function. Thirty‐four trials with a total of 985 participants measured upper extremity function at the end of the intervention period, revealing no evidence of an effect in favour of tDCS (SMD 0.17, 95% CI ‐0.05 to 0.38 for studies presenting absolute values; moderate‐quality evidence, and SMD 0.33, 95% CI ‐0.12 to 0.79 for studies presenting change values; 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) and three studies with 72 participants (change scores) that showed no evidence of an effect (SMD ‐0.00, 95% CI ‐0.39 to 0.39; moderate‐quality evidence and SMD 1.07, 95% CI 0.04 to 2.11; low‐quality evidence, respectively). Twelve studies with 258 participants examined the effect of tDCS on lower extremity function, but did not show evidence of an effect (moderate‐quality evidence). Eighteen studies with 551 participants reported outcome data for muscle strength at the end of the intervention period, but in the corresponding meta‐analysis there was no evidence of an effect (high‐ and moderate‐quality evidence, respectively). 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).

Six studies with 116 participants examined the effects of tDCS on cognitive abilities (including spatial neglect). Two studies with 56 participants showed no evidence of an effect on cognitive abilities (SMD 0.46, 95% CI ‐0.10 to 1.02; low‐quality evidence) and another three studies, which could not be included in meta‐analysis reported evidence of an effect. One study with 30 participants showed evidence of effect on spatial neglect in meta‐analysis (MD 4.80 points in the line‐bisection test, 95% CI 0.13 to 9.47: very low‐quality evidence) and we identified another randomised cross‐over trial with 15 participants that examined the effects of tDCS on neglect (but could not be included in meta‐analysis); this trial reported evidence of an effect of tDCS on neglect.

Forty‐one of 60 studies (74%) reported data on dropouts, and 33 of 60 studies (55%) reported data on adverse events. In 25 of 60 studies (42%), dropouts, adverse events or deaths that during the intervention period occurred. We found no evidence of an effect regarding differences in dropouts, adverse effects and deaths between intervention and control groups (RR 1.25, 95% CI 0.74 to 2.13; Mantel‐Haenszel method with random‐effects model; analysis based only on studies that reported either on dropouts or on adverse events or on both; moderate‐quality evidence).

A summary of this comparison's main findings can be found in summary of findings Table 1 and summary of findings Table 3.

Comparison 2: tDCS versus any type of active control intervention

We identified seven studies with 209 participants comparing active tDCS with an active control intervention (physiotherapy or virtual reality).

We found three studies with 121 participants examining the effects of tDCS on our primary outcome measure, ADL, after stroke. We found low‐quality evidence of effect regarding ADL performance at the end of the intervention period (MD 6.59 BI points, 95% CI 1.26 to 11.91). There were no studies examining the effect of tDCS at follow‐up.

One of our secondary outcome measures was upper extremity function: five trials with a total of 124 participants measured upper extremity function at the end of the intervention period, revealing evidence of an effect in favour of tDCS (SMD 0.84, 95% CI 0.2 to 1.48 for studies presenting absolute values; low‐quality evidence, and SMD 0.51, 95% CI 0.2 to 1.22 for studies presenting change values; low‐quality evidence). Regarding the effects of tDCS on upper extremity function at the end of follow‐up, we identified one study with a total of 32 participants presenting change values that showed no evidence of an effect (MD 10% change in JTT‐time, 95% CI ‐0.07 to 20.07; moderate‐quality evidence). Three studies with 66 participants examined the effect of tDCS on lower extremity function, but did not show evidence of an effect (moderate‐quality evidence). Two studies with 57 participants reported outcome data for muscle strength at the end of the intervention period, but in the corresponding meta‐analysis there was no evidence of an effect (low‐quality evidence). We could not identify any study examining the effects of tDCS on muscle strength at follow‐up and no studies examining the effects of tDCS on cognitive abilities and spatial neglect. We identified one study with a total of 12 participants presenting change values that showed no evidence of an effect, but no meta‐analysis was possible.

Seven of seven studies (100%) reported data on dropouts, and four of seven studies (57%) reported data on adverse events. In two of seven studies (29%), dropouts, adverse events or deaths occurred during the intervention period. We found no evidence of an effect regarding differences in dropouts, adverse effects and deaths between intervention and control groups (RR 1.76, 95% CI 0.43 to 7.17; Mantel‐Haenszel method with random‐effects model; analysis based only on studies that reported either on dropouts or on adverse events or on both; moderate‐quality evidence).

A summary of this comparison's main findings can be found in summary of findings Table 2 and summary of findings Table 4.

Overall completeness and applicability of evidence

The results of this review appear to be generalisable to other settings in industrialised countries. However, some factors suggest uncertainty in generalisations. These include the following.

Most of the studies included participants with first‐time stroke.
Most participants suffered from ischaemic stroke.

Hence, the results may be of limited applicability for people with recurrent and haemorrhagic strokes. Moreover, completeness of evidence is lacking regarding studies on the effects of tDCS on lower limb function, cognitive abilities (including spatial neglect), and the reporting of adverse events.

The physiological mechanisms of tDCS are not yet fully understood (Buch 2017). Included studies are heterogeneous in terms of type, location and duration of stimulation, amount of direct current delivered, electrode size and positioning, and participants with cortical and subcortical stroke. For example, recent research suggests that the effectiveness of C‐tDCS over the contralesional M1 depends on corticospinal tract integrity, thus implicating that this is not a 'one size fits all' intervention (Byblow 2011). Hence, it could be that this heterogeneity, even in the absence of excess heterogeneity in our analyses, produces a false‐negative finding (Antal 2015). It also has been proposed to conduct pragmatic and large RCTs in order to better identify treatment responders (Grefkes 2016).

Forty‐eight of 67 studies (74%) reported data on dropouts, and 36 of 67 studies (55%) reported data on adverse events. In our analyses of adverse events, we therefore decided to include only studies that reported either on dropouts, or on adverse events, or on both. However, it could be that the real risk of dropouts or adverse events is underestimated in our analysis, since the analysis could be prone to reporting bias.

Quality of the evidence

Based on our assessments of the evidence provided in summary of findings Table 1, summary of findings Table 2, summary of findings Table 3 and summary of findings Table 4, we downgraded evidence quality due to several included studies with high risk of bias and the imprecision of effect estimates that included the effect size of no difference in the comparators. We also found heterogeneity regarding trial design (parallel‐group or cross‐over design, two or three intervention groups), therapy variables (type of stimulation, location of stimulation, dosage of stimulation) and participant characteristics (age, time post‐stroke, severity of stroke/initial functional impairment).

Potential biases in the review process

The methodological rigour of Cochrane Reviews minimises bias during the process of conducting systematic reviews. However, some aspects of this review represent an 'open door' to bias, such as eliminating obviously irrelevant publications according to titles and abstracts, based on the determination of only one review author (BE). This encompasses the possibility of unintentionally ruling out relevant publications. Another possibility is that publication bias could have affected our results. With the funnel plot for our main outcome of ADL (at the end of the intervention period) showing no asymmetry, a small‐study effect or publication bias nevertheless could exist, resulting in overestimation of the effects (Figure 4) (Sterne 2011).

Another potential source for the introduction of bias is that two of the review authors (JM and MP) were involved in conducting and analysing the largest of the included trials (Hesse 2011). However, in our review, they did not participate in extracting outcome data and determining risk of bias for Hesse 2011. They were replaced by another review author (JK), so that the introduction of bias is unlikely in this case.

We had to exclude nine trials from quantitative synthesis (meta‐analysis) because of missing information regarding treatment order (i.e. the first intervention period of the cross‐over trial) (Au‐Yeung 2014; Fregni 2005a; Jo 2008a; Kang 2008b; Kim 2009; Klomjai 2018; Ko 2008a; Mahmoudi 2011; Sohn 2013; Sunwoo 2013a). However, the results of these trials regarding upper and lower extremity function and spatial neglect but not on cognitive abilities are mostly consistent with the results of comparisons made in our meta‐analyses, and it is therefore unlikely that the results of these studies would have substantially altered our results.

Agreements and disagreements with other studies or reviews

As far as we know, there are several systematic reviews on the effects of tDCS on function after stroke: Tedesco Triccas 2015a included true RCTs with multiple sessions of tDCS. They included nine studies with 371 participants and showed no evidence of effect at the end of the intervention period (SMD 0.11, 95% CI ‐0.17 to 0.38) or at long‐term follow‐up (SMD 0.23, 95% CI ‐0.17 to 0.62). These results are similar to the results of our analyses regarding the effects of tDCS (combined) on upper limb function.

Another systematic review of quasi‐randomised and properly randomised controlled trials has examined the effects of A‐tDCS on upper limb motor recovery in stroke patients (Butler 2013). The review authors included eight trials with 168 participants, and their analysis revealed evidence of an effect of tDCS on upper limb function (SMD 0.49, 95% CI 0.18 to 0.81), mainly measured by the JTT. This is different to our results, which may be explained by a different search strategy, different selection criteria and a different outcome measure.

In another systematic review on the effects of tDCS, Adeyemo 2012 included 50 non‐randomised trials and RCTs with 1314 participants (1282 people with stroke and 32 healthy volunteers) on the pooled effects of tDCS and rTMS on motor outcomes after stroke. With their analysis based on change values, they revealed an effect of SMD 0.59, 95% CI 0.42 to 0.76). These results differ from the results of our analyses, perhaps because the review by Adeyemo 2012 included non‐randomised studies, which tend to overestimate treatment effects (Higgins 2011a), and because of that review's statistical pooling of tDCS data with trials examining the effects of rTMS on motor outcomes after stroke.

Two other systematic reviews included meta‐analyses dealing with the topic of tDCS for improving function after stroke (Bastani 2012; Jacobson 2012). Bastani 2012 examined the effects of A‐tDCS on cortical excitability (as measured by transcranial magnetic stimulation (TMS)) and upper extremity function (mainly measured by JTT) in healthy volunteers and people with stroke. Their analysis of the effects of A‐tDCS over the lesioned hemisphere, based mainly on results of randomised cross‐over studies, yielded no evidence of effect (SMD 0.39, 95% CI ‐0.17 to 0.94). Jacobson 2012, a review about the effects of A‐tDCS and C‐tDCS on healthy volunteers, stated that the anodal‐excitation and cathodal‐inhibition (AeCi) dichotomy is relatively consistent regarding the effects of tDCS on function in healthy volunteers. However, we found no evidence of effect for A‐tDCS over the lesioned hemisphere in our planned subgroup analysis, which is consistent with the findings of Bastani 2012, but not with the findings of Suzuki 2012. In contrast to that, we found evidence of an effect of tDCS on ADL for C‐tDCS over the non‐lesioned hemisphere, which in turn is consistent with the findings of Suzuki 2012. However, when compared with the subgroups, A‐tDCS over the lesioned hemisphere and dual‐tDCS, the subgroup C‐tDCS over the non‐lesioned hemisphere has the highest statistical power.

O'Brien 2018 and colleagues performed a systematic review with meta‐analysis of RCTs examining the effect of tDCS and rTMS on fine motor improvement after stroke and in healthy volunteers. There was evidence of an effect of tDCS (SMD 0.31, 95% CI 0.08 to 0.55, 18 studies), which is comparable to our findings. Another published systematic review with meta‐analysis dealt with the effects of tDCS on walking ability after stroke (Li 2018). The authors included 10 RCTs with 194 participants and showed evidence of effect of tDCS on mobility (SMD 0.44, 95% CI 0.01 to 0.87) and muscle strength of the lower limb (SMD 1.54, 95% CI 0.29 to 2.78), but not on walking endurance (SMD 0.28, 95% CI ‐0.28 to 0.84) and balance function (SMD 0.44, 95% CI ‐0.06 to 0.94). In our analyses, there was no evidence of effect regarding lower limb function and muscle strength, which may be due to a different search strategy, different selection criteria and a different approach to statistical analysis. A published systematic review with meta‐analysis on motor‐learning after stroke showed that there is evidence of a longer‐term retention effect of tDCS (SMD 0.59, 95% CI 0.40 to 0.79; mean retention interval 44 days) (Kang 2016). It also showed evidence of an effect of A‐tDCS (SMD 0.59, 95% CI 0.20 to 0.97) and C‐tDCS (SMD 0.60, 95% CI 0.15 to 1.04) as well as Dual tDCS (SMD 0.68, 95% CI 0.37 to 0.99), which is not consistent with our findings, since in our subgroup analysis there was only evidence of an effect of C‐tDCS. This difference may be explained by a different search strategy and different selection criteria. Another published systematic review with meta‐analysis about the use of tDCS in post‐stroke upper extremity motor recovery found evidence of an effect of tDCS (SMD 0.61, 95% CI 0.08 to 1.13, eight studies with 213 participants), which principally is in accordance with our findings, although the effect size in our analyses was smaller (Chhatbar 2016). Furthermore, they found a relatively large effect size for tDCS in people with chronic stroke (SMD 1.23, 95% CI 0.20 to 2.25), which is not consistent with our findings. This discrepancy may be explained by a different search strategy and different selection criteria. Another systematic review found evidence of an effect of tDCS on motor‐evoked potentials (MEP), but not on physiologic parameters, which is not in accordance with our findings (Horvath 2015). Most of the published systematic reviews to date have focused on the effects of tDCS on function and ADL. A systematic review with meta‐analysis on the efficacy of non‐invasive brain stimulation on spatial neglect after stroke concluded that there is evidence of effect of tDCS for improving neglect of the stroke (SMD 0.51, 95% CI 0.01 to 1.02), which is consistent with our findings (Fan 2018).

There is also a comprehensive published guideline on the therapeutic use of tDCS, which also covers the application in people with stroke In order to improve motor function (Lefaucheur 2017). The guideline states that there is insufficient evidence to either refute or recommend tDCS in routine clinical practice for improving motor function after stroke and hence gives no recommendation regarding its use.

Further research for optimising stimulation parameters is needed. Further directions in tDCS research should aim at identifying the patients who may benefit the most from tDCS by, for example high definition (HD)‐tDCS to increase focality, tDCS during MRI to increase spatial resolution and tDCS with concomitant EEG to increase temporal resolution (Elsner 2018). Future research should adhere to the Stroke Recovery and Rehabilitation Roundtable's core recommendations regarding the development, monitoring and reporting of stroke rehabilitation research (Walker 2017).

To our knowledge, our review, including 67 true RCTs with a total of 1729 participants, is the most comprehensive review about the effects of tDCS on ADL, function, muscle strength and cognitive abilities (including spatial neglect) in stroke.

Figure 1

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 4

Figure 5

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.2 Primary outcome measure: ADL until the end of follow‐up.

Figure 6

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.3 Secondary outcome measure: upper extremity function at the end of the intervention period.

Figure 7

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.4 Secondary outcome measure: upper extremity function to the end of follow‐up.

Figure 8

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.5 Secondary outcome measure: lower extremity function at the end of the intervention period.

Figure 9

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.6 Secondary outcome measure: muscle strength at the end of the intervention period.

Figure 10

Funnel plot of comparison: 1 tDCS versus any type of placebo or passive control intervention, outcome: 1.7 Secondary outcome measure: muscle strength at the end of follow‐up.

Figure 11

Analysis 1.1

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 1: Primary outcome measure: ADL at the end of the intervention period

Analysis 1.2

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 2: Primary outcome measure: ADL until the end of follow‐up

Analysis 1.3

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 3: Secondary outcome measure: upper extremity function at the end of the intervention period

Analysis 1.4

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 4: Secondary outcome measure: upper extremity function to the end of follow‐up

Analysis 1.5

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 5: Secondary outcome measure: lower extremity function at the end of the intervention period

Analysis 1.6

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 6: Secondary outcome measure: muscle strength at the end of the intervention period

Analysis 1.7

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 7: Secondary outcome measure: muscle strength at the end of follow‐up

Analysis 1.8

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 8: Secondary outcome measure: cognitive abilities at the end of the intervention period

Analysis 1.9

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 9: Secondary outcome measure: hemispatial neglect at the end of intervention period

Analysis 1.10

Comparison 1: tDCS versus any type of placebo or passive control intervention, Outcome 10: Secondary outcome measure: dropouts, adverse events and deaths during the intervention period

Analysis 2.1

Comparison 2: tDCS versus any type of active control intervention, Outcome 1: Primary outcome measure: ADL at the end of the intervention period, absolute values

Analysis 2.2

Comparison 2: tDCS versus any type of active control intervention, Outcome 2: Secondary outcome measure: upper extremity function at the end of the intervention period

Analysis 2.3

Comparison 2: tDCS versus any type of active control intervention, Outcome 3: Secondary outcome measure: upper extremity function to the end of follow‐up

Analysis 2.4

Comparison 2: tDCS versus any type of active control intervention, Outcome 4: Secondary outcome measure: lower extremity function at the end of the intervention period

Analysis 2.5

Comparison 2: tDCS versus any type of active control intervention, Outcome 5: Secondary outcome measure: muscle strength at the end of the intervention period

Analysis 2.6

Comparison 2: tDCS versus any type of active control intervention, Outcome 6: Secondary outcome measure: spatial neglect at the end of the intervention period

Analysis 2.7

Comparison 2: tDCS versus any type of active control intervention, Outcome 7: Secondary outcome measure: dropouts, adverse events and deaths during the intervention period

Analysis 3.1

Comparison 3: Subgroup analyses for primary outcome measure: ADL at the end of the intervention period, Outcome 1: Planned analysis: duration of illness ‐ acute/subacute phase versus postacute phase for ADL at the end of the intervention period

Analysis 3.2

Comparison 3: Subgroup analyses for primary outcome measure: ADL at the end of the intervention period, Outcome 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)

Analysis 3.3

Comparison 3: Subgroup analyses for primary outcome measure: ADL at the end of the intervention period, Outcome 3: Planned analysis: type of control intervention (sham tDCS, conventional therapy or nothing)

Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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 1)
		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 1)
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.
^aDowngraded one level because 95% CI contains effect size of the minimal important difference. ^bDowngraded one level because the total sample size is less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^cDowngraded one level due to study ratings with 'high' risk of bias ^dDowngraded one level because 95% CI contains effect size of no difference and the minimal important difference. ^ePublication bias strongly suspected by visual inspection of funnel plot. Downgraded one level.

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

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)
CI: confidence interval SMD: standardised mean difference

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

Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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.
^aDowngraded one level due to total sample size being less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^bDowngraded one level due to several study ratings with 'high' risk of bias.

Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)	Comments
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
	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.
^aDowngraded one level due to study ratings with 'high' risk of bias. ^bDowngraded one level because the total sample size is less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria). ^cDowngraded one level because publication bias strongly suspected.

Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)
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
Patient or population: people with stroke Settings: inpatient and outpatient Intervention: tDCS
	Assumed risk	Corresponding risk
	Control	tDCS
Primary outcome measure: mean ADL at the end of follow‐up Scale from: 0 to 100.	No evidence available
Secondary outcome measure: mean upper extremity function to the end of follow‐up per cent change in Jebsen‐Taylor‐Test		Mean in the intervention groups was 10 higher (0.07 lower to 20.07 higher)	32 (1 study)	⊕⊕⊕⊝ moderate^a
Secondary outcome measure: lower extremity function at the end of follow‐up	No evidence available
Secondary outcome measure: muscle strength at the end of follow‐up	No evidence available
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 life; 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.
^aDowngraded one level due to total sample size being less than 400 (as a rule of thumb for implementing GRADE 'optimal information size' criteria).

Table 2. Patient characteristics

Study ID	Experimental: age, mean (SD)	Control: age, mean (SD)	Experimental: time post stroke, mean (SD)	Control: time post stroke, mean (SD)	Experimental: sex, n (%)	Control: sex, n (%)	Experimental: lesioned hemisphere, n (%)	Control: lesioned hemisphere, n (%)	Experimental: severity, mean (SD)	Control: severity, mean (SD)	Experimental: lesion cause/ location, n (%)	Control: lesion cause/ location, n (%)	Handedness, n (%)
Allman 2016	60 (12) years	67 (10) years	51 (33) months	57 (40) months	3 (27) female	4 (31) female	3 (27) left	4 (31) left	UE‐FM 39 (16)	UE‐FM 36 (17)	2 (18) cortical	4 (31) cortical	Not stated
Andrade 2017	54 (4) years	55 (4) years	2 (2) months	2 (1) months	18 (45) female	8 (40) female	20 (50) left	10 (50) left	NIHSS 17 (1)	NIHSS17 (1)	15 (38) haemorrhagic, 17 (43) cortical	5 (25) haemorrhagic, 8 (40) cortical	Not stated
Ang 2012	52 (12) years	56 (10) years	3 (2) years	3 (1) years	4 (40) female	1 (11) female	5 (50) left	6 (67) left	UE‐FM 35 (8)	UE‐FM 33 (8)	6 (60) ischaemic; 1 (10) cortical, 9 (90) subcortical	7 (78) ischaemic; 9 (100) subcortical	Not stated
Au‐Yeung 2014	63 (6) years		8 (3) years		0 female		5 (50) left		UE‐FM 58 (8); MMSE 29 (2)		8 (80) ischaemic		10 (100) right‐handed
Bang 2015	66 (4) years	66 (5) years	7 (2) weeks	7 (1) weeks	2 (50) female	2 (50) female	6 (100) right	6 (100) right	MBI 51 (5)	MBI 50 (6)	Not described		Not stated
Boggio 2007a	56 (11) years	75 (NA) years	33 (34) months	39 months	3 (100) male	1 (100) male	2 (67) left	1 (100) left	MRC 4.2 (0.53)	MRC 4.7 (NA)	3 (100) ischaemic and subcortical	1 (100) ischaemic and subcortical	12 (100) right‐handed
Bolognini 2011	43 (13) years	51 (15) years	44 (31) months	26 (18) months	4 (57) female	5 (71) female	4 (57) left	4 (57) left	BI 18.13 (2.42)	BI 14.33 (5.46)	2 (29) haemorrhagic, 5 (71) ischaemic	7 (100) ischaemic	14 (100) right‐handed
Cha 2014	60 (11) years	58 (10) years	14 (5) months	15 (4) months	Not stated	Not stated	4 (40) left	5 (50) left	Brunnstrom 5 (1)	Brunnstrom 5 (1)	Not stated	Not stated	Not stated
Chang 2015	60 (10) years	66 (11) years	16 (6) days	17 (5) days	9 (38) female		6 (50) left	5 (42) left	NIHSS 7 (4)	NIHSS 9 (5)	24 (100) ischaemic/11 (46) corona radiata, 7 (29) MCA, 4 (17) MCA border zone, 2 (8) internal capsule		Not stated
Chelette 2014	58(7) years	62 (5) years	5 (2) years	5 (1) years	9 (45) female	1 (17) female	14 (70) left	1 (17) left	SIS 62 (13)	SIS 57 (18)	16 (80) ischaemic/17 (81) cortical	6 (100) ischaemic/4 (67) cortical	16 (80) right‐handed
Cho 2017	61 (9) years	58 (13) years	14 (6) days	14 (5) days	6 (40) female	7 (47) female	7 (47) left	7 (47) left	UE‐FM 50 (19)	UE‐FM 41 (13)	12 (75) ischaemic/5 (33) cortical	13 (87) ischaemic/4 (27) cortical	Not stated
Cunningham 2015	64 (8) years	59 (10) years	63 (81) months	37 (27) months	2 (33) female	2 (33) female	2 (33) left	4 (67) left	UE‐FM 41 (14)	UE‐FM 47 (11)	2 (33) haemorrhagic	2 (33) haemorrhagic	Not stated
D'Agata 2016	57 (12) years	65 (12) years	41 (39) months	37 (32) months	8 (33) female	3 (39) female	12 (50) left	6 (60) left	Not described clearly		17 (71) ischaemic/6 (25) cortical, 15 (62) subcortical, 3 (13) corticosubcortical	7 (70) ischaemic/1 (10) cortical, 8 (80) subcortical, 1 (10) corticosubcortical	Not stated
Danzl 2012	65 (15) years	71 (11) years	57 /55) months	39 (33) months	1 (25) female	3 (75) female	4 (100) left	4 (100) left			2 (50) ischaemic/not described	4 (100) ischaemic/not described	Not stated
Di Lazzaro 2014a	66 (16) years	71 (14) years	3 (1) days	3 (1) days	2 (29) female	3 (43) female	3 (43) left	3 (43) left	NIHSS 7 (5)	NIHSS 7 (4)	7 (100) ischaemic; 3 (43) subcortical; 4 (57) corticosubcortical	7 (100) ischaemic; 2 (29) subcortical, 5 (71) corticosubcortical	Not stated
Di Lazzaro 2014b	61 (16) years	69 (12) years	3 (2) days	3 (1) days	4 (40) female	6 (60) male	2 (20) left	6 (60) left	NIHSS 6 (3)	NIHSS 6 (2)	10 (100) ischaemic; 4 (40) subcortical, 6 (60) corticosubcortical	10 (100) ischaemic; 4 (40) subcortical, 6 (60) corticosubcortical	Not stated
Fregni 2005a	54 (17) years		27 (24) months		2 (33) female		3 (50) left		MRC 4.18 (0.37)		Cause not clearly stated by the authors		6 (100) right‐handed
Fusco 2013a	44 (16) years	65 (22) years	31 (13) days	25 (5) days	3 (60) female	1 (25) female	3 (60) left	2 (50) left	Grasp force 17.83 (7.45) kg		5 (100) ischaemic	3 (75) ischaemic, 1 (25) haemorrhagic	9 (100) right‐handed
Fusco 2014	56 (15) years	60 (12) years	19 (8) days		3 (60) female	3 (50) female	2 (40) left	2 (33) left	BI 33 (22)	BI 51 (34)	5 (100) ischaemic	6 (100) ischaemic	9 (73) right‐handed
Geroin 2011	64 (7) years	63 (6) years	26 (6) months	27 (5) months	2 (20) female	4 (40) female	Not stated by the authors	Not stated by the authors	ESS 79.6 (4.1)	ESS 79.6 (2.7)	10 (100) ischaemic; 4 (40) cortical, 3 (30) corticosubcortical, 3 (30) subcortical	10 (100) ischaemic; 5 (50) cortical, 3 (30) corticosubcortical, 2 (20) subcortical	Not stated by the authors
Hamoudi 2018	62 (13) years	62 (13) years for sham tDCS and 65 (2) for passive control group	48 (80) months	44 (51) months for sham tDCS and 23 (4) months for passive control group	6 (33) female	3 (17) and 6 (43) female	9 (50) left	8 (44) and 7 (50) left	UE‐FM 59 (4)	UE‐FM 59 (4) and 59 (4)	18 (100) ischaemic/9 (50) subcortical	18 (100) ischaemic/9 (50) subcortical and 14 (100) ischaemic/7 (50) subcortical	EHI 78 in the Exp group, EHI 84 in the Sham group and EHI 90 in the Ctl group
Hathaiareerug 2019	56 (8) years	59 (10) years	6 (4) months	5 (3) months	1 (11) female	2 (22) female	4 (44) left	2 (22) left	UE‐FM 38 (17)	UE‐FM 32 (14)	6 (67) ischaemic/1 (11) cortical, 4 (44) subcortical, 4 (44) corticosubcortical	7 (77) ischaemic/1 (11) cortical, 3 (33) subcortical, 5 (55) corticosubcortical	89% right‐handed
Hesse 2011	65 (10) years	66 (10) years	4 (2) weeks	4 (2) weeks	26 (41) female	11 (34) female	35 (55) left	16 (50) left	BI 34.15 (6.97); UE‐FM 7.85 (3.58)	BI 35.0 (7.8); UE‐FM 8.2 (4.4)	64 (100) ischaemic; 29 (45) TACI, 20 (31) PACI, 15 (23) LACI	32 (100) ischaemic; 13 (41) TACI, 13 (41) PACI, 6 (18) LACI	Not stated by the authors
Ilić 2016	58 (8) years	62 (4) years	41 (24) months	37 (21) months	10 (71) female	7 (58) female	13 (50) left		UE‐FM 47 (8)	UE‐FM 51 (6)	26 (100) ischaemic/26 (100) subcortical		24 (92) right‐handed
Jo 2008a	48 (9) years		2 (1) months		3 (30) female		10 (100) right		Not reported		4 (40) ischaemic		Not stated by the authors
Kang 2008b	70 (3) years		544 (388) days		4 (40) female		7 (70) right		21 (1) MMSE		7 (70) ischaemic		Not stated by the authors
Khedr 2013	59 (9) years	57 (8) years	13 (5) days	13 (5) days	9 (33) female	5 (38) female	12 (44) left	6 (46) left	BI 32.76 (10.75)	BI 31.1 (12.6)	27 (100) ischaemic; 12 (44) cortical, 5 (19) corticosubcortical, 10 (37) subcortical	13 (100) ischaemic; 6 (42) cortical, 3 (23) corticosubcortical, 4 (31) subcortical	Not stated by the authors
Kim 2009	63 (13) years		6 (3) weeks		7 (70) female		8 (80) left		MRC between 3 and 5 for the all paretic finger flexors and extensors		8 (80) infarction, 2 (20) haemorrhage		Not stated by the authors
Kim 2010	54 (15) years	63 (9) years	27 (21) days	23 (8) days	2 (18) female	3 (43) female	7 (64) left	2 (29) left	BI 71.77 (23.86) UE‐FM 34.7 (15.0)	BI 67.9 (22.4) UE‐FM 41.0 (13.0)	11 (100) ischaemic; 3 (27) cortical, 3 (27) corticosubcortical, 5 (71) subcortical	7 (100) ischaemic; 2 (29) cortical, 1 (14) corticosubcortical, 4 (57) subcortical	Not stated by the authors
Kim 2016	59 (13) years	52 (11) years	15 (6) months	15 (7) months	5 (33) female	6 (40) female	8 (53) left	7 (47) left	FIM 67 (10)	FIM 80 (11)	4 (27) ischaemic/not stated	10 (67) ischaemic/not stated	Not stated by the authors
Ko 2008a	62 (9) years		29‐99 days		5 (33) female		15 (100) right		19 per cent deviation (11)		10 (66) ischaemic		15 (100) right‐handed
Koo 2018	52 (3) years	59 (3) years	19 (8) months	20 (8) months	7 (58) female	6 (50) female	6 (50) left	8 (75) left	MBI 35 (16)	MBI 38 (20)	4 (33) ischaemic; 3 (25) cortical, 9 (75) subcortical	7 (58) ischaemic;2 (17) cortical, 8 (67) subcortical, 2 (17) brain stem	24 (100) right handed
Klomjai 2018	57 (12) years		3 (2) months		5 (26) female		12 (63) right		TUG 21 (13) s	TUG 20 (13) s	19 (100) ischaemic		16 (84) right‐handed
Lee 2014	62 (11) years	61 (14) years	18 (8) days	17 (6) days	17 (44) female	9 (45) female	19 (49) left	13 (65)	UE‐FM 37 (23)	UE‐FM 35 (22)	21 (54) ischaemic; 21 (54) cortical; 18 (46) subcortical	14 (70) ischaemic; 10 (50) cortical; 10 (50) subcortical	Not stated by the authors
Lindenberg 2010	62 (15) years	56 (13) years	31 (21) months	40 (23) months	2 (20) female	3 (30) female	6 (60) left	7 (70) left	UE‐FM 38.2 (13.3)	UE‐FM 39.8 (11.5)	10 (100) ischaemic	10 (100) ischaemic	19 (95) right‐handed, 1 (5) both‐handed
Mahmoudi 2011	61 (14) years		8 (5) months		3 (33) female		6 (60) left, 3 (30) right, 1 (10) brainstem		JTT (without handwriting): 12.3 (7.3) s		10 (100) ischaemic		Not stated by the authors
Manji 2018	62 (10) years	64 (11) years	4 (2) months	5 (1) months	5 (33) female	4 (27) female	Not reported		FIM 107 (10)	FIM 104 (10)	9 (60) ischaemic	8 (16) ischaemic	Not stated by the authors
Mazzoleni 2019	68 (16) years	69 (16) years	Not reported		12 (60) female	12 (63) female	11 (55) left	11 (58) left	CMMSA 4.3 (1.4)	CMMSA 5.1 (1.1)	13 (65) ischaemic	16 (84) ischaemic	38 (97) right‐handed
Mortensen 2016	66 (11) years	61 (10) years	32 (16) months	29 (15) months	4 (50) female	2 (29) female	4 (50) left	4 (57) left	JTT 69 (29) s	JTT 55 (18) s	0 ischaemic	0 ischaemic	Not stated by the authors
Nair 2011	61 (12) years	56 (15) years	33 (20) months	28 (28) months	2 (29) female	3 (43) female	3 (43) left	5 (71) left	UE‐FM 30 (11)	UE‐FM 31 (10)	7 (100) ischaemic; 5 (71) cortical and corticosubcortical, 2 (29) subcortical	7 (100) ischaemic; 4 (56) cortical and corticosubcortical, 3 (43) subcortical	14 (100) right‐handed
Nicolo 2017	65 (12) years	64 (17) years	1 (0.4) months	1 (0.3) months	13 (46) female	5 (38) female	4 (29) left	5 (36) left	NIHSS 13 (6)	NIHSS 12 (5)	13 (46) ischaemic; 4 (14) cortical, 16 (67) corticosubcortical, 8 (29) subcortical	10 (71) ischaemic; 1 (8) cortical, 6 (46) corticosubcortical, 6 (46) subcortical	39 (95) right‐handed
Park 2013	65 (14) years	66 (11) years	29 (19) days	25 (17) days	6 (67) female	2 (40) female	2 (33) left	2 (40) left	NIHSS 8 (3)	NIHSS 10 (3)	4 (67) ischaemic	3 (60) ischaemic	Not stated by the authors
Park 2015	59 (6) years	60 (13) years	19 (12) months	24 (16) months	Not reported		9 (56) left	3 (19) left	Gait speed 0.7 (0.3) m/s	Gait speed 0.6 (0.3) m/s	4 (25) ischaemic	4 (50) ischaemic	Not stated by the authors
Picelli 2015	64 (9) years	61 (7) years	57 (35) months	55 (33) months	6 (30) female	2 (20) female	Not reported		6MWT 181 (79) m	6MWT 183 (51) m	7 (35) cortical; 7 (35) corticosubcortical; 6 (30) subcortical	4 (40) cortical; 4 (40) corticosubcortical; 2 (20) subcortical	Not stated by the authors
Qu 2009	45 (11) years	45 (14) years	6 (range 3 to 36) months	4 (range 3 to 12) months	4 (16) female	3 (12) female	14 (56) left	13 (52) left	BI 64 (17)	BI 72 (22)	10 (40) haemorrhagic, 15 (60) infarction	10 (40) haemorrhagic, 15 (60) infarction	Not stated by the authors
Qu 2017	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described	Not described
Rabadi 2017	62 (11) years	63 (6) years	7 (4) days	6 (3) days	0 female	0 female	4 (50) left	2 (25) left	FIM 61 (17)	FIM 59 (12)	8 (100) ischaemic	8 (100) ischaemic	15 (94) right‐handed
Rocha 2016	58 (range 41‐71) years	59 (range 46‐70) years	31 months (range 9‐67)	27 months (6‐46)	3 (21) female	3 (43) female	8 (57) left	3 (43) left	UE‐FM 48 (6)	UE‐FM 51 (9)	Not stated by the authors		21 (100) right‐handed
Rossi 2013	66 (14) years	70 (14) years	2 days	2 days	13 (52) female	11 (44) female	18 (72) left	16 (64) left	UE‐FM 4.1 (6.4)	FM 4.6 (7.8)	25 (100) ischaemic; 1 (4) cortical, 17 (68) corticosubcortical, 7 (28) subcortical	25 (100) ischaemic; 2 (8) cortical, 18 (72) corticosubcortical, 5 (20) subcortical	Not stated by the authors
Saeys 2015	62 (10) years	65 (7) years	46 (22) days	38 (15) days	7 (44) female	7 (47) female	11 (92) left	6 (55) left	Tinetti 8 (7)	Tinetti 9 (6)	15 (94) ischaemic	11 (73) ischaemic	Not stated by the authors
Salazar 2019	60 (10) years	56 (16) years	21 months (range 6‐59)	23 months (range 8‐59)	5 (33) female	5 (33) female	8 (53) left	8 (53) left	median UE‐FM 25 points (range 9‐46)	median UE‐FM 29 (range 16‐46)	14 (93) ischaemic	11 (73) ischaemic	27 (90) right handed
Sattler 2015	68 (10) years	63 (12) years	5 (3) days	6 (4) days	3 (30) female	3 (30) female	Not exactly described		NIHSS 3 (1); UE‐FM 47 (3)	NIHSS 3 (2), UE‐FM 49 (3)	Not exactly described		All patients were right handed
Seo 2017	61 (9) years	63 (9) years	76 (83) months	153 (123) months	2 (18) female	3 (30) female	6 (55) left	2 (20) left	MRS 3 (0.5)	MRS 3 (0.4)	9 (82) ischaemic	7 (70) ischaemic	Not stated by the authors
Shaheiwola 2018	49 (9) years	52 (11) years	18 (15) months (median(IQR))	16(13) months (median(IQR))	1 (7) female	2 (13) female	7 (47) left	9 (60) left	UE‐FM 16 (9)	UE‐FM 18 (13)	Not exactly described
Sik 2015	60 (IQR 54‐68) years	60 (IQR 55‐67) years	22 (32) months (median(IQR))	18 (19) months (median(IQR))	10 (50) female	3 (27) female	10 (50) left	5 (45) left	Not exactly described		19 (95) ischaemic	10 (91) ischaemic	Not stated by the authors
Sohn 2013	58 (15) years		63 (17) days		2 (18) female		6 (55) left		Not stated by the authors		4 (36) ischaemic		Not stated by the authors
Straudi 2016	53 (16) years	64 (10) years	41 (35) weeks	78 (62) weeks	7 (58) female	4 (36) female	9 (75) left	6 (55) left	UE‐FM 28 (19)	UE‐FM 37 (14)	7 (83) ischaemic; 9 (75) cortical, 3 (25) subcortical	9 (82) ischaemic; 5 (45) cortical, 6 (55) subcortical	Not stated by the authors
Sunwoo 2013a	63 (13) years		28 (60) months		6 (60) female		10 (100) left		MMSE 28 (2)		7 (70) ischaemic		10 (100) right‐handed
Tahtis 2012	67 (12) years	56 (12) years	20 (5) days	25 (11) days	2 (29) female	1 (14) female	3 (43) left	3 (43) left	MRS 2 (1)	MRS 3 (1)	7 (100) ischaemic; 4 (57) cortical, 3 (43) subcortical	7 (100) ischaemic; 3 (43) cortical; 4 (57) subcortical	Not stated by the authors
Tedesco Triccas 2015b	64 (10) years	63 (14) years	25 (31) months	13 (16) months	5 (42) female	4 (33) female	6 (50) left	5 (45) left	UE‐FM 28 (19)	UE‐FM 37 (14)	3 (25) ischaemic; 3 (25) cortical, 9 (75) subcortical	9 (81) ischaemic; 4 (36) cortical; 7 (64) subcortical	22 (96) right‐handed
Utarapichat 2018	57 (12) years		34 (19) months		4 (40) female		5 (50) left		MRC knee extensor 4		10 (100) ischaemic		Not stated by the authors
Viana 2014	56 (10) years	55 (12) years	32 (18) months	35 (20) months	1 (10) female	3 (30) female	5 (50) left	3 (30) left	UE‐FM 41 (16)	UE‐FM 39 (17)	9 (90) ischaemic	10 (100) ischaemic	19 (95) right‐handed
Wang 2014	54 (14) years	52 (9) years	Not explicitly stated, but all participants were enrolled between 1 and 4 weeks post stroke		1 (16) female	1 (33) female	2 (33) left	0 left	FIM 59 (18)	FIM 74 (8)	6 (100) ischaemic	3 (100) ischaemic	Not stated by the authors
Wong 2015	69 (10) years		11 (5) days		11 (65) female		Not explicitly stated		Not explicitly stated		Not stated by the authors		Not stated by the authors
Wu 2013a	46 (11) years	49 (13) years	5 (3) months	5 (3) months	11 (24) female	10 (22) female	24 (53) left	23 (51) left	BI 55 (range 0 to 85) UE‐FM 12.3 (5.5)	BI 55 (range 25 to 95) UE‐FM 11.8 (8.2)	27 (60) ischaemic, 18 (40) haemorrhagic	26 (58) ischaemic, 19 haemorrhagic (42)	Not stated by the authors
Yi 2016	62 (11) years	62 (10) years	Not stated		5 (25) female	4 (40) female	None	None	Not stated	Not stated	Not explicitly stated	Not explicitly stated	Not stated by the authors
Yun 2015	60 (14) years	69 (15) years	1.5 (1) months	1.5 (1) months	17 (57) female	8 (53) female	11 (37) left	4 (27) left	Not explicitly stated	Not explicitly stated	Not explicitly stated	Not explicitly stated	Not stated by the authors
BBT: Box and Block Test BI: Barthel Index CMMSA: Chedoke McMaster Stroke Assessment ESS: European Stroke Scale IQR: Interquartile Range JTT: Jebsen Taylor Hand Function Test LACI: lacunar stroke MRC: Medical Research Council NA: not applicable NIHSS: National Institute of Health Stroke Scale PACI: partial anterior circulation stroke SD: standard deviation TACI: total anterior circulation stroke UE‐FM: Upper Extremity Fugl‐Meyer Score

Table 2. Patient characteristics

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

Study ID	Type of intervention/ stimulation (polarity)	Electrode position and size	Reference electrode position	Treatment intensity		Base treatment	Dropouts	Adverse events	Source of information
Allman 2016	A‐tDCS	5 x 7–cm electrodes, encased in saline‐soaked sponges with the anode placed over ipsilesional primary motor cortex (5 cm lateral to Cz: C3) and the cathode over the contralateral supraorbital ridge		1 mA for 20 minutes	Base treatment plus 20 minutes of A‐tDCS or sham tDCS	Daily self‐administered Graded Repetitive Arm Supplementary Program (GRASP) training for 60 minutes over 9 days	2 (15%) in the EXP group due to organizational issues	Not described explicitely	Published
Allman 2016	Sham tDCS			1 mA for 10 seconds	Base treatment plus 20 minutes of A‐tDCS or sham tDCS		2 (15%) in the EXP group due to organizational issues	Not described explicitely	Published
Andrade 2017	A‐tDCS	6.4 x 2.5 cm anode over premotor cortex	On the supraorbital region in the contralateral hemisphere	0.7 mA (duration not described)	Base treatment plus unknown duration of A‐tDCS over PMC or M1 or sham tDCS	CIMT on a 3‐hour daily protocol of motor skills training for two weeks, supervised by a blinded physiotherapist (restriction of 90% of waking hours)	None	16 out of 60 patients reported mild side effects after stimulation (7 in the M1 group, 6 in PMC group, and 3 in the sham group): skin redness under the site of stimulation (5 in M1 group, 4 in PMC group, and 3 in sham group), mild headache (3 in M1 group and 2 in PMC group), and sleepiness (1 in PMC group). In all groups some subjects experienced multiple adverse effects.	Published
	A‐tDCS	6.4 x 2.5 cm anode over M1
	Sham tDCS	Not described
Ang 2012	Dual‐tDCS	Saline‐soaked sponge electrodes with the anode placed over M1 of the affected hemisphere and the cathode placed over M1 the unaffected hemisphere (size not stated)		1 mA for 20 minutes	20 minutes of dual‐tDCS or sham tDCS followed by 8 minutes of evaluation prior to base treatment	60 minutes of therapy using EEG‐based MI‐BCI with robotic feedback with the MIT‐Manus 5 times a week for 2 weeks	None	Unclear	Published
Ang 2012	Sham tDCS			1 mA for 30 seconds			None	Unclear	Published
Au‐Yeung 2014	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS, C‐tDCS and sham tDCS once in random order with at least 5 days wash‐out period	None	None	Unclear	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of both hemispheres		1 mA for 10 seconds
Bang 2015	Dual tDCS	Anodal sponge electrode of 35cm² was attached to the right posterior parietal cortex (P4) and accompanied by cathode tDCS of the second circuit was positioned over the left posterior parietal cortex (P3). Therefore, in the first tDCS circuit, the anode was placed over P4 and the cathode was placed over the left supraorbital area		1 mA for 20 minutes	Base treatment either with or without Dual tDCS	Mirror‐based feedback training	Not described	Unclear	Published
Bang 2015	Feedback training			NA	Base treatment either with or without Dual tDCS	Mirror‐based feedback training	Not described	Unclear	Published
Boggio 2007a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS, C‐tDCS or sham tDCS 4 days once a day	None	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Bolognini 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes; with the anode placed over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Base treatment + A‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks	CIMT up to 4 hours/day for 5 days a week for 2 consecutive weeks	7 (33%) due to frustration and tiredness during assessments (Bolognini 2013 [pers comm]); these participants have been excluded from analysis and presentation of results	None	Published and unpublished
Bolognini 2011	Sham tDCS			2 mA for 30 seconds				None	Published and unpublished
Cha 2014	A‐tDCS	Water‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base treatment + A‐tDCS for 20 minutes	Basic training for improving function of upper and lower extremities for 30 minutes per day, 5 times a week for four weeks	None	Unclear	Published
Cha 2014	PT	NA		NA	NA		None	Unclear	Published
Chang 2015	A‐tDCS	Saline‐soaked sponge surface electrodes with the 7 cm² anode over the tibialis anterior area of precentral gyrus of affected hemisphere	Saline‐soaked sponge surface electrodes with the 28 cm² cathode over the contralateral supraorbital area	2 mA for 10 minutes	Base treatment + either A‐tDCS or sham tDCS for 20 minutes	Conventional physical therapy	Not reported	Unclear	Published
Chang 2015	Sham tDCS			2 mA for 15 seconds	Base treatment + either A‐tDCS or sham tDCS for 20 minutes	Conventional physical therapy	Not reported	Unclear	Published
Chelette 2014	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1.4 mA for 20 minutes	Either A‐tDCS, C‐tDCS, dual tDCS or sham tDCS prior to base treatment	3 hours of intensive, task‐oriented UE motor training (a modified constraint‐based protocol)	Not reported	Unclear	Published
	C‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode contralesional supraorbital	35 cm² saline‐soaked sponge electrodes with the cathode over contralesional M
	Dual tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode over contralesional M1
	Sham tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1.4 mA for 30 seconds
Cho 2017	C‐tDCS	35 cm² wet sponge electrodes with the cathode over contralesional M1	35 cm² wet sponge electrodes with the anode contralesional supraorbital	2 mA for 20 minutes	Either base treatment plus C‐tDCS or base treatment only daily for 2 weeks	10 Hz and 90% rMT for 5 seconds with a 55‐second inter‐train interval, 90% of rMT intensity	None	No serious adverse events occured	Published
Cho 2017	rTMS	rTMS over ipsilesiona lM1 of the hand		1000 pulses over 20 min			None	No serious adverse events occured	Published
Cunningham 2015	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over ipsilesional PMC and SMA, identified with neuronavigation	35 cm² saline‐soaked sponge electrodes with the cathode contralesional supraorbital	1 mA for 30 minutes	A‐tDCS or sham tDCS during each rehabilitation session	CIMT for 30 minutes, 3 times per week for 5 weeks with supervision from a physical therapist. Intensive functional exercises were performed via a graded, regimented, feedback‐driven approach. Patient‐specific goals were emphasized. Patients were asked to restrain the non‐paretic upper limb by placing it in a mitt for 2 hours every weekday while performing home exercises. Exercise log was monitored at each session	None	Unclear	Published
Cunningham 2015	Sham tDCS			1 mA for 30 seconds	A‐tDCS or sham tDCS during each rehabilitation session		None	Unclear	Published
D'Agata 2016	rTMS + dual tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 20 minutes	1a. 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	rTMS@1Hz at 80% rMT for 15 min (900 stimuli) over the non‐lesioned M1 of the hand area	Not clearly stated	Unclear	Published
	Dual tDCS + mirror therapy			1.5 mA for 20 minutes		Mirror box training with the plegic hand (3 series of 25 repetitions of 6 different movements)
	Sham tDCS + mirror therapy			Not described
Danzl 2012	A‐tDCS	25 cm² saline‐soaked sponge electrodes with the anode over ipsilesional M1 of the leg and the anode over the contralateral supraorbital forehead		2 mA for 20 minutes	A‐tDCS or sham tDCS prior to base treatment	Robot‐assisted walking training (20 to 40 minutes) 3 times per week for 4 weeks	2 (20%): 1 in the A‐tDCS and 1 in the sham group due to knee pain and contractures	None	Published
Danzl 2012	Sham tDCS			2 mA for 75 seconds	A‐tDCS or sham tDCS prior to base treatment			None	Published
Di Lazzaro 2014a	Dual‐tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Dual‐tDCS or sham tDCS on 5 continuous days	None	None	Unclear	Published
Di Lazzaro 2014a	Sham tDCS			2 mA for 30 seconds	Dual‐tDCS or sham tDCS on 5 continuous days	None	None	Unclear	Published
Di Lazzaro 2014b	Dual‐tDCS	Anode over M1 of the lesioned hemisphere and cathode over M1 of the non‐lesioned hemisphere		2 mA for 40 minutes	Base treatment + dual‐tDCS or sham tDCS on 5 continuous days	CIMT for at least 90% of waking hours, including 1.5 hours per day arm training	None	Unclear	Published
Di Lazzaro 2014b	Sham tDCS			2 mA for 30 seconds			None	Unclear	Published
Fusco 2013a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 15 minutes	1 active tDCS (A‐tDCS, C‐tDCS, dual‐tDCS) and 1 sham tDCS session in 2 consecutive days	None	None	None	Published and unpublished
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 15 minutes
	Dual‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 15 minutes
	Sham tDCS	Not described by the authors
Fusco 2014	C‐tDCS	Saline‐soaked 35 cm² gel‐sponge electrodes with the cathode over M1 of the non‐lesioned hemisphere	Above the right shoulder	1.5 mA for 10 minutes	Each participant underwent C‐tDCS and sham tDCS on 5 consecutive days each week for 2 weeks prior to a rehabilitative session in randomised order	Patient‐tailored motor rehabilitation focusing on recovery of upper limb for 45 minutes twice a day	2 (14%); reasons not described by the authors	Unclear	Published
Fusco 2014	Sham tDCS		Above the right shoulder	Not described			1 (7%); emergency transfer to another hospital	Unclear	Published
Fregni 2005a	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent A‐tDCS, C‐tDCS and sham tDCS once, separated by at least 48 hours of rest	None	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over the M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Geroin 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 7 minutes	Base treatment + A‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks	50‐minute training sessions 5 days a week for 2 consecutive weeks, consisting of 20 minutes of robot‐assisted gait training and 30 minutes of lower limb strength and joint mobilisation training	None	None	Published
Geroin 2011	Sham tDCS		Over the contralateral supraorbital forehead	0 mA for 7 minutes			None	None	Published
Hamoudi 2018	A‐tDCS	25 cm² anode over ipsilesional M1 hotspot	25 cm² cathode over the contralateral supraorbital forehead	1.2 mA for 20 minutes	Either base treatment + A‐tDCS or sham tDCS or passive control group	Computerised grip strength training for 45 minutesper day for 5 days	No dropouts during intervention phase	1 (6) migraine, 1 (6) tingling sensation of the unaffected hand	Published
	Sham tDCS	25 cm² anode over ipsilesional M1 hotspot	25 cm² cathode over the contralateral supraorbital forehead	1.2 mA for 30 seconds				3 (17) mild headache, 1 (6) phosphene, 1 (6) abdominal pain, 1 (6) retching
	Passive control group	NA				No base treatment		None
Hathaiareerug 2019	Dual tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere	Saline‐soaked 35 cm² sponge electrodes with the cathode over M1 of the non‐lesioned hemisphere	2 mA for 20 minutes	Base treatment + either dual tDCS or electro‐acupuncture once a week for 3 weeks	Intensive physical therapy and occupational therapy performed in hourly sessions 3 times per week for 3 weeks	1 (11) dropped out during follow‐up	Unclear	Published
Hathaiareerug 2019	Electro‐acupuncture	NA					None	Unclear	Published
Hesse 2011	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS 5 days a week for 6 consecutive weeks	20 minutes of robot‐assisted arm training 5 days a week for 6 consecutive weeks	11 (11%); 7 dropouts in the EXP‐groups: 1 (14%) during intervention period due to pneumonia and 6 (86%) until 3 months of follow‐up (2 deaths due to myocardial infarction and stent surgery, 3 due to being unavailable and 1 due to refusal of further enrolment); 4 dropouts in the CTL group: 3 (75%) due to being not available and 1 (25%) due to refusal of further enrolment	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the non‐lesioned hemisphere		2 mA for 20 minutes
	Sham tDCS	As in the A‐tDCS or the C‐tDCS group (changing consecutively)		0 mA for 20 minutes
Ilić 2016	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 hand area of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + either A‐tDCS or sham tDCS prior	Intensive task oriented training, delivered by OT and consisting of strength training, ROM exercises, manipulation exercises, pinch grip, grasp, release and simulating ADL	1 dropout in the sham group (reason not stated)	None	Published
Ilić 2016	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 60 seconds	Base treatment + either A‐tDCS or sham tDCS prior		1 dropout in the sham group (reason not stated)	None	Published
Jo 2008a	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over DLPFC of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 30 minutes	A‐tDCS once and sham tDCS once or vice versa, separated by at least 48 hours of resting period	None	None	6 Quote: "Transient aching or burning sensations were reported in six cases, and transient skin redness at the electrode contact site was reported in three cases."	Published
Jo 2008a	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 10 seconds		None	None		Published
Kang 2008b	A‐tDCS	25 cm² electrodes over the left DLPFC	Over the contralateral supraorbital forehead	2 mA for 20 minutes	A‐tDCS and sham tDCS or vice versa, separated by at least 48 hours of resting period	None	Not described	Unclear	Published
Kang 2008b	Sham tDCS	25 cm² electrodes over the left DLPFC	Over the contralateral supraorbital forehead	2 mA for 1 minute		None	Not described	Unclear	Published
Khedr 2013	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 25 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS for 6 consecutive days after	Rehabilitation program within 1 hour after each tDCS session, starting with passive movement and range of motion exercise up to active resistive exercise	None	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes, cathode over M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 25 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 2 minutes
Kim 2009	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent A‐tDCS and sham tDCS, separated by at least 24 hours of rest	None	None	None	Published and unpublished
Kim 2009	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds		None	None	None	Published and unpublished
Kim 2010	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the lesioned hemisphere (as confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS, C‐tDCS or sham tDCS 5 days a week for 2 consecutive weeks at the beginning of each therapy session	Occupational therapy according to a standardised protocol aimed at improving paretic hand function for 30 minutes 5 days a week for 2 consecutive weeks	2 of 20; 1 participant discontinued treatment because of dizziness and another because of headache (authors did not state corresponding groups)	Two	Published
	C‐tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the non‐lesioned hemisphere (confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 20 minutes
	Sham tDCS	Saline‐soaked 25 cm² sponge electrodes over M1 of the lesioned hemisphere (confirmed by MEP)	Over the contralateral supraorbital forehead	2 mA for 1 minutes
Kim 2016	A‐tDCS	Saline‐soaked 24 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base treatment + either A‐tDCS or sham tDCS	Traditional occupational therapy treatment	Not described	Unclear	Published
Kim 2016	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds	Base treatment + either A‐tDCS or sham tDCS	Traditional occupational therapy treatment	Not described	Unclear	Published
Ko 2008a	A‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes over right (lesioned) PPC	Over the contralateral supraorbital forehead	2 mA for 20 minutes	A‐tDCS once and sham tDCS once or vice versa, separated by at least 48 hours of resting period	None	Not described	None	Published
Ko 2008a	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 10 seconds		None	Not described	None	Published
Koo 2018	A‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes with the anode over S1 of the affected hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	A‐tDCS or sham tDCS during 10 stimulation sessions over 10 days	None	Not described	None	Published
Koo 2018	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 20 seconds		None	Not described	None	Published
Klomjai 2018	Dual tDCS	Saline‐soaked sponge‐pad electrodes with 35cm² surface and electroconductive gel	Anodal tDCS over the M1 of the affected hemisphere and cathodal tDCS over the M1 of the unaffected hemisphere	2 mA for 20 minutes	Dual tDCS once prior to base treatment and sham tDCS once prior to base treatment or vice versa, separated by at least 7 days of resting period	Dose‐matched physical therapy for 60 minutes under expert supervision, aiming at improving strength in the lower extrimity	Not described	Unclear	Published
Klomjai 2018	Sham tDCS			2 mA for 120 seconds			Not described	Unclear	Published
Lee 2014	C‐tDCS	Saline‐soaked 25 cm² surface sponge electrodes over hand area of M1 of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	20 minutes per day, 5 times per week for 3 weeks	Occupational therapy for 30 minutes per day, 5 times per week for 3 weeks	3 of 42 (7%); 2 medical problems; 1 refused to participate	No major adverse events	Published
	C‐tDCS		Over the contralateral supraorbital forehead	2 mA for 20 minutes	20 minutes per day, 5 times per week for 3 weeks	Virtual reality therapy for 30 minutes per day, 5 times per week for 3 weeks	3 of 42 (7%); 2 medical problems; 1 refused to participate
	Virtual reality	NA		NA	NA	Virtual reality only for 30 minutes per day, 5 times per week for 3 weeks	2 of 22 (9%); 1 refused to participate; 1 early discharge
Lindenberg 2010	Dual‐tDCS	Saline‐soaked 16.3 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1.5 mA for 30 minutes	Base treatment + dual‐tDCS or sham tDCS at 5 consecutive sessions on 5 consecutive days	Physical and occupational therapy sessions at 5 consecutive sessions on 5 consecutive days for 60 minutes, including functional motor tasks	None	None	Published
Lindenberg 2010	Sham tDCS			1.5 mA for 30 seconds			None	None	Published
Mahmoudi 2011	A‐tDCS1	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	Over the contralateral orbit	1 mA for 20 minutes	Each participant underwent A‐tDCS1, A‐tDCS2, C‐tDCS, dual‐tDCS and sham tDCS once with a wash‐out period of at least 96 hours	None	None	Unclear	Published
	A‐tDCS2	Saline‐soaked 35 cm² sponge electrodes, anode over M1 of the lesioned hemisphere	On the contralateral deltoid muscle	1 mA for 20 minutes
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes, cathode over M1 of the non‐lesioned hemisphere	Over M1 of the lesioned hemisphere	1 mA for 20 minutes
	Dual‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode over M1 of the lesioned hemisphere and the cathode over M1 of the non‐lesioned hemisphere		1 mA for 20 minutes
	Sham tDCS	Not described by the authors		1 mA for 30 seconds
Manji 2018	A‐tDCS	25 cm² saline‐soaked sponge electrodes with the anode over the SMA of the lesioned hemisphere	Over the inion	1 mA for 20 minutes	Each participant underwent A‐tDCS + base treatment or sham tDCS + base treatment in a random order, each once a day for a week	Body‐weight‐supported treadmill training (BWSTT) with 20% of body weight support for 20 minutes once a day for a week	None	Unclear	Published
Manji 2018	Sham tDCS		Over the inion	1 mA for 30 seconds			None	Unclear	Published
Mazzoleni 2019	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + 20 minutes either A‐tDCS or sham tDCS 5 times a week for 6 weeks	Robotic wrist‐training with appr. 1000 repetitions per session. The robot provided assistance, if necessary	1 out of 20 (5) in the CTL group dropped out due to robot failure	None	Published
Mazzoleni 2019	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 5 seconds				None	Published
Mortensen 2016	A‐tDCS	35 cm² saline‐soaked sponge electrodes with the anode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1.5 mA for 20 minutes	Base treatment + 20 minutes either A‐tDCS or sham tDCS on 5 consecutive days	30 minutes of home‐based occupational therapy, aiming at activities and functional tasks	1 out of 8 (13) in the CTL group dropped out during worsening of hand function	There were 6 moderate or severe adverse events (3 in the EXP group and 3 in the CTL group, respectively)	Published
Mortensen 2016	Sham tDCS		Over the contralateral supraorbital forehead	1.5 mA for 30 seconds					Published
Nair 2011	C‐tDCS	Saline‐soaked sponge electrodes with the cathode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 30 minutes	Base‐treatment + C‐tDCS or sham tDCS for 5 consecutive daily sessions, each at the beginning of the base treatment sessions	Occupational therapy (PNF; shoulder abduction, external rotation, elbow extension, forearm pronation) for 5 consecutive daily sessions (60 minutes each)	None	None	Published
Nair 2011	Sham tDCS	Not described by the authors		For 30 minutes			None	None	Published
Nicolo 2017	C‐tDCS	35 cm² saline‐soaked sponge electrodes with the cathode over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 25 minutes	Base therapy + brain stimulation 3 times per week for 3 weeks during upper extremity functional motor training sessions	30 minutes of active functional motor practice, consisting of patient‐tailored exercises	None	None	Published
	Sham (tDCS, cTBS)		Over the contralateral supraorbital forehead	1 mA for 30 seconds
	cTBS	Over non‐lesioned M1	N/A	267 bursts, each consisting of 3 pulses at 30 Hz, repeated at inter‐burst intervals of 167 ms); 2 stimulation trains of 30 seconds (separated by 15 minutes)
Park 2013	A‐tDCS	Sponge electrodes with the anode positioned over the bilateral prefrontal cortex	At the non‐dominant arm	2 mA for 30 minutes	Base‐treatment + A‐tDCS or sham tDCS for 5 days a week for approximately 18 days	Computer‐assisted cognitive rehabilitation (CACR) with the ComCog program (15 minute attention and 15 minute memory training)	Unclear	None	Published
Park 2013	Sham tDCS		At the non‐dominant arm	2 mA for 30 seconds			Unclear	None	Published
Park 2015	A‐tDCS	Anode over Cz area of the left parietal lobe [sic]	Over the contralateral supraorbital forehead	2 mA for 15 minutes	Physiotherapy + either A‐tDCS or sham tDCS for 3 days a week during 4 weeks	Task related training for weight support ability improvement and stepping strategy Quote: "(1) lifting and maintaining the lower extremity; (2) lifting the heels; (3) lifting the lower extremity over the footstool followed by lowering; (4) lifting the lower extremity and lowering in onto a footstool; (5) walking back and forth over a 3‐m distance to a chair; and (6) going back and forth at a constant pace over 10‐m distance. The tasks were conducted one‐on‐one with a physical therapist."	None	None	Published
	Sham tDCS	Anode over Cz area of the left parietal lobe [sic]	Over the contralateral supraorbital forehead	Not described
	PT	N/A
Picelli 2015	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Base treatment + A‐tDCS with either cathodal transcutaneous spinal direct current stimulation (tsDCS) or with sham tsDCS	Robot‐assisted gait training on a G‐EO for 20 minutes, 5 times per week for 2 weeks	None	None	Published
Picelli 2015	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 2 minutes	Base treatment + sham tDCS and cathodal tsDCS		None	None	Published
Qu 2009	C‐tDCS	Saline‐soaked 18 cm² sponge electrodes over primary sensorimotor cortex of the lesioned hemisphere	Unclear	0.5 mA for 20 minutes, once a day for 5 consecutive days for 4 weeks		NA	None	None	Published
Qu 2009	PT	NA		Physical therapy according to the Bobath, Brunnstrom and Rood approaches for 40 minutes twice a day for 5 consecutive days for 4 weeks		NA	None	None	Published
Qu 2017	C‐tDCS	Not described	Not described	1.0 mA cathodal tDCS for 2 weeks, once a day, once for 20 minutes, 5 days a week	Not described	Not described	Not described	Unclear	Published
	C‐tDCS	Not described	Not described	2.0 mA cathodal DCS for two weeks, once a day, once for 20 minutes, 5 days a week
	Sham tDCS	Not described	Not described	Sham tDCS for 2 weeks, once a day, once for 20 minutes, 5 days a week
Rabadi 2017	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over PMC of the non‐lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 30 minutes	Base therapy + C‐tDCS or sham tDCS 30 minutes a day on 5 consecutive days for 2 weeks	4 hours of standard occupational and physical therapy	There were no drop‐outs during intervention phase. Until 3 months follow‐up 3 dropouts (38) occured in the EXP group and 1 (13) in the CTL group. Reasons were not stated by the authors.	None	Published
Rabadi 2017	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 30 seconds		4 hours of standard occupational and physical therapy		None	Published
Rocha 2016	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	1 mA for 13 minutes	A‐tDCS, C‐tDCS or sham tDCS 3 times a week for 4 consecutive weeks prior to base therapy	mCIMT (total immobilisation of the non‐paretic upper limb and intensive training of the paretic upper limb) for 6 continuous hours each day over 4 weeks plus 1 hour gross and fine motor activities training per day	There were 2 drop‐outs in each group (28%) due to unknown reasons	None	Published
	C‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the non‐lesioned hemisphere		1 mA for 9 minutes
	Sham tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere		1 mA for 30 seconds
Rossi 2013	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes over M1 of the lesioned hemisphere	Over the contralateral supraorbital forehead	2 mA for 20 minutes	Once a day for 5 consecutive days	Not described by the authors	None	None	Published
Rossi 2013	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds	Once a day for 5 consecutive days	Not described by the authors	None	None	Published
Saeys 2015	A‐tDCS	Over the motor cortex (on C4 or C3 of the 10–20 EEG system)	over the intact hemisphere	1.5 mA for 20 minutes	16 x 20‐minute sessions (4 times a week for 4 weeks)	Both groups received multidisciplinary regular physical and occupational therapy mainly focused on the neurodevelopmental treatment concept (1 hour daily)	None	None	Published
Saeys 2015	Sham tDCS	Over the motor cortex (on C4 or C3 of the 10–20 EEG system)	over the intact hemisphere	Stimulation turned off after 30 seconds	16 x 20‐minute sessions (4 times a week for 4 weeks)		None	None	Published
Salazar 2019	Dual‐tDCS	Over the the M1 area (C3 and C4 of the EEG system) Anode electrodes were positioned over the ipsilesional M1 and cathodes over the contralesional M1		Both groups received 10 sessions of concurrent tDCS and FES or sham tDCS and FES during 30 minutes, 5 times a week for 2 weeks		Before each stimulation session, participants had scapular, shoulder, elbow, wrist and finger passive mobilization for approximately 10 min	None	None	Published
Salazar 2019	Dual sham tDCS						None	None	Published
Sattler 2015	A‐tDCS	Over the M1 area (at the hotspot of the extensor carpi radialis muscle	Cathode over the contralesional supraorbital region	1.2 mA anodal tDCS	5 consecutive daily sessions for 13 minutes each	rPNS (5 Hz) was delivered to the radial nerve through bipolar round brass electrodes placed in the spiral grove of the paretic side and was applied at the same time as the real or sham tDCS stimulation. It was applied similarly in both active and sham conditions for 13 minutes. The intensity of was adjusted to be below the threshold for direct M response (0.7 x MT).	None	None	Published
Sattler 2015	Sham tDCS		Cathode over the contralesional supraorbital region	Stimulation (same site and same parameters) was turned off after 60 seconds of stimulation	5 consecutive daily sessions for 13 minutes each		None	None	Published
Seo 2017	A‐tDCS	Over the presumed leg area of the lesioned hemisphere, just lateral to the Cz position according to the 10–20 system	Cathode on the forehead above the contralateral orbit	2 mA for 20 minutes	20 minutes of tDCS for every weekday during 2 weeks (total 10 sessions)	RAGT for 45 minutes after tDCS	None at first follow‐up	None	Published
Seo 2017	Sham tDCS		Cathode on the forehead above the contralateral orbit	Stimulation intensity was slowly tapered down from 2 to 0 mA over several seconds after initial minute		RAGT for 45 minutes after tDCS	None at first follow‐up	None	Published
Shaheiwola 2018	A‐tDCS	Primary motor cortex using (abductor pollicis brevis) hot spot)	Cathode on the contralateral symmetrical area of non‐lesioned hemisphere	2.0 mA, time of ramp‐up: 10 seconds, time of ramp‐down: 10 seconds, 20 minutes	5 sessions per week on workdays and a total of 20 sessions during the 4 weeks	60 minutes FES each day	None	None	Published
Shaheiwola 2018	Sham tDCS					60 minutes FES each day	None	None	Published
Sik 2015	A‐tDCS g	Anodal tDCS over C3‐C4 area of the affected hemisphere	Opposite supraorbital region	2 mA, 20 mintes in patients with anodal stimulation ‐‐‐ 2 mA, 40 minutes in the bihemispheric‐treated patients (20 minutes anodal tDCS to the lesional hemisphere/20 minutes cathodal tDCS to the non‐lesional hemisphere)		tDCS application was started simultaneously with occupational therapy (15 sessions for 3 weeks) Physiotherapy and occupational therapy, (2 hours, including range of motion exercises, strengthening exercises, outreach activities)	5 (2 in A‐tDCS group and 2 in bihemispheric group and 1 in sham group)	None	Published
	Dual‐tDCS	Dual‐TDCs active electrode to the C3‐C4 area of the unaffected hemisphere in addition to its anodal application
	Sham tDCS	Sham: electrodes were placed as in the anodal group
Sohn 2013	A‐tDCS	25 cm² sponge electrodes over M1 of the affected hemisphere	Not described	2 mA for 10 minutes	A‐tDCS or sham tDCS once	None	Unclear	Unclear	Published
Sohn 2013	Sham tDCS	25 cm² sponge electrodes over M1 of the affected hemisphere	Not described	2 mA for 20 seconds	A‐tDCS or sham tDCS once	None	Unclear	Unclear	Published
Straudi 2016	Dual‐tDCS	Anode was placed on the M1 of the affected hemisphere. Electrodes were located at C3 and C4 according to the 10/20 international EEG system	Cathode on the contralateral M1 area	1 mA for 30 minutes, during RAT		Upper Extremity Robot‐Assisted Training	None	No severe adverse events (10 out of 23 reported mild adverse events)	Published
Straudi 2016	Sham tDCS		Cathode on the contralateral M1 area	Current was delivered for only 30 seconds and then the current was discontinued, but the tDCS apparatus was left in place for the same time as active tDCS (30 minutes)		Upper Extremity Robot‐Assisted Training	None		Published
Sunwoo 2013a	Dual‐tDCS	Saline‐soaked 25 cm² sponge electrodes over the right posterior parietal cortex (PPC) plus cathodal tDCS over the left PPC	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Each participant underwent dual‐tDCS, A‐tDCS and sham tDCS once with a wash‐out period of at least 24 hours	None	None	3 (30%) suffered from mild headache after dual‐tDCS, which disappeared spontaneously	Published
	A‐tDCS	Saline‐soaked 25 cm² sponge electrodes over the right PPC plus sham tDCS over the left PPC		1 mA for 20 minutes
	Sham tDCS	Saline‐soaked 25 cm² sponge electrodes over the right PPC plus sham tDCS over the left PPC		1 mA for 10 seconds
Tahtis 2012	Dual‐tDCS	Saline‐soaked 25 cm² electrodes with the anode placed over the leg area of the lesioned hemisphere and the cathode placed over leg area of the non‐lesioned hemisphere	Not described	2 mA for 15 minutes	Dual‐tDCS or sham tDCS once	None	Unclear	None	Published
Tahtis 2012	Sham tDCS		Not described	2 mA for < 30 seconds	Dual‐tDCS or sham tDCS once	None	Unclear	None	Published
Tedesco Triccas 2015b	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	1 mA for 20 minutes	Base therapy plus tDCS or sham tDCS for 18 sessions during 8 weeks (approximately 2 to 3 sessions per week)	Robotic arm training with the ArmeoSpring device (60 minutes per session) for 18 sessions during 8 weeks (approximately 2 to 3 sessions per week)	1 out of 12 (8%) in the A‐tDCS group due to a skin reaction after receiving four sessions of A‐tDCS	6 out of 12 (50%) in the A‐tDCS group reported adverse events such as pain, burning or headache after receiving A‐tDCS	Published/unpublished
Tedesco Triccas 2015b	Sham tDCS		Over the contralateral supraorbital forehead	1 mA for 20 seconds					Published/unpublished
Utarapichat 2018	A‐tDCS	Saline‐soaked 10 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	2 mA for 10 minutes	Not described	Not described	None	Unclear	Published
Utarapichat 2018	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds	Not described	Not described	None	Unclear	Published
Viana 2014	A‐tDCS	Saline‐soaked 35 cm² sponge electrodes with the anode placed over M1 of the affected hemisphere	Over the contralateral supraorbital forehead	2 mA for 13 minutes	Base therapy + A‐tDCS or sham tDCS 3 times a week for 5 weeks	Virtual reality training using Nintendo Wii (Games used: Wii Sports resort, Wii Play Motion, Let's Tap) aiming at movements of shoulder, elbow, wrist, hand and fingers; each game was played for 15 minutes (total time per training session: 60 minutes); passive stretching exercises were performed before and after each training session	None	None	Published
Viana 2014	Sham tDCS		Over the contralateral supraorbital forehead	2 mA for 30 seconds			None	None	Published
Wang 2014	Dual‐tDCS	35 cm² electrodes with the anode placed over M1 of the affected hemisphere	Over contralateral M1	1 mA for 20 minutes	Dual‐tDCS or sham‐tDCS once	Placebo methylphenidate 1 hour prior to stimulation	Unclear	No major adverse events; 3 participants (50%) from the dual‐tDCS group reported mild tingling sensation with tDCS stimulation	Published
	Dual‐tDCS			1 mA for 20 minutes		20 mg MP 1 hour prior to stimulation
	Sham‐tDCS			1 mA for 10 seconds		20 mg MP 1 hour prior to stimulation
Wong 2015	A‐tDCS	Over the hand area of primary motor cortex of the affected hemisphere	Cathodal electrode was placed over the contralateral supraorbital area	1 mA tDCS for 20 minutes	Not described	Not described	Not described	Unclear	Published
Wong 2015	5 consecutive sessions of intensive physiotherapy upper limb training			1 mA tDCS for 20 minutes	Not described	Not described	Not described	Unclear	Published
Wu 2013a	C‐tDCS	Saline‐soaked 24.75 cm² sponge electrodes over primary sensorimotor cortex of the lesioned hemisphere	Over the shoulder on the unaffected side	1.2 mA for 20 minutes	Once daily 5 days a week for 4 weeks	Quote: "Both groups received a conventional physical therapy program for 30 minutes twice daily, including maintaining good limb position, chronic stretching via casting or splinting, physical modalities and techniques, and movement training"	None	None	Published
Wu 2013a	Sham tDCS		Over the shoulder on the unaffected side	1.2 mA for 30 seconds	Once daily 5 days a week for 4 weeks		None	None	Published
Yi 2016	A‐tDCS	Over the right PPC (5 cm x 5 cm)	Over Cz	2 mA for 30 minutes	5 sessions per week for 3 weeks	Conventional physical therapy throughout the duration of the 3 weeks	2 out 32 (6%)	None	Published
	C‐tDCS	Over the left PPC		2 mA for 30 minutes
	Sham tDCS	Sham tDCS was performed in the same way as for anodal group		2 mA for 30 minutes
	Sham tDCS	Sham tDCS was performed in the same way as for anodal group		Stimulator was turned off after 30 seconds
Yun 2015	A‐tDCS left	At T3 for the left‐group and	Unclear	2 mA for 30 minutes	5 times a week for 3 weeks	Not described	None	None	Published
	A‐tDCS right	At T3 for the left‐group and	Unclear
	Sham tDCS	At T4 for the right‐group	Unclear
	Sham tDCS	Using the same method as for the left‐ group,	Unclear
A‐tDCS: anodal direct current stimulation C‐tDCS: cathodal direct current stimulation CIMT: constraint‐induced movement therapy cTBS: Continuous Theta Burst Stimulation Dual‐tDCS: A‐tDCS and C‐tDCS simultaneously EEG: electroencephalography FES: Functional electrical stimulation M1: primary Motor Cortex MEP: motor‐evoked potentials MI‐BCI: motor imagery brain‐computer interface MP: methylphenidate NA: not applicable PNF: proprioceptive neuromuscular facilitation PPC: posterior parietal cortex PT: physical therapy RAGT: robotic‐assisted gait training rPNS: Repetitive electrical stimulation SD: standard deviation tDCS: transcranial direct current stimulation tsDCS: transcutaneous spinal direct current stimulation

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

Table 4. Sensitivity analyses for comparison 1.2: primary outcome of ADL performance at the end of follow‐up at least 3 months after the end of the intervention period

Sensitivity analysis	Studies included in analysis	Effect estimate
All studies with proper allocation concealment for primary outcome absolute values	Hesse 2011; Khedr 2013; Kim 2010; Tedesco Triccas 2015b	(SMD 0.30, 95% CI ‐0.15 to 0.75; participants = 199; studies = 4; I² = 51%; inverse variance method with random‐effects model)
All studies with proper allocation concealment for primary outcome change scores	Rabadi 2017	(SMD 0.19, 95% CI ‐0.27 to 0.64; participants = 16; studies = 1; I² = 0%; inverse variance method with random‐effects model)
All studies with proper blinding of outcome assessor for primary outcome	Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Kim 2010; Rossi 2013; Tedesco Triccas 2015b	(SMD 0.31, 95% CI 0.01 to 0.62; participants = 269; studies = 6; I² = 27%; inverse variance method with random‐effects model)
All studies with intention‐to‐treat analysis	Di Lazzaro 2014b; Hesse 2011; Khedr 2013; Rossi 2013	(SMD 0.38, 95% CI 0.05 to 0.70; participants = 205; studies = 4; I² = 16%; inverse variance method with random‐effects model)
CI: confidence interval SMD: standardised mean difference

Table 4. Sensitivity analyses for comparison 1.2: primary outcome of ADL performance at the end of follow‐up at least 3 months after the end of the intervention period

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

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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

1.10 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period Show forest plot	47	1330	Risk Ratio (M‐H, Random, 95% CI)	1.25 [0.74, 2.13]

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

Comparison 2. tDCS versus any type of active control intervention

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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	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 Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

2.7 Secondary outcome measure: dropouts, adverse events and deaths during the intervention period Show forest plot	7	209	Risk Ratio (M‐H, Random, 95% CI)	1.76 [0.43, 7.17]

Comparison 2. tDCS versus any type of active control intervention

Comparison 3. Subgroup analyses for primary outcome measure: ADL at the end of the intervention period

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 Show forest plot	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) Show forest plot	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) Show forest plot	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]

Comparison 3. Subgroup analyses for primary outcome measure: ADL at the end of the intervention period