Transcranial Direct Current Stimulation in Pediatric Motor Disorders: A Systematic Review and Meta-analysis

Abstract

Objective:

To systematically examine the safety and effectiveness of transcranial direct current stimulation (tDCS) interventions in pediatric motor disorders.

Data Sources:

PubMed, EMBASE, Cochrane, CINAHL, Web of Science, and ProQuest databases were searched from inception to August 2018.

Study Selection:

tDCS randomized controlled trials (RCTs), observational studies, conference proceedings and dissertations in pediatric motor disorders were included. Two authors independently screened articles based on predefined inclusion criteria.

Data Extraction:

Data related to participant demographics, intervention, and outcomes were extracted by two authors. Quality assessment was independently performed by two authors.

Data Synthesis:

Twenty-three studies involving a total of 391 participants were included. There was no difference in drop-out rates between active (1/144) and sham (1/144) tDCS groups, risk difference 0.0, 95% CI [−.05, .04]. Across studies, the most common adverse effects in the active group were tingling (17.2%), discomfort (8.02%), itching (6.79%), and skin redness (4%). Across 3 studies in children with cerebral palsy, tDCS significantly improved gait velocity (MD = .23; 95% CI [0.13, 0.34], p < .0005), stride length (MD = 0.10; 95% CI [0.05, 0.15], p< .0005), and cadence (MD = 15.7; 95% CI [9.72, 21.68], p< .0005). Mixed effects were found on balance, upper-extremity function, and overflow movements in dystonia.

Conclusion:

Based on the studies reviewed, tDCS is a safe technique in pediatric motor disorders and may improve some gait measures and involuntary movements. Research to date in pediatric motor disorders shows limited effectiveness in improving balance and upper-extremity function. tDCS may serve as a potential adjunct to pediatric rehabilitation; to better understand if tDCS is beneficial for pediatric motor disorders, more well-designed RCTs are needed.

Pediatric motor disorders encompass a heterogeneous group of movement disorders that vary in symptoms based on their etiology and age of onset [1]. The most common cause of pediatric motor disorders is cerebral palsy (CP), affecting 2.5 in 1000 newborns in the United States [2]. Neurorehabilitation of pediatric motor disorders primarily employs motor learning approaches to enhance motor behaviors and encourage performance of functional tasks [3,4].

Recently, tDCS is gaining popularity in neurorehabilitation because of its potential to alter cortical excitability and enhance motor performance in adults with physical impairments [5–7]. In adult neurorehabilitation, tDCS has been successfully used as a priming intervention, or add-on approach, to increase the responsiveness of cortical regions to accompanying therapy [8]. However, the effect of tDCS in pediatric neurorehabilitation as a priming or standalone intervention has yet to be systematically examined.

tDCS appears more favorable than other noninvasive brain stimulation methods due to its easy application, low cost, and portability [9]. Generally, a tDCS apparatus is comprised of a low-level current stimulator and two surface electrodes that deliver direct current (0.5–2 mA) to the scalp via saline-soaked sponges [5,9] and can modulate brain functioning to elicit neuroplasticity effects that may translate into improved motor functioning [3]. Importantly, tDCS can produce long-lasting after-effects beyond the period of stimulation; typically the duration of these after-effects is dependent on the duration of stimulation [5,6]. One potential mechanism of tDCS’ effect is alteration of neuronal resting membrane potential [3]. Current is transferred from the positive electrode (anode) to the negative electrode (cathode), and either electrode can be placed over the presumed cortical region of interest, resulting in either anodal or cathodal tDCS polarity for the targeted region [6]. The position of the reference electrode varies by purpose/protocol [3]. tDCS can also be delivered through dual montage, consisting of simultaneous placement of the anode and cathode over ipsilesional and contralesional cortex or vice versa [7]. Although it is presumed that anodal tDCS increases cortical excitability while cathodal tDCS decreases excitability; in reality, these two electrodes can have mixed effects based on their position and orientation and the state of the underlying neurons [10].

In adults with neuromotor disorders, tDCS has been shown to be safe and moderately effective in improving performance and participation [11,12]. Although limited data suggest that tDCS is safe for children with developmental, motor and neuropsychiatric disorders [13,14], recent narrative reviews in children have not adequately reported on the adverse effects of tDCS in patients with pediatric motor disorders [13–15]; the increasing use of tDCS in pediatric motor disorders demands systematic evaluation of its safety and tolerability in this population [4]. Of note, tDCS parameters (e.g., current intensity) can affect the frequency and severity of adverse effects (e.g., itching/tingling) [5,16]. While systematic reviews have investigated the effectiveness of tDCS in adults with neuromotor disorders [12,17], no systematic review has comprehensively synthesized the findings of randomized controlled trials (RCTs) and observational studies in pediatric motor disorders. Given the anatomical, neurophysiological, and pathophysiological differences in children and adults, a systematic review examining tDCS safety and effectiveness in pediatric motor disorders is needed.

The primary objective of this systematic review and meta-analysis was to empirically and quantitatively analyze evidence related to the safety and effectiveness of tDCS in pediatric motor disorders. The secondary objective was to identify tDCS parameter guidelines for future tDCS intervention studies. The research questions were:

1. What proportion of children with motor disorders receiving tDCS experience adverse effects?

2. What are the effects of tDCS on motor outcomes in pediatric motor disorders?

3. What are the methodological considerations for tDCS use in pediatric motor disorders?

Material and methods

Study selection

A systematic review and meta-analysis were conducted following Cochrane recommendations and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [18]. PubMed, EMBASE, Cochrane, CINAHL, Web of Science, and ProQuest were searched. To reduce publication bias, we included unpublished and ongoing studies from conference abstracts, dissertations, and Letters to the Editor (grey-literature) [19]. The search was conducted based on the Patient, Intervention, Comparison, and Outcome framework [20], and search terms included keywords related to children and motor disorders (Participant), tDCS (Intervention) and motor function (Outcome) (Appendix S.1). For the purposes of this review, motor disorders were defined by the presence of atypical, involuntary, or rigid movements; postural instability; muscle tone abnormalities; or delayed motor development [2,21]. The inclusion criteria were 1) English language and 2) original articles reporting tDCS parameters, safety, and/or utilizing at least one motor outcome measure, ranging from body structure to participation, in children (≤21 years of age) with motor disorders. The review protocol was registered with PROSPERO, the international prospective register of systematic reviews.

The literature search (RCTs, observational studies, and conference abstracts) was first conducted on October 20^th, 2017 and December 5^th, 2017 for dissertations and last updated on August 29^th, 2018 with no filter applied to the start date of the search. The systematic review software Covidence was used. Two authors independently reviewed titles and abstracts and ranked the studies as relevant, possibly relevant, or irrelevant according to the inclusion/exclusion criteria. Studies ranked irrelevant by both authors were eliminated. The full-text of studies deemed relevant or possibly relevant was independently assessed by two authors for eligibility; discrepancies were resolved through discussion with the senior authors. Abstracts from conference proceedings that met the inclusion criteria were analyzed separately for narrative review [19].

Data extraction

The following data were independently extracted by two authors: 1) Metadata (authorship, publication date, region); 2) participant characteristics (sample size, age, gender, diagnosis); 3) methods (study design, any outcome measure evaluating motor function); 4) tDCS dosage parameters (current intensity, duration, current type, electrode montage, electrode placement; mode of targeting the primary motor cortex, e.g., using EEG 10–20 coordinates, number of sessions); and 5) report of adverse effects (including dropouts and report of symptoms such as tingling).

Quality assessment

The methodological quality of RCTs and observational studies was assessed using the Cochrane Risk of Bias tool [22]. The following biases were assessed: 1) sequence generation; 2) allocation concealment; 3) blinding of participants and personnel; 4) blinding of outcome assessor; 5) incomplete outcome data and 6) selective outcome reporting.

Outcomes

To assess safety and tolerability of tDCS, dropout rates and presence of adverse effects were quantified. To examine the effectiveness of tDCS, motor outcomes were considered post-tDCS and at follow-up when available. Either post-intervention means and standard deviations or change scores for both active and sham-tDCS groups were required for meta-analysis of studies. Effectiveness was measured as the difference in post-intervention scores or change scores between active versus sham-tDCS groups. A meta-analysis was conducted when at least 2 studies provided the necessary outcome data points utilizing the same motor outcome measure. Authors of RCTs were contacted to obtain missing information. A narrative summary was provided for other motor outcomes not included in the meta-analysis. Due to the subjective nature of the Canadian Occupational Performance Measure (COPM), 2 studies using this outcome were also summarized narratively.

Statistical analysis

Forest plots were created using Review Manager 5.3 employing a random-effects model according to the inverse-variance method [23,24]. The random-effects model was chosen because it allows for treatment effects to vary across studies [25]. For the meta-analysis of dropout rates, risk difference was calculated (RD, difference between the risk of the event in the active versus sham groups), as it can be calculated even when there are no events in either group. A funnel plot to assess publication bias was created for studies reporting dropouts. The mean difference (absolute difference between mean values in two groups) was used for meta-analysis of motor outcomes. There were fewer than 10 studies for all motor outcome measures, so funnel plots were not created [26]. All results are presented with 95% confidence intervals. Alpha was set at p<.05. Results with .05 < p < .10 were considered trending towards significance.

Results

Description of studies

Of the initial 1779 references identified, 23 full-texts, 10 abstracts, and 1 letter to the editor were included (Figure 1). Of the 23 full-texts, 12 were RCTs, 7 were observational studies, 2 were methodological studies, and 2 studies investigated biomarkers for tDCS effectiveness. Seventeen studies included a total of 373 participants with CP; 3 studies included a total of 33 participants with dystonia; 3 were single case studies (one with involuntary movements, one with delayed neuromotor development, one described as minimally conscious with spastic quadriplegia due to anoxic brain injury). Mean study sample size was 16 participants (range 1–56). Participants’ ages ranged from 4–21 years. Fourteen studies used anodal stimulation, 9 studies used cathodal stimulation, 2 studies used a bi-hemispheric montage,1 study used anodal and cathodal stimulation, 1 study used both cathodal stimulation and a bi-hemispheric montage, and 1 study investigating electrode placement did not deliver tDCS. Of the 19 intervention studies, 11 used a concurrent intervention (e.g., treadmill training) with tDCS delivered during the intervention (see Tables 1 and 2 for study characteristics).

Figure 1.

PRISMA flow-chart of study selection

Table 1.

Characteristics of included studies

Author/Year	Diagnosis	Age (M, SD)	Design	Sample Size	Outcome Measures
Aree-uea 2014 [38]	Perinatal stroke	13.5 (3.09)	RCT	46	Modified Ashworth Scale and PROM
Auvichayapat 2017 [37]	Spastic CP	10.4 (1.65)	Observational study (pre-post)	10	Proton magnetic resonance spectroscopy, Tardieu Scale, Quest
Bhanpuri 2015 [44]	Primary or secondary dystonia	15.3 (4.2)	Observational study (pre-post)	9	Barry Albright Dystonia Scale and EMG (step and continuous tracking)
Carlson 2018 [47]	Perinatal stroke	12.3 (3.3)	RCT	34	COPM, AHA, MA, BBT
Dimitri 2017 [39]	Minimally conscious state and spastic quadriplegia	20-year-old	Case-report	1	DOCS and the Ashworth Scale
Duarte 2014 [30]	CP	7.9 (1.75)	RCT	24	Stabilometric analysis, PBS, and PEDI
Gillick 2014 [48]	Arterial perinatal ischemic stroke	10-year-old	Case-report	1	Not applicable
Gillick 2018 [34]	Unilateral CP	12.9 (3.55)	RCT	20	COPM, AHA, Grip Strength
Gillick 2015 [40]	Congenital hemiparesis	14.0 (3.5)	RCT	13	Modified pediatric stroke outcome measure, token intelligence test, Dynamometer, BBT
Grecco 2014 a [36]	Delayed neuromotor development	3-year-old	Case-report	1	Spatiotemporal gait variables
Grecco 2014 b [29]	Spastic CP	15.8 (2.6)	RCT	24	Spatiotemporal gait variables, 6-min walk test, GMFM-88, treadmill test, cortical excitability
Grecco 2014 c [28]	Spastic CP	7.5 (1.65)	RCT	20	Spatiotemporal gait variables, balance using force plate
Grecco 2015 [27]	Spastic diparetic CP	8.5 (2.7)	RCT	20	Spatiotemporal gait variables, GMFM-88, and PEDI
Grecco 2016 [46]	Spastic hemiparetic or diparetic CP	7.8 (7.7)	Observational study (retrospective)	56	Gait Profile Score, 6-min walk test, GMFM-88, gait speed
Grecco 2017 [35]	Ataxic CP	7.1 (2.1)	Observational study (crossover)	6	PBS; Timed Up & Go test, PEDI
Kirton 2017 [33]	Hemiparetic CP	11.5 (3.3)	RCT	23	COPM, AHA
Lazzari 2015 [31]	CP	4–12 years old	RCT	20	Force plate displacement
Lazzari 2017 [32]	CP	7.5 (2.1)	RCT	20	Force plate, Pediatric balance scale, Timed up & go test,
Moura 2017 [41]	Hemiparetic CP	10.7 (2.6)	RCT	20	Spatiotemporal variables
Nagai 2018[45]	Involuntary movement	8-year-old	Case-report	1	Accelerometer data (to measure movement)
Rich 2017 [49]	Hemiparesis (Periventricular leukomalacia or Perinatal stroke)	13.9 (3.0)	Observational study (case-control)	16	Not applicable
Young 2013 [42]	Primary or secondary dystonia	13.1 (4.1)	Observational study (pre-post)	10	EMG measured Error and overflow
Young 2014 [43]	Primary or secondary dystonia	12.6 (3.7)	Observational study (crossover)	14	EMG measured Error and overflow

Note: RCT (Randomized Control Trial), PROM (Passive range of motion), QUEST (Quality of Upper Extremity Skills Test), EMG (Electromyography), COPM (Canadian Occupational Performance Measure), AHA (Assisting Hand Assessment), MA (Melbourne Assessment), BBT (Box and Blocks Test), DOCS (Disorders of Consciousness Scale), PBS (Pediatric Balance Scale), PEDI (Pediatric Evaluation of Disability Inventory), GMFM-88 (Gross Motor Function Measure).

Table 2.

Characteristics of tDCS parameters used in intervention studies

Author/Year	Concurrent Intervention	Anode/Cathode	Current Density	Session Duration	Number of sessions	Electrode placement	Electrode Size
RCTs
Aree-uea 2014 [38]	No	Anode	1 mA	20 minutes	5 sessions	Left M1	5 cm × 7 cm
Duarte 2014 [30]	Treadmill training	Anode	1 mA	20 minutes	10 sessions	Non-dominant M1	5 cm × 5 cm
Gillick 2015 [40]	No	Bihemispheric montage	0.7 mA	10 minutes	1 session	Bilateral motor cortices	5 cm × 7 cm
Gillick 2018[34]	Constraint induced movement therapy	Cathode	0.7 mA	20 minutes	10 sessions	Contralesional M1	Not reported
Grecco 2014 b [29]	Treadmill training	Anode	1 mA	20 minutes	10 sessions	Dominant M1	5 cm × 5 cm
Grecco 2015 [27]	Virtual reality training	Anode	1 mA	20 minutes	10 sessions	Ipsilesional M1	Not reported
Grecco 2014 c [28]	No	Anode	1 mA	20 minutes	1 session	Dominant M1	5 cm × 5 cm
Kirton 2017 [33]	Intense motor learning, individualized therapy, and group activities, constraint-induced movement therapy	Cathode	1 mA	20 minutes	10 sessions	Contralesional M1	5 cm × 5 cm
Lazarri 2015 [31]	Virtual reality training	Anode	1 mA	20 minutes	1 session	M1	5 cm × 5 cm
Lazarri 2017 [32]	Virtual reality training	Anode	1 mA	20 minutes	10 sessions	M1	5 cm × 5 cm
Moura 2017 [41]	Functional training of upper limb	Anode	1 mA	20 minutes	1 session	Contralesional C3/C4	5 cm × 5 cm
Observational Studies
Auvichayapat 2017 [37]	No	Anode		20 minutes	5 sessions	Left M1	5 cm × 7 cm
Bhanpuri 2015 [44]	No	Anode & Cathode	2 mA	9 minutes	10 sessions	Either C3 or C4	4 cm × 7 cm
Dimitri 2017 [39]	Pyscho-sensory stimulation	Cathode	1.5 mA	20 minutes	180 sessions	C4	4 cm × 5 cm
Grecco 2014 a [36]	Treadmill training	Anode& Cathode	1 mA	20 minutes	10 sessions	C3	5 cm × 5 cm
Grecco 2017 [35]	Treadmill training	Anode	1 mA	20 minutes	5 sessions	Cerebellum	5 cm × 7 cm
Nagai 2018[45]	No	Cathode	1 mA	20 minutes	3 sessions	SMA	5 cm × 7 cm
Young 2013 [42]	No	Cathode	1 mA	9 minutes, 20-minutes pause, 9 minutes	1 session	Contralesional C3 or C4	5 cm × 7 cm
Young 2014 [43]	No	Cathode	1 mA	9 minutes, 20-minute pause, 9 minutes	1 session	Contralesional C3 or C4	5 cm × 7 cm

Thirteen studies were included in the meta-analysis examining safety/tolerability, and 7 studies in children with CP were included in the meta-analyses examining tDCS effectiveness on motor outcomes. Meta-analyses are presented below for the motor outcome measures of spatiotemporal gait variables, balance as measured by anterior-posterior (AP) and medio-lateral (ML) sway, Assisting Hand Assessment (AHA), Gross Motor Function Measure (GMFM-88), Pediatric Balance Scale, and Pediatric Evaluation of Disability Inventory (PEDI). A narrative summary is presented below for studies examining spatiotemporal and balance variables which could not be included in the meta-analyses due to missing outcome data points and motor outcome measures of upper extremity tone, hand function, and self-perception of performance and participation.

Risk of bias

The Cochrane Risk of Bias tool revealed that most RCTs presented with a low risk of bias across all domains assessed. (Appendix S.2 and S.3).

Safety and tolerability

Of the 23 studies, 20 reported on dropouts or adverse events. Across 13 studies included in the meta-analysis, there was no difference in dropouts in active (1/144) versus sham (1/144) groups (Figure 2). The reason for one dropout was discomfort, and the reason for the other was not reported. Three observational studies without a sham group (total of 22 participants) reported 2 dropouts due to discomfort (summary of all dropouts, Appendix S.4). The funnel plot suggested low publication bias for dropouts (Appendix S.5). The most common adverse effects in the active group across the 20 studies were tingling (18%), discomfort (8%), and itching (7%) (Appendix S.6).

Figure 2.

Forest plot of dropouts in tDCS trials. The diamond at the bottom of the plot summarizes the best estimate results of the meta-analysis with the width representing the corresponding 95% CI.

Effects of intervention

Meta-analyses

Spatiotemporal gait variables

Across three studies (64 participants) [27–29] (one study with no concurrent motor training [28], one study with treadmill training[29], and one study utilizing virtual reality[27]), anodal tDCS compared to sham stimulation resulted in significant improvements in velocity, stride length, and cadence (p’s<.0005) but not step length and step width (Figures 3 and 4). Two RCTs (44 participants) also reported outcomes at 1-month follow-up post-tDCS [27,29]; significant improvements for active compared to sham-tDCS were observed for cadence (p=.03, Figure 4), and effects on velocity trended towards significance (p=.06, Figure 5).

Figure 3.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for spatiotemporal gait variables immediately post-tDCS from the Inverse-variance random effects model in children with cerebral palsy.

Figure 4.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for Cadence, immediately post-tDCS and at 1-month follow-up, from the Inverse-variance random effects model in children with cerebral palsy.

Figure 5.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for spatiotemporal gait variables at 1-month follow-up, from the Inverse-variance random effects model in children with cerebral palsy.

Balance

tDCS effects on balance as assessed by AP and ML sway while standing (eyes open and eyes closed) were examined by 3 RCTs (64 participants) [30–32]. No significant effects were identified in these studies which used anodal tDCS in conjunction with treadmill [30] or virtual reality training [31,32] (Figure 6).

Figure 6.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for anterior-posterior (AP) and medio-lateral (ML) sway in eyes open (EO) and eyes closed (EC) conditions from the Inverse-variance random effects model in children with cerebral palsy.

Assisting Hand Assessment (AHA)

Two RCTs (43 participants) found no significant effects on the AHA; both studies used cathodal tDCS with constraint-induced movement therapy (CIMT) [33,34] (Figure 7).

Figure 7.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for the Assisting Hand Assessment, 1-week post-tDCS, from the Inverse-variance random effects model in children with cerebral palsy.

Gross Motor Function Measure (GMFM-88)

Two RCTs (44 participants) evaluated gross motor function (GMFM-88) [27,29]. There was a significant effect on the standing domain (p=.04) but not the walking domain; both RCTs used anodal tDCS along with treadmill [29] or virtual reality training [27] (Figure 8).

Figure 8.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for the Gross Motor Function Measure for the standing (GMFM-D) and walking (GMFM-E) domains from the Inverse-variance random effects model in children with cerebral palsy.

Pediatric Balance Scale (PBS)

Two RCTs (44 participants) reported on the PBS and found an effect trending toward significance immediately post-anodal tDCS combined with treadmill [30] or virtual reality training [32] (p=.051) and at 1-month follow-up (p=.09) (Figure 9).

Figure 9.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for the Pediatric Balance Scale, immediately post-tDCS and at 1-month follow-up, from the Inverse-variance random effects model in children with cerebral palsy.

Pediatric Evaluation of Disability (PEDI)

Two studies (44 participants) reported on the PEDI [27,30]. The mobility subdomain, but not the self-care domain, trended toward significance (p=.08) immediately post-anodal tDCS combined with treadmill [30] or virtual reality training [27]. Similar effects were observed at 1-month follow-up (mobility p=0.09, self-care p=.13) (Figure 10).

Figure 10.

Forest plot showing the effect sizes from the comparison between active vs. sham tDCS for the Pediatric Evaluation of Disability for the mobility and self-care domains, immediately post-tDCS and at 1-month follow-up, from the Inverse-variance random effects model in children with cerebral palsy.

Qualitative Analysis

Gait and balance

In a crossover study, 6 children with CP improved significantly in AP sway with eyes closed following 10 sessions of anodal tDCS along with treadmill training, but no effects were observed in the other measures of balance such as the Timed Up and Go Test [35]. A case-report of a child with delayed neuromotor development showed clinically meaningful improvement in spatiotemporal gait variables and balance following 10 anodal tDCS sessions (without concurrent motor training) [36].

Upper-extremity tone

One study reported significant reductions in spasticity (Tardieu scale) in the shoulder, elbow and wrist muscles following one session of anodal tDCS (without concurrent motor training) in 10 children with CP [37]. Another study using 5 sessions of anodal tDCS (without concurrent motor training), in 46 children with CP showed significant reductions in spasticity (Modified Ashworth Scale) immediately post-tDCS compared to sham stimulation for shoulder, wrist, and fingers without improvement in passive range of motion [38]; improvements were maintained only for wrist tone at 48-hour follow-up. Similarly, a case-report using a cathodal montage and 180 sessions of tDCS combined with psycho-sensory stimulation in a 20-year old female with minimally conscious state and spastic quadriplegia found improvements in upper limb hypertonia (Modified Ashworth Scale) [39].

Hand function

A study using a bi-hemispheric montage with one-time tDCS application (no concurrent motor training) in 11 children with CP found no effect on grip strength (hand dynamometer or Box and Blocks test) [40]. A different study using cathodal stimulation with constraint-induced movement therapy found no significant improvement on grip strength in 20 children with CP [34]. However, an observational study using anodal tDCS, without concurrent motor training, found significant improvements in hand function in 10 children with CP [Quality of Upper Extremity Skills Test (QUEST)] [37]. In another study of 20 children with CP, a single-session of anodal tDCS with constraint-induced movement therapy compared to sham resulted in significant improvements in movement duration during a reaching task [41].

Self-perception of performance and satisfaction

Two studies using 10-sessions of cathodal tDCS with constraint-induced movement therapy reported on the COPM; in one study the treatment group showed significant improvements in the performance and satisfaction domains of the COPM 1 week and 2 months post-tDCS [33], while no group difference was found in the second study [34].

tDCS effects in children with primary/secondary dystonia and involuntary movement

Three studies with 33 participants with dystonia reported on outcome measures of EMG tracking and muscle overflow [42–44]; effects were limited to the hand contralateral to the hemisphere receiving stimulation. An open-label study using cathodal montage (without concurrent motor training) found no group effect after 1 session of tDCS, but 4 of 10 participants showed significant improvement in tracking or overflow [42]. Another study reported significant reductions in overflow following 1 session of cathodal tDCS (without concurrent motor training) compared to sham in 14 children with dystonia [43]. An observational study investigating the effects of 10 sessions of cathodal (n=7) or anodal (n=6) tDCS without concurrent motor training demonstrated that of 7 participants receiving cathodal stimulation, 3 showed minor improvements and 2 showed worsening in overflow or tracking [44]; 5 of the 6 participants receiving anodal tDCS showed symptom worsening [44]. A case-study of an 8-year old child reported significant reduction in involuntary movements of the head and neck following 3 sessions of cathodal stimulation (no concurrent motor training) compared to sham [45].

Biomarkers of tDCS responsiveness

The presence of baseline motor evoked potentials (MEP) recorded from the quadriceps was reported to be predictive of gains on the 6-minute walk test and gait speed post-anodal tDCS, while subcortical injury predicted improvements in gait kinematics and gross motor function [46]. In another study of anodal tDCS, concentrations of neurotransmitters including N-acetylaspartate (NAA), glutamine-glutamate, choline, and creatine increased in children with CP following tDCS; further, higher level of NAA/creatine was associated with reduced spasticity [37]. In a study of cathodal tDCS, higher levels of baseline choline and creatine compounds in the affected hemisphere of children with CP were reported to be predictive of improvements in motor performance and satisfaction [47].

tDCS parameters

A study investigated child-specific dosage parameters and found that peak current intensity of 0.7 mA in a child would result in a peak electric field resembling 1 mA current intensity in an average adult, providing that the distribution of current flow is equivalent in both cases [48]. A study examined optimal electrode placement for tDCS intervention in children (8–17 years old) and reported that the 10/20 EEG-based and the TMS-guided method for localizing M1 region produced variable results irrespective of child’s age across typically-developing children and children with hemiparesis [49].

Findings from grey-literature

With regard to safety, one Letter to the editor reported a seizure following anodal tDCS (without concurrent motor training) in a 4-year-old boy with CP and epilepsy who had been seizure-free on anti-epileptic medication leading up to and during tDCS sessions [50]. A study evaluating safety and feasibility in 10 children with unilateral CP reported only mild and transient adverse effects with 2 sessions of anodal tDCS (without concurrent motor training) [51].

Five studies reported the effects of tDCS on gait outcomes. A case report in a child with hemiparetic CP tested the effectiveness of an anodal montage and a bi-hemispheric anodal-cathodal montage with concurrent treadmill training compared with sham stimulation on gait; while gains were reported with both tDCS montages, gains were greater using the bi-hemispheric montage [52]. A study retrospectively analyzed data from two RCTs using tDCS and found that, in 105 children with CP, concurrent treadmill training was superior to concurrent virtual reality training for gait improvements and that real tDCS predicted maintenance of improvements achieved during gait training [53]. Similarly, a case-report examined the effectiveness of a bi-hemispheric anodal-cathodal montage combined with gait training in a child with CP and found that tDCS improved the gait profile score of both paretic and non-paretic limbs [54]. Another case-report using a bilateral montage (ipsilesional anodal and contralesional cathodal) combined with treadmill training in an 8-year old child with CP found that tDCS improved scores on balance and gait profile measures [55]. In a retrospective analysis of 105 children with CP who underwent either anodal or sham stimulation, tDCS amplified the effect of the concurrent gait training regardless of the child’s initial motor impairment level [56].

A case series of 6 children with CP found significant improvements in upper extremity function and reach following 10 sessions of anodal tDCS applied concurrently with motor therapy [57]. An RCT in 10 participants with hemiparetic CP found improvements in speed of the affected hand on the Jebsen Taylor Test of Hand Function (montage not reported) [58].

A case-report in a child with Down Syndrome found a positive effect of a single-session of anodal tDCS (without concurrent motor training) in reducing the adjustment sway of the left arm and the jerking movement of the right arm compared to sham stimulation [59].

With regard to tDCS parameters, a study modeling the electric field in children to individualize tDCS dosage showed that tDCS generated highest electric fields in brain regions located directly under the electrodes and that cortical tissues supported the spread of the electric field to distant areas. Further, electric field strength and location differed in children with arterial ischemic stroke compared to typically-developing children [60].

Discussion

This systematic review included 23 studies reporting on tDCS safety, effectiveness, and parameters in pediatric motor disorders. Meta-analysis of dropouts demonstrated that tDCS is safe and tolerable in children with motor disorders. Anodal tDCS, both in conjunction with motor training and alone, appeared to be effective in improving some spatiotemporal gait characteristics in children with CP. Mixed-effects were seen on balance, upper-extremity function, daily task performance, and participation. Quality assessment revealed that most RCTs presented with low risk of bias.

tDCS safety

Consistent with prior literature, we used total number of dropouts as a surrogate for feasibility [61,62]. Our meta-analysis of 13 studies (1–10 sessions) revealed no difference in dropouts between active and sham tDCS groups, with only 1 dropout in each (active or sham) group. The low number of dropouts may have been influenced by small sample sizes and use of cautious inclusion criteria [62]. Additionally, adverse effects in active tDCS groups were limited to mild skin problems. Most studies in CP excluded children with epilepsy, and seizure was only reported in a Letter to the editor regarding a child with a seizure 4 hours after the third application of tDCS [50]. This child’s epilepsy had been well-controlled on two anti-epileptic medications, one of which was discontinued two-weeks prior to the start of tDCS. The child received 2.5 mg of escitalopram 2 hours before each tDCS session; seizures are listed among the serious reactions associated with this medication [63]. The multiple possible contributors to seizure in this child make it difficult to clearly distinguish the role of tDCS in precipitating the seizure. We were unable to quantitatively analyze the findings of adverse effects because of the lack of or insufficient reporting of adverse events and inconsistent descriptions when reported. Overall, our results reinforce the findings of existing literature in pediatrics demonstrating tDCS to be safe and tolerable in children when standard exclusion criteria are observed [64] and safety guidelines are followed [14,16,65,66].

tDCS effectiveness

Overall, tDCS appeared to have a favorable response on motor outcomes in pediatric motor disorders; however, divergent results were also encountered. The primary outcomes included spatiotemporal gait variables, balance and mobility, upper-extremity function, EMG tracking, muscle overflow, self-care, and self-perception of performance and satisfaction.

Consistent with existing meta-analyses showing positive effects of tDCS on motor learning in adult stroke [17], we found improvements in cadence, velocity, and stride length immediately post-tDCS and at 1-month follow-up [12,17,27,28], while no effects of tDCS were observed on step length and step width [27,28,36]. Walking velocity directly influences cadence and stride length [67]; thus if velocity was primarily improved, secondarily this would lead to improvements in cadence and stride length but not in step length and step width. The significant effect on stride length but not step length may also be attributed to variability in step length in children with hemiplegia compared to diplegia [68]. Of note, spatiotemporal gait variables improved both with tDCS-alone or combined with motor training.

Our meta-analysis revealed mixed effects of tDCS on balance and motor function. Although tDCS effects were not observed on laboratory-based measures such as sway [30–32], we observed significant effects on the GMFM-88 and trend-level findings on the PBS and PEDI [27,28,30,32]. The significant effects on GMFM-88 were observed both with tDCS used alone and in combination with other motor training [27,28]. A meta-analysis of walking ability in stroke had similar findings, with improvements in mobility but not balance [69]. The PBS and PEDI balance measures assess more global parameters such as motor learning and control in addition to balance, and tDCS may have had influenced these parameters [17]. With respect to tDCS montage, studies examining balance used a consistent anodal montage, and thus variability in the responses of these outcomes cannot be attributed to the use of a specific stimulation technique [64]. The positive effects of tDCS on motor performance and trend-level effects of tDCS on balance in children with motor disorders are promising and require future investigation.

Compared to gait and mobility domains, fewer studies have examined tDCS effects on upper extremity function. In contrast to a meta-analysis of adult stroke [70], we found significant beneficial effects of tDCS on upper-limb spasticity [37–39]; benefit was found with either anodal [37,38] or cathodal [39] stimulation and with tDCS used alone [37,38] or along with other motor training [39]. There may be an age- or injury-dependent mechanism that impacts effect of tDCS on tone in children versus adults; however, studies with larger sample sizes are required to validate this finding. Hand function improvements following tDCS were observed as measured by the QUEST or a reaching task [37,41] but not the AHA, Box and Blocks Test, or hand dynamometer [33,34,40]. The variability in efficacy cannot be attributed to tDCS methodology, as these studies used uniform tDCS parameters (contralesional cathode, 1 mA for 20 minutes; 5–10 sessions). However, the QUEST and reaching task primarily assess proximal function whereas the other assessments primarily measure fine motor control. It appears that tDCS may differentially impact gross and fine motor hand function. Overall, our findings are consistent with systematic reviews in adult stroke with either a small-to-moderate change [13,71] or no effect of tDCS on upper limb function [12].

Though a meta-analysis in adult stroke reported improvements in activities of daily living following tDCS [12], our meta-analysis did not indicate significant improvements in the PEDI self-care domain [27,30]. These discrepant findings could be explained by the small sample sizes in our meta-analysis. Our review found positive effects of tDCS on the COPM performance and satisfaction domains [33,34]. Improvements on the COPM but not the PEDI may reflect individualized, client-centered goal setting of the COPM compared to global performance on the PEDI. Additionally, studies investigating self-care and self-perception of performance and satisfaction used different montages (both anodal and cathodal) which may have resulted in divergent outcomes.

The studies in dystonia and involuntary movements showed inconsistent results of tDCS on tracking and overflow measures. Based on these studies, it appears that cathodal tDCS may be more beneficial than anodal tDCS in modulating overflow and involuntary movement. However, given the limited number of studies with small sample sizes, evidence for the effectiveness of tDCS in dystonia and involuntary movement is inconclusive, and further trials are required.

Across 19 intervention studies in the systematic review, tDCS alone versus tDCS coupled with motor training did not seem to differentially impact spatiotemporal, balance, spasticity, and hand-function outcomes. This finding is in contrast to literature from adult neurorehabilitation where more pronounced effects of tDCS were observed when tDCS was combined with motor training [72]. While preliminary, this suggests that tDCS may impact developing brains differently from mature brains and that in children there may be a role for tDCS as a standalone intervention, in addition to the use of tDCS for its priming effects when combined with another motor intervention. The efficacy of tDCS as a standalone intervention versus in combination with concurrent interventions must be empirically studied in children with motor disorders. If effective, standalone tDCS may be particularly useful for children who are unable to consistently engage in traditional motor interventions.

Biomarkers of tDCS responsiveness and tDCS parameters

Anatomical and physiological differences in children compared to adults must be considered when designing pediatric tDCS rehabilitation protocols. The reviewed studies suggest that tDCS current intensity differentially impacts depolarization in children compared to adults, with a smaller intensity producing a larger effect and a similar intensity producing a stronger and a more expansive spread in children [40,73,74]. However, similar to adult stroke [74], baseline MEP and post-tDCS metabolite concentrations were predictive of motor improvements [46,47]. Additionally, M1 localization differed based on the methodology used (10/20 EEG vs. TMS) which can impact the peak electric field [75]. The 10/20 EEG system is generally utilized to determine the electrode placement though greater precision can be obtained with neuronavigation and TMS [75]. The effects of tDCS on function in reference to electrode placement need to be systematically studied.

Only one study in this review reported the participants’ medication use during the study [38]. Neurostimulant medications can impact the effects of tDCS on behavior in adults [76]. The effect of neurostimulant medications on tDCS should thus be explored in children, especially given the frequent use of these medications in children with neurodevelopmental disorders [77].

Limitations

The RCTs in this meta-analysis contained small sample sizes, and thus, the interpretation of results warrant caution. This review analyzed primary outcomes and is not a comprehensive review of all the outcomes in the included studies. The generalizability of the findings is limited to clinical populations included in this review. We were unable to conduct sub-group analysis due to the limited number of available studies. The inclusion of studies published only in English may have resulted in bias [78]. Due to the small sample sizes of studies in the meta-analysis, we have included trend-level findings which may have led to over-estimation of the true effects of tDCS. Additionally, due to a limited number of included studies, we were unable to systematically analyze the effects of tDCS when used alone versus with other interventions. We have included abstracts and a Letter to the editor to reduce publication bias [19], however these data must be interpreted with caution.

Directions for future studies

Given the increasing application of tDCS, several research groups have proposed regulatory and clinical guidelines for tDCS use [79,80]. These guidelines include detailed reporting on stimulation parameters and adverse effects, sham-controlled trials, larger sample sizes, and longer follow-ups. In addition to these recommendations, we propose the following considerations when designing clinical trials in pediatric motor disorders.

1. Use standardized reporting, adhering to Consolidated Standards of Reporting Trials (CONSORT) guidelines [81].

2. Use a standardized tDCS adverse effect reporting measure [82].

3. Report outcome data necessary for inclusion in meta-analyses (e.g., means and standard deviation).

4. Use measures that assess change in functional performance.

5. Use common data elements as outlined by the National Institutes of Health (NIH) [83].

6. Use response and predictive biomarkers according to NIH guidelines [84].

7. Report the concurrent use of neurostimulant medications.

Conclusions

This is the first systematic review and meta-analysis investigating the safety and effectiveness of tDCS in pediatric motor disorders. Based on the studies included in this review, tDCS is safe and tolerable and may enhance functional outcomes in children with motor disorders and potentially serve as a viable additive to motor learning interventions.

Supplementary Material

1

Appendix S1. A list of the search terms included in the systematic review and meta-analysis.

2

Appendix S2–6. Figures and Tables depicting the Risk of Bias, Adverse Effects, and Publication Bias.

List of abbreviations

tDCS

Transcranial direct current stimulation

RCT

Randomized controlled trials

CP

Cerebral palsy

PROSPERO

The international prospective register of systematic reviews

COPM

Canadian Occupational Performance Measure

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-analyses

AHA

Assisting Hand Assessment

AP

Anterior-posterior

ML

Medio-lateral

GMFM-88

Gross Motor Function Measure

PEDI

Pediatric Evaluation of Disability Inventory

PBS

Pediatric Balance Scale

QUEST

Quality of Upper Extremity Skills Test

MEP

Motor evoked potentials

NAA

N-Acetylaspartate

CONSORT

Consolidated Standards of Reporting Trials