Effects of Transcranial Direct Current Stimulation Combined with Physiotherapy on Gait Pattern, Balance, and Functionality in Stroke Patients. A Systematic Review

Background: The effectiveness of transcranial direct current stimulation (tDCS) together with conventional physiotherapy in motor rehabilitation after stroke has been widely studied. Despite this, few studies have focused on its application in gait and balance rehabilitation. This review aimed to determine the efficacy of transcranial direct current stimulation combined with conventional physiotherapy on gait, balance, and the functionality of the lower limb after stroke. Methods: This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Four electronic databases were systematically searched for relevant articles. Randomized clinical trials in English or Spanish that evaluated the use of the transcranial direct current stimulation, combined with physiotherapy, to improve gait, balance, and lower limb functionality after stroke were included. Main results: 10 articles were included, with a total of 222 subjects. Data about population, assessment tools, protocols, sessions, and results were extracted. The methodological quality of the included studies ranged between 3 and 5. Conclusion: The use of transcranial direct current stimulation combined with physiotherapy improves gait parameters, static and dynamic balance, and lower limb functionality in stroke patients. Long-term effects have not yet been demonstrated.


Introduction
Gait impairments are a common entity in stroke survivors. It is estimated that 50% of patients in the first phase of the pathology will not have the ability to walk; 20-40% will not be able to walk in the chronic phase of the disease (>6 months), and 11% will only be able to walk with the use of aids [1,2]. Gait patterns are heterogeneous in these subjects, with disturbances in different phases of the gait cycle, which differ from healthy controls, in addition to disturbances in temporospatial parameters even in those who walk independently [3][4][5]. To achieve an independent gait is one of the objectives of people who have suffered a stroke [6]. However, this does not only encompass the achievement of a household walk but also community ambulation (which oscillates between 0.66 and 1.31 m/s) [7], since this influences the quality of life perceived by the subjects at a psychological level [8]. Rehabilitation programs have been found to improve quality of life in this regard [9], so there have been many methods used in gait training. However, none of these have been proven to be superior to another [10]. Repetitive gait training is a method with great results in the early stages of stroke [11,12]. Examples of these therapies are bodyweight supported treadmill training (BWSTT) and robot assisted gait training (RAGt) [12]. According to the most current therapies, noninvasive brain stimulation (NIBS) has shown promising results in the treatment of stroke survivors [13]. One of the most widely used methods is transcranial direct current stimulation (tDCS) [14]. According to several studies, tDCS can promote motor recovery in stroke patients by modifying cortical excitability (it appears to modify the discharge threshold of cortical neurons, using lowamplitude currents (0.5-2.0 milliamps or mA)) [15]. The way in which tDCS modulates cortical excitability is polarity-dependent, with several application models: (a) anodic stimulation, which increases cortical excitability by decreasing the excitability threshold; this model in stroke is placed in the injured cerebral hemisphere, (b) cathodic stimulation, which decreases cortical excitability by increasing the excitability threshold; this model in stroke is placed in the cerebral hemisphere contralateral to the stroke, (c) dual stimulation, which is a combination of anodic stimulation on the affected cerebral hemisphere and cathodic stimulation on the non-affected [16,17].
It is important to consider that tDCS is a technique with great inter-individual variability between subjects [15]. Even so, its efficacy in functional lower limb recovery remains unclear, and studies are limited [18]; those that do exist claim that more careful protocols are necessary due to the lower limb motor cortex representation proximity of both hemispheres and may produce the same sign modulation; in which both hemispheres would be stimulated with the same type of current, without producing the wanted therapeutic effects [19]. In addition, 2 milliamps (mA) stimulations will be necessary to achieve activity in the cortical representation of the lower limbs since these are located in the interhemispheric fissure, at a greater depth in the cerebral cortex compared to the cortical representation of the upper limbs, which will need lower intensity stimuli to be excited (1-1.5 mA) [20].
Although there are not many studies that support the use of tDCS in gait, there are several investigations focused on demonstrating its effectiveness in both gait training and lower limb rehabilitation, especially in combination with other therapies. Some recent systematic reviews and meta-analyses [21][22][23] studied the effectiveness of tDCS in the rehabilitation of gait and lower limb functionality. Other reviews [24] focused on the efficacy of tDCS in balance rehabilitation. These reviews analyzed articles that combined the tDCS with conventional physical therapy (PT) or robotic gait training. In this sense, the objective of the present study was to determine the efficacy of the tDCS combined with conventional PT on gait, balance, and lower limb functionality after stroke due to the limitations of some clinicians in the application of instrumental PT methods.

Clinic Question
The clinic question was posed according to the "population, intervention, comparison, and results" (PICO) model. Studies were included if they involved patients diagnosed with stroke who underwent an intervention using tDCS, combined with conventional PT, focused on the rehabilitation of locomotor gait patterns and balance, comparing its application to conventional PT alone. Conventional therapy was understood as any intervention of PT that did not use an instrumental process. The outcomes found in these studies had to be evaluated by quantitative or qualitative methods that measured gait coordination, gait spatial-temporal parameters, balance, functionality, and quality of life.

Search Strategy and Database
The following databases were searched in March 2020: MEDLINE Complete, PubMed, Scopus, and SciELO. Some medical subject headings, such as "Stroke", "Transcranial Direct Current Stimulation", "Gait Disorders, Neurologic", "Postural Balance", or "Physical Therapy Modalities", were used along with other related terms and were combined with the Boolean operators AND and OR. For the present review, experts were also consulted; the bibliography of the reviewed studies, gray literature, in addition to searching for protocols published in the registries, was reviewed for the inclusion of possible studies. Medline complete search is summarized in Table 1.

Eligibility Criteria
We included randomized clinical trials that investigated the effects of tDCS on gait and balance of stroke patients. We included studies with at least one group treated with tDCS combined with conventional PT and one control group treated with conventional PT treatment with sham tDCS or without tDCS. We included studies that applied any tDCS protocol, with intensities from 0.5 to 2 mA. Studies had to include measures of gait coordination, gait speed, or balance. If the studies showed measures of lower limb functionality, lower limb strength, and risk of falls, that would be included as secondary outcomes. Finally, only articles in English or Spanish were included, without limiting the year published.

Study Selection and Data Extraction
The clinical question was identified according to the PICO format, extracting the population included in the study (n), the inclusion criteria applied, the measures used for the participant evaluation, the description of the intervention carried out, and the results obtained. Titles and abstracts were screened by two independent authors to select which studies met the inclusion criteria; disagreements related to the eligibility criteria were solved by a third researcher. For the potential studies that met the inclusion criteria, full text articles were obtained. A new screening was performed independently by two researchers to choose the studies included in the review.

Methodological Quality Assessment and Risk of Bias
To analyze their methodological quality and guarantee their objectivity, the Oxford quality scoring system (Oxford, UK, Jadad, 1996) [25] was used. We divided the articles into three scores: (<3) indicated poor methodological quality, (3-4) indicated a moderate methodological quality, and (5) indicated a high methodological quality.
To assess the risk of bias of each randomized clinical trial, the Cochrane library criteria were used, based on the criteria of random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Study data were scored according to three criteria: "Low risk", "High risk", or "Unclear risk".
The review was carried out following the PRISMA statement (checklist for systematic reviews and meta-analysis) [26]. Both methodological processes were carried out by two independent researchers; in case of different scores, a third researcher resolved the differences.

Study Selection
A total of 377 articles were found, duplicated studies were eliminated, leaving a total of 245 original articles. A standard screening checklist based on the eligibility criteria was used; studies other than randomized clinical trials in a non-relevant population that had a different intervention than tDCS or with a different objective on gait were not selected. The abstracts of the potentially relevant articles were read to see if they answered the research question and met the inclusion criteria. Ten studies were included in this review [27][28][29][30][31][32][33][34][35][36], with a total of 222 subjects. The selection process is shown in the flowchart (Figure 1). The 25 excluded articles and their reason for exclusion are shown in the Appendix A.

Study Characteristics and Result Synthesis
According to the primary outcomes, six studies evaluated gait patterns and temporalspatial parameters, three studies used three-dimensional systems, one study used the Rivermead Motor Assessment (RMA), two studies used the functional ambulatory category (FAC), one study used the 10-m walking test (10MWT), and three studies used the 6-min walking test (6MWT). Nine studies evaluated balance, four studies used the timed upand-go test (TUG), two studies used the Tinetti assessment tool (POMA), two studies used the Berg Balance Scale (BBS), one study used posturography, and one study used balance platforms.
According to the secondary outcomes, two studies evaluated functionality, one used the Fugl-Meyer Assessment lower limb sub-scale (FMA-LE), one used the Rivermead Mobility Index (RMI), three studies evaluated lower limb strength, two used dynamometry, one study used electromyography, six studies evaluated the risk of falls, one study used the rate of falls (RF), two studies used the Tinetti assessment tool, and four studies used the TUG.
Four studies assessed the application of dual tDCS [27,30,33,35]. Three studies found significant improvement in gait and balance [27,33,35], and one study found no significant improvement [30]. Andrade et al. [27] showed significant improvements in gait measured by 6MWT and balance measured by BBS and RF. Saeys et al. [33] showed significant improvements in balance measured by POMA and in the lower limb functionality measured by RMA, and Tahtis et al. [35] showed significant improvements in dynamic balance measured with the TUG.
Concerning the stroke phase of the participants recruited, there were seven studies based on acute and sub-acute stroke subjects [27][28][29][30][33][34][35] and four studies based on chronic stroke patients [29,31,32,36]. On the subject of the studies that evaluated individuals in the acute/subacute stages, four studies found significant improvements in terms of gait and balance [27,28,34,35], and three studies did not show significant differences between the groups [29,30,33]. Long-term efficacy was measured by two studies [27,29]. Among them, the study carried out by Andrade et al. [27] showed significant improvements at 1 to 3 months of follow-up.
Group B: more improvements in BBS (p = 0.001-p = 0.007), and in 6MWT (no significant) than A-C, respectively. T0 baseline, T1 9 days after first treatment, T2 9 days after second treatment. 6MWT significantly improved on real tDCS after 1 h after stimulation (p = 0.038). No significant differences were observed for the other evaluation.

Discussion
Similar studies in stroke patients were grouped in which the same parameters were compared whenever possible. Regarding the risk of bias, the item with the highest risk of bias was the blinding of participants and staff. Moreover, we found three articles with a high risk due to not blinding the therapists.
The sessions lasted between 10 and 20 min, according to what was established by Jeffery et al. [20]. The authors suggested that to achieve excitation in the lower limb motor cortex areas, 2 mA intensities are required for at least 10 min, these intensities being higher than those used to stimulate the upper limb motor cortex (1 mA). The stimulation intensities in the included studies were between 1.5 and 2 mA. In the studies that worked with intensities lower than 2 mA [29,33], no significant differences were observed between groups. It seems that currents of 2 mA should be used to achieve significant improvements in applications aimed at improving parameters related to the lower limb, such as gait, balance, risk of falls, or functionality and strength of the lower limbs.
The number of sessions ranged from 1 to 16, leading us to establish two groups: shortterm and long-term studies. Similar improvements were found in both groups. The studies carried out by Ojardias et al. [31], Sohn et al. [34], and Tahtis et al. [35] found significant improvements in terms of balance, static balance (eyes closed and opened), quadriceps isometric strength, and gait after one tDCS session. Furthermore, Madhavan et al. [37] measured the precision of ankle movements following visual feedback with the foot, finding that the group that received a single session of tDCS improved more rapidly than the group that received sham stimulation (5 min vs. 10 min). Similar results were also found by Kaski et al. [38] in leukoaraiosis subjects after one tDCS session, where the authors observed improvements in gait velocity, stride length, stride length variability, and dynamic balance, assessed by video analysis and the TUG test. Five studies evaluated the results of several treatment sessions [27][28][29]32,33]. These included the studies carried out by Andrade et al. [27], Chang et al. [28], and Park et al. [32], who found significant improvements in terms of gait parameters, lower limb functionality, balance, and RF. The studies that did not find improvements were those that worked with lower intensities than 2 mA. Among these studies, two followed the sample for 1 to 3 months [27,29], finding maintenance of improvements at three months follow-up in the study carried out by Andrade et al. [27]. However, the studies carried out by Danzl et al. [39] and Geroin et al. [40], both using RAGt, did not find significant improvements between the study groups. The study of Geroin et al. [40] worked with intensities and times lower than those recommended. In contrast, the study of Seo et al. [41] found improvements in the FAC test and the 6MWT in 21 subjects with chronic stroke treated by RAGt. Although a single session and several treatment sessions have shown statistically significant improvements in favor of tDCS, the difference between the applications could be the duration of the improvements. Single sessions of tDCS have modified cortical excitability for a few minutes, while several treatment sessions have modified cortical excitability for hours, even a day [42]. This is clinically relevant for use in combination with physical therapy, as a longlasting state of increased (or decreased) excitability will be vital to improve the neuroplastic brain changes.
Regarding the application mode, tDCS seems to be effective in the rehabilitation of gait and balance, showing promising results in an anodic and dual application. Regarding cathodic application, its efficacy could not be determined due to the inclusion of only two studies of these characteristics. However, one of these studies stated that it is just as effective as anodic application. Khedr et al. [43], in a sample of 40 stroke patients, observed no differences between anodic or cathodic stimulation; both significantly improved function compared to sham-treated patients, but there were no significant differences between the two groups. The differences between Khedr et al. [43] and Fusco et al. [29] could be explained by the use of lower intensities than recommended and the small sample size in the Fusco study. The study by Andrade et al. [27] compared different application modes (anodic, cathodic, dual, or sham) and concluded that all groups showed a significant improvement compared to the sham group and the group that received the dual-tDCS stimulation showed significant improvements in respect of all types of stimulation in terms of gait and balance. Future research will be necessary to affirm which type of current is most effective in the rehabilitation of parameters related to the lower limbs; dual stimulation (anode in ipsilateral hemisphere to stroke, cathode in the contralateral hemisphere) appears to be a promising treatment option.
Regarding the stroke stage, in the acute and sub-acute stage, Tahatis et al. [35], Sohn et al. [34], and Andrade et al. [27] found significant improvements in balance; Chang et al. [28] found significant improvements in the functionality of the lower limb, and Andrade et al. [27] found improvements in gait and the RF. However, Leon et al. [44] found no significant improvement in the study of 49 subjects with subacute stroke in which they studied the application of tDCS combined with RAGt. No further studies were found evaluating gait and balance in these stages of stroke, so they were compared with articles that studied the application of tDCS in the upper limb. In the study carried out by Di Lazzaro et al. [45], significant changes were found in the inter-hemispheric imbalance in acute stroke subjects when they studied the use of the tDCS combined with constraint-induced movement therapy measured by evoked motor potentials. However, they did not find significant differences in the user manual function clinical scale. The included studies that did not find significant improvements were those that worked with intensities lower than recommended [29,33], so tDCS seems to be effective in the rehabilitation of the lower limbs in patients in acute and subacute stages. Klomjai et al.'s [30] study did not show significant improvements, although the parameters used were those recommended.
Regarding chronic stroke patients, Park et al. [32] and Ojardias et al. [31] found significant improvements in gait parameters. These results are consistent with the changes found by Danzl et al. [39] in the cortical excitability measured by evoked motor potentials and Seo et al. [41] in terms of gait and functionality in chronic stroke patients treated by RAGt and tDCS. There were two studies that did not show significant improvements [29,36]. Differences could be the application time since Utarapichat et al. [36] applied only 10 min of stimulation compared to 15 and 20 that applied the studies that found significant improvements, and intensities since Fusco et al. [29] applied a lower intensity than recommended. Thus, tDCS seems to be an effective treatment in acute, sub-acute, and chronic stroke patients.
For future research and clinical application in relation to stroke patients and the parameters related to the lower limb, such as gait, balance, functionality, or strength, it seems that there are improvements in all stages of the disease when applying tDCS and conventional PT. Several treatment sessions seem to achieve long-lasting effects on cortical excitability, being essential applications of 2 mA on the region of the leg of the primary motor area to achieve excitability brain changes. Dual tDCS appears to be a promising treatment option, although it needs further study.

Study Limitations
There were limitations related to the methodology and the low number of studies that compared the use of tDCS together with conventional PT. Regarding the methodology, only articles in English or Spanish were chosen, without limiting the methodological quality (three studies presented a methodological quality classified as "poor"), so the results of the studies have to be read carefully. Three studies were included which did not report whether their subjects performed any type of PT; these were included because they met the rest of the criteria and used only tDCS and no other intervention.
The included studies showed differences in the number of sessions, ranging from 1 to 16, but the cases that performed interventions of more than one session were homogeneous in their frequency (between 4 days/week and 5 days/week). A small number of studies were found to carry out the discussion. Because of this, we included a study on a pathology other than stroke and a study that carried out an intervention other than gait, balance, or lower limb training.

Conclusions
The use of tDCS combined with PT improves gait parameters, static and dynamic balance, and lower limb functionality in stroke patients. The parameters that have shown improvements are 2 mA for at least 10 min, with anodic or bihemispheric stimulation. These parameters have shown improvements at any stage of stroke in single or multiple session protocols, but their long-term efficacy has not been demonstrated. A greater number of comparative studies of modalities will be necessary to determine which method is more effective.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
Articles excluded in the systematic review: