Noninvasive Brain Stimulations for Unilateral Spatial Neglect after Stroke: A Systematic Review and Meta-Analysis of Randomized and Nonrandomized Controlled Trials

Background Unilateral spatial neglect (USN) is the most frequent perceptual disorder after stroke. Noninvasive brain stimulation (NIBS) is a tool that has been used in the rehabilitation process to modify cortical excitability and improve perception and functional capacity. Objective To assess the impact of NIBS on USN after stroke. Methods An extensive search was conducted up to July 2016. Studies were selected if they were controlled and noncontrolled trials examining transcranial direct current stimulation (tDCS), repetitive transcranial magnetic stimulation (rTMS), and theta burst stimulation (TBS) in USN after stroke, with outcomes measured by standardized USN and functional tests. Results Twelve RCTs (273 participants) and 4 non-RCTs (94 participants) proved eligible. We observed a benefit in overall USN measured by the line bisection test with NIBS in comparison to sham (SMD −2.35, 95% CI −3.72, −0.98; p = 0.0001); the rTMS yielded results that were consistent with the overall meta-analysis (SMD −2.82, 95% CI −3.66, −1.98; p = 0.09). The rTMS compared with sham also suggested a benefit in overall USN measured by Motor-Free Visual Perception Test at both 1 Hz (SMD 1.46, 95% CI 0.73, 2.20; p < 0.0001) and 10 Hz (SMD 1.19, 95% CI 0.48, 1.89; p = 0.54). There was also a benefit in overall USN measured by Albert's test and the line crossing test with 1 Hz rTMS compared to sham (SMD 2.04, 95% CI 1.14, 2.95; p < 0.0001). Conclusions The results suggest a benefit of NIBS on overall USN, and we conclude that rTMS is more efficacious compared to sham for USN after stroke.


Background
Stroke is the second leading cause of death worldwide and the primary cause of chronic disability in adults [1]. In the United States, it is the fourth leading cause of death overall [2]. Among people who survive a stroke, unilateral spatial neglect (USN) is the most frequent disorder for right hemisphere lesions [3].
The incidence of USN varies widely from 10% to 82% [4,5]. USN is characterized by the inability to report or respond to people or objects presented on the side contralateral to the lesioned side of the brain and has been associated with poor functional outcomes and long stays in hospitals and rehabilitation centers [6].
Pharmacological interventions such as dopamine and noradrenergic agonists or procholinergic treatment have been used in people affected by USN after stroke, but the evidence derived from a Cochrane systematic review that included only two available RCTs was very low and inconclusive [7].
Other nonpharmacological rehabilitation techniques have been explored for USN with the aim to facilitate the recovery of perception and behavior, which include right half-field eye-patching [8], mirror therapy [9], prism adaptation [10], left-hand somatosensory stimulation with visual scanning training [11], contralateral transcutaneous electrical nerve stimulation and optokinetic stimulation [12], trunk rotation [13], repetitive transcranial magnetic stimulation [14], galvanic vestibular stimulation [15], and dressing practice [16]. However, their results do not support the use of these techniques in isolation for improvement of secondary outcomes such as performance and sensorimotor functions, activities of daily living (ADLs), or quality of life [9,14,17]. Noninvasive brain stimulations (transcranial direct current stimulation (tDCS) and repetitive transcranial magnetic stimulation (rTMS)) have already shown their ability to modify cortical excitability [18]. tDCS is a noninvasive method used to modulate cortical excitability by applying a direct current to the brain that is less expensive than repetitive transcranial magnetic stimulation (rTMS). The latter is an electric current that creates magnetic fields that penetrate the brain and can modulate cortical excitability by decreasing or increasing it and potentially improve perceptual and cognitive abilities [19,20].
A previous Cochrane systematic review summarized results about the effects of tDCS versus control (sham/any other intervention) on activities of daily living (ADLs) among stroke survivors. The authors included 32 randomized controlled trials (RCTs) and concluded that tDCS might enhance ADLs, but upper and lower limb function, muscle strength, and cognitive abilities should be further explored [21]. Another Cochrane systematic review assessed the efficacy of repetitive transcranial magnetic stimulation (rTMS) compared to sham therapy or no therapy for improving function in people with stroke. The 19 included trials showed that rTMS was not associated with a significant increase in ADLs or in motor function; therefore, the authors do not support the use of rTMS for the treatment of stroke, and they plan to complete further trials to confirm their findings [22]. Previous reviews were, however, limited in that they did not include non-RCT studies nor did they evaluate the newest noninvasive brain stimulation-theta burst. We therefore conducted a systematic review of RCT and non-RCT studies that assessed the impact of tDCS, rTMS, and TBS for unilateral spatial neglect after stroke.

Methods
We adhered to methods described in the Cochrane Handbook for Intervention Reviews [23]. Our reporting also adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [24] and Meta-Analysis of Observational Studies in Epidemiology (MOOSE) statements [25].
2.1. Eligibility Criteria. The eligibility criteria are as follows: (1) Study designs: RCTs, quasi-RCTs, and non-RCTs (2) Participants: adults over 18 years of age, regardless of gender and the duration of illness or severity of the initial impairment, with USN after any type of stroke diagnosis (ischemic or intracranial hemorrhage) measured by clinical examination or radiographically by computed tomography (CT) or magnetic resonance imaging (MRI), regardless of whether they were included after evaluation by standardized USN tests.
(3) Interventions: any noninvasive brain stimulations such as tDCS, rTMS, and including theta burst (continuous TBS (cTBS) or intermittent theta burst (iTBS)) (we considered evaluating both the different types of stimulations (i.e., cathodal tDCS versus anodal tDCS versus dual tDCS) and types of frequency (i.e., high-frequency versus low frequency)) (4) Comparators: interventions were to be compared against sham stimulation or any conventional stroke rehabilitation (e.g., pharmacological therapy or nonpharmacological therapy such as right half-field eyepatching, mirror therapy, prism adaptation, left-hand somatosensory stimulation, and visual scanning training or other conventional treatment) We also considered noninvasive brain stimulations as an adjunct to any type of conventional stroke rehabilitation.
(5) Outcomes: (i) Overall USN measured by any paper-andpencil tests, such as the line cancellation task [26], the line bisection test [27], or the star cancellation test [28], and by any validated specific instrument, such as the Catherine Bergego Scale [29], and the Behavioral Inattention Test [30] (ii) Disability in neurological and functional abilities as measured by any validated specific instrument, such as the National Institutes of Health Stroke Scale and the Modified Rankin Scale [31], the box and block test [32], or the Fugl-Meyer Assessment [33]

Selection of Studies.
Two pairs of reviewers independently screened all titles and abstracts identified by the literature search, obtained full-text articles of all potentially eligible studies, and evaluated them for eligibility. Reviewers resolved disagreement by discussion or, if necessary, with third party adjudication. We also considered studies reported only as conference abstracts.

Data Extraction.
Reviewers underwent calibration exercises and worked in pairs to independently extract data from included studies. They resolved disagreement by discussion or, if necessary, with third party adjudication. They abstracted the following data using a pretested data extraction form: study design, participants, interventions, comparators, outcome assessed, and relevant statistical data.
2.5. Risk of Bias Assessment. Reviewers, working in pairs, independently assessed the risk of bias of included RCTs using a modified version of the Cochrane Collaboration's instrument (http:/distillercer.com/resources/) [39]. That version includes nine domains: adequacy of sequence generation, allocation sequence concealment, blinding of participants and caregivers, blinding of data collectors, blinding for outcome assessment, blinding of data analysts, incomplete outcome data, selective outcome reporting, and the presence of other potential sources of bias not accounted for in the previously cited domains [40]. For incomplete outcome data in individual studies, we stipulated as low risk of bias for loss to follow-up as less than 10% and a difference of less than 5% in missing data between intervention/exposure and control groups.
When information regarding risk of bias or other aspects of methods or results was unavailable, we attempted to contact study authors for additional information.
2.6. Certainty of Evidence. We summarized the evidence and assessed its certainty separately for bodies of evidence from RCT and non-RCT studies. We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology to rate certainty of the evidence for each outcome as high, moderate, low, or very low [41]. In the GRADE approach, RCTs begin as high certainty and non-RCT studies begin as moderate certainty. Detailed GRADE guidance was used to assess overall risk of bias [42], imprecision [43], inconsistency [44], indirectness [45], and publication bias [46] and to summarize the results in an evidence profile (Table 1).
We planned to assess publication bias through visual inspection of funnel plots for each outcome in which we identified 10 or more eligible studies; however, we were not able to do so because there were an insufficient number of studies to allow for this assessment.
2.7. Data Synthesis and Statistical Analysis. We calculated pooled inverse variance standardized mean difference (SMD) and associated 95% CIs using random-effects models. We addressed variability in results across studies by using I 2 statistic and the P value obtained from the Cochran chi square test. Our primary analyses were based on eligible patients who had reported outcomes for each study (complete case analysis We planned to conduct subgroup analyses only when five or more studies were available, with at least two in each subgroup. We planned to synthesize the evidence separately for bodies of evidence from RCT and non-RCT studies by a sensitivity analysis.
None of the included studies evaluated noninvasive brain stimulations as an adjunct to any type of conventional stroke rehabilitation. Figure 2 describes the risk of bias assessment for the RCTs and non-RCTs, respectively. The major issue  [51] study (newest trial in the meta-analysis). 1 The majority of the studies were ranked as high risk of bias for both allocation sequence and allocation concealment. 2 There was a substantial heterogeneity (I 2 > 70%). 3 There was no substantial difference related to the mean age and eligibility criteria throughout the six included studies. 4 95% CI for absolute effects includes clinically important benefit and no benefit.
Certainty in evidence was rated as moderate because of risk of bias due to the studies that were ranked as high risk of bias for both allocation sequence and allocation concealment (Figure 2).   (Figure 8). However, in the subgroup analysis with the use of 1 Hz rTMS, we found a statistically significant difference compared to sham (SMD 2.04, 95% CI 1.14, 2.95; p < 0 00001; I 2 = not applicable). Regarding the use of TBS, there was no benefit compared to sham (SMD −0.01, 95% CI −0.89, 0.87; p = 0 98; I 2 = not applicable). Certainty in evidence was rated as low because of inconsistency and risk of bias due to the studies that were ranked as high risk of bias for both allocation sequence and allocation concealment ( Figure 2).  [64] was the only study that reported on adverse events; no significant adverse effect of tDCS was reported, except only a few cases of minimal irritation of the skin beneath the electrodes.
None of the included studies reported on the following outcomes: neurological and functional disabilities, loss of balance, depression or anxiety and evaluation of poststroke fatigue, quality of life, and death.

Synthesized
Results from Non-RCTs. The non-RCTs did not report data in a usable way to allow for any statistical analysis.

Main Findings.
Based on pooled data from six randomized trials with 116 participants, we found evidence for a benefit in overall USN with noninvasive brain stimulation, especially with the use of rTMS in comparison to the sham (Figures 5, 7, and 8). The evidence is from moderatequality evidence because of risk of bias due to the studies that were ranked as high risk of bias for both allocation sequence and allocation concealment (Figure 2). Non-RCT studies provided no evidence, suggesting that future trials should adhere to CONSORT guidelines to ensure clarity and reproducibility in the reporting of methods.
We presented the results of overall USN in a forest plot, which showed a statistically significant difference between the noninvasive brain stimulations and sham in the following tests: line bisection test, Motor-Free Visual Perception Test, and Albert's test and line crossing test. Nevertheless, the study also showed a nonsignificant difference between the noninvasive brain stimulations and sham on the star cancellation test.
Several noninvasive brain stimulations have been explored to determine whether some of these techniques might be useful in promoting recovery from USN after stroke. The lesion of the right parietal cortex after stroke causes disinhibition of the left hemisphere and thus a  pathological overactivation of the latter. This overactivation in the left depresses the neural activity by an increased inhibition on the right hemisphere, aggravating the perception. The rTMS can generate currents capable of depolarizing cortical neurons, and tDCS changes cortical activity by means of small electric currents and does not evoke action potentials. The tDCS has the advantage that the device is inexpensive, portable, and easy to use, but rTMS presented more activation of the neural network and induced a neuroplastic response for a long-term potentiation [65].
In three of four meta-analyses, rTMS was responsible for the improvement of overall USN, revealing that an electric current is an effective strategy for generating lasting promising effects in the brain. Unfortunately, we did not find any significant TBS or tDCS effects compared to sham procedures.

Strengths and Limitation.
Strengths of our review include a comprehensive search; assessment of eligibility, risk of bias, and data abstraction independently and in duplicate; assessment of risk of bias that included a sensitivity analysis addressing loss to follow-up; and use of the GRADE approach for rating the certainty of evidence for each outcome (Table 3). Furthermore, there were no language restrictions, and translations of non-English trials were obtained whenever possible.
The primary limitation of our review is the low certainty consequent to study limitations. We identified a small number of RCTs with a modest number of participants resulting in wide confidence intervals. The total number of participants was relatively very low (RCTs n = 278, non-RCTs n = 94) due to the small sample sizes of individual trials, which led to downgrading the quality of evidence in some instances because underpowered trials are likely to have a greater degree of imprecision.
Moreover, selection bias and unblinding were substantial. Another limitation of this review was having an insufficient number of included studies to allow for the complete statistical analysis that we had planned. We were not able to assess publication bias because there were fewer than 10 eligible studies addressing the same outcome in a meta-analysis. We also planned to perform subgroup analyses according to the characteristics of stroke type, type of stimulation, type of frequency, and comparators (type of control intervention, i.e., pharmacological therapy versus nonpharmacological). However, we also were not able to conduct these analyses because they did not meet our minimal criteria, which was at least five studies available with at least two in each subgroup.   Although this review presents several limitations, the issue is whether one should dismiss these results entirely or consider them bearing in mind the limitations. The latter represents our view of the matter.

Relation to Prior
Work. The research question we investigated in our review has been addressed before from different perspectives using our population of interest but with a different intervention (i.e., pharmacological intervention) [7] or investigating either the intervention or the control arms explored in this review but with a different population (e.g., idiopathic Parkinson's disease (IPD) [48], panic disorder in adults [66], or amyotrophic lateral sclerosis or motor neuron disease) [13].
Two Cochrane reviews [21,49] evaluated the effect of tDCS in people after stroke but not in comparison with rTMS; instead, the authors compared tDCS with placebo, sham tDCS, no intervention, or conventional motor rehabilitation. The first review's [49] authors found evidence of effect regarding activities of daily living performance at the end of the intervention period and at the end of follow-up. However, the results did not persist in a sensitivity analysis including only trials of good methodological quality. In the second review [21], the authors found that there were no studies examining the effect of tDCS on cognition in stroke patients with aphasia.
Another Cochrane review [22] that addressed the use of rTMS compared to sham treatment or other conventional treatment for improving function after stroke revealed that rTMS treatment was not associated with improved activities of daily living, nor did it have a statistically significant effect on motor function.
Three additional Cochrane reviews also discussed the effects of both tDCS [48] and rTMS [13,66] but in different populations-in Parkinsonism [48], in patients with amyotrophic lateral sclerosis or motor neuron disease [13], and in adults with panic disorder [66]. All reviews suffered from poor methodological quality, imprecision, and hence low confidence in the estimate of the true effect to draw a consistent conclusion on the effects of noninvasive brain stimulations.

Implications
. Moderate-quality evidence shows that rTMS, at 1 Hz, is more efficacious than sham for unilateral spatial neglect after stroke measured by Motor-Free Visual Perception Test. Furthermore, low-quality evidence also suggests a benefit of noninvasive brain stimulation, particularly with the use of rTMS, for overall USN measured by the line bisection test, Albert's test, and the line crossing test. Future trials should adhere to CONSORT guidelines to ensure clarity and reproducibility in the reporting of methods [67].

Disclosure
The funding agencies played no role in conducting the research or preparing the manuscript.