Secondary Logo

Journal Logo

Original Research Articles

Noninvasive Brain Stimulation Does Not Improve Neuropathic Pain in Individuals With Spinal Cord Injury

Evidence From a Meta-Analysis of 11 Randomized Controlled Trials

Yu, Binbin MD; Qiu, Huaide MD; Li, Jiahui MD; Zhong, Caiyun PhD; Li, Jianan MD

Author Information
American Journal of Physical Medicine & Rehabilitation: September 2020 - Volume 99 - Issue 9 - p 811-820
doi: 10.1097/PHM.0000000000001421


What Is Known

  • Noninvasive brain stimulation has been increasingly adopted in many rehabilitation facilities for treating neuropathic pain (NP) in individuals with spinal cord injury (SCI), yet its effects remain controversial.

What Is New

  • With a meta-analysis of 11 randomized controlled trials, we found that noninvasive brain stimulation had no significant effect on pain reduction in individuals with post-SCI NP, but cranial electrotherapy stimulation might be useful in the management of anxiety in these individuals. These findings do not support the routine use of noninvasive brain stimulation for NP in individuals with SCI.

Spinal cord injury (SCI) is a devastating neurological condition, resulting in many physiologic changes including loss of motor function and sensation, changes in sexual function, as well as development of spasticity. According to the World Health Organization 2013 report, approximately 250,000–500,000 people worldwide experience SCI annually due to accidents, falls, violence, sports, and other causes.1 In addition to the severe sensory and motor function loss below the injury plane, SCI can also be associated with chronic neuropathic pain (NP).2 More than 40% of individuals with SCI develop severe NP.3 Neuropathic pain is characterized by pain at or below the level of injury, which may be associated with allodynia or hyperalgesia.4–6 It can negatively affect the performance of daily activities, sleep, mood, and lead to social communication barriers and poor quality of life.7

Currently, the most common treatments for SCI-related NP are pharmacological, such as antidepressants, anticonvulsants, and opioids.8 However, the pharmacological effect remains unsatisfactory over the long-term because of the refractoriness of NP.9 In addition, consistent administration of medication leads to many adverse effects, such as constipation, toxicity, and increased risk of addiction or abuse. Under such circumstance, nonpharmacological strategies, especially noninvasive brain stimulation (NIBS) including repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), cranial electrotherapy stimulation (CES), and theta burst stimulation (TBS) have been increasingly adopted to treat NP in individuals with SCI. There have been few systematic reviews and meta-analyses examining the effects of specific NIBS in treating post-SCI NP, and their results have been inconclusive. As reported in a previous systematic review,10 significant pain reduction was observed in two of three studies on tDCS for post-SCI NP, although this study was limited by lack of quantitative analysis. A meta-analysis by Mehta et al.11 showed significant effects of tDCS in reducing NP after SCI posttreatment but the effect was not maintained at follow-up. The conclusions of this study are limited by the fact that a fixed effect model was applied, whereas the forest plot examining effects of tDCS on pain intensity showed significant heterogeneity. According to the Cochrane Handbook 5.10,12 a fixed effect model is inappropriate when significant heterogeneity exists as it does not account for between-study variability. Hence, it remains elusive whether tDCS is effective for post-SCI NP.

Besides, there has been no meta-analysis synthesizing evidence on the effects of overall NIBS on post-SCI NP. Therefore, the aim of our meta-analysis was to examine the effectiveness of NIBS in the treatment of NP among individuals with SCI.


Data Sources

The present review was performed and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (see Supplemental Checklist, Supplemental Digital Content 1, The protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO), CRD42019128161. Relevant articles published up to January 31, 2019, were searched in the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (PubMed), Embase (OvidSP), PsycINFO (OvidSP), and Physiotherapy Evidence Database (PEDro). The following search terms were used for NIBS: repetitive transcranial magnetic stimulation, transcranial magnetic stimulation, cranial electrotherapy stimulation, transcranial direct current stimulation, and theta burst stimulation. Other keywords included spinal cord injury and pain. We combined multiple terms on NIBS with “OR” and then combined other keywords with “AND.” All potentially eligible studies will be considered with no language limitation.

Study Selection

Studies were included if they met the following criteria: (1) randomized controlled trials (RCTs) comparing noninvasive brain stimulation with sham stimulation for NP in SCI; (2) NP or central pain in SCI had to be clearly defined; and (3) studies had to report pain intensity as an outcome at baseline and end of treatment. Studies were excluded if they were (1) studies involving participants without a clear diagnosis of NP in SCI13 and (2) studies without accessible data and contacts of the authors. All the references of the articles were evaluated for eligibility as well.

Quality Assessment of Selected Studies

Study quality was assessed using the PEDro scoring system, which was designed to measure the methodological quality of randomized controlled trials.14 The PEDro scale is an 11-item scale with a maximum score of 10. If a study was rated at least 4, it would be considered as moderate to high quality.15 Quality assessment for each study was done by two independent reviewers, and any disagreement was resolved upon discussion involving a third reviewer.

Data Extraction and Analysis

We extracted the following data from the selected studies: study design, intervention, number of participants, characteristics of participants, outcome measures (mean, standard deviation) for pain intensity, as well as depression and anxiety of treatment group and control group. Two independent reviewers searched for articles relevant to this meta-analysis and extracted information from the studies independently. A third reviewer participated in discussions and made decisions on any discrepancies. The reviewers were among the authors, whereas the author in charge of statistical analysis did not participate in the literature screening to avoid manipulation of data.

Analyses were conducted for the outcomes at posttreatment and follow-up. I2 statistic was used to measure heterogeneity between included studies, and the value of I2 > 50% indicates significant statistical heterogeneity.16 To adequately account for the additional uncertainty associated with variability between studies, a random effect model was applied. We used the standardized mean difference (SMD) to report treatment effect as there were different assessment scales for pain intensity, depression, and anxiety.12 A predetermined sensitivity analysis was performed by excluding studies one by one to explore the effect of each selected study on overall estimates.17 If the pooled effect remained unaltered after each study was removed respectively, the results will be considered relatively credible. Begg’s test,18,19 a regular statistical test for detecting publication bias, was used to identify the publication bias. All P values were calculated using a two-sided test and P < 0.05 was considered statistically significant. Data analysis was performed using STATA 12.0 (StataCorp LP, College Station, TX).


A total of 318 potentially eligible studies were identified, of which 11 studies met the inclusion criteria, including eight RCTs20–27 and three cross-over RCTs.28–30 The flowchart of study selection is presented in Figure 1. All selected studies compared an active NIBS group with a sham group. The intervention of these studies included rTMS (four trials),20,22,27,28 tDCS (six trials),21,23,24,26,29,30 and CES (one trials).25 The characteristics of selected studies concerning design, participants, intervention, and outcome are presented in Table 1. Soler et al.26 randomized subjects into four groups: tDCS, tDCS plus visual illusion, visual illusion only, and sham tDCS, and we only extracted the data from the tDCS group and sham tDCS group. There were different washout periods for cross-over trials, 12 wks in Kang et al.,28 4 wks in Wrigley et al.,29 and 1 wk in Ngernyam et al.30 Sample sizes ranged from 10 to 105 participants, and males predominated in them. The average age ranged from 35.7 to 56.1 yrs. Details of treatment protocol are showed in Table 1. All of the included studies on rTMS used a figure-of-eight coil on the head with different stimulation parameters and duration, whereas the six tDCS studies, except Thibaut et al.,23 applied the same positioning of stimulation and parameters. All included studies measured pain intensity as the primary outcome, and five studies measured depression and/or anxiety as secondary outcomes.20–22,25,29

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2009 flow diagram.
Profiles of included studies

The pooled analysis demonstrated no significant effect of rTMS, tDCS, or CES on pain reduction after treatment in individuals with NP after SCI (Fig. 2A). Four rTMS studies20,22,27,28 used different scales to measure pain intensity, three20,22,27 used visual analog scale (VAS) and the other28 applied numeric rating scale (NRS). No significant effect was detected in the effect of rTMS versus sham groups on pain intensity immediately after treatment (SMD = −0.03; 95% confidence interval [CI] = −1.45 to 1.39). Likewise, data synthesis of six tDCS studies21,23,24,26,29,30 showed no significant difference in pain reduction (SMD = −0.37; 95% CI = −0.84 to 0.10). Meanwhile, only one study25 examined the effect of CES versus sham group, and the result indicated no significant effect (SMD = 0.00; 95% CI = −0.39 to 0.39). Similarly, no beneficial effect was found in the effect of NIBS on pain reduction at follow-up (Fig. 2B).

Forest plot of SMD in pain intensity for comparing NIBS versus control in subjects with post-SCI NP immediately after stimulation (A) and follow-up (B).

Our data in Figure 3 showed no heterogeneity with I2 of 0% in each pooled analysis, which means the effects of different NIBS on depression and anxiety are statistically consistent. Therefore, we combined all forms of NIBS in the data synthesis while labeling the type of each study for subgroup analysis (Fig. 3). As a result, NIBS showed no significant effect on depression after treatment (SMD = 0.21; 95% CI = −0.10 to 0.52, Fig. 3A), but it revealed a significant effect on anxiety levels after treatment (SMD = 0.47; 95% CI = 0.12 to 0.82; Fig. 3A). Subgroup analysis suggested that only CES showed significant improvement of anxiety levels immediately after treatment (SMD = 0.45; 95% CI = 0.06 to 0.84), whereas rTMS and tDCS did not yield positive results. In addition, no significant difference was observed on depression (SMD = 0.62; 95% CI = −0.22 to 1.47; Fig. 3B) and anxiety (SMD = 0.44; 95% CI = −0.32 to 1.20; Fig. 3B) at follow-up.

Forest plot of SMD in depression and anxiety for comparing NIBS versus control in subjects with post-SCI NP immediately after stimulation (A) and follow-up (B).

As shown in Figure 4, the point estimates after excluding one study at a time were unanimously within the pooled effect of all studies, indicating that these data are relatively credible. The results of Begg’s test (Fig. 5) indicates a low risk of publication bias (P > 0.1). Individual P value for each Begg’s test was presented in Table 2.

Sensitivity analysis of each included study. A, Pain intensity immediately after stimulation. B, Pain intensity at follow-up. C, Depression and anxiety immediately after stimulation. D, Depression and anxiety at follow-up.
Begg’s funnel plots of the publication bias. A, Pain intensity immediately after stimulation. B, Pain intensity at follow-up. C, Depression and anxiety immediately after stimulation. D, Depression and anxiety at follow-up.
Results of Begg's test for each meta-analysis


Results of our study indicate that three types of NIBS (rTMS, tDCS, and CES) have no significant effects over sham stimulation on post-SCI NP reduction. Meanwhile, the current meta-analysis showed that showed that overall NIBS may not improve the levels of depression in these individuals. However, CES might be beneficial for improving anxiety immediately after treatment. These results challenged the routine use of NIBS in post-SCI NP.

In a previous meta-analysis of three trials,31 no benefits of rTMS over sham control was observed in treating chronic pain after SCI. An updated meta-analysis published in 201732 showed that rTMS had a beneficial effect on pain reduction over sham stimulation, but the results were not statistically significant. Likewise, our meta-analysis of four selected RCTs concluded that rTMS may not help reduce NP after SCI. Among the selected studies, only one study showed a significant effect of rTMS versus sham control after treatment,22 whereas greater pain reduction was observed in the sham stimulation group in the study by Defrin et al.20 Despite discrepancies across individual studies, the results of these two meta-analyses showed no favorable effect of rTMS on pain reduction in subjects with NP after SCI. Consistent with previous studies, our data do not support the routine use of rTMS for post-SCI NP in clinical settings.

The meta-analysis by Boldt et al.31 showed favorable effects of tDCS over sham stimulation in pain reduction at short- and mid-term (follow-up of 16–38 days after intervention), but the meta-analysis included only two studies.21,26 Mehta et al.11 included five studies21,24,26,29,30 and conducted an updated meta-analysis on the effects of tDCS. As a result, a moderate effect of tDCS was observed in reducing NP among individuals with SCI posttreatment, but not maintained at follow-up. However, the study used the fixed effect model without considering data variability across different studies, which compromised the reliability of the significant effect of tDCS. According to the Cochrane Handbook 5.10,12 a fixed effect model is only appropriate under the circumstance that no significant heterogeneity was present across studies. With a random effect model, the standard errors of the study-specific estimates can be adjusted to incorporate a measure of heterogeneity, resulting in a more conservative pooled effect.12 Given that Mehta et al.11 did not specify the I2 statistic, which represents heterogeneity in the forest plot, we extracted the data (SMD and CI) from the published figure examining the immediate effect of tDCS on pain reduction and reproduced the meta-analytic result in Stata 12.0. Then, we found that there was significant heterogeneity (I2 > 50%) among the data across included trials (Supplementary Fig. 1A, Supplemental Digital Content 2, Interestingly, no significant effect of tDCS was observed after we took the heterogeneity into consideration and applied a random effect model for recalibration (Supplementary Fig. 1B, Supplemental Digital Content 2, Transcranial direct current stimulation was not beneficial over control in treating NP after SCI. Consistent with previous data, our study showed that tDCS has no significant effect in reducing NP after SCI. Likewise, only one study25 explored the effectiveness of CES for NP in persons with SCI in this meta-analysis and the result had no significant effects of CES on pain reduction.

In addition, we investigated the effects of NIBS on depression and anxiety, which has been long thought to be closely related to chronic pain.33,34 We performed data synthesis with all forms of NIBS included because there was no heterogeneity detected (I2 = 0). Our consistent data indicated that different forms of NIBS do not improve depression, but it had a significant effect on anxiety posttreatment. Subgroup analysis revealed that only CES significantly improved the anxiety levels in individuals with post-SCI NP as compared with sham stimulation, whereas other forms of NIBS (rTMS and tDCS) did not. Although the CES study involved a total of 105 individuals, the positive effect of CES is subject to further validation. The mechanism by which CES reduced anxiety levels may be the result of the increased release of serotonin, endorphins, melatonin, and the concentration of γ-aminobutyric acid after stimulation.35,36

The present study has been the first meta-analysis investigating the effects of overall NIBS in the treatment of post-SCI NP. Within the present study, we also conducted updated meta-analyses for rTMS, tDCS, and CES, respectively. A random effect model was applied in our analysis to account for the heterogeneity across different studies. Furthermore, a sensitivity analysis that excluded each trial at a time did not alter the results. The results of Begg’s test indicated a low risk of publication bias among included studies with P values higher than 0.1. Therefore, our results were statistically robust.

However, there are some limitations in the current meta-analysis. In total, 11 studies were included but only a very small number was evaluated for each type of stimulation, limiting the statistical power of the results. Meanwhile, variability existed on stimulation parameters and treatment sessions, which may impact the pooled effect of NIBS to some extent. Participants in most of the included studies were predominantly male, so that the generalizability of the results across gender may be compromised. Furthermore, only studies that contained pain as an outcome were included, whereas those only examined anxiety and depression may have been missed.


In the current meta-analysis, no beneficial effect of NIBS was observed on pain relief in individuals with NP after SCI. Likewise, NIBS showed no beneficial effect in reducing depression levels, while it might be useful for short-term management of anxiety in these individuals. Among the three forms of NIBS, only CES yielded a significant immediate effect on the reduction of anxiety levels. These findings do not support the routine use of NIBS for NP in individuals with SCI.


1. World Health Organization: International Perspectives on Spinal Cord Injury[C]. Malta: © World Health Organization, 2013. Available at: Accessed February 15, 2016
2. Sezer N, Akkus S, Ugurlu FG: Chronic complications of spinal cord injury. World J Orthop 2015;6:24–33
3. Margot-Duclot A, Tournebise H, Ventura M, et al.: What are the risk factors of occurrence and chronicity of neuropathic pain in spinal cord injury patients? Ann Phys Rehabil Med 2009;52:111–23
4. Heutink M, Post MW, Luthart P, et al.: Long-term outcomes of a multidisciplinary cognitive behavioural programme for coping with chronic neuropathic spinal cord injury pain. J Rehabil Med 2014;46:540–5
5. Bryce TN, Biering-Sorensen F, Finnerup NB, et al.: International spinal cord injury pain classification: part I. Background and description. March 6-7, 2009. Spinal Cord 2012;50:413–7
6. Burke D, Fullen BM, Stokes D, et al.: Neuropathic pain prevalence following spinal cord injury: a systematic review and meta-analysis. Eur J Pain 2017;21:29–44
7. Kim JY, Choi GS, Cho YW, et al.: Attenuation of spinal cord injury-induced astroglial and microglial activation by repetitive transcranial magnetic stimulation in rats. J Korean Med Sci 2013;28:295–9
8. Dworkin RH, O’Connor AB, Backonja M, et al.: Pharmacologic management of neuropathic pain: evidence-based recommendations. Pain 2007;132:237–51
9. Siddall PJ, Loeser JD: Pain following spinal cord injury. Spinal Cord 2001;39:63–73
10. David M, Moraes AA, Costa MLD, et al.: Transcranial direct current stimulation in the modulation of neuropathic pain: a systematic review. Neurol Res 2018;40:555–63
11. Mehta S, McIntyre A, Guy S, et al.: Effectiveness of transcranial direct current stimulation for the management of neuropathic pain after spinal cord injury: a meta-analysis. Spinal Cord 2015;53:780–5
12. Higgins JPT, Green S (eds): Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available at: Accessed February 10, 2019
13. Treede RD, Jensen TS, Campbell JN, et al.: Neuropathic pain: redefinition and a grading system for clinical and research purposes. Neurology 2008;70:1630–5
14. Moseley AM, Herbert RD, Sherrington C, et al.: Evidence for physiotherapy practice: a survey of the Physiotherapy Evidence Database (PEDro). Aust J Physiother 2002;48:43–9
15. Foley NC, Bhogal SK, Teasell RW, et al.: Estimates of quality and reliability with the physiotherapy evidence-based database scale to assess the methodology of randomized controlled trials of pharmacological and nonpharmacological interventions. Phys Ther 2006;86:817–24
16. Higgins J, Thompson SG, Deeks JJ, et al.: Measuring inconsistency in meta-analyses. BMJ 2003;327:557–60
17. Copas J, Shi JQ: Meta-analysis, funnel plots and sensitivity analysis. Biostatistics 2000;1:247–62
18. Begg CB, Mazumdar M: Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088–101
19. Zhang Y, Cai J, An L, et al.: Does music therapy enhance behavioral and cognitive function in elderly dementia patients? A systematic review and meta-analysis. Ageing Res Rev 2017;35:1–11
20. Defrin R, Grunhaus L, Zamir D, et al.: The effect of a series of repetitive transcranial magnetic stimulations of the motor cortex on central pain after spinal cord injury. Arch Phys Med Rehabil 2007;88:1574–80
21. Fregni F, Boggio PS, Lima MC, et al.: A sham-controlled, phase II trial of transcranial direct current stimulation for the treatment of central pain in traumatic spinal cord injury. Pain 2006;122:197–209
22. Nardone R, Höller Y, Langthaler PB, et al.: RTMS of the prefrontal cortex has analgesic effects on neuropathic pain in subjects with spinal cord injury. Spinal Cord 2016;55:20–5
23. Thibaut A, Carvalho S, Morse LR, et al.: Delayed pain decrease following M1 tDCS in spinal cord injury: a randomized controlled clinical trial. Neurosci Lett 2017;658:19–26
24. Yoon EJ, Kim YK, Kim HR, et al.: Transcranial direct current stimulation to lessen neuropathic pain after spinal cord injury: a mechanistic PET study. Neurorehabil Neural Repair 2014;28:250–9
25. Tan G, Rintala DH, Jensen MP, et al.: Efficacy of cranial electrotherapy stimulation for neuropathic pain following spinal cord injury: a multi-site randomized controlled trial with a secondary 6-month open-label phase. J Spinal Cord Med 2011;34:285–96
26. Soler MD, Kumru H, Pelayo R, et al.: Effectiveness of transcranial direct current stimulation and visual illusion on neuropathic pain in spinal cord injury. Brain 2010;133:2565–77
27. Yílmaz B, Kesikburun S, Yaşar E, et al.: The effect of repetitive transcranial magnetic stimulation on refractory neuropathic pain in spinal cord injury. J Spinal Cord Med 2014;37:397–400
28. Kang BS, Shin HI, Bang MS: Effect of repetitive transcranial magnetic stimulation over the hand motor cortical area on central pain after spinal cord injury. Arch Phys Med Rehabil 2009;90:1766–71
29. Wrigley PJ, Gustin SM, McIndoe LN, et al.: Longstanding neuropathic pain after spinal cord injury is refractory to transcranial direct current stimulation: a randomized controlled trial. Pain 2013;154:2178–84
30. Ngernyam N, Jensen MP, Arayawichanon P, et al.: The effects of transcranial direct current stimulation in patients with neuropathic pain from spinal cord injury. Clin Neurophysiol 2015;126:382–90
31. Boldt I, Eriks-Hoogland I, Brinkhof MW, et al.: Non-pharmacological interventions for chronic pain in people with spinal cord injury. Cochrane Database Syst Rev 2014;Cd009177
32. Gao F, Chu H, Li J, et al.: Repetitive transcranial magnetic stimulation for pain after spinal cord injury: a systematic review and meta-analysis. J Neurosurg Sci 2017;61:514–22
33. Fishbain DA, Cutler BR, Rosomoff HL, et al.: Comorbidity between psychiatric disorders and chronic pain. Curr Rev Pain 1998;2:1–10
34. Zhuo M: Neural mechanisms underlying anxiety-chronic pain interactions. Trends Neurosci 2016;39:136–45
35. Kennerly R: QEEG analysis of cranial electrotherapy: a pilot study. J Neurother 2004;8:112–3
36. Smith RB: Microcurrent therapies: emerging theories of physiological information processing. NeuroRehabilitation 2002;17:3–7

Neuropathic Pain; Spinal Cord Injury; Noninvasive Brain Stimulation; Meta-analysis

Supplemental Digital Content

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.