The effects of anodal tDCS on pain reduction in people with knee osteoarthritis: A systematic review and meta-analysis

Objectives: To synthesise the literature on the ef ﬁ cacy of primary motor cortex anodal transcranial direct current stimulation (M1-a-tDCS), as a standalone or priming technique, for pain reduction in people with knee osteoarthritis (KOA). Methods: The systematic literature search was conducted in MEDLINE, CINAHL, Embase and CENTRAL according to PRISMA statement. Results: Fourteen studies involving 740 people with KOA were included. In the meta-analysis, six studies compared a-tDCS alone with sham stimulation, and ﬁ ve studies compared a-tDCS combined with other methods with sham stimulation. We found positive effect of a-tDCS alone on pain in KOA (standard mean difference (SMD) (cid:1) 0.52; 95% CI, (cid:1) 0.78 to (cid:1) 0.25; P=0.001; I2 = 69%). Further, a-tDCS with other treatments showed positive effect (SMD (cid:1) 1.23; 95% CI, (cid:1) 1.59 to (cid:1) 0.88; P < 0.001; I2 = 48%) on pain in people with KOA. This evidence showed low certainty due to a high risk of bias and imprecision.


Introduction
Osteoarthritis (OA) is the most common joint disease in older adults, affecting more than 500 million people worldwide [17].The number of people with OA has risen by 48% from 1990 to 2019 [22].The knee is the most frequently affected site, accounting for approximately 61% of the total cases in 2019.Globally, the average annual total cost, including direct and indirect costs per patient living with knee and hip OA is estimated to be US$ 11,700.00[35].Due to the progressive nature of knee OA [8], early identification and management are essential to mitigate the burden of this condition.
Knee OA has no cure, and the management methods focus primarily on reducing pain and improving function.Conservative pain management methods largely comprise pharmacological interventions, including local and systemic analgesia.These pharmacological interventions have shown improvements in knee pain; however, they often carry side effects, including fatigue, sore throat, abdominal pain and heartburn that outweigh their benefits [29,33,38,40].As a result, there is growing research interest in novel non-pharmacological pain management approaches for people with knee OA.
Transcranial direct current stimulation (tDCS) is a noninvasive and non-pharmacological approach with minimal side effects.It has received considerable attention for pain management in people with knee OA.This technique involves the delivery of a weak current (1-2 mA) via two surface electrodes (anode and cathode) placed on different areas of the brain [20].The current enters the brain and modulates transmembrane neuronal potentials in the brain regions below the electrodes, leading to changes in neuronal excitability (Chib et al., 2013).tDCS is applied using one of the following two types of electrode configurations.Cathodal stimulation (c-tDCS) hyperpolarises neurons, thereby diminishing cortical excitability.In contrast, anodal stimulation (a-tDCS) causes neuronal depolarisation and increases the cortical excitability [6,24,27].The pain modulation action of tDCS is thought to be related to a complex interplay of different mechanisms, including modulation of cortical excitability, activation of endogenous pain control systems, modulation of neurotransmitters and inducing neural plasticity [25,39].
To date, the capability of a-tDCS to modulate pain in knee OA has been explored in numerous studies, although results have been mixed [4,9,23,31,44].Several systematic reviews and meta-analyses have concluded that standalone or adjunct a-tDCS applied to the primary motor cortex significantly improved knee pain and function [12,26,45].However, the analyses were limited to a small number of studies or restricted only to randomised controlled trials (RCTs).A substantial number of additional studies on the effect of a-tDCS have since been published, and no study has provided a comprehensive meta-analytic review that includes all experimental studies.On these grounds, a carefully constructed systematic review and meta-analysis can provide a complete picture of the evidence base for the effect of a-tDCS on pain in knee OA.Furthermore, a comprehensive review may help to understand the real-world effectiveness of a-tDCS on knee pain management and help inform clinical practice and future research directions.Therefore, the primary aim of this review was to synthesise and critically evaluate the most up-to-date literature reporting on the efficacy of a-tDCS as a standalone or adjunct therapy for pain management in people with knee OA.

Protocol registration
We used the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines for conducting and reporting this systematic review [30,37].The protocol for this systematic review and meta-analysis was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO, Registration number: CRD42021255114).

Deviations from protocol
Our protocol stated that this review would primarily carry out subgroup analysis based on the stimulation parameters (current intensity and duration).However, to aid with interpretation, we conducted subgroup analysis based on the intervention time points, including short-term (0 to < 1 week post intervention), medium-term ( 1 to 6 weeks post intervention), and long-term ( 6 weeks post intervention) as per O'Connell et al. [28].

Search strategy
The electronic databases of MEDLINE, CINAHL, Embase and CENTRAL (Cochrane Central Registered of Control Trials) were searched from inception through February 2023.We used the following search terms: "Knee" OR "Knee osteoarthritis" OR "Osteoarthritis" AND "transcranial direct current stimulation" OR "tDCS" OR "anodal transcranial direct current stimulation" OR "cathodal transcranial direct current stimulation" OR "Non-invasive brain stimulation" (Supplementary file 1).These terms were used in various combinations and adapted for each database, as necessary.

Outcome measure
The outcome of interest was pain intensity scales post-intervention, assessed by visual analogue scale, numerical rating scale, Western Ontario and McMaster Universities Arthritis Index (WOMAC_pain) or Knee Injury and Osteoarthritis Outcome Score (KOSS_pain).

Selection criteria
A study was included if it: 1) included participants who had knee OA for > 3 months; 2) used the a-tDCS alone or in combination with other non-pharmacological interventions; 3) reported pain intensity at baseline and post-intervention, and 4) was either an RCT or used a non-randomised design.A study was excluded if it: 1) involved participants who had undergone total knee arthroplasty, 2) focused on other types of osteoarthritis conditions such as hip and hand 3) was published in a language other than English, or 4) was a review, thesis, conference poster abstract, letter to the editor, qualitative paper or commentary.
Titles and abstracts of retrieved studies were initially screened by two reviewers (TDD and PN) to identify potentially relevant studies for the review.The full text of the selected studies was then independently reviewed by two reviewers (TDD and PN) to identify eligible studies.Disagreements were resolved by consensus, or by a third author (ATH) when agreement could not be reached.

Assessment of risk of bias and quality of evidence
Two review authors (KM and PN) independently assessed the risk of bias in the included studies.The risk of bias in RCTs was assessed according to the Cochrane risk of bias tool (RoB 2.0).This tool covers five domains, including the randomisation process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results [42].The risk of bias in nonrandomised studies was assessed using the Risk of Bias in Non-randomised Studies -of Interventions (ROBINs 1.0).The ROBINs 1.0 assesses bias across seven domains (confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement outcome, and selection of the reported results) [41].Disagreements were resolved by a third reviewer (TDD).The quality of evidence at the outcome level was assessed using GRADE methodology.

Data extraction
Data were extracted using an excel form by two independent authors (TDD and PN).Any disagreements were resolved by an additional author (KM).The data extraction sheet was developed for descriptive information, including authors, year of publication, country, study design, sample size, percentage of females, mean age, description of the intervention (intensity, duration, electrode size and placement), pain scale, time of outcome measurement and the main results related to pain outcome.Data were extracted from each available time point.
Post-intervention pain intensity was collected as mean § standard deviation [SD].When the pain was measured using different scales, the outcomes were prioritised based on the hierarchy proposed by Juhl et al. [19] (Supplementary file 2).The preferred pain outcome measure was the pain scale of the WOMAC; followed by the visual analogue scale (VAS); and then any other pain measures.We used Plot digitiser software [18] to obtain the mean and SD for studies reporting these estimates graphically rather than numerically.This Java-based software converts the plotted values into a numerical format.When the standard error (SE) was reported, the SD was estimated using the formula SD = SE € On (n = number of subjects in each group) [15].If numerical data were not reported and could not be extracted from graphs, an email request for information was sent to the corresponding author.Study results were synthesised narratively when pooling data was not possible.Based on a previous review by O'Connell et al. [28], the pain outcome was categorised according to three time groups: i) short-term (< 1 week post intervention); ii) medium-term ( 1 to 6 weeks post intervention); and iii) long-term ( 6 weeks post intervention).Where more than one data point was available for short-term outcomes, the first post stimulation measure was used, and where multiple treatments were given, the first outcome at the end of the treatment period was considered.For medium-term outcomes where more than one data point was available, the midpoint was used.When more than one data point was available for long-term outcomes, the longest point was used.The WOMAC knee pain subscale scores were normalised to a scale from 0 to 100 to provide a measure comparable with the VAS scale.

Meta-analysis
Results were pooled using RevMan 5 software (version 5.2) (RevMan, 2012).We assumed that clinical and methodological heterogeneity was likely to exist in the included studies, and so we used a random effects meta-analysis [7] to account for differences in outcome measures between trials, we used standardised mean difference to measure the effect size.The minimum clinically important difference threshold of 15% for pain was used in this study [28].
To evaluate the robustness of the results, leaveÀoneÀout analyses were performed by excluding each study in turn to expose the potential influence of the individual studies on the pooled estimate.The funnel plot was not developed since the meta-analysis included less than ten studies [16].

Description of studies
A total of 431 studies were retrieved, of which 14 (n = 740) met eligibility criteria, and eight were included in the metaanalysis.The other six studies were not included in the meta-analysis for the following reasons: i) the necessary data were not available in the study report or upon author request by the submission date of this review (n = 1); ii) they contained pre-post study designs without a control group (n = 2); iii) they used another treatment as the control (n = 1); or iv) they assessed experimental pain (n = 2).These studies were instead synthesised narratively (Fig. 1).The excluded studies and the reason for exclusion, are available in the supporting information (Supplementary file 3).

Study characteristics
A total of 740 participants with knee OA were included in the 14 eligible studies (Table 1).Sample sizes ranged from 10 to 120.Among the included studies, twelve were RCTs, and two were quasi-experimental pre-post studies.All studies used a-tDCS using a similar electrode montage, with an anode over the M1 of the hemisphere contralateral to the affected knee, and a cathode on the supraorbital area ipsilateral to the affected knee.Eligible studies applied a-tDCS alone (n = 6) or in combination with exercise (n = 2), transcutaneous electrical nerve stimulation (TENS; n = 3), mindfulness-based meditation (MBM; n = 1), intramuscular electrical stimulation (EIMS; n = 1), or physiotherapy, including a combination of TENS, ultrasound, patellofemoral and tibiofemoral mobilisation (grade 1 and 2) and exercise therapy (PT; n = 1).The selected studies used various stimulation parameters in terms of stimulation intensity, electrode size and the number of sessions.All the studies stimulated M1 for 20 min, and all the studies applied a-tDCS for multiple sessions ranging from 5 to 20.Thirteen studies assessed the effect of a-tDCS on clinical pain, and one focused on both clinical and experimental pain.Eligible studies were conducted in the United States of America (n = 6) [1][2][3][4]23,31], Iran (n = 3) [5,32,34], India (n = 2) [10,11], Brazil (n = 2) [13,44], and Australia (n = 1) [9].The mean age of participants' age ranged from 54.8 years to 74.8 years.While one study recruited only females [13], all the other studies (n = 13) included both males and females.

Effect of a-tDCS alone on knee pain
Six RCTs were included and the meta-analysis (n = 361) showed strong evidence that active a-tDCS reduced pain in patients with knee OA compared to sham a-tDCS (SMD -0.52, 95% confidence interval (95% CI) -0.78, -0.25, P = 0.001) and substantial heterogeneity (I2 = 69%) (Fig. 2).The SMD was back-transformed to a mean difference using the mean standard deviation of the post treatment sham group scores of the studies included in this analysis (2.09).This was used to estimate the real percentage change on a 0 to 10 pain intensity scale of active stimulation compared with the mean post stimulation score from the sham groups of the included studies (4.01).This equated to a reduction of points 1.08 (95% CI 0.52 to 1.63), or a percentage change of 26% (95% CI 12% to 40%) of the control group outcome.This estimate reached the pre-established criteria for a minimal clinically important difference ( 15%).Further, the leave-one-out analysis did not alter the results of this analysis.

Effect of a-tDCS with other treatments on knee pain
Five RCTs were included, and the overall meta-analysis found strong evidence that active a-tDCS reduced pain in people with knee OA (SMD -1.23, 95% CI -1.59, -0.88, P<0.001).There was, however, also evidence for low heterogeneity (I2 = 48%) (Fig. 3).The SMD was back-transformed to a mean difference, using the mean standard deviation of the post-treatment sham group scores of the studies included in this analysis (1.31).This was used to estimate the real percentage change on a 0 to 10 pain intensity scale of active stimulation compared with the mean poststimulation score from the sham groups of the included studies (4.41).Compared to the control group, a-tDCS reduced knee pain by 1.61 points (95% CI 1.15 to 2.08), and equivalent to a percentage change of 36% (95% CI 26% to 47%).This estimate reaches and indeed exceeds the pre-established criteria for a minimal clinically important difference ( 15%).

Medium-term effect ( 1 to 6 weeks post intervention)
Three studies provided data on the medium-term effect of a-tDCS (n = 93).Pooled analysis found strong evidence that a-tDCS reduced knee pain at 1-6 weeks post intervention (SMD -1.60, 95% CI -2.08, -1.13, P<0.001, I2 = 0%).The leave-one-out sensitivity analysis indicated that the pooled estimates were not dependent on any single study (Supplemental file 8).Long-term effect ( 6 weeks post intervention) Only two studies assessed the long-term effect of a-tDCS on knee pain (Table 4).Pooling these studies found strong evidence that active a-tDCS reduced knee pain (SMD -1.79, 95% CI -2.81, -0.77, P = 0.001).The leave-one-out sensitivity analysis indicated that the pooled estimates were not dependent on any single study (Supplemental file 9).
Fig. 3 Effect of active transcranial direct current stimulation as an adjunct technique on pain in people with knee osteoarthritis.

Narrative synthesis
Four RCTs [1,3,4,34] and two pre-post studies [2,31] assessed the effect of a-tDCS on clinical pain in knee OA.Additionally, one RCT [1] investigated the effect of active a-tDCS on clinical and experimental pain.Similar to the findings of the meta-analysis, narrative synthesis showed evidence that a-tDCS alone or with other treatments reduced knee pain.One of the RCTs was conducted in adults with knee OA (n = 40) [4].This study investigated whether five consecutive days of active a-tDCS for 20 minutes would reduce pain compared to sham tDCS.The results of this double-blinded, randomised controlled pilot study showed no evidence for an effect on pain at the end of the fifth day of treatment compared to sham tDCS.
The same research group investigated the effect of 10 sessions of home-based a-tDCS combined with MBM on pain in knee OA people in a double-blinded, randomised, shamcontrolled study (n = 30).The results indicated that the effect of active a-tDCS combined with active mindfulnessbased meditation reduced pain scores measured using the WOMAC questionnaire.
Sajadi et al. [34] found a contrasting result.In a doubleblinded randomised controlled study (n = 40), the authors explored the effect of six sessions of active a-tDCS in combination with exercise on knee pain compared to TENS with exercise.Although there was evidence for a within group improvement, no difference was found between groups.
Among the two studies using pre-post designs, the first study (n = 20) to apply active a-tDCS over the M1 to improve knee pain was conducted by Ahn et al. [2].This single group pre-post study comprised of 10 sessions of active a-tDCS and found improvement in pain at the end of treatment compared to baseline.In addition, another pre-post study involving people with knee OA (n = 10) using the same experimental design found a reduction in pain post-intervention compared to baseline [31].
Ahn et al. [1] used a double-blinded RCT (n = 40) to investigate the effect of a-tDCS on clinical and experimental pain compared to sham tDCS.The results showed a difference in clinical pain between groups for numeric rating scale scores but not WOMAC scores.While the experimental pain sensitivity was reduced with active a-tDCS, this study showed an association between experimental pain changes and clinical pain reduction.

Risk of bias and quality assessment
Of the included studies, 42.9% (6/14) are at low risk of bias, 50% (7/14) are at unclear risk of bias, and 7.1% (1/14) are at high risk of bias (Tables 2 and 3).In addition, the GRADE ratings revealed that the quality of evidence in the meta-analysis of the effect of a-tDCS alone and with other treatments on pain intensity was low.The evidence was downgraded owing to the proportion of data from the studies at high risk of bias and imprecision due to the small sample size of the included studies (Table 4).

Discussion
This is the first comprehensive systematic review and metaanalysis evaluating the effects of active a-tDCS, either as a standalone or adjunct treatment, on knee pain in people with knee OA.Despite the low certainty of the evidence, the pooled analysis of active a-tDCS as standalone or with other treatment methods revealed a reduction in pain compared to sham a-tDCS, which was higher than the preestablished minimal clinically important difference.The additional narrative synthesis of studies not eligible for the meta-analysis showed that active a-tDCS alone or in combination with other interventions, including TENS, exercise, or mindfulness-based meditation, improved pain in people with knee OA.Overall, the evidence suggests a positive effect of active a-tDCS as a standalone and an adjunct therapy on knee pain management, albeit with low certainty of evidence.
The meta-analysis showed strong evidence of the shortterm (end of intervention or <1 week), but not medium (>1 week to <6 weeks) or long term (6 weeks) effect of active a-tDCS as a standalone treatment for knee pain.However, the narrative synthesis showed some evidence of a positive effect.This finding is consistent with a recent review by Chaturvedi et al. [12], which demonstrated a strong effect of a-tDCS alone or with other treatments compared to sham treatment on knee pain reduction.It is worth noting that their review included only two studies.Evidence supporting the reduction in pain and the longevity of its effects primarily comes from studies that administered active a-tDCS over a range of 5 to 20 sessions, each lasting 20 minutes.While some studies demonstrate a lasting effect with a higher number of sessions, others have failed to do so.This observation highlights the potential cumulative impact achievable through the application of multiple sessions of active a-tDCS.However, the available evidence does not provide concrete support for establishing a standard number of sessions for a long-lasting effect.
The present meta-analyses highlighted the potential short-term (end of intervention or <1 week), medium (>1 week to <6 weeks) and long-term (6 weeks) efficacy of a-tDCS combined with other treatments for improving pain in people with knee OA compared to sham stimulation or other interventions.The studies that employed active a-tDCS in combination with other interventions used a real sham, while some utilized sham tDCS in conjunction with another intervention.Additionally, the majority of studies assessed the effect of individual treatments on pain intensity.Interestingly, when these treatments were administered individually, they showed a significant effect compared to sham interventions.However, when combined with active a-tDCS, the effect was significantly enhanced, indicating the additional value of adding active a-tDCS to the other treatment regimen.
A more recent meta-analysis (n=2 studies) revealed a positive effect of a-tDCS in combination with peripheral stimulation, on pain reduction in people with knee OA [26].The current review's findings are also in line with a meta-analysis by Teixeira et al., [45] (n=2 studies), which showed a positive effect of a-tDCS combined with physical therapy on pain in people with knee OA.However, our work has important implications for clinical practice due to the pooled results of a comparatively higher number of studies (n=6 vs n=2 of previous reviews) revealing a clinically important difference in pain reduction.This is consistent with a Cochrane systematic review and meta-analysis of non-invasive brain stimulation on chronic pain not limited to OA, which highlighted the strong and clinically important reduction of pain with a-tDCS [28].Additionally, the present study extends findings from these earlier reviews by including all published experimental studies investigating the effect of a-tDCS on knee OA.We also included a narrative synthesis when the original data could not be extracted from published graphs, or the authors did not provide the raw data.Our methods collectively provide a comprehensive overview of the relationship between active a-tDCS as an adjunct therapy in reducing pain in people with knee OA.
Although the precise underlying mechanisms that may help explain the effect of a-tDCS as a standalone or adjunct treatment for knee OA pain are still not completely understood, various hypotheses exist.The effects of a-tDCS as a stand-alone technique could be explained based on the electrode montage.All of the included studies in this review used a cephalic montage, with active electrode over M1 and the return electrode over contralateral supra orbital area.It has been suggested that this montage modifies the intracortical motor connectivity and thalamic activity, which ultimately modulates descending pain processing pathways [14].Indeed, the effect of a-tDCS combined with other therapies can be attributed to both priming and summative effects.The application of active a-tDCS may prime the associated brain area, rendering it more responsive to subsequent intervention.Additionally, a-tDCS application can modulate the top-down mechanism [21], while other treatments, such as peripheral nerve stimulation [36] may activate bottom-up mechanisms.This synergistic or combined process eventuates in corticospinal changes with inhibition of pain decedent pathways [13,43].
The findings of the current meta-analysis should be interpreted in the context of the following limitations.First, the *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).CI: confidence interval; SMD: standardised mean difference.GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
Explanations a.The mixture of unclear and high RoB studies.b.Small size.12 T. Dissanayaka, P. Nakandala, K. Malwanage et al.
pooled studies were limited to peer-reviewed English language articles, possibly leading to a smaller number of studies to pool in the analysis.Furthermore, the pooled results are limited by low certainty of evidence as determined by the GRADE framework and the results of this review should be interpreted with caution.Moreover, the publication bias assessment using a funnel plot was possible due to less than ten studies included in the meta-analysis.Further, the leave-one-out analyses found that some pooled estimates were sensitive to individual studies, suggesting that further studies are required to improve the robustness of the estimated effect.Considering these, carefully planned RCTs studies with larger sample sizes need to be conducted to better assess and have greater confidence in the efficacy of a-tDCS on knee OA pain.

Conclusion
Overall, the results of the present systematic review and meta-analysis showed that a-tDCS as is effective as a standalone treatment on pain in people with knee OA.Despite the low certainty of the evidence, the results were positive in the short-, medium-and long-term effects of a-tDCS as an adjunct treatment in improving knee pain.Future studies should consider larger sample sizes to confirm the findings of this meta-analysis.

Fig. 1
Fig. 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Fig. 2
Fig. 2 Effect of active transcranial direct current stimulation as a standalone technique on pain in people with knee osteoarthritis.

Table 1
Characterises of the included studies.

Table 2
Risk of bias of included randomised controlled studies.

Table 3
Risk of included non-randomised interventional studies.

Table 4
Summary of findings.Patient or population: pain in knee osteoarthritis Setting: various rehabilitation, orthopaedic, or rheumatology clinics Intervention: Transcranial direct current stimulation Comparison: sham/no/other treatments