Effects of transcranial direct current stimulation on pain and function in patients with knee osteoarthritis: A systematic review and meta‐analysis

Abstract

Objective

The aim of this systematic review and meta‐analysis is to determine the effects of transcranial direct current stimulation (TDCS) on pain (considering the different types of pain measured) and function in adults with knee osteoarthritis (OA).

Literature Survey

The Cochrane Library, MEDLINE, Web of Science, Embase, Scopus, Physiotherapy Evidence Database, Literatura Latinoamericana y del Caribe en Ciencias de la Salud, and Science Direct were searched from inception to December 2023. Reference lists of selected trials and previously published and cited reviews were also examined. Results were restricted to randomized controlled trials on adults with knee OA, with TDCS as the primary intervention. Studies that included other knee issues or that compared TDCS with other therapies without a control group were excluded.

Methodology

Study characteristics, such as author, year of publication, study design, intervention description, sample characteristics and outcomes, were described. The Cochrane risk‐of‐bias tool and the Grading of Recommendations Assessment, Development, and Evaluation system were used to assess the individual and global risk of bias and quality of evidence. To analyze the effects of the interventions on clinical outcomes, a meta‐analysis was performed. The outcomes included were self‐reported pain and function, pain pressure threshold (PPT), and conditioned pain measure (CPM). Intervention comparisons were grouped as “active TDCS” and “sham TDCS”.

Synthesis

A total of 1045 participants completed the interventions. The meta‐analysis demonstrated significant improvement in self‐reported pain intensity measured with the visual analogue scale or numerical rating scale (Cohen's d = −1.05 [95% CI, −1.40 to −0.70]), pain measured according to PPT (Cohen's d = −0.77 [95% CI, −1.45 to −0.09]) and function (Cohen's d = −0.29 [95% CI, −0.54 to −0.04]) with active TDCS as compared with sham TDCS. There was no difference in CPM between interventions.

Conclusions

TDCS intervention alone or in combination with other techniques improves self‐reported pain intensity (as measured by visual analogue scale, numerical rating scale, or PPT) and function in knee OA but has no impact on the endogenous pain inhibition system as measured with CPM.

INTRODUCTION

Osteoarthritis (OA) is a degenerative chronic disease in which the dynamic equilibrium between the formation and breakdown of the cartilaginous matrix is disrupted, and the metabolic activity is unable to compensate for the deterioration. ¹ Although its prevalence varies across studies, it is one of the major conditions reported in older populations. ² , ³ Michael et al. ¹ reported that 6% of adults had radiologic knee OA, rising to a prevalence of 40% of adults over 70 years old. However, not all the adults presenting radiologic signs of OA experience symptoms. ¹

Individuals who experience these symptoms often report pain with movement—especially when they have not warmed up—most commonly at the start of activity or during the first few steps of walking. ⁴ But as OA progresses, pain may appear even without movement of the joint. ⁵ However, people with OA do not always display the same pattern. ⁶ There are patients with extensive radiologic evidence of OA who do not experience any pain. ⁶ In this regard, recent evidence suggests that pain intensity in knee OA is a complex process that is not completely associated with the severity of the damage. ⁷ , ⁸ This may be partially explained by altered sensory processing in the central nervous system due to the chronicity of the pain syndrome. ⁷

Because a cure is not yet available for OA, the primary goals of treatment are to slow disease progression and relieve symptoms. ⁵ , ⁹ A variety of interventions have been studied to achieve these goals, including physical therapy focused on therapeutic exercise (land‐ or water‐based), ¹⁰ , ¹¹ transcutaneous electrical nerve stimulation, ¹² weight reduction, ⁹ and opioid‐based pharmacotherapy. ¹³ Among these, the Osteoarthritis Research Society International guidelines specifically recommend structured land‐based exercise, mind–body exercise, aquatic exercise, and the use of gait aids. ¹⁴ Although therapeutic exercise can slow progression, its effectiveness is limited by factors such as poor adherence and interactions with opioid medications, particularly in older adults. ¹⁵ As a result, interventions that target central nervous system processing—including noninvasive brain stimulation techniques such as transcranial direct current stimulation (TDCS)—offer promising alternatives. ¹⁶ TDCS is a noninvasive technique that delivers a low electric current through two or more surface electrodes. ¹⁷ This current can modulate the neural resting membrane potential and endogenous excitability of targeted brain tissue. ¹⁸ In general, anodal TDCS seems to increase the excitability of the underlying cortex, whereas cathodal TDCS tends to decrease it when applied to the motor cortex (M1). ¹⁹ Moreover, TDCS can result in longer‐term effects by inducing changes in neuroplasticity. ¹⁷ , ²⁰

Due to these changes in neuroplasticity, TDCS has demonstrated benefits for pain and function in other chronic conditions associated with central pain, such as fibromyalgia or multiple sclerosis. ²⁰ , ²¹ , ²² However, its efficacy remains unclear given the methodologic limitations that some studies have in blinding. ²³ Recent clinical trials have been performed to analyze the effects of TDCS on knee OA as an isolated technique ³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² or in combination with meditation, ³³ therapeutic exercise ³⁴ or peripheral currents. ³⁵ , ³⁶ , ³⁷ Because of this, the need for a systematic review and meta‐analysis that analyzes the effectiveness of TDCS on knee OA becomes evident.

Some previous systematic reviews and meta‐analyses have studied the effects of TDCS on knee OA symptoms, especially on pain and function. These studies provide valuable evidence regarding the potential efficacy of TDCS as a nonpharmacological intervention for chronic pain management in this population. For example, the study of Chaturvedi et al., ³⁶ which included four articles, found a significant pain reduction for the active TDCS group compared to the sham condition. Since then, more authors have published systematic reviews and meta‐analyses, increasing the number of included articles and supporting the previous results. This is the case of Yang et al., ³⁸ Dai et al., ³⁹ Lozano‐Meca et al. ⁴⁰ and Comino‐Suárez et al. ⁴¹ Moreover, other reviews such as those published by Dissanayaka et al. ⁴² or Wu et al. ⁴³ have already demonstrated that these effects on pain are more effective for short‐term, compared with long‐term, assessments. However, previously published revisions do not emphasize the role of central pain processing. This is particularly relevant considering that, in many cases, knee OA‐related pain cannot be fully explained by structural findings or joint degeneration. It could be explained, at least in part, by mechanisms of central sensitization.

None of the previous reviews has compared conventionally reported questionnaires such as visual analogue scale (VAS) or numerical rating scale (NRS) with different methods for quantifying pain sensitivity such as pain pressure threshold (PPT), which identifies the threshold of pain onset caused by pressure on the injured body segment, or conditioned pain modulation (CPM), which examines the PPT difference between the measurement before and after a painful cold stimulus is applied on a different noninjured body part. ⁴⁴ Only Lozano‐Meca et al. and Wu et al. reported the results of these measures, but without determining their relevance. ¹² , ⁴⁰ Although this evidence has not been demonstrated in humans, experimental animal studies suggest that CPM could be a centrally processed measure of the net effect of the descending pain pathway (nociplastic pain). In addition, this measure could provide information about the activation of endogenous analgesia in experimental and clinical pain. ⁴⁵

The pain perception threshold can also affect quality of life, as a low threshold is associated with lower levels of physical activity and, consequently, with poorer functional status that may impact independence. Although previous reviews have considered the effects of TDCS on patients' reported function, there is inconsistency in their results. Some reported that the application of TDCS improves reported function, ⁴¹ , ⁴⁶ whereas others argue that the evidence published to date is not sufficient to support the use of TDCS to improve function in patients with OA. ³⁸ , ⁴³ Even when applied in combination with physiotherapy, long‐term results appear to be inconsistent. ⁴⁰

Given these factors, the aims of this systematic review and meta‐analysis are (1) to determine the effects of TDCS on reported pain and function in adults with knee OA; (2) to determine the effects of TDCS on different measures of pain such as the PPT and CPM; (3) and, to qualitatively identify the most widely used dosage and type of TDCS that have demonstrated improvements in pain and function.

METHODS

This systematic review and meta‐analysis were conducted based on the Preferred Reporting Items for Systematic Review and Meta‐Analyses. ⁴⁷ The protocol has been registered at the International Prospective Register of Systematic Reviews (http://www.crd.york.ac.uk/PROSPERO), under number: CRD42023489572.

Search strategy

A combination of keywords from Medical Subject Headings was used to identify relevant studies. Moreover, previous systematic reviews were analyzed to identify relevant bibliography. The following electronic databases were consulted until December 2023: the Cochrane Library, MEDLINE (PubMed), Web of Science, Embase, Scopus, Physiotherapy Evidence Database (PEDro), Literatura Latinoamericana y del Caribe en Ciencias de la Salud, and Science Direct. Reference lists of selected trials and previously published and cited reviews were also examined. No temporal filters were included.

Terms were divided into five categories relating to the PICOS strategy (Population, Intervention, Comparison, Outcomes, Study design). Categories were related using the Boolean AND. All terms related to the same category were linked using the Boolean OR. PICOS categories were as follows: Population: knee osteoarthritis; Intervention: TDCS; Comparison: control, sham; Outcome: pain, balance, gait, walk, function, security, quality of life; Study Design: randomized controlled trial. Results were restricted to human participants only.

Eligibility criteria

Articles that met the following criteria were included: (1) randomized controlled trials (RCT) on patients with knee OA, (2) TDCS as the primary intervention, (3) having a control group that performed sham TDCS intervention, and (4) participants over 18 years old. The following exclusion criteria were used: (1) Studies that included other knee issues, and (2) studies that compared TDCS with other therapies without a control group.

After discarding duplicates, articles not related to the topic were removed after reading the titles, while studies identified from the references of previous systematic reviews and meta‐analyses related to the topic were included. A total of 560 abstracts were independently assessed by two investigators (N.M.S. and S.M.C.), resulting in the exclusion of 501 records. A total of 59 studies were read entirely, and 43 were excluded for the reasons provided in Figure 1. In case of disagreement, a third researcher (P.S.A.) made the final decision.

FIGURE 1.

Study selection represented by PRISMA flow chart. LILACS, Literatura Latinoamericana y del Caribe en Ciencias de la Salud; PEDro, Physiotherapy Evidence Database; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta‐Analyses; RCT, randomized controlled trial.

Data extraction

A Microsoft Excel (2023, Microsoft Corporation, Redmond, WA, USA) table was designed to collect data from included studies regarding samples, groups, interventions, assessed outcomes, and results. Specifically, the following data were included: author and year of publication, study design and intervention (TDCS and control intervention description), sample characteristics (number, gender, and age) and outcomes (measures and main results, mainly outcomes related to pain and function).

Methodologic quality and risk of bias assessment

The PEDro scale ⁴⁸ was used to evaluate the methodological quality of the studies. ⁴⁸ Moreover, version 2 of the Cochrane risk‐of‐bias tool was used to assess the individual and global risk of bias. ⁴⁹ This tool assesses selection bias, performance bias, detection, attrition, and reporting bias. It urges users to rate the risk of bias as “high,” “low,” or “unclear” and to document the basis of their rating. All the analyzed studies were scored by two independent reviewers (N.M.S. and S.M.C.) and then compared. A third reviewer (P.S.A.) decided in cases of disagreement.

Additionally, we evaluated the quality of evidence using the Grading of Recommendations Assessment, Development, and Evaluation system. ⁵⁰ This approach examines eight key domains: risk of bias, directness of the evidence, consistency and precision of results, publication bias, effect size, dose–response relationships, and the impact of confounding factors. Each domain for each outcome is rated as “high” strength of evidence, “moderate” strength of evidence, “low” strength of evidence, or “very low” strength of evidence. The analysis was performed by two independent reviewers (N.M.S. and S.M.C.). A third reviewer (P.S.A.) decided in cases of disagreement.

Statistical synthesis method

To analyze the effects of the interventions on the clinical outcomes, a quantitative synthesis was performed by conducting a meta‐analysis using the RevMan 5.4 tool. ⁵¹ The assessed outcomes included in the meta‐analysis were (1) self‐reported pain intensity, which included all pain assessments conducted using questionnaires such as VAS or NRS; (2) PPT, which is a test that determines the threshold pressure (kPa) at which pain is experienced in the affected area; (3) CPM, which is a measure of the net effect of the descending pain pathway (comprising the facilitatory and the inhibitory effect) and calculated by the difference between PPT in a noninjured body region before and after cold stimuli (kPa) ⁴⁴ ; and (4) functionality, assessed by the function subscale of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), because all the included studies that measured reported function employed this scale. Functionality assessed with performance tests was excluded, as the tests employed measured heterogeneous functional conditions.

Only studies that compare the effects of active TDCS with sham TDCS were included in the meta‐analysis. To report the effect of the interventions, Cohen's d and the 95% confidence interval (95% CI) were used as indicators of the standardized mean difference (to compare different metrics) or mean differences (to compare the same metrics) and significance. For the assessment of study heterogeneity, the I ² statistical test was used. As recommended by the Cochrane Handbook, a heterogeneity range of 0%–40% might not be important; 30%–60% may represent moderate heterogeneity; 50%–90% may represent substantial heterogeneity; and 75%–100% represents considerable heterogeneity. ⁵² No differentiation has been made in terms of intervention duration and dose, as there are not enough studies to do it. This is addressed at length in the Discussion section.

Differences between baseline and immediate postintervention were used for the analysis. When a study did not report a difference or SD of the difference, Cochrane Handbook instructions were used for its calculation ⁵³ :

$${\mathit{SD}}_{\text{Change}}=\sqrt{{\mathit{SD}}_{\text{baseline}}^2+{\mathit{SD}}_{\text{post}}^2-\left(2*0.5*{\mathit{SD}}_{\text{baseline}}*{\mathit{SD}}_{\text{post}}\right)}$$

According to the Cochrane manual, under certain circumstances, an analysis based on changes from baseline can be more efficient and have greater statistical power than a comparison of endpoint values. This increased efficiency arises because analyzing change scores removes the component of between‐person variability from the analysis. Moreover, change from baseline results may also be more convenient if they have a less skewed distribution than the final measurement results.

Calculating a change score requires measuring the outcome twice and, in practice, may be less efficient for outcomes that are unstable or difficult to measure accurately, where the measurement error may be greater than the true between‐person variability at baseline.

RESULTS

From the total of 560 articles identified in all databases, 150 were removed due to duplication between databases and 351 were discarded because they did not comply with the inclusion criteria. Then, 59 studies were read in‐depth. Through this reading, it was seen that 8 of them were performed in patients with other pathologies, 23 were not RCTs, 2 were performed in healthy patients, 5 of the studies had no control group, and 5 applied therapies other than TDCS. Finally, 16 studies were included in the qualitative review analysis, and 7 studies were included in the quantitative analysis (meta‐analysis) (Figure 1).

Study assessment

All included studies were RCTs with fair to high methodologic quality levels (>3) according to the PEDro scale (Table 1). Most of the included studies obtained a PEDro score of ≥8 points, which indicates good to excellent methodological quality.

TABLE 1.

PEDro scale scoring.

Articles	0	1	2	3	4	5	6	7	8	9	10	Total
Ahn et al. 3	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Ahn et al. 24	Yes	Yes	Yes	No	Yes	Yes	No	Yes	Yes	Yes	Yes	8
Ahn et al. 54	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Azizi et al. 25	Yes	Yes	Yes	Yes	Yes	No	No	Yes	Yes	Yes	Yes	8
Brandão Tavares et al. 26	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes	9
Chang et al. 55	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes	9
Chaturvedi et al. 35	Yes	Yes	Yes	No	Yes	No	No	No	No	No	Yes	4
Chaturvedi et al. 56	Yes	Yes	Yes	No	Yes	No	No	Yes	Yes	Yes	Yes	7
Graca‐Tarragó et al. 37	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes	9
Martorella et al. 27	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Martorella et al. 28	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Montero‐Hernández et al. 29	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Pollonini et al. 30	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	No	Yes	No	7
Rahimi et al. 57	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes	9
Suchting et al. 31	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	9
Suchting et al. 32	Yes	Yes	Yes	Yes	Yes	No	No	Yes	Yes	No	No	6

Note: 0: eligibility criteria were specified; 1: participants were randomly allocated to groups; 2: allocation was concealed; 3: the groups were similar at baseline regarding the most important prognostic indicators; 4: there was blinding of all participants; 5: there was blinding of all therapists who administered the therapy; 6: there was blinding of all assessors who measured at least one key outcome; 7: measures of at least one key outcome were obtained from >85% of the participantsinitially allocated to groups; 8: all participants for whom outcome measures were available received the treatment or control condition as allocated or, where this was not the case, data for at least one key outcome was analyzed by “intention to treat”; 9: the results of between‐group statistical comparisons are reported for at least one key outcome; 10: the study provides both point measures and measures of variability for at least one key outcome.

Abbreviation: PEDro, Physiotherapy Evidence Database.

The risk of bias was assessed using the Cochrane risk‐of‐bias tool and is presented in Figure 2, with “Blinding of the assessor” item being the most frequently observed bias.

FIGURE 2.

Risk of bias assessment (+: low risk of bias; −: high risk of bias;?: unclear risk of bias).

The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system was used to assess the reliability of the results from the included studies (Table 2). Regarding risk of bias for pain and functional outcomes, no major concerns were identified, as validated and reliable instruments were consistently applied across study arms. For function, the certainty of evidence was downgraded by one level because, although the WOMAC is a validated measure, it reflects self‐reported rather than objectively tested functional performance.

TABLE 2.

GRADE recommendation.

Outcomes	No. of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Absolute effect	GRADE recommendation
Perceived pain	5	RCT	0	0	0	0	0	SMD −1.05 (−1.40 lower to −0.70 higher)	High
Punctuate pain threshold	4	RCT	0	−1	0	0	0	SMD −0.72 (−1.15 lower to −0.29 higher)	Moderate
Conditioned pain modulation	4	RCT	0	−1	0	−1	0	SMD −0.04 (−0.51 lower to 0.44 higher)	Moderate
Functionality	5	RCT	−1	0	0	0	0	SMD −3.96 (−7.45 lower to −0.47 higher)	Low

Abbreviations: GRADE, Grading of Recommendations Assessment Development and Evaluation; RCT, randomized controlled trial; SMD, standardized mean difference.

In terms of consistency, results across studies were generally aligned in both direction and magnitude of effect. Heterogeneity was low for self‐reported pain (I ² = 32%) and function (I ² = 28%), with overlapping confidence intervals. In contrast, PPT and CPM demonstrated substantial heterogeneity (I ² > 64%), resulting in a one‐level downgrade for each.

All included studies directly addressed the review question based on the PICOS criteria, meeting the standard for directness. With respect to imprecision, the confidence intervals for perceived pain, PPT, and function were narrow, did not cross the line of no effect, and were supported by adequate sample sizes. However, the certainty for CPM was downgraded by one level due to imprecision in the estimates.

Overall, the certainty of evidence was rated as high for perceived pain, moderate for PPT and CPM, and low for function.

Study characteristics and qualitative results

A total of 1045 participants (746 women and 299 men) completed the interventions with a mean (SD) age of 61.5 (8.2) years. All the included study participants reported pain in one or both knees due to at least moderate OA as defined by the Kellgren and Lawrence radiologic classification, ⁵⁸ without other comorbidities that could affect knee function or pain. The sample size of the clinical trials ranged from a minimum of 25 participants to a maximum of 120 participants. The sample sizes ranged from 12 to 60 participants per group. All the studies were RCTs with parallel‐group designs ranging from 2 (12/16) to 4 groups (4/16) with at least one group that received active TDCS and one control group that received sham TDCS.

Regarding TDCS electrode placement, the interventions across trials were largely consistent, with electrodes positioned over the M1–supraorbital region (C3 or C4 according to the 10–20 system) and the supraorbital region contralateral to the affected knee. The total number of sessions was different between studies. Six studies performed five sessions, ³ , ²⁴ , ²⁵ , ³¹ , ³² , ³⁷ three studies performed 10 sessions, ³³ , ⁵⁷ , ⁵⁹ four studies performed 15 sessions, ²⁶ , ²⁷ , ²⁸ , ²⁹ one study performed 16 sessions ⁵⁵ and two studies performed 30 sessions. ³⁵ , ⁵⁶ In all of the studies, TDCS was applied for 20 minutes, except Graca‐Tarragó et al. ³⁷ in which TDCS was applied for 30 minutes per session. Most of the studies conducted their sessions in a laboratory (12/16) and four were carried out in a domicile (4/16). All studies used 2 mA intensity current except for those by Rahimi et al. ⁵⁷ and Chang et al., ⁵⁵ which used 1 mA current. Most of the studies compared active TDCS with sham TDCS alone (9/16), but three added other therapies such as transcutaneous electrical nerve stimulation or electrical intramuscular stimulation (EIMS), two studies added meditation to active and sham TDCS interventions, and two studies added TDCS to conventional physical therapy. Additionally, one study compared the standard electrode placement with alternative placements and with sham TDCS. Specifically, it evaluated anodal TDCS over the left M1 (10–20 system) with the cathode on the supraorbital region at 1 mA for 20 minutes, versus anodal stimulation over the left S1, anodal stimulation over the left dorsolateral prefrontal cortex, and sham TDCS. The complete information about the studies is summarized in Table 3.

TABLE 3.

Trial characteristics and results of the studies.

Author	Study design and intervention	Sample characteristics	Outcomes
Ahn et al. & Ahn et al. 3 , 24	RCT: Double blind, 2 parallel groups. EG (n = 20). TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 2 mA with sponge electrodes (0.3 cm, 5 × 7 cm). 5 consecutive sessions, 20 min. CG (n = 20). Sham TDCS with the same electrode positions (only 30 sec of TDCS 2 mA). 5 consecutive sessions, 20 min.	Sample: 40 Gender: 19 M/21 F Age: 59.95 (9.2)	Measures: Preintervention (T0), postintervention (T1), at wk 1 (T2), at wk 2 bT3. Pain (NRS, and SF‐MPQ‐2), Multimodal quantitative sensory test: HPTs (TSA‐II neurosensory analyzer), mechanical pain (PPT by digital pressure algometer and punctate mechanical pain), CPM, function (6MWT, SPPB, and WOMAC), security (secondary effects and VAS), PROMIS. Main results: Pain: EG < CG in NRS pain (p = .03) in T0‐T1‐T2‐T3 (p < .02). EG = CG in SF‐MPQ‐2 (p > .05). Multimodal quantitative sensory test: EG > CG on toleration to heat pain, PPT, punctate mechanical pain and CPM (p > .05 in T1 vs. T0). Function: EG and CG increased in 6MWT and SPPB but without differences between them (p > .05). EG = CG in WOMAC (p > .05). Security: TDCS was well tolerated in both groups (EG = CG, p > .05).
Ahn et al. 33	RCT: Double blind, 2 parallel groups. EG (n = 15). Domiciliary active TCDS + Meditation, 20 min, 5 d/wk, 2 wk. TDCS was applied on M1‐SO (C3 or C4 according to 10–20 system), S0 (ipsilateral to the affected knee), 2 mA with sponge electrodes (0.3 cm, 5 × 7 cm). CG (n = 15). Domiciliary sham TDCS + Meditation. TDCS was applied with the same electrode positions (30 sec of TDCS 2 mA). 20, 5 d/wk, 2 wk.	Sample: 30 Gender: 12 M/18 F Age: 59.47 (6.91)	Measures: Preintervention (T0), postintervention (T1). Pain (NRS, PPT, CPM), patients' satisfaction (0–10 Likert) and secondary effects and function (WOMAC) Pain: EG and CG improved in NRS and WOMAC (p < .02), but EG > CG (p < .001). PPT and CPM improved in EG and reduced in CG. Main results. Patients' satisfaction: EG = CG in patients' satisfaction. Function: EG and CG improved in WOMAC (p < .02), but EG > CG (p < .001).
Azizi et al. 25	RCT: Single‐blind, 2 parallel groups. EG (n = 27): TDCS applied on M1‐SO (C3 or C4 according to 10–20 system), S0 (ipsilateral to the affected knee), 2 mA with 16 cm2 sponge electrodes. 5 consecutive se sessions, 20 min. CG (n = 27). Sham TDCS with the same electrode positions (10 sec of TDCS 2 mA). 5 consecutive sessions, 20 min.	Sample: 54 Gender: 15 M/39 F Age: 58.85 (12.6)	Measures: Preintervention (T0), 3 mo postintervention (T1). Pain (VAS) and function (KOOS). Main results: Pain: EG and CG reduced pain measures but EG = CG (p = .226). Function: EG and CG increased KOOS values (p = .005).
Brandão Tavares et al. 26	RCT: Double‐blind, 2 parallel groups. EG (n = 52). TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 2 mA with 35 cm2 sponge electrodes. 20′, 5 d/wk, 3 wk. CG (n = 52). Sham TDCS with the same electrode positions (10 sec of TDCS 2 mA). 20 min, 5 d/wk, 3 wk.	Sample: 104 Gender: 16 M/88 F Age: 73.96 (7.98)	Measures: Preintervention (T0), at first wk (T1), at second wk (T2), posttreatment (T3), 2 wk after treatment (T4), 4 wk after treatment (T5), 8 wk after treatment (T6). Pain (VAS, PPT, CPM), quality of life (SF‐12), cognition, mood, security (adverse effects questionnaire) and function (TUG, one‐leg stance test and WOMAC). Main results: Pain: EG < CG in VAS (T0 vs. T3, p < .001). The effects were not maintained at long‐term (T3‐T6). Security: CG > EG in security items “headache” (p < .01) and “tingling” (p < .001). EG > CG in security items “pain in the scalp” (p = .03) and “redness of skin” (p < .01). Quality of life, cognition, mood and function: EG = CG.
Chang et al. 55	RCT: Single‐blind, 2 parallel groups. EG (n = 13). Active TDCS 20 min before exercise +5 Therapeutic quadriceps exercise (3 × 10–30 sec rest). TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 1 mA with 35 cm2 sponge electrodes. 2 d/wk, 8 wk. CG (n = 12). Sham TDCS with the same electrode positions (15 sec of TDCS 1 mA) + Therapeutic quadriceps exercise. 2 d/wk, 8 wk.	Sample: 25 Gender: 10 M/20 F Age: 61.95 (10.1)	Measures. Preintervention (T0), postintervention (T1). Adherence (sessions assisted, dropouts, etc.), pain (VAS, PPT, HPTs, CPM, NFR), function (WOMAC) and security (adverse effects). Main results. Pain: EG and CG decreased VAS values T0 vs. T1 (p < .05), but EG = CG (p = .08). Rest of variables: EG = CG (p > .05).
Chaturvedi et al. 35	RCT: 4 parallel groups, single‐blinded. EG 1 (n = 18). Active TDCS (applied on M1‐SO, C3 or C4, S0 [ipsilateral to the affected knee], 2 mA, 20 min, 5 d/wk) + active TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk. EG 2 (n = 18). Active TDCS (applied on M1‐SO, C3 or, S0 [ipsilateral to the affected knee], 2 mA, 20 sec, 5 d/wk) + sham TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk. EG 3 (n = 18). Sham TDCS (same electrode positions but only 30 sec) + active TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 5 d/wk, 6 wk. CG (n = 18). Sham TDCS (same electrode positions but only 30 sec) + sham TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 5 d/wk, 6 wk.	Sample: 71 Gender: 23 M/48 M Age: 53.55 (5.96)	Measures. Preintervention (T0), wk 1 (T1), wk 2 (T2), posttreatment (T3). Pain (VAS) and function (6MWT, KOOS and WOMAC). Main results: Pain: At T1, T2 and T3 EG1, EG2 and EG3 < CG in VAS. EG2 > CG in KOOS pain (p > .05) at T1, T2 and T3. Function: EG1 = EG2 = EG3 = CG in 6MWT in all the times. At T1 EG2 > CG in KOOS, at T2 EG1 and EG2 > CG, at T3 EG1 and EG2 > CG. EG2 > CG at T1, T2 and T3, and EG1 > CG at T3 in KOOS function. Without differences in KOOS quality of life and the rest of variables in any group.
Chaturvedi et al. 56	RCT: 4 parallel groups, single‐blinded. EG 1 (n = 18). Active TDCS (applied on M1‐SO, C3 or, S0 [ipsilateral to the affected knee], 2 mA, 20 min, 5 d/wk) + active TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk. EG 2 (n = 18). Active TDCS (applied on M1‐SO, C3 or C4 according to 10–20 system, S0 [ipsilateral to the affected knee], 2 mA, 20 min, 5 d/wk) + sham TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk. EG 3 (n = 18). Sham TDCS (same electrode positions but only 30 sec) + active TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk. CG (n = 18). Sham TDCS (same electrode positions but only 30 sec) + sham TENS + conventional physiotherapy (hot packs, quadriceps stretching and strengthening). 6 wk.	Sample: 71 Gender: 23 M/48 M Age: 53.55 (5.96)	Measures: Preintervention (T0), wk 1 (T1), wk 2 (T2), posttreatment (T3). Pain (VAS) and function (6MWT, KOOS, WOMAC). Main results: Pain: EG2 < CG at T1 (p = .011); EG1, EG2 and EG3 < CG at T2 (p < .006); EG1, EG2 and EG3 < CG at T3 (p < .015). Function: EG1 = EG2 = EG3 = CG in all the times on the 6MWT (p > .05). EG1 and EG2 > CG at T2 in KOOS symptoms (p < .032). EG1 > CG at T3 in KOOS symptoms (p = .046). EG1 = EG2 = EG3 = CG (p > .05) in KOOS pain, function, and sports in any time. EG1 > CG at T3 in KOOS quality of life (p = .04). No results showed in WOMAC.
Graca‐Tarragó et al. 37	RCT: 4 parallel groups, double‐blinded. EG 1 (n = 15). Active TDCS (applied on M1‐SO, C3 or C4, S0 [ipsilateral to the affected knee], 2 mA with 35 cm2 sponge electrodes, 30 min, 5 d/wk) + EIMS 30 min (2 Hz, acupuncture needles on spine segments related to knee). 5 d/wk, 1 wk. EG 2 (n = 15). Active TDCS + sham EIMS (the same as active but without current). 5 d/wk, 1 wk. EG 3 (n = 15). Sham TDCS (only 30 sec of TDCS 2 mA) + active EIMS 30 min (2 Hz, acupuncture needles on spine segments related to knee). 5 d/wk, 1 wk. CG (n = 15). Sham TDCS + sham EIMS (the same as active but without current). 5 d/wk, 1 wk.	Sample: 60 Gender: 0 M/60 F Age: 64.60 (8.00)	Measures: Pretreatment (T0), posttreatment (T1). Pain (VAS, NRS, CPM, PPT, and analgesic use), function (WOMAC) and serum BDNF levels. Main results: Pain: EG1 reduced VAS, CPM, and analgesics use (p < .03). EG1, EG2 and EG3 > CG in PPT. Function: EG1 > EG2, EG3, and CG in WOMAC. EG2 and EG3 > CG in WOMAC. BDNF: EG1 = EG2 = EG3 = CG (p > .05).
Martorella et al. 27	RCT: Double‐blind, 2 parallel groups. EG (n = 60). Domiciliary TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 2 mA with 16 cm2 sponge electrodes. 20 min, 5 d/wk, 3 wk. CG (n = 60). Domiciliary sham TDCS with the same electrode positions (only 30 sec of TDCS 2 mA). 20 min, 5 d/wk, 3 wk.	Sample: 120 Gender: – Age: 65.95 (8.4)	Measures: Preintervention (T0), at wk 1, (T1), at wk 2, (T2), postintervention (T3), 3 mo postintervention (T4). Pain (NRS), function (WOMAC), acceptability, and viability. Main results: Pain: The EG show a reduction in pain T4 < T3, T4 and T3 < T0 (p > .05). CG did not show differences. Function: EG = CG (p > .05) in WOMAC. Acceptability and viability were high in both groups.
Martorella et al. 28	RCT: Double‐blind, 2 parallel groups. EG (n = 60). Domiciliary TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 2 mA with 16 cm2 sponge electrodes. 20 min, 5 d/wk, 3 wk. CG (n = 60). Domiciliary sham TDCS with the same electrode positions (only 30 sec of TDCS 2 mA). 20 min, 5 d/wk, 3 wk.	Sample: 120 Gender: 36 M/84 F Age: 66 (8.41)	Measures. Preintervention (T0), at week 1, (T1), at week 2, (T2), postintervention (T3), Pain (NRS, HPTh, HPTo, PPT, CPM). Main results At T3, EG > CG (p > .05) in HPTh, PPT, HPTo, and CPM. A significative correlation between HPTh (p = .03), HPTo, (p < .01) and PPT (p < .001) with NRS were found. Not for CPM (p = .08).
Montero‐Hernández et al. 29	RCT: Double‐blind, 2 parallel groups. EG (n = 60). Domiciliary active TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee), 20 min, 2 mA. 5 d/wk, 3 wk. CG (n = 60). Sham TDCS simulated (same protocol but only 30 sec of TDCS), 20 min. 5 d/wk, 3 wk.	Sample: 120 Gender: 36 M/84 F Age: 66 (8.3)	Measures. Pre‐intervention (T0), at wk 1, (T1), at wk 2, (T2), post‐intervention (T3). Pain (fNIRS with thermal and mechanical stimulation). Main results: EG intervention modulated the pain significantly but not the CG. Only the EG showed a significant reduced number and strength of functional connections evoked during nociception in the prefrontal cortex (M1 and S1).
Pollonini et al. 30	RCT: Double‐blind, 2 parallel groups. EG (n = 15) Active TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee) applied with sponge electrodes + guided meditation (at the same time). 20 min, 5 d/wk, 2 wk. CG (n = 15). Sham TDCS (same protocol but only 30 sec of TDCS) + breathing and relaxation. 20 min, 5 d/wk, 2 wk.	Sample: 30 Gender: – Age: –	Measures. Preintervention (T0), postintervention (T1). Pain (oxyhemoglobin motor cortex activation, NRS) and function (WOMAC). Main results: Pain: NRS values reduced significantly in EG (p < .001) but not in CG (p > .05). EG showed a significant oxyhemoglobin motor cortex activation but not in the CG. Function: No differences were found in WOMAC (p = .11) in any group.
Rahimi et al. 57	RCT: Double‐blind, 4 parallel groups. EG 1 (n = 20). Anodic active TDCS on left M1, with 10/20 system and cathode on SO (1 mA, 20 min) + Physiotherapy (TENS, pulsed ultrasound, Infrared, patello‐femoral mobilizations, strength quadriceps exercises and stretching). 5 d/wk, 2 wk. EG 2 (n = 20). Anodic active TDCS on left S1 (same protocol) + physiotherapy. 5 d/wk, 2 wk. EG 3 (n = 20). Anodic active TDCS on left DLPFC (same protocol) + physiotherapy, 5 d/wk, 2 wk. CG (n = 20). Sham TDCS (same protocol but only 30 sec) + Physiotherapy. 5 d/wk, 2 wk.	Sample: 80 Gender: 8 M/72F Age: 58.8 (3.3)	Measures: Preintervention (T0), postintervention (T1), 1‐mo follow‐up (T2). Pain (VAS), function (KOOS, ROM, and strength, stepping 15 sec, chair stand test, 10‐m walking test). Main results. Pain: EG1, EG2, EG3, and CG reduced their punctuation in T1 and T2 vs. T0 (p < .001). EG1 and EG2 < CG on VAS (at T1 and T2). EG3 < CG at T2. EG1 > EG2 on VAS (at T2). EG2 > EG3 on VAS at T2. Function: EG1 > CG on KOOS (at T1 and T2), walking test (at T2) and chair stand test (at T1 and T2) (p < .05). EG2 > CG on KOOS (at T2), strength (at T2) and functional tests (at T2) (p < .05). EG3 > CG on chair stand test (at T1 and T2). EG1 > EG2 on VAS (at T2), KOOS and functional tests (at T2). EG1 > EG3 on 10‐m walk mining test at T2. EG2 > EG3 on KOOS at T1 and T2, and on strength and functional tests at T2.
Suchting et al. 31	RCT: Single‐blind, 2 parallel groups. EG (n = 20). Active TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee) applied with sponge electrodes, 2 mA, 20. 5 d/wk, 1 wk. CG (n = 20). Sham TDCS. 5 d/wk, 1 wk.	Sample: 40 Gender: 19 M/21 F Age: 59.9 (9.13)	Measures: Preintervention (T0), at D 1 (T1), postintervention (T2). Inflammatory and stress markers (B‐endorphin, PCR, cortisol, TNF‐α, IL‐6 and IL‐10). Main results: The EG < CG in inflammatory cytokines and B‐endorphins. EG = CG in PCR and cortisol.
Suchting et al. 32	RCT: Single‐blind, 2 parallel groups. EG (n = 20). Active TDCS applied on M1‐SO (C3 or C4), S0 (ipsilateral to the affected knee) applied with sponge electrodes, 2 mA, 20 min. 5 d/wk, 1 wk. CG (n = 20). Sham TDCS. 5 d/wk, 1 wk.	Sample: 40 Gender: 18 M/20 F Age: 60 (9.1)	Measures. Preintervention (T0) and postintervention (T1). BDNF levels, osteocalcin, adiponectin, irisin. Main results EG < CG in BDNF levels at T1 (p < .05). EG = CG (p > .05) in adiponectin and osteocalcin. EG increased in irisin in EG (p < .05) T0 vs. T1.

Abbreviations: 6MWT, 6‐Minute Walk Test; BDNF, brain‐derived neurotrophic factor; CG, control group; CPM, conditioned pain modulation; DLPFC, dorsolateral prefrontal cortex; EIMS, electrical intramuscular stimulation; EG, experimental group; F, female; fNIRS, functional near‐infrared spectroscopy; HPTs, heat pain thresholds; IL, interleukin; KOOS, Knee injury and Osteoarthritis Outcome Score; M, male; M1‐SO, primary motor cortex‐ipsilateral supraorbital region; N, participants per group; NFR, nociceptive flexion reflex; NRS, numerical rating scale; PCR, polymerase chain reaction; PPT, pressure pain threshold; PROMIS, Patient‐Reported Outcomes Measurement Information System; RCT, randomized controlled trial; ROM, range of movement; SF‐12, 12‐Item Short‐Form Health Survey; SF‐MPQ‐2, Short‐Form McGill Pain Questionnaire; SPPB, Short Physical Performance Battery; TDCS, transcranial direct current stimulation; TENS, transcutaneous electrical nerve stimulation; T0, preintervention; T1, postintervention; TNF‐α, tumor necrosis factor alpha; TSA‐II, Advanced Thermosensory Stimulator of pain; TUG, Timed Up and Go Test; VAS, visual analogue scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index; –, data not available.

Fourteen of the 16 selected studies measured pain. The most commonly reported scales were the NRS (7/14) ³ , ²⁶ , ²⁷ , ²⁸ , ³⁰ , ³³ , ³⁷ and VAS (7/14). ²⁵ , ²⁶ , ³⁵ , ³⁶ , ³⁷ , ⁵⁵ , ⁵⁷ Other pain assessment scales used were the subscale of pain of the WOMAC questionnaire (3/14) ³ , ²⁶ , ³³ and the Short‐form McGill Pain Questionnaire 2 (1/14). ³ As the studies that reported these two questionnaires also used NRS or VAS, only the NRS and VAS were included in the meta‐analysis. Additionally, multimodal quantitative sensory assessment was performed in 6 of the 14 studies to assess a possible nociplastic component of pain. ²⁴ , ²⁶ , ²⁸ , ³³ , ³⁷ , ⁵⁵ The multimodal assessment included tests such as heat pain thresholds (HPTs) (3/14) ²⁴ , ²⁸ , ⁵⁵ or the maximum temperature at which the person begins to feel pain, PPTs (6/14) ²⁴ , ²⁶ , ²⁸ , ³³ , ³⁷ , ⁵⁵ and CPM (6/14). ²⁴ , ²⁶ , ²⁸ , ³³ , ³⁷ , ⁵⁵ Because only three studies reported HPT, and these studies also performed PPT procedures, only PPT outcomes were included in the meta‐analysis. As indirect outcomes of pain, 2 studies measured peripheral brain‐derived neurotrophic factor levels ³² , ³⁷ and 1 study measured inflammatory markers. ³¹ Twelve of the 14 pain studies showed significant improvements in pain after active TDCS intervention (alone [those included in the meta‐analysis] or combined with other therapies) compared to control group. ³ , ²⁴ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³³ , ³⁵ , ³⁶ , ³⁷ , ⁵⁷

Six of 16 studies measured functional performance employing functional performance tests such as 6‐Minute‐Walk‐Test, ³ , ³⁶ , ⁵⁶ Timed Up and Go, ²⁶ One Leg Stance Test (5/6), ²⁶ or the Knee injury and Osteoarthritis Outcome Score (4/6). ²⁵ , ³⁵ , ³⁶ , ⁵⁷ Three of the six studies showed improvements compared to the control group, ²⁹ , ³⁵ , ³⁷ and three of them did not. ³ , ²⁵ , ²⁶ Additionally, 9 of the 16 studies ³ , ²⁶ , ²⁸ , ³⁰ , ³³ , ³⁵ , ³⁶ , ³⁷ , ⁵⁵ measured self‐reported functional performance using the WOMAC questionnaire and its subscale of functional impairment (2/9 increased their functionality perception after TDCS therapy ³³ , ³⁷ ). All of the studies that measured function using functional performance tests also assessed self‐reported function with the WOMAC.

Safety of the intervention was reported in four studies. ³ , ²⁶ , ³³ , ⁵⁵

Quantitative results (meta‐analysis)

Only studies comparing the effects of active TDCS with sham TDCS were included in the meta‐analysis. Of the 1045 participants included in the qualitative analysis, 738 participants were included in the quantitative analysis. To address the main objectives of the study, meta‐analyses related to the two main outcomes of pain and function were performed. The meta‐analysis of pain was grouped in three subcategories: self‐reported pain intensity, PPT and CPM. In the meta‐analysis for function, only studies that reported the WOMAC were included.

Pain

Self‐reported pain intensity, measured using standardized scales, showed a significant reduction in the active TDCS group (Cohen's d = −1.05 [95% CI, −1.40 to −0.70]). Pain assessed using PPT also demonstrated a significant improvement (Cohen's d = −0.72 [95% CI, −1.15 to −0.29]). In contrast, CPM outcomes did not differ between the active and sham TDCS groups. Detailed results and corresponding forest plots are presented in Figures 3, 4, 5. Heterogeneity was low for self‐reported pain measured by scales (I ² = 32%), moderate to high for pain assessed using PPT (I ² = 64%), and moderate for CPM outcomes (I ² = 68%).

FIGURE 3.

Results of perceived pain scales meta‐analysis. CI, confidence interval; df, degree of freedom; IV, inverse variance; TDCS, transcranial direct current stimulation.

FIGURE 4.

Results of punctuated pain threshold meta‐analysis. CI, confidence interval; df, degree of freedom; IV, inverse variance; TDCS, transcranial direct current stimulation.

FIGURE 5.

Results of conditioned pain modulation meta‐analysis. CI, confidence interval; df, degree of freedom; IV, inverse variance; TDCS, transcranial direct current stimulation.

Function

The meta‐analysis showed significant differences between interventions (active TDCS and sham TDCS) in the measures of function (Cohen's d = −3.96 [95% CI, −7.45 to −0.47]). The active TDCS group showed higher scores on WOMAC (an increase of functionality) than the sham TDCS group. The heterogeneity was low [I ² = 28%]. The complete information and the forest plot are presented in Figure 6.

FIGURE 6.

Results of functionality meta‐analysis. CI, confidence interval; df, degree of freedom; IV, inverse variance; TDCS, transcranial direct current stimulation.

DISCUSSION

The primary aim of this systematic review was to determine the effects of TDCS as compared to sham TDCS intervention on pain and function in individuals with knee OA. Additionally, this systematic review and meta‐analysis explores the effects of TDCS on different self‐reported measures of pain, including conventional patient‐reported pain intensity scales ⁶¹ ; PPT, which indicates the pain threshold when applying pressure to a specific point of the affected knee; and CPM, which measures the difference in pain threshold before and after the application of a cold stimulus. CPM provides insight into the functioning of endogenous pain inhibition or modulation mechanisms, which are thought to be impaired in individuals with chronic pain. Both PPT and CPM are laboratory‐based assessments that have been linked to central sensitization, although this relationship remains debated in recent literature. ⁶² , ⁶³ , ⁶⁴ Because pain measurement inherently includes a subjective component, we aimed to more fully understand how TDCS therapy might influence multiple dimensions of the pain experience.

This systematic review included 16 randomized controlled trials with a total of 1045 adults with knee OA. Our findings showed that TDCS interventions may have a beneficial effect on pain experienced by patients with OA. Specifically, TDCS improves self‐reported pain level assessed through conventional patient‐reported scales and PPT, when compared to sham interventions. However, this improvement was not reflected in the CPM assessment. The results also showed that TDCS could improve the function of people with knee OA.

The results on pain are consistent with previous systematic reviews, ³⁸ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁶ , ⁵⁶ which concluded that TDCS improved pain in the short‐ and medium‐term in patients with knee OA (compared to the effect of a sham intervention). Moreover, the results of the meta‐analysis are supported by those studies that qualitatively assessed pain using indirect measures (such as decrease in inflammatory markers or an increase in brain‐derived neurotrophic factor) to evaluate the TDCS effects. ³² , ⁶⁵ In this same line, other studies in other populations have demonstrated that TDCS also exhibits benefits, such as those achieved in fibromyalgia, chronic low back pain, or myofascial pain, where central sensitization may play an important role. ²² , ⁶⁶ , ⁶⁷

The observed decrease in perceived knee pain can be attributed to multiple underlying mechanisms. TDCS, both anodal and cathodal, purportedly attenuates gamma‐aminobutyric acid neurotransmission and engenders biochemical changes, encompassing alterations in ion channel abundance and activity, thereby facilitating electrical propagation (non‐synaptic plasticity). ⁶⁸ Furthermore, beyond localized effects, TDCS has been evidenced to activate diverse brain regions and their interconnections, using magnetic resonance and imaging studies. ²⁹ , ⁶⁹

Although TDCS produces these changes in the central nervous system, our results showed that, overall, no impact was obtained on pain modulation assessed by CPM. This discrepancy may stem from various factors. There is heterogeneity in the procedures measuring the central modulation of pain. Those that registered the differences in pressure thresholds in the injured knee did not achieve TDCS improvement. Nevertheless, the study of Martorella et al., who assessed this central nervous system capacity in a healthy body region (ie, trapezius), did obtain an improvement of the central pain mechanisms.

CPM is posited as a measure of nociplastic pain that informs about an altered function of pain‐related sensory pathways in the periphery and central nervous system, causing increasing sensitivity. ⁷⁰ Nevertheless, when the measurement is conducted on an injured body region, intrinsic injury mechanisms may mediate the overall central response to pain. Therefore, the central pain modulation assessment (ie, the reorganization of central pain pathways discussed previously) would be conducted differently, where other peripheral pain mechanisms do not influence the result, that is, only an evaluation assessment of an unaffected area. Further, there is a study that showed increased CPM values. ²⁸ This particular study employed more sessions than the others (15 sessions of TDCS versus 5 or 10 sessions). ²⁴ , ³³ Therefore, due to the heterogeneity of the protocols, further research is needed to explore whether is a minimum number of sessions required to affect the endogenous pain inhibition system, or which body region is most suitable when exploring the response to painful stimuli. Indeed, another study that also used 15 sessions found no improvement in the aged population. ²⁶ Therefore, not only the number of sessions and the body region but also the characteristics of the sample must be considered, as older adults may present changes in neuroplasticity compared to younger populations. ⁷¹ Moreover, Graca‐Tarragó et al. ³⁷ performed only five sessions of TDCS combined with EIMS and found differences with sham conditions in CPM. However, these benefits could be attributed to the EIMS, because a previous study that applied EIMS alone also obtained improvements in CPM. ⁷² Accordingly, future research endeavors should prioritize delineating the optimal dosing regimen requisite to assuaging pain in patients with knee OA, taking into consideration anthropometric and sociodemographic factors and should consider the combination of TDCS with other techniques.

This systematic review and meta‐analysis demonstrated an improvement in function with TDCS. This is the first systematic review that analyzes the benefits of TDCS on function in patients with knee OA. Only the systematic review conducted by Elsner et al. ⁷³ analyzed the effects of TDCS therapy among individuals post stroke and documented the absence of functional improvements post anodal and/or post cathodal TDCS therapy. Nevertheless, these results are not entirely comparable, as the population with stroke manifests disruption of motor pathways that may hinder functional recovery.

The improvements obtained in our study could be multifactorial. On the one hand, a reduction in knee pain could lead to increased function as not experiencing pain may promote greater physical activity and participation in daily life activities. ⁷⁴ This hypothesis is supported by the results of the meta‐analysis, which showed that studies with better results in pain also achieved better results in perceived function ³⁰ , ³³ whereas studies with poorer improvement in pain did not significantly improve their perceived function. ³ , ²⁶ , ²⁷ Alternatively, improved function could be due to the promotion of new motor learning via TDCS influences on neurophysiological mechanisms such as cerebral excitability, neurotransmitters, synaptic plasticity, and brain network functional connections in multiple brain regions. ⁶⁰ , ⁷⁵ Therefore, future studies should investigate these relationships.

Regarding the optimal dosage and type of TDCS needed to improve the outcomes described herein, it is important to acknowledge the substantial heterogeneity across studies. This variability makes it difficult to establish firm recommendations for achieving improvements in pain or function. The most commonly used protocol for knee OA applies 2 mA for 20 minutes, with electrodes positioned over the primary motor cortex (M1) and the ipsilateral supraorbital region. ¹³

Meta‐analytic findings suggest that at least 5 sessions are required to achieve a large effect on self‐reported pain (Cohen's d = −1.05), whereas medium effects on functional outcomes (Cohen's d = 0.29) generally require at least 10 sessions. These thresholds may also vary by population; for example, older adults may demonstrate smaller treatment effects with the same number of sessions, indicating that dosage may need to be adjusted accordingly. For example, Ahn et al. ³³ obtained pain improvements with 10 sessions of TDCS, but the mean age of the population was 59  years. This improvement was lower in the studies by Brandão Tavares et al. ²⁶ and the Martorella et al. ²¹ which used 15 sessions, but their samples were aged 74 years and 66 years, respectively. To date, the evidence is scarce and further studies with larger samples and different dosages are needed. What is known is that this technique is safe, as no adverse effects have been reported in any of the studies included in this systematic review and meta‐analysis. ³ , ²⁶ , ⁵⁵

Although the methodologic quality of the included studies was generally moderate to high, the certainty of evidence varied across outcome measures. High certainty was found for perceived pain, indicating strong confidence in the effect estimates. For PPT and CPM, the certainty was moderate, suggesting some limitations but still providing a reasonable degree of confidence. In contrast, the evidence for functional outcomes was rated as low certainty, indicating that current findings are less reliable and that additional rigorous studies are needed to better understand the impact of TDCS on function.

Our systematic review and meta‐analysis has several strengths as well as limitations. Among its strengths, we conducted a comprehensive search across eight databases and additional sources to identify all relevant studies that could inform clinicians and physiotherapists about the potential benefits of TDCS for pain and function in individuals with knee OA. Most of the included studies demonstrated moderate to high methodological quality. Furthermore, despite variability in TDCS protocols across studies, we were able to identify and summarize the most commonly used and clinically relevant parameters.

However, our review also has limitations. Although the meta‐analysis yielded consistent findings overall, substantial methodological heterogeneity—particularly in CPM assessment—was observed across studies, which may affect the comparability and interpretation of results.

This could be explained by the fact that, although the patients included had the same pathology, differences in OA stage, population age, and comorbidities could have influenced the patient's experience of OA as well as the results of the interventions. Additionally, it is important to acknowledge that the studies by Ahn et al. ³ , ²⁴ , ³³ had substantial influence on the overall results and conclusions of this meta‐analysis. Their relative impact contributed to shaping the aggregate effect sizes and interpretations. Future research incorporating a wider range of studies may clarify the robustness and generalizability of the observed effects. Future studies should also standardize measurements and compare TDCS interventions with other approaches to better understand its impact on pain and function.

CONCLUSION

Moderate to high‐quality evidence suggests that TDCS used alone or in combination with other interventions can improve self‐reported pain intensity and PPT in adults with knee OA. However, TDCS does not appear to affect endogenous pain inhibition as measured by CPM and its effect on function remains uncertain.

Future TDCS studies should standardize intervention protocols. To date, the most commonly used and effective approach to improve pain and function is to deliver at least 15 sessions of active TDCS at 2 mA for 20 minutes targeting M1–supraorbital (C3 or C4) and S0 ipsilateral to the affected knee.

FUNDING INFORMATION

This work was supported in part by Spanish Government, co‐financed by the European Union (EU) European Regional Development Fund (ERDF) Funds under Grant PID2021‐125694OB‐I00; and from Generalitat Valenciana, Conselleria d'Innovació, Universitats, Ciència i Societat (CIAICO/2024/107).

DISCLOSURE

None of the authors have a conflict of interest to disclose.

DATA AVAILABILITY STATEMENT

Research data are not shared.