Introduction
Major depressive disorder (MDD) remains a public health concern, affecting millions of individuals worldwide [Reference Herrman, Patel, Kieling, Berk, Buchweitz and Cuijpers1, Reference Diseases and Injuries2]. Beyond mood symptoms, MDD greatly impairs cognitive functions, encompassing memory, attention, and executive functions, further worsening the quality of life in individuals with this disorder [Reference Elliott3, Reference Veiel4]. Its impact extends beyond the individual, causing substantial societal and economic burdens [5, Reference Lam, McIntosh, Wang, Enns, Kolivakis and Michalak6]. Conventional treatments, such as antidepressant medications and psychotherapy, often have limited effectiveness, and many patients do not respond adequately [Reference Rush, Trivedi, Wisniewski, Nierenberg, Stewart and Warden7–Reference Leichsenring, Steinert, Rabung and Ioannidis9]. Notably, cognitive impairment in MDD can persist even after the alleviation of other symptoms [Reference Snyder10, Reference Hasselbalch, Knorr and Kessing11]. Furthermore, these treatments can be burdened by adverse effects and require considerable time commitments [Reference Parikh, Quilty, Ravitz, Rosenbluth, Pavlova and Grigoriadis12, Reference Kennedy, Lam, McIntyre, Tourjman, Bhat and Blier13].
In response to these challenges, transcranial direct current stimulation (tDCS), a non-invasive neuromodulation technique, has emerged as a promising alternative or adjunctive treatment for MDD [Reference Chen, Lyu, Wang, Huang, Yang and Zhang14]. Recent meta-analyses suggest its effectiveness in treating MDD, not only improving depressive symptoms [Reference Mutz, Edgcumbe, Brunoni and Fu15, Reference Brunoni, Moffa, Fregni, Palm, Padberg and Blumberger16] but also enhancing cognitive functions, particularly working memory and attention [Reference Begemann, Brand, Ćurčić-Blake, Aleman and Sommer17, Reference Wolkenstein and Plewnia18]. Moreover, tDCS offers potential advantages over other techniques, including excellent safety and tolerability profiles, cost-effectiveness, and user-friendly application [Reference Moffa, Brunoni, Fregni, Palm, Padberg and Blumberger19, Reference Brunoni, Amadera, Berbel, Volz, Rizzerio and Fregni20]. Given these advantages, with simple training and guidance, a potential for home-based tDCS has been suggested [Reference Loo, Alonzo and Fong21, Reference Cappon, den Boer, Jordan, Yu, Lo and LaGanke22], including enhanced accessibility and convenience [Reference Carvalho, Brietzke, Gasparin, Dos Santos, Vercelino and Ballester23, Reference Park, Oh, Chung, Kim, Kim and Park24].
Despite these advantages, the efficacy of home-based tDCS as an adjunct to antidepressants in treating MDD, encompassing both mood and cognitive symptoms, remains underexplored. Furthermore, since a non-linear relationship between tDCS intensity and efficacy has been consistently reported [Reference Batsikadze, Moliadze, Paulus, Kuo and Nitsche25, Reference Jamil, Batsikadze, Kuo, Labruna, Hasan and Paulus26], the comparative effects of different tDCS intensities and their safety profiles in a home-based context need comprehensive evaluation.
We designed a double-blind, sham-controlled, randomized clinical trial to evaluate the efficacy and safety of home-based high- and low-intensity tDCS as adjunctive treatments for MDD. We hypothesized that there would be differences in improvement among the three groups (high-intensity, low-intensity, and sham) after 6 weeks of treatment. Additionally, we hypothesized that both active tDCS groups would demonstrate greater improvements in the depressive symptoms (HAM-D), anxiety symptoms (HAM-A), and quality of life (QLES) than those in the sham. Finally, we expected that adverse effect rates would be similar across all three groups.
Methods
The study was a multicenter, double-blind, randomized controlled trial conducted at Seoul National University Bundang Hospital and Samsung Medical Center from March 2021 to January 2023. The study enrolled 141 patients diagnosed with MDD who were receiving antidepressants. Participants were randomly allocated to one of three groups: sham (n = 45), low-intensity (1 mA, n = 47), or high-intensity (2 mA, n = 49) tDCS. The intervention lasted for 6 weeks, during which participants self-administered tDCS once daily for a total of 42 sessions. Each session lasted 30 minutes and was performed every day, including weekends and holidays. Outcomes were assessed during in-person visits at baseline (week 0), week 2, week 4, and week 6. The primary outcome was cognitive improvement assessed by changes in total accuracy on the 2-back test from baseline to week 6. Secondary outcomes included changes in depressive symptoms (HAM-D scores), anxiety levels (HAM-A scores), and quality of life (QLES scores). The randomization process used a computer-generated list with block sizes. The study protocol was approved by the Institutional Review Boards of both Seoul National University Bundang Hospital (approval number: B-2011-651-006) and Samsung Medical Center (approval number: 2020–06-143) and registered at ClinicalTrials.gov (NCT04709952). All participants provided written informed consent.
Participants
Participants were recruited through physician referrals and advertisements. All participants were screened by trained, board-certified psychiatrists using the modified Mini-International Neuropsychiatric Interview [Reference Amorim27] and a case review of their medical records to diagnose MDD. In-person interviews were conducted for pre-screening, selecting participants aged 19–65. The study included patients with residual symptoms after antidepressant treatment who did not achieve remission. Information on the antidepressants used by the participants is included in Table 1. All the patients receiving treatment with antidepressants approved in South Korea were included, and no specific antidepressants were excluded from the study. The specific inclusion and exclusion criteria are as follows: (1) diagnosed with MDD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [28]; (2) experiencing an ongoing depressive episode with a Clinical Global Impression Severity score of ≥3; (3) ongoing use of psychiatric medication for depressive symptoms at the time of the participation; and (4) able to understand the research details and consent documentation and capable of responding to questionnaires. Pharmacologic interventions adhered to the Canadian Network for Mood and Anxiety Treatments 2016 depression guidelines [Reference Parikh, Quilty, Ravitz, Rosenbluth, Pavlova and Grigoriadis12, Reference Kennedy, Lam, McIntyre, Tourjman, Bhat and Blier13]. Participants were excluded if they had a history of neurological disorders, head injuries, high suicide risk, psychotic symptoms, head-implanted metal devices, dermatological issues affecting electrode placement, were pregnant, or refused to use medically approved contraception (up to 24 weeks after tDCS use). In accordance with the exclusion criteria, participants with a history of dementia or neurodegenerative diseases were not included in this study. To assess potential cognitive impairments that might affect comprehension, participants received a thorough explanation and demonstration of device usage during their first session. This included step-by-step instructions on how to apply the electrodes and operate the device. Participants were required to demonstrate their ability to independently understand and operate the device after the explanation. Those who were unable to do so after repeated demonstrations were excluded from the study. However, no such cases were encountered during recruitment. Participants were considered lost to follow-up if they (1) did not visit the hospital at weeks 2, 4, or 6; (2) reported severe worsening of their psychiatric condition or serious adverse events during the trial; or (3) voluntarily withdrew their participation.
Clinical and demographic characteristics of the study sample at baseline
Abbreviations: SGAs, second-generation antipsychotics; SSRIs, selective serotonin reuptake inhibitors; HAMD, Hamilton Depression Rating Scale; K-BDI-II, Korean Beck Depression InventoryII; HAMA, Hamilton Anxiety Rating Scale; QLES, Quality of Life Enjoyment and Satisfaction Questionnaire.
a Escitalopram, fluoxetine, paroxetine, sertraline, and vortioxetine.
b Venlafaxine, desvenlafaxine, and duloxetine.
c The p-value was 0.363.
d Clonazepam, lorazepam, alprazolam, etiazolam, bromazepam, diazepam, and flunitrazepam
e Gabapentin and topiramate.
f Total accuracy was calculated as the sum of hits and correct rejections, divided by the total number of trials.
g Millisecond.
* The raw data for the 2-back test from one participant is not included due to technical error related to equipment malfunction.
Intervention
Following detailed instructions, participants used a home-based tDCS device (MINDD STIM). Participants did not meet with the researchers in person before each session but underwent comprehensive training prior to and during the intervention to ensure proper use of the device. Initial training was conducted at baseline, during which participants received step-by-step instructions on self-applying the tDCS device. This included live demonstrations by trained staff and opportunities for participants to practice under supervision. The training emphasized correct electrode placement, which followed a “one-size-fits-all” approach. Participants were instructed to position the headset approximately 1 cm above the highest point of their eyebrows to target the left and right dorsolateral prefrontal cortex. To reinforce this training and address any challenges encountered during the intervention, additional training sessions were conducted at 2 and 4 weeks. Participants who faced difficulties applying the device at home were encouraged to contact the research team for support. Guidance was provided through photo reviews, ensuring adherence to the protocol and consistent application of the device throughout the study. Researchers recorded application duration and timing. Anode and cathode electrodes were utilized to deliver stimulation following the protocol described in a previous study [Reference Park, Choi, Kim, Kim, Son and Roh29]. During the sessions, the participants received no specific instructions; they were allowed to read or use their smartphones but were not allowed to fall asleep. Participants attached 28.26 cm2 round electrodes in a “bifrontal” setup (F3-Anode, F4-Cathode). The anode and cathode were positioned over the left and right dorsolateral prefrontal cortex, respectively. Participants used single-use sponges soaked in saline solution to ensure proper electrode conductivity. They were instructed to use a mirror to confirm correct alignment of the headset on their scalp, and a photo guide was provided to assist with accurate placement. The tDCS device was programmed to monitor impedance levels in real time, allowing stimulation only when impedance was within an acceptable range of 500 Ω to 10 kΩ. If impedance exceeded this range, or patches were detached or misaligned, stimulation was blocked or stopped immediately. Participants were encouraged to use the device at approximately the same time each day for consistency in treatment schedules. To further support this, the tDCS device was programmed to block additional stimulations if insufficient time (less than 12 hours) had passed since the previous session, minimizing deviations from the recommended daily schedule. Participants were required to complete at least 9 out of 14 sessions (to complete at least 60% of the total sessions) within each 2-week period to avoid being classified as dropouts. The treatment groups received a constant current of 2 mA (high intensity) or 1 mA (low intensity) for 30 minutes, whereas the sham stimulation group received 0.1 mA of current for 30 seconds upon device activation. Participants reported adverse events using an adverse effects questionnaire [Reference Lee, Lee, Jang, You, Park and Ji30]. The questionnaire assessed symptoms as mild, moderate, severe, or none (Supplementary Table S5). A double-blind design was maintained throughout the study, ensuring that the rater, operator, and participants did not know which tDCS settings were administered. To ensure blinding, a designated administrator assigned stimulation settings to the device based on group allocation. The tDCS device automatically recorded usage history to monitor compliance.
Outcomes
All assessments were conducted by trained researchers blinded to the study conditions. Participants underwent outcome assessments at baseline and weeks 2, 4, and 6 to document cognitive function, depressive symptoms, anxiety symptoms, quality of life, and adverse events. All the assessments were conducted in person at the hospital to ensure consistency and standardization. Online or remote assessments were not conducted during the study. The N-back test is widely used to assess the working memory components of executive functions. In the 2-back test, participants respond to sequences of stimuli and identify if each item matches the one presented two positions back. The 2-back task was used to assess working memory performance. A total of 400 stimuli (letters) were presented on a computer monitor. Each stimulus was displayed for 0.5 seconds, followed by an interstimulus interval of 1.5 seconds during which a fixation cross was shown. Participants were instructed to press a button only when the current letter matched the letter presented two trials earlier (target stimuli). Non-target stimuli required no response. The task was developed using the Unity game engine and distributed as a Windows-based executable file provided by Ybrain. All the assessments were conducted in person. Task performance included three key outcomes: reaction time, omission errors (failing to respond to a target stimulus), and commission errors (responding to a distractor stimulus). Total accuracy was determined using the formula: (hits + correct rejections) / total stimuli, where hits represent targets – omission errors and correct rejections are distractors – commission errors. Increased accuracy indicates improved cognitive function. Secondary outcomes included changes in scores between groups over time on HAM-D [Reference Hamilton31] and HAM-A [Reference Hallit, Haddad, Hallit, Akel, Obeid and Haddad32]. We consistently used the 17-item HAM-D throughout the study, a clinician-rated assessment with 17 items on a 3- or 5-point Likert scale, for its reliability in assessing core depression symptoms and its widespread use in clinical trials [Reference Carrozzino, Patierno, Fava and Guidi33]. The HAM-A is a clinician-administered questionnaire used to assess anxiety severity. It comprises 14 items with a composite score ranging from 0 to 56. Scores of ≤17 suggest mild anxiety, 18–24 indicate mild-to-moderate anxiety, and 25–30 indicate moderate-to-severe anxiety, with lower scores reflecting better symptom control. Additionally, changes in scores on the Korean Beck Depression Inventory II (K-BDI-II) [Reference Lee34] and QLES [Reference Endicott, Nee, Harrison and Blumenthal35] were assessed.
Statistical analysis
Statistical analyses were performed using R, version 4.3.2 (lme4 package; R Foundation for Statistical Computing, Vienna, Austria). The sample size was estimated for a repeated-measures analysis of variance to achieve a statistical power of 90% with a 2-tailed α level of 5% and effect size of 0.20. The estimated sample size based on these assumptions was n = 111. Assuming a 25% attrition rate, 47 participants per group were required, resulting in a total sample size of 141 participants. The calculation was conducted using G*power software [Reference Faul, Erdfelder, Lang and Buchner36], informed by previous research [Reference Rauh, Müller, Nolte, Haaf, Mußmann and Steinmann37]. We performed an intention-to-treat analysis. Missing data were imputed using the last observation carried forward method. Differences in baseline clinical and demographic variables between groups were analyzed using the F-test or χ2 test for continuous and categorical variables, respectively. The Wilcoxon rank-sum test was used for non-parametric data. Mean changes in total accuracy scores from baseline to week 6 were calculated and compared between the three groups using an F-test. If a significant difference was found, then a Tukey’s post-hoc analysis was conducted. Linear mixed-effects models were used to analyze the interactive effects of group (three levels: high-intensity, low-intensity, and sham) and time (four levels: baseline; weeks 2, 4, and 6) on total accuracy, HAM-D score, and secondary outcomes. Participants were included as a random-effects variable. Additional analyses were performed after excluding outliers from the n-back task. Outliers were identified and removed using Tukey’s Fences, which define them as data points outside the range of Q1–1.5 × IQR to Q3 + 1.5 × IQR.
We carried out a subgroup analysis based on onset age (<29 or ≥29 years), current episode duration (<13 months or ≥13 months), age (<45 or ≥45 years), sex, HAM-A score (≤median or >median), and medication usage (current use of selective serotonin reuptake inhibitors, serotonin norepinephrine reuptake inhibitors, and antipsychotics). Total accuracy and HAM-D scores were used as the dependent variables in these models. Potential confounding variables, including sex, age, age at onset, current episode duration, and employment status, were adjusted for in the models. Owing to the regular 2-week assessment intervals, we applied an autoregressive covariance structure as the working correlation matrix. The frequency and severity of adverse events at weeks 2, 4, and 6 were compared between groups using the χ2 test or Fisher’s exact test (for categorical variables) and the Kruskal–Wallis test (for continuous variables). The χ2 test was also used to assess the integrity of blinding by comparing the frequency of correct guesses about treatment allocation in each group.
Results
Participants
The proportion of participants with moderate to severe depression based on baseline HAM-D score (>17) [Reference Ma, Yang, Yang, Kang, Wang and Zhang38] was 62.4%. The clinical and demographic characteristics of the study participants are presented in Table 1. Out of 141 participants, 102 (72.3%) were female. The mean age (standard deviation) was 35.7 (12.7) years. The study flowchart is shown in Figure 1 and Supplementary Table S1. Participants were divided into three groups: low-intensity (1 mA, n = 47), high-intensity (2 mA, n = 49), and sham (n = 45). These numbers reflect the actual participants enrolled in each group, as noted in Table 2. Additionally, one participant’s 2-back test data in the sham group was excluded due to a technical error related to equipment malfunction. Ninety-five (30 in the high-intensity group, 33 in the low-intensity group, and 32 in the sham group) completed the final assessment. Dropout rates were comparable across groups (38.8% in high-intensity tDCS, 29.8% in low-intensity tDCS, and 29.6% in sham; p = 0.549). Reasons for dropout included voluntary withdrawal (14 in high-intensity, 12 in low-intensity, and 12 in sham). Specific reasons included difficulty attending bi-weekly in-person visits due to scheduling conflicts, and inconvenience associated with daily device application at home. Non-compliance with protocol was defined as failing to complete at least 60% of the scheduled sessions or attending fewer than 8 sessions by the 2-week midpoint. This led to a dropout in two participants from the low-intensity group and one participant from the sham group. Clinical decisions for discontinuation were made when participants experienced significant discomfort or side effects, such as persistent headaches (n = 2) or skin irritation under the electrodes (n = 3), which occurred in five participants from the high-intensity group.
Trial flowchart. ITT, intention-to-treat.
Primary and secondary outcomes at week 6
Note: The raw data for the 2-back test from one participant are not included due to technical error related to equipment malfunction.
* p < 0.05, **p < 0.01, *** p < 0.001.
a Mean (SD).
Efficacy of tDCS on modulating cognitive function: Primary outcome
The mean change in total accuracy score from baseline to week 6 was 0.022 (0.096) in the high-active group, −0.009 (0.324) in the low-active group, and 0.105 (0.229) in the sham group (F = 2.9, p = 0.058). No significant differences were observed among groups (as presented in Table 2). We conducted an additional exploratory analysis using a linear mixed model (LMM), which revealed a significant time-group interaction for total accuracy (F = 5.236, p < 0.05, Supplementary Figure S1). However, interestingly, the sham group showed a more significant increase in the test scores than the low-intensity group at week 6. This result remained unchanged even after adding the baseline HAM-D scores as a covariate to control for depression severity (F = 5.236, p < 0.05), and removing outliers using Tukey’s Fences (F = 5.544, p < 0.05). For reaction time, the analysis showed no significant time-group interaction (F = 0.178, p = 0.672).
Efficacy of tDCS on depressive symptoms: Secondary outcomes
No significant treatment effects were observed for depressive symptoms. Across the high-intensity group, low-intensity group, and the sham group, the mean change in HAM-D scores from baseline to week 6 was −8.59 (9.18), −7.40 (8.51), and − 6.96 (6.82), respectively, (F = 0.497, p = 0.609, Table 2). The LMM analysis showed that there was no significant time-group interaction for the HAMD score (F = 1.409; p = 0.236, Figure 2). Similarly, we did not observe significant treatment effects for the secondary outcomes (F = 1.147, p = 0.321 for BDI; F = 0.164, p = 0.849 for HAM-A; F = 0.68, p = 0.508 for QLES [quality of life], Table 2). The results of the subgroup analysis for total accuracy and HAM-D scores are shown in Supplementary Tables S2 and S3.
Changes in HAM-D score over 6 weeks.
Frequency of adverse events
The mean number of reported adverse events per participant did not differ significantly between the groups (p = 0.07, Table 3). However, analysis by severity revealed a significantly higher mean number of mild adverse events in the active group than in the sham group (p < 0.01). The burning and skin redness frequencies were significantly higher in the high-activity group during weeks 0–2 (p < 0.05 and p < 0.01, respectively).
Frequency of adverse events a
a Adverse events were assessed using an adverse effects questionnaire.
b P values were determined with X 2 or Fisher exact test.
c P values were determined with X 2 or Fisher exact test and Kruskal-Wallis test.
Blinding integrity
Blinding integrity was assessed at weeks 2 and 6. A significant difference was not observed between the active and sham groups regarding the likelihood of participants correctly guessing their assigned group at week 2 (χ2 = 5.33, p = 0.07). However, by week 6, significant differences in group allocation guesses were observed (χ2 = 9.65, p < 0.05, Supplementary Tables S4), suggesting partial unblinding by the end of treatment.
Discussion
This study investigated home-based tDCS for its effects on working memory in MDD patients, with cognitive improvement designated as the primary outcome based on study registration and trial design. While no significant differences were observed among groups for cognitive or depressive outcomes, these findings contribute to understanding tDCS’s safety profile and feasibility in home-based settings. Previous studies have reported positive effects of home-based tDCS in cognitive impairments in some neurodegenerative diseases, such as mild cognitive impairment (MCI), or Alzheimer’s disease (AD). These studies have shown efficacy of home-based tDCS in improving general cognition and memory domains, while demonstrating no significant effects on executive function [Reference Park, Oh, Chung, Kim, Kim and Park24, Reference Satorres, Escudero Torrella, Real, Pitarque, Delhom and Melendez39]. The discordance in the efficacy of home-based tDCS across diseases might be associated with differences in the mechanisms underlying cognitive impairment. While neurodegenerative changes and the loss of functional connectivity between brain areas are associated with cognitive impairment in neurodegenerative diseases [Reference Liang, Wang, Yang, Jia and Li40, Reference Pelucchi, Gardoni, Di Luca and Marcello41] a combination of neurochemical imbalances, neuroplasticity deficits, and brain structure alterations, exacerbated by stress and inflammation, is thought to be the main cause of cognitive impairment in depression [Reference Hasler42–Reference Castrén, Võikar and Rantamäki44]. The difference in the tests used to measure cognition could also be a contributing factor. While the 2-back test employed in our study primarily assesses working memory, it also engages executive functions, such as attention, as part of task performance. Moreover, individuals with depression often have greater difficulty maintaining attention or executive function, whereas those with neurodegenerative diseases such as MCI or AD are more affected by memory impairments [Reference Wang, Zhou and Zhu45–Reference Lanza, Sejunaite, Steindel, Scholz and Riepe47]. It is possible that poor performance on the 2-back test is associated with impaired attention and executive function in MDD patients, which home-based tDCS did not significantly improve. Although some participants reported mild adverse events, such as burning sensations and erythema, the incidence of moderate-to-severe adverse events did not differ significantly between groups. This indicates that the home-based tDCS intervention demonstrated an acceptable safety profile for MDD treatment. Additionally, we observed no significant differences in the depression and anxiety symptoms, as measured by the HAM-D, K-BDI-II, HAM-A, and QLES (quality of life) scores, between the groups. Our findings align with recent randomized controlled trials investigating the efficacy and safety of tDCS for MDD [Reference Loo, Husain, McDonald, Aaronson, O’Reardon and Alonzo48, Reference Burkhardt, Kumpf, Crispin, Goerigk, Andre and Plewnia49]. These studies similarly reported no significant differences between active and sham groups in terms of symptom improvement. The consistent results from these trials suggest that home-based tDCS may not be superior to sham stimulation for MDD treatment.
Several factors may have contributed to the lack of significant efficacy in this study. First, the severity of the patients’ condition may have influenced the outcomes. Studies suggest that tDCS may be less effective in treating more chronic, recurrent, and severe forms of MDD. Our participants had an average of three major depressive episodes, with approximately 20% reporting previous suicide attempts and 35% having a history of hospitalization, suggesting a more severe form of MDD. This could explain the limited responsiveness to tDCS, and investigating milder cases might yield different results. However, evaluating tDCS as an adjunct treatment when medications alone are insufficient aligns with its potential clinical utility. Second, the concomitant medications taken by participants may have affected the efficacy of tDCS. Certain medications have been reported to modulate the effects of tDCS. For example, changes in the GABA/glutamate system, induced by benzodiazepines are thought to decrease tDCS effects. Additionally, reduced neuronal excitability resulting from the blockade of voltage-gated sodium channels by mood stabilizers, such as lithium, valproic acid, and lamotrigine, has been proposed to diminish the antidepressant effects of tDCS. Considering that more than 50% of participants were taking benzodiazepines and over 20% were taking mood stabilizers, the efficacy of tDCS could have been significantly influenced. Although randomization was employed, the higher rate of lithium use in the high-intensity group may have affected the observed effects. Future studies restricting concomitant medication use are necessary to elucidate the specific effects of tDCS. Third, our study did not comprehensively assess the various cognitive domains affected by depression. Therefore, future research exploring cognitive domains beyond working memory is warranted.
Home-based tDCS interventions present unique challenges. While previous tDCS studies have reported mixed results, a recent meta-analysis of tDCS combined with antidepressants showed a larger effect size for tDCS on depression scores and response rates compared to sham stimulation [Reference Wang, Luo, Schülke, Geng, Sahakian and Wang50]. However, a recent home-based tDCS trial reported negative results [Reference Burkhardt, Kumpf, Crispin, Goerigk, Andre and Plewnia49, Reference Borrione, Cavendish, Aparicio, Luethi, Goerigk and Carneiro51]. Limitations of tDCS include low spatial resolution and challenges in defining precise electrode localization [Reference Kropotov and Kropotov52]. Similar to a prior study with negative results, our unsupervised, home-based approach raises concerns about inadequate stimulation delivery due to a one-size-fits-all electrode montage.
We conducted subgroup analyses to explore the potential efficacy of tDCS within subgroups defined by clinical variables, such as the age of onset (<29 or ≥29 years), duration of the current episode (<13 months or ≥13 months), age (<45 or ≥45 years), sex, HAM-A score (≤median or >median), and concomitant medication usage. Although some significant group-by-time interactions on total accuracy scores were observed, these findings were not robust. Contrary to our expectations, the sham group, but not the active tDCS groups, demonstrated sustained improvement in total accuracy over time. This observation aligns with those of previous reports suggesting that sham stimulation in tDCS studies can be inconsistent and act as a hidden confounding variable [Reference Loo, Husain, McDonald, Aaronson, O’Reardon and Alonzo48]. Our sham stimulation method (0.1 mA for 30 seconds upon device activation) was designed to mitigate the placebo effects and maintain blinding integrity. While blinding was preserved at week 2, partial unblinding occurred by week 6, as participants in the active groups were more likely to correctly guess their allocation. This partial unblinding may have influenced participant behavior or expectations, potentially affecting cognitive performance outcomes. Future studies should consider alternative sham protocols or additional measures to enhance blinding integrity over longer treatment durations.
Our study had some limitations. First, the 6-week intervention period might have been insufficient to induce substantial changes or observe the effects of tDCS, particularly on cognitive function. Previous studies have reported significant efficacy of tDCS with an extended 10-week treatment duration [Reference Brunoni, Moffa, Sampaio-Junior, Borrione, Moreno and Fernandes53]. Additionally, meta-analyses suggest that longer treatment periods could elicit more pronounced tDCS effects, potentially needed for the persistent cognitive impairments observed in MDD compared with mood symptoms [Reference Nikolin, Moffa, Razza, Martin, Brunoni and Palm54]. Second, concomitant medication use by the participants may have influenced the effects of tDCS, as certain medications have been reported to modulate tDCS responses. Moreover, the HAM-D scores were not measured prior to the participants’ initiation of antidepressant treatment. As a result, we could not determine whether participants were responders or non-responders to their current medication regimens. This lack of data limits our ability to assess whether tDCS effects vary based on prior treatment response status. Future studies should include serial HAM-D assessments before and during antidepressant treatment to better stratify participants into responder and non-responder groups. Third, the sham stimulation setup used in our study may have compromised blinding integrity, potentially affecting the results. Fourth, our assessment was limited to working memory, neglecting evaluating other cognitive domains affected by depression. Finally, a key limitation of this study is the fixed electrode positioning, which does not account for individual anatomical variability and raises concerns about whether the targeted cortical areas were accurately stimulated. This lack of precision may contribute to the observed null results. To address this, future research should consider incorporating supervision methods, such as real-time remote monitoring systems [Reference Charvet, Shaw, Bikson, Woods and Knotkova55], or utilizing advanced tDCS devices tailored to individual neuroanatomy to ensure more accurate targeting [Reference Nikolin, Moffa, Razza, Martin, Brunoni and Palm54, Reference Woodham, Selvaraj, Lajmi, Hobday, Sheehan and Ghazi-Noori56]. These strategies could help overcome the limitations of the one-size-fits-all approach and better elucidate the therapeutic potential of home-based tDCS. Additionally, although participants were encouraged to apply tDCS at approximately the same time each day, adherence to this instruction was not strictly monitored. Variability in the session timing may have introduced inconsistencies in the stimulation effects, as prior studies suggest that circadian rhythms and time-of-day effects could influence the neuromodulation outcomes [Reference Salehinejad, Ghanavati, Kuo and Nitsche57]. The exploratory analysis using an LMM revealed a significant time-group interaction for total accuracy, but unexpectedly, the sham group showed greater improvement than the low-intensity tDCS group. The results in Supplementary Figure S1 reveal inconsistent trends in total accuracy across the four time points (baseline, weeks 2, 4, and 6) among the three groups. These inconsistencies suggest potential variability in response to active tDCS, which may be influenced by individual differences or external factors such as adherence to the protocol. The variability observed across time points might also reflect limitations in the study design, such as insufficient treatment duration or variability in home-based device application. These findings highlight the need for further research with more robust protocols and monitoring to better understand the mechanisms underlying these outcomes.
Conclusions
This randomized controlled clinical trial investigated the effects of adjunctive home-based tDCS with medication in patients with MDD. We found no significant differences between tDCS and sham stimulation for cognitive function or mood symptoms. Further research is needed to address the technical limitations, evaluate broader cognitive functions, and extend study durations to thoroughly assess the efficacy of tDCS as a potential therapeutic option for MDD.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2024.1811.
Data availability statement
The data supporting this study are available from the authors upon reasonable request to the corresponding authors (W.M. [wmyung@snu.ac.kr] or J.H.B. [jihyunbeak@skku.edu]).
Acknowledgments
None.
Author contribution
WM and JHB had full access to all data in the study and take responsibility for the integrity of the data and accuracy of the data analysis. Concept and design: WM and JHB. Data acquisition: CWL, JEA, YJ, YSP, HY, DL, HKI, JL, SSK, HYP, JK, YJIL, SL, THH, IYY, WM, and JHB. Interpretation of data: All authors. Drafting of the manuscript: CWL, KP, and YJ. Writing of the manuscript: CWL and KP. Critical revision of the manuscript for important intellectual content: WM and JHB. Statistical analysis: CWL, JK, and SL. Obtained funding: WM and JHB. Supervision: WM and JHB.
Financial support
This work was supported by a Korea Medical Device Development Fund grant funded by the Korean government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health & Welfare, the Ministry of Food and Drug Safety, Project Number: NTIS 9991006915, KMDF_PR_20200901_0250).
Competing interest
Y.I.L. and S.L. are currently employed by Ybrain. The other authors declare no conflicts of interest.
Open access
Comments
No Comments have been published for this article.