Efficacy and Safety of Home-Based Transcranial Direct Current Stimulation for Major Depressive Disorder: A Randomized, Sham-Controlled Trial

• The table demonstrates that the two treatment groups are relatively balanced at baseline on most demographic and clinical characteristics, except for ethnicity.
• The high percentage of participants on antidepressants suggests that the study sample includes a substantial proportion of individuals with treatment-resistant or persistent depression.
• The baseline data provides a clear picture of the study population's clinical profile, enabling readers to assess the generalizability and relevance of the findings.
• The lack of information on psychotherapy type limits the ability to assess the potential influence of this variable on treatment outcomes. Future studies could collect more detailed information on psychotherapy to address this limitation.

• Mean HDRS score: 19.07 ± 2.73 indicates moderate to severe depression at baseline in both groups.
• Mean age of onset of MDD: ~22 years in both groups highlights the relatively early onset of this condition.
• Percentage taking antidepressants: ~63% reflects the real-world population of MDD patients, many of whom are already receiving medication.
• Significant difference in ethnicity (P=0.012) between groups raises a potential confound that warrants further investigation and discussion.

This table presents the baseline demographic and clinical characteristics of the 174 participants enrolled in the study, stratified by treatment group (active tDCS and sham tDCS). It includes information on age, sex, ethnicity, education level, age of MDD onset, number of prior episodes, suicide attempts, clinical ratings (HDRS, MADRS, MADRS-s, HAM-A, YMRS, EQ-5D-3L, RAVLT, SDMT), and current treatment status (antidepressant medication, psychotherapy). The table uses mean ± standard deviation for continuous variables and number (percentage) for categorical variables. Median and interquartile range are provided for number of previous episodes and suicide attempts. The caption clarifies how sex was determined (self-report) and inclusion criteria related to treatment status. A footnote explains the scoring ranges for the clinical rating scales and provides additional details on the presentation of data. The reference text further explains the inclusion criteria and notes that meeting MDD criteria while on antidepressants for at least 6 weeks has been used as a criterion for treatment-resistant depression in other studies.

• The choice of baseline characteristics is relevant and comprehensive, covering demographics, clinical history, and current treatment status. This allows for assessment of potential confounding factors and subgroup analyses.
• The table is mostly transparent, providing clear definitions and units for most variables. However, more detail on how 'ethnicity' was assessed and categorized would be beneficial. Additionally, specifying the type of psychotherapy received would enhance transparency.
• The statistical significance of the difference in ethnicity between groups is noted (P=0.012), but the practical significance of this difference isn't discussed. It's unclear if this imbalance might influence the results.
• The table adheres to standard practices for presenting baseline characteristics in randomized controlled trials. Reporting both descriptive statistics and inferential statistics (p-value for ethnicity) is appropriate.

• The table is generally clear, but some terms could be more accessible to a broader audience (e.g., explaining acronyms like HDRS, MADRS, etc.). The footnote helps with interpretation, but integrating some of this information into column headers might improve readability.
• The visual organization is effective, with clear separation between groups and variable categories. However, the footnote is dense and could be visually improved by breaking it into smaller, more digestible chunks.
• The table is appropriate for a scientific audience, but a simplified version with fewer variables and clearer explanations might be beneficial for a broader audience or patient information materials.
• The table generally adheres to field conventions, but clarifying the specific versions of the rating scales used (e.g., HDRS-17) would enhance precision.

• The CONSORT diagram reveals a substantial number of individuals were screened but excluded (194/368 = 52.7%), emphasizing the importance of careful participant selection in clinical trials.
• The similar attrition rates between treatment groups strengthen the internal validity of the study by reducing the risk of bias due to differential dropout.
• The relatively high retention rate suggests good acceptability of the home-based tDCS intervention, which is an important consideration for real-world implementation.
• Providing more detailed information about exclusion and discontinuation reasons would enhance transparency and allow for a more comprehensive understanding of the study population and potential biases.

• Enrolled and randomly assigned: 174 participants represents the initial sample size after meeting inclusion/exclusion criteria.
• Excluded: 194 participants highlights the stringent screening process and the proportion of individuals who did not meet the study criteria.
• Discontinued intervention: 13 (active) and 12 (sham) shows similar attrition rates between groups, suggesting that discontinuation was not differentially influenced by treatment.
• Included in modified ITT analysis: 87 (active) and 86 (sham) represents the final sample size used for the primary analysis. This demonstrates a good retention rate overall.

The figure is a CONSORT flow diagram depicting the participant flow through the different stages of the randomized controlled trial. It starts with the initial screening and enrollment process, followed by randomization into either the active tDCS or sham tDCS group. It then tracks the number of participants who discontinued the intervention in each group, specifying reasons for discontinuation. Finally, it shows the number of participants included in the modified intention-to-treat (ITT) analysis. The caption identifies the figure as a CONSORT diagram and briefly describes its purpose. The reference text mentions the randomization process and provides the initial group sizes and mean ages.

• Using a CONSORT diagram is a standard and appropriate methodological choice for reporting participant flow in a randomized controlled trial. It enhances transparency and allows readers to understand how the final analyzed sample relates to the initially enrolled participants.
• The diagram provides a reasonable level of detail regarding the reasons for exclusion and discontinuation. However, it could be improved by providing more specific information about the 'Other medical conditions or circumstances' category that led to 4 exclusions. Also, clarifying the specific 'Change in treatment during the trial' would be helpful.
• The numbers presented in the diagram support the claims made in the reference text about the initial randomization numbers. The diagram visually represents the flow of participants, making it easier to understand the impact of exclusions and dropouts on the final sample size.
• The diagram generally aligns with CONSORT guidelines. However, it could be further improved by explicitly stating the number of participants who completed the study in each arm, separate from those included in the modified ITT analysis. This would provide a clearer picture of study completion rates.

• The diagram is generally clear and easy to follow. The use of standard CONSORT terminology and clear labeling of each stage facilitates understanding. However, using more concise language for some exclusion reasons (e.g., instead of "Not having at least a moderate depressive severity," perhaps "Mild Depression") could improve readability.
• The visual organization is effective, with a clear flow from left to right. The use of boxes and arrows clearly depicts the different stages and transitions. Color-coding the treatment arms could further enhance visual clarity.
• The diagram is appropriate for a scientific audience familiar with clinical trials and CONSORT guidelines. For a broader audience, a simplified version with less technical jargon might be beneficial.
• The diagram adheres to the standard conventions for CONSORT flow diagrams. The inclusion of specific numbers at each stage and clear labeling of reasons for exclusion/discontinuation are consistent with best practices.

Fig. 2 | Change in depressive severity ratings over time. Estimated mean 17-item HDRS rating scores from baseline to week 10 in the modified ITT analysis sample (n = 173) for the active and sham tDCS treatment arms. The error bars represent ± 1 s.e. The HDRS scores range from 0 to 52, with higher values indicating more severe depressive symptoms. A significant improvement was observed in the change in HDRS ratings from baseline to week 10 in the active tDCS treatment arm, that is, an HDRS decrease of 9.41 ± 6.25 (s.d.) (mean HDRS at week 10 = 9.58 ± 0.70 (s.e.)), compared to the sham tDCS treatment arm (HDRS decrease = 7.14 ± 6.10 (s.d.)) (mean HDRS at week 10 = 11.66 ± 0.69 (s.e.)) (95% CI = 0.5–4.0, P = 0.012). The difference in change scores was also significant at week 4 (95% CI = 0.2–3.4, P = 0.03), with a greater score decrease in the active treatment arm. A fully conditional specification (FCS) approach was used to produce 20 multiply imputed complete datasets. The FCS approach accommodates nonmonotonicity in the pattern of missing data and requires regression models to be specified for each variable, with missing values needing imputation. All models included age, sex, undergoing psychotherapy at baseline, use of any antidepressants at baseline and treatment group. The resulting complete datasets were combined using Rubin’s rules. *P < 0.05.

• The figure demonstrates that active tDCS is more effective than sham tDCS in reducing depressive symptoms over 10 weeks of treatment, supporting the primary hypothesis of the study.
• The significant difference at week 4 suggests that the treatment effect of active tDCS emerges relatively early in the course of treatment.
• The finding that sham tDCS also leads to some improvement highlights the importance of controlling for placebo effects in depression research.
• While the figure provides strong evidence for the efficacy of active tDCS, the relatively small difference between groups at week 10 suggests that further research is needed to optimize treatment parameters and maximize clinical benefits.

• HDRS decrease (active): 9.41 ± 6.25 (s.d.) represents a substantial reduction in depressive symptoms in the active tDCS group.
• HDRS decrease (sham): 7.14 ± 6.10 (s.d.) indicates a smaller but still notable improvement in the sham group, likely due to placebo effects.
• Mean HDRS at week 10 (active): 9.58 ± 0.70 (s.e.) suggests that active tDCS reduced HDRS scores to a level considered mild depression.
• Mean HDRS at week 10 (sham): 11.66 ± 0.69 (s.e.) indicates that sham tDCS resulted in a less pronounced reduction in symptoms, remaining in the moderate range.
• P = 0.012 (week 10) and P = 0.03 (week 4) demonstrate statistically significant differences between the active and sham groups at these time points.

This figure displays the change in Hamilton Depression Rating Scale (HDRS) scores over time for participants with Major Depressive Disorder (MDD) receiving either active or sham transcranial Direct Current Stimulation (tDCS). The line graph plots the estimated mean HDRS scores at baseline, week 1, week 4, week 7, and week 10 for both groups. Error bars represent standard error. The caption details the statistical analysis, including the use of a fully conditional specification (FCS) approach for multiple imputation to handle missing data, and the variables included in the regression models. It also highlights the significant difference between groups at week 10 and week 4.

• Plotting the change in HDRS scores over time is a standard and effective method for visualizing the treatment effect in depression studies. The use of a mixed model for repeated measures (MMRM) is appropriate for analyzing longitudinal data with missing values.
• The caption provides a comprehensive description of the statistical methods, including the use of FCS for multiple imputation and the variables included in the model. This level of detail enhances transparency and allows for replication. However, it could be improved by specifying the type of mixed model used (e.g., random intercept, random slope).
• The figure and caption clearly present the significant findings, including the difference in HDRS change scores at week 10 and week 4. The use of error bars and p-values strengthens the evidence supporting the claims of treatment efficacy.
• The methodology and reporting align with standard practices for clinical trials, including the use of intention-to-treat analysis and appropriate statistical methods for handling missing data.

• The figure is generally clear and easy to understand. The axes are clearly labeled, and the lines representing each group are distinct. The caption provides a detailed explanation of the figure and the statistical analysis. However, the caption is quite lengthy and dense; separating the detailed statistical explanation into a separate methods section might improve readability.
• The visual organization is effective, with clear differentiation between the active and sham groups. The use of error bars provides a visual representation of the variability within each group. However, adding a visual cue highlighting the statistically significant time points (e.g., asterisks) directly on the graph would improve its clarity.
• The figure and caption are appropriate for a scientific audience familiar with statistical methods and clinical trials. For a broader audience, a simplified figure focusing on the key findings and a less technical caption would be more accessible.
• The figure adheres to field conventions for presenting longitudinal data. The inclusion of error bars and clear labeling of axes and groups are standard practices.

Table 2 | Primary and secondary outcomes: changes in depressive severity as measured using the HDRS, MADRS and MADRS-s, and quality of life as measured using the EQ-5D-3L after a 10-week course of active or sham tDCS

• The table provides strong evidence for the efficacy of active tDCS compared to sham tDCS in reducing depressive symptoms and achieving clinical response and remission.
• The substantial difference in clinical response rates between groups, as measured by the MADRS, is a particularly compelling finding that supports the potential clinical utility of tDCS.
• The lack of significant difference in EQ-5D-3L scores suggests that while tDCS improves depressive symptoms, it may not have a substantial impact on broader quality of life measures in this study. This warrants further investigation.
• The table would benefit from clearer presentation of the statistical methods and more detailed explanations of key findings to enhance its accessibility and impact.

• HDRS decrease: 9.41 ± 6.25 (active) vs. 7.14 ± 6.10 (sham) demonstrates a greater reduction in depressive symptoms with active tDCS.
• MADRS decrease: 11.31 ± 8.81 (active) vs. 7.74 ± 8.47 (sham) further supports the efficacy of active tDCS.
• Clinical response (MADRS): 64.2% (active) vs. 32.3% (sham), OR = 3.76 (95% CI: 1.83-7.74, p = 0.0002) indicates a substantially higher response rate with active tDCS. This is a key finding highlighted in the reference text.
• Clinical remission (MADRS): 57.5% (active) vs. 29.4% (sham) shows a similar pattern of greater remission rates with active tDCS.
• EQ-5D-3L change: 0.07 ± 0.15 (active) vs. 0.07 ± 0.17 (sham) suggests no significant difference in quality of life between groups.

This table presents the primary and secondary outcome measures of the study, comparing active tDCS and sham tDCS treatment for depression. It includes the change in HDRS, MADRS, and MADRS-s scores from baseline to week 10, as well as the clinical response and remission rates for each scale. Additionally, it presents the change in EQ-5D-3L score, a measure of quality of life. The table provides means and standard deviations for continuous variables and percentages with odds ratios (ORs), confidence intervals (CIs), and p-values for categorical variables. A footnote explains the definitions of clinical response and remission for each scale and provides details about the statistical methods used, including the use of a Fully Conditional Specification (FCS) approach for multiple imputation.

• Presenting the primary and secondary outcomes in a single table is a clear and efficient way to summarize the main findings of the study. The use of appropriate statistical tests (e.g., MMRM for continuous outcomes, ORs for categorical outcomes) is methodologically sound.
• The table is generally transparent, providing clear definitions for clinical response and remission and explaining the statistical methods used for multiple imputation. However, it lacks details about the specific type of MMRM model used (e.g., random intercepts, random slopes). More information on how EQ-5D-3L change scores are interpreted would also be beneficial.
• The data presented in the table support the claims made in the reference text and caption regarding the significantly greater response rate in the active tDCS group. The inclusion of confidence intervals and p-values strengthens the evidence presented.
• The table adheres to standard reporting practices for clinical trials, including the presentation of effect sizes (Cohen's d or Number Needed to Treat (NNT)) and confidence intervals. The use of a modified intention-to-treat (ITT) analysis is also appropriate.

• The table is generally well-organized and easy to read. The use of clear headings and subheadings facilitates navigation. However, the footnote is quite dense and could be improved by breaking it into smaller, more digestible chunks. Explaining acronyms like HDRS, MADRS, and MADRS-s in the table itself, rather than just in the footnote, would improve accessibility.
• The visual presentation is effective, with clear separation between the active and sham groups and different outcome measures. However, visually highlighting the statistically significant findings (e.g., using bold font for p-values < 0.05) would enhance readability.
• The table is appropriate for a scientific audience familiar with clinical trial terminology and statistical methods. For a broader audience, a simplified table with fewer outcome measures and a less technical explanation in the footnote would be more accessible.
• The table generally adheres to field conventions for presenting clinical trial results. However, including the actual sample sizes (n) for each outcome measure within the table would enhance clarity and transparency.

• The higher rate of skin and subcutaneous tissue disorders in the active tDCS group suggests a potential safety concern that warrants further investigation and monitoring. The paper mentions two cases of reported "burns" that resolved without scarring, which likely contribute to this finding.
• The absence of serious device-related adverse events and mania/hypomania provides reassurance about the overall safety of tDCS treatment.
• The table highlights the importance of monitoring and reporting all adverse events in clinical trials, even those that are not considered serious or related to the intervention.
• Providing more detailed information about the specific adverse events within each category would enhance the transparency and clinical relevance of the safety data.

• Skin and subcutaneous tissue disorders: 17 (19.5%) in active tDCS vs. 7 (8.1%) in sham tDCS suggests a higher rate of skin-related issues with active treatment. This is a statistically significant difference (p=0.05).
• Nervous system disorders: 7 (8.0%) in active tDCS vs. 8 (9.3%) in sham tDCS shows a similar rate of nervous system events between groups.
• Serious adverse events: One hospitalization for hypertension in the active group, but this was deemed unrelated to the intervention. No deaths or new-onset mania/hypomania were observed in either group.
• No serious adverse events related to the device itself were reported, indicating a good safety profile for the tDCS device.

This table reports the unanticipated adverse events observed at 10 weeks in the active and sham tDCS groups. It lists various event categories (e.g., ear and labyrinth disorders, eye disorders, etc.) and presents the number and percentage of participants in each group experiencing at least one event within each category. It also includes the number of participants with mild, moderate, and severe adverse events, and lists any serious adverse events that occurred during the trial. A footnote clarifies how adverse events were determined and assessed, and the reference text notes the absence of serious adverse events related to the device and the absence of mania/hypomania.

• Presenting adverse events by category and severity is a standard and appropriate method for reporting safety data in clinical trials. The inclusion of both the number and percentage of participants experiencing events is helpful for understanding the frequency and distribution of adverse events.
• The table provides a reasonable level of detail about the types of adverse events observed. However, it lacks specific information about the nature and severity of individual events within each category. For example, knowing the specific types of "skin and subcutaneous tissue disorders" would be more informative. The footnote could also be more explicit about how participant-reported adverse events were adjudicated by the investigators.
• The table supports the reference text's claim about the absence of serious device-related adverse events and mania/hypomania. However, the table itself doesn't provide data on mania/hypomania; this information presumably resides in the referenced supplementary tables.
• The table generally aligns with standard reporting practices for adverse events in clinical trials. However, it would be strengthened by including a more comprehensive description of the methods used to collect and assess adverse events, including the timing and frequency of assessments.

• The table is generally clear and easy to follow. The use of categories and subheadings makes it easy to locate specific information. However, some medical terms (e.g., "infections and infestations," "injury, poisoning and procedural complications") might be difficult for a lay audience to understand.
• The visual organization is adequate, but could be improved. Visually highlighting statistically significant differences between groups (e.g., using bold font or asterisks) would enhance readability. Additionally, presenting the data for "Number of participants with adverse events" as percentages, consistent with the rest of the table, would improve consistency.
• The table is appropriate for a scientific audience familiar with medical terminology and clinical trial reporting. For a broader audience, a simplified table with clearer explanations of medical terms and a focus on the most common or relevant adverse events would be more accessible.
• The table generally adheres to field conventions for reporting adverse events. However, it could be improved by adding a column indicating the total number of adverse events reported in each group, regardless of category. This would provide a clearer overall picture of the safety profile of each treatment.

Table 4 | Anticipated adverse events at 10 weeks as measured using the tDCS Adverse Events Questionnaire 39

• The significantly higher rate of skin redness in the active tDCS group confirms that this is a common side effect of active stimulation, consistent with previous research. The paper mentions two specific cases of skin "burns" that resolved, which likely contribute to this finding.
• The increased incidence of trouble concentrating in the active group raises a potential cognitive side effect of tDCS that warrants further investigation, although the relatively low incidence and lack of statistical significance in other cognitive events (e.g., acute mood change) suggest it may not be a major concern.
• The lack of significant differences in other common tDCS side effects (e.g., headache, itching) between groups suggests that these events may not be specifically related to active stimulation and could be due to other factors.
• The table provides valuable information about the safety and tolerability of tDCS treatment, which is important for informing clinical practice and future research. However, more detailed information about the assessment and interpretation of adverse events would strengthen the findings.

• Skin redness: 63.5% (active) vs. 18.5% (sham), p<0.001 indicates a significantly higher rate of skin redness in the active tDCS group.
• Trouble concentrating: 14.1% (active) vs. 3.7% (sham), p=0.03 suggests a higher rate of trouble concentrating in the active group, although the absolute difference is relatively small.
• Itching: 50.6% (active) vs. 43.2% (sham), p=0.08 shows a numerically higher rate of itching in the active group, but the difference is not statistically significant.
• Headache: 42.4% (active) vs. 35.8% (sham), p=0.43 indicates a similar rate of headaches between groups.

This table presents the anticipated adverse events at 10 weeks, assessed using the tDCS Adverse Events Questionnaire. It lists common side effects associated with tDCS (e.g., headache, neck pain, scalp pain, itching, burning sensation, skin redness, sleepiness, trouble concentrating, acute mood change) and reports the number and percentage of participants in both the active and sham tDCS groups experiencing each event, broken down by severity (total, mild, moderate, severe). P-values are provided for each adverse event, comparing the total incidence between groups. The caption specifies the questionnaire used, and the reference text reiterates the absence of serious adverse events related to the device and the lack of mania/hypomania.

• Using a standardized questionnaire (tDCS Adverse Events Questionnaire) is a methodologically sound approach for collecting data on anticipated adverse events in a tDCS study. Reporting the frequency and severity of each event allows for a comprehensive assessment of the safety profile of the intervention.
• The table is fairly transparent, providing clear definitions of the adverse events assessed and reporting the data by severity. However, it would be more informative to include the specific questions or criteria used to define each adverse event within the questionnaire. The footnote could also be more explicit about how adverse event severity was determined (participant-reported or investigator-assessed).
• The data presented in the table support the general claim in the reference text about the safety of the tDCS device, as no serious device-related adverse events were reported. The table also provides evidence for a higher incidence of certain adverse events (skin redness, trouble concentrating) in the active tDCS group, which is important for understanding the potential risks associated with active treatment.
• The table generally aligns with standard reporting practices for adverse events in clinical trials. However, it would be strengthened by providing more details about the administration and scoring of the tDCS Adverse Events Questionnaire, including the timing of assessment and any specific instructions given to participants.

• The table is generally clear and well-organized, with clear headings and subheadings. However, the presentation could be improved by using bold font or other visual cues to highlight statistically significant differences between groups. Also, some terms (e.g., "acute mood change") could be more clearly defined.
• The visual organization is adequate, but could be enhanced. Using color-coding or other visual cues to differentiate between the active and sham groups would improve readability. Additionally, reordering the rows to group similar adverse events together (e.g., all skin-related events) might be helpful.
• The table is suitable for a scientific audience familiar with tDCS and adverse event reporting. For a broader audience, a simplified table focusing on the most common or relevant adverse events, with clearer explanations of medical terms, would be more accessible.
• The table generally adheres to field conventions for reporting adverse events. However, it would be beneficial to include a column with the total number of participants assessed for adverse events in each group (n=87 for active, n=86 for sham). This would clarify the denominator used for calculating percentages and facilitate interpretation.

Extended Data Fig. 1 | Change in Montgomery-Åsberg Depression Rating Scale (MADRS) ratings over time. Estimated mean MADRS rating scores from baseline to week 10 in the modified intention-to-treat analysis sample (n = 173) in active tDCS and sham tDCS treatment arms. Error bars represent ± 1 standard error (SE). MADRS scores range from 0 to 60 with higher values indicating more severe depressive symptoms. A significant improvement was observed in the change in MADRS ratings from baseline to week 10 in the active tDCS treatment arm, MADRS change 11.31 ± 8.81 (standard deviation (SD)) (mean week 10 MADRS 12.46 ± 1.09 (SE)) as compared to sham tDCS treatment arm, MADRS change 7.74 ± 8.47 (SD) (mean week 10 MADRS 15.30 ± 1.07 (SE)) (95% CI 1.1 to 6.1, p = 0.006). The difference in change scores was also significant at week 4 (95% CI 1.2 to 5.5, p = 0.003) and week 7 (95% CI 1.1 to 5.8, p = 0.005) with a greater score decrease in the active treatment arm. Fully Conditional Specification (FCS) approach was used to produce 20 multiply imputed completed data sets. The FCS approach accommodates nonmonotonicity in the pattern of missing data and requires regression models to be specified for each variable with missing values needing imputation. All models included age, sex, in psychotherapy at baseline, use of any antidepressants at baseline and treatment group. The resulting completed datasets were combined using Rubin's Rules. ** = p < 0.01.

• The figure provides strong evidence for the efficacy of active tDCS compared to sham in reducing depressive symptoms as measured by the MADRS, consistent with the findings using the HDRS.
• The significant differences at weeks 4 and 7 suggest that the treatment effect of active tDCS emerges relatively early and is sustained throughout the 10-week trial.
• The finding that sham tDCS also leads to some improvement reinforces the importance of placebo control in depression research.
• The figure would benefit from some presentational improvements to enhance its clarity and accessibility for a wider audience.

• MADRS change (active): 11.31 ± 8.81 (SD) represents a substantial improvement in depressive symptoms from baseline to week 10 in the active tDCS group.
• MADRS change (sham): 7.74 ± 8.47 (SD) indicates a smaller improvement in the sham group, suggesting a placebo effect.
• Mean week 10 MADRS (active): 12.46 ± 1.09 (SE) suggests that active tDCS reduced MADRS scores to a level generally considered mild depression.
• Mean week 10 MADRS (sham): 15.30 ± 1.07 (SE) indicates that sham tDCS resulted in less improvement, with scores remaining in the moderate range.
• p = 0.006 (week 10), p = 0.003 (week 4), p = 0.005 (week 7) demonstrate statistically significant differences between the active and sham groups at these time points.

This figure presents a line graph illustrating the change in MADRS scores over time for both the active and sham tDCS groups. The x-axis represents time in weeks (0, 1, 4, 7, and 10), and the y-axis represents the estimated MADRS score. Error bars indicate standard error. The caption provides detailed information about the significant findings, including the mean change in MADRS score, the estimated mean MADRS score at week 10 for both groups, 95% confidence intervals, and p-values. It also explains the statistical method used (FCS for multiple imputation) and the variables included in the imputation model.

• Using a line graph to visualize change over time is appropriate and allows for easy comparison of treatment effects between groups. The use of MMRM (implied by the use of FCS multiple imputation) is a valid approach for analyzing longitudinal data with missing values.
• The caption is thorough in explaining the statistical methods, including the FCS approach and the variables included in the imputation model. This enhances transparency and reproducibility. However, it could be strengthened by explicitly stating that a mixed model for repeated measures (MMRM) was used for the analysis, as FCS is a method for handling missing data within that framework. Also, the caption mentions "Rubin's Rules" but doesn't explain what they are; a brief explanation would be helpful for readers unfamiliar with multiple imputation.
• The figure and caption clearly present the significant findings, including the differences in MADRS change scores at weeks 4, 7, and 10. The use of error bars, confidence intervals, and p-values supports the claims of a treatment effect.
• The methodology and reporting generally align with standard practices for clinical trials. However, clarifying the specific type of MMRM model used (e.g., random intercepts, random slopes) would enhance the methodological rigor.

• The figure is generally clear and easy to interpret. The axes are labeled, and the lines for each group are distinct. The caption provides a comprehensive description of the findings and statistical methods. However, the caption is quite lengthy and dense. Breaking it down into smaller paragraphs or moving some of the statistical details to the methods section would improve readability.
• The visual presentation is effective in showing the trend of MADRS scores over time. The use of error bars provides a visual representation of variability. However, adding visual cues directly on the graph to highlight statistically significant time points (e.g., asterisks) would improve clarity.
• The figure and caption are appropriate for a scientific audience familiar with statistical methods and clinical trials. For a broader audience, a simplified figure focusing on the key findings and a less technical caption would be more accessible.
• The figure generally adheres to field conventions for presenting longitudinal data. However, adding the sample sizes (n) for each time point directly on the graph would enhance transparency.

Extended Data Fig. 2 | Change in Montgomery-Åsberg Depression Rating Scale-Selfreport (MADRS-s) ratings over time. Estimated mean MADRS-s rating scores from baseline to week 10 in the modified intention-to-treat analysis sample (n = 173) for the active tDCS and sham tDCS treatment arms. Error bars represent ± 1 standard error (SE). MADRS-s scores range from 0 to 60 with higher values indicating more severe depression. A significant improvement was observed in the change in MADRS-s ratings from baseline to week 10 in the active tDCS treatment arm, MADRS-s change 9.90 ± 8.94 (standard deviation (SD)) (mean week 10 MADRS-s 16.60 ± 1.18 (SE)) as compared to sham tDCS treatment arm, MADRS-s change 6.23 ± 9.13 (SD) (mean week 10 MADRS-s 19.55 ± 1.16 (SE)) (95% CI 0.9 to 6.4, p = 0.009). The difference in change scores was also significant at week 4 (95% CI 0.3 to 4.9, p = 0.030) with a greater score decrease in the active treatment arm. Fully Conditional Specification (FCS) approach was used to produce 20 multiply imputed completed data sets. The FCS approach accommodates nonmonotonicity in the pattern of missing data and requires regression models to be specified for each variable with missing values needing imputation. All models included age, sex, in psychotherapy at baseline, use of any antidepressants at baseline and treatment group. The resulting completed datasets were combined using Rubin's Rules. *= p <0.05, **= p < 0.01.

• The figure supports the study's primary hypothesis, showing active tDCS is more effective than sham in reducing self-reported depressive symptoms (MADRS-s).
• The significant difference at week 4 suggests a relatively early onset of the treatment effect.
• The persistent moderate symptom levels at week 10, even in the active group, suggest a need for further research to optimize tDCS treatment and achieve greater symptom reduction.
• The figure's presentation could be improved for broader accessibility.

• MADRS-s change (active): 9.90 ± 8.94 (SD) shows a significant reduction in self-reported depressive symptoms in the active tDCS group.
• MADRS-s change (sham): 6.23 ± 9.13 (SD) indicates a smaller reduction in the sham group, likely due to placebo effects.
• Mean week 10 MADRS-s (active): 16.60 ± 1.18 (SE) suggests a moderate level of residual depressive symptoms, even with active treatment.
• Mean week 10 MADRS-s (sham): 19.55 ± 1.16 (SE) indicates higher residual symptoms in the sham group.
• p = 0.009 (week 10) and p = 0.030 (week 4) demonstrate statistically significant differences between groups at these time points.

This figure displays the change in self-reported MADRS (MADRS-s) scores over 10 weeks for participants with MDD receiving active or sham tDCS. It's a line graph with time (weeks) on the x-axis and estimated MADRS-s scores on the y-axis. Error bars represent standard error. The caption details the results, including mean change in MADRS-s, estimated mean score at week 10, confidence intervals, and p-values. It also describes the statistical approach (FCS for multiple imputation) and lists variables in the imputation model.

• Visualizing change in self-reported symptom scores (MADRS-s) over time via a line graph is a standard and appropriate method. Using FCS multiple imputation within an MMRM framework is a valid way to handle missing data in longitudinal studies.
• The caption thoroughly describes the statistical procedures, including FCS and the variables considered for imputation. However, explicitly mentioning "mixed model for repeated measures (MMRM)" would enhance clarity, as FCS is a component of that broader analytical framework. A brief explanation of "Rubin's Rules" would also benefit readers unfamiliar with multiple imputation.
• The figure and caption clearly present the significant findings, including the changes in MADRS-s at week 4 and week 10. Error bars, confidence intervals, and p-values strengthen the evidence presented.
• The methodology and reporting are generally consistent with established practices for clinical trials. Specifying the type of MMRM (e.g., random intercepts, random slopes) would further improve the methodological rigor.

• The figure is generally clear and understandable, with labeled axes and distinct lines for each group. The caption provides a detailed explanation. However, the caption's length and density could be improved by condensing it or moving some statistical details to the methods section.
• The visual organization effectively shows the MADRS-s trends over time. Error bars represent variability. Adding visual cues (e.g., asterisks) directly on the graph to indicate statistically significant time points would enhance clarity.
• The figure and caption suit a scientific audience familiar with statistical methods and clinical trials. A simplified version with key findings and less technical language would be more accessible to a broader audience.
• The figure adheres to conventions for longitudinal data presentation. Including sample sizes (n) for each time point on the graph would improve transparency.