Cerebellar Transcranial Direct Current Stimulation and Aphasia Treatment: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cerebellar Transcranial Direct Current Stimulation and Aphasia Treatment (NCT02901574), a randomized na study. The primary outcome is Change in Accuracy of Naming Untrained Pictures (Philadelphia Naming Test:PNT) Pre-treatment to Post-treatment.

2. Clinical Question

3. Trial Design

The registry record identifies this as a randomized, quadruple-masked crossover study, with an enrollment of 32 and a completed status. The sponsor is Johns Hopkins University. The recorded start date is 2016-09-19 and the primary completion date is 2022-05-20. The study is listed as phase NA.

The two treatment conditions paired computerized naming therapy with either anodal or cathodal cerebellar tDCS, or sham tDCS. The active condition is described as stimulation to the right cerebellum at 2 milliamps (mA), alongside 15 sessions of naming treatment over 3-5 weeks. Each treatment session lasted 45 minutes, including 20 minutes of computerized naming treatment. The sham condition used the same naming treatment schedule, with a brief ramp-like application of current that was then decreased to 0 mA.

In general, a crossover design allows comparisons between conditions within participants, but the sequence of conditions, any washout period, and the timing between treatment periods are not reported in the registry record. The record labels the study crossover but does not provide enough detail to describe how that feature was implemented or how the analysis accounted for period or sequence effects. The meaning of quadruple masking in practice, including which groups of people were masked, is also not reported in the registry record.

The primary endpoint was change in accuracy of naming untrained pictures, measured before and after 15 sessions. The registry lists further outcomes concerning naming accuracy and functional communication at post-treatment and follow-up assessments. The analysis population for the primary endpoint is described as completers, with missing data not imputed. The registry record does not report the allocation procedure, a sample-size justification, or a fuller statistical analysis plan.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 3 participants who were randomized did not start the intervention due to the COVID-19 pandemic and 1 did not start due to transportation issues.

Intervention Period 1 (up to 5 Weeks)

Wash Out ( 8 Weeks)

Intervention Period 2 (up to 5 Weeks)

Analysis population for the primary outcome, as defined in the registry: Completers sample. Missing data is not imputed. Data was missing for 3 participants for tDCS and 1 participant for sham

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Change in Accuracy of Naming Untrained Pictures (Philadelphia Naming Test:PNT) Pre-treatment to Post-treatment, measured over Before and after 15 sessions (3-5 sessions per week; up to 5 weeks) of treatment..

The primary endpoint is the change in accuracy of naming untrained pictures on the Philadelphia Naming Test (PNT), measured before and after treatment. The PNT is described as a 175-item picture-naming test. A correct answer earns one point, scores range from 0-175, and higher scores are associated with better performance. The endpoint therefore captures performance on naming items that were not trained during the computerized therapy, rather than only performance on practiced items.

Change from before to after treatment is used to describe how naming accuracy moved over the treatment period. Comparing that change between the tDCS-plus-therapy and sham-plus-therapy conditions addresses the study's primary question: whether adding cerebellar stimulation was associated with a different change than naming therapy paired with sham stimulation. Both conditions include naming therapy, so the comparison concerns the additional stimulation condition rather than therapy versus no therapy.

The endpoint is a continuous score, and the registry reports group means with 95% confidence intervals. Those summaries describe average change and its uncertainty; they do not show how individual participants' scores varied unless individual-level distributions are also reported. The registry record does not give further detail on the handling of scores at the scale limits or on other endpoint-processing decisions.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a continuous primary outcome, arm-level means, 95% confidence intervals, and a mixed-model comparison. Each method answers a different statistical question. The arm summaries describe average change within each treatment condition. The model-based contrast addresses the comparison between conditions under the model used by the study analysis. Confidence intervals communicate uncertainty in estimates, while the p-value summarizes evidence against a specified null model under the analysis assumptions.

What does a mean change describe?

A mean is the arithmetic average of participants' endpoint values or changes. For a change score, the average summarizes how scores moved between the pretreatment and post-treatment assessments in the analyzed group. It is not a description of every participant's response: participants may have changes both larger and smaller than the average, and some may change in the opposite direction. The registry's arm summaries do not include a measure of spread around the mean beyond their confidence intervals.

The reported group means are 5.76 for tDCS plus computerized naming therapy and 1.88 for sham plus computerized naming therapy. The primary endpoint is a score on a scale, and the PNT description says higher scores indicate better performance. The meaning of a group average should therefore be kept tied to this score and this assessment period, rather than interpreted as a measure of all aspects of language or communication.

What does a confidence interval convey?

A confidence interval gives a range of values compatible with the observed data under the statistical procedure and its assumptions. The 95% confidence intervals for the arm-level means are 1.2 to 10.32 for tDCS plus therapy and -1.95 to 5.72 for sham plus therapy. The 95% confidence interval for the reported model-based difference is -2.55 to 10.30. These intervals describe uncertainty in estimates, not the range of individual participants' outcomes.

For the comparison, the interval includes zero and extends to values on either side of zero. That pattern indicates that the estimate is not precise enough to rule out a range of possible comparative results. A confidence interval should not be read as assigning a probability to the true value lying within that particular calculated interval. Its long-run interpretation depends on the statistical procedure and assumptions being appropriate.

How should the mixed-model result be read?

The registry names the method as a Mixed Models Analysis and identifies its parameter as a mean difference in final values. The reported estimate is 3.87, with a 95% confidence interval from -2.55 to 10.30 and p = 0.24. A mixed model is a general class of models that can represent relationships among observations and incorporate modeled sources of variation. The registry does not report the model formula, covariance structure, fixed or random effects, covariates, or exact handling of the crossover periods. It is therefore not possible to determine from the record which particular model features were used.

In a crossover study, observations from the same participant under different conditions may be related. General statistical practice accounts for that dependence rather than treating all observations as independent. Period and sequence effects can also matter in crossover studies. The registry record does not report whether or how these issues were modeled, so no specific claim about the treatment of within-participant correlation or period effects can be made from the posted description.

What does the p-value add?

The p-value is a summary calculated under a null hypothesis and the assumptions of the selected analysis. Here, the reported value is 0.24. It does not give the probability that the null hypothesis is true, the probability that the result occurred by chance in an unrestricted sense, or the size of any treatment effect. Its interpretation depends on the model and the assumptions behind the test. The estimate and confidence interval are needed to understand the direction and precision of the comparison.

A p-value alone cannot distinguish a small effect estimated precisely from a larger effect estimated imprecisely. In this result, the reported interval itself includes both negative and positive values, making the uncertainty in the estimated comparison directly visible. The registry record does not report an adjustment for multiple comparisons or the primary analysis's prespecified significance threshold.

Why can over-dispersion matter?

For count outcomes, over-dispersion means that variation is greater than a simple model assumes; ignoring it can make uncertainty estimates too small. This primary endpoint is a continuous PNT score, not a count of events for the treatment comparison, so over-dispersion is not the central issue for interpreting the posted primary result. General statistical concepts should be matched to the type of outcome rather than applied mechanically.

The sample-size assumptions recorded for planning illustrations use a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the effect to detect defined as a stated fraction of the observed effect. Such illustrations are planning exercises, not observed power and not evidence that the trial was designed using those assumptions. The registry record does not report the trial's own sample-size justification. Any planning numbers associated with those assumptions should be read as hypothetical illustrations rather than as observed trial findings.

10. Analyses Reported by the Sponsor

The registry's reported analysis is a Mixed Models Analysis, with the parameter described as a mean difference in final values. It reports an estimate of 3.87, a 95% confidence interval from -2.55 to 10.30, and p = 0.24. Separately, the arm summaries report mean changes of 5.76 for tDCS plus therapy and 1.88 for sham plus therapy. A comparison of those arm summaries is not necessarily the same estimand as a model-based difference in final values, particularly in a crossover design.

The registry record does not provide enough model detail to reproduce the mixed-model result from the arm summaries. A simple unadjusted recalculation, when available, should be understood as a check based on posted summary statistics rather than as a replacement for the study's analysis. If a discrepancy between such a recalculation and the sponsor's result were flagged, the sponsor's own analysis would take precedence. No discrepancy is flagged in this record.

The distinction does not remove the uncertainty in the sponsor-reported result. Its confidence interval includes negative and positive values, and the reported p-value is 0.24. The registry record does not provide further detail about model specification, covariate adjustment, or crossover-period handling that would support a more complete reconstruction.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies the primary endpoint as change in naming accuracy from pretreatment to post-treatment. It also lists secondary outcomes concerning naming accuracy and functional communication, including assessments at post-treatment and follow-up. The record identifies 12 outcome measures and indicates 11 secondary outcomes posted. When several outcomes or assessment times are analyzed, the chance of finding an apparently notable result somewhere among them can increase. Statistical approaches to multiplicity can address this issue, but the registry record does not report whether any adjustment or hierarchy was prespecified or used.

The record lists one analysis for the primary result. That does not, by itself, establish that only one statistical test was conducted across all outcomes, assessment times, or exploratory analyses. The full statistical analysis plan and any interim-analysis procedures are not reported in the registry record. No assumptions about those procedures can therefore be made.

The primary endpoint analysis is described as using a completers sample, and missing data were not imputed. The registry reports missing data for 3 participants in the tDCS condition and 1 participant in the sham condition. The reported arm summaries each have n = 23. The registry record does not explain the relationship between those analysis counts, the enrollment of 32, and the missing observations; it also does not report reasons for missing data or sensitivity analyses evaluating their impact.

Complete-case analysis can be affected when the availability of outcome data is related to participants' outcomes or characteristics. The absence of imputation means missing endpoint values were not filled in by an imputation model for this analysis, but it does not establish that missingness had no effect. The record does not report whether the missingness mechanism was assessed, or whether other analysis populations were used. These limitations matter when interpreting the precision and applicability of the primary estimate.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

For serious adverse events, the registry reports 0 affected participants among 26 at risk in the tDCS-plus-therapy group and 0 among 24 at risk in the sham-plus-therapy group. Deaths are reported as 0 in both groups. No top serious-event terms are listed. These figures describe the events recorded for the stated groups and denominators; they do not establish that the risk of a rare event is zero.

For other adverse events, 9 participants are reported as affected among 26 at risk in the tDCS-plus-therapy group, and 9 among 24 at risk in the sham-plus-therapy group. The registry record does not give the event terms in this summary, so the nature, severity, timing, and relation of these events to treatment cannot be characterized from these figures alone. Nor does the record specify whether an affected participant could contribute more than one event.

The adverse-event denominators are distinct from the primary endpoint analysis counts. The safety counts should therefore be interpreted using their own stated groups and denominators, not substituted for the outcome-analysis sample. A comparison of these totals alone does not establish whether the conditions differ in safety, particularly without event details and a prespecified comparative analysis.

13. What the Result Does, and Does Not, Mean

The posted primary results describe average PNT score changes during the study's treatment period and a model-based comparison between tDCS paired with naming therapy and sham paired with naming therapy. The tDCS group mean change was 5.76, and the sham group mean change was 1.88. The registry's mixed-model comparison reports a mean difference in final values of 3.87, with a 95% confidence interval from -2.55 to 10.30 and p = 0.24. These are the results of the recorded analysis, not a guarantee about what would be observed in other participants or settings.

The confidence interval spans negative and positive differences. The result therefore does not precisely determine the direction or size of the comparative effect. The estimate should be interpreted together with that uncertainty, the completers analysis population, and the un-imputed missing data. It should not be reduced to a binary label based only on the p-value.

The comparison is between two conditions that both include computerized naming therapy. It addresses the added stimulation condition in the context of that therapy; it does not compare naming therapy with no treatment. The outcome is accuracy on a picture-naming test for untrained pictures. The registry lists functional communication outcomes separately, so the primary naming result should not be treated as a direct measure of every communication ability or of all possible clinical outcomes.

The study is recorded as a crossover trial, but sequence, washout, and analysis details are not reported in the registry record. Those details would help readers assess whether the model appropriately reflects the design. The available figures also do not reveal individual variation, reasons for missing outcomes, or the effect of missingness on the estimate.

Accordingly, the posted result is best read as an uncertain estimate from this study's recorded analysis. It neither proves a comparative benefit nor establishes the absence of a difference. It also does not, by itself, answer questions beyond the measured outcome, assessment period, and study population.

14. Why This Trial Matters Statistically

This study illustrates why a treatment comparison should be read through both its estimate and its uncertainty. The group means describe average change, while the mixed-model estimate describes a comparison under the analysis used by the registry. The reported confidence interval extends across negative and positive values, so the point estimate alone would give an incomplete account of the evidence.

The crossover designation makes design details particularly relevant. In general, crossover studies can use within-participant comparisons, but interpretation depends on how condition order, periods, and possible carryover are addressed. The registry record does not report these details or the model specification, leaving readers unable to fully evaluate how the stated design relates to the reported comparison.

The endpoint also illustrates the importance of matching conclusions to what was measured. The PNT assesses picture-naming accuracy for untrained items, while the registry lists functional communication as a separate outcome area. A change in one measure does not automatically describe change in the other. The shared naming-therapy component also means the primary comparison concerns stimulation added to therapy, rather than therapy versus no therapy.

Finally, the completers analysis and the reported missing observations show why analysis populations matter. The primary result is based on available completer data without imputation, and the registry does not report reasons for missingness or sensitivity analyses. These details shape how precisely and broadly the result can be interpreted. Together, the design, endpoint, model-based comparison, and missing-data record provide a useful example of how registry results require more than a single estimate or p-value for careful statistical reading.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record leaves important design and analysis details unspecified. Although the study is identified as randomized, quadruple-masked, and crossover, it does not report the randomization procedure, treatment sequence, washout interval, masking roles, or how period and sequence effects were handled. It also does not provide the mixed-model specification. These omissions limit independent assessment of whether the analysis reflects the crossover design and how the reported estimate was obtained.

The primary analysis used a completers sample without imputation. Data were missing for 3 participants in the tDCS condition and 1 participant in the sham condition, but reasons for missingness and sensitivity analyses are not reported in the registry record. The record also does not explain how the reported primary analysis counts relate to enrollment of 32. The consequences of missing data therefore cannot be evaluated from the posted summary alone.

The primary outcome is a score on the PNT, and the registry gives a description of its scale and direction. It does not report fuller information about score distributions, individual participant changes, or any assessment of the assumptions underlying the model. The mixed-model result is labelled as a mean difference in final values, whereas the arm summaries are changes; further detail would be needed to understand the exact relationship between these summaries.

The registry lists secondary outcomes and reports 11 secondary outcomes posted, but does not describe a multiplicity strategy, a complete analysis plan, or interim analyses. A sample-size justification for the trial is not reported in the registry record. Safety summaries provide counts but no serious-event terms and no detail on the other adverse events. These limitations do not invalidate the posted numbers, but they restrict the conclusions that can be drawn from them.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Cerebellar Transcranial Direct Current Stimulation and Aphasia Treatment — NCT02901574.
• Registry record last updated 2023-04-21; status last verified 2023-02.
• Results first posted 2023-04-21.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry record describes NCT02901574, “Cerebellar Transcranial Direct Current Stimulation and Aphasia Treatment,” as a completed randomized, quadruple-masked crossover study sponsored by Johns Hopkins University. It reports enrollment of 32 and dates of 2016-09-19 for study start and 2022-05-20 for primary completion.

The study compared computerized naming therapy paired with anodal or cathodal cerebellar tDCS against naming therapy paired with sham stimulation. The primary endpoint was change in accuracy on the Philadelphia Naming Test for untrained pictures. The reported arm mean changes were 5.76 and 1.88; the mixed-model comparison was 3.87, with a 95% confidence interval from -2.55 to 10.30 and p = 0.24.

The primary endpoint analysis used a completers sample without imputation. The registry reports missing data for 3 participants in the tDCS condition and 1 participant in the sham condition. The record does not report a trial sample-size justification or sufficient detail to reconstruct the crossover model.