Effect of tDCS Timing on Safety Memory in PTSD: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effect of tDCS Timing on Safety Memory in PTSD (NCT04152772), a randomized na study. The primary outcome is Skin Conductance Responses.

2. Clinical Question

3. Trial Design

The registry record describes a randomized, triple-masked, parallel-group study sponsored by Butler Hospital. Its phase is listed as NA, and its status is completed. The recorded start date is 2019-11-22 and the primary completion date is 2024-01-31. The registry lists enrollment as 62. The brief summary describes a plan to randomize 90 participants equally across three groups. The record does not explain the difference between that planned number and the listed enrollment.

The three main-protocol groups differed in when active versus sham tDCS was administered. One received active stimulation during extinction learning and sham stimulation during consolidation. A second received sham during extinction and active stimulation during consolidation. A third received sham during both stages. The outcome record lists group sizes of 19, 19, and 21 for the primary skin-conductance measure. These outcome counts are separate from the enrollment field; the registry record does not explain their relationship.

The brief summary also describes an online sub-study, identified as Aim 2, on contextual processing across the PTSD spectrum. It was intended to examine associations between threat and non-threat learning in contextual and non-contextual situations, with a maximum of 500 participants recruited online through a panel-based platform. The posted primary outcome concerns the main protocol, not those online questions, and no secondary outcomes are posted for this record.

Although randomization, masking, and parallel allocation are stated, the registry record does not report the details of sequence generation, concealment, masking procedures, or deviations from assigned stimulation. It also does not report the original sample-size justification or the planned statistical analysis in enough detail to assess them.

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.

Overall Study

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 Skin Conductance Responses, measured over Study visit day 4: during administration of fear extinction memory..

The primary endpoint is titled Skin Conductance Responses. It is reported as a mean, with skin conductance measured in microsiemens, and with standard errors as the accompanying measure of variability. The assessment took place during administration of fear-extinction memory on study visit day 4.

In the main protocol description, skin conductance is used to quantify threat reactivity: responses to a conditioned stimulus that was subsequently extinguished are considered in relation to responses to a never-conditioned stimulus. In general, skin conductance is a physiological measure of autonomic arousal. It offers a measurable response during a learning task, rather than relying only on a participant’s verbal account of fear or safety learning.

The posted outcome is a group mean, not an individual-level record or a full description of response patterns. The registry record does not report separate numerical results for the conditioned-and-extinguished and never-conditioned stimuli, nor does it specify in the outcome description how their responses were combined into the reported measure. It also does not provide a clinically meaningful threshold for interpreting a difference in microsiemens. Consequently, the means describe the recorded measure, but their clinical significance cannot be determined from the reported endpoint alone.

7. Results Posted in the Registry

• Sham tDCS During Extinction Followed by Active tDCS During Consolidation: mean 0.09 versus 0.07 on control; difference 0.01 (95% CI -0.06 to 0.08), standardized effect d = 0.12; z-test p = 0.719.
• Sham tDCS During Extinction Followed by Sham tDCS During Consolidation: mean 0.06 versus 0.07 on control; difference -0.02 (95% CI -0.09 to 0.05), standardized effect d = -0.16; z-test p = 0.611.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a group mean represent?

A mean is a summary of the observed endpoint values in a group. In this trial, the outcome is skin conductance in microsiemens, summarized as a mean for each stimulation schedule. A mean can make groups easier to compare, but it conceals the spread and shape of individual observations. The registry reports standard errors alongside the means. In general, a standard error describes the uncertainty in an estimated mean, rather than the range containing most individual participants’ measurements. It should not be read as a standard deviation or as a confidence interval by itself.

The posted means are 0.075, 0.088, and 0.057 for the three stimulation schedules, with reported standard errors of 0.025, 0.026, and 0.025. The group sizes for this endpoint are 19, 19, and 21, respectively. These summaries do not reveal individual values, distributions, missing observations, or whether unusual measurements influenced the means. Those details are not reported in the registry record.

What does a mean difference describe?

A mean difference expresses one group mean relative to another on the outcome’s original scale. The recalculated comparisons use the active-during-extinction and sham-during-consolidation group as the reference. The comparison with sham during extinction followed by active during consolidation is reported as 0.013. The comparison with sham at both stages is reported as -0.018. In general, a positive difference means the first named comparison group’s mean is higher than the reference mean when the difference is calculated in that direction; a negative value means it is lower.

These are simple, unadjusted recalculations from the posted summary statistics. They are not the sponsor’s mixed-model estimates. They do not account for any model structure, covariates, or other adjustments that might have been used in the sponsor’s analysis. Details of those model features are not reported in the registry record. The recalculated p-values are also explicitly unadjusted: they are 0.719 and 0.611 for the two comparisons.

How should a confidence interval be read?

In general, a confidence interval expresses the range of values reasonably compatible with data under a specified statistical procedure. A narrower interval indicates greater precision for that procedure; a wider interval indicates less precision. It is not a guarantee that the true value lies within the particular interval, nor does it measure the probability that a result is clinically important.

The recalculated intervals for both pairwise mean differences extend from negative to positive values and include zero. That means the calculations do not precisely establish the direction of either pairwise difference. The registry record does not report the confidence level for these intervals. It also does not report an interval or effect estimate for the sponsor’s mixed-model result, so the sponsor analysis’s precision cannot be assessed from the posted record.

What do the p-values address?

In general, a p-value measures how compatible data are with a statistical null model, given the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not quantify the size or practical value of an effect. The recalculated pairwise p-values, 0.719 and 0.611, are unadjusted results from comparisons based on the posted summary statistics. The sponsor reports p = 0.02 for a mixed models analysis across all three groups, classified as superiority.

Those p-values answer different statistical questions because the sponsor’s model and the recalculations are not the same analysis. The sponsor’s result takes precedence where the record indicates a discrepancy. However, without the model specification, estimated contrast, and confidence interval, the registry record does not permit a detailed assessment of the sponsor’s effect magnitude or its uncertainty.

Why can model choice matter?

General statistical principle: a mixed model can account for structure in data, such as repeated observations or grouping, when that structure is specified in the model. The name of the analysis alone does not disclose which structure, covariance assumptions, covariates, or contrasts were used. Those details are not reported in the registry record. The primary outcome description refers to responses to different stimulus types, but it does not say whether the sponsor’s model used repeated stimulus-level observations or how those observations were represented.

General statistical principle: when data are over-dispersed relative to a model’s assumptions, standard errors and tests can be misleading if the extra variation is not handled appropriately. Over-dispersion is most directly discussed for count outcomes; the posted endpoint is a continuous mean. The registry record does not report diagnostics or model checks, so there is not enough information to assess whether the assumptions of the sponsor’s model were evaluated or whether any particular variance issue arose.

Why are standardized effects not a substitute?

The recalculation includes standardized effect estimates, but a standardized effect is not the same as the endpoint’s mean difference. In general, standardization expresses a difference relative to a measure of variability, which can help compare results measured on different scales. It may be less directly interpretable than the original microsiemens scale, and it depends on the variability estimate used. The exact standardized estimates in this record are not needed to interpret the means and are not repeated. They also do not resolve the discrepancy with the sponsor’s mixed-model analysis.

10. Confidence Intervals and Precision

The recalculated mean differences are 0.013 for sham during extinction followed by active during consolidation versus active during extinction followed by sham during consolidation, and -0.018 for sham at both stages versus the same reference group. The registry reports interval bounds for each recalculated contrast. In both cases, the bounds extend across zero, indicating that the recalculated results are compatible with differences in either direction under the interval method.

In general, an interval is more informative than a point estimate alone because it communicates uncertainty about the estimated contrast. Here, the intervals do not pin down a clear direction, and their width indicates that the pairwise estimates are imprecise. The confidence level used for the posted intervals is not reported in the registry record. The sponsor’s mixed-model analysis has a reported p-value of 0.02, but its effect estimate and confidence interval are not reported, so its magnitude and precision cannot be assessed from the registry result.

The uncertainty is relevant to interpretation: the recalculated point estimates are small on the reported microsiemens scale, but the registry record does not supply a clinically meaningful threshold for this outcome. A statistical interval cannot establish whether a difference matters in practice without that context. The posted means and standard errors also do not show the distribution of individual responses or the precision of stimulus-specific measurements.

These precision statements apply to the simple recalculations from the posted summary statistics, not to the sponsor’s mixed model. Because the record identifies a discrepancy, the sponsor’s analysis takes precedence for the trial’s own reported analysis. Its uncertainty remains difficult to judge because the corresponding estimate and interval are not reported in the registry record.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed pairwise effects, not the original design calculation for this trial. For this continuous outcome, the figures are participants per arm, rather than total events. The registry record’s planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The assumed effect to detect is the stated fraction of the observed effect.

The 80% power column gives the illustrative number of participants per arm that the planning calculation estimates would be needed for an 80% chance of detecting the specified effect under those assumptions. The 90% power column gives the corresponding per-arm number for a 90% chance. These are planning probabilities conditional on the assumed effect, significance level, allocation, and calculation method; they do not describe the probability that the trial’s result is correct.

The planning rows consider 100%, 75%, and 50% of the observed effect. Smaller assumed effects require more participants to detect them at the same stated power and significance level. The two pairwise comparisons have different observed effects and therefore different planning figures. The figures are illustrations based on those recalculated effects, not evidence that the original study was designed around them. The trial’s own original sample-size assumptions are not reported in the registry record.

This is not observed power. Observed power is not a useful substitute for examining the effect estimate and its uncertainty after results are available; the planning figures instead show how sample-size requirements change under stated assumptions. They should be read as hypothetical design guidance, not as a retrospective rating of the completed trial.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.013; pooled standard deviation σ = 0.111.
2. n = 2 × (1.96 + 0.842)² × 0.111² / 0.013² = 1,148.0, rounded up to 1,149 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.013                      # observed difference in means
sigma <- 0.111173288158622                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(4152772)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 1,149 (80% power), 1,537 (90% power)
 75% of observed effect: n per arm = 2,041 (80% power), 2,733 (90% power)
 50% of observed effect: n per arm = 4,593 (80% power), 6,148 (90% power)
power.t.test: 1149 
Simulated power at n = 1,149 per arm: 0.794 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The sponsor reports a mixed models analysis across the three groups, classified as superiority, with p = 0.02. The registry record does not report the model’s estimated effect, confidence interval, or detailed specification. As a result, the direction and magnitude associated with that reported test cannot be fully assessed from the posted sponsor result alone.

The recalculated pairwise comparisons are simple, unadjusted comparisons based on the reported group summary statistics. They give p = 0.719 for sham during extinction followed by active during consolidation versus active during extinction followed by sham during consolidation, and p = 0.611 for sham at both stages versus that reference group. Their confidence-interval bounds span zero. These recalculations do not match the sponsor’s analysis and should not be represented as the sponsor’s estimates.

The record flags a discrepancy, so the sponsor’s own analysis takes precedence. The difference cannot be explained from the registry record: the exact mixed-model specification, analyzed population, handling of missing observations, and planned contrasts are not reported. The appropriate reading is therefore to distinguish the sponsor’s three-group model result from the two unadjusted recalculations, rather than treating them as interchangeable or combining their conclusions.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists one primary outcome measure and no posted secondary outcomes. It also lists one reported analysis, the sponsor’s mixed models analysis across the three groups. These entries do not establish that only one statistical test was conducted, because the registry record does not describe every analysis that may have been planned or performed. The number and handling of interim analyses are not reported in the registry record.

Multiplicity arises when several outcomes, comparisons, time points, or analyses are considered: repeated testing can increase the chance of a nominally notable result even when no underlying effect exists. General statistical practice may address this with a prespecified testing strategy or adjustment, but the registry record does not report whether a multiplicity plan was specified or applied. The sponsor’s p = 0.02 should therefore be understood as the reported result, without assuming that it was adjusted or unadjusted for any unreported testing.

The two posted pairwise recalculations are explicitly unadjusted. Their p-values are 0.719 and 0.611. They are not a substitute for the sponsor’s model, and the record does not report a multiplicity correction for them. The sponsor’s contrast structure and whether its reported test was an overall test or another model-based test are not detailed beyond the mixed-model method and superiority classification.

Missing-data handling and the analysis population are also not reported in the registry record. It does not state whether all randomized participants were included, whether the outcome was available for every participant, how missing measurements were addressed, or whether the analysis followed an intention-to-treat approach. The outcome-group sizes and safety at-risk counts differ, but the record does not explain those count differences or identify the reason for them. Without these details, it is not possible to evaluate potential bias from missing outcome data or departures from the assigned groups.

Other statistical-plan details—including covariate adjustment, model diagnostics, prespecified contrasts, and the handling of protocol deviations—are not reported in the registry record. These gaps limit reproducibility and make it difficult to determine how closely the posted analysis followed a prespecified plan.

15. 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

The registry reports zero serious adverse events in each of the three safety groups. The numbers at risk for serious adverse events are 19 in the active-during-extinction and sham-during-consolidation group, 20 in the sham-during-extinction and active-during-consolidation group, and 23 in the sham-both group. Deaths are also recorded as zero in each group.

These figures describe what was recorded in the registry’s safety fields. A zero count of reported serious events does not establish that an intervention is free of risk, particularly when the groups are limited in size and the record does not describe event ascertainment in detail. The registry does not list serious-event terms in its posted summary.

The record separately lists other adverse events affecting 16, 18, and 19 participants across the three groups, with respective at-risk counts of 19, 20, and 23. These are in the non-serious adverse-event fields and should not be confused with the serious-event counts. The record does not provide further detail about event types, severity beyond that classification, or causal attribution. The safety counts alone do not determine whether an event was caused by stimulation.

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

The posted outcome describes average skin-conductance responses during administration of fear-extinction memory on study visit day 4. The three group means are 0.075 for active tDCS during extinction followed by sham during consolidation, 0.088 for sham during extinction followed by active during consolidation, and 0.057 for sham at both stages. The sponsor reports a three-group mixed models superiority analysis with p = 0.02.

Those results do not, by themselves, establish the clinical importance of the group differences. The sponsor’s model estimate and confidence interval are not reported, and the registry does not define a meaningful threshold on the skin-conductance scale. The separate pairwise recalculations have p-values of 0.719 and 0.611, with intervals spanning zero; because the record flags a discrepancy, the sponsor’s analysis takes precedence. The different analyses should remain clearly distinguished.

The outcome is a physiological measure, and the record does not provide separate values for responses to the conditioned-and-extinguished stimulus and the never-conditioned stimulus. It therefore does not show which component, if any, accounts for the reported group means. Nor does the result alone establish a change in symptoms or functioning: those outcomes are not posted as secondary measures in this record.

The record describes a randomized, triple-masked, parallel-group design, but it does not report enough detail about the analysis population, missing data, model specification, or multiplicity procedures to fully evaluate the sponsor’s result. The appropriate interpretation is limited to the posted endpoint summaries and reported analyses. No treatment recommendation or conclusion about clinical benefit follows from these statistics alone.

17. Why This Trial Matters Statistically

This trial raises a statistical question about timing: active tDCS was assigned during extinction learning in one group and during consolidation in another, with a sham-both group as a further comparator. That structure makes it important to distinguish comparisons between particular schedules from an overall analysis across all three groups. A single overall model result and pairwise contrasts answer different questions and need not produce the same p-values.

The record illustrates why an effect estimate and its precision matter alongside a test result. The sponsor’s mixed-model analysis is reported with p = 0.02, but its estimated contrast and confidence interval are not reported. The recalculated pairwise differences are small on the measured scale and their intervals include zero, but those calculations are not the sponsor’s analysis. The record flags the difference, so the sponsor analysis takes precedence while its magnitude remains difficult to assess.

The outcome also highlights the need for clear endpoint definitions. The protocol description refers to responses to a conditioned, then extinguished stimulus versus a never-conditioned stimulus, while the posted endpoint is a group mean. Without separate stimulus-level results or an explanation of the calculation, readers cannot determine how the two response types contribute to the summary.

Finally, the difference between the planned enrollment described in the summary and the registry enrollment field, as well as differing outcome and safety counts, makes transparent participant accounting important. Those discrepancies do not by themselves establish a flaw, but the registry record does not reconcile them. Clear reporting of the analysis population, missing-data approach, model specification, and planned contrasts would make the results easier to evaluate and reproduce.

18. 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 contains several count fields that do not align transparently. It lists enrollment as 62, while the brief summary describes a plan to randomize 90 participants equally among three groups. The endpoint group sizes are 19, 19, and 21, and the safety at-risk counts are 19, 20, and 23. The record does not explain how these figures relate to one another, whether they reflect different analysis populations, or why the listed counts differ. They should not be combined or treated as interchangeable.

The sponsor’s result is reported only as a mixed models superiority analysis with p = 0.02. An effect estimate, confidence interval, model specification, and analysis population are not reported. The unadjusted recalculations do not match the sponsor analysis; the record identifies this discrepancy, so the sponsor’s analysis takes precedence. The available information does not explain why the results differ.

The endpoint description does not provide numerical results separately for the conditioned-and-extinguished and never-conditioned stimuli, and it does not detail how those observations were summarized. A clinically meaningful threshold for skin conductance is not reported. Confidence levels for the recalculated intervals are also not reported.

The registry does not report missing-data procedures, multiplicity handling, interim analyses, detailed eligibility or participant characteristics, or the original sample-size justification. Although an online sub-study is described in the summary, no secondary outcomes are posted. These gaps restrict assessment of generalizability, analysis choices, and the broader learning questions described in the study summary.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Effect of tDCS Timing on Safety Memory in PTSD — NCT04152772.
• Registry record last updated 2025-10-21; status last verified 2025-10.
• Results first posted 2025-10-21.
• 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.

22. Record Summary

NCT04152772 is a completed, randomized, triple-masked, parallel-group study sponsored by Butler Hospital. It examines tDCS timing in PTSD, comparing active stimulation during extinction followed by sham during consolidation, sham during extinction followed by active during consolidation, and sham at both stages.

The primary outcome is mean skin conductance during fear-extinction memory administration on study visit day 4. The sponsor reports a mixed models superiority analysis with p = 0.02, without an estimate or confidence interval. Unadjusted recalculations from the posted group summaries differ from that analysis; the registry record flags a discrepancy, so the sponsor’s analysis takes precedence.

The registry lists enrollment as 62, while the brief summary describes a plan for 90 participants. The record does not reconcile those figures or the separate endpoint and safety counts. It also does not report the sponsor model details, analysis population, missing-data approach, or multiplicity procedures. An online contextual-processing sub-study is described, but no secondary results are posted.