Neurofeedback of Amygdala Activity for Post-traumatic Stress Disorder: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Neurofeedback of Amygdala Activity for Post-traumatic Stress Disorder (NCT03574974), a randomized na study. The primary outcome is Change in Control Over the Amygdala.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group study with quadruple masking. The registry lists the phase as NA and the enrollment as 27. The study began on 2018-06-01, had a primary completion date of 2022-03-08, and is listed as terminated. The registry record does not report the reason for termination.

Participants were assigned to Experimental Feedback Intervention or Control Feedback Intervention. Both groups received real-time feedback during fMRI scans; the registry describes the experimental feedback as one the investigators believed might help participants learn to control PTSD symptoms, and the control feedback as one they did not believe could help those symptoms. The record identifies the study as randomized and quadruple-masked, but does not describe the allocation procedure or specify which people were masked.

The primary outcome results include 14 participants in the experimental group and 11 in the control group. These are the outcome-specific group counts, and the record does not explain why they differ from one another or from the listed enrollment. The safety fields list 15 and 12 participants at risk in the two groups. The registry record does not report how the analysis populations for efficacy and safety were defined or reconciled.

The study also had secondary outcomes relating to PTSD symptoms and resting-state connectivity to the amygdala. The registry lists four outcome measures and one analysis. It does not report the original statistical design assumptions, a sample-size justification, or a full protocol for outcome assessment.

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 Change in Control Over the Amygdala, measured over baseline; 30 day-day follow up.

The primary endpoint is titled “Change in Control Over the Amygdala.” It measures change in the BOLD signal in an individually localized region of interest in the amygdala. Specifically, the measure concerns down-regulation blocks following symptom provocation, relative to rest. The registry describes the score as based on beta values and therefore unitless.

The comparison is the change from baseline to the 30-day post-treatment follow-up. The endpoint is intended to quantify amygdala regulation in response to trauma reminders, rather than directly quantify PTSD symptoms. The study description also identifies fMRI as a way to investigate brain function and amygdala connectivity in people with PTSD.

As a general statistical point, a physiological measure can address a different question from a symptom scale: it may capture activity in a targeted brain region, while a symptom measure captures reported or assessed clinical features. Neither measure alone establishes how meaningful a change is to participants. The registry record does not report a threshold for a clinically meaningful change, nor does it describe how a particular numerical direction should be interpreted clinically. The endpoint therefore needs to be read as the study’s operational measure of amygdala control, not as a direct measure of symptom improvement.

7. Results Posted in the Registry

• Experimental Feedback Intervention: mean -0.28 versus 0.20 on control; difference -0.48 (95% CI -0.92 to -0.05), standardized effect d = -0.88; z-test p = 0.030.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Improvements in PTSD Symptoms

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Experimental Feedback Intervention: mean 8.79 versus 10.54 on control; difference -1.76 (95% CI -11.20 to 7.68), standardized effect d = -0.15; z-test p = 0.715.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Experimental Feedback Intervention and the control. The interval gives the range of effects that remain compatible with the data.

The listed secondary outcomes concern PTSD symptoms and resting-state connectivity to the amygdala. The registry lists “Improvements in PTSD Symptoms” twice, while one numerical set of symptom results is posted. A numerical result for changes in resting-state connectivity to the amygdala is not reported in the registry record. The secondary entries therefore do not provide a complete numerical account of every listed outcome.

The available symptom comparison does not show a clear statistical separation between groups: its unadjusted p-value is 0.715 and its recalculated confidence interval includes zero. This is not evidence that the groups are equivalent. The interval allows a range of possible differences, and the record does not report a threshold for interpreting symptom changes as clinically meaningful.

The primary outcome concerns a unitless fMRI measure of amygdala regulation; the symptom measure addresses a different aspect of the study question. A difference in one does not automatically imply a difference in the other. The posted pattern should therefore be read as endpoint-specific rather than as a single summary of all effects. In particular, the primary comparison cannot substitute for the unreported connectivity result, and the available symptom result does not resolve the primary physiological comparison.

As a general statistical point, secondary outcomes often involve multiple opportunities to observe a nominally notable result. When p-values are unadjusted, they do not account for that multiplicity. The registry record does not report a correction procedure or a hierarchy for the secondary outcomes. Those results consequently carry less confirmatory weight than a clearly prespecified primary analysis, especially when the number and relationship of all tested outcomes are not fully described.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Improvements in PTSD Symptoms
• Improvements in PTSD Symptoms
• Number of Participants With Changes in Resting State Connectivity to the Amygdala

9. Statistical Interpretation

10. Statistical Methods Explained

The registry reports a two-sided t-test for the primary comparison. The recalculated summaries treat the outcome as continuous and compare group means. The estimated mean difference is defined as the experimental group mean minus the control group mean. For this endpoint, the estimate is -0.481, with an unadjusted p-value of 0.03. The reported sponsor analysis gives p = 0.047. The posted record does not give the test statistic or degrees of freedom for the sponsor analysis.

What does a two-sided test ask?

As a general statistical explanation, a two-sided test evaluates departures in either direction from a null hypothesis of no difference. It does not assume in advance that the experimental group must have a higher or lower mean. A p-value describes how unusual data at least as inconsistent with the null would be under the test’s assumptions; it is not the probability that the null hypothesis is true, and it does not measure clinical importance.

What does the mean difference represent?

The mean difference puts the groups on the endpoint’s original scale. Its sign depends on the subtraction order: the experimental group is subtracted from the control group as experimental minus control. A negative estimate therefore means that the experimental group’s observed mean change was lower than the control group’s. The record does not define whether a lower value on this endpoint is clinically preferable, so the sign should not be translated into a claim of benefit.

The primary outcome is reported as a mean with standard error. As a general explanation, standard error describes uncertainty in an estimated mean, whereas standard deviation describes variation among individual observations. They are not interchangeable. The primary arm summaries list standard errors of 0.146 and 0.167. The secondary symptom summaries instead report standard deviations of 9.744 and 13.44. The recalculation uses the posted summary information; it is not an adjusted model estimate.

Why report both an estimate and a p-value?

A p-value compresses evidence against a specified null into a single measure, while the estimated difference gives the direction and scale of the observed contrast. Neither conveys uncertainty as fully as a confidence interval. A confidence interval gives a range of values compatible with the estimate and model assumptions. In general, a wider interval indicates less precision. Statistical significance alone does not establish that an effect is large, clinically important, or reproducible.

The sponsor’s analysis is reported as a two-sided t-test and has p = 0.047. The recalculated unadjusted test based on posted summaries has p = 0.03. These are distinct entries in the registry record. The record does not give enough methodological detail to explain the difference, so the recalculation should not be presented as if it were the sponsor’s analysis or as a replacement for it.

What can the recalculation establish?

As a general point, recalculations from group summaries can check the rough relationship between the reported means, uncertainty, and contrast, but they do not reproduce an individual-participant analysis. They cannot account for adjustments, covariates, missing-data methods, or other features unless those are represented in the posted summaries. The registry record does not report whether such features were used in the sponsor’s analysis.

The primary endpoint is a continuous measure, so the mean difference is a natural way to describe the contrast. General statistical considerations still apply: means can be affected by unusual values, and a t-test’s behavior depends on its assumptions and the data. The registry record does not report distribution checks or robust alternatives. With modest group counts, the reported summaries alone cannot show whether those assumptions held.

For the secondary symptom outcome, the posted comparison is also a mean difference. Its p-value is unadjusted. As a general matter, when several outcomes or analyses are examined, unadjusted p-values do not account for the increased chance of at least one apparently notable result. The record does not report a multiplicity procedure, so the secondary results warrant more caution than a single prespecified primary test would receive. The record also does not report a detailed statistical analysis plan, interim analyses, or the method used to handle missing outcome data.

11. Confidence Intervals and Precision

The primary estimated mean difference is -0.481. The recalculated confidence interval lies entirely below zero, extending from a value close to -0.916 to a value close to -0.046. This interval conveys uncertainty about the size of the difference: values across a noticeable range remain compatible with the recalculation. The interval does not cross zero, but that fact alone does not establish clinical importance or resolve the appropriate interpretation of a lower endpoint value.

As a general statistical explanation, a confidence interval is a range produced by a method that, under its assumptions and repeated use, has a specified coverage property. It is not a guarantee that the true effect lies within the interval, nor is it a range of likely individual outcomes. Its width reflects both the variability of the data and the amount of information available. Here, the group counts are 14 and 11, and the uncertainty should be considered alongside the estimated contrast.

The sponsor-reported result gives p = 0.047 but no confidence interval or estimate in the analysis entry. The recalculated contrast and interval come from the posted group summaries and should be understood as an unadjusted reconstruction. The registry record does not report a sponsor-analysis interval that could be compared directly with the recalculated one.

For the secondary symptom comparison, the recalculated interval runs from -11.2 to 7.683, spanning zero and a range of possible directions. As a general interpretation, an interval that includes zero is not proof of no difference; it indicates that the data are compatible with zero as well as with effects on either side. The registry record does not report a clinically meaningful threshold for either outcome, so precision cannot be judged against a stated clinical margin.

12. 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 display gives planning illustrations in participants per arm, not total events. Its assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect equal to a stated fraction of the observed effect. The original trial’s own sample-size assumptions and justification are not reported in the registry record.

For the primary outcome, planning for 100% of the observed effect gives 21 participants per arm for 80% power and 28 participants per arm for 90% power. Planning for 75% of the observed effect gives 37 participants per arm for 80% power and 49 participants per arm for 90% power. Planning for 50% of the observed effect gives 82 participants per arm for 80% power and 110 participants per arm for 90% power.

As a general explanation, 80% power means that a study planned under the stated assumptions would have an 80% chance of detecting the specified effect at the chosen significance level, if that effect and assumptions were correct. The 90% column uses a higher target chance of detection. Smaller effects require more participants because they are harder to distinguish from random variation. The displayed counts are planning illustrations based on the observed effect, not observed power and not a reconstruction of the original design.

13. 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 = -0.481; pooled standard deviation σ = 0.550.
2. n = 2 × (1.96 + 0.842)² × 0.550² / -0.481² = 20.5, rounded up to 21 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 <- -0.481                      # observed difference in means
sigma <- 0.549596769182327                     # 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(3574974)
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 = 21 (80% power), 28 (90% power)
 75% of observed effect: n per arm = 37 (80% power), 49 (90% power)
 50% of observed effect: n per arm = 82 (80% power), 110 (90% power)
power.t.test: 21.5 
Simulated power at n = 21 per arm: 0.795 (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.

14. Analyses Reported by the Sponsor

The registry’s sponsor-reported primary analysis is a two-sided t-test labeled as a superiority analysis, with p = 0.047. It does not include an estimate or confidence interval in the analysis entry. The posted arm summaries allow a simple unadjusted recalculation: the mean difference is -0.481, the recalculated confidence interval lies below zero, and the recalculated p-value is 0.03.

The sponsor-reported and recalculated p-values are not identical. The registry record does not report enough detail about the sponsor analysis to explain the difference, including whether it used adjustments or other features not represented in the posted summaries. The recalculation should therefore be read as an independent, summary-based check, not as a reproduction of every aspect of the sponsor’s analysis.

The registry record does not flag a discrepancy. Both entries are part of the posted record, and the sponsor-reported result remains the trial’s reported analysis. Neither p-value establishes clinical importance, and interpretation should also consider the estimated difference, its uncertainty, group counts, endpoint meaning, and limitations in the available analysis details.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary endpoint and lists secondary outcomes, but it does not report a detailed statistical analysis plan or a multiplicity adjustment. It is therefore not reported in the registry record whether the primary outcome was the sole confirmatory test, whether secondary outcomes followed a prespecified hierarchy, or whether any family-wise error or false-discovery procedure was planned. The sponsor analysis is labeled as a superiority analysis, but that label does not supply the missing details.

As a general statistical explanation, testing several outcomes can increase the chance of obtaining at least one small p-value even when none of the underlying effects is real. Adjustments or a prespecified testing sequence can address that issue, but no such approach is described in the registry record. The unadjusted secondary p-value should be interpreted in that context, rather than as independent confirmation of the primary result.

The record does not report how missing outcome data were handled, whether all randomized participants were included in the efficacy analysis, or what analysis population was used. It also does not describe imputation, exclusions, adherence requirements, or sensitivity analyses. The primary-outcome counts are 14 and 11, while the listed enrollment is 27; the record does not explain the difference. The safety at-risk counts are 15 and 12, and the record does not explain how those populations relate to the outcome analysis populations.

It is also not reported in the registry record whether there were interim analyses, stopping rules, covariate adjustments, or checks of statistical assumptions. The study is listed as terminated, but the reason and any relationship to analysis decisions are not reported. These omissions limit assessment of whether the posted results reflect the planned analysis and how robust they are to alternative reasonable choices. They do not establish that any particular procedure was or was not used; they identify details unavailable in the registry record.

16. 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 5% reporting threshold.

The registry safety fields list 15 participants at risk in the Experimental Feedback Intervention group and 12 in the Control Feedback Intervention group. For serious adverse events, the number affected is 0 in each group. For other adverse events, the number affected is also 0 in each group, and deaths are recorded as 0 in both groups. No serious adverse-event terms are listed.

These are the counts recorded in the safety fields; they do not describe event types or severity beyond the categories shown. The at-risk counts differ from the primary-outcome group counts of 14 and 11. The registry record does not explain that difference or report how safety follow-up was defined. As a general statistical point, zero recorded events in a study does not establish that an event is impossible or that a treatment has no safety risks; the observation is limited to the participants and reporting period represented in the record.

The safety figures should be interpreted as descriptive registry counts, not as a comparative safety analysis. No inferential comparison or detailed account of adverse-event collection is reported in the registry record.

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

The posted primary data show different mean changes between the two feedback groups on the unitless measure of amygdala control. The estimated experimental-minus-control difference is -0.481. The sponsor-reported two-sided test has p = 0.047, while the unadjusted recalculation from posted summaries has p = 0.03. The recalculated interval lies below zero, but the estimate’s clinical meaning is not established by statistical significance alone.

The measure is based on BOLD activity in an individually localized amygdala region during down-regulation following symptom provocation, relative to rest. It is not itself a direct score of PTSD symptoms. The available secondary symptom comparison does not show a clear statistical separation, and a numerical connectivity result is not reported in the registry record. These distinct outcomes should not be collapsed into one claim about the study’s overall effect.

The study’s results do not establish that neurofeedback improves PTSD symptoms, nor do they establish that it does not. The outcome group counts are limited, the statistical detail is incomplete, and the registry does not state what magnitude of change would be clinically meaningful. The numerical direction of the primary endpoint also cannot be labeled favorable or unfavorable from the available description alone.

As a general interpretation, a test result is evidence about a particular outcome, comparison, population, and analysis under assumptions. It does not automatically generalize to other settings or answer questions that were not measured. The registry results should be understood as a record of this trial’s posted comparisons, with uncertainty and missing methodological detail kept in view. They are not a substitute for a complete account of the protocol, individual-level data, or broader evidence.

18. Why This Trial Matters Statistically

This trial illustrates why a biological endpoint and a symptom endpoint need to be distinguished. The primary measure captures a targeted fMRI signal during a defined regulation task, while the secondary symptom measure addresses a clinical construct. A statistical difference on one does not guarantee a corresponding difference on the other, and the posted results show why interpretation should remain specific to each outcome.

It also illustrates the value of reading an estimated contrast with its uncertainty rather than relying on a p-value alone. The primary contrast is negative, and the recalculated confidence interval lies below zero, but the record does not define the clinical meaning of that direction or give a threshold for meaningful change. The sponsor-reported test and simple summary-based recalculation also have distinct p-values, while the record does not provide enough analysis detail to reconcile them.

Finally, the trial highlights how small outcome counts and incomplete reporting constrain statistical interpretation. The record does not explain the analysis populations, missing-data approach, multiplicity handling, or the trial’s original sample-size assumptions. The planning illustrations can show how required group sizes change under specified assumptions, but they cannot retrospectively establish the design rationale or the certainty of the observed result. These are general lessons in evaluating clinical trial evidence, not conclusions about the merits of any particular intervention.

19. 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 primary-outcome groups include 14 and 11 participants, despite a listed enrollment of 27. The registry record does not explain the participant flow, the relationship between enrollment and outcome counts, or whether any participants lacked usable data. The safety at-risk groups are listed as 15 and 12, with no explanation of how those counts relate to the efficacy populations.

The record does not report the original sample-size justification, detailed analysis plan, missing-data methods, analysis population, interim analyses, stopping rules, or multiplicity adjustment. It also does not provide a numerical result for the listed resting-state connectivity outcome. The reason the study is listed as terminated is not reported in the registry record.

The primary measure is unitless and the record gives no clinically meaningful threshold or explicit interpretation of whether a higher or lower change is preferable. The sponsor’s reported analysis entry provides a p-value but no estimate or confidence interval; the summary-based recalculation cannot supply undisclosed features of that analysis. The record also does not specify which roles were covered by quadruple masking. These limitations restrict evaluation of the size, clinical relevance, and robustness of the posted findings.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Neurofeedback of Amygdala Activity for Post-traumatic Stress Disorder — NCT03574974.
• Registry record last updated 2024-02-15; status last verified 2024-02.
• Results first posted 2023-10-18.
• 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.

23. Record Summary

NCT03574974 is a randomized, quadruple-masked, parallel-group study of real-time amygdala neurofeedback during fMRI for PTSD. The registry lists enrollment of 27, phase NA, and a terminated status. The primary outcome is change in control over the amygdala from baseline to 30-day follow-up.

The posted primary group means were -0.279 for Experimental Feedback Intervention and 0.202 for Control Feedback Intervention, with an estimated mean difference of -0.481. The sponsor-reported two-sided t-test has p = 0.047; the summary-based unadjusted recalculation has p = 0.03. A secondary symptom result is posted, but a numerical connectivity result is not reported in the registry record.

The record does not report the original sample-size rationale, detailed missing-data or multiplicity methods, or the reason for termination. Interpretation should account for the outcome group counts and the limits of the available analysis details.