Cognitive Training for PTSD: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cognitive Training for PTSD (NCT03316196), a randomized na study. The primary outcome is Clinician Administered PTSD Scale - Reexperiencing.

2. Clinical Question

This is a question about a specific symptom domain and its measured change over the intervention period, not a direct test of every PTSD symptom or of longer-term outcomes. The posted comparison is a mean difference between the two groups. Its interpretation depends on the scale’s direction, the observed values, and the uncertainty around the difference. Random assignment is intended to support a fair comparison, while masking can reduce some sources of bias; the registry identifies the trial as randomized and triple masked but does not report who was masked.

3. Trial Design

The registry describes NCT03316196, “Cognitive Training for PTSD,” as a completed, randomized, triple-masked, parallel-group trial. Its phase is listed as NA. The listed sponsor is the VA Office of Research and Development. The registry gives a start date of 2018-06-13 and a primary completion date of 2023-05-31.

The two behavioral conditions were COGENT cognitive training and a non-training computer condition. The latter was described as matched for time and memory demands. The brief summary frames the intervention as computer-based exercises targeting aspects of cognitive control, with a proposed schedule of twice per week for eight weeks. These descriptions characterize the intended programs; further details about adherence, delivery, or deviations from the schedule are not reported in the registry record.

The registry’s enrollment is 81. The outcome-specific groups have different reported sample sizes: the primary outcome lists 35 for COGENT and 28 for Non-Training. Those outcome counts should not be treated as interchangeable with enrollment or with the denominators used for other outcomes. The record does not explain why those denominators differ.

Triple masking is recorded, but the registry record does not identify which participants, investigators, assessors, or other roles were masked. The record also does not report the original sample-size justification, the detailed allocation procedure, or an analysis plan. The trial description therefore establishes the broad randomized comparison and intervention structure, but it does not resolve all questions about implementation or analysis.

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 Clinician Administered PTSD Scale - Reexperiencing, measured over Pre to Post Intervention (baseline to week 8).

The primary endpoint is the Clinician Administered PTSD Scale (CAPS) Reexperiencing score. The registry describes it as a semi-structured interview measure of PTSD diagnostic status and symptom severity. The reexperiencing score ranges from 0 to 20, with higher scores indicating greater severity. The listed timeframe is pre to post intervention, from baseline to week 8, and the registry says change over time will be evaluated.

Reexperiencing is a defined symptom domain rather than a measure of every aspect of PTSD. A domain-specific score can make an intervention question more focused: the study can assess whether the measured symptom result differs between assigned conditions over the stated period. That focus does not establish that the endpoint captures all clinically relevant changes, nor does it answer whether any change persists beyond the stated assessment window.

The posted primary outcome is summarized as a mean and standard deviation for each group, with a mean difference comparison. These summaries describe the group-level results, not the experience of every individual participant. The registry record does not provide a fuller explanation of the endpoint’s selection, scoring procedures, or the exact statistical model intended for the pre-to-post evaluation.

7. Results Posted in the Registry

• Non-Training: mean 2.96 versus 2.60 on control; difference 0.36 (95% CI -1.17 to 1.89), standardized effect d = 0.12; z-test p = 0.644.

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.

Reading Span fMRI Task

• 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:
  • Non-Training: mean 2.65 versus 0.47 on control; difference 2.18 (95% CI -0.21 to 4.57), standardized effect d = 0.67; z-test p = 0.074.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Non-Training and the control. The interval gives the range of effects that remain compatible with the data.

Emotion Regulation fMRI Task

• 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:
  • Non-Training: mean 0.06 versus -0.01 on control; difference 0.07 (95% CI -0.15 to 0.29), standardized effect d = 0.25; z-test p = 0.525.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Non-Training and the control. The interval gives the range of effects that remain compatible with the data.

Clinician Administered PTSD Scale Diagnosis

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Non-Training: 41.0% (16/39) versus 50.0% (21/42) on control; risk difference -9.0 percentage points (95% CI -30.6 to 12.6), risk ratio 0.82 (0.51–1.33); unadjusted two-proportion z-test p = 0.418.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Non-Training and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover Reading Span, Emotion Regulation, and CAPS Diagnosis, adding cognitive, task-related, and diagnostic measures to the symptom-focused primary outcome. Their posted estimates do not form a uniform pattern across measures: the directions vary, and the displayed confidence intervals include zero. This does not provide clear evidence of a consistent separation between the assigned conditions across these different domains.

These outcomes provide context for the primary result rather than replacing it. They address related but distinct questions and use different measurement scales and comparison metrics. A secondary estimate may be suggestive, but its interpretation depends on uncertainty, measurement properties, and the analysis used. The registry record lists no reported analyses for these secondary outcomes, so the posted comparisons should be understood as unadjusted recalculations rather than as a complete account of a prespecified secondary analysis plan.

Unadjusted secondary results carry less weight when multiple outcomes are considered because each additional test creates another opportunity for a chance finding. In general, multiplicity procedures can address that issue, but the registry record does not report whether any adjustment was planned or performed. Without that information, isolated secondary results should be treated as exploratory rather than as independent confirmation of the primary comparison. The overall pattern is best read cautiously: it adds descriptive information, but does not resolve the uncertainty in the primary estimate.

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

• Reading Span fMRI Task
• Emotion Regulation fMRI Task
• Clinician Administered PTSD Scale Diagnosis

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: each participant’s reexperiencing score is measured on a scale, and the registry summarizes each group with a mean and standard deviation. A mean describes the group’s average score, while a standard deviation describes how spread out individual scores are around that average. These summaries can make a group comparison readable, but they do not show the full distribution of participant-level results.

What does the mean difference represent?

The posted comparison subtracts the COGENT mean from the Non-Training mean. Its estimate is 0.36. A positive value therefore indicates a higher average score in Non-Training on the posted scale, where higher scores indicate greater severity. This is a difference between group means; it is not a percentage change or a statement about how many individuals improved.

The registry lists the primary timeframe as pre to post intervention, from baseline to week 8, and says change over time will be evaluated. The record does not report the exact analysis model or specify how baseline values were handled in the reported summary comparison. Consequently, the comparison should be read as the posted pre-to-post result, not as evidence of a particular adjusted model.

How should the p-value be read?

The unadjusted p-value is 0.644. In general, a p-value describes how unusual data at least as incompatible with a specified null hypothesis would be under the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. Because the analysis is unadjusted, this value does not account for consideration of multiple outcomes or analyses.

The registry record does not list a sponsor-reported primary analysis. The comparison estimates and p-values are unadjusted recalculations from the posted summary statistics. They should therefore not be described as adjusted or as results from a model whose details are not recorded. The discrepancy flag is false, so there is no recorded conflict requiring a sponsor analysis to take precedence over the recalculation.

Why does the confidence interval matter?

A confidence interval gives a range of values compatible with the observed data under the statistical method used. For the primary mean difference, the interval runs from -1.166 to 1.886. The range includes zero, the value representing no difference between group means, and it also includes values in either direction. This makes the direction and size of the underlying difference uncertain. A confidence interval is not a guarantee that the true value lies within its endpoints, and its interpretation depends on the method and assumptions used.

The posted group standard deviations, 2.9 for COGENT and 3.2 for Non-Training, describe variability in scores within the groups. When participant-level data are unavailable, a recalculation based on summary statistics cannot examine distribution shape, influential observations, or whether the outcome behaves similarly across subgroups. The registry record does not report diagnostic checks or a method for handling departures from model assumptions.

How do the secondary outcome types differ?

Reading Span and Emotion Regulation are also continuous outcomes and are posted as means and standard deviations. Their comparisons are expressed as mean differences. CAPS Diagnosis is a count of participants meeting the recorded outcome, and its comparison is expressed as a risk difference. In general, a risk difference compares the proportions with an outcome between groups; it is distinct from a mean difference on a continuous scale. The posted CAPS Diagnosis interval is for the risk difference, not a difference in mean symptom scores.

For continuous outcomes, over-dispersion is not the central issue in the way it can be for count outcomes modeled with a Poisson distribution. In general, over-dispersion means that observed count variability exceeds what a model assumes; ignoring it can make uncertainty estimates too narrow. No count-regression model or over-dispersion assessment is reported in the registry record, and the posted diagnosis comparison is a risk difference rather than a count-model estimate.

What analysis details remain unknown?

The registry record does not report the prespecified statistical analysis plan, the handling of missing outcome data, the analysis population, interim analyses, or adjustments for multiple comparisons. It also does not provide participant-level data or a detailed account of outcome completeness. These omissions limit the ability to determine whether the posted calculations match the trial’s intended primary analysis or how robust the results are to alternative assumptions. The results can be described accurately as posted summary statistics and unadjusted comparisons, but should not be presented as a complete account of the trial’s analytic methods.

11. Confidence Intervals and Precision

The primary mean difference is 0.36, and its 95% confidence interval ranges from -1.166 to 1.886. This interval is wide enough to include zero and values on both sides of zero. It therefore leaves uncertainty about whether the average pre-to-post result was higher in one condition, higher in the other, or similar. The estimate’s direction alone is not a precise description of the possible group difference.

In general, a confidence interval communicates both an estimate and the uncertainty around it. A narrower interval would indicate greater precision under the same method; a wider interval indicates less precision. The interval does not show the range of individual participant scores, and it is not a prediction interval for a future participant or trial. Its interpretation also depends on the analysis assumptions, which are not fully described in the registry record.

The observed primary outcome groups are reported as 35 for COGENT and 28 for Non-Training. Their standard deviations are 2.9 and 3.2, respectively. The trial’s registry enrollment is 81, but the record does not explain the differing outcome denominators or report how missing outcome data were handled. These details matter when judging precision and the scope of the comparison.

The interval should be read alongside the point estimate and the study context, not as proof of no effect. It permits uncertainty in both directions and does not identify a single most likely clinical interpretation. The posted result is therefore informative about the range of values compatible with the summary comparison, but it does not provide a precise determination of the size or direction of a difference.

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 planning figures are participants per arm, not total events. They are illustrative calculations based on the observed effect and are not the original trial’s design justification. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set at the stated fraction of the observed effect. The original sample-size assumptions are not reported in the registry record.

The 80% power column indicates the 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 expresses the corresponding planning target for a 90% chance. These are prospective planning quantities, not measures of what the completed trial achieved.

The rows use the full observed effect and smaller fractions of it. Smaller effects are harder to distinguish from random variation, so the planning calculation requires more participants per arm as the target effect becomes smaller. The displayed values are specific to the observed estimates and stated assumptions; they should not be treated as universal sample-size requirements. They also do not establish whether the trial’s own enrollment was adequate for its original aims.

These figures are not observed power. They do not measure the probability that the completed trial found an effect, and they should not be used to reinterpret a non-significant result as proof of no difference. The trial’s original design assumptions and sample-size justification are not reported in the registry record.

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 = 0.360; pooled standard deviation σ = 3.036.
2. n = 2 × (1.96 + 0.842)² × 3.036² / 0.360² = 1,116.8, rounded up to 1,117 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 <- 0.36                      # observed difference in means
sigma <- 3.0364452901378                     # 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(3316196)
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 = 1,117 (80% power), 1,496 (90% power)
 75% of observed effect: n per arm = 1,986 (80% power), 2,658 (90% power)
 50% of observed effect: n per arm = 4,468 (80% power), 5,981 (90% power)
power.t.test: 1117.7 
Simulated power at n = 1,117 per arm: 0.802 (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. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary outcome and several secondary outcomes, but it does not report a statistical analysis plan describing how they were prioritized or analyzed together. In particular, multiplicity adjustment is not reported in the registry record. When several outcomes or comparisons are examined, the chance of obtaining at least one apparently notable result by chance can increase. In general, prespecifying a primary outcome and using a stated multiplicity strategy can help keep interpretation focused; the record does not say whether such a strategy was used.

The posted p-values are unadjusted. They should not be interpreted as accounting for the full set of outcomes or analyses. The registry reports no analyses for the secondary outcomes, and no sponsor-reported analysis details are available to clarify whether these recalculations match the planned approach.

Missing-data handling is also not reported in the registry record. The listed denominators differ across outcomes, and the record does not explain the reasons for those differences, how missing values were treated, or whether a particular imputation or complete-case approach was used. It does not report the analysis population, such as whether participants were analyzed according to assigned condition regardless of adherence. Therefore, the results cannot be confidently characterized as intention-to-treat, per-protocol, or another specified population.

The record also does not report interim analyses or the trial’s original sample-size justification. These are unknowns, not evidence that a particular method was or was not used. Without the protocol-level details, readers should interpret the posted results as outcome summaries and unadjusted comparisons, with uncertainty about how multiplicity and missingness were addressed in the trial’s intended analysis.

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 5% reporting threshold.

The registry lists safety counts separately for COGENT and Non-Training. For serious adverse events, the groups at risk are reported as 42 and 39, respectively, and the number affected is zero in each group. For other adverse events, the groups at risk are also 42 and 39, with zero affected in each group. Deaths are listed as zero in both groups.

These are the counts recorded in the registry, not evidence that no adverse events of any kind occurred outside the reported categories or reporting period. The registry record does not provide event terms, additional descriptions, or a detailed safety-analysis method. The safety denominators differ from the primary outcome denominators, and the record does not explain why. These figures should therefore be read as the posted safety counts for the listed groups, without treating them as a comparison of risk adjusted for follow-up time or other factors.

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

The posted primary result is a mean difference of 0.36 for Non-Training compared with COGENT on the pre-to-post CAPS Reexperiencing score. The confidence interval ranges from -1.166 to 1.886, and the unadjusted p-value is 0.644. The estimate is compatible with uncertainty in either direction and does not establish a clear difference between the groups.

This finding does not prove that the conditions are equivalent, that there is no effect, or that a potentially meaningful effect has been ruled out. It also does not establish that cognitive training improves or worsens PTSD. The estimate is specific to the measure, comparison, and timeframe posted in the registry. It should not be extended to all PTSD symptoms, other populations, different forms of cognitive training, or outcomes beyond the reported assessment period.

The registry describes random assignment, which supports a comparative design, but the analytic details available in the record are limited. The precise masking roles, original analysis plan, handling of missing outcomes, analysis population, and multiplicity strategy are not reported. These gaps matter when judging how the posted summaries relate to the trial’s intended analyses.

The secondary outcomes and safety counts add descriptive context but do not remove the primary result’s uncertainty. The secondary comparisons are unadjusted and do not show a consistent direction across all measures. The safety record lists zero affected participants in the reported serious and other adverse-event categories, but it does not supply enough detail to make broader claims about safety. Overall, the posted data describe this trial’s recorded outcomes; they do not support a definitive conclusion about treatment benefit, absence of benefit, or effects outside the measured setting.

17. Why This Trial Matters Statistically

This trial is statistically informative because it connects a defined cognitive-training intervention with a symptom-focused primary measure and additional cognitive and diagnostic outcomes. Its randomized, parallel-group design gives the comparison a clear structure. The posted results also illustrate why an estimate should be considered together with its uncertainty: the primary point estimate is modest, while the confidence interval spans values on both sides of no difference.

The distinction between a non-significant test and evidence of equivalence is important in clinical research. A p-value of 0.644 does not show that the groups had identical outcomes, and the confidence interval does not narrowly constrain the possible difference. Readers can use the posted estimate and interval to see both the observed direction and the limits on precision without turning the result into a yes-or-no verdict.

The record also highlights the role of analysis transparency. The trial is marked completed, and results are posted, but details about the prespecified model, missing data, analysis populations, multiplicity, and original sample-size rationale are not reported. Those details are needed to evaluate whether the displayed summaries align with the intended confirmatory analysis and how to interpret multiple outcomes.

Finally, planning illustrations based on observed effects can help show how sample-size requirements depend on the size of the effect targeted. They are not evidence of the trial’s original design reasoning or its achieved power. Separating these planning calculations from the observed outcome analysis helps prevent a common statistical misreading.

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.

Several trial-specific limitations affect interpretation. The primary outcome groups are reported as 35 for COGENT and 28 for Non-Training, while the safety groups at risk are 42 and 39. Secondary outcome denominators vary as well. The registry record does not explain these differences, so it is not possible to determine from the record alone whether they reflect missing measurements, different assessment subsets, or another reason.

The result is limited to the CAPS Reexperiencing measure and the stated pre-to-post intervention timeframe. The registry does not report longer-term follow-up results, so persistence beyond week 8 cannot be assessed from these data. The record also does not provide participant-level distributions, detailed outcome completion information, or subgroup results.

Although the trial is described as triple masked, the registry does not identify which roles were masked or describe how masking was maintained. It also does not report the original statistical analysis plan, missing-data strategy, analysis population, multiplicity adjustment, or interim analyses. Without those details, the posted unadjusted comparisons cannot be confirmed as matching a prespecified primary or secondary analysis.

The brief summary describes training twice per week for eight weeks, but the record does not provide adherence or treatment-fidelity results. The registry reports a completed study, yet the available outcome summaries do not establish how the intervention was delivered to each participant. These limitations do not invalidate the posted statistics, but they constrain the conclusions that can be drawn from them.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Cognitive Training for PTSD — NCT03316196.
• Registry record last updated 2024-09-19; status last verified 2024-08.
• Results first posted 2024-09-19.
• 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

NCT03316196 is a completed, randomized, triple-masked, parallel-group study sponsored by the VA Office of Research and Development. It compared COGENT cognitive training with a non-training computer condition in Veterans with PTSD. The registry lists enrollment as 81 and describes the intended training schedule as twice per week for eight weeks.

The primary outcome was the pre-to-post CAPS Reexperiencing score. The posted mean difference was 0.36, with a 95% confidence interval from -1.166 to 1.886 and an unadjusted p-value of 0.644. The interval includes zero and allows uncertainty in either direction. Secondary outcomes and safety counts are also posted, but analysis-plan, missing-data, multiplicity, and analysis-population details are not reported in the registry record.