Cognitive Training Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cognitive Training Trial (NCT01133418), a randomized na study. The primary outcome is Total ADHD Symptom Score From Vanderbilt ADHD Parent Rating Scale.

2. Clinical Question

The central contrast is between two forms of computerized cognitive training, not between training and no training. That distinction matters for interpretation: the registry describes the comparator as participating in the same tasks, with the same number of blocks and trials and the same trainers. The treatment contrast therefore concerns progressive versus non-progressive training as described, rather than exposure to training versus no exposure.

The intervention rationale in the summary focuses on reaction-time variability, while the primary endpoint is a parent-rated symptom score. The secondary outcomes include reaction-time variability on a Go/No-Go task and a web-based reading score. These measures address different domains, so they should not be treated as interchangeable indicators of a single outcome.

3. Trial Design

The registry describes a randomized, triple-masked, parallel-group trial. Random allocation is intended to make assignment to the two groups independent of participant characteristics, on average. A parallel design means participants are assigned to one group rather than receiving both training approaches in sequence. The registry gives the masking designation but does not identify which roles were masked or describe how masking was maintained.

The recorded enrollment is 72. The trial summary says 64 children were to be randomly assigned, whereas the registry enrollment field records 72. The reason for the difference, and whether enrollment refers to a screened, enrolled, randomized, or another population, is not reported in the registry record. The displayed primary outcome groups have 29 and 32 participants, respectively. The registry does not give a participant-flow account that connects these figures.

The study is marked completed and its phase is listed as not applicable. The record gives a start of 2010-09 and primary completion of 2014-08. It identifies the sponsor as Children's Hospital Medical Center, Cincinnati. These administrative details do not establish when individual participants were assessed or how long follow-up lasted beyond the outcome timeframe stated for each measure.

The progressive intervention is named Computerized Progressive Attention Training. The control arm is named Non-progressive Cognitive Training. The summary describes 8 weeks of intervention for the progressive-training group. A complete description of the control schedule and of the intervention procedures is not reported in the registry record. The record also does not report the randomization method, allocation concealment procedures, trial-specific sample-size justification, or a full protocol description.

The primary outcome is listed at 2 months. The two posted secondary outcomes are also listed at 2 months. How the timing of that assessment relates precisely to the training period, and whether there were other planned assessment times, is not reported in the registry record.

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

Reasons for not completing this period:

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 Total ADHD Symptom Score From Vanderbilt ADHD Parent Rating Scale, measured over 2 months.

The primary endpoint is the total ADHD symptom score from the Vanderbilt ADHD Parent Rating Scale at 2 months. It is reported as a mean on a scale with a range of 0-54. The score is computed by summing 18 ADHD symptom items, each rated from 0 to 3; higher scores indicate higher symptom severity. The unit is described as units on a scale, and the registry reports a standard deviation for each group.

A parent rating is used to quantify reported ADHD symptoms across the listed items. Summing the items produces one overall score that can be compared between the randomized groups. In general, a continuous endpoint such as this retains information about the range of scores rather than reducing each participant to a category. Comparing group means describes the average score in each arm, while the standard deviations indicate how much scores vary within the arms.

The endpoint is relevant to the study question because the summary frames the intervention as potentially affecting behavioral outcomes, and the measure captures parent-rated ADHD symptoms. It is distinct from the intervention's stated training target of reducing variability in responses. A change in a training task or reaction-time measure would not, by itself, be the same as a difference in parent-rated symptoms; the trial records these as separate outcomes.

The registry provides the score description and timeframe, but does not report whether the primary endpoint was assessed as a change from baseline or as a score at follow-up adjusted for baseline. It also does not report the primary analysis population or a missing-data procedure. Those details affect how a numerical comparison should be understood and cannot be inferred from the endpoint definition alone.

7. Results Posted in the Registry

• Non-progressive Cognitive Training: mean 26.63 versus 27.00 on control; difference -0.37 (95% CI -5.27 to 4.53), standardized effect d = -0.04; z-test p = 0.882.

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.

Intra-individual Variability on Go/No-Go 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-progressive Cognitive Training: mean 236.60 versus 242.24 on control; difference -5.64 (95% CI -64.41 to 53.13), standardized effect d = -0.05; z-test p = 0.851.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Non-progressive Cognitive Training and the control. The interval gives the range of effects that remain compatible with the data.

Academic Improvement Measurement System - Web-based (AIMSWEB) Reading Score (Proportion Accurate)

• 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-progressive Cognitive Training: mean 0.97 versus 0.97 on control; difference 0.00 (95% CI -0.02 to 0.02), standardized effect d = 0.00; z-test p = 1.000.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Non-progressive Cognitive Training and the control. The interval gives the range of effects that remain compatible with the data.

The two posted secondary outcomes cover reaction-time variability on a Go/No-Go task and reading accuracy on the AIMSWEB web-based reading measure. Together, they span a task-related measure connected to the intervention rationale and an academic performance measure. Neither is the same construct as the parent-rated symptom score used as the primary endpoint.

Across the secondary outcomes, the recalculated differences are small relative to their respective intervals, and both intervals include zero. The registry also reports General Linear Model p-values for the secondary analyses, but does not give model estimates, confidence intervals, or model specifications. Those reported analyses and the unadjusted recalculations therefore should not be collapsed into one result or assumed to use identical methods.

The secondary pattern is broadly consistent with the primary comparison in that the posted unadjusted intervals do not exclude zero. This is a description of the results, not proof that the outcomes are equivalent or that there is no effect. The measures have different units and assess different domains, so their numerical estimates cannot be pooled or directly ranked by size without an appropriate framework.

Unadjusted secondary results generally carry less evidential weight than a prespecified primary result because multiple outcomes create more opportunities for apparently notable findings. In general, testing several outcomes without accounting for multiplicity can increase the chance of at least one false-positive finding. The registry record does not report a multiplicity adjustment, a hierarchy for the secondary outcomes, or whether they were prespecified in the analysis plan. Their findings are best treated as supporting, exploratory, or descriptive evidence rather than as independent confirmation of the primary result.

The registry lists four outcome measures and four analyses overall, but only two secondary outcomes are posted in the record. The remaining outcome measures and their results are not reported in the registry record. That incomplete reporting limits a full assessment of the range of outcomes considered.

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

• Intra-individual Variability on Go/No-Go Task
• Academic Improvement Measurement System - Web-based (AIMSWEB) Reading Score (Proportion Accurate)

9. Statistical Interpretation

10. Statistical Methods Explained

The registry lists a General Linear Model for the sponsor-reported analysis of the primary endpoint. It provides a p-value of .36 but no model estimate, confidence interval, covariate list, or description of the model's exact form. The registry does not report whether the analysis used baseline adjustment, how repeated observations were handled, or which participants were included. A method name by itself is not enough to reconstruct the analysis.

What does a comparison of means describe?

For a continuous measure, the mean is the arithmetic average of the observed scores in a group. The reported primary means are 27.0 for cognitive training and 26.63 for non-progressive cognitive training. The reported standard deviations, 9.46 and 10.08, describe the spread of individual scores around each group mean. In general, a standard deviation is not a measure of uncertainty in the mean; it describes variability among observations. The uncertainty in a comparison also depends on the number of observations and the analysis method.

The unadjusted recalculation uses the posted group means and standard deviations to estimate a difference between group means. Its direction is defined in the comparison as non-progressive cognitive training minus cognitive training. Thus, the reported estimate of -0.37 is the unadjusted difference in that order. The recalculation does not incorporate participant-level data, covariates, baseline scores, or model adjustments that may be part of a separate analysis.

What does a confidence interval add?

A confidence interval gives a range of estimates compatible with the data and the method used to construct it. In general, a narrower interval indicates greater precision, while a wider interval indicates more uncertainty. The primary unadjusted interval runs from -5.274 to 4.534. Because it spans zero, it includes both directions of difference under the stated subtraction order. The registry does not state the confidence level for this recalculated interval, so no level should be inferred from the record.

A confidence interval is not a list of outcomes that each have equal probability, and it is not proof that the true effect lies within the printed endpoints. It is best read alongside the estimate and the study design. Here, the estimate is close to zero relative to the interval's range, and the interval includes values that would represent differences in either direction. The registry does not give a threshold for deciding which differences would be clinically important.

How should the p-values be read?

In general, a p-value measures how incompatible the observed data are with a specified null model, under the assumptions of the analysis. It does not measure the probability that the null hypothesis is true, the size of an effect, or the probability that a result will replicate. The unadjusted recalculation reports p = 0.882. The registry's General Linear Model result is p = .36. Since the model details and estimate are not posted, these two p-values should not be treated as if they came from the same calculation. Their difference cannot be explained from the information in the registry record.

The sponsor-reported p-value is not accompanied by a direction or magnitude estimate. It therefore cannot, by itself, tell a reader how far apart the groups were on the scale. The group means and recalculated interval help describe the unadjusted contrast, but they do not supply the missing model estimate. In general, an estimate with an interval is more informative about the size and precision of a contrast than a significance test alone.

What does the planning illustration represent?

The sample-size planning illustration is based on the observed effect and uses a two-sided significance level of 0.05, equal allocation between the arms, and a normal approximation. Its rows consider the stated fractions of the observed effect, and the columns report participants per arm for 80% and 90% power. These are planning calculations, not the original trial design. The registry does not report the trial's own sample-size justification or its original design assumptions.

In general, over-dispersion means that variability is greater than a model assumes. If a model understates variability, it can make uncertainty appear smaller than it is. The posted results describe continuous outcomes and standard deviations; the registry does not report a particular over-dispersion assessment or a model adjustment for it. The term is therefore a general statistical consideration, not a reported feature of this trial.

Randomization supports a causal comparison when assignment is implemented as intended and the analysis appropriately handles the observed data. The registry says the study was randomized, but it does not describe allocation concealment, adherence, crossover, or participant flow. Those omissions limit what can be assessed about implementation and analysis. Likewise, triple masking is recorded, but the masked groups and procedures are not described. No stronger inference about masking or bias control can be made from the designation alone.

11. Confidence Intervals and Precision

The primary unadjusted mean difference is -0.37, with a confidence interval from -5.274 to 4.534. The interval crosses zero and extends in both directions. This indicates that the point estimate alone does not capture the uncertainty in the comparison. The interval's endpoints are on the Vanderbilt symptom-score scale, and the registry's description says higher scores indicate higher severity.

In general, precision depends on the amount of information in the data and on variability. The primary groups include 29 and 32 participants, and their standard deviations are 9.46 and 10.08. The trial's recorded enrollment is 72, but the registry does not provide a participant-flow account explaining how the outcome group counts relate to enrollment. It also does not report a confidence level for the recalculated interval, so the level should not be assumed.

The interval includes zero but that does not demonstrate that the groups are equivalent. Equivalence requires a prespecified margin defining what differences are sufficiently small, together with an analysis designed to evaluate that margin. Such a margin and an equivalence analysis are not reported in the registry record. The interval also does not establish that every value it spans is clinically important; the record supplies no threshold for clinical importance.

The sponsor-reported General Linear Model result is p = .36, but its estimate and interval are not reported. That analysis therefore cannot be assessed for precision from the registry entry. Its p-value should not be used as a substitute for an interval around the model's estimated contrast. The recalculated interval describes an unadjusted comparison from posted summary statistics and may not represent the sponsor model's uncertainty.

For the Go/No-Go variability outcome, the recalculated mean difference is -5.64 milliseconds, with an interval from -64.41 to 53.13. The interval also spans zero. For the reading score, the recalculated difference is 0.0, with an interval from -0.018 to 0.018. Those secondary intervals have their own outcome scales and should not be compared directly with the primary interval's numerical width.

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 for 80% power and participants per arm for 90% power; these are not total event counts. The illustration assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It uses the observed effect as a starting point and considers the stated fractions of that effect. The original trial's own sample-size assumptions and justification are not reported in the registry record.

In general, power is the probability that a study will detect an effect of a specified size under stated assumptions, if that effect is present. The 80% power column gives the planned number of participants in each arm for the calculation to have 80% power under those assumptions. The 90% power column gives the corresponding number for 90% power. A higher power target generally requires more participants when other assumptions are held fixed.

The rows for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect show how the planning calculation changes when the effect to detect is set at different fractions of the observed estimate. Smaller effects need more participants to detect under otherwise identical assumptions. The figures rise sharply as the planning effect becomes smaller, which illustrates the sensitivity of sample-size calculations to the assumed effect size.

For the primary endpoint, planning for 100% of the observed effect gives 10,992 participants per arm for 80% power and 14,715 participants per arm for 90% power. Planning for 75% of the observed effect gives 19,541 participants per arm for 80% power and 26,160 participants per arm for 90% power. Planning for 50% of the observed effect gives 43,967 participants per arm for 80% power and 58,859 participants per arm for 90% power.

These numbers are planning illustrations derived from the observed effect, not observed power and not evidence about the probability that this trial detected an effect. Because the observed primary effect is very small, calculations anchored to it yield very large planning counts. The registry does not report that these calculations were used to design the trial.

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 = -0.370; pooled standard deviation σ = 9.791.
2. n = 2 × (1.96 + 0.842)² × 9.791² / -0.370² = 10,991.5, rounded up to 10,992 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 <- -0.370000000000001                      # observed difference in means
sigma <- 9.79065898649476                     # 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(1133418)
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 = 10,992 (80% power), 14,715 (90% power)
 75% of observed effect: n per arm = 19,541 (80% power), 26,160 (90% power)
 50% of observed effect: n per arm = 43,967 (80% power), 58,859 (90% power)
power.t.test: 10992.5 
Simulated power at n = 10,992 per arm: 0.799 (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 sponsor-reported primary analysis is identified as a General Linear Model and has p = .36. The registry does not report its estimated group contrast, confidence interval, covariates, or other model details. The posted summary statistics permit a separate unadjusted recalculation: a mean difference of -0.37, a confidence interval from -5.274 to 4.534, and p = 0.882.

These results should be kept distinct. The recalculation uses only the posted group means, standard deviations, and sample sizes; it is not a reconstruction of the General Linear Model. Different analysis methods or adjustments can produce different estimates of uncertainty and different p-values, but the registry does not explain which features account for the values. It would therefore be speculative to attribute the difference to a particular covariate, adjustment, or model choice.

The two analyses point to different levels of statistical evidence as reflected by their p-values, but neither p-value alone gives a complete description of effect size and precision. The unadjusted estimate and interval show the scale and range of that particular comparison. The sponsor's model estimate and interval are not reported, so its magnitude and precision cannot be described from the registry entry. The registry does not mark a discrepancy, but that does not supply the missing model detail or establish that the two procedures are equivalent.

For interpretation, the sponsor's reported result is the analysis identified by the registry as the trial's General Linear Model analysis. The unadjusted recalculation is supplementary context based on the arm summaries, not a replacement for the model result.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity concerns arise when investigators evaluate several outcomes, time points, subgroups, or statistical contrasts. In general, each additional test can add another opportunity to observe a small p-value by chance under a null setting. A prespecified strategy can control or organize this risk, for example by defining a primary outcome, setting a testing sequence, or applying an adjustment. The registry identifies a primary endpoint and lists secondary outcomes, but it does not report a multiplicity procedure or a hierarchy for testing them.

The record lists four outcome measures and four analyses, while two secondary outcomes are posted. It does not report the results for the other listed measures. The absence of those results in the registry entry does not establish why they are absent, whether they were measured, or whether they were analyzed. The complete outcome reporting and any prespecified distinctions between confirmatory and exploratory analyses are not reported in the registry record.

Missing-data methods are also not reported. The primary endpoint group sizes are 29 and 32, whereas the recorded enrollment is 72. The registry does not describe participant flow, reasons for missing measurements, exclusions, adherence, or whether the analysis followed intention-to-treat principles. The posted group sizes alone cannot establish how missing observations were handled or why the outcome counts differ from the enrollment field.

In general, missing data can affect a comparison if the probability of being observed is related to outcome values or group assignment. Different assumptions and methods can lead to different results, so a complete report would describe missingness by group, analysis populations, and any imputation or sensitivity analyses. These details are not reported in the registry record, and no particular missing-data mechanism should be assumed.

The registry states triple masking, but does not identify the roles masked, describe masking procedures, or report whether masking was assessed. It also does not provide the full statistical analysis plan, the model specification, or a trial-specific sample-size justification. These omissions limit evaluation of prespecification, analysis choices, and potential selective reporting. They do not establish that a particular problem occurred; they identify aspects of the evidence that cannot be evaluated from the record.

16. Safety Results

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

Other (non-serious) adverse events

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

For serious adverse events, the registry reports 34 participants at risk in the cognitive-training group and 38 at risk in the non-progressive cognitive-training group. The number affected is recorded as 0 in each group. The registry also records 34 and 38 participants at risk, respectively, for other adverse events, with 0 affected in each group. These are the posted counts and should be read in relation to the listed at-risk group sizes.

The record does not list serious adverse event terms. Death counts are not reported in the registry record. Definitions, collection procedures, observation periods, and the completeness of adverse-event reporting are not described, so the counts cannot establish what was or was not captured outside those registry fields.

Zero recorded affected participants is a statement about the adverse-event results posted in the registry, not proof that no events occurred under any definition or during any period. These figures also do not allow a meaningful comparative safety estimate by themselves. The registry does not provide event-level descriptions or a safety analysis beyond the counts.

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

The primary outcome data show mean Vanderbilt scores of 27.0 in the cognitive-training group and 26.63 in the non-progressive cognitive-training group at 2 months. The unadjusted mean difference is -0.37, with a confidence interval from -5.274 to 4.534. The recalculated unadjusted p-value is 0.882, while the registry reports p = .36 for its General Linear Model analysis. These are distinct analyses, and details needed to reconcile them are not reported.

The findings do not establish that the interventions are equivalent. A failure to show a difference under a statistical test is not the same as demonstrating that any difference is small enough to be unimportant. An equivalence or non-inferiority margin and an analysis designed for that question are not reported. The interval's inclusion of zero also does not mean that the true difference is exactly zero.

The trial compares progressive with non-progressive computerized cognitive training. It does not provide a comparison with no training or with another treatment modality. The posted results therefore address only the contrast between the two recorded training arms. They do not establish the effect of training generally, the effect on all children with ADHD, or effects beyond the outcomes and timeframes recorded.

The intervention rationale describes training response consistency, while the primary endpoint measures parent-rated ADHD symptoms. Secondary outcomes address reaction-time variability and reading accuracy. The measures are related to different study questions, and one should not be treated as a substitute for another. The registry does not report a threshold for clinical importance on the primary score, nor does it report enough model detail to interpret the sponsor analysis's effect size.

In general, trial results are interpreted in light of precision, design, missing data, analysis choices, and the prespecified question. The registry provides some of these elements, including randomization, masking designation, group summaries, and timeframes, but not a complete protocol or analysis account. The results describe the posted data; they do not support individualized conclusions or medical advice.

18. Why This Trial Matters Statistically

This trial is statistically instructive because its design compares two active training conditions that share several features in the registry description. The control participants did the same tasks, had the same number of blocks and trials, and worked with the same trainers. That makes the contrast more specific than a comparison with no intervention, while also requiring careful attention to what the comparator actually represents.

The study also places outcomes from different domains alongside one another. The primary endpoint is parent-rated ADHD symptoms, while the secondary measures include task-based reaction-time variability and reading accuracy. This highlights a general issue in clinical research: an intervention's proposed mechanism, a proximal task measure, and a patient- or parent-reported outcome are not interchangeable. Each answers a different question.

The record offers a useful example of why effect estimates and intervals matter in addition to p-values. The primary unadjusted estimate is -0.37, and the interval extends on both sides of zero. The sponsor's General Linear Model p-value is posted without its estimate or interval. Readers therefore have different levels of information for the two analyses, and should not infer an effect magnitude from a p-value alone.

The planning illustration also demonstrates how strongly required sample size can depend on the effect chosen for planning. The calculations use fractions of the observed effect, with smaller assumed effects requiring more participants. Because these are not the original design assumptions, they should not be mistaken for evidence that the trial was designed to detect those effects.

Finally, the difference between recorded enrollment and endpoint group sizes, alongside the absence of a participant-flow and missing-data account, illustrates why a randomized design alone is not enough to evaluate every source of uncertainty. Clear reporting of analysis populations, missingness, model specifications, and outcome handling is important for interpreting trial findings and assessing their reproducibility.

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 registry record does not report a full protocol, statistical analysis plan, or trial-specific sample-size justification. The planning illustration is based on the observed effect and is not evidence about the original design. The record also does not describe the randomization procedure or allocation concealment, or identify which parties were masked despite the triple-masking designation.

There is a difference between the enrollment field of 72 and the primary outcome group sizes of 29 and 32. The summary describes 64 children as intended for random assignment. The record does not explain how these figures relate, provide participant flow, or state how many participants were randomized, completed training, or contributed data to each analysis. Missing-data handling and analysis populations are not reported.

The sponsor's primary General Linear Model result includes a p-value but no effect estimate, confidence interval, covariates, or model specification. The unadjusted recalculation is therefore not directly comparable to a fully described sponsor model. The registry does not explain the difference between p = .36 and the recalculated p = 0.882.

Only two secondary outcomes are posted, although the record lists four outcome measures and four analyses. The reason for the absent outcome results is not reported. Multiplicity handling, secondary-outcome hierarchy, and the distinction between confirmatory and exploratory tests are also not reported. These omissions limit assessment of the full analysis and reporting context.

The control-arm description is incomplete, and the registry does not report whether the primary score was analyzed as a follow-up value or as change from baseline. It also does not state a threshold for clinical importance on the primary scale. These limitations concern what can be assessed from the registry entry; they do not establish that bias, missingness, or selective reporting occurred.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Cognitive Training Trial — NCT01133418.
• Registry record last updated 2017-05-30; status last verified 2017-04.
• Results first posted 2017-05-01.
• 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

The registry record describes a completed, randomized, triple-masked, parallel-group trial of Computerized Progressive Attention Training versus Non-progressive Cognitive Training in children with ADHD. Enrollment is recorded as 72. The primary outcome is the Vanderbilt ADHD Parent Rating Scale total symptom score at 2 months. The primary group means are 27.0 and 26.63, with a recalculated unadjusted difference of -0.37 and a confidence interval from -5.274 to 4.534.

The registry reports a General Linear Model p-value of .36 for the primary analysis, without an estimate or confidence interval. A separate unadjusted recalculation gives p = 0.882. Two secondary outcomes are posted, with unadjusted intervals that include zero. Serious adverse events affected 0 participants in each group among the listed at-risk counts.

Participant flow, missing-data methods, the full analysis plan, multiplicity handling, and the sponsor model details are not reported in the registry record. The posted results should therefore be interpreted as a partial statistical account of the randomized comparison.