The Impact of Reactivation During Sleep on the Consolidation of Abstract Information in Humans: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Impact of Reactivation During Sleep on the Consolidation of Abstract Information in Humans (NCT05746299), a randomized na study. The primary outcome is Structure Knowledge for a New Modular Category in Stage 2.

2. Clinical Question

The study’s rationale is that category learning may encode relationships among features, rather than only the individual features themselves. The investigators ask whether experience with a structure can support learning about a different category that shares that structure, and whether offline time, including sleep with targeted reactivation, may be relevant to that transfer. The reported group means describe accuracy under the listed conditions; they do not by themselves isolate why groups differ. In particular, the registry record does not report a direct statistical contrast between every pair of congruent and incongruent conditions, or between the sleep stages.

3. Trial Design

The registry describes a randomized, single-masked, parallel behavioral study, marked completed. The listed phase is NA. Enrollment is reported as 194. The sponsor is the University of Pennsylvania. The recorded start date is 2023-03-29 and primary completion date is 2024-06-30. The identity of the masked parties, the randomization procedure, and the original allocation plan are not reported in the registry record.

The design is organized around two aims and six groups with posted primary outcome results. Aim 1 compares immediate learning conditions: Immediate Incongruent and Immediate Congruent. The latter learned two categories with the same structure, whereas the former learned categories with different structures. Aim 2 includes Awake Incongruent, Awake Congruent, Sleep Incongruent, and Sleep Congruent. The awake conditions include a 2.5-hour break before learning and testing a novel category. The sleep descriptions include a 2-hour nap opportunity and a 30-minute break; targeted memory reactivation is described for the sleep procedure. Some arm descriptions are incomplete in the registry record, so details beyond the text available there cannot be established.

The endpoint results give group sizes of 47.0, 47.0, 26.0, 25.0, 22.0, and 27.0, respectively, for the six reported outcome groups. Those values describe the reported groups for this endpoint, not necessarily everyone enrolled. The record also lists a Sleep Congruent (REM) group in the safety information, but no separate REM outcome mean is posted. The registry record does not report analysis populations, participant flow, exclusions, or reasons for any difference between enrollment and endpoint group sizes.

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.

Population: Since only a small percentage of participants enter REM sleep within the 2-hour nap opportunity, if we were to randomly assign participants to either non-REM vs. REM conditions, there would be an extremely high likelihood that participants assigned to the REM condition would never enter that sleep phase, rendering that participant's data unusable and resulting in a high attrition rate. In order to generate as much usable data as possible, we therefore decided to remove the REM condition.

Only the first 6 of 8 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Structure Knowledge for a New Modular Category in Stage 2, measured over In Aim 1, accuracy is collected in a missing feature task 25 min. into the experiment, taking 25 min. In Aim 2, accuracy is collected in a missing feature task over 25 min in Stage 2.

The primary outcome is titled “Structure Knowledge for a New Modular Category in Stage 2.” It measures the proportion of correct responses on a missing feature task. Participants use memory from earlier in the experiment to guess which features complete category exemplars. The registry describes accuracy on a 0-100% scale and reports group means as proportions, with standard deviations.

This outcome is relevant to the study question because it tests use of learned structure in a new category, rather than only recall of the original category’s individual examples. Accuracy can therefore serve as a behavioral indicator of whether participants learned information that supports generalization to missing features. It is not a direct measure of an abstract representation itself, nor does a group mean reveal the reasoning used by any individual participant.

The timeframe describes accuracy collection in a missing feature task during Stage 2, over 25 min; it also describes an Aim 1 task 25 min. into the experiment, taking 25 min. The outcome title focuses on a new modular category in Stage 2. The registry record does not report a fuller operational definition of modular structure, an individual-level scoring procedure, or a threshold for interpreting a particular accuracy value. The posted results are group means and standard deviations, so they describe average performance and its spread rather than the full distribution of responses.

7. Results Posted in the Registry

• Aim 1 Immediate Congruent: mean 0.57 versus 0.56 on control; difference 0.01 (95% CI -0.03 to 0.05), standardized effect d = 0.10; z-test p = 0.614.
• Aim 2 Awake Incongruent: mean 0.63 versus 0.56 on control; difference 0.07 (95% CI 0.02 to 0.11), standardized effect d = 0.69; z-test p = 0.004.
• Aim 2 Awake Congruent: mean 0.63 versus 0.56 on control; difference 0.06 (95% CI 0.01 to 0.11), standardized effect d = 0.63; z-test p = 0.012.
• Aim 2 Sleep Incongruent: mean 0.63 versus 0.56 on control; difference 0.07 (95% CI 0.01 to 0.12), standardized effect d = 0.63; z-test p = 0.020.

8. Statistical Interpretation

9. Statistical Methods Explained

The outcome is continuous: each group has a mean proportion correct and a standard deviation. A mean describes the group’s average task accuracy. A standard deviation describes how dispersed participant scores are around that average; it is not the uncertainty in the mean itself. For example, the Immediate Incongruent mean is 0.563 and its standard deviation is 0.098. The standard deviation should not be read as a confidence interval or as a range containing every participant’s score.

What the mean differences represent

The posted contrasts express the mean in each comparison group minus the mean in the Immediate Incongruent reference group. A positive estimate means that the comparison group’s posted mean is higher. The Immediate Congruent estimate is 0.010; the estimates for the four Aim 2 groups are 0.067, 0.064, 0.066, and 0.060. These are unadjusted, summary-based comparisons. They are not adjusted model estimates, and the registry record does not report a sponsor analysis for them.

An interval around a mean difference conveys uncertainty about the estimate under the assumptions of the calculation. When the interval includes zero, the data are compatible with no mean difference under that method, as well as with values on either side of zero. When the whole interval lies above zero, the estimated difference is positive across the interval’s bounds. Interval level and the exact method used for the posted bounds are not reported in the registry record, so those details should not be inferred. An interval is not a guarantee that the true value lies within its limits.

How to interpret the p-values

A p-value describes how unusual the observed data, or data more extreme, would be under a specified null model. It is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of an effect. The posted p-values are unadjusted: 0.614 for Immediate Congruent and 0.004, 0.012, 0.02, and 0.012 for the Aim 2 groups. They should be considered alongside the estimated differences and interval bounds. The registry record reports zero analyses, so these p-values should not be presented as registry-posted sponsor inferential results.

Because the comparisons use summary means and standard deviations, they do not account for participant-level covariates or other design features not represented in those summaries. The recalculations are unadjusted. The discrepancy indicator is false, so the record does not flag a conflict requiring a sponsor analysis to take precedence. Still, the registry record does not identify the sponsor’s planned model or establish that these recalculations reproduce the investigators’ analysis.

General: why variability and sample size matter

In general, uncertainty in a group mean depends on both the spread of individual observations and the amount of information contributing to the estimate. Greater variability can make it harder to distinguish a real group difference from sampling fluctuation. Unequal group sizes can also yield different precision across comparisons. The posted standard deviations and group sizes help readers understand why estimates should be considered with their interval bounds rather than judged by the means alone.

In general, a randomized design can support causal interpretation when allocation, follow-up, adherence, outcome measurement, and analysis are handled appropriately. Randomization is reported, but the record does not report allocation procedures, masking details, missing-data handling, or analysis populations. Thus the degree to which those safeguards were achieved cannot be assessed from the available registry record. The record also does not give an inferential contrast that directly answers whether TMR during a particular sleep stage changes structure transfer.

General: limits of a proportion-correct outcome

Accuracy proportions condense responses into a single summary. This is useful for comparing average task performance, but it can obscure whether errors cluster in particular categories or features. If responses are constrained or correlated within participant or task item, a model that accounts for those structures may be relevant in general. Whether the trial used such a model, or checked assumptions about the outcome distribution, is not reported in the registry record. No additional model or adjustment should be presumed from the posted means.

10. Confidence Intervals and Precision

The comparison estimates have interval bounds that express statistical uncertainty, while each group’s standard deviation describes variation in participant scores. These are different quantities. The registry record does not report the confidence level or construction method for the displayed interval bounds, so their exact coverage interpretation cannot be specified from this record.

For Immediate Congruent versus Immediate Incongruent, the estimated mean difference is 0.010 and the bounds are -0.028 to 0.047. Because the bounds extend across zero, the estimate is imprecise with respect to whether the group difference is positive or negative. The unadjusted p-value is 0.614. It would be inappropriate to interpret this result as demonstrating equivalence; no equivalence margin or equivalence analysis is reported.

The Aim 2 comparisons have positive estimates and bounds entirely above zero: Awake Incongruent, 0.067 with bounds 0.021 to 0.113; Awake Congruent, 0.064 with bounds 0.014 to 0.113; Sleep Incongruent, 0.066 with bounds 0.010 to 0.121; and Sleep Congruent, 0.060 with bounds 0.013 to 0.107. The bounds still span a range of plausible magnitudes, particularly in relation to the point estimates. They convey uncertainty and should not be reduced to a yes-or-no decision based only on whether zero is included.

Precision differs across contrasts, in part because group sizes and standard deviations differ. The Immediate Congruent group has 47.0 participants, whereas Aim 2 comparison groups have 26.0, 25.0, 22.0, and 27.0. The listed standard deviations also vary. These features help explain why the estimates should be interpreted with their respective bounds. Precision does not establish causality, and a narrow interval would not correct design or analysis limitations.

11. Planning a Similar Trial: Sample Size

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

The planning figures are participants per arm for 80% power and participants per arm for 90% power. They are not total event counts; the special event-count convention for time-to-event outcomes does not apply to this continuous outcome. The calculations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect. These figures are planning illustrations, not observed power.

In plain language, the 80% power column gives the number of participants per arm that the planning calculation estimates would be needed for an 80% chance of detecting an effect of the specified size under its assumptions. The 90% power column makes the analogous calculation for a 90% chance. These are prospective planning quantities, not properties observed in the completed study.

The rows consider planning for 100%, 75%, and 50% of the observed effect. Smaller effects need more participants to detect under the same assumptions. For example, for the Immediate Congruent comparison, the planning figures for 100% of the observed effect are 1,449 participants per arm for 80% power and 1,940 for 90% power; planning for 50% of that effect gives 5,794 and 7,757. Such figures depend strongly on the observed effect, which can be unstable, especially when estimates are uncertain.

The original trial’s design assumptions, including its own intended power target and sample-size justification, are not reported in the registry record. These calculated planning illustrations should not be treated as evidence that the trial was designed to detect the posted differences or that its actual power can be inferred from the results.

12. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Observed difference Δ = 0.010; pooled standard deviation σ = 0.092.
2. n = 2 × (1.96 + 0.842)² × 0.092² / 0.010² = 1,448.4, rounded up to 1,449 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 <- 0.00960000000000005                      # observed difference in means
sigma <- 0.0922154542362613                     # 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(5746299)
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 = 1,449 (80% power), 1,940 (90% power)
 75% of observed effect: n per arm = 2,576 (80% power), 3,448 (90% power)
 50% of observed effect: n per arm = 5,794 (80% power), 7,757 (90% power)
power.t.test: 1449.4 
Simulated power at n = 1,449 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.

13. Multiplicity, Missing Data and Analysis Populations

The registry record reports one primary outcome measure and zero analyses. The posted recalculations include five comparisons against the Immediate Incongruent reference group, with unadjusted p-values. No multiplicity adjustment is reported. When several hypotheses are examined, the chance of at least one small p-value under a collection of null hypotheses can increase; in general, multiplicity procedures address that issue. Whether the investigators planned other comparisons or how they handled multiple testing is not reported in the registry record.

The five comparisons do not cover every scientific question suggested by the design. In particular, the listed results are not direct contrasts of congruent versus incongruent learning within Aim 2, awake versus sleep, or sleep-stage-specific reactivation. The registry record does not report those contrasts or an analysis plan that would identify primary and secondary comparisons. Therefore, the unadjusted results should not be used to claim that one of these specific mechanisms was established.

Methods for missing outcome data are not reported in the registry record. The available information does not state whether all randomized participants completed the task, how missing measurements were handled, or whether any imputation or sensitivity analysis was used. The analysis population is also not reported: it is not possible to tell from the record whether the posted group summaries follow an intention-to-treat, per-protocol, or other approach. The endpoint group sizes are available, but the record does not provide participant flow or reasons for exclusions.

These omissions matter because missingness and analysis-population choices can affect the interpretation of group comparisons. Randomization alone does not show that all participants were retained or analyzed as assigned. The record’s completed status describes trial status, but it does not fill in the missing analysis details. No assumptions about the investigators’ handling of multiplicity or missing data should be substituted for the absent registry information.

14. Safety Results

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

Deaths

Other (non-serious) adverse events

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

The safety record lists serious adverse events, other adverse events, and deaths by group. For the six listed groups with participants at risk, the serious adverse event count is 0 in each group, the other adverse event count is 0 in each group, and deaths are 0. The Sleep Congruent (REM) safety entry records 0 at risk and 0 events, so it does not provide an observed safety count for participants in that group.

Zero reported events means no events were recorded in the relevant registry fields for the listed at-risk groups; it does not establish that the risk is absent. The record does not report event definitions, surveillance procedures, or further detail on how safety information was collected. The safety figures should not be interpreted as evidence about outcomes beyond the events and groups recorded.

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

The posted outcome means show accuracy on a task intended to test use of previously learned structure when completing features for a new modular category. All Aim 2 groups have higher posted means than the Immediate Incongruent reference group, and the unadjusted summary-based intervals for those comparisons lie above zero. The Immediate Congruent comparison has a smaller estimated difference, with bounds that cross zero. These are descriptions of the posted data and their uncertainty, not conclusions that a particular treatment works or fails.

The data do not establish that sleep itself caused the higher Aim 2 means. The direct comparisons shown use Immediate Incongruent as the reference, and the groups vary in timing, category structure, and other procedures. The record does not report a direct sleep-versus-awake contrast or a direct test of targeted reactivation during SWS versus REM. Although the safety record names a Sleep Congruent (REM) group, the primary outcome results do not include a separate REM mean.

The findings also do not show whether structure knowledge became abstract in a psychological or neural sense. Accuracy on the missing feature task is a behavioral measure. The task may be consistent with transfer, but the group averages alone cannot reveal the representation participants used, identify a causal mechanism, or describe individual performance.

Interpretation is further limited by the absence of a registry-reported analysis plan, sponsor inferential analyses, confidence-interval level, multiplicity procedure, missing-data method, and analysis population. The unadjusted recalculations offer one way to contextualize the group summaries, but they should not be mistaken for the investigators’ complete analysis. The posted results therefore support a cautious account of group-level task performance, not a definitive claim about causal effects or the specific role of sleep and reactivation.

16. Why This Trial Matters Statistically

This study raises a useful statistical distinction between learning individual category examples and learning relations among features that may transfer to a new category. Its primary behavioral outcome targets this transfer question rather than simply reporting recall of the training material. For analysis, that means the comparison must match the scientific question: a contrast against Immediate Incongruent is not automatically a test of congruence, sleep, or sleep-stage reactivation.

The results also illustrate why point estimates, uncertainty, and design need to be read together. The Aim 2 estimates are positive relative to the reference, but their interpretation depends on group sizes, score variability, the unadjusted nature of the comparisons, and the absence of direct contrasts for several mechanistic questions. A p-value alone cannot express the magnitude or practical meaning of the accuracy differences.

The distinction between posted summaries and a prespecified analysis is important in a multi-condition study. The registry record reports no analyses and does not provide the analysis plan, so readers cannot determine how the investigators prioritized comparisons or managed multiplicity. Transparent reporting of those choices would help clarify which comparisons were confirmatory and which were exploratory. Likewise, explicit participant-flow and missing-data information would make the connection between enrollment and analyzed endpoint groups easier to assess.

Finally, the planning illustrations show how quickly required sample sizes can grow when planning for effects smaller than the observed estimate. That is a general design lesson, not evidence about the trial’s original planning or its observed power. Together, these statistical issues define what the posted evidence can say: it describes performance across the listed conditions, while leaving important causal and mechanistic comparisons unresolved.

17. 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 results include six groups, but the registry’s safety information includes a separate Sleep Congruent (REM) entry without a corresponding outcome mean. The available results therefore do not support a direct comparison of the reported sleep-stage conditions. Some arm descriptions are truncated, limiting the detail available about the procedures.

The comparisons are against Immediate Incongruent rather than matched contrasts designed to isolate one feature of the protocol. Groups differ in timing and category-structure conditions, and the posted results do not provide direct comparisons between awake and sleep conditions or between congruent and incongruent conditions within Aim 2. Consequently, differences cannot be attributed specifically to sleep, congruence, or targeted memory reactivation based on these comparisons alone.

Important design and analysis details are not reported in the registry record, including participant characteristics, eligibility criteria, allocation procedures, the identity of the masked parties, analysis populations, participant flow, missing-data handling, and the sponsor’s statistical analysis. The interval level and method are not stated. Multiplicity handling is also not reported, despite multiple unadjusted comparisons being available.

The outcome is a group-level accuracy summary for a behavioral task. It cannot by itself establish what kind of representation participants formed or how they reached their responses. The posted means and standard deviations also do not show the full score distributions or performance at the feature level. Finally, sample-size planning figures are illustrations based on observed effects and specified assumptions, not evidence of the original design rationale or observed power. These limits call for cautious interpretation of the estimates.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: The Impact of Reactivation During Sleep on the Consolidation of Abstract Information in Humans — NCT05746299.
• Registry record last updated 2025-07-14; status last verified 2025-06.
• Results first posted 2025-07-14.
• 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.

21. Record Summary

The registry record describes a completed randomized, single-masked, parallel behavioral study with enrollment of 194. It examines learning of semantic category structures and accuracy on a Stage 2 missing feature task for a new modular category. Six groups have posted primary outcome means and standard deviations; the listed comparisons use Aim 1 Immediate Incongruent as the reference.

Compared with that reference, the four Aim 2 groups have positive unadjusted mean-difference estimates with interval bounds above zero. The Immediate Congruent comparison has an estimate of 0.010 and bounds that cross zero. These summary-based recalculations are not sponsor-reported analyses; the registry record lists zero analyses. It does not report multiplicity or missing-data methods, analysis populations, or a direct sleep-versus-awake or sleep-stage contrast. Safety entries record no adverse events or deaths for groups with participants at risk, while the REM entry records no participants at risk.