Impact of Theta Burst Stimulation Over the Left DLPFC on Language Switching: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Impact of Theta Burst Stimulation Over the Left DLPFC on Language Switching (NCT03148470), a randomized na study. The primary outcome is RTs for Picture Naming With the Factors Language (L1, L2) and Block (Switching, Non-Switching).

2. Clinical Question

3. Trial Design

The registry describes a completed, randomized, double-masked study with a factorial model designation. It lists an enrollment of 41 and four device interventions: Sham iTBS, Sham cTBS, iTBS, and cTBS. The intervention descriptions pair the active stimulation with the corresponding sham condition. The registry does not describe the allocation sequence, the factorial factors in detail, or how masking was maintained, so those aspects cannot be reconstructed from the record.

The study began in 2016-09, and primary completion was recorded in 2018-07. Its stated purpose was to investigate whether theta burst stimulation over the left dorsolateral prefrontal cortex influences language switching and cognitive flexibility in healthy bilinguals. The primary outcome was assessed within 30 minutes after intervention. It measured speech-onset response times during picture naming, with language and block type as factors. A secondary outcome concerned nonverbal switching, also assessed within 30 minutes after intervention.

The primary outcome description identifies the intended active-condition comparison as iTBS versus cTBS and says sham stimulation was used for baseline control between the two groups. The posted primary outcome summaries contain 20 participants for each active stimulation group. Summaries by language and block list 40 participants each. The record does not explain these different summary counts in relation to the enrollment of 41. The registry reports mixed models analysis for the primary and secondary outcomes, but does not give the model formula, covariates, handling of repeated observations, or analysis population definition. Those design and analysis details are 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

Analysis population for the primary outcome, as defined in the registry: As a primary outcome we compared RTs following iTBS vs cTBS. Sham-TBS is only used to have a baseline control between the two groups

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 RTs for Picture Naming With the Factors Language (L1, L2) and Block (Switching, Non-Switching), measured over Within 30 minutes after intervention.

The primary endpoint was picture-naming response time, measured in milliseconds as the mean and accompanied by a standard deviation. The registry describes it as speech-onset time. Participants named pictures in L1, their mother tongue, and L2, their later-acquired second language. Naming was assessed in switching blocks, where picture naming alternated between languages, and non-switching blocks, where naming took place in one language.

Response time is a continuous measure: it records the amount of time taken to produce a response, rather than classifying a response into a category. The endpoint is suited to describing performance across the language and block conditions named in the study question. The registry’s outcome description therefore connects the measure to both language switching and non-switching picture naming. It does not, by itself, establish why any observed response-time difference occurred or whether that difference has practical importance.

The outcome timeframe was within 30 minutes after intervention, so the posted endpoint addresses performance in that specified period. The registry record does not state how response times were processed, whether incorrect responses or unusually long times were excluded, or whether the primary endpoint was a particular interaction among language, block, and stimulation. Those analytic details are not reported in the registry record.

7. Results Posted in the Registry

• RTs for PN After Continuous Theta Burst Stimulation: mean 1,107 versus 1,103 on control; difference 3.94 (95% CI -75.72 to 83.60), standardized effect d = 0.03; z-test p = 0.923.
• RTs for Picture Naming in L1: mean 1,099 versus 1,103 on control; difference -4.50 (95% CI -141.62 to 132.62), standardized effect d = -0.01; z-test p = 0.949.
• RTs for Picture Naming in L2: mean 1,159 versus 1,103 on control; difference 55.59 (95% CI -89.23 to 200.41), standardized effect d = 0.15; z-test p = 0.452.
• RTs for Picture Naming in the Non-switching Blocks: mean 1,081 versus 1,103 on control; difference -22.18 (95% CI -161.86 to 117.50), standardized effect d = -0.06; z-test p = 0.756.

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.

Nonverbal Switching

• 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:
  • Continuous Theta Burst Stimulation: mean 739.32 versus 772.07 on control; difference -32.75 (95% CI -124.89 to 59.39), standardized effect d = -0.22; z-test p = 0.486.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Continuous Theta Burst Stimulation and the control. The interval gives the range of effects that remain compatible with the data.

The registry posts one secondary outcome, nonverbal switching, measured in milliseconds within 30 minutes after intervention. Its reported analysis is mixed models analysis with p = 0.642. The record does not provide an estimate or confidence interval for that model. The simple recalculation from the posted group summaries has a mean difference of -32.75 milliseconds and a 95% confidence interval from -124.9 to 59.39; the interval includes no difference and is broad relative to the point estimate.

At a broad level, this secondary result shares the primary analysis’s limitation: the posted model-based result is a p-value without an accompanying effect estimate or interval, while the summary-based recalculation is unadjusted and cannot reproduce the mixed model. It therefore adds an additional outcome perspective but does not resolve uncertainty in the primary language-related question. Its measurement is related to cognitive flexibility, but it is not the same endpoint as picture-naming response time across language and block conditions.

Secondary outcomes generally carry less evidential weight when they are not designated as the primary question, particularly when multiple outcomes or contrasts may be examined. Testing multiple hypotheses can increase the chance of finding an apparently notable result by chance unless the analysis plan addresses multiplicity. The registry record does not report a multiplicity procedure, so it is not possible to determine whether or how any adjustment was applied. The secondary result should consequently be treated as supportive or exploratory context rather than as an independent confirmation of a primary effect.

The limited number of posted secondary outcomes does not remove these concerns: the registry does not state the full set of outcomes considered, the prespecified hierarchy, or whether analyses were selected for reporting. The secondary findings are best read with the same emphasis on the specified outcome, analysis method, and uncertainty as the primary findings. No stronger conclusion about a general effect on cognitive flexibility follows from the posted secondary analysis alone.

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

• Nonverbal Switching

9. Statistical Interpretation

10. Statistical Methods Explained

The registry identifies mixed models analysis for both the primary and secondary outcomes. A mixed model is a general statistical approach for outcomes that may have related observations, such as repeated measurements from the same participant or measurements organized by several experimental factors. It can represent both fixed effects, which describe the contrasts of interest, and random effects, which account for sources of variation or grouping. This is general statistical context; the registry record does not report which fixed or random effects were included in this study’s model.

What does a mean difference describe?

A mean difference expresses the difference between two group means on the original measurement scale. For the primary endpoint, that scale is milliseconds. The recalculated comparisons use the posted group means and standard deviations to produce simple unadjusted estimates. They are not adjusted for other factors and should not be confused with the registry’s reported mixed-model result. In particular, the comparisons across language and block summaries do not reveal the full interaction structure that may have been considered in the mixed model.

The direction of a mean difference depends on the order of the groups in the comparison. A positive estimate indicates that the second listed mean is higher than the first under the displayed contrast, while a negative estimate indicates the reverse. Interpretation should focus on the contrast definition as well as the sign. These are response-time measurements, so a difference describes the numerical separation of average times; it does not by itself identify a mechanism or establish practical importance.

How should a confidence interval be understood?

A confidence interval is a range of values compatible with the data and the calculation method at the stated confidence level. It communicates uncertainty around an estimate, not the probability that a fixed effect lies in the particular interval. Wider intervals indicate less precision than narrower intervals. When an interval includes no difference, the data do not distinguish the estimate from no difference at the interval’s confidence level; that does not establish that the groups are equal. The recalculated intervals are based on simple comparisons of summary statistics, whereas the registry’s mixed-model analysis has no posted interval.

The interval’s width depends on both the variability in the measurements and the amount of information available. The posted standard deviations for response time are substantial in relation to the estimated differences. In general, high variability makes it more difficult to estimate a group contrast precisely when the sample is limited. A standard deviation describes spread among measurements; it is not the same as the standard error, which describes uncertainty in an estimated mean. The recalculations report standard errors for the group means, but the registry record does not report the corresponding model-based uncertainty for its mixed-model contrasts.

Why might a mixed model matter?

When each participant contributes measurements under more than one language or block condition, those observations may be correlated. Treating related observations as independent can misrepresent uncertainty. Mixed models can account for such dependence, depending on the model structure. This is a general reason to use a mixed model in a repeated-measures setting, not a claim about the exact structure used in this study. The registry lists language and block factors and names mixed models analysis, but does not report whether participants contributed all combinations, what covariance structure was used, or which interaction terms were tested.

Factorial designs can examine more than a single average contrast. For example, a model may consider whether an intervention contrast varies by a second factor. That is a general possibility, not a description of this study’s specified tests. Although the registry labels the design factorial and identifies language and block factors for the primary outcome, it does not report the precise factor coding, primary model contrast, or hierarchy of hypotheses. Consequently, the reported overall mixed-model p-value cannot be mapped confidently to one specific effect from the registry record alone.

What can and cannot be inferred from the reported p-values?

A p-value measures how incompatible the observed data are with a specified statistical model under a null hypothesis. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. The registry reports p = 0.932 for the primary mixed-model analysis and p = 0.642 for the secondary mixed-model analysis. Because estimates, confidence intervals, and model details are not posted for those analyses, the p-values alone give limited information about the magnitude and precision of the model-based effects.

The recalculated p-values are unadjusted and derive from simple summary-statistic comparisons. They do not account for the model structure, and they do not incorporate any multiplicity adjustment. A p-value should therefore be read alongside its estimate, interval, analysis method, and the question being tested. For this study, the model specification and multiplicity procedures are not reported in the registry record, so neither can be inferred from the p-values.

What does over-dispersion mean in general?

Over-dispersion occurs when the variability in data is greater than a statistical model assumes. It is especially discussed for count outcomes, where an overly restrictive variance assumption can make standard errors too small. The primary and secondary outcomes in this study are continuous response times, and the registry does not report an over-dispersion assessment. The general lesson is that a model’s variance assumptions should fit the outcome and data structure; the record does not give enough detail to evaluate those assumptions for this analysis.

Overall, the posted information supports a distinction between the registry’s named model-based analysis and the unadjusted recalculations based on group summaries. The former has a reported method and p-value but lacks an estimate and interval. The latter gives interpretable mean contrasts and intervals, but does not reproduce the reported model. The registry does not report enough methodological detail to reconcile those two presentations beyond that distinction.

11. Confidence Intervals and Precision

The recalculated confidence intervals communicate substantial uncertainty around the primary response-time contrasts. For the active stimulation comparison, the estimated mean difference is 3.94 milliseconds and the 95% confidence interval extends from -75.72 to 83.6. The interval includes no difference and spans both negative and positive values. The point estimate is therefore not a precise account of the possible contrast, and the interval does not support a conclusion of equivalence.

The language- and block-specific recalculations likewise have intervals that include no difference. For L1, the estimate is -4.5 milliseconds, with an interval from -141.6 to 132.6. For L2, the estimate is 55.59 milliseconds, with an interval from -89.23 to 200.4. The non-switching block estimate is -22.18 milliseconds, with an interval from -161.9 to 117.5; the switching block estimate is 71.92 milliseconds, with an interval from -69.47 to 213.3. These ranges indicate that the estimates are not tightly determined by the posted summaries.

The secondary nonverbal switching comparison also has a broad interval: its recalculated mean difference is -32.75 milliseconds, with a 95% confidence interval from -124.9 to 59.39. As with the primary comparisons, this interval is not the registry’s mixed-model interval. The registry does not report confidence intervals for its model-based analyses, so the precision of those analyses cannot be read directly from the posted p-values.

Confidence intervals describe uncertainty under their calculation assumptions. The recalculated intervals are unadjusted and based on summary statistics, not on a fully specified mixed model. Their breadth is consistent with limited precision in these comparisons; it should not be interpreted as proof that all differences within the ranges are equally plausible or equally important. The registry record does not report a threshold for a meaningful response-time difference, which also limits interpretation of the endpoints of these intervals.

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 illustrations give participants per arm for 80% power and participants per arm for 90% power. They are based on two-sided significance level 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. The numbers are planning illustrations based on the observed effect, not a reconstruction of the trial’s original design assumptions. The original sample-size assumptions are not reported in the registry record.

The 80% power column gives the number of participants per arm that the stated planning calculation would require to have an 80% chance of detecting the specified effect under its assumptions. The 90% power column gives the corresponding requirement for a 90% chance. These are prospective design quantities: they describe hypothetical sample sizes for a planned study, not the probability that the completed study detected an effect. This is not observed power.

Each planning row uses a different fraction of the observed effect. Planning for 100% of the observed effect uses the full estimate as the target; planning for 75% or 50% uses a smaller target. Smaller effects need more participants to detect under the same assumptions, which is why the displayed requirements rise as the target effect becomes smaller. The requirements can vary markedly across outcomes because the observed estimates and variability differ.

These calculations assume equal allocation between two arms and a normal approximation. They do not incorporate any unreported design features or the mixed-model structure. The figures are participants per arm, not total participants, and they are not counts of events. The registry’s own justification for its enrolled sample is 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 = 3.940; pooled standard deviation σ = 128.524.
2. n = 2 × (1.96 + 0.842)² × 128.524² / 3.940² = 16,703.8, rounded up to 16,704 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 <- 3.94000000000005                      # observed difference in means
sigma <- 128.52431676535                     # 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(3148470)
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 = 16,704 (80% power), 22,362 (90% power)
 75% of observed effect: n per arm = 29,696 (80% power), 39,755 (90% power)
 50% of observed effect: n per arm = 66,816 (80% power), 89,447 (90% power)
power.t.test: 16704.8 
Simulated power at n = 16,704 per arm: 0.792 (20000 replicates)

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

14. Analyses Reported by the Sponsor

The registry reports a mixed models analysis for the primary outcome, with p = 0.932. It does not provide the model-based effect estimate, confidence interval, or detailed model specification. This is the sponsor’s reported analysis as recorded in the registry, and it addresses an analysis across the listed outcome groups rather than presenting a single simple contrast with an interval.

The recalculated comparisons are simple, unadjusted comparisons of posted means and standard deviations. For the active stimulation comparison, the recalculated mean difference is 3.94 milliseconds, with a 95% confidence interval from -75.72 to 83.6 and an unadjusted p-value of 0.923. This calculation is not a re-fit of the mixed model and cannot supply the model’s missing estimate or interval. The small difference between the two p-values should not be interpreted as a contradiction, because the methods and scope of the analyses differ and the registry does not give enough model detail for a direct reproduction.

The record marks no discrepancy between the posted results and the recalculation. Even so, the recalculation should be read as a supplementary unadjusted description rather than as a replacement for the stated mixed-model analysis. The registry result is the only reported model-based inferential analysis, while the recalculation makes the group-summary contrast and its uncertainty more explicit. Neither presentation establishes a narrowly precise effect, and the absence of the model estimate limits comparison between them.

15. Multiplicity, Missing Data and Analysis Populations

The primary outcome involves language and block factors, and the registry also lists a secondary outcome. This creates more than one possible contrast or question of interest, but the record does not describe which hypotheses were designated as primary within the factorial structure or how the family of tests was defined. A multiplicity procedure, such as an adjustment or a prespecified testing hierarchy, is not reported in the registry record.

Multiplicity matters because the chance of observing at least one small p-value can increase when many hypotheses are tested. This is a general statistical issue, not evidence that the study made an inappropriate choice. Without the analysis plan or a multiplicity statement, it is not possible to tell whether the reported p-values were adjusted, whether particular contrasts were prioritized, or how exploratory analyses were separated from confirmatory ones. The recalculated p-values are explicitly unadjusted and should not be treated as multiplicity-controlled results.

The registry names mixed models analysis but does not report the model formula, estimation method, covariance structure, missing-data approach, or assumptions used for inference. It does not identify an analysis population, such as all randomized participants or participants with complete outcome data. Those details are important for understanding which observations contributed to each estimate and whether any departures from the intended allocation were handled in a particular way. They are not reported in the registry record.

The overall enrollment is 41, while the posted primary outcome summaries list 20 participants in each active stimulation group and 40 for the language and block summaries. The record does not explain the relationship among those counts, describe participant flow, or state whether any outcome data were missing. It is therefore not possible to infer whether unreported participants lacked measurements, were excluded from an analysis, or were handled in another way. No explanation should be inferred from the numerical difference alone.

The registry record also does not report interim analyses, stopping rules, or the original sample-size justification. Their absence from the record does not establish that such procedures were or were not used; it means they cannot be assessed from the posted information. Overall, the missing analysis details limit independent evaluation of multiplicity, missing-data handling, and the exact populations contributing to the model-based results.

16. Safety Results

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

Deaths

Other (non-serious) adverse events

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

Most frequent non-serious adverse event terms

The adverse-event summaries list serious adverse events and deaths as 0 in each of the four recorded groups: cTBS, iTBS, Sham iTBS, and Sham cTBS. The numbers at risk are 20 for cTBS, 20 for iTBS, 21 for Sham iTBS, and 20 for Sham cTBS. These are the registry’s posted counts; they should be read as event counts within the listed groups, not as proof that no unrecorded event occurred.

For other, non-serious adverse events, the registry records 1 affected participant among 21 at risk in the Sham iTBS group and 0 affected participants in the other listed groups. The record does not provide further details about the event or its seriousness beyond its classification in the summary. No serious adverse-event terms are listed.

These summaries are descriptive safety reporting within this study and do not establish a comparative safety profile. The counts are small, and the registry does not report further event descriptions, duration of follow-up for safety, or a formal comparison between groups. The absence of recorded serious events in these summaries should therefore be interpreted narrowly as what the registry reports, not as a broader conclusion about safety.

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

The posted results describe response times after iTBS and cTBS, along with summaries by language and block condition. For the primary active-condition comparison, the group means are close, and the unadjusted recalculation gives a small mean difference relative to a wide confidence interval. The registry’s mixed-model result is p = 0.932, but its estimate and interval are not reported. Taken together, the available statistics do not specify a precise difference between these stimulation conditions.

That interpretation is not the same as saying the conditions are equivalent. Demonstrating equivalence requires an appropriately designed analysis and a prespecified margin defining which differences would count as sufficiently small. Such a margin and an equivalence analysis are not reported in the registry record. A non-significant p-value or interval containing no difference cannot establish equivalence.

The results also do not establish that theta burst stimulation has no effect on language switching or cognitive flexibility. The study’s estimates are uncertain, the model details are incomplete, and the registry does not report a threshold for a meaningful change in response time. Nor can these results be generalized beyond the population and measurements described in the registry. The study concerned healthy bilinguals and picture-naming response time within 30 minutes after intervention.

The primary and secondary outcomes address specific laboratory tasks. They do not directly describe all aspects of language use or cognition. The registry’s intention was to investigate whether stimulation influences language switching and cognitive flexibility, but the posted statistics only support a cautious account of the measured response-time contrasts. They do not identify a mechanism, establish persistence beyond the stated timeframe, or show practical importance.

Accordingly, the most defensible reading is that the record reports no statistically clear separation in its stated analyses, while the recalculated intervals remain wide enough to leave substantial uncertainty about the possible differences. The data are informative as estimates from this study, but not as definitive evidence of no effect or of a particular effect size.

18. Why This Trial Matters Statistically

This study illustrates a central statistical challenge in small clinical and experimental studies: a point estimate can appear close to no difference while the uncertainty around it remains broad. The active-condition means are 1103.0 and 1107.0 milliseconds, but the recalculated interval extends from -75.72 to 83.6. The interval is essential to interpretation because the point estimate alone would conceal how many different underlying contrasts remain compatible with the posted summary data.

The record also makes clear why an analysis label is not enough to reproduce a result. It names mixed models analysis, which can be appropriate when outcomes are organized by multiple factors or may be correlated within participants. Yet the model formula, factor structure, and effect estimates are not reported. Readers can see the reported p-value but cannot fully establish which contrast it represents or compare its precision with a model-based interval.

Another useful statistical lesson is the distinction between a reported model-based result and a simple recalculation from means and standard deviations. The recalculation offers an interpretable unadjusted comparison, but it does not replace the model or account for the same structure. Presenting both kinds of information can help readers understand what is directly available and what remains uncertain, provided the different analyses are not treated as interchangeable.

Finally, planning calculations based on an observed effect can be strikingly sensitive to the effect chosen for future detection. They are useful for illustrating how target effect size and desired power relate to sample requirements, but they are not a retrospective verdict on the completed study. The original design rationale and sample-size assumptions are not reported in the registry record. Together, these points show how estimates, intervals, analysis details, and planning assumptions each answer different statistical questions.

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 leaves important details unresolved for this specific study. It identifies a randomized, double-masked factorial design and lists four stimulation or sham interventions, but it does not explain the allocation sequence, the factorial structure, or precisely how the active and sham conditions were combined in the primary analysis. The number enrolled is 41, yet the outcome summaries use different group counts. The record does not explain participant flow or the relation between these counts.

The primary outcome is described as response time across language and block factors, but the exact model contrast of interest is not stated. The reported mixed-model analysis has a p-value without an estimate or confidence interval, and the model specification is not reported. This prevents a detailed assessment of how within-participant dependence, interactions, or other covariates were handled. The unadjusted recalculations provide additional context but cannot fill those gaps.

The registry does not report missing-data methods, analysis populations, multiplicity procedures, or the original sample-size justification. It also does not state whether response-time observations were screened or transformed, how incorrect responses were handled, or whether exclusions were made. These are specific concerns for interpreting a task-based response-time outcome, but no particular handling method can be inferred.

The primary measurement was taken within 30 minutes after intervention, so the posted outcome addresses only that timeframe. The registry does not report a threshold for a meaningful response-time change. The posted secondary outcome is nonverbal switching, but its model-based estimate and interval are likewise absent. Safety summaries list few recorded events and do not provide detailed event descriptions. These limitations constrain interpretation of both the magnitude and broader relevance of the results.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Impact of Theta Burst Stimulation Over the Left DLPFC on Language Switching — NCT03148470.
• Registry record last updated 2020-04-22; status last verified 2020-04.
• Results first posted 2020-04-10.
• 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

NCT03148470 is a completed randomized, double-masked study of theta burst stimulation over the left dorsolateral prefrontal cortex in healthy bilinguals. It lists an enrollment of 41 and compares iTBS and cTBS, with sham stimulation included as baseline control. The primary endpoint is mean picture-naming response time in milliseconds across language and switching-block conditions, assessed within 30 minutes after intervention.

The registry reports mixed models analysis for the primary outcome with p = 0.932, but no model-based effect estimate or confidence interval. The unadjusted recalculation for the active-condition comparison estimates a mean difference of 3.94 milliseconds, with a 95% confidence interval from -75.72 to 83.6. The secondary nonverbal switching analysis reports p = 0.642, also without a model-based estimate or interval. The record does not report the model details, multiplicity procedure, missing-data handling, or original sample-size assumptions.