Atomoxetine to Treat Korean Children and Adolescents With Attention-Deficit/Hyperactivity Disorder (ADHD): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Atomoxetine to Treat Korean Children and Adolescents With Attention-Deficit/Hyperactivity Disorder (ADHD) (NCT00568685), a randomized phase 3 study. The primary outcome is Change From Baseline to Day 42 Endpoint in Attention-Deficit/Hyperactivity Disorder Rating Scale-IV-Parent Version: Investigator Administered and Scored (ADHDRS-IV-Parent:Inv) Total Score.

2. Clinical Question

3. Trial Design

The registry describes a Phase 3b, randomized, open-label, parallel-group study of atomoxetine in Korean pediatric outpatients with ADHD. The study was completed and enrolled 153 participants. The recorded study start was 2007-11, and primary completion was 2008-11. Participants were assigned to one of three atomoxetine dose groups: 0.2 mg/kg/day, 0.5 mg/kg/day, or 1.2 mg/kg/day. The primary endpoint analysis included all randomized participants, which the record identifies as the intent-to-treat population.

Randomization means treatment assignment was made according to an allocation process rather than being selected solely by the participant or investigator. In general, randomization can help balance characteristics between groups on average, although it does not guarantee identical groups. The registry does not report details of the randomization sequence, allocation concealment, or the way assignment was implemented.

Open-label means that masking was not used. The registry lists masking as none. In general, knowledge of assigned treatment can influence participant reports, investigator ratings, treatment decisions, or the recording of outcomes, depending on how those outcomes are collected. The endpoint was investigator administered and scored, and the record does not indicate that the assessor was masked.

The parallel model means participants were assigned to a dose group rather than being described as crossing between groups. No placebo or untreated group is listed. The study purpose in the registry is to provide information on the relative effectiveness of three atomoxetine doses. The record does not report the protocol’s original sample-size justification, detailed eligibility criteria, treatment adherence, or the planned duration of dosing. The listed primary outcome is assessed from baseline to Day 42, and the registry also lists secondary assessments at several times and for safety-related outcomes.

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:

Analysis population for the primary outcome, as defined in the registry: Analysis included all randomized participants (intent-to-treat population)

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline to Day 42 Endpoint in Attention-Deficit/Hyperactivity Disorder Rating Scale-IV-Parent Version: Investigator Administered and Scored (ADHDRS-IV-Parent:Inv) Total Score, measured over Baseline, Day 42.

The primary endpoint is change from baseline to the Day 42 endpoint in the total score on the Attention-Deficit/Hyperactivity Disorder Rating Scale-IV-Parent Version, investigator administered and scored. The registry describes the scale as measuring the 18 symptoms contained in the DSM-IV diagnosis of ADHD. Individual item scores range from 0, described as none/never or rarely, to 3, described as severe/very often. Total scores range from 0, no symptoms, to 54, highly symptomatic.

A change score represents the difference between the baseline measurement and the later measurement. In general, change from baseline is used to describe how an outcome has shifted during follow-up; when groups are compared, the between-group contrast addresses whether the recorded change differs across assignments. Because higher values on this scale represent more symptoms, a negative change corresponds to a lower score at follow-up than at baseline. That direction describes the score, not by itself the clinical importance of the change.

The endpoint is a continuous measure, and the posted group estimates are least-squares means with 95% confidence intervals. Least-squares means are model-based group estimates rather than simple raw averages. The record does not provide the fitted model’s full specification, so the adjustment variables and other modeling details are not reported in the registry record. A Day 42 score can help describe change over the specified follow-up, but it does not by itself establish how durable any difference may be beyond that endpoint.

7. Results Posted in the Registry

• Atomoxetine 0.5 mg/kg/Day: mean -12.31 versus -9.55 on control; difference -2.76 (95% CI -6.25 to 0.73), standardized effect d = -0.31; z-test p = 0.121.
• Atomoxetine 1.2 mg/kg/Day: mean -14.51 versus -9.55 on control; difference -4.96 (95% CI -8.46 to -1.46), standardized effect d = -0.55; z-test p = 0.006.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a numerical total score, and the registry reports a mean-based estimate for each dose group. The posted group values are least-squares means, accompanied by 95% confidence intervals. The pairwise recalculations use mean differences derived from the group summaries. These are different statistical summaries: the sponsor’s reported result comes from an ANCOVA, while the recalculated results are simple, unadjusted contrasts.

What does ANCOVA test?

ANCOVA is analysis of covariance. In general, it compares group outcome means while accounting for one or more covariates in a statistical model. In a change-from-baseline setting, baseline score is often a potentially relevant covariate, but the registry does not specify whether it was included in this particular model. The covariates, model form, and other implementation details are not reported in the registry record. The reported ANCOVA result covers all three dose groups and has p = 0.024. Without model details, the p-value can be described as the reported overall result but not reconstructed from the posted summaries.

An overall test asks whether the data are compatible with a common group mean under the model, against an alternative in which group means are not all the same. It does not, on its own, say which dose groups differ. Separate pairwise tests ask narrower questions about particular pairs. Those questions may require a planned procedure for controlling the chance of false-positive findings when several comparisons are made. Whether such a procedure was specified or applied is not reported in the registry record.

What do the recalculated contrasts represent?

The posted recalculations compare each of the 0.5 mg/kg/day and 1.2 mg/kg/day groups with the 0.2 mg/kg/day group. The estimated mean differences are -2.76 and -4.96, respectively. The contrast is expressed as the higher-dose group’s mean change minus the 0.2 mg/kg/day group’s mean change; a negative estimate therefore indicates a more negative mean change in the compared group. The estimates are unadjusted recalculations from posted summary statistics, not the sponsor’s fitted ANCOVA contrasts.

The recalculations use pooled standard deviations estimated approximately from the reported standard errors and sample sizes. The record marks this approximation as used. Such a calculation can be useful for a transparent comparison based on summaries, but it may not reproduce the original model’s estimate, standard error, or p-value. It does not incorporate covariate adjustment unless that information is contained in the summary values, which is not established. The presence of a discrepancy between the sponsor’s analysis and the recalculations means the sponsor’s own analysis takes precedence.

How should a confidence interval be understood?

In general, a confidence interval gives a range of values compatible with the data and the method used, while reflecting sampling uncertainty. It is not a guarantee that the true effect lies in the interval, and it does not measure every source of uncertainty. The pairwise 95% interval for the 0.5 mg/kg/day comparison includes zero, while the interval for the 1.2 mg/kg/day comparison does not. The intervals describe the recalculated contrasts, not the sponsor’s adjusted model result. Their interpretation also depends on the assumptions behind the calculation.

A p-value describes how unusual the data, or more extreme data, would be under a specified null model and its assumptions. It is not the probability that the null hypothesis is true, nor a measure of clinical importance. The recalculated p-values are unadjusted; their use does not account for multiple comparisons. The sponsor-reported p-value is from the overall ANCOVA analysis and should not be treated as the p-value for either individual pairwise contrast.

Why do analysis populations and missing observations matter?

The primary analysis population is identified as all randomized participants, or the intent-to-treat population. In general, this approach aims to retain participants in the groups to which they were assigned. However, the population label alone does not explain how participants without a Day 42 measurement were included in a model or how missing data were handled. The registry lists a secondary endpoint using last observation carried forward, but that does not establish that the primary endpoint used the same method. Primary endpoint missing-data handling is not reported in the registry record.

Variability also affects precision. Greater variation among participant scores makes it harder to estimate a group mean difference precisely at a fixed sample size. In general, over-dispersion means observed variability exceeds what a model assumes; if present and not accommodated, uncertainty can be understated. The posted summary statistics do not establish whether over-dispersion was assessed or whether it affected this analysis. Model diagnostics and residual checks are not reported in the registry record.

The endpoint is an investigator-administered and scored symptom scale, and the study was open-label. In general, a lack of masking can matter when outcomes involve judgment or reporting, although the registry does not establish that such influence occurred. Interpretation therefore rests on the recorded design and outcome, without assuming an unreported safeguard or bias correction.

10. Confidence Intervals and Precision

The group-level 95% confidence intervals describe uncertainty around the estimated mean change in each dose group. For 0.2 mg/kg/day, the estimate is -9.55, with a 95% confidence interval from -12.03 to -7.07. For 0.5 mg/kg/day, the estimate is -12.31, with a 95% confidence interval from -14.76 to -9.85. For 1.2 mg/kg/day, the estimate is -14.51, with a 95% confidence interval from -16.99 to -12.04. These intervals concern each group’s estimated change, not the differences between groups.

The pairwise recalculated interval for 0.5 mg/kg/day versus 0.2 mg/kg/day is -6.25 to 0.730. Because it includes zero, the summaries are compatible with a range that includes no mean difference as well as differences in either direction. The corresponding unadjusted p-value is 0.121. For 1.2 mg/kg/day versus 0.2 mg/kg/day, the recalculated interval is -8.464 to -1.456, and the unadjusted p-value is 0.006. This interval excludes zero under the recalculation. These are approximate, unadjusted pairwise results and are not the sponsor’s ANCOVA contrasts.

In general, a narrower confidence interval indicates greater precision for the estimate under the statistical method used; a wider interval indicates more uncertainty about the range of compatible values. Precision is not the same as importance. A precisely estimated small difference need not be important, and an imprecise estimate does not establish absence of a difference. The registry does not report a threshold for a clinically important change in this scale, so the intervals cannot be judged against such a threshold from the record.

Confidence intervals also reflect only the uncertainty represented by their method and assumptions. They do not automatically account for open-label outcome assessment, missing data, model misspecification, or multiple testing. The reported overall ANCOVA result has a p-value but no group contrast estimates or confidence intervals in the registry record. Its precision and the size of individual adjusted differences therefore cannot be assessed from the reported overall test alone.

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 sample-size entries are planning illustrations based on the observed pairwise effect, not the trial’s original design justification. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and a target effect equal to the stated fraction of the observed effect. The listed figures are participants per arm for 80% power and participants per arm for 90% power. They are not total sample sizes.

For the comparison of 0.5 mg/kg/day with 0.2 mg/kg/day, the planning entries give 167 participants per arm for 80% power and 224 participants per arm for 90% power when planning for 100% of the observed effect. For 75% of the observed effect, the corresponding figures are 297 and 397 participants per arm. For 50% of the observed effect, they are 667 and 893 participants per arm.

For the comparison of 1.2 mg/kg/day with 0.2 mg/kg/day, the entries are 52 participants per arm for 80% power and 70 participants per arm for 90% power when planning for 100% of the observed effect. For 75% of the observed effect, they are 93 and 124 participants per arm; for 50% of the observed effect, they are 208 and 279 participants per arm.

In general, 80% power means that, under the planning assumptions and for the specified effect, the design would have an 80% chance of detecting an effect at the chosen significance level. The 90% column uses a higher target chance of detection under those same assumptions. Smaller effects require more participants to achieve a given power target, as the entries illustrate. These calculations are not observed power: they are planning illustrations based on the observed effect, not a measure of what the completed trial achieved. The original trial’s sample-size assumptions are not reported in the registry record.

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 = -2.760; pooled standard deviation σ = 8.991.
2. n = 2 × (1.96 + 0.842)² × 8.991² / -2.760² = 166.6, rounded up to 167 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 <- -2.76                      # observed difference in means
sigma <- 8.99082885911818                     # 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(568685)
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 = 167 (80% power), 224 (90% power)
 75% of observed effect: n per arm = 297 (80% power), 397 (90% power)
 50% of observed effect: n per arm = 667 (80% power), 893 (90% power)
power.t.test: 167.5 
Simulated power at n = 167 per arm: 0.804 (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. Analyses Reported by the Sponsor

The registry reports the sponsor’s own primary analysis as an ANCOVA across the three dose groups, with p = 0.024. That is an overall analysis. No adjusted pairwise estimates or confidence intervals are reported alongside it, so this result should not be assigned to a specific dose-pair contrast.

The pairwise mean differences, confidence intervals, and p-values discussed in the recalculations are simple unadjusted calculations from posted summary statistics. The comparison of 0.5 mg/kg/day with 0.2 mg/kg/day has an estimated mean difference of -2.76, a 95% confidence interval from -6.25 to 0.730, and an unadjusted p-value of 0.121. The comparison of 1.2 mg/kg/day with 0.2 mg/kg/day has an estimated mean difference of -4.96, a 95% confidence interval from -8.464 to -1.456, and an unadjusted p-value of 0.006.

The sponsor’s result and these recalculations do not represent the same analysis: the sponsor reports a model-based overall test, while the recalculations are approximate, unadjusted pairwise contrasts. The registry flags a discrepancy. Accordingly, the sponsor’s own analysis takes precedence as the reported trial analysis. The recalculations can provide context about the posted group summaries, but they should not replace or be presented as a reconstruction of the sponsor’s ANCOVA.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and multiple secondary outcomes, including symptom-rating assessments at several time points, discontinuation due to adverse events, completion of a suicide and self-harm assessment, and changes in heart rate, temperature, blood pressure, and weight. Eight secondary outcomes are posted. The presence of multiple outcomes and multiple possible group comparisons raises a general multiplicity question: when many statistical tests are conducted, the chance of at least one apparently unusual result can increase unless the testing strategy accounts for the number and structure of comparisons.

The registry reports an overall ANCOVA p-value of 0.024 for the three groups. It does not report whether that test was part of a prespecified hierarchy, whether follow-up pairwise comparisons were planned, or whether any adjustment for multiple comparisons was used. The recalculated pairwise p-values are explicitly unadjusted and should be read as exploratory summaries rather than as multiplicity-controlled findings. The protocol’s statistical analysis plan is not reported in the registry record.

The primary analysis population is described as all randomized participants, or the intent-to-treat population. That identifies who the analysis was intended to include, but it does not specify the handling of missing measurements. The registry lists a secondary endpoint using last observation carried forward, but it does not state that this approach was used for the primary outcome. The number of missing Day 42 primary scores, the method used for any missing values, and sensitivity analyses for missing data are not reported in the registry record.

In general, missing-data methods rely on assumptions about why outcomes are missing and how missing values relate to observed information. Different assumptions can affect estimates and uncertainty. Likewise, analysis-population definitions can affect interpretation if participants discontinue, do not adhere to treatment, or lack endpoint measurements. The registry does not report enough detail to evaluate those issues for the primary endpoint.

The record also does not report interim analyses, stopping boundaries, or whether the study’s analysis plan was finalized before outcomes were examined. These details should not be inferred from the randomized design or the presence of a completed trial. Without them, the posted analysis can be described, but the full prespecification and multiplicity context cannot be independently assessed.

15. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: road traffic accident (2.1%), appendicitis (1.9%).

The serious adverse event figures are reported for dose bands, rather than the three primary efficacy groups exactly as labeled. In the 0.00-0.35 mg/kg/day band, 51 participants were at risk and 0 had a serious adverse event. In the 0.36-0.85 mg/kg/day band, 54 were at risk and 1 had a serious adverse event. In the >0.85 mg/kg/day band, 48 were at risk and 1 had a serious adverse event.

The registry lists Road traffic accident at 2.1% and Appendicitis at 1.9% among the top serious adverse event terms. It does not identify a dose band alongside those terms in the posted information. These percentages should therefore not be assigned to an individual group on the basis of the record alone. The deaths field is not reported in the registry record.

These are descriptive event counts. In general, a serious adverse event count does not by itself establish whether an event was caused by treatment, and small counts provide limited information about how frequently events might occur beyond the observed participants. The registry does not report a causal attribution for the listed serious events. Other adverse events affected 15 of 51, 22 of 54, and 27 of 48 participants in the respective dose bands; those are distinct from the serious adverse event counts.

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

The posted primary outcome estimates show a decrease from baseline in the ADHD symptom score in each of the three atomoxetine dose groups by the Day 42 endpoint. The estimated mean changes were -9.55 for 0.2 mg/kg/day, -12.31 for 0.5 mg/kg/day, and -14.51 for 1.2 mg/kg/day. Since the scale’s higher scores indicate more symptoms, each estimate describes a lower score at follow-up than at baseline.

These within-group estimates do not establish that atomoxetine caused the changes: the study has no placebo or non-atomoxetine group listed, and changes over time cannot be separated from other influences using within-group comparisons alone. The study’s randomized comparisons are among doses. The unadjusted recalculation shows a pairwise contrast for 1.2 mg/kg/day versus 0.2 mg/kg/day whose confidence interval excludes zero, while the recalculated interval for 0.5 mg/kg/day versus 0.2 mg/kg/day includes zero. Those calculations are not substitutes for the sponsor’s analysis.

The sponsor-reported ANCOVA result is an overall p-value across the three groups, p = 0.024. It does not specify which pairwise comparison accounts for the overall result. Because the registry flags a discrepancy between the sponsor analysis and the recalculations, the sponsor’s own analysis takes precedence. The record does not report adjusted pairwise estimates, a clinically important difference threshold, or the details needed to reconstruct the model.

The findings are specific to the recorded study: Korean pediatric outpatients with ADHD, three atomoxetine dose groups, and the Day 42 endpoint. The record does not establish results beyond that endpoint, comparisons against other treatments, or outcomes for populations not represented by the study. Statistical evidence of a difference is also distinct from clinical importance. The registry does not report a threshold that would allow the score changes to be classified by clinical importance.

The open-label design and investigator-administered, investigator-scored endpoint are relevant context for interpretation. The record does not establish whether assessment was influenced by knowledge of assignment. The data describe recorded outcomes and their statistical uncertainty; they do not, on their own, settle questions about durability, comparative benefit against treatments not studied, or the causes of individual changes.

17. Why This Trial Matters Statistically

This trial is statistically informative because it compares three randomized dose groups on a continuous symptom scale and reports both group-level changes and a sponsor-reported overall analysis. The group estimates allow readers to distinguish within-group change from between-group comparisons, an important distinction in trials without a placebo or untreated arm. A decrease within each arm is not itself evidence that one dose is better than another.

The difference between an overall test and pairwise comparisons is also central. The registry’s ANCOVA result, p = 0.024, concerns all three groups collectively. The unadjusted recalculations address two individual contrasts with the 0.2 mg/kg/day group. These analyses ask different questions and use different methods, so their results should not be conflated. The flagged discrepancy makes it especially important to keep the sponsor’s reported analysis separate from simple recalculations based on summary statistics.

The confidence intervals make uncertainty visible rather than relying on point estimates alone. In the 0.5 mg/kg/day comparison, the recalculated interval includes zero; in the 1.2 mg/kg/day comparison, it does not. Neither interval is the sponsor’s adjusted pairwise interval, and neither automatically answers whether a difference is clinically important. The registry provides no clinical importance threshold.

The record also illustrates how design and reporting shape statistical interpretation. Randomization supports comparisons among assigned groups, while open-label conduct means masking was absent. Missing-data procedures, model details, multiplicity handling, and original sample-size justification are not reported in the registry record. Those omissions do not erase the results, but they limit how fully the analyses can be reproduced or evaluated. The sample-size illustrations can help describe planning under stated assumptions, but they do not reveal the study’s original design rationale or the probability that this completed study would produce a particular result.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The primary outcome is measured at Day 42, so the posted estimate addresses change through that endpoint only. Longer-term results are not established by these figures. The registry lists secondary measurements at other times for some outcomes, but those results are not included in the primary endpoint summaries. The duration of treatment beyond the endpoint and any later follow-up are not reported in the registry record.

The trial is open-label, with no masking. The primary outcome is an investigator-administered and scored symptom measure, and the record does not indicate that the assessor was masked. In general, outcome ratings involving judgment can be vulnerable to influences beyond the measured symptoms; the record does not establish that such influence occurred or quantify its possible effect.

No placebo or non-atomoxetine comparator is listed, so the randomized evidence is about relative results among the three atomoxetine dose groups rather than a comparison with no atomoxetine. The registry does not report a clinically important difference threshold for the ADHD rating scale. Consequently, the statistical contrasts cannot be classified by clinical importance from this record alone.

The reported sponsor analysis is an overall ANCOVA test without posted adjusted pairwise estimates or confidence intervals. The model specification, covariates, missing-data handling for the primary outcome, and multiplicity procedures are not reported in the registry record. The pairwise estimates are unadjusted recalculations based on summary statistics and are approximate; the record flags a discrepancy, so the sponsor’s analysis takes precedence.

The serious adverse event figures are presented by dose bands, which do not exactly match the primary efficacy arm labels. The listed serious event terms are not linked to a particular dose band in the information reported. Deaths are not reported in the registry record. The registry also does not report the original sample-size justification, detailed randomization implementation, or interim-analysis plans, so those aspects cannot be assessed.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Atomoxetine to Treat Korean Children and Adolescents With Attention-Deficit/Hyperactivity Disorder (ADHD) — NCT00568685.
• Registry record last updated 2010-01-26; status last verified 2010-01.
• Results first posted 2009-12-14.
• 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.

22. Record Summary

NCT00568685 is a completed Phase 3b, randomized, open-label, parallel-group study sponsored by Eli Lilly and Company. It enrolled 153 Korean pediatric outpatients with ADHD and compared atomoxetine doses of 0.2 mg/kg/day, 0.5 mg/kg/day, and 1.2 mg/kg/day. The primary endpoint was change from baseline to Day 42 in the ADHD Rating Scale-IV-Parent Version, investigator administered and scored.

The posted least-squares mean changes were -9.55, -12.31, and -14.51 across the respective dose groups. The registry reports an overall ANCOVA p-value of 0.024. Approximate unadjusted recalculations give pairwise mean differences of -2.76 for 0.5 mg/kg/day versus 0.2 mg/kg/day and -4.96 for 1.2 mg/kg/day versus 0.2 mg/kg/day. Because the registry flags a discrepancy, the sponsor’s own analysis takes precedence. The record does not report the original sample-size justification, full model specification, primary endpoint missing-data method, or multiplicity procedure.