Treatment of ADHD With Atomoxetine in Children & Adolescents With ADHD & Comorbid Dyslexia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of ADHD With Atomoxetine in Children & Adolescents With ADHD & Comorbid Dyslexia (NCT00607919), a randomized phase 4 study. The primary outcome is Change From Baseline in Attention-Deficit/Hyperactivity Disorder Rating Scale (ADHDRS) Total Score - Parent Version at Week 16 Endpoint.

2. Clinical Question

The endpoint focused on ADHD symptoms rather than reading performance. The registry also lists reading, phonological processing, working memory, and participation measures among the secondary outcomes, but the primary question was the difference in change on the parent-version ADHD Rating Scale total score.

The study was registered as a Phase 4 randomized, double-blind, parallel-group trial. Its enrollment was 209. The registry record states that the study evaluated atomoxetine for ADHD symptoms in children and adolescents with ADHD and comorbid reading disability; it does not establish a separate primary question about dyslexia symptoms.

3. Trial Design

The registry describes a randomized, double-blind, parallel-group, Phase 4 study sponsored by Eli Lilly and Company. The study is listed as completed. Its start is recorded as 2008-03 and its primary completion as 2010-11. Enrollment was 209.

Participants were assigned to atomoxetine or placebo. Atomoxetine was described as an oral once-daily morning treatment at 1.0 to 1.4 mg/kg/day. Placebo was also taken orally once daily in the morning and packaged in the same way as the experimental drug. The stated masking was double, and the parallel model indicates that participants were allocated to one study group rather than crossing between the blinded groups during the main comparison.

The primary endpoint was assessed at the Week 16 endpoint. The registry describes an optional 16-week open-label extension for eligible participants who completed the double-blind study period. The arm descriptions do not give a complete account of the extension, and extension results are not reported in the registry record addressed by the primary comparison.

The primary endpoint analysis population was defined as all randomized participants with a baseline result and at least one post-baseline result. The registry record does not report the original design assumptions, including the planned effect size, power target, or sample-size justification. Those design details should not be inferred from the planning illustrations based on the observed estimate.

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.

Before assignment: The study comprised a 16-week placebo-controlled, double-blind acute phase and followed by an optional 16-week open-label in which all participants were treated with Atomoxetine.

Acute Phase

Reasons for not completing this period:

Open-Label Phase

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants with a baseline and at least one post-baseline result.

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 in Attention-Deficit/Hyperactivity Disorder Rating Scale (ADHDRS) Total Score - Parent Version at Week 16 Endpoint, measured over Baseline, 16 weeks.

The primary endpoint was change from baseline in the total score on the parent version of the Attention-Deficit/Hyperactivity Disorder Rating Scale (ADHDRS) at the Week 16 endpoint. The scale measures the 18 symptoms included in the DSM-IV-TR diagnosis of ADHD. Each item ranges from 0, meaning none, never, or rarely, to 3, meaning severe or very often. The total score ranges from 0 to 54.

Change from baseline expresses how the recorded score at follow-up differs from the participant’s baseline score. Because higher scores indicate more frequent or severe symptoms on the described scale, a negative change represents a decrease in the score. The posted group estimates are least-squares means rather than simple raw group averages.

This endpoint was used to quantify the trial’s primary symptom question with a standardized parent-rated measure. Parent ratings provide one perspective on symptoms; the registry separately lists a teacher-version rating scale among the secondary outcomes. The primary endpoint does not directly measure reading performance, even though the participants had comorbid dyslexia and the registry lists reading-related secondary measures.

7. Results Posted in the Registry

• ADHD+ Dyslexia (D): Atomoxetine: mean -20.01 versus -12.27 on control; difference -7.74 (95% CI -11.70 to -3.78), standardized effect d = -0.76; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary result uses a least-squares mean change from baseline estimated with a restricted maximum likelihood-based mixed model repeated-measures analysis. This is a model-based approach for repeated measurements, rather than a report of a simple unadjusted mean change in each group. The registry identifies the analysis population as all randomized participants with a baseline and at least one post-baseline result.

What does a least-squares mean represent?

In general, a least-squares mean is an estimated group mean from a statistical model. It reflects the model’s specified terms, rather than only the raw distribution of scores in a group. In this trial, the model was described as including treatment, investigative site, baseline, visit, treatment-by-visit interaction, and baseline-by-visit interaction. The estimate therefore represents group change adjusted within that specified model. The registry does not provide enough information to reconstruct each participant’s contribution or the fitted model in full.

Why use a repeated-measures model?

In general, repeated-measures methods analyze outcomes recorded across visits while accounting for the fact that measurements from the same participant are related. A model can use the pattern of available measurements rather than treating every visit as unrelated. The registry names a mixed model repeated-measures analysis and identifies visit and interactions with visit as model terms. It does not report the visit schedule beyond the primary endpoint timeframe of baseline and 16 weeks in the endpoint description, nor does the registry record state the covariance structure or other implementation details.

Restricted maximum likelihood is a method for estimating model variance components. In general, it is used in mixed models to estimate variability while accounting for the model’s fixed effects. The registry names restricted maximum likelihood as the estimation approach but does not report the software, estimation diagnostics, or detailed model output beyond the posted summaries.

How should the comparison be read?

The reported group results are -20.01 for atomoxetine and -12.27 for placebo, each a least-squares mean change. Their estimated difference is -7.74 units, with a 95% confidence interval from -11.7 to -3.776. The direction follows the stated comparison: atomoxetine relative to placebo. A negative difference indicates a greater decrease in the atomoxetine group’s score. The interval describes uncertainty about the estimated difference under the analysis; it is not a range containing every participant’s change.

The reported p-value is from the registry’s mixed-model analysis. A separate unadjusted recalculation from posted summary statistics is a simpler comparison and does not reproduce all model adjustments. Such a recalculation is useful as a check on the general scale and direction of a result, but it is not a replacement for the sponsor’s model-based analysis. The registry record reports p < 0.001 for the mixed-model analysis, and the unadjusted recalculation also yields p < 0.001.

What does the standard error tell the reader?

In general, a standard error describes uncertainty in an estimate, not the spread of individual participant outcomes. The posted standard errors for the group least-squares mean changes are 1.45 and 1.41. They reflect the precision of those estimated group changes under the reported analysis. They should not be interpreted as standard deviations of participant scores.

The primary analysis population requires baseline and at least one post-baseline result. Mixed models can, in general, make use of incomplete repeated measurements under assumptions about the data and model. The registry describes the analysis method, but the missing-data assumptions, sensitivity analyses, and detailed handling of missing observations are not reported in the registry record. It is therefore not possible to assess from the reported description whether alternative missing-data approaches would produce similar estimates.

10. Confidence Intervals and Precision

The estimated mean difference in Week 16 change was -7.74 units on the parent-rated ADHDRS total score. Its reported 95% confidence interval extends from -11.7 to -3.776. The interval expresses uncertainty in the estimated group difference under the reported analysis. It is compatible with a range of negative differences, rather than identifying one exact value as the population effect.

Because both interval limits are below zero, the interval does not include no difference for this comparison. This is consistent with the mixed-model p-value of p < 0.001. Statistical compatibility with a nonzero difference does not establish that the difference is clinically important; the registry record does not give a threshold for interpreting the size of a score change as clinically meaningful.

Precision also depends on the analysis and its assumptions. The posted group changes have standard errors of 1.45 for atomoxetine and 1.41 for placebo. The confidence interval is for the estimated contrast, not for each group’s individual scores. It should not be read as a prediction interval for the outcomes of future participants or as a description of how much scores varied between participants.

The estimate is reported to two decimal places and the upper interval limit to three decimal places. The displayed precision does not remove uncertainty about the model, missing observations, or the applicability of the result outside this study population. The registry does not report further precision analyses or sensitivity intervals for the primary result.

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 illustrations express required sample sizes as participants per arm, not as total events. They are calculations based on the observed effect and are not the trial’s original sample-size justification. The registry record’s original design assumptions are not reported.

The stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The scenarios target 100%, 75%, or 50% of the observed effect. For planning for 100% of the observed effect, the illustration gives 28 participants per arm for 80% power and 37 participants per arm for 90% power. For planning for 75% of the observed effect, the corresponding figures are 49 and 65 participants per arm. For planning for 50% of the observed effect, they are 109 and 146 participants per arm.

The 80% power column describes the number of participants per arm in the assumed design that would give an 80% chance of detecting the specified effect under the calculation’s assumptions. The 90% power column describes the analogous target for a 90% chance. These are planning targets, not guarantees that a future study will detect an effect.

Smaller effects require more participants to detect under otherwise unchanged assumptions. The illustrations show that pattern by increasing the planned number per arm as the target effect is reduced from the full observed effect to a smaller fraction. This is not observed power; it is a planning illustration based on the observed effect, and it does not describe the probability that the completed trial’s result is true.

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 = -7.740; pooled standard deviation σ = 10.168.
2. n = 2 × (1.96 + 0.842)² × 10.168² / -7.740² = 27.1, rounded up to 28 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 <- -7.74                      # observed difference in means
sigma <- 10.1680561395267                     # 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(607919)
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 = 28 (80% power), 37 (90% power)
 75% of observed effect: n per arm = 49 (80% power), 65 (90% power)
 50% of observed effect: n per arm = 109 (80% power), 146 (90% power)
power.t.test: 28.1 
Simulated power at n = 28 per arm: 0.795 (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 primary analysis as a mixed-model analysis, with p < 0.001. Its method is described as a restricted maximum likelihood-based mixed model repeated-measures analysis that includes treatment, investigative site, baseline, visit, and the stated interactions. The reported primary result is therefore a model-based comparison.

The recalculated comparison is an unadjusted calculation from the posted summary statistics. It does not reproduce all terms in the mixed model. It gives a mean difference of -7.74, and its p-value is p < 0.001. The sponsor’s analysis also reports p < 0.001. The registry’s discrepancy indicator is false, so the posted summaries and the sponsor’s reported test are not flagged as discrepant.

The recalculation is best understood as a check based on group-level summary values, not as an alternative analysis with the same modeling detail. The sponsor’s mixed-model result remains the registry’s reported analysis. Neither p-value alone conveys the clinical importance of the score difference or resolves limitations that are not addressed in the registry record.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 21 outcome measures and reports 20 secondary outcomes posted. The secondary outcomes include parent- and teacher-version ADHD Rating Scale measures and reading, phonological processing, working memory, and life participation measures. The primary endpoint is the parent-version total score at the Week 16 endpoint.

The record reports one analysis and gives the primary mixed-model p-value, but a multiplicity adjustment method or testing hierarchy is not reported in the registry record. When a study examines multiple outcomes or comparisons, the chance of at least one small p-value can increase. In general, a prespecified testing strategy or adjustment may address that issue; no such strategy should be assumed. The primary p-value should therefore be read as the reported test for this endpoint, without claiming that it was adjusted for all outcomes.

The primary analysis population is stated as all randomized participants with a baseline and at least one post-baseline result. This identifies who was included under the endpoint definition, but does not establish that every randomized participant contributed to the analysis. The registry record does not report a detailed participant flow, reasons for missing observations, or a comparison of included and excluded participants.

The analysis description names a mixed model repeated-measures approach, but the missing-data assumptions, sensitivity analyses, and imputation procedures are not reported in the registry record. In general, conclusions from a model using incomplete repeated measurements depend in part on its assumptions about why data are missing. Without the additional details, the robustness of the primary estimate to other missing-data approaches cannot be assessed from this record.

The registry record also does not report the original sample-size justification, interim analysis plans, or a complete statistical analysis plan. These omissions limit assessment of how the endpoint, analysis population, multiplicity approach, and missing-data procedures were set before results were known.

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: wound infection staphylococcal (0.8%).

The registry lists serious adverse events for the Atomoxetine/Atomoxetine group as 1 affected participant among 120 at risk, and for the Placebo/Atomoxetine group as 0 among 89 at risk. The group labels are reproduced as recorded; the registry summary does not provide further detail about timing or the relationship of these groups to periods of treatment.

For other adverse events, the record lists 108 affected participants among 120 at risk in the Atomoxetine/Atomoxetine group and 71 among 89 at risk in the Placebo/Atomoxetine group. These are counts of participants with the listed category, not evidence by themselves that an event was caused by a study drug. The registry identifies wound infection staphylococcal at 0.8% as the top serious adverse event term.

These figures should be read as the registry’s reported safety summaries, not as a complete account of event severity, duration, or attribution. Deaths are not reported in the registry record. The posted counts alone do not establish a comparative safety conclusion.

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

The primary result describes a difference in change on a parent-rated ADHD symptom score over the stated baseline-to-Week 16 period. Both groups had negative least-squares mean changes, and the estimated difference was -7.74 units in the direction of a larger decrease for atomoxetine than placebo. The 95% confidence interval was -11.7 to -3.776, and the registry reports p < 0.001 for the mixed-model analysis.

This is a statistical result for the specified scale, time point, and primary analysis population. It is not a direct measurement of reading performance, even though participants had comorbid dyslexia and reading-related outcomes were listed as secondary measures. Nor does the primary estimate describe the teacher-rated ADHD score or results during the optional open-label extension.

The scale records symptom ratings and its total ranges from 0 to 54. The registry does not report a threshold for judging whether a particular change is important to participants or families. Statistical evidence for a difference should therefore be distinguished from evidence about its practical or clinical importance. The confidence interval describes uncertainty in the estimated average contrast, not the range of responses among individual participants.

The randomized, double-blind design provides a structured comparison between the assigned groups, but the result should not be extended beyond what the record documents. The reported analysis depends on the model and the defined analysis population. Details about missing-data sensitivity analyses and adjustment for multiple outcomes are not reported in the registry record. The result also does not establish what would occur in populations or settings not described by the study.

17. Why This Trial Matters Statistically

This trial offers a useful statistical example of how a continuous, patient-related symptom score can be analyzed in a randomized parallel-group study. The primary endpoint is a change score, and the reported group estimates are least-squares means from a repeated-measures model rather than raw averages. That distinction matters when interpreting a group comparison because the model incorporates specified factors and interactions.

The result also illustrates why an estimate and confidence interval add information beyond a p-value. The estimated difference is expressed in units of the ADHDRS total score, while the interval communicates uncertainty about that estimate. The reported p-value indicates statistical evidence under the model, but does not determine whether the size of the difference is meaningful to participants.

There is a further methodological distinction between the sponsor’s model-based analysis and an unadjusted recalculation from posted summary statistics. The recalculation can offer a basic check on the reported direction and scale, but it cannot reproduce every feature of the mixed model. The registry’s record gives both a primary analysis and enough summary information to see why those approaches should not be treated as interchangeable.

Finally, the study makes clear that a significant primary result does not answer every question raised by a clinical trial. Interpretation also depends on the analysis population, missing-data handling, multiplicity, clinical meaning of the scale, and the scope of the reported outcomes. Several of those details are not reported in the registry record, so they remain part of the uncertainty surrounding a full appraisal.

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 result addresses a parent-rated ADHD symptom score at the Week 16 endpoint. It does not, by itself, provide evidence about the listed secondary reading measures, teacher ratings, or the optional extension. Although the registry lists 20 secondary outcomes posted, their detailed results are not reported in the registry record addressed by this primary comparison.

The registry reports least-squares means and standard errors for the groups, and a confidence interval for their difference. It does not provide a full model output, model diagnostics, or the covariance structure used for repeated measurements. The primary analysis population is defined, but detailed participant flow and reasons for missing outcomes are not reported in the registry record. This limits assessment of how incomplete data might affect the estimate.

The registry record does not report a multiplicity adjustment or testing hierarchy for the outcome measures. It also does not report the trial’s original sample-size justification or design assumptions. Planning illustrations based on the observed effect do not substitute for that information.

The endpoint description gives the score range and item anchors, but a threshold for clinically meaningful change is not reported in the registry record. The available summary therefore supports interpretation of the estimated score difference and its statistical precision, but not a conclusion about the practical importance of that difference. The optional open-label extension is described as available to eligible completers, but its results and detailed conduct are not reported in the primary result information.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Treatment of ADHD With Atomoxetine in Children & Adolescents With ADHD & Comorbid Dyslexia — NCT00607919.
• Registry record last updated 2011-12-15; status last verified 2011-11.
• Results first posted 2011-12-15.
• 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

NCT00607919 is a completed Phase 4 randomized, double-blind, parallel-group study of atomoxetine versus placebo in children and adolescents with ADHD and comorbid dyslexia. Enrollment was 209. The registry records a start of 2008-03 and primary completion of 2010-11.

The primary endpoint was change from baseline to the Week 16 endpoint in the parent-version ADHDRS total score. The least-squares mean changes were -20.01 for atomoxetine and -12.27 for placebo, with a mean difference of -7.74 and a 95% confidence interval from -11.7 to -3.776. The sponsor-reported mixed-model analysis gives p < 0.001.

The registry lists secondary ADHD, reading, phonological processing, working memory, and participation outcomes, and describes an optional open-label extension. Important design and analysis details, including the original sample-size justification, multiplicity plan, and detailed missing-data procedures, are not reported in the registry record.