Evaluation of an Intervention for Improving Community-based Pediatric Attention-Deficit Hyperactivity Disorder (ADHD) Care: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of an Intervention for Improving Community-based Pediatric Attention-Deficit Hyperactivity Disorder (ADHD) Care (NCT01143701), a randomized phase 4 study. The primary outcome is Parent-rated ADHD Symptoms.

2. Clinical Question

The clinical question concerns both a care-delivery intervention and a patient-reported symptom outcome. The intervention was designed to influence how pediatric practices organize and deliver ADHD care; the primary endpoint then assessed symptoms as rated by parents. These are distinct levels of the study: the practices were the units assigned to intervention conditions, while the symptom measure came from individual participants and their parents.

The registry record describes the study as completed and identifies ADHD as the condition under study. It does not report additional eligibility details or the precise characteristics of the enrolled children. Accordingly, the findings address the recorded comparison and endpoint, but the record alone does not establish how broadly the result applies to other settings or populations.

3. Trial Design

The registry record identifies a randomized, single-masked, parallel-group phase 4 trial, with 577 participants enrolled. It describes the design as cluster randomized: pediatric practices, rather than individual patients, were assigned to ADHD Collaborative or typical ADHD care. The record states that 32 pediatric practices were to be randomly assigned. This structure matters because patients cared for within the same practice may share providers and clinical routines, so their outcome measurements cannot automatically be treated as independent observations.

The intervention arm was the ADHD Collaborative. Its components included academic detailing, quality improvement methods, and tools intended to support systematic use of the American Academy of Pediatrics recommendations. Physicians in the comparator group were to provide typical ADHD care. The record does not specify the allocation ratio between arms, nor does it identify which role or individuals were masked under the single-masking designation.

The study start is recorded as 2010-05, and primary completion as 2015-05. The status is completed. The registry record classifies the intervention as phase 4 and reports two outcome measures and two analyses, but does not supply further detail about the second outcome or analysis in the registry record for this interpretation.

Because assignment occurred by practice while the primary outcome was measured at the participant level, the analysis needed to account for the nested data structure. The registry describes hierarchical modeling that accounted for patients nested within providers and providers nested within practices. A simple analysis that treats all participant observations as independent would not reflect that design. The registry specifically notes that a simple individual-level re-analysis would be misleading; consequently, the reported sponsor analysis is the relevant estimate to interpret.

Details of the randomization procedure, masking procedures, follow-up completeness, and operational conduct are not reported in the registry record. The record also does not explain whether participants or parents were aware of their practice’s assignment. Those omissions limit assessment of how assignment and measurement were implemented, beyond the design labels and analysis description that are recorded.

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: Hierarchical Modeling accounted for clustering of patients nested within providers and providers nested within practices.

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 Parent-rated ADHD Symptoms, measured over 12 months.

The primary endpoint was parent-rated ADHD symptoms at 12 months. It used the Total Symptom Score on the Parent-Rated Vanderbilt ADHD Rating Scale, described as a score on a scale ranging from 0 to 54. Higher scores represent more severe ADHD symptom presentation. The outcome is continuous, and the registry identifies the target summary as a mean with standard deviation.

General: a symptom scale turns a set of reported behaviors into a structured measure that can be compared across study groups. A parent rating captures observations made in the child’s home or day-to-day life, while a consistent instrument provides a common scoring framework. A continuous score also allows an analysis to estimate a difference in average final values rather than reducing the outcome to a simple yes-or-no category.

The choice of this endpoint fits a study intended to evaluate whether a practice-level care intervention is associated with differences in children’s ADHD symptom presentation. It is nevertheless a specific measure, from a specific rater, at a specified follow-up time. The result does not by itself describe every aspect of ADHD care, functioning, or family experience. The registry record does not report group-specific endpoint means or standard deviations in the registry record for this interpretation; it reports a mean difference in final values.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary result is described as a mixed models analysis of mean differences in final values. The registry also states that hierarchical modeling accounted for patients nested within providers and providers nested within practices. These descriptions are consistent with the need to recognize the cluster-randomized structure, in which the practice is the assignment unit and the symptom outcome is measured for individual participants.

Why account for clustering?

General: observations from people treated by the same practice may be more alike than observations from unrelated practices. Providers may share training, workflows, or local procedures, and patients within a practice encounter some of the same care environment. If an analysis assumes that every participant contributes independent information, it may understate uncertainty when outcomes are correlated within clusters. A cluster-aware model represents some of this shared structure instead of treating each score as entirely independent.

Clustering affects the precision of an estimate, not just its presentation. The amount of information in a cluster trial depends on both the number of people measured and how much their outcomes resemble one another within clusters. The registry record does not report the correlation estimates, model details, or variance components, so their contributions to the reported interval cannot be examined from the recorded information.

What does a mixed model do?

General: mixed models can represent both average relationships of interest and variation associated with groups or clusters. In a nested study, the model may account for observations grouped within providers and practices. This helps align the uncertainty calculation with the way participants were assigned and cared for. The registry names mixed models for the reported comparison and says hierarchical modeling accounted for the nesting, but further specifications are not reported in the registry record. The exact fixed effects, random effects, covariance structure, and any covariate adjustments are therefore unknown.

The reported quantity is a mean difference in final values. That is a contrast between group averages at the endpoint, not a change-from-baseline estimate unless the analysis description says so. The registry names final values and does not state that baseline scores were included as covariates. It also does not report group-specific means or standard deviations in the registry record for this interpretation. The estimate should therefore be read as the sponsor’s reported contrast, not reconstructed from group summaries.

How should the confidence interval be read?

General: a confidence interval communicates uncertainty around an estimated parameter under the chosen model and its assumptions. Its endpoints describe plausible values for the mean-difference parameter under that framework. The 95% interval reported for this analysis is -5.89 to -1.19, around the estimate of -3.54. The endpoints do not describe the spread of individual symptom scores, and they do not mean that most participants had score changes within that range.

A confidence interval is tied to the analysis. If the clustering model, outcome definition, or assumptions changed, the interval could also change. For that reason, the registry’s warning against simple individual-level re-analysis is important: an alternative calculation that ignores the practice structure would not be a faithful check of the sponsor’s model-based uncertainty.

How should the p value be read?

The reported p value is <.05. General: a p value measures how unusual the observed data, or data more extreme, would be under a specified null hypothesis and the model assumptions. It is not an effect-size measure and does not describe the probability that the result is due to chance. The p value should be read alongside the estimated mean difference and confidence interval, rather than as a stand-alone verdict.

The registry does not state the exact null hypothesis formulation, whether tests were one- or two-sided, or whether the p value is adjusted for multiple comparisons. It reports two outcome measures and two analyses, but that count alone does not establish how multiplicity was handled. Any adjustment procedure, if used, is not reported in the registry record.

What can a planning illustration show?

General: a sample-size calculation can explore how many participants per arm might be needed for a future study under explicit assumptions about the target effect, significance level, allocation, and model. Such an illustration is planning information, not a description of what the trial observed and not a measure of the probability that the completed trial detected an effect. The registry’s stated planning assumptions describe a two-sided significance level of 0.05, equal allocation, a normal approximation, and an effect expressed as a fraction of the observed effect. No planning sample-size values are available in the registry record for this interpretation. The registry record does not report that these assumptions were the trial’s original design rationale.

More broadly, no protocol-level sample-size justification, interim-analysis plan, or detailed statistical analysis plan is reported in the registry record. The model’s detailed specification, missing-data handling, and analysis population are also not reported. Those omissions do not invalidate the sponsor’s recorded estimate, but they constrain independent assessment of how the estimate was produced and how sensitive it may be to modeling choices.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("ADHD Collaborative Intervention", "Typical ADHD Care")
n   <- c(235, 287)
m   <- c(20.5, 24.1)
sp  <- c(9, 9.4)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  ADHD Collaborative Intervention  235  20.500  9.000  19.343  21.657
CARM  Typical ADHD Care  287  24.100  9.400  23.008  25.192

11. Analyses Reported by the Sponsor

The registry reports the sponsor’s own analysis as a mixed models analysis, with a mean difference in final values of -3.54, a 95% confidence interval from -5.89 to -1.19, and p <.05. The primary outcome description says that hierarchical modeling accounted for patients nested within providers and providers nested within practices. These details matter because the trial was cluster randomized.

No simple unadjusted re-calculation is presented. The registry explicitly notes that a simple individual-level re-analysis would be misleading, so the sponsor’s cluster-aware analysis takes precedence. There is no separate recalculated estimate in the registry record for comparison, and the record does not identify a discrepancy between a sponsor result and a recalculation. Accordingly, there is no basis to characterize the sponsor result as agreeing or disagreeing with an independent arithmetic reconstruction.

General: recalculations from summary statistics can serve as checks when the relevant group summaries and analysis structure are available. In a cluster trial, however, an individual-level calculation that ignores dependence would not reproduce the model-based uncertainty. The reported estimate should therefore be interpreted as the sponsor’s model-based comparison, while recognizing that detailed model specifications are not reported in the registry record.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures and two analyses, but the registry entries available for this interpretation identify only the primary endpoint and its reported analysis. They do not describe the second outcome, its analysis, or whether either result was designated as part of a confirmatory testing sequence. An adjustment for multiple testing, such as a procedure that controls the chance of false-positive findings across outcomes, is not reported in the registry record. The reported p value should therefore be understood as presented, without assuming a multiplicity adjustment or assuming that none was used.

Details on missing data are also not reported in the registry record. The available information does not state how many participants had an observed 12-month score, how missing endpoint measurements were handled, or whether the model relied on a particular missing-data assumption. It also does not describe sensitivity analyses for missing outcomes. These details matter because estimates can be affected if outcome availability is related to participant characteristics or symptom scores.

The analysis population is not reported in the registry record. The record does not say whether the primary analysis followed an intention-to-treat principle, included all randomized practices or participants, or used another defined population. Although enrollment is recorded as 577, that number alone does not show how many participants contributed data to the primary analysis.

Likewise, the registry does not provide enough detail to assess adherence to assigned practice conditions, crossovers, or exclusions after randomization. These are not evidence that a problem occurred; they are unreported aspects of the analysis and conduct. Interpretation should distinguish the reported model-based result from questions that cannot be resolved from the registry record, including multiplicity control, missingness, and the exact analysis population.

13. Safety Results

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

Other (non-serious) adverse events

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

The serious adverse event figures list 0 serious adverse events among 258 participants at risk in the ADHD Collaborative Intervention group and 0 among 319 participants at risk in the Typical ADHD Care group. The other adverse event figures also list 0 in each group, with the same respective numbers at risk. These are the recorded event counts and denominators; they do not establish that no adverse experiences of any kind occurred outside the categories or reporting process represented in the registry.

Deaths are recorded as null for both groups. A null value should not be read as a reported count of zero: death data are not stated as a numeric count in the record. The registry also lists no top serious-event terms.

General: safety figures are most informative when their definitions, collection period, severity criteria, and completeness are clear. Those details are not reported in the registry record in the registry record for this interpretation. The figures therefore describe the posted counts, but they do not support broader conclusions about comparative safety or the absence of unrecorded events.

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

The primary finding is a sponsor-reported mean difference in final parent-rated ADHD symptom scores of -3.54 at 12 months, with a 95% confidence interval from -5.89 to -1.19 and p <.05. The analysis used mixed models, and the registry says the hierarchical modeling accounted for patients nested within providers and providers nested within practices. The estimate is therefore a cluster-aware comparison as described in the record.

The numerical sign alone does not identify which named group had the lower score, because the subtraction order is not stated clearly enough in the analysis description. The endpoint’s direction is known: higher scores mean more severe symptom presentation. Linking that scoring direction to a particular arm requires knowing how the reported contrast was ordered, which is not reported in the registry record.

The result does not, by itself, establish the clinical importance of the difference. The record does not state a threshold for a meaningful change on this scale. Nor does a p value below .05 establish that the effect is large, durable, or relevant to every child or practice. The reported endpoint concerns parent-rated symptoms at 12 months; it does not establish effects on other clinical, educational, or family outcomes.

The study evaluates a practice-level intervention that included academic detailing, quality improvement methods, and tools to support guideline use. Its design and outcome do not isolate which component, if any, accounts for the reported contrast. The registry record also does not provide the detailed model specification, outcome completeness, analysis population, or multiplicity approach needed to answer several questions about robustness.

These qualifications do not negate the reported result. They define its scope: a model-based comparison of final symptom scores in a completed cluster-randomized trial, with uncertainty summarized by the reported confidence interval. Broader claims require information beyond the reported estimate and the registry entries.

15. Why This Trial Matters Statistically

This trial is statistically notable because the level of assignment and the level of outcome measurement differ. Pediatric practices were assigned to intervention conditions, while parent-rated symptom scores were measured for participants. That arrangement creates nested data: patients are associated with providers, and providers with practices. Accounting for those relationships is central to estimating uncertainty appropriately.

The registry identifies hierarchical modeling for the nested structure and a mixed models analysis for the reported mean difference. It also warns that a simple individual-level re-analysis would be misleading. This is a useful reminder that the number of participant records alone does not describe the amount of independent information in a cluster trial. Shared practice context can make observations statistically dependent, and analysis should reflect the design.

The endpoint also illustrates the difference between statistical evidence and clinical interpretation. The registry reports a mean difference, confidence interval, and p value for a parent-rated continuous symptom score. Those statistics describe an estimated contrast and its precision under the model, but they do not themselves define a clinically important change or explain how the score maps to a participant’s experience.

Finally, this record shows why transparency about analysis details matters. The nested structure and broad method are reported, while specific model choices, missing-data methods, multiplicity handling, and analysis populations are not. Those details influence how fully readers can evaluate the result and its limitations. The reported evidence is most responsibly interpreted by preserving both what the cluster-aware analysis estimates and what the registry record leaves unresolved.

16. 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.

Several limitations specific to this trial follow from the information recorded. The design is cluster randomized, but the registry does not report the exact randomization procedure, the allocation ratio, or practice-level characteristics. It identifies 32 pediatric practices as the planned randomized clusters, yet further information on cluster participation and retention is not reported in the registry record.

The primary analysis is described as mixed models, and the primary outcome description says hierarchical modeling accounted for patients nested within providers and providers nested within practices. The registry does not report the detailed model specification, such as its covariates, variance structure, or other modeling choices. Nor does it report group-specific means or standard deviations alongside the contrast in the registry record for this interpretation.

The subtraction order for the mean difference is not made clear enough to assign its negative direction to a named group. The outcome’s score direction is reported, but the contrast convention is not. This limits a direct statement about which arm had the lower final mean based solely on the result as recorded.

Follow-up completeness, missing-data handling, analysis-population definitions, and sensitivity analyses are not reported in the registry record. The record lists two outcomes and two analyses but does not describe any multiplicity strategy. It also does not provide the detailed masking procedures or identify who was masked. These are reporting gaps, not evidence that the methods were inadequate.

Finally, the endpoint is a parent-rated symptom measure at 12 months. The registry entries do not report a threshold for clinical importance, results for other outcomes, or longer-term findings. Accordingly, the recorded estimate should be interpreted narrowly as a cluster-aware comparison of the specified endpoint and timeframe.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Evaluation of an Intervention for Improving Community-based Pediatric Attention-Deficit Hyperactivity Disorder (ADHD) Care — NCT01143701.
• Registry record last updated 2016-11-08; status last verified 2016-09.
• Results first posted 2016-11-08.
• 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.

20. Record Summary

The registry record describes a completed, randomized, single-masked, parallel, phase 4 cluster trial of the ADHD Collaborative versus typical ADHD care. It reports enrollment of 577 participants and states that 32 pediatric practices were to be randomized. The study start is recorded as 2010-05 and primary completion as 2015-05.

The primary endpoint was the parent-rated Vanderbilt ADHD Rating Scale Total Symptom Score at 12 months, ranging from 0 to 54, with higher scores indicating more severe symptoms. The sponsor-reported mixed models analysis estimated a mean difference in final values of -3.54, with a 95% confidence interval from -5.89 to -1.19 and p <.05. Hierarchical modeling accounted for patients nested within providers and providers nested within practices.

The record lists zero serious and other adverse events in each arm, while death counts are null. Further details on the contrast ordering, model specification, missing data, analysis populations, and multiplicity are not reported in the registry record.