An Open-Label Trial of Aripiprazole in Autism Spectrum Disorders: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for An Open-Label Trial of Aripiprazole in Autism Spectrum Disorders (NCT00308074), a na phase 2 study. The primary outcome is Clinical Global Impressions-Improvement.

2. Clinical Question

The trial's stated hypothesis concerned possible reductions in anxiety and aggression associated with autism spectrum disorder and improvement in global outcome. That is the rationale recorded for the study, not a reported finding. The record also lists the Yale-Brown Obsessive Compulsive Scale and the Brief Psychiatric Rating Scale for Children as secondary outcomes. It does not report their numerical results or explain how those outcomes were analyzed.

Because participants and clinicians were not masked and there was no control group, observed change would not by itself distinguish a treatment-associated change from other influences, such as expectations, the passage of time, or changes in clinical assessment. The study can describe what was recorded in this group, but the registry information does not establish a comparative treatment effect.

3. Trial Design

This is a completed phase 2 trial sponsored by Cambridge Health Alliance. The registry lists a single-group model, no masking, and allocation as NA. Enrollment is 14. The registry gives a start of 2006-02 and primary completion of 2008-12. These design features describe a small, open-label study without a concurrent control arm, rather than a randomized comparison.

All participants were assigned to aripiprazole monotherapy as described in the intervention record. The starting dose could be 2.5 mg or 5.0 mg, based on clinical impression and the severity of aggression and agitation. The dose could be adjusted weekly by increments of no more than 5 mg, up to a maximum daily dose of 20 mg; the record states that the lowest effective dose would be used. It does not report the dose actually received by each participant, treatment duration by participant, or adherence.

The summary describes a 12-week study period. The primary outcome timeframe is described as baseline and endpoint using last observation carried forward (LOCF) at weeks 3, 5, 7, 9, 11, and 13. The record does not explain the relationship between the 12-week description and the listed week 13 assessment, so that timing cannot be reconciled from the registry information alone.

Four outcome measures are listed, while the registry records zero analyses. This does not establish whether analyses were conducted outside the registry; it means analysis details and numerical outcome results are not reported in the registry record. Eligibility criteria, recruitment procedures, participant flow, and the statistical analysis plan are also 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.

Before assignment: All individuals enrolled will have Diagnostic and Statistical Manual Fourth Edition (DSM-IV) diagnoses of Autism, Asperger's Disorder or Pervasive Developmental Disorder, not otherwise specified (PDD NOS) and will be enrolled in this 12-week open-label study

Overall Study

Reasons for not completing this period:

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 Clinical Global Impressions-Improvement, measured over Baseline, Endpoint using last observation carried forward (LOCF) at weeks 3,5,7,9,11 and 13..

The primary endpoint is Clinical Global Impressions-Improvement (CGI-I), a clinician-rated measure of how much a patient's illness has improved or worsened relative to baseline. Its seven categories range from 1, very much improved, through 4, no change, to 7, very much worse. Thus, lower ratings correspond to greater improvement, while higher ratings correspond to worsening.

The registry classifies the outcome as a mean measured in units on a scale, with standard deviation as the dispersion measure. A mean can offer a compact description of a group's average rating, while the standard deviation describes variation among ratings around that average. In general, the mean alone does not show how many participants fell into each CGI-I category; participants with different individual ratings can contribute to the same group mean.

A clinician-rated global measure can capture an overall judgment of change rather than focusing on a single symptom. That makes it relevant to a study whose stated interest includes global outcome. However, interpretation depends on the scale, assessment procedures, and actual recorded results. The registry gives the scale description and the planned assessment timeframe, but the numerical endpoint result, its standard deviation, and the number of participants contributing to it are not reported in the registry record. Consequently, the observed average change and its precision cannot be assessed from this record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes the primary outcome as a mean on a continuous measure and names standard deviation as its dispersion statistic. CGI-I itself is a seven-category rating scale. In general, treating a rating scale as a numeric outcome permits calculation of a group mean and standard deviation, but a mean compresses the distribution: it does not reveal the frequency of each response category or whether ratings were concentrated or spread across the scale. The registry record does not report the actual mean, standard deviation, or category counts.

What does a mean describe?

In general, a mean is an average across the observations included in an analysis. It is interpretable only alongside information about who contributed observations, how missing values were handled, and how the scale is being treated analytically. A standard deviation describes dispersion among observations; it is not the same as the uncertainty in the estimated mean. The standard deviation and the number of analyzed participants are both needed to understand variation and estimate precision. Those outcome details are not reported in the registry record.

How should the scale direction be read?

The CGI-I scale runs from 1, very much improved, to 7, very much worse, with 4 indicating no change. Lower ratings therefore indicate improvement and higher ratings indicate worsening. This direction matters when reading any mean or graphical display: a lower average rating would indicate greater improvement on the stated scale, not deterioration. The registry does not report a numerical average, so no such interpretation can be applied to an observed result from this record.

What does LOCF mean?

The endpoint timeframe specifies last observation carried forward (LOCF) for assessments at weeks 3, 5, 7, 9, 11, and 13. In general, LOCF substitutes a participant's last available measurement for a later measurement that is missing. It creates a value for an assessment that was not observed, based on an earlier observation. This approach relies on assumptions about what the unobserved later outcome would have been; carrying an earlier value forward does not establish that the participant's condition stayed unchanged. The registry does not report how many values were carried forward, why measurements were missing, or whether another missing-data analysis was conducted.

What does the study design permit?

This study is identified as single-group, with no masking and allocation listed as NA. In general, a single-group study can describe outcomes observed among its participants, but does not supply a concurrent estimate of outcomes without the intervention. Without random allocation to a comparator, differences from baseline cannot be interpreted as a direct randomized treatment contrast. Open-label assessment can also leave ratings susceptible to expectations or knowledge of treatment. These are design considerations, not evidence that a particular observed rating was biased.

What is the role of the planning illustration?

The record describes a sample-size planning illustration based on a fraction of an observed effect, with a two-sided significance level of 0.05, equal allocation, and a normal approximation. Such an illustration is a planning exercise: it asks how many participants per arm might be needed under stated assumptions to detect an assumed effect. It is not the trial's observed enrollment, and it is not observed power. Because this trial is single-group and the numerical observed effect is not reported in the registry record, the illustration cannot be treated as a result of the trial or as a justification for its design.

In general, power is a design property calculated for a specified effect, variability, analysis, and sample size. Calculating power after observing a result does not provide a substitute for an effect estimate and its confidence interval. A planning calculation based on an observed effect can also be sensitive to uncertainty in that effect. The registry does not report a sample-size justification for the trial, so the planning illustration should not be read as the original design rationale.

Why do uncertainty and assumptions matter?

Any estimate depends on the data included and the analysis choices. A confidence interval, when reported, helps describe statistical uncertainty around an estimate; it does not address all sources of bias or establish clinical importance. The registry record does not report an interval, a test statistic, or a p-value for the primary endpoint. It also does not report a model or adjusted analysis. No inferential conclusion can be reconstructed from the endpoint definition alone.

In general, over-dispersion means that observed variability is greater than a model assumes. If a model underestimates variability, its uncertainty estimates can be too small and inferential results can appear more precise than warranted. That concept is relevant when selecting models for data whose variability exceeds model assumptions, but the registry does not describe an analysis model or indicate that over-dispersion was assessed in this trial. It should not be inferred that this issue arose.

The record identifies four outcome measures and lists two secondary outcomes, but does not report a statistical analysis plan, outcome-specific numerical results, or methods for handling multiple outcomes. Therefore, the method used for the primary endpoint and the status of any additional analyses cannot be determined from the registry record.

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("Aripiprazole")
n   <- c(14)
m   <- c(1.8)
sp  <- c(1.6)
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  Aripiprazole  14  1.800  1.600  0.876  2.724

11. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures and identifies Clinical Global Impressions-Improvement as the primary outcome. It also names the Yale-Brown Obsessive Compulsive Scale and the Brief Psychiatric Rating Scale for Children as secondary outcomes. The other outcome measures are not described in the record. A statistical plan specifying which comparisons were confirmatory, how outcomes were prioritized, or whether multiplicity adjustments were used is not reported in the registry record.

In general, evaluating several outcomes or conducting several tests can increase the chance that at least one result appears noteworthy by chance. Approaches to multiplicity depend on the aims and analysis plan; they cannot be inferred from the number of listed outcomes alone. Because there are no numerical outcome analyses recorded, the effect of any multiplicity strategy cannot be assessed.

The primary timeframe names LOCF at weeks 3, 5, 7, 9, 11, and 13. This indicates an intended method for carrying forward the last available observation, but does not reveal how many participants had missing assessments or what their missingness patterns were. Reasons for missing data, sensitivity analyses, and other imputation methods are not reported in the registry record.

Enrollment is listed as 14, but the primary analysis population and the number contributing to each outcome are not reported. The record does not specify whether all enrolled participants were analyzed, whether analyses followed an intention-to-treat principle, or how withdrawals were handled. The zero analyses recorded should be understood as an absence of analysis details in the registry record, not as proof that no analysis was performed elsewhere.

12. 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.

Most frequent non-serious adverse event terms

The registry lists 14 participants at risk for serious adverse events and zero participants affected in the aripiprazole group. It lists no serious adverse-event terms. For other adverse events, 14 participants were at risk and one was affected. The deaths entry is null, so deaths are not reported in the registry record.

These counts describe the registry's recorded categories. They do not identify the other adverse event or indicate its timing, severity, duration, or relationship to aripiprazole. The record does not provide participant-level details, exposure time, or a fuller breakdown of event counts. A count of participants affected is not necessarily the same as a count of events if an individual can experience more than one event; the record does not explain its counting procedures.

In general, absence of a recorded serious event in a small study does not establish that serious events cannot occur. Safety counts from a study without a comparator also do not show whether events were more or less frequent than they would have been under another treatment or no treatment. The figures should therefore be read as the recorded counts for this group, not as a comparative risk estimate or a complete safety profile.

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

The record establishes that a completed phase 2, open-label, single-group trial evaluated aripiprazole monotherapy in a study described as involving children and adolescents with autism spectrum disorder. It specifies a clinician-rated primary outcome, the CGI-I scale, and an assessment timeframe. It also lists two secondary measures and provides limited adverse-event counts. These details define the study question and some of the information recorded.

What the record does not establish is the numerical result for the primary outcome. Without the reported mean, standard deviation, analysis population, or uncertainty interval, the observed magnitude and precision of CGI-I ratings cannot be characterized. The stated hypothesis about anxiety, aggression, and global outcome is not a result. The record likewise does not provide numerical secondary outcome results from which changes on those measures could be described.

Even if a before-and-after change were available, the single-group and open-label design would not provide a randomized comparison. In general, change over time can reflect multiple influences, and a study without a control arm cannot separate those influences from the intervention using a direct between-group contrast. This is a limit on causal interpretation, not a claim that any specific alternative explanation occurred.

The safety record reports zero participants affected by serious adverse events among 14 at risk and one affected by other adverse events among 14 at risk. It does not report the event description or deaths. These counts are too limited to establish comparative safety or to describe a complete range of possible events.

Accordingly, the record is informative about trial structure, the intended clinical question, and the recorded safety counts. It does not support a quantitative estimate of the primary outcome, a conclusion about the size of any change, or a comparison with another intervention. Those limits reflect missing result and design information in the registry record.

14. Why This Trial Matters Statistically

This trial is statistically notable for the gap between a clearly stated clinical question and the amount of outcome analysis recorded in the registry. The record names a primary measure, its scale direction, its summary statistic, and its assessment timeframe. But it lists zero analyses and does not report a numerical primary result. Without results, readers cannot assess the observed mean, variation, precision, or the pattern of individual responses.

The single-group, open-label design also illustrates the distinction between observing change and estimating a comparative effect. A group can be followed over time and measured on a defined scale, but without a concurrent comparator the design does not show what would have happened to a similar group under another condition. In general, this distinction is central to interpreting clinical studies: a measured outcome and a causal treatment comparison are not interchangeable.

The LOCF specification is another useful statistical detail to recognize. It makes explicit that an earlier measurement could stand in for a later missing one, while leaving unanswered how often that occurred and how robust results might be to other assumptions. Reporting missingness and sensitivity analyses is important because an analysis method cannot reveal the unobserved outcomes themselves.

Finally, planning illustrations should be distinguished from trial findings. A calculation based on a fraction of an observed effect is not evidence that the study had a particular power, and is not proof that the original enrollment target was justified by that calculation. The numerical outcome needed to evaluate such an illustration is not reported in the registry record. Together, these points show why statistical interpretation depends not just on the name of an endpoint, but on complete results, transparent analysis methods, and a design suited to the intended comparison.

15. 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 trial enrolled a listed 14 participants and had no comparator group. As a result, the record does not offer a direct comparative estimate. The open-label design, with masking recorded as none, also means that clinician ratings were made without masking to treatment. In general, this may affect subjective assessments, although the registry information does not establish whether or how much it influenced ratings in this study.

The primary endpoint result is not reported in the registry record. The number analyzed, mean CGI-I, standard deviation, category distribution, and uncertainty estimate are also absent. Consequently, the actual outcome and its precision cannot be assessed. The registry's zero analyses entry further leaves statistical methods and analysis populations unspecified; it does not show whether analyses were conducted outside the record.

The stated 12-week study period and the primary timeframe listing endpoint assessments through week 13 are not reconciled in the registry. The record also does not give participant-level dose exposure, adherence, withdrawals, missingness reasons, eligibility criteria, or demographics beyond its description of children and adolescents in the summary. These omissions restrict assessment of who contributed data and how closely the analyzed group represents the enrolled group.

The record lists two secondary outcome measures but does not report their results or the handling of multiple outcomes. It does not report a statistical analysis plan, sample-size justification, or details of any sensitivity analyses. The safety information gives limited counts without event descriptions, and deaths are not reported in the registry record. These trial-specific gaps constrain interpretation of both outcomes and safety.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: An Open-Label Trial of Aripiprazole in Autism Spectrum Disorders — NCT00308074.
• Registry record last updated 2017-02-15; status last verified 2016-12.
• Results first posted 2017-01-02.
• 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.

19. Record Summary

NCT00308074 is listed as a completed phase 2 study sponsored by Cambridge Health Alliance. It is described as an open-label, single-group trial of aripiprazole monotherapy, with enrollment of 14 and no comparator or masking. The summary states an aim of evaluating efficacy, safety, and tolerability over 12 weeks in children and adolescents with autism spectrum disorder.

The primary outcome is clinician-rated CGI-I, described as a seven-category scale comparing illness with baseline. The registry specifies a mean and standard deviation and lists a timeframe using LOCF, with assessments at weeks 3, 5, 7, 9, 11, and 13. Numerical primary and secondary outcome results are not reported in the registry record.

For safety, the record lists zero participants affected by serious adverse events among 14 at risk and one affected by other adverse events among 14 at risk. Deaths are not reported in the registry record. Analysis populations, missing-data counts, and detailed statistical methods are not reported.