Effectiveness of Aripiprazole for Improving Side Effects of Clozapine in the Treatment of People With Schizophrenia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effectiveness of Aripiprazole for Improving Side Effects of Clozapine in the Treatment of People With Schizophrenia (NCT00345033), a randomized phase 4 study. The primary outcome is Change in Total Cholesterol.

2. Clinical Question

The registry describes a study of aripiprazole for clozapine-associated medical morbidity. Its brief summary says the study evaluates insulin resistance, blood fat levels, and weight gain in people diagnosed with schizophrenia. The posted primary result addressed total cholesterol, one blood-fat measure. Other outcome results are not detailed in the record information summarized for this analysis.

3. Trial Design

The registry record identifies NCT00345033 as a Phase 4, randomized, triple-masked, parallel-group study. Massachusetts General Hospital is listed as the sponsor. The record reports enrollment of 38 and a completed status. The study started in 2005-03, with primary completion in 2009-10. These registry entries describe the trial’s broad design and status; they do not supply a full protocol or a detailed account of how each design feature was implemented.

The two interventions listed are aripiprazole and placebo. The arm descriptions specify aripiprazole 15mg/day for 8 weeks in one arm and placebo for 8 weeks in the other. The brief summary describes combination treatment with aripiprazole and clozapine, but the arm descriptions do not state a clozapine dose, treatment history, or additional medication details. Those details are not reported in the registry record.

The primary outcome is change in total cholesterol, measured at baseline and Week 8. The outcome description says the comparison is between the aripiprazole and placebo groups. It reports an intent-to-treat analysis population as participants who completed the study, N = 30. That wording gives the registry’s stated population but does not explain how participants who did not complete the study were handled in the analysis.

Random assignment is intended to support a fair comparison between groups, while masking is intended to limit knowledge of assigned treatment from the people covered by that procedure. The record labels the study randomized and triple-masked, but does not specify which parties were masked, how randomization was generated, whether allocation was concealed, or how masking was maintained. Further details of the protocol, including the original design assumptions and any interim analyses, are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: After providing written informed consent, subjects underwent a diagnostic evaluation by a research psychiatrist using the Structured Clinical Interview for DSM-IV (SCID). All subjects were screened and enrolled based on eligibility criteria. Baseline study assessments were completed prior to intervention.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The number of participants for analysis (intent to treat) were those that completed the study (N = 30).

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 in Total Cholesterol, measured over Measured at Baseline and Week 8.

The primary endpoint is change in total cholesterol, measured at baseline and Week 8, with results expressed in mg/dL. Total cholesterol is a continuous measure, so its group results are reported as means with standard deviations. The endpoint asks whether the average change over the measurement period differed between participants assigned to aripiprazole and those assigned to placebo.

Measuring change uses each participant’s baseline measurement as a reference for interpreting the Week 8 value. A comparison of group changes can describe whether the groups moved in different directions or by different average amounts during follow-up. The registry’s group results report a mean change of -15.3 in the aripiprazole group and 5.6 in the placebo group. The unadjusted recalculation reports a mean difference of -20.9 for aripiprazole relative to placebo.

As a general statistical point, a continuous endpoint can retain information about the range of measured outcomes rather than reducing them to categories. Total cholesterol is relevant to the study’s stated focus on blood fat levels and medical morbidity associated with clozapine. However, the registry result for this endpoint alone does not summarize every outcome mentioned in the study summary, and it does not establish what happened to insulin resistance, weight, or other health measures. Those outcomes and their results are not reported in the registry record information addressed.

7. Results Posted in the Registry

• Aripiprazole: mean -15.30 versus 5.60 on control; difference -20.90 (95% CI -45.05 to 3.25), standardized effect d = -0.62; z-test p = 0.090.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous, and the registry reports each group’s mean change and standard deviation. A mean represents the group’s average change. A standard deviation describes how spread out participant changes are around that average. It is not the uncertainty interval for the mean difference: the standard deviations of 33.3 and 34.0 describe participant-level variation, while the confidence interval for the between-group contrast describes uncertainty in the estimated difference between group means.

The recalculated comparison is an unadjusted difference in means. It compares the reported mean change of -15.3 in the aripiprazole group with the reported mean change of 5.6 in the placebo group, with the contrast defined as aripiprazole minus placebo. The reported estimate is -20.9, and its confidence interval is -45.05 to 3.254. “Unadjusted” means this recalculation does not account for other variables through a regression model. It is a straightforward comparison of the posted group summaries, not a replacement for the sponsor’s analysis.

A confidence interval gives a range of values compatible with the observed data under the method’s assumptions. In repeated use of the same method, intervals would have a stated coverage property; an individual interval is not a probability statement that the true value lies inside it. Here the interval includes zero, so it does not rule out no difference in average change. Its width also signals that the exact size of the group difference is imprecisely estimated. The interval should be interpreted alongside the point estimate rather than as a separate verdict.

The registry’s own reported analysis is ANCOVA, or analysis of covariance. As a general statistical concept, ANCOVA models an outcome while adjusting for one or more covariates. In a baseline-to-follow-up setting, such a model can account for baseline outcome values if they are included, but the registry record does not identify the covariates used in this ANCOVA. It also does not report the model estimate or confidence interval. The only numerical result for that analysis is p = 0.125, and the record categorizes its type as non-inferiority or equivalence. The margin and decision criteria for such an analysis are not reported in the registry record, so the p-value alone cannot establish an equivalence or non-inferiority conclusion.

A p-value is a measure of how surprising data at least as incompatible with a specified null hypothesis would be if that hypothesis and the model assumptions were true. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The unadjusted recalculation reports p = 0.09, whereas the sponsor’s ANCOVA reports p = 0.125. These values come from different analyses, so they need not match. The record does not provide enough model detail to reproduce the ANCOVA result.

Standard deviation and standard error address different sources of variation. In general, a standard error expresses uncertainty in an estimated quantity, such as a mean, while a standard deviation describes variation among individual observations. The posted group standard deviations are 33.3 and 34.0. The recalculation also reports standard errors of 9.087 and 8.325 for the group means. These summaries contribute to the uncertainty in the mean difference; the confidence interval is the more direct description of precision for that contrast.

General statistical explanation: if outcome variation is greater than a simple model expects, a model that assumes less variation can understate uncertainty and make evidence appear more precise than it is. This is often described as over-dispersion. The registry identifies the primary outcome as continuous with standard deviation, but does not report an over-dispersion assessment or a model of that kind. The issue is therefore a general consideration rather than a reported feature of this trial’s analysis.

The outcome description labels the analyzed population “intent to treat” and says the participants were those that completed the study, N = 30. In general, an intent-to-treat analysis is intended to preserve participants in their randomized groups, regardless of treatment adherence, but the record’s wording does not describe the actual handling of non-completers. The record does not say whether missing outcomes were imputed, whether analyses used only observed measurements, or whether other analysis populations were used. Those details are not reported in the registry record.

The reported analyses count is 6, and the record lists 6 outcome measures. However, the information addressed does not identify all results, outcome-specific analysis plans, or whether any adjustment for multiple comparisons was used. Multiple testing can increase the chance of at least one apparently unusual result when many outcomes or analyses are considered. As a general concept, multiplicity procedures may control that risk, but no particular procedure should be assumed for this study.

10. Confidence Intervals and Precision

The recalculated mean difference in change in total cholesterol is -20.9 mg/dL for aripiprazole relative to placebo. Its 95% confidence interval is -45.05 to 3.254 mg/dL. The interval is wide enough to include zero, so the estimate is not precise enough to distinguish a negative group difference from no difference or a small difference in the opposite direction. The data also remain compatible with a more substantial negative difference.

The point estimate is only one plausible summary of the group contrast. The confidence interval communicates how much uncertainty remains when the observed group sizes and variation are taken into account. A range including zero does not prove the groups had identical outcomes, and a p-value does not resolve the uncertainty about effect size. The unadjusted p-value is 0.09, while the sponsor’s ANCOVA p-value is 0.125; these arise from different analyses and should not be treated as interchangeable precision measures.

The group standard deviations, 33.3 for aripiprazole and 34.0 for placebo, are large relative to the observed group means and indicate variation among participant changes. The analysis includes 16 and 14 participants in the respective groups. In general, limited observations combined with substantial variation can leave a group difference estimated imprecisely. The registry does not report a sponsor-model confidence interval or its adjusted estimate, so the precision of the ANCOVA contrast cannot be described from the record.

Confidence intervals are about statistical uncertainty under a method’s assumptions; they do not incorporate every possible source of uncertainty, such as missing-data assumptions, measurement issues, or limitations in trial conduct. Details on these matters are not reported in the registry record.

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 figures are planning illustrations based on the observed effect, not the trial’s original design assumptions. The original design assumptions and sample-size justification are not reported in the registry record. The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect.

For the comparison of continuous outcomes, the figures are participants per arm, not total participants. The entry for planning to detect 100% of the observed effect gives 41 participants per arm for 80% power and 55 participants per arm for 90% power. Planning for 75% of the observed effect gives 73 participants per arm for 80% power and 97 participants per arm for 90% power. Planning for 50% of the observed effect gives 163 participants per arm for 80% power and 218 participants per arm for 90% power.

In plain terms, 80% power is the planned chance of detecting an effect of the stated size under the listed assumptions if that effect is present; the 90% column uses a higher target chance of detection. Higher target power generally requires more participants under otherwise similar assumptions. The entries also illustrate that smaller effects need more participants: detecting a smaller fraction of the observed effect requires a larger planned sample in each arm.

These figures are not observed power. They are prospective illustrations derived from the observed effect and the stated calculation assumptions. They do not establish that the completed study was designed with these sample sizes, nor do they show the probability that the observed 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 = -20.900; pooled standard deviation σ = 33.627.
2. n = 2 × (1.96 + 0.842)² × 33.627² / -20.900² = 40.6, rounded up to 41 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 <- -20.9                      # observed difference in means
sigma <- 33.6268122188232                     # 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(345033)
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 = 41 (80% power), 55 (90% power)
 75% of observed effect: n per arm = 73 (80% power), 97 (90% power)
 50% of observed effect: n per arm = 163 (80% power), 218 (90% power)
power.t.test: 41.6 
Simulated power at n = 41 per arm: 0.790 (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 analysis as ANCOVA, with p = 0.125. It classifies the analysis as non-inferiority or equivalence. The record does not provide the ANCOVA effect estimate, confidence interval, covariates, margin, or decision criteria. The reported p-value alone therefore does not show the magnitude or precision of the sponsor’s adjusted comparison, and it is not enough to establish the meaning of the non-inferiority or equivalence classification.

The recalculated result is an unadjusted comparison of the posted group means: the mean difference is -20.9, with a 95% confidence interval from -45.05 to 3.254 and an unadjusted p-value of 0.09. It is based on summary statistics and does not reproduce an ANCOVA model. The two p-values can differ because the analyses use different methods and may address different statistical formulations.

The sponsor’s reported analysis is the trial’s own analysis, while the recalculation offers a simple check based on the posted summaries. Because the required model details are absent, the sponsor’s estimate cannot be reconstructed or compared numerically with an adjusted recalculation. Neither p-value by itself describes clinical importance.

14. Multiplicity, Missing Data and Analysis Populations

The record lists 6 outcome measures and 6 analyses. The primary outcome addressed is change in total cholesterol. The registry information does not identify all outcomes and their results, specify which were designated primary or secondary beyond the cholesterol outcome description, or describe a multiplicity adjustment. Whether the analyses were prespecified in a statistical analysis plan is not reported in the registry record.

As a general statistical concept, examining multiple outcomes or conducting multiple analyses can raise the chance of finding at least one result that appears unusual by chance. Methods for addressing multiplicity can adjust the interpretation or decision thresholds, but no adjustment method should be assumed when it is not described. The registry’s ANCOVA analysis is categorized as non-inferiority or equivalence, but the margin and associated decision rules are not reported. The stated analysis category does not supply those missing details.

The primary-outcome description says that the participants for analysis, labeled intent to treat, were those who completed the study, N = 30. The record lists enrollment of 38, but does not describe why some enrolled participants were not in the primary analysis population, the amount or pattern of missing outcome data, or any imputation or sensitivity analyses. It is therefore not possible to assess whether missingness could have affected the reported comparison.

In general, an intent-to-treat approach aims to analyze participants according to randomized assignment and helps preserve the benefits of randomization. The registry’s wording links its intent-to-treat label to study completers, but does not explain how treatment assignment and incomplete follow-up were handled. No further claim about the analysis population is warranted from the recorded description. Interim analyses, stopping rules, and the broader statistical analysis plan are not reported in the registry record.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports zero serious adverse events in the aripiprazole group among 16 participants at risk and zero in the placebo group among 14 participants at risk. The record lists no top serious adverse-event terms. This means no serious adverse events are recorded in these posted group figures; it does not establish that every possible safety outcome was assessed or that no unrecorded event occurred.

The record separately reports other adverse events in 4 participants at risk of 16 in the aripiprazole arm and 3 participants at risk of 14 in the placebo arm. These counts are distinct from the serious adverse-event counts. The record does not provide event descriptions in the information, so the nature, severity, timing, or relationship to assigned intervention cannot be inferred from the counts alone.

Deaths are not reported in the registry record. The posted counts should be read as the adverse-event information recorded for the groups, rather than as a complete characterization of comparative safety. Group size, follow-up, event definitions, and reporting methods affect how much can be concluded from adverse-event counts. Further safety details are not reported in the registry record.

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

The recorded primary-outcome results show that the mean change in total cholesterol was -15.3 mg/dL in the aripiprazole group and 5.6 mg/dL in the placebo group. The unadjusted difference, aripiprazole minus placebo, was -20.9 mg/dL. This describes a lower average change in the aripiprazole group over the measurement period. The estimate is uncertain: its 95% confidence interval ranges from -45.05 to 3.254 mg/dL and includes zero.

The registry also reports an ANCOVA p-value of 0.125, categorized as non-inferiority or equivalence. Without an estimate, confidence interval, margin, or model details, that result cannot establish how large an adjusted difference was or whether specified non-inferiority or equivalence criteria were met. The unadjusted recalculation and sponsor analysis should be read as distinct analyses, not as duplicate versions of one statistic.

The result does not establish a clinical benefit or lack of benefit. The registry provides no threshold for a clinically meaningful change in total cholesterol, and the uncertainty interval allows a range of possible group differences. The result also does not establish effects on insulin resistance, weight gain, or other outcomes mentioned in the study summary; their results are not reported in the registry record addressed.

The analysis concerns participants recorded as study completers, N = 30, despite the registry listing enrollment of 38. The handling and outcomes of participants not included in that primary analysis are not described. The study’s follow-up for this endpoint was measured at Week 8, so these results do not describe longer-term outcomes. Conclusions should remain limited to the reported comparison, measurement period, analysis population, and uncertainty.

17. Why This Trial Matters Statistically

This trial illustrates why a point estimate, its uncertainty interval, and the analysis method all matter. The posted group means produce an unadjusted mean difference of -20.9, while the confidence interval extends from -45.05 to 3.254 and includes zero. The estimate gives the direction and scale of the observed contrast; the interval shows that the data do not pin down its size precisely. A p-value alone would not convey this range.

The record also provides a useful distinction between a sponsor analysis and a simple recalculation from group summaries. The sponsor reports ANCOVA with p = 0.125, while the unadjusted recalculation has p = 0.09. Different analytical approaches can yield different results, and missing model details limit independent interpretation of the sponsor’s result. Clear reporting of estimates, confidence intervals, covariates, and decision criteria is important for understanding what an analysis supports.

Planning illustrations further show how sample-size requirements depend on the effect a study aims to detect and the target power. The listed per-arm figures increase when planning for smaller fractions of the observed effect or for the higher power target. They are illustrative calculations, not evidence of the original design or of the probability that the observed result is correct.

Finally, the analysis population and missing-data description affect interpretation. The registry calls the population intent to treat but describes it as participants who completed the study. Without information on non-completion and missing measurements, readers cannot fully assess how the observed comparison relates to all randomized participants. These statistical reporting issues are relevant even when the primary endpoint is clearly defined.

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 registry reports enrollment of 38 and an analysis population of N = 30 described as participants who completed the study. It does not explain the difference between enrollment and analysis, give reasons for non-completion, or describe missing-data methods. The reported group sizes are 16 for aripiprazole and 14 for placebo. This leaves uncertainty about how representative the analyzed participants are of everyone enrolled and how missing outcomes may have affected the comparison.

The sponsor’s ANCOVA result is reported with p = 0.125, but its effect estimate, confidence interval, covariates, and model details are not reported in the registry record. The analysis is categorized as non-inferiority or equivalence, yet the relevant margin and decision criteria are also absent. This limits interpretation of the sponsor’s statistical conclusion and prevents reproduction of that analysis from the recorded information.

The primary endpoint is a change measured at Week 8. Longer-term results for total cholesterol are not reported in the registry record. Although the study summary mentions insulin resistance, blood fat levels, and weight gain, the registry information addressed does not provide results for those other measures. The record also does not provide a threshold for a clinically meaningful total-cholesterol change.

Further details about randomization, masking implementation, the original sample-size justification, interim analyses, multiplicity methods, and the statistical analysis plan are not reported in the registry record. These gaps do not alter the numerical results but constrain assessment of the trial’s conduct and analysis choices.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Effectiveness of Aripiprazole for Improving Side Effects of Clozapine in the Treatment of People With Schizophrenia — NCT00345033.
• Registry record last updated 2014-06-23; status last verified 2014-06.
• Results first posted 2012-09-13.
• 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

NCT00345033 is a completed Phase 4 randomized, triple-masked, parallel-group study sponsored by Massachusetts General Hospital. The registry lists enrollment of 38 and conditions of schizophrenia and insulin resistance. The study compared aripiprazole 15mg/day with placebo for 8 weeks.

The primary outcome was change in total cholesterol from baseline to Week 8. The reported mean changes were -15.3 mg/dL for aripiprazole and 5.6 mg/dL for placebo. The unadjusted mean difference was -20.9, with a 95% confidence interval of -45.05 to 3.254. The sponsor reported an ANCOVA p-value of 0.125, categorized as non-inferiority or equivalence, without an estimate or confidence interval in the registry record.

The primary analysis population is described as study completers, N = 30. Further details on missing data, model specification, and other outcomes are not reported in the registry record.