A Phase 2 Efficacy and Safety Study of TAK-063 in Participants With an Acute Exacerbation of Schizophrenia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase 2 Efficacy and Safety Study of TAK-063 in Participants With an Acute Exacerbation of Schizophrenia (NCT02477020), a randomized phase 2 study. The primary outcome is Change From Baseline in the Positive and Negative Symptom Scale (PANSS) Total Score at Week 6.

2. Clinical Question

The comparison concerns the average change in symptom severity, not whether every participant improved or whether individual responses were alike. The primary endpoint is continuous, so the estimated contrast retains information about the scale of change rather than reducing results to a responder category. The registry also lists secondary outcomes involving PANSS subscales and response, clinical global ratings, and cognition. Those outcomes provide other planned views of symptoms and functioning, but the primary statistical question is the Week 6 PANSS total-score contrast.

3. Trial Design

This was a Phase 2, randomized, double-blind, placebo-controlled, parallel-group study of TAK-063 in participants with an acute exacerbation of schizophrenia. The registry lists enrollment of 164 and a study duration of 6 weeks. It records the study as completed, with a start date of 2015-07-01 and a primary completion date of 2016-07-13. Takeda is listed as the sponsor.

Participants were assigned to TAK-063 20 mg or matching placebo. Both were oral tablets taken once daily for up to 6 weeks. The TAK-063 dose could be reduced to 10 mg/day if intolerable; the placebo description does not specify a corresponding dose adjustment. The trial used a parallel-group model, meaning the treatment groups were followed as separate arms rather than participants crossing between treatments. The registry describes masking as quadruple and the allocation as randomized.

The primary endpoint was change from baseline in PANSS total score at Week 6. The registry reports 55 participants analyzed in the TAK-063 arm and 51 in the placebo arm for this outcome. These are outcome-specific analyzed counts; the registry’s total enrollment is a separate record-level figure. The primary efficacy population required randomization, at least one dose, a baseline value, and at least one valid postbaseline primary efficacy value.

The registry lists 12 outcome measures and 86 analyses, with secondary outcomes covering symptom scores, clinical global ratings, responder measures, and cognition. It does not report the original sample-size justification or the trial’s original design assumptions for power. The later sample-size illustration is a planning exercise based on the observed effect, not a statement of what the original trial was designed to detect.

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: Participants with a diagnosis of acute exacerbation of schizophrenia were enrolled in two treatment groups TAK-063 or placebo.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The full analysis set included all participants who were randomized, received at least one dose of study drug and have a baseline value and at least one valid postbaseline value for assessment of primary efficacy. Here number of participants analyzed are participants evaluable for this outcome measure.

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.

Population: All randomized participants who received at least 1 dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in the Positive and Negative Symptom Scale (PANSS) Total Score at Week 6, measured over Baseline and Week 6.

The primary endpoint is change from baseline in total score on the Positive and Negative Syndrome Scale (PANSS) at Week 6. PANSS assesses positive symptoms, negative symptoms, and general psychopathology associated with schizophrenia. It consists of 30 items, each rated from 1, symptom not present, to 7, symptoms extremely severe. The total is the sum of those item scores and ranges from 30 to 210; a higher score indicates greater severity. A negative change from baseline therefore indicates improvement.

A total score offers a single continuous outcome that combines several symptom domains. Analyzing change from baseline makes the question about how the total score shifted over the study period, rather than only its final value. The Week 6 assessment is the primary time point in the registry record; other PANSS assessments appear among the secondary outcomes.

The reported group estimates are least-squares means with standard errors, obtained using a mixed model for repeated measures. Least-squares means are model-based estimates of average change, and the treatment contrast compares those estimates. This differs from simply comparing raw averages, because the reported model uses repeated observations over time. The registry’s endpoint description does not specify additional details of the model beyond that method and the reported outcome.

Clinical meaning should be read alongside statistical precision. The estimated contrast describes the average difference in change, while its confidence interval indicates a range of effect sizes compatible with the reported analysis. Neither the average nor the interval describes how any particular participant will respond.

7. Results Posted in the Registry

• TAK-063: mean -19.54 versus -14.08 on control; difference -5.46 (95% CI -12.23 to 1.31), standardized effect d = -0.31; z-test p = 0.114.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a change in PANSS total score. The analysis reports least-squares means for each arm and a least-squares mean difference between arms. These are model-based estimates rather than simple unadjusted averages. The registry identifies the method as a mixed model for repeated measures (MMRM), with baseline and postbaseline assessments contributing to the analysis.

What does a mixed model for repeated measures do?

Generally, repeated-measures methods analyze observations collected from the same participant at multiple time points while recognizing that measurements from one participant are related. A mixed model can represent that within-participant correlation and estimate average patterns over time. It can also use available observations when some scheduled measurements are missing, subject to the model and missing-data assumptions. This general property does not establish which specific assumptions or model details were used in this trial.

The registry reports the MMRM method and the Week 6 endpoint, but does not report the full model specification, covariance structure, covariates, handling of intercurrent events, or detailed missing-data assumptions. Those details are not reported in the registry record. It is therefore not possible to reconstruct the sponsor’s model from the posted summary statistics alone.

How should the least-squares means be read?

Generally, a least-squares mean is an estimate adjusted according to the model’s specified structure. It should not automatically be interpreted as the raw mean of the observed values. Here, the posted arm estimates are -19.54 for TAK-063 and -14.08 for placebo, and the reported contrast is -5.46. The contrast is the model’s estimate of the difference in average change, with negative values indicating a greater decrease in the TAK-063 arm.

The registry reports standard errors of 2.412 and 2.474 for the arm estimates. A standard error describes uncertainty in an estimated mean; it is not the standard deviation of individual changes. The comparison’s confidence interval is the more direct summary of uncertainty in the difference between arms.

What does the p-value test?

Generally, a p-value measures the compatibility of the data with a null model, conditional on the analysis assumptions. The sponsor reports p = 0.115 for the primary comparison. It is not a measure of effect size, a probability that the null hypothesis is true, or a measure of clinical importance. Interpreting it alongside the estimated difference and confidence interval is more informative than using it alone.

The registry record does not report whether this p-value was adjusted for multiple outcomes or analyses. The study lists 12 outcome measures and 86 analyses, but the number of listed outcomes does not establish the multiplicity procedure. No statistical plan for controlling error across analyses is reported in the registry record.

Why can a recalculation differ from the reported model result?

Generally, a simple recalculation from arm-level summary statistics cannot reproduce every feature of a repeated-measures model. A recalculation based on the posted estimates and standard errors can provide an unadjusted reference, but it does not replace the sponsor’s model-based analysis. The recalculated contrast is -5.46, with a 95% confidence interval from -12.23 to 1.312 and p = 0.114. The sponsor’s reported MMRM result is -5.46, with a 95% confidence interval from -12.26 to 1.35 and p = 0.115. The registry record does not mark a discrepancy.

The recalculation uses an approximate pooled standard deviation of 17.78 and an effect-size estimate of -0.307. These are derived reference quantities, not additional trial endpoints. In particular, the approximate standard deviation is not a direct reported measure of the individual outcome distribution. Such recalculations are useful for orientation, but the sponsor’s reported analysis remains the trial’s own model-based result.

What cannot be inferred from the reported method?

The method name alone does not reveal the assumptions about missing observations, the model’s covariance structure, the treatment of protocol deviations, or the handling of multiple testing. Details about interim analyses, an analysis population beyond the stated full analysis set definition, and a prespecified sample-size rationale are not reported in the registry record. General statistical explanations describe how methods can work; they do not fill in these trial-specific omissions.

10. Confidence Intervals and Precision

The sponsor’s reported 95% confidence interval for the primary difference is -12.26 to 1.35. It places the point estimate of -5.46 within a range that includes zero. Under the model and assumptions used for the reported analysis, this interval conveys uncertainty about the treatment contrast. It is not a range containing a known proportion of individual participants’ responses.

In general, a confidence interval is a method for expressing uncertainty in an estimate. A narrower interval indicates greater precision on the outcome scale; a wider interval leaves more possible effect sizes compatible with the data and analysis. Here, the interval extends from a larger decrease in PANSS total score with TAK-063 to a small positive difference in the opposite direction. Thus, the estimated direction favors a greater average decrease with TAK-063, but the interval does not exclude no difference or the reverse direction.

The arm-level standard errors are 2.412 for TAK-063 and 2.474 for placebo. They describe the estimated uncertainty around each least-squares mean, not participant-to-participant variability. The sponsor’s interval for the contrast is the appropriate reported summary for precision of the between-arm comparison. Its limits are not identical to the recalculated interval, which is -12.23 to 1.312; the recalculation is an unadjusted approximation and should not be substituted for the sponsor’s MMRM interval.

The registry record does not report a clinical threshold for interpreting the size of a PANSS difference. Statistical precision and clinical importance are distinct questions: the interval indicates uncertainty in the numerical effect, while a clinical threshold would be needed to judge whether particular values represent a meaningful difference. That threshold is 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 display is a planning illustration based on the observed effect, not the trial’s original design justification. The figures are participants per arm, not total participants or total events. The calculation assumptions listed in the registry analysis 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. The original trial’s own sample-size assumptions are not reported in the registry record.

The 80% power column gives the planned participants per arm for a hypothetical study designed to have an 80% chance of detecting the specified effect, if the assumptions used in the calculation hold. The 90% power column expresses the corresponding planning target for a 90% chance. These are design probabilities under a specified effect and analysis framework, not probabilities that the present result is correct. This is not observed power.

The rows consider planning for 100%, 75%, and 50% of the observed effect. Planning for 100% of the observed effect gives 167 participants per arm for 80% power and 223 for 90% power. Planning for 75% gives 297 participants per arm for 80% power and 397 for 90% power. Planning for 50% gives 667 participants per arm for 80% power and 892 for 90% power. Smaller effects need more participants to detect under the same stated assumptions.

These figures are conditional illustrations, not a recommendation for a future study and not evidence that the trial was underpowered or adequately powered. Their interpretation depends on treating the observed effect as a planning reference, which can be uncertain. A planning curve can help visualize how required sample size changes as the assumed effect changes; it does not add independent evidence about the treatment effect.

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 = -5.460; pooled standard deviation σ = 17.782.
2. n = 2 × (1.96 + 0.842)² × 17.782² / -5.460² = 166.5, rounded up to 167 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -5.46                      # observed difference in means
sigma <- 17.7824523163881                     # 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(2477020)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 167 (80% power), 223 (90% power)
 75% of observed effect: n per arm = 297 (80% power), 397 (90% power)
 50% of observed effect: n per arm = 667 (80% power), 892 (90% power)
power.t.test: 167.5 
Simulated power at n = 167 per arm: 0.807 (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 sponsor’s own primary analysis is the MMRM comparison reported in the registry: a least-squares mean difference of -5.46, with a 95% confidence interval from -12.26 to 1.35 and p = 0.115. This is the analysis most directly tied to the trial’s stated repeated-measures method.

The simple recalculation from the posted summary statistics gives the same displayed point estimate, -5.46, but a 95% confidence interval from -12.23 to 1.312 and p = 0.114. It uses an approximate pooled standard deviation and is an unadjusted recalculation, so it is not a reproduction of the sponsor’s full MMRM. The registry does not mark a discrepancy. The small differences in the interval limits and p-value should therefore be read as differences between the reported model analysis and an approximate calculation, not as two independent trial findings.

Both presentations have a confidence interval that includes zero. The sponsor’s estimate remains the primary reported result; the recalculation provides a secondary statistical check based on the posted summaries. Neither should be converted into a claim of clinical importance, since the registry record does not report a threshold for that judgment.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 12 outcome measures and 86 analyses. The primary endpoint is change from baseline in PANSS total score at Week 6. The secondary outcomes include PANSS total scores at earlier time points, PANSS subscales, clinical responder measures, clinical global ratings, and cognition assessments. This breadth means that the primary comparison sits within a wider collection of outcomes and analyses.

Multiplicity arises generally when several hypotheses are tested: even if each test uses the same significance threshold, the chance of at least one false-positive result across a family of tests can increase. A prespecified procedure may address this, for example by defining a testing sequence or adjusting significance thresholds. The registry record does not report whether the primary p-value was adjusted, which outcomes formed any testing families, or what multiplicity strategy was used. The reported p-value should therefore be interpreted as presented, without assuming that adjustment was or was not performed.

The primary efficacy population is described as participants who were randomized, received at least one dose, and had a baseline value and at least one valid postbaseline value for primary efficacy assessment. The outcome-specific analysis counts are 55 for TAK-063 and 51 for placebo. The registry does not provide further detail in this record about exclusions, protocol deviations, or how those counts relate to missing Week 6 measurements.

MMRM can generally use available repeated measurements under its modeling assumptions, but the method name does not establish that missing data are ignorable or explain how missingness was handled in this trial. Detailed missing-data assumptions, sensitivity analyses, and imputation procedures are not reported in the registry record. The effect of missing observations on the result therefore cannot be assessed from these details alone.

The registry also does not report interim analyses, a full statistical analysis plan, or the original sample-size justification. These are distinct from the planning illustration based on the observed effect. Without the analysis plan and missing-data details, the scope of confirmatory inference and the robustness of the result to alternative assumptions remain uncertain.

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 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: schizophrenia (2.4%), cellulitis (1.2%), agitation (1.2%), suicidal ideation (1.2%).

The registry reports serious adverse events among 81 participants at risk in the placebo group, with 2 affected, and among 83 participants at risk in the TAK-063 group, with 4 affected. These are serious-adverse-event counts and denominators; they should not be confused with the primary efficacy analysis counts. The record does not report deaths, so death information is not reported in the registry record.

The listed serious adverse event terms and percentages are Schizophrenia, 2.4%; Cellulitis, 1.2%; Agitation, 1.2%; and Suicidal ideation, 1.2%. These are the top serious-event terms recorded in the registry, not a complete description of every possible safety outcome. The registry also reports other adverse events affecting 42 of 81 at risk in the placebo group and 58 of 83 at risk in the TAK-063 group.

These counts describe events observed during the trial and do not by themselves establish that treatment caused an event. They also do not determine comparative safety, because interpretation depends on event definitions, exposure, follow-up, severity, and the broader clinical context. Further detail on those matters is not reported in the registry record. The figures should be read as descriptive safety information from this study, not as medical advice or a prediction for an individual.

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

The primary estimate is a -5.46-point difference in mean PANSS total-score change at Week 6, with the point estimate indicating a larger average decrease in the TAK-063 arm than in the placebo arm. The sponsor’s 95% confidence interval, -12.26 to 1.35, includes zero, and the reported p-value is 0.115. The posted result therefore leaves substantial uncertainty about the size and direction of the average difference.

This result does not establish that the groups were equivalent. A confidence interval that includes zero can be compatible with no difference, but equivalence requires a prespecified equivalence margin and an analysis designed for that question. No such margin is reported in the registry record. Nor does the result prove that one treatment is superior: the estimated difference is not sufficiently precise to exclude zero under the reported analysis.

The point estimate and interval concern group averages, not the likely response of a particular participant. The trial’s primary endpoint is a symptom scale, and the registry does not report a threshold for a clinically important difference. Statistical evidence about an average score change therefore cannot, on its own, establish how meaningful a difference would be to participants or clinicians.

The finding is specific to the registered Phase 2 study: participants with an acute exacerbation of schizophrenia, oral TAK-063 with a possible dose reduction, matching placebo, and assessment at Week 6. The available record does not support extending the estimate to other populations, durations, doses, or clinical circumstances. The reported safety counts likewise describe events recorded in this study and do not establish causation or predict individual risk.

In short, the posted data show a negative estimated average contrast, but the confidence interval includes effects in both directions. The registry result should be understood as an uncertain estimate, not as proof of benefit, proof of no effect, or evidence of equivalence.

17. Why This Trial Matters Statistically

This trial illustrates why a point estimate, confidence interval, and p-value should be interpreted together. The point estimate of -5.46 indicates the direction and scale of the average modeled difference. The confidence interval from -12.26 to 1.35 shows that the estimate is imprecise enough to include zero and values on either side. The p-value of 0.115 is consistent with that uncertainty; it does not replace the interval or measure clinical importance.

It also demonstrates the distinction between a model-based primary analysis and a summary-statistic recalculation. The sponsor used MMRM, while the recalculation uses an approximate pooled standard deviation and unadjusted summary-statistic methods. Their point estimates match, but their uncertainty summaries are not identical. Reporting both can help readers understand what can and cannot be recovered from summary data, while keeping the trial’s own analysis central.

The registry’s list of multiple outcomes and analyses raises a further statistical question: how should evidence be interpreted when many comparisons are available? The record does not describe a multiplicity strategy, so the primary p-value cannot be placed in a broader testing framework from the registry entry alone. This absence does not show that no strategy existed; it means the strategy is not reported in the registry record.

Finally, planning calculations based on the observed effect illustrate how the assumed target effect influences participant requirements. Smaller target effects need more participants under the stated assumptions. Those calculations are planning illustrations, not observed power and not evidence about the original trial’s design. Together, these points show why transparent reporting of methods, uncertainty, missing-data handling, and planning assumptions is essential for interpreting a clinical trial result.

18. Limitations

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

Several details needed for a fuller statistical appraisal are not reported in the registry record. These include the original sample-size justification, full MMRM specification, covariance structure, model covariates, detailed missing-data assumptions, and sensitivity analyses. The registry also does not state whether the primary p-value was adjusted for multiplicity or how the 86 listed analyses were organized for inference.

The primary efficacy population requires a baseline value and at least one valid postbaseline value, rather than simply including every randomized participant in the endpoint analysis. The record does not describe the reasons participants may have been excluded from this outcome-specific population or provide further detail on missing Week 6 assessments. That limits assessment of how missing observations or population selection may affect the estimate.

The sponsor’s reported confidence interval includes zero and extends to values in both directions. The registry does not provide a minimal clinically important difference, so the clinical relevance of values within that interval cannot be assessed from the endpoint estimate alone. The secondary outcomes could add context, but their results and the approach to interpreting multiple analyses are not detailed in the registry record for this account.

The safety figures are descriptive counts. Detailed exposure-adjusted rates, event timing, event-level narratives, and further information about deaths are not reported in the registry record. The posted counts cannot establish causality or fully characterize comparative safety. Finally, the sample-size illustration uses the observed effect and specified planning assumptions; it does not reveal the original design rationale and should not be used as a retrospective measure of study quality.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Phase 2 Efficacy and Safety Study of TAK-063 in Participants With an Acute Exacerbation of Schizophrenia — NCT02477020.
• Registry record last updated 2026-05-08; status last verified 2026-04.
• Results first posted 2017-09-29.
• 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

NCT02477020 is a completed, randomized, double-blind, placebo-controlled, parallel-group Phase 2 study sponsored by Takeda. It enrolled 164 participants and evaluated oral TAK-063 20 mg, with a possible reduction to 10 mg/day if intolerable, against matching placebo for up to 6 weeks.

The primary outcome was change from baseline in PANSS total score at Week 6. The sponsor-reported MMRM contrast was -5.46, with a 95% confidence interval from -12.26 to 1.35 and p = 0.115. The interval includes zero, leaving uncertainty about the direction and size of the average difference. Serious adverse events were reported for 2 of 81 participants at risk with placebo and 4 of 83 at risk with TAK-063. Further details about the statistical plan, missing-data handling, multiplicity, and original sample-size rationale are not reported in the registry record.