Adjunctive Pimavanserin in Subjects With Major Depressive Disorder and Inadequate Response to Antidepressant Treatment: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Adjunctive Pimavanserin in Subjects With Major Depressive Disorder and Inadequate Response to Antidepressant Treatment (NCT03968159), a randomized phase 3 study. The primary outcome is Change From Baseline to Week 5 in Hamilton Depression Scale (17 Items) (HAMD-17) Total Score.

2. Clinical Question

3. Trial Design

This was a Phase 3, multicenter, randomized, double-blind, placebo-controlled parallel-group study of adjunctive pimavanserin for participants with major depressive disorder and an inadequate response to antidepressant treatment. The registry record describes masking as quadruple. It does not say which participant, clinician, assessor, or other roles were masked, so the exact meaning of that label cannot be assessed from the record.

The intervention was pimavanserin 34 mg tablets; the comparator was placebo tablets. The study was intended to evaluate efficacy and safety in the context of inadequate response to antidepressant therapy. The background antidepressant drugs, rules about their continuation or adjustment, and the duration of treatment beyond the primary Week 5 assessment are not reported in the registry record.

The record lists 298 enrolled participants. For the primary endpoint, the stated Full Analysis Set included randomized participants who received at least one dose of study drug and had a baseline HAMD-17 value and at least one postbaseline value. The reported arm summaries list 148 participants for pimavanserin and 149 for placebo. The record does not explain the relationship between these arm counts and the overall enrollment figure, so they should be read as reported for the relevant endpoint rather than reconciled by assumption.

The study start date was 2019-04-25, primary completion was 2020-04-30, and the registry status is completed. Allocation ratio, site locations, detailed eligibility criteria, the rationale for the planned enrollment, and the trial's original sample-size assumptions are not reported in the registry record. The record lists multiple secondary outcomes, including symptom severity, functioning, sexual functioning, sleepiness, anxiety/somatization, impulsiveness, response, and remission measures. Numerical results are not available in the posted record for every listed outcome.

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: During the screening period, patients were assessed for study eligibility, and prohibited medications were discontinued when medically appropriate.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set: all subjects who were randomized, received at least one dose of study drug, and had a baseline value and at least one postbaseline value for HAMD-17 total score.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline to Week 5 in Hamilton Depression Scale (17 Items) (HAMD-17) Total Score, measured over Baseline, Week 5.

The primary endpoint was change from baseline to Week 5 in total score on the 17-item Hamilton Depression Scale, or HAMD-17. The registry describes the scale as having 8 items scored on a 3 point scale and 9 items scored on a 5 point scale. Their sum ranges from 0 to 52, and higher total scores denote more severe depression. Change is interpreted in relation to that direction: a more negative change corresponds to a larger reduction in the score over the assessment period.

A symptom scale provides a structured way to quantify the measured construct at the start of treatment and at a specified follow-up. The change endpoint focuses on movement from baseline rather than the follow-up score alone. Comparing the average changes between randomized groups asks whether the observed score trajectory differed between the assigned study treatments, while retaining the clinical meaning of the HAMD-17 scale.

The registry reports least-squares means for the arms and a least-squares mean difference from a mixed-effects model for repeated measures. Least-squares means are model-based estimates, rather than necessarily simple arithmetic averages of every observed score. Details of the model specification beyond its stated method are not reported in the registry record. The endpoint describes change through Week 5; it does not, on its own, establish the durability of any difference beyond that assessment or address outcomes not measured by the scale.

7. Results Posted in the Registry

• Pimavanserin: mean -9.00 versus -8.10 on control; difference -0.90 (95% CI -2.51 to 0.71), standardized effect d = -0.13; z-test p = 0.273.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change From Baseline to Week 5 in Clinical Global Impression-Severity (CGI-S) Score for Depressive Symptoms

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Pimavanserin: mean -1.40 versus -1.10 on control; difference -0.30 (95% CI -0.58 to -0.02), standardized effect d = -0.25; z-test p = 0.034.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Pimavanserin than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline to Week 5 in Sheehan Disability Scale (SDS) Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Pimavanserin: mean -2.50 versus -2.10 on control; difference -0.40 (95% CI -1.01 to 0.21), standardized effect d = -0.15; z-test p = 0.199.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Pimavanserin and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline to Week 5 in the Changes in Sexual Functioning Questionnaire Short Form

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Pimavanserin: mean 3.40 versus 2.50 on control; difference 0.90 (95% CI -0.93 to 2.73), standardized effect d = 0.11; z-test p = 0.335.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Pimavanserin and the control. The interval gives the range of effects that remain compatible with the data.

The numerical secondary results included in the posted summaries show a mixed pattern rather than a uniform shift across measures. The pimavanserin arm has a more negative mean change than placebo on the CGI-S and SDS measures, while the sexual-functioning measure has a positive difference. The intervals for the SDS and sexual-functioning comparisons include zero; the CGI-S interval does not. These are unadjusted re-calculations from the posted summary statistics, not sponsor-reported model comparisons in the record.

The primary HAMD-17 comparison is also uncertain: its sponsor-reported confidence interval includes zero. The secondary summaries therefore do not change the central interpretation of the primary endpoint. The numerical secondary results may offer context about other measured domains, but they do not independently confirm or override the primary result. The registry lists additional secondary outcomes, while numerical arm-level comparisons for all listed outcomes are not reported in the registry record.

Unadjusted secondary results carry less weight because the trial evaluates multiple outcomes. In general, when many statistical comparisons are made, the chance of at least one apparently notable result arising by chance can increase. A planned multiplicity strategy can address that issue, but whether such a strategy was used, what comparisons it covered, and how it affected interpretation are not reported in the registry record. The unadjusted p-values should therefore be read as exploratory summaries rather than as a set of definitive independent findings.

The outcomes also measure different constructs and use different scales. A numerical difference on symptom severity, disability, or sexual functioning does not automatically imply a corresponding difference on another measure. Without a stated hierarchy for the secondary endpoints or a reported multiplicity approach, the overall pattern is best described cautiously: some estimates lean in one direction, others are less conclusive or point in another direction, and the available secondary results do not provide a consistent basis for changing the primary interpretation.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Change From Baseline to Week 5 in Clinical Global Impression-Severity (CGI-S) Score for Depressive Symptoms
• Change From Baseline to Week 5 in Sheehan Disability Scale (SDS) Score
• Change From Baseline to Week 5 in the Changes in Sexual Functioning Questionnaire Short Form
• Change From Baseline to Week 5 in Karolinska Sleepiness Scale (KSS) Score
• Change From Baseline to Week 1 in the HAMD-17 Total Score
• Treatment Responder and Treatment Remission Rates at Week 5
• Change From Baseline to Week 5 in the Hamilton Depression (HAMD) Anxiety/Somatization Factor Score
• Change From Baseline to Week 5 in the Barratt Impulsiveness Scale (BIS-11)

9. Statistical Interpretation

10. Statistical Methods Explained

The posted primary analysis is described as a mixed-effects model for repeated measures, with a least-squares mean difference as the comparison measure. Repeated-measures models are used when an outcome is measured at more than one time point. In general, such a model can use the pattern of available repeated observations to estimate group outcomes over time and compare groups at a specified assessment. That is a general description of the method; the registry record does not report the exact model terms, covariance structure, estimation procedure, or missing-data assumptions used in this trial.

What is a least-squares mean?

In general, a least-squares mean is a model-based estimate of a group mean. Depending on the model, it may account for specified terms or covariates, whereas a raw mean is calculated directly from observed values. The registry gives the primary arm values as least-squares means and identifies the sponsor comparison as an LSM difference. The details needed to reconstruct those model-based estimates are not reported in the registry record, so the reported values should not be treated as simple raw averages.

The primary outcome summaries list a standard error of 0.58 for each arm. A standard error describes the estimated uncertainty of a mean, rather than the spread of individual participant scores. In general, standard errors tend to be smaller when estimates are more precise, but their size also depends on the variability of observations and the analysis. The registry does not provide the underlying participant-level observations, so the posted standard errors cannot be independently verified from individual data.

What does the re-calculation do?

The re-calculated comparison uses the posted arm-level summary statistics to form a simple unadjusted contrast and its interval. It uses an approximate standard deviation inferred from the reported standard errors, as indicated in the posted analysis details. This is a summary-statistic calculation, not a reproduction of the sponsor's mixed-effects model. The re-calculated primary estimate is -0.9, with a 95% confidence interval from -2.508 to 0.708 and p = 0.273. Because this approach does not incorporate the trial model's unreported details, it is best used as a transparent check on the broad direction and uncertainty, not as a replacement for the reported analysis.

General statistical explanation: an unadjusted comparison uses the group summaries without additional model-based adjustment. A model-based comparison may use information across repeated measurements and specified covariates or terms. If the model specification differs from a simple contrast, estimates of uncertainty and associated p-values can differ. The registry record does not report enough detail to attribute the differences between the posted sponsor result and the re-calculation to a particular modeling choice.

How should the confidence interval be read?

In general, a confidence interval expresses uncertainty around an estimate under the assumptions of the analysis. Its endpoints represent values reasonably compatible with the observed data and method; the interval is not a guarantee that the true value lies within it. For the sponsor-reported primary comparison, the 95% confidence interval extends from -2.5 to 0.8. Because it includes zero, the data do not clearly separate a negative difference from no difference or a positive difference under that analysis. The interval also indicates that the point estimate should not be read as an exact or highly precise effect size.

The p-value is another summary of the compatibility of the data with a specified null hypothesis under the statistical model. In general, a larger p-value does not prove that there is no effect, and a smaller p-value does not measure the size or importance of an effect. The primary p-value reported by the sponsor is 0.296. The unadjusted re-calculation gives p = 0.273. The sponsor result is the official reported analysis; the re-calculation is a separate, approximate summary-based calculation.

Why can over-dispersion matter?

General statistical explanation: over-dispersion occurs when observed variability is greater than a model assumes. If a model understates variability, standard errors can be too small and intervals too narrow; if variability is modeled appropriately, uncertainty estimates better reflect the observed data. The primary endpoint is continuous, and the posted summary includes standard errors, but the registry does not report an over-dispersion assessment or a specific correction. It would therefore be inappropriate to claim that over-dispersion was present or that any adjustment was made.

Interpretation also depends on who was included in the analysis. The stated Full Analysis Set required randomization, at least one dose, a baseline HAMD-17 value, and at least one postbaseline value. This is a defined analysis population, not necessarily every person enrolled or randomized. The registry record does not describe the number excluded at each step, how missing observations were handled, or whether sensitivity analyses were conducted. Those omissions limit independent assessment of the analysis assumptions.

11. Confidence Intervals and Precision

Precision is reflected in the width of the interval around an estimate: narrower intervals indicate less statistical uncertainty, while wider intervals leave a broader range of effects compatible with the data and analysis. The sponsor-reported primary difference was -0.9, with a 95% confidence interval from -2.5 to 0.8. The interval crosses zero, so it does not establish a clear direction for the between-group contrast at the stated analysis. It includes the possibility of a negative difference, no difference, and a positive difference.

The point estimate is only one part of the result. It is the model's estimated difference, but uncertainty around it matters for judging how much weight to place on its direction and size. The confidence interval does not say that every value inside it is equally plausible, nor does it state the probability that the true effect equals any particular value. It is an interval constructed using a specified statistical method; its interpretation depends on that method and its assumptions.

The approximate unadjusted re-calculation gives a 95% confidence interval from -2.508 to 0.708. This interval likewise includes zero. It is not identical to the sponsor's interval, which is expected to be treated as the primary reported analysis rather than replaced by an approximate calculation. The re-calculation relies on summary statistics and an approximate standard deviation, whereas the sponsor result is reported from a mixed-effects model for repeated measures. Detailed model specifications are not reported in the registry record.

A confidence interval that includes zero does not establish equivalence or prove an absence of a potentially meaningful difference. Equivalence or noninferiority requires an appropriate design and prespecified margin, neither of which is reported for the primary question in the registry record. The record also does not provide a threshold for a clinically meaningful HAMD-17 contrast. Consequently, precision can be described statistically, but clinical importance cannot be settled by the interval alone.

12. 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 a report of the trial's original design calculation. The figures are participants per arm, rather than total participants or total events. The underlying 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. The trial's own original design assumptions and sample-size justification are not reported in the registry record.

The rows describe planning for 100%, 75%, and 50% of the observed effect. The 80% power column gives the estimated participants per arm for a design with an 80% chance of detecting the specified effect under the stated assumptions, if that effect is real and the assumptions hold. The 90% power column describes the corresponding planning target for a 90% chance. These are design probabilities, not guarantees for a particular future study.

Smaller effects are harder to distinguish from random variation, so planning to detect a smaller fraction of the observed effect requires more participants per arm under the same assumptions. The sample-size curve expresses that relationship visually; it should be read as a planning aid rather than as evidence about the actual trial's power. These figures are not observed power. Observed power is not a useful substitute for reading the estimated effect and its confidence interval after a study has been completed.

The calculations assume equal allocation, but the registry does not report that these were the assumptions used when the trial was originally planned. Nor do the illustrations account for unreported aspects of the trial's model, missing observations, or any other design considerations. Their role is to show how planning requirements change when the target effect is varied, not to judge whether the completed trial was adequately designed or whether its result is true.

13. 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 = -0.900; pooled standard deviation σ = 7.068.
2. n = 2 × (1.96 + 0.842)² × 7.068² / -0.900² = 968.1, rounded up to 969 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 <- -0.9                      # observed difference in means
sigma <- 7.06795374698303                     # 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(3968159)
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 = 969 (80% power), 1,297 (90% power)
 75% of observed effect: n per arm = 1,722 (80% power), 2,305 (90% power)
 50% of observed effect: n per arm = 3,873 (80% power), 5,185 (90% power)
power.t.test: 969.1 
Simulated power at n = 969 per arm: 0.797 (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.

14. Analyses Reported by the Sponsor

The sponsor-reported primary analysis is a mixed-effects model for repeated measures. It gives a least-squares mean difference of -0.9, a 95% confidence interval from -2.5 to 0.8, and p = 0.296. This is the trial's own reported primary comparison and should be distinguished from the summary-statistic re-calculation.

The re-calculation gives the same rounded point estimate of -0.9, with a 95% confidence interval from -2.508 to 0.708 and p = 0.273. It is an unadjusted calculation based on the posted arm summaries, with an approximate standard deviation inferred from the standard errors. It does not reproduce the mixed-effects model. Both analyses yield an interval that includes zero, but their uncertainty estimates and p-values are not identical.

The registry record does not provide the full model specification or sufficient details to explain precisely why the intervals and p-values differ. It would be speculative to attribute the difference to particular covariates, correlations, or missing-data procedures. The sponsor analysis is the direct report of the trial's primary method; the re-calculation provides a transparent, approximate check rather than an alternative official result.

There is no flagged discrepancy between the posted arm summaries and the sponsor-reported analysis. Even so, the two forms of analysis answer the comparison using different levels of information. Readers should use the sponsor estimate and its stated interval for the main interpretation, while treating the re-calculation as a limited summary-statistic exercise.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists multiple secondary outcomes in addition to the primary HAMD-17 endpoint. When a study tests many outcomes or comparisons, the chance of at least one result appearing unusual by chance can increase. Statistical approaches to multiplicity may adjust the interpretation or testing threshold, but the registry record does not report whether a multiplicity procedure was planned or applied, which outcomes it covered, or whether any secondary analyses were designated as confirmatory.

The primary result should therefore be distinguished from the secondary comparisons. The sponsor-reported primary analysis is the central analysis described for the primary endpoint. Secondary p-values from summary-based recalculations are unadjusted, and they do not account for the number of outcomes examined. They can be useful for describing the posted estimates, but their evidential weight is limited when the overall testing strategy is not reported.

The primary analysis population is defined as randomized participants who received at least one dose of study drug and had a baseline value and at least one postbaseline HAMD-17 value. The registry record does not report the number excluded from that population, reasons for missing measurements, or how missing data were handled in the model. It also does not report sensitivity analyses that would show how the primary conclusion changes under alternative assumptions about missing observations. Those details matter because repeated-measures analyses depend on assumptions about the data available for estimation.

The record does not describe the complete statistical analysis plan, the covariates or repeated-measures structure used in the model, interim analyses, or whether any analysis decisions were made before or after examining outcomes. These items are not reported in the registry record; no particular procedure should be inferred. In general, prespecification helps readers assess whether analyses were selected independently of observed results, but the available record does not allow that assessment for these details.

Consequently, the primary comparison can be interpreted from its reported estimate and interval, while the robustness of the result to missing-data assumptions and the confirmatory status of secondary findings remain uncertain. This limitation does not itself show that the analysis is biased; it identifies information needed to assess the analysis more fully.

16. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: atrial fibrillation (0.7%), lower gastrointestinal haemorrhage (0.7%), multiple sclerosis (0.7%), gastritis haemorrhagic (0.7%).

The registry reports serious adverse events in 2 participants at risk in the pimavanserin group and 2 participants at risk in the placebo group. It reports 0 deaths in each group. These counts describe the serious adverse event information recorded for the trial; they do not by themselves establish whether an event was caused by study treatment. The registry record does not provide additional detail about event timing, severity criteria, or individual clinical circumstances.

The record lists atrial fibrillation, lower gastrointestinal haemorrhage, multiple sclerosis, and gastritis haemorrhagic among the top serious event terms, each reported as 0.7%. The terms are not linked to treatment arms in the posted summary, so they should not be assigned to a group based on this information.

The record also gives other adverse event figures of 54 among 148 at risk for pimavanserin and 43 among 150 at risk for placebo. These are separate from the serious-event counts in the registry summary. The record does not provide enough detail to assess event duration, recurrence, or the extent of follow-up for safety. Comparisons of event counts should be read descriptively, not as proof of a causal difference or of safety equivalence.

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

The primary result describes the difference between groups in HAMD-17 score change from baseline to Week 5. Both reported arm-level least-squares means are negative, indicating lower scores at Week 5 on average than at baseline. Since higher HAMD-17 totals denote more severe depression, lower scores represent a reduction on that measured scale. The between-group estimate is -0.9, and its 95% confidence interval includes zero.

This result does not establish that pimavanserin produced a specific benefit, nor does it establish that the groups were equivalent. It indicates that the estimated difference was uncertain under the reported analysis. The point estimate favors a greater reduction in the pimavanserin arm numerically, but the confidence interval also permits no difference and a difference in the opposite direction. The record does not provide a threshold for a clinically meaningful HAMD-17 difference, so the importance of the estimate cannot be judged by such a benchmark.

The finding is bounded by the trial's setting and outcome. It concerns adjunctive treatment in participants with major depressive disorder and an inadequate response to antidepressant treatment, using the stated analysis population and a Week 5 HAMD-17 change endpoint. It does not establish results for people outside that population, for other treatment settings, or for time points not represented by the primary endpoint. Detailed eligibility and background-treatment information are not reported in the registry record.

The trial's secondary measures and safety counts address other questions, but they do not transform the primary estimate into a broader claim. The numerical secondary summaries are mixed and unadjusted, while serious adverse event counts are descriptive. Interpretation of long-term outcomes, subgroup effects, or the relationship between symptom scores and other aspects of health would require information not reported in the registry record.

The appropriate conclusion is therefore limited: the posted primary analysis estimates a modest negative score difference at Week 5, with uncertainty that includes no between-group difference. The result alone does not prove efficacy, prove lack of efficacy, demonstrate equivalence, or settle the balance of benefits and harms. It is one trial result within the measures and follow-up reported in the registry.

18. Why This Trial Matters Statistically

This trial illustrates why a point estimate should be read together with its uncertainty. The primary estimate is in one numerical direction, but its reported confidence interval crosses zero. Focusing only on the sign or size of the point estimate would understate what the data leave unresolved. The interval communicates that the analysis does not sharply distinguish a between-group difference from no difference.

It also illustrates the distinction between a trial's model-based result and a recalculation from summary data. The sponsor reports a mixed-effects model for repeated measures, while the recalculation uses posted arm summaries and an approximate standard deviation. Their point estimates are similar, but the confidence limits and p-values differ. This makes clear why an approximate calculation can serve as a check without becoming a substitute for the prespecified or sponsor-reported analysis.

The primary outcome is a continuous score measured over time, and the analysis population has a specific definition based on treatment exposure and available HAMD-17 measurements. Those choices affect what the estimate describes. The registry does not provide all details needed to assess the model or missing-data assumptions, which shows why statistical interpretation depends not only on results but also on a transparent account of analysis methods.

The secondary outcomes provide a further lesson about multiplicity. Several domains were listed, and the numerical summaries that are posted do not all point in the same direction or have intervals with the same degree of uncertainty. Without a reported multiplicity strategy, isolated secondary findings should not be elevated above the primary result. Together, these features make the trial useful for understanding effect estimates, uncertainty, model-based analyses, and the limits of interpreting multiple outcomes from a registry record.

19. 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 primary endpoint addresses change through Week 5. It does not describe longer-term outcomes, and the registry record does not report a later primary assessment. That limits conclusions about whether any observed score changes persisted beyond the stated endpoint.

The record does not provide detailed eligibility criteria, the background antidepressant regimens, or information about how those regimens were managed. This limits assessment of how closely the trial population and treatment context match other clinical settings. The record identifies the study as multicenter but does not report site locations or the distribution of participants across sites.

The primary result is reported from a mixed-effects model for repeated measures, but the model specification, covariates, covariance structure, and missing-data assumptions are not reported in the registry record. The Full Analysis Set requires baseline and postbaseline HAMD-17 data, but counts and reasons for exclusion are not stated. Without those details, readers cannot fully assess the sensitivity of the estimate to missing observations or reproduce the sponsor analysis.

The registry record also does not report the original sample-size justification, any interim analyses, the full analysis plan, or a multiplicity procedure for the secondary outcomes. The sample-size illustrations based on the observed effect are not a substitute for the original design rationale. Secondary findings are therefore harder to classify as confirmatory or exploratory, and the unadjusted comparisons warrant caution.

Finally, the posted numerical secondary summaries cover only some of the listed outcomes, and the serious-event terms are not linked to treatment arms in the summary. The available safety counts do not establish causality, and the outcome summaries do not address every potential clinical domain. These limitations narrow what can be concluded from the registry record, without proving that any particular result is biased or incorrect.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Adjunctive Pimavanserin in Subjects With Major Depressive Disorder and Inadequate Response to Antidepressant Treatment — NCT03968159.
• Registry record last updated 2021-11-17; status last verified 2021-10.
• Results first posted 2021-11-17.
• 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.

23. Record Summary

The registry record describes a completed Phase 3, randomized, quadruple-masked, parallel-group, placebo-controlled study of adjunctive pimavanserin in participants with major depressive disorder and an inadequate response to antidepressant treatment. Pimavanserin was described as 34 mg tablets, with placebo tablets as the comparator. The record lists 298 enrolled participants.

The primary outcome was change from baseline to Week 5 in HAMD-17 total score. The sponsor-reported mixed-effects model for repeated measures estimated a least-squares mean difference of -0.9, with a 95% confidence interval from -2.5 to 0.8 and p = 0.296. The interval includes zero, so the primary comparison remains uncertain. The registry also contains secondary outcome and safety summaries, but the full analysis plan, missing-data details, multiplicity strategy, and original sample-size rationale are not reported in the registry record.