A Cariprazine Study in the Prevention of Relapse in Bipolar I Disorder Patients Whose Current Episode is Manic or Depressive, With or Without Mixed Features: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Cariprazine Study in the Prevention of Relapse in Bipolar I Disorder Patients Whose Current Episode is Manic or Depressive, With or Without Mixed Features (NCT03573297), a randomized phase 3 study. The primary outcome is Time to First Relapse of Any Mood Episode During the Double-Blind Treatment Period.

2. Clinical Question

The trial’s stated purpose was to evaluate efficacy and safety in preventing relapse among participants with bipolar I disorder whose index episode was manic or depressive, with or without mixed features. It used a randomized withdrawal design: an open-label treatment period preceded double-blind treatment. This design asks about relapse after participants have entered the study’s treatment sequence; it is not simply a comparison of starting cariprazine or placebo in an unselected group at the beginning of the trial.

The two dose descriptions address distinct target-dose comparisons. The brief summary specifically links the 1.5 mg/day comparison to participants initially stabilized on a 3.0 mg/day target dose. Details of how participants qualified for randomization, and of the assignment process beyond the registry’s description of the study as randomized, are not reported in the registry record. The primary endpoint is time to first relapse, but its observed group estimates and the sponsor’s comparative result are not reported in the registry record. Consequently, the clinical question can be described, but the posted results do not allow a conclusion about the direction or precision of either comparison.

3. Trial Design

This was a phase 3, randomized, double-blind, parallel-group, multicenter clinical trial sponsored by AbbVie. The registry lists an enrollment of 901 and a completed status. The recorded start date is 2018-06-15, and the primary completion date is 2022-09-05. The official title describes the study as placebo-controlled and as using a dose-reduction paradigm in the prevention of relapse.

The treatment sequence included an open-label period followed by a double-blind treatment period. During open-label treatment, participants started cariprazine at 1.5 mg once daily, with a target dose of 3.0 mg once daily, for up to 16 weeks. The double-blind arms are described as cariprazine 3.0 mg once daily, cariprazine 1.5 mg once daily, and placebo once daily, each for up to 39 weeks. The registry’s brief summary says the 1.5 mg/day target-dose comparison was for participants initially stabilized on a 3.0 mg/day target dose. It does not report further details of eligibility for randomization or the dose-reduction procedures.

Randomized allocation means treatment assignment was determined by chance, while double masking means the study was designed to keep treatment assignment concealed from participants and relevant study personnel. The registry identifies a parallel model, in which assigned groups are followed as separate groups during the double-blind period. It does not report the randomization method, allocation concealment procedures, or details of who remained masked.

The primary endpoint timeframe is stated as from Week 16 to Week 55, while the double-blind period is described as lasting up to 39 weeks. The endpoint population is the Double-Blind Intent-to-Treat Population, defined by treatment exposure and post-randomization assessment or relapse. Although the registry reports overall enrollment, it does not report the number randomized to each arm in the primary analysis or provide the primary outcome results. The precise relation between overall enrollment and inclusion in each analysis should therefore not be inferred from the enrollment figure alone.

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: A total of 901 participants were enrolled; 5 participants did not enter the open label treatment period.

Open Label Treatment Period

Reasons for not completing this period:

Double-Blind Treatment Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Double-Blind Intent-to-Treat Population: All participants who took at least 1 dose of DB IP and had at least 1 post-randomization assessment of the YMRS, MADRS or CGI-S scores or relapsed during the 39-week DBTP of the study.

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: Safety Population: all participants who received at least 1 dose of study medication.

6. Primary Endpoint

The registry lists the primary outcome as Time to First Relapse of Any Mood Episode During the Double-Blind Treatment Period, measured over From Week 16 to Week 55.

The primary endpoint is time to first relapse of any mood episode during the double-blind treatment period. It is measured in days, with time calculated from randomization to the first relapse. Participants who did not meet relapse criteria were censored at completion or discontinuation from the double-blind treatment period. The registry says percentiles and their 95% confidence intervals are based on Kaplan-Meier estimates.

Relapse could be identified through several clinical or treatment events: a YMRS total score of at least 17, a MADRS total score of at least 20, a CGI-S score of at least 4, initiation of additional psychiatric medication, psychiatric hospitalization, or investigator judgment that the illness had exacerbated. The scales capture different aspects of mood symptoms and clinical status, while the other criteria capture a treatment change, hospitalization, or clinical judgment. Thus, the endpoint is a composite definition rather than a single symptom score.

Time-to-event endpoints retain information about when an event occurs, rather than recording only whether it occurred by a fixed final visit. Censoring allows participants without a recorded relapse to contribute follow-up until their completion or discontinuation, while acknowledging that their later relapse status is not observed in the study period. Kaplan-Meier estimates are commonly used to describe the event-time distribution under censoring. The registry record does not report the estimated median time to relapse, its confidence interval, or the comparative result for the treatment groups, so the endpoint definition alone does not establish how the groups differed.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Time-to-event methods address both whether an event occurred and when it occurred. In this trial the event is the first relapse that meets the registry’s composite criteria. The time origin is randomization, and the endpoint is expressed in days. This makes the analysis different from a simple comparison of symptom scores at a single visit: participants can have different follow-up times, and the timing of a first event contributes to the result.

What does censoring mean?

A participant who has not met the relapse criteria when follow-up ends does not have an observed relapse time. The registry treats that participant as censored at completion or discontinuation from the double-blind treatment period. In general, censoring means the event time is only known to extend beyond the last observed time; it does not mean that the participant experienced no relapse after leaving observation. A valid interpretation depends on suitable assumptions about censoring and on how much follow-up is available. The registry record does not report the number censored by group, reasons for discontinuation by group, or analyses assessing the effect of censoring.

How does Kaplan-Meier estimation help?

Kaplan-Meier estimation is a nonparametric method for describing the probability of remaining free of the event over time while accounting for censored observations. At each observed event time, the estimate reflects the participants still under observation and at risk immediately before that time. Censored participants contribute information up to their censoring time, but are not counted as having had the event. This produces an estimated event-free curve from which percentiles, including a median when it can be estimated, may be obtained.

A median time to event is the time by which the estimated event-free probability has reached the median point of its distribution. It is not an average event time, and it may not be estimable if the observed curve does not reach that point. The registry identifies the median and 95% confidence interval as the primary outcome measure, but it does not report their values. Readers therefore cannot assess whether a median was estimable for each group or how uncertain an estimate was.

What would a hazard ratio describe?

A hazard ratio compares the instantaneous event rates between groups over follow-up, conditional on participants remaining event-free up to each time. It is not a ratio of median times and does not directly state the probability that an individual participant will relapse. In general, interpreting a single hazard ratio relies on assumptions about how the relative event rates behave over time. The registry entries identify sponsor-reported hazard ratios as the source used for time-to-event comparisons, but do not include any hazard ratio, confidence interval, or associated test result. No comparative estimate can be reconstructed from the endpoint definition or adverse-event counts.

What can a confidence interval tell the reader?

A confidence interval represents statistical uncertainty around an estimate under the analysis method and its assumptions. A narrower interval indicates greater precision than a wider interval for the same type of estimate, but the interval does not capture every source of uncertainty, such as limitations in study conduct or outcome measurement. For a treatment comparison, the estimate and its interval need to be read together; a point estimate alone can conceal substantial uncertainty. The primary outcome’s 95% confidence intervals are not reported in the registry record, so their precision cannot be evaluated.

Why does the relapse definition matter?

The composite outcome combines scale thresholds with additional medication, hospitalization, and investigator judgment. Composite outcomes can capture different ways a clinically important event may present, but their interpretation depends on which components occurred and how often. A combined time-to-first-event result does not by itself reveal which component most often triggered relapse. Component-specific counts and analyses are not reported in the registry record. The threshold rules are part of the endpoint definition; the record does not provide enough outcome results to assess how the composite was distributed across components.

How should the planning illustrations be read?

The stated sample-size assumptions describe a planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. For time-to-event tables, the stated convention is that totals refer to events. Such illustrations are not the trial’s observed power and do not replace the study’s actual design justification. The observed treatment effect and numerical planning outputs are not included in the registry entries, and the trial’s own sample-size justification is not reported in the registry record. The illustration should therefore be treated as a hypothetical planning aid, not as a result of the trial or evidence that the trial was adequately powered.

General statistical interpretation also depends on the analysis population, missing outcome information, handling of protocol deviations, and prespecified analysis choices. The endpoint population is defined in the registry, but additional analysis details are not reported. Those omissions limit independent assessment of whether the reported methods, if any, were implemented as planned.

10. Multiplicity, Missing Data and Analysis Populations

The primary outcome is identified as time to first relapse of any mood episode, and the study description includes comparisons involving cariprazine 3.0 mg/day and 1.5 mg/day against placebo. Evaluating more than one dose comparison can raise multiplicity concerns: when several hypotheses are tested, the chance of at least one apparently positive result can increase unless the testing strategy accounts for the number and structure of comparisons. The registry record does not report a multiplicity adjustment, a testing hierarchy, or a prespecified strategy for controlling the overall type I error rate. It is therefore not possible to determine from this record how multiple comparisons were handled.

The endpoint definition specifies censoring at completion or discontinuation for participants without a qualifying relapse. It does not report the amount or pattern of missing outcome information, the reasons participants discontinued, or any sensitivity analyses addressing missingness and censoring. General statistical interpretation of censored time-to-event data depends on assumptions about how censoring relates to the event process; the record does not report assessments of those assumptions.

The registry defines the Double-Blind Intent-to-Treat Population as participants who took at least one dose of double-blind investigational product and had at least one post-randomization YMRS, MADRS, or CGI-S assessment, or relapsed during the double-blind treatment period. This is the stated primary analysis population definition. The record does not report the number of participants included in that population by randomized group, the number excluded, or results for other analysis populations. It also does not report how intercurrent treatment changes or protocol deviations were handled in the efficacy analysis. These gaps restrict assessment of how closely the analysis population and procedures corresponded to the randomized groups and the stated treatment question.

11. 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: bipolar i disorder (1.4%), appendiceal abscess (0.7%), appendicitis perforated (0.7%), coronavirus pneumonia (0.7%), sacral radiculopathy (0.7%).

The serious adverse event counts are presented for the open-label treatment period and for each double-blind group. In open-label treatment, 19 of 896 participants at risk had a serious adverse event, and 1 death was recorded. During double-blind treatment, the serious adverse event counts were 5 of 145 in the placebo group, 7 of 144 in the cariprazine 1.5 mg QD group, and 3 of 147 in the cariprazine 3.0 mg QD group. The corresponding death counts were 1, 0, and 0. These are counts within the stated groups and periods, not estimates of causal effects.

The record also reports other adverse events: 256 of 896 participants in open-label treatment, 27 of 145 in the placebo group, 34 of 144 in the cariprazine 1.5 mg QD group, and 34 of 147 in the cariprazine 3.0 mg QD group. Serious adverse events and other adverse events are distinct categories in the record and should not be added together. The listed serious event terms and percentages include bipolar I disorder at 1.4%, with appendiceal abscess, perforated appendicitis, coronavirus pneumonia, sacral radiculopathy, and bipolar disorder each at 0.7%.

These figures do not show event timing, duration, severity beyond the serious-event classification, or whether events were judged treatment-related. The open-label and double-blind periods also differ in treatment context and have different group denominators, so their counts should be read separately. Without further comparative analyses, the figures do not establish that a treatment caused or prevented an adverse event.

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

The registry record establishes the study question, the treatment sequence, the relapse criteria, and the stated primary analysis population. It also reports safety counts for open-label treatment and the double-blind groups. Those details help explain what was intended to be measured and what kinds of events were recorded.

The record does not include the primary time-to-relapse results. It gives no group-specific median estimates, confidence intervals, hazard ratios, or comparative test results. As a result, the posted information cannot establish whether the groups differed in time to first relapse, the size or direction of any difference, or the precision of a treatment comparison. The fact that a trial is completed does not itself reveal its outcome.

The safety counts should be understood as descriptive counts, not as proof of a causal relationship. The registry entries do not provide enough information to determine whether an adverse event was attributed to study treatment, and the open-label period should not be treated as equivalent to the randomized double-blind comparison. Nor can the primary efficacy question be answered by comparing adverse-event counts.

The endpoint is a composite, so even a reported overall time-to-relapse estimate would not, by itself, explain which relapse criteria contributed most. Component-specific results are not reported in the registry record. Further, the record does not report the analysis details needed to assess multiplicity, missingness, or sensitivity to censoring. Those omissions matter when interpreting the certainty and scope of a result.

Accordingly, the defensible reading is limited: the trial was designed to evaluate relapse prevention in a defined bipolar I disorder population using randomized, double-blind treatment after an open-label phase, and the registry lists adverse-event counts. The registry information does not support a conclusion that cariprazine did or did not change relapse timing, nor does it establish a comparative safety conclusion. Such conclusions would require the relevant outcome estimates, uncertainty intervals, and analysis details.

13. Why This Trial Matters Statistically

This trial is statistically notable for pairing a time-to-event primary endpoint with a randomized withdrawal design. Time-to-first-relapse analysis uses the timing of events and accommodates participants whose event status is not observed through the end of follow-up by censoring them at completion or discontinuation. That approach makes the handling of follow-up and censoring central to interpretation, rather than treating every participant as if they had the same observed time.

The endpoint is also composite. It combines rating-scale thresholds, additional psychiatric medication, hospitalization, and investigator judgment. Such a definition can capture several forms of relapse, but a single overall result may conceal variation among components. Understanding the clinical and statistical meaning of a composite therefore benefits from component-level reporting, which is not included in the registry entries.

The dose comparisons and the stated sequence from open-label treatment to double-blind treatment create a further interpretive distinction: the question concerns relapse during double-blind follow-up after an earlier treatment period, not simply the effect of initiating treatment in all enrolled participants. The registry’s defined analysis population and dose descriptions are important context for understanding what population the primary result would represent.

Finally, the registry presents a completed phase 3 trial but does not report the primary time-to-relapse estimates or comparative results in the available record. Statistical design and endpoint definitions alone cannot substitute for effect estimates and uncertainty. The missing results prevent assessment of the magnitude, direction, and precision of the principal comparison, while the absence of multiplicity and missing-data details further limits appraisal. This distinction between a clearly described question and unavailable outcome evidence is essential when reading trial registries.

14. 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 main limitation for interpretation is that the registry record does not report the primary outcome estimates. Although the endpoint, timeframe, analysis population, and Kaplan-Meier method are described, the median time-to-relapse values and their 95% confidence intervals are absent. The sponsor’s hazard ratios and comparative statistical results are also not reported in the registry record. Without these results, the central efficacy question cannot be evaluated.

The available description does not report the number of primary-analysis participants by randomized group, the number of relapses, the number censored, or the reasons for discontinuation. It also does not report whether the relapse criteria were adjudicated or provide results by component of the composite endpoint. These omissions make it difficult to assess the completeness and composition of endpoint ascertainment.

Details of the randomization procedure, allocation concealment, masking maintenance, and dose-reduction procedures are not reported in the registry record. The record also does not report the trial’s own sample-size justification, a prespecified multiplicity strategy, or sensitivity analyses for missing data and censoring. The planning assumptions described for illustrative sample-size calculations do not fill these gaps or establish the trial’s observed power.

Safety information is limited to event counts and selected serious-event terms. The record does not provide treatment attribution, event timing, exposure-adjusted comparisons, or further detail on severity and outcomes. Open-label and double-blind periods have different contexts and should not be pooled for interpretation. These limitations mean that the listed counts are descriptive and cannot independently establish comparative safety.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Cariprazine Study in the Prevention of Relapse in Bipolar I Disorder Patients Whose Current Episode is Manic or Depressive, With or Without Mixed Features — NCT03573297.
• Registry record last updated 2023-09-28; status last verified 2023-09.
• Results first posted 2023-09-28.
• 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.

18. Record Summary

This completed phase 3, randomized, double-blind, parallel-group, multicenter trial evaluated cariprazine in preventing relapse among people with bipolar I disorder whose current episode was manic or depressive, with or without mixed features. The study included an open-label treatment period followed by double-blind treatment with cariprazine 1.5 mg QD, cariprazine 3.0 mg QD, or placebo. The registry lists enrollment of 901.

The primary outcome was time to first relapse of any mood episode during the double-blind treatment period, defined using rating-scale thresholds and additional clinical or treatment events. Kaplan-Meier estimates were specified for percentiles and confidence intervals. The registry record does not report the primary outcome estimates or comparative results, so the efficacy question remains unanswered by the information.

Safety counts are reported separately for open-label treatment and the double-blind groups. They are descriptive and do not establish causation or comparative safety. Details about multiplicity, the trial’s sample-size justification, and several aspects of missing-data and analysis procedures are not reported in the registry record.