A Study of Cariprazine in the Prevention of Relapse of Symptoms in Participants With Schizophrenia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Cariprazine in the Prevention of Relapse of Symptoms in Participants With Schizophrenia (NCT01412060), a randomized phase 3 study. The primary outcome is Time From Baseline to the First Symptom Relapse During the Double-blind Phase.

2. Clinical Question

The study addressed relapse prevention rather than change in a single symptom score. Its primary endpoint was time from baseline to the first qualifying relapse, reported in days. The record reports the 25th percentile for time to relapse in each arm and a log-rank hazard ratio for the comparison. The endpoint timeframe was up to 34 weeks, with bi-weekly assessments thereafter until Week 92.

The main question is therefore about the timing of a first event under the two assigned treatment strategies. It is not a direct comparison of average symptom scores, nor does the endpoint count every relapse after the first. The registry describes a randomized, double-blind, placebo-controlled, parallel-group Phase 3 study. Its posted results are for the double-blind intent-to-treat population, not every participant enrolled in the study.

3. Trial Design

The registry describes a randomized, quadruple-masked, parallel-group Phase 3 study sponsored by Forest Laboratories. The study enrolled 765 participants and is listed as completed. The recorded study start was 2011-09-27, and primary completion was 2014-09-03. Its objective was to evaluate the efficacy and safety of cariprazine relative to placebo in preventing relapse of symptoms in participants with schizophrenia.

The study had an open-label phase followed by a double-blind treatment phase and safety follow-up. In the open-label phase, participants received cariprazine orally once a day for 20 weeks. The dose could be modified during the first 6 weeks, after which it was fixed at 3, 6, or 9 mg for the last 14 weeks. During the double-blind treatment phase, participants received either placebo or cariprazine at 3, 6, or 9 mg orally once a day for 26 to 72 weeks. The registry also identifies open-label, placebo and cariprazine safety follow-up groups.

The primary endpoint uses the double-blind phase, and the primary analysis population is specifically defined as participants who received at least one dose of double-blind investigational product and had at least one post-randomization PANSS or CGI-S assessment during that phase. This definition matters: the reported analysis does not simply represent everyone enrolled or everyone who started the open-label phase.

Although randomization, masking and the parallel-group model are recorded, further operational details are not reported in the registry record. In particular, details of sequence generation, allocation concealment and the procedures used to maintain masking are not reported in the registry record. The registry record also does not report the rationale for the enrollment target or a sample-size justification. Such omissions limit what can be inferred about the design process, but they do not alter the endpoint definition or the results recorded for the primary comparison.

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 765 participants were enrolled and received cariprazine in Run-in Phase; 751 of these, had at least 1 postbaseline Positive and Negative Syndrome Scale (PANSS) evaluation and 364 entered Stabilization Phase. Only 200 participants who completed Open-label phase, received either placebo (n=99) or cariprazine (n=101) in Double-Blind Phase.

Open-label Run-in Period (RIP)

Reasons for not completing this period:

Period Between RIP and SP

Reasons for not completing this period:

Open-label Stabilization Period (SP)

Reasons for not completing this period:

Period Between SP and DBP

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 received at least 1 dose of double-blind investigational product and who had at least 1 post-randomization assessment of PANSS or CGI-S during the Double-blind Treatment Phase.

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: Run-in Phase safety population: All participants who received at least 1 dose of open-label cariprazine during the Run-in Phase of the study.

6. Primary Endpoint

The registry lists the primary outcome as Time From Baseline to the First Symptom Relapse During the Double-blind Phase, measured over Up to 34 Weeks and Bi-Weekly thereafter until Week 92.

The primary endpoint was time from baseline to the first symptom relapse during the double-blind phase, measured in days. The registry reports the 25th percentile for time to relapse. A percentile describes a point in the event-time distribution; it is not the average time to relapse and is not a median. The registry timeframe was up to 34 weeks, with bi-weekly assessments thereafter until Week 92.

Relapse was defined as meeting at least one of several criteria. These included hospitalization due to worsening condition; an increase in PANSS total score of at least 30% for participants whose score was at least 50, or an increase of at least 10 points for participants whose score was below 50 at randomization; and an increase of at least 2 points in CGI-S score at Week 20. Other criteria were deliberate self-injury or aggressive behaviour, investigator-judged clinically significant suicidal or homicidal ideation, or a score greater than 4 on one or more specified PANSS items. The record says that a second assessment was not performed at the investigator's discretion.

PANSS is a 30-item rating scale, with each item scored on a 7-point scale and a total score ranging from 30 to 210; lower scores indicate fewer schizophrenic symptoms. CGI-S is a 7-point measure of illness severity compared with others with the same diagnosis; lower scores indicate less severe illness.

Time to first relapse is useful when the clinical question concerns how long participants remain without a qualifying event. It incorporates event timing rather than treating all participants as if follow-up ended at the same time. However, interpretation depends on the operational relapse definition, assessment schedule and handling of participants whose event status is not observed throughout follow-up. Details of those handling procedures are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a time-to-event outcome measure?

Time-to-event analysis considers both whether an event occurred and when it occurred. Here, the event was the first relapse meeting the registry's definition, and the time scale was days from baseline. Participants without a recorded relapse may contribute information until their observed follow-up ends. General: this approach can use information from participants whose observation periods differ, provided the statistical method and its assumptions are appropriate.

The registry gives the 25th percentile of time to relapse for each arm. A percentile is a point in the event-time distribution, not an arithmetic average. It should not be called a median: the median is a different percentile. The cariprazine estimate has no reported upper confidence limit, so the registry does not supply a bounded interval on both sides for that percentile.

What does the hazard ratio mean?

The registry's log-rank analysis reports a hazard ratio of 0.45, with a 95% confidence interval from 0.28 to 0.73. General: a hazard is the instantaneous rate of experiencing an event among participants who have not yet experienced it, at a given point in follow-up. A hazard ratio compares those rates between groups over time. It is not a ratio of cumulative risks, and it does not directly state how many days are gained or how many participants avoid relapse.

The reported value is below 1 for the cariprazine-versus-placebo comparison, which is consistent with a lower event hazard in the cariprazine group under the reported analysis. Interpretation of a single hazard ratio as a constant relative effect over follow-up relies on assumptions about how the relative hazards behave over time. Whether those assumptions were assessed, and whether the hazards were proportional, are not reported in the registry record.

What does the log-rank test assess?

The log-rank method compares time-to-event experience between groups across follow-up, taking account of the ordering of observed events and the risk sets at those event times. The registry reports p = 0.001 for this analysis. General: a p-value measures how compatible the observed data are with a specified null model under the test's assumptions. It is not the probability that the null hypothesis is true, nor is it a measure of clinical importance.

The confidence interval gives a range of values compatible with the data and method at the stated confidence level. It reflects sampling uncertainty under that analysis; it does not capture all sources of uncertainty, such as uncertainty from endpoint assessment, missing follow-up or departures from model assumptions. The interval from 0.28 to 0.73 excludes 1, the value representing equal hazards in the ratio scale.

How should the percentile estimates be read?

The placebo arm's reported 25th percentile is 92 days, with a confidence interval from 44 to 151 days. The cariprazine arm's is 224 days, with a lower confidence limit of 99 days and no upper confidence limit reported. These are separate estimates of event-time distribution points. The registry does not report a formal comparison between these percentiles, and a difference between their point estimates should not be treated as a separately tested result.

The registry's summary does not include enough arm-level event-time information to reproduce the analysis. It reports no event counts or censoring details. The posted inferential result is therefore the log-rank analysis in the record, rather than a new calculation from the percentile summaries.

Why can over-dispersion matter for count outcomes?

General: count outcomes often vary more between participants than a simple statistical model assumes. This extra variation is called over-dispersion. If ignored, standard errors may be too small and confidence intervals too narrow. Over-dispersion is relevant to some count-data analyses, but the primary endpoint in this study is time to first relapse, not a count of repeated events. The registry's result is presented as a log-rank hazard ratio, not a count-model estimate. No over-dispersion adjustment is reported for this endpoint.

What can be said about planning calculations?

General: a planning illustration based on an observed effect is not observed power and does not show the probability that the completed trial detected an effect. It answers a hypothetical design question under stated assumptions. The record describes a planning approach using a two-sided significance level of 0.05, equal allocation, a normal approximation and an effect set to a stated fraction of the observed effect; for time-to-event outcomes, the calculation concerns a total number of events. Those illustrations are not the trial's own sample-size justification. The registry does not report the original sample-size rationale, and no planning result should be interpreted as an observed trial outcome.

For this endpoint, the registry flags the arm-level values as missing or not estimable for a two-arm comparison of the count-other type. Consequently, no simple unadjusted recalculation of a comparable two-arm effect is available from those posted values. This does not replace or invalidate the registry's separately reported log-rank analysis.

10. Analyses Reported by the Sponsor

The registry reports the sponsor's log-rank analysis as a hazard ratio of 0.45, with a 95% confidence interval from 0.28 to 0.73 and p = 0.001. This is the principal inferential comparison recorded for the primary endpoint. The record does not flag a discrepancy between that analysis and a recalculation.

A simple unadjusted recalculation is not available for this endpoint from the posted arm-level summary values: the record identifies the values as missing or not estimable for a two-arm comparison of this type. The percentile estimates alone do not provide the event-time and censoring information needed to reproduce a log-rank analysis. The sponsor's log-rank result should therefore be read as the reported analysis, rather than as a result independently reconstructed from the posted percentile summaries.

Even where a recalculation is possible in other settings, it answers a narrower question based on summary statistics and may not reproduce an analysis using participant-level time-to-event information. Here, the registry record does not provide the information needed for such a comparison. The reported hazard ratio and confidence interval describe the sponsor's analysis; they are not a recalculated estimate.

The record does not report further detail on adjustment variables, assessment of proportional hazards, or how incomplete follow-up was handled. Those details would help readers assess the analysis assumptions and reproducibility, but their absence does not justify substituting an unsupported calculation for the reported result.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure and one analysis, and reports no secondary outcomes. This limits what can be said about the breadth of posted outcome testing. However, a single listed primary analysis does not establish that no other outcomes, time points or interim looks were examined. The registry record does not report a multiplicity adjustment, a hierarchy of hypotheses, or whether any interim analyses were conducted. Those details are not reported in the registry record, so the degree to which multiplicity was planned or controlled cannot be determined from the entry.

The primary analysis population is the double-blind intent-to-treat population: participants who received at least one dose of double-blind investigational product and had at least one post-randomization PANSS or CGI-S assessment during that phase. This definition is narrower than the overall enrollment description. It excludes, by definition, participants lacking either of those specified features. The registry does not report the reasons participants may have been excluded from this analysis population or how those exclusions affected the results.

Time-to-event analysis also requires a way to handle participants whose first relapse is not observed before their follow-up ends. The registry does not provide the number of events, the number censored, reasons for censoring, or the details of the censoring rules. It also does not describe how missing assessments were handled, beyond noting that a second assessment was not performed at investigator discretion for the relapse definition. The impact of missing outcome information and the assumptions used to address it are therefore not reported in the registry record.

The available results do not establish whether sensitivity analyses were conducted, whether the analysis population was altered in additional analyses, or how protocol deviations were handled. These elements are important for judging robustness, especially when the primary outcome depends on clinical assessments and time-to-event follow-up. Their absence from the registry record should be treated as an information limitation, not as evidence that a particular procedure was or was not used.

12. 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: schizophrenia (7.1%), psychotic disorder (2.0%), schizophrenia, paranoid type (2.0%), atrial fibrillation (1.0%), cholecystitis acute (1.0%).

The registry reports serious adverse events separately for the open-label phase, double-blind treatment groups and safety follow-up phases. In the open-label phase, 50 serious adverse events were reported among 765 at risk. During double-blind treatment, 14 were reported among 99 in the placebo group and 14 among 101 in the cariprazine group. In safety follow-up, 6 were reported among 765 in the open-label group, 2 among 99 in the placebo group and 0 among 101 in the cariprazine group. The record reports 0 deaths in each listed group.

These are counts associated with the registry's phase-specific safety groups, not rates adjusted for time under observation. The double-blind treatment period is described as lasting 26 to 72 weeks, so participants' time at risk may vary. The registry does not provide person-time, details about event overlap, or an adjusted comparative safety analysis. The serious-event counts should not be treated as a direct estimate of relative risk or as a test of a difference between groups.

The registry lists Schizophrenia (7.1%), Psychotic disorder (2.0%), Schizophrenia, paranoid type (2.0%), Atrial fibrillation (1.0%), Cholecystitis acute (1.0%) and Generalised tonic-clonic seizure (1.0%) among the top serious-event terms. These terms are reported as listed and are not a complete account of all adverse events. The safety findings are descriptive registry data and should be interpreted alongside the phase structure and limits of the reported information.

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

The posted primary result is a log-rank hazard ratio of 0.45 for cariprazine compared with placebo, with a 95% confidence interval from 0.28 to 0.73 and p = 0.001. The interval excludes 1, and the reported test provides evidence of a difference in time-to-first-relapse experience under the analysis. The reported point estimates for the 25th percentile were 92 days for placebo and 224 days for cariprazine; the cariprazine estimate has no reported upper confidence limit.

These results do not give the absolute probability of relapse at a particular time, the average time to relapse, or the number of events prevented. The hazard ratio is a relative time-to-event measure; it is not a risk ratio or a difference in days. The percentile estimates are not medians, and the registry does not report a formal test comparing those percentiles. The clinical significance of the estimates cannot be determined solely from the p-value.

The result applies to the registry's defined endpoint and analysis population. Relapse could be triggered by different clinical events or symptom-rating thresholds, and the record allows investigator discretion regarding a second assessment. The analysis population required both exposure to double-blind investigational product and at least one post-randomization PANSS or CGI-S assessment. The registry does not report the full event, censoring or missing-data information needed to evaluate how those features affected the estimates.

The study was randomized and double-blind, but the registry entry alone does not resolve every question about conduct, analysis assumptions or generalizability. The results do not guarantee a particular outcome for an individual and should not be interpreted as medical advice. They describe the statistical evidence recorded for this trial, not a universal effect across all participants or settings.

14. Why This Trial Matters Statistically

This trial illustrates why a relapse endpoint can be framed as time to first event rather than as a simple end-of-study score. Participants may have different lengths of observation, and event timing carries information. A time-to-event analysis can incorporate that timing, while the interpretation still depends on how relapse is defined, when assessments occur and how observation ending before an event is handled.

It also shows the distinction between a hazard ratio and an event-time percentile. The registry's log-rank hazard ratio summarizes a relative comparison over follow-up, whereas the 25th percentile describes a point in each arm's event-time distribution. These estimates are related to the same endpoint but do not answer identical questions. Neither can be converted into an absolute risk difference using the posted summaries alone.

The endpoint definition combines hospitalization, changes in PANSS or CGI-S scores, safety-related behaviours and investigator-judged ideation. This makes the operational definition central to interpretation: a change in criteria or assessment procedures could change which events qualify. The registry's note about investigator discretion for a second assessment is therefore relevant when considering reproducibility and outcome classification.

Finally, the record illustrates a common reporting limitation. It includes a sponsor-reported inferential analysis but not the event counts, censoring details or participant-level information needed to reproduce it. Clear reporting of analysis populations, missing-data procedures and model checks would enable a fuller appraisal. The posted confidence interval and p-value remain interpretable as the reported analysis, but they do not remove the need to consider these uncertainties.

15. Limitations

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

The registry record provides limited information for independently assessing the primary time-to-event analysis. It reports the hazard ratio, confidence interval and p-value, but not the numbers of relapses, censoring patterns, reasons for censoring, or follow-up duration by participant. The record does not report whether proportional-hazards assumptions were assessed or provide other diagnostic details. These omissions limit assessment of the model and the precision of the event-time comparison.

The relapse definition includes several distinct criteria, including symptom-score thresholds and clinical events. The registry notes that a second assessment was not performed based on investigator discretion. It does not report how often that discretion affected classification, how assessors were trained, or whether adjudication was used. Such details matter because event classification is part of the endpoint itself.

The posted primary analysis population requires at least one dose of double-blind investigational product and at least one post-randomization PANSS or CGI-S assessment. The record does not report participant flow into this population, reasons for not meeting the definition, or sensitivity analyses using other populations. Missing-data methods and handling of protocol deviations are not reported in the registry record.

The registry does not report the original sample-size justification, multiplicity plan, or interim-analysis procedures. It also does not give detailed methods for randomization and masking beyond identifying the study as randomized and quadruple-masked. Finally, the double-blind treatment period varied from 26 to 72 weeks, and the posted summary does not provide individual exposure time. These limitations do not negate the reported result, but they constrain independent evaluation and interpretation.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Cariprazine in the Prevention of Relapse of Symptoms in Participants With Schizophrenia — NCT01412060.
• Registry record last updated 2018-07-06; status last verified 2018-06.
• Results first posted 2018-07-06.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT01412060 is a completed Phase 3 randomized, quadruple-masked, parallel-group study of cariprazine versus placebo for prevention of relapse in participants with schizophrenia. The registry records enrollment of 765 participants and identifies Forest Laboratories as sponsor. The study began on 2011-09-27, with primary completion on 2014-09-03.

The primary endpoint was time from baseline to first symptom relapse during the double-blind phase, reported in days. The registry reports a 25th percentile of 92 days for placebo and 224 days for cariprazine, and a log-rank hazard ratio of 0.45 with a 95% confidence interval from 0.28 to 0.73 and p = 0.001. The cariprazine percentile has no reported upper confidence limit.

The analysis population required at least one dose of double-blind investigational product and at least one post-randomization PANSS or CGI-S assessment. The registry reports safety events by study phase but does not provide detailed event-time or censoring information for independently reproducing the primary analysis. No secondary outcomes are posted in the registry record.