Bipolar Maintenance Study of Lurasidone Adjunctive to Lithium or Divalproex: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bipolar Maintenance Study of Lurasidone Adjunctive to Lithium or Divalproex (NCT01358357), a randomized phase 3 study. The primary outcome is Time to Recurrence of Mood Event During the Double Blind Treatment Phase.

2. Clinical Question

This was a maintenance-treatment question: the endpoint focused on how long participants remained without a defined mood event during follow-up, rather than only on a change in symptom score at a single visit. A qualifying event could be a manic, mixed manic, hypomanic, or depressive episode, but the registry definition also included certain treatment interventions, psychiatric hospitalization, specified rating-scale thresholds at repeated assessments, or discontinuation because of a mood event.

The record reports a hazard ratio of 0.71 for lurasidone versus placebo, with a 95% confidence interval from 0.49 to 1.04 and a reported p-value of <0.078. The point estimate is below 1, while the interval includes 1. The posted results therefore do not pin down the size of any difference with high precision. The registry reports a placebo median time of 207 days; a lurasidone median is not reported in the registry record.

3. Trial Design

The registry describes a completed phase 3, randomized, double-blind, placebo-controlled, flexible-dose, parallel-group study. It was multicenter. The registry categorizes masking as triple, but the identities of the masked parties are not reported in the registry record. Enrollment was 965. The study began in 2011-06, and primary completion was in 2015-04.

Participants were assigned to lurasidone 20-80 mg flexible dose or placebo, alongside lithium or divalproex, for maintenance treatment of bipolar I disorder. The primary endpoint was assessed during a double-blind treatment phase with a stated timeframe of 28 weeks (up to 33 weeks). The registry describes the study population as including people with or without rapid cycling and/or psychotic features. The precise dosing schedule, rules for dose adjustment, and detailed background-treatment procedures are not reported in the registry record.

The primary efficacy analysis used the intent-to-treat population defined in the registry as all randomized participants who received at least one dose of study medication in the double-blind phase. Participants were analyzed according to randomized assignment. The primary results list 246 participants in the lurasidone group and 250 in the placebo group. These analysis counts are not the same as total enrollment; the registry does not provide a full account of how all enrolled participants were distributed across analysis populations.

The registry lists 14 outcome measures and 13 analyses. It names several secondary measures, including recurrence, discontinuation, symptom-rating scales, and clinical-global-impression scores. Details of the original sample-size justification, the planned statistical analysis strategy beyond the reported model names, interim analyses, and stopping rules are not reported in the registry record.

4. Participant Flow and Accounting

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

Before assignment: There were 2 phases in this study. In Phase 1 (Open-label Stabilization Phase), there was 1 reporting group. In phase 2 (Double-blind Maintenance Phase), there were 2 reporting groups. 7 subjects completed the open-label but were not randomized DB(3- mood episode,3-not meeting the criteria for the DB phase,and 1-insufficient clinical response)

Open Label Phase

Reasons for not completing this period:

Double-Blind Phase

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT (Intent to treat) population: all subjects who were randomized and received at least one dose of study medication in the double-blind phase. Subjects were analyzed based on the treatment they were randomized.

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 Time to Recurrence of Mood Event During the Double Blind Treatment Phase, measured over 28 weeks (up to 33 weeks).

The primary endpoint was time to recurrence of a mood event during the double-blind treatment phase. It measures the duration from the relevant study baseline until a qualifying event, with participants who have not experienced an event contributing information about how long they remained event-free. The registry reports this endpoint using a median and a Cox-model hazard ratio.

The event definition was broader than a formal episode diagnosis alone. It included meeting DSM-IV-TR criteria for a manic, mixed manic, hypomanic, or depressive episode; specified treatment intervention for symptoms; psychiatric hospitalization for a bipolar mood episode; rating-scale thresholds at two consecutive assessments no more than 10 days apart; or study discontinuation because of a mood event as determined by the investigator. The specified rating thresholds were a YMRS or MADRS total score at least 18 or a CGI-BP-S score at least 4.

General statistical context: time-to-event outcomes can retain information about both whether an event occurred and when it occurred, rather than reducing follow-up to a single yes-or-no outcome. They can also accommodate participants with differing observed follow-up by accounting for event-free time until the end of observation. The registry record does not report detailed censoring rules or the handling of particular follow-up situations.

7. Results Posted in the Registry

• Lurasidone 20-80 mg Flexible Dose vs Placebo: hazard ratio 0.71 (95% CI 0.49–1.04), reported p < 0.078 (as reported in the registry).

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.

Time to Recurrence of a Manic, Mixed Manic, Hypomanic, or Depressed Episode

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Cox Proportional hazard Model); it was not recalculated.
• Result:
  • Lurasidone 20-80 mg Flexible Dose vs Placebo: hazard ratio 0.72 (95% CI 0.48–1.08), reported p = 0.113 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Lurasidone 20-80 mg Flexible Dose vs Placebo and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Subjects Who Experience a Recurrence of a Manic, Mixed Manic, Hypomanic, or Depressed Episode

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Lurasidone 20-80 mg Flexible Dose: 16.7% (41/246) versus 21.6% (54/250) on control; risk difference -4.9 percentage points (95% CI -11.8 to 2.0), risk ratio 0.77 (0.54–1.11); unadjusted two-proportion z-test p = 0.163.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lurasidone 20-80 mg Flexible Dose and the control. The interval gives the range of effects that remain compatible with the data.

Change From Double-blind Baseline to Week 28 (LOCF) in CGI-BP-S Overall 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:
  • Lurasidone 20-80 mg Flexible Dose: mean 0.40 versus 0.49 on control; difference -0.09 (95% CI -0.33 to 0.15), standardized effect d = -0.07; z-test p = 0.454.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lurasidone 20-80 mg Flexible Dose and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover several aspects of maintenance treatment, including time to recurrence, whether recurrence occurred, discontinuation, symptom ratings, and clinical-global-impression scores. The posted secondary findings do not form a uniformly conclusive pattern. The time-to-recurrence estimate is directionally similar to the primary time-to-event estimate, while its confidence interval also includes 1. The other detailed secondary outcomes use different outcome scales and analysis approaches, so they are not interchangeable confirmations of the primary endpoint.

The registry lists 13 secondary analyses, but detailed numerical results are not reported in the registry record for every named secondary measure. The posted recurrence percentage and continuous score findings should be understood as separate outcomes with their own measurement and modeling choices, rather than as substitutes for the primary time-to-event endpoint. The primary outcome asks when a qualifying event occurs; a percentage outcome asks whether recurrence occurred over its stated period; a symptom-score outcome tracks a scale-based change.

Secondary outcomes are often useful for context, but multiple outcomes create opportunities for chance findings when each is assessed separately. Unless a prespecified multiplicity adjustment or hierarchy controls the overall false-positive risk, nominal or unadjusted results carry less confirmatory weight. The registry record does not report a multiplicity procedure or the hierarchy used to interpret these outcomes. Therefore, secondary findings can inform the overall pattern, but should not overturn or independently settle an uncertain primary result.

The record also does not report full details for the analyses of all listed secondary measures, including whether every planned comparison was reported. This restricts assessment of the complete secondary-outcome pattern and the role of selective reporting. The available results are best treated as supportive descriptive evidence, with the primary endpoint remaining the main comparison specified in the registry results.

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

• Time to All-cause Discontinuation
• Time to Recurrence of a Manic, Mixed Manic, Hypomanic, or Depressed Episode
• Percentage of Subjects Who Experience a Recurrence of a Manic, Mixed Manic, Hypomanic, or Depressed Episode
• Change From Double-blind Baseline to Week 28 (LOCF) in CGI-BP-S Overall Score
• Change From Double -Blind Baseline to Week 28 (LOCF) in CGI-BP-S Mania Score
• Change From Double-blind Baseline to Week 28 (LOCF) in CGI+-BP-S Depression Score
• Change From Double-blind Baseline to Week 28 (LOCF) in YMRS Total Score
• Change From Double-blind Baseline to Week 28 (LOCF) in MADRS Total Score

9. Statistical Interpretation

10. Statistical Methods Explained

The primary analysis used a Cox proportional hazards model, as reported by the sponsor. General statistical explanation: this model compares event hazards over time while using information from participants who have not yet experienced the event. Its hazard ratio is a relative measure of the event rate at a given point in follow-up among participants still at risk. It is not a risk ratio, an odds ratio, or a difference in median survival time.

What the hazard ratio conveys

The reported hazard ratio of 0.71 compares lurasidone with placebo. A value of 1 would represent equal estimated hazards under the model; values below or above 1 indicate a lower or higher estimated hazard for the first group, respectively. This is a general interpretation of the measure, not proof that the difference is causal in every context. Randomization supports a comparison of assigned groups, while the estimate remains subject to sampling uncertainty and the study’s analysis choices.

The proportional hazards model’s name refers to an assumption about how the relative hazards behave over time. The registry record names the model but does not report whether that assumption was assessed, how it was assessed, or what diagnostic results were obtained. The detailed model specification, covariates, stratification, and censoring procedures are also not reported in the registry record. Those omissions limit independent appraisal of the modeling choices.

How to read the confidence interval and p-value

A confidence interval gives a range of values reasonably compatible with the data and the model at the stated confidence level. The primary 95% interval runs from 0.49 to 1.04. Because it includes 1, the results do not rule out equal hazards at the interval’s level of uncertainty. The reported p-value is <0.078. A p-value is a measure of how incompatible the observed data are with a specified null model, assuming the model and analysis conditions; it is not the probability that the null hypothesis is true, nor is it a measure of clinical importance.

General statistical context: a point estimate alone can obscure uncertainty. An estimate below 1 may look directional, but its interval describes the range of effects the data can support with the chosen method. Here, the interval extends on both sides of 1, so a single directional reading would overstate what the analysis establishes.

Median time and event definition

The registry reports a placebo median time of 207 days, but no lurasidone median is reported in the registry record. The median is the time by which half of the relevant group has experienced the event in the analyzed time-to-event distribution, subject to the analysis method and follow-up. A median by itself does not describe the full shape of event timing or the comparison between groups. The reported hazard ratio is the available comparative summary.

The event definition combines episode criteria with treatment interventions, hospitalization, rating-scale thresholds, and investigator-determined discontinuation for a mood event. This broad definition can capture several ways recurrence may be recognized in the trial. It also means that the endpoint is not limited to a single diagnostic assessment. The registry provides the definition but does not report the number of primary events by group or further information on how each event type contributed.

Other statistical summaries

The secondary recurrence endpoint also used a Cox proportional hazards model, with a reported hazard ratio of 0.72 and a 95% confidence interval from 0.48 to 1.08. Another secondary endpoint reports recurrence as a percentage. General statistical context: an unadjusted difference in percentages is a simple comparison of observed proportions and does not account for covariates or other features of the analysis. A continuous secondary endpoint reports least-squares means, which are model-based group summaries; its sponsor analysis used ANCOVA.

Recalculated estimates for some non-primary outcomes are simple, unadjusted calculations based on posted summary statistics. They do not reproduce every feature of the sponsor’s model. For the continuous CGI-BP-S outcome, the recalculated interval and p-value differ from the sponsor’s ANCOVA results, so the sponsor’s own analysis is the appropriate reference for that outcome. The registry does not report enough model detail to independently reproduce the sponsor’s adjusted analyses.

The registry lists 13 analyses, but the original statistical plan, multiplicity procedure, missing-data rules beyond the explicitly named LOCF endpoint, and interim-analysis arrangements are not reported in the registry record. These are important for understanding how outcome selection, repeated testing, and incomplete observations may affect interpretation.

11. Confidence Intervals and Precision

Precision is conveyed chiefly by the primary 95% confidence interval, which spans 0.49 to 1.04 around a hazard ratio of 0.71. General statistical explanation: a narrower interval indicates that the data constrain the estimate more tightly, while a wider interval leaves more plausible values. This interval covers values consistent with a lower hazard in the lurasidone group and values close to, or slightly above, equal hazards.

The interval includes 1, the reference value for equal hazards. That means the primary result does not provide a precise distinction between a reduction in hazard and no difference. It does not mean every value in the interval is equally likely, and it does not establish that the groups are identical. The reported p-value, <0.078, also does not specify a more exact value in the registry record.

The placebo median is reported as 207 days, whereas a lurasidone median is not reported in the registry record. This prevents a direct comparison of the two medians from the posted primary results. No confidence interval for the lurasidone median is given. The hazard-ratio interval remains the available measure of comparative precision.

Precision also depends on event information, follow-up, and the analysis model, not only on the number enrolled. The primary analysis included 246 lurasidone participants and 250 placebo participants, but the registry does not report the primary event counts or detailed censoring information. Without those details, the sources of uncertainty cannot be separated further. The secondary time-to-recurrence estimate likewise has an interval that includes 1, from 0.48 to 1.08, indicating that its relative effect is also uncertain.

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 illustrations are planning scenarios based on an effect fraction of the observed effect; they are not the trial’s original design justification. For time-to-event outcomes, the figures are total events, not participants per arm. For other outcomes, the figures are participants per arm. The assumptions used for the illustrations are a two-sided significance level of 0.05, equal allocation between 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 are not reported in the registry record.

The 80% power column describes the planned number of events, or participants per arm as applicable, under which the specified scenario would have an 80% chance of detecting the assumed effect using those planning assumptions. The 90% power column gives the corresponding planning quantity for a 90% chance. These are prospective planning interpretations, not descriptions of what happened in the trial. This is not observed power.

Rows that plan for 100% of the observed effect use the full observed effect as the target for the illustration. Rows planning for 75% or 50% use smaller target effects. Smaller effects are harder to distinguish from random variation, so the illustrations require more events or more participants per arm. These scenarios can help readers see how planning needs change with the assumed effect size, but they do not show that the trial was designed around those assumptions or that the listed targets were reached.

For the primary time-to-event analysis, the relevant planning quantity is total events. For binary and continuous outcomes, it is participants per arm. Reading the units correctly is essential: a total-event figure should not be interpreted as an enrollment target, and a per-arm figure should not be interpreted as a combined two-arm total.

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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 0.710, so ln(HR) = -0.3425.
2. d = 4 × (1.96 + 0.842)² / [ln(0.710)]² = 267.7, rounded up to 268 events in total 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
hr <- 0.71                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(1358357)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 268 (80% power), 359 (90% power)
 75% of observed effect: events = 476 (80% power), 637 (90% power)
 50% of observed effect: events = 1,071 (80% power), 1,434 (90% power)
Simulated power with 268 events: 0.782 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The primary hazard ratio, confidence interval, and reported p-value are sponsor-reported results from a Cox proportional hazards model: 0.71, 0.49 to 1.04, and <0.078. The time-to-event result is not independently recalculated from the posted summaries; the registry’s reported analysis is the relevant estimate. The median time reported for placebo is 207 days, while the lurasidone median is not reported in the registry record.

For the secondary recurrence time-to-event outcome, the sponsor reported a hazard ratio of 0.72, a 95% confidence interval from 0.48 to 1.08, and p = 0.113. This is also a Cox-model result. It should be interpreted using its own interval and analysis rather than treated as a recalculation of the primary outcome.

For the continuous CGI-BP-S Overall Score outcome, the sponsor reported an ANCOVA least-squares mean difference of -0.09, with a 95% confidence interval from -0.29 to 0.12 and p = 0.406. A separate unadjusted recalculation uses posted summary statistics and yields a different interval and p-value. That difference reflects the distinction between the sponsor’s model-based analysis and a simple calculation from summaries; the sponsor’s own analysis takes precedence for interpreting the trial’s reported result.

The registry does not provide enough detail to reconstruct the sponsor’s full model specifications. The methods named for these analyses are reported, but covariate choices and other implementation details are not reported in the registry record.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists 14 outcome measures and 13 analyses, including a primary time-to-event endpoint and multiple secondary outcomes. Testing several outcomes raises a multiplicity issue: even when each test is evaluated separately, the chance of at least one apparently notable result can increase across a family of tests. General statistical practice may address this through adjustment, a prespecified hierarchy, or a clear distinction between confirmatory and exploratory outcomes.

The multiplicity procedure, if any, and the planned hierarchy of outcomes are not reported in the registry record. It is therefore not possible to determine from the posted information whether the reported secondary p-values were adjusted, how their confirmatory status was defined, or whether the overall type I error rate was controlled. Secondary findings should be interpreted cautiously in the absence of those details.

The primary analysis population is described as all randomized participants who received at least one dose during the double-blind phase, analyzed according to randomized assignment. This is the registry’s stated population definition; it is not an all-randomized definition because receipt of at least one dose is required. The record does not report a complete accounting of exclusions or missing endpoint data within this population.

The CGI-BP-S Overall Score outcome is described as change from baseline to Week 28 using LOCF. LOCF carries the last observed value forward when a later value is unavailable, but the registry record does not report how much missing data occurred, why data were missing, or whether alternative missing-data analyses were performed. Other missing-data methods and the detailed handling of incomplete time-to-event observations are not reported in the registry record.

Interim analyses, stopping rules, and the original statistical analysis plan are also not reported in the registry record. Their absence from the record does not establish that such procedures were or were not used; it means their role cannot be evaluated from the posted information.

16. 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 3% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: mania (1.2%), depression (1.0%), suicidal ideation (0.8%), bipolar 1 disorder (0.8%), cholelithiasis (0.4%).

In the safety results, 13 of 246 participants in the lurasidone group and 11 of 250 in the placebo group were reported as having a serious adverse event. The registry also reports other adverse events affecting 81 participants in each of those groups. Across all subjects, 41 of 962 were reported with a serious adverse event and 474 of 962 with other adverse events. These are counts among the listed groups; they should be read as descriptive safety summaries, not as adjusted comparisons.

The serious-event terms listed among the most frequent include Mania at 1.2%, Depression at 1.0%, Suicidal Ideation at 0.8%, Bipolar 1 Disorder at 0.8%, Cholelithiasis at 0.4%, and Appendicitis at 0.4%. The registry excerpt does not provide group-specific counts for these listed terms, so they should not be assigned to a particular treatment group based on this information.

Deaths are not reported in the registry record. Differences in counts alone do not establish a treatment-related cause or a definitive safety difference. The registry results do not include enough detail about event severity, timing, exposure, or statistical comparisons to support broader conclusions about comparative safety.

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

The posted primary result is a comparative estimate of time to a defined mood event during the double-blind phase. Its point estimate is below 1, but the 95% confidence interval includes 1. The result therefore leaves uncertainty about whether the underlying event hazard differed between groups and about the size of any difference. It does not establish that the groups were identical, and it does not establish a precise reduction in recurrence risk.

The endpoint incorporates several types of events: diagnostic episode criteria, treatment interventions, hospitalization, scale thresholds at repeated assessments, and discontinuation attributed to a mood event by the investigator. This makes the outcome broader than a single symptom score or a diagnosis recorded at one assessment. The resulting estimate applies to this registry-defined endpoint and the analyzed trial population over the stated follow-up, not to every possible definition of recurrence.

The registry reports a placebo median time but not a lurasidone median. The hazard ratio is the primary comparative summary available in the posted results. It should not be translated into an absolute probability, a number of days gained, or an individual prediction without additional data and a suitable analysis.

These results do not answer every question about longer-term outcomes, other treatment settings, or people outside the study population. The registry record does not report all details needed to assess the study’s original design assumptions, multiplicity handling, or missing-data sensitivity. Such limits matter when judging how far the reported estimate can be generalized.

General statistical context: an uncertain estimate is not the same as evidence of no effect. It means the data and analysis do not sharply distinguish among several plausible values. The appropriate reading is to retain both the direction of the point estimate and the uncertainty expressed by its interval, without converting either into a categorical claim about treatment.

18. Why This Trial Matters Statistically

This trial illustrates why time-to-event results require more than reading the point estimate. The primary hazard ratio is below 1, yet its confidence interval extends across the reference value of 1. The point estimate and interval together convey both the direction suggested by the data and the uncertainty that remains. A conclusion based on the point estimate alone would omit an important part of the result.

The endpoint also shows how clinical event definitions shape statistical outcomes. Recurrence included formal episode criteria but also specified treatment actions, hospitalization, rating-scale thresholds, and investigator-determined discontinuation. Such a composite definition can capture clinically relevant developments in several ways, while making it important to understand exactly what counted as an event.

The distinction between a sponsor’s adjusted analysis and a simple recalculation is another useful statistical lesson. A summary-statistic recalculation can be informative, but it does not reproduce a model such as ANCOVA or the full details of a Cox analysis. When the estimates differ because the methods differ, the sponsor’s reported analysis is the primary source for the trial’s stated result.

Finally, the registry lists multiple outcomes, while the multiplicity procedure and original analysis plan are not reported. This limits the strength of conclusions that can be drawn from secondary findings. The study is therefore a useful example of reading an estimate, its interval, the endpoint definition, and the analysis context together rather than treating a single statistic as the whole result.

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 registry’s primary results provide a hazard ratio and confidence interval, but do not report primary event counts by group, detailed censoring rules, or the full Cox model specification. The lurasidone median time is not reported, while the placebo median is reported as 207 days. These omissions prevent a fuller reconstruction of event timing and the analysis.

The original sample-size justification, detailed statistical analysis plan, any interim analyses, stopping rules, and multiplicity procedure are not reported in the registry record. The sample-size illustrations are planning scenarios based on the observed effect and specified assumptions, not evidence of the trial’s original design choices. The registry also does not report whether the proportional hazards assumption was assessed.

The primary population excludes randomized participants who did not receive at least one dose during the double-blind phase, according to the stated definition. A complete participant flow and detailed reasons for missing endpoint data are not reported in the registry record. LOCF is named for the CGI-BP-S Overall Score outcome, but the extent and pattern of missingness and sensitivity analyses are not reported.

The study used a broad mood-event definition, with more than one route for an event to qualify. The registry does not give a breakdown of primary events by event type. Safety results are also limited: deaths are not reported, and the listed serious adverse-event terms are not accompanied by group-specific counts. These limits constrain interpretation of both efficacy and safety beyond the reported summaries.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Bipolar Maintenance Study of Lurasidone Adjunctive to Lithium or Divalproex — NCT01358357.
• Registry record last updated 2016-09-07; status last verified 2016-07.
• Results first posted 2016-07-26.
• 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

NCT01358357 is a completed phase 3 randomized, parallel-group, placebo-controlled study of flexible-dose lurasidone adjunctive to lithium or divalproex for maintenance treatment of bipolar I disorder. Enrollment was 965, and the registry categorizes masking as triple. The primary endpoint was time to recurrence of a defined mood event during 28 weeks (up to 33 weeks) of double-blind treatment.

The primary Cox-model result was a hazard ratio of 0.71, with a 95% confidence interval from 0.49 to 1.04 and reported p <0.078. The interval includes 1, so the estimate remains uncertain. The registry reports a placebo median time of 207 days; a lurasidone median is not reported in the registry record. Details of the original design assumptions, multiplicity handling, and several analysis procedures are not reported in the registry record.