Bipolar Disorder in Late Life: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bipolar Disorder in Late Life (NCT00787930), a registered phase not stated study. The primary outcome is Boyko DWM Hyperintensity Value >2 in Subjects Who Received Acute Treatment for Mania.

2. Clinical Question

3. Trial Design

This is a completed study listed as “Bipolar Disorder in Late Life.” Its official title describes an evaluation of MRI hyperintensities and diffusion tensor imaging in bipolar disorder. The listed condition is bipolar disorder, the sponsor is Duke University, and enrollment is 19. The study start is recorded as 2005-10 and primary completion as 2010-09. The registry record does not report further details about recruitment, eligibility criteria, or follow-up after the stated primary completion.

The study description focuses on structural brain changes in people with bipolar disorder or a history of the condition. The intervention is listed as valproic acid. For the acute-mania endpoint, the description refers to treatment for 3 weeks using the STEP-BD protocol, with valproic acid as the primary intervention. Arm descriptions refer to naturalistic treatment starting with valproic acid and characterize participants using Boyko deep white matter ratings and a week 3 YMRS criterion. These descriptions do not establish that participants were assigned randomly to distinct treatment strategies.

The registry record explicitly cautions that this is not a randomized comparison and that the arms are not a like-for-like control comparison. The allocation method is not stated. As a result, any contrast between the listed groups would need to be interpreted as a comparison of defined participant groups, not as an estimate of the effect of assigning one treatment rather than another.

Five outcome measures and five analyses are listed. The primary outcome is a participant count based on a Boyko deep white matter hyperintensity threshold. Secondary outcomes include a Boyko subcortical hyperintensity value and fractional anisotropy measures for the left and right orbitofrontal cortex among subjects who received continuation treatment. The registry record does not report a complete statistical analysis plan, analysis populations, or results for these outcomes.

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: Only subjects currently in a manic episode were recruited for the acute treatment protocol substudy.

Acute Treatment -- 3 Weeks

Continuation Treatment -- up to 12 Month

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All subjects in an acute manic state who were treated with specified protocol.

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 Boyko DWM Hyperintensity Value >2 in Subjects Who Received Acute Treatment for Mania, measured over 3 weeks.

The primary endpoint is the number of participants with a Boyko deep white matter hyperintensity value above 2 after acute treatment for mania. The timeframe is 3 weeks. The endpoint is a count of participants meeting a stated imaging classification threshold; it is not a measure of symptom improvement or a direct comparison of treatment strategies.

The description explains that the Boyko classification assesses deep white matter hyperintensities from absent findings through punctate, rounded, irregular, and confluent lesions. A threshold-based endpoint can make a specific imaging feature easier to summarize, but it also compresses the underlying classification into whether a participant falls above or does not fall above the chosen threshold. The result can therefore depend on how the rating is made and on the threshold used.

As a general statistical point, an endpoint defined as a participant count can be reported as a count and, where denominators are available, as a proportion. Uncertainty matters because a count from a small study population may be imprecise. The registry record does not report the primary count, its denominator, or an uncertainty interval. It is therefore not possible to describe the observed frequency or its precision from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a binary classification summarized as a participant count: whether the Boyko deep white matter hyperintensity value is above 2. In general, a binary endpoint has two possible states for each assessed participant, such as meeting or not meeting a specified criterion. A count gives the number in one state. To understand how common that state was in a group, the count must be considered alongside the number assessed in that group. The registry record does not report those primary endpoint values.

What does the imaging threshold represent?

The endpoint uses a threshold on the Boyko classification system. The description lists categories from absent findings through punctate, rounded, irregular, and confluent lesions. Converting an ordered classification into a threshold-based outcome simplifies reporting, but it removes distinctions among values on the same side of the threshold. It also makes results dependent on the selected cutoff and on the consistency of image assessment. The registry record does not report details about assessor training, blinding, repeat readings, or agreement between readers.

Why do denominators matter?

A count alone does not show the frequency of an outcome unless the number assessed is also known. In general, a proportion uses the count meeting the criterion relative to the number with an observed assessment. If some participants lack an assessment, the denominator and the reason for missingness affect interpretation. Enrollment is listed as 19, but the number with a primary endpoint assessment and the analysis denominator are not reported in the registry record. It would not be appropriate to assume that every enrolled participant contributed to every outcome.

What does a confidence interval add?

In general, a confidence interval communicates uncertainty around an estimate under a stated statistical method and its assumptions. A wide interval indicates less precision than a narrow interval. An interval does not remove bias from participant selection, measurement, or group differences, and it does not turn an observational contrast into a randomized comparison. The registry record does not report a confidence interval for the primary endpoint or identify an interval method.

How should the group descriptions be read?

The acute-mania arm descriptions classify participants using Boyko deep white matter ratings and a YMRS value below 15 at week 3. The registry record characterizes treatment as naturalistic and notes that allocation was not stated and the arms are not a like-for-like control comparison. In general, comparisons of groups defined by observed characteristics can be affected by differences between the participants in those groups. Without random assignment and adequate information about the analysis, a group contrast cannot be interpreted as the causal effect of treatment.

The intervention description refers to treatment starting with valproic acid, while the primary endpoint description refers to the STEP-BD protocol and valproic acid as the primary intervention. The registry record does not report the full protocol, dosing, treatment changes, adherence, or concomitant care. Those details would matter for understanding actual treatment exposure, but should not be inferred from the brief descriptions.

What can planning calculations tell readers?

The stated planning assumptions describe an illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The stated effect to detect is a fraction of the observed effect. For non-time-to-event outcomes, planning quantities are described as participants per arm. These assumptions characterize planning illustrations, not observed study results or observed power. The registry record does not report the numerical planning values, the effect used for those illustrations, or a sample-size justification from the study’s own design.

In general, a sample-size calculation asks how much information would be needed to detect a chosen effect under specified assumptions. Its result depends on the assumed effect, variability, significance level, power target, and analysis method. A calculation based on an observed effect is especially dependent on that estimate, which may be imprecise. Such planning values do not establish that a study was adequately powered, and they do not provide evidence about the endpoint in the absence of its observed result.

Why can over-dispersion matter?

As a general concept, over-dispersion occurs when observed variation is greater than a statistical model expects. It is relevant to some count models and can make standard errors too small if the extra variability is ignored. The primary endpoint in this record is described as a participant count for a binary classification, not as a rate or recurrent-event count. The registry record does not report a model, a dispersion assessment, or a method for adjusting for over-dispersion. No claim about the use or need for such an adjustment can therefore be made for this analysis.

The record lists five outcome measures and five analyses. It does not report the model specifications, covariates, missing-data methods, multiplicity strategy, or whether analyses were prespecified. These details are important for evaluating reproducibility and the scope of any statistical claims. They are not established by the endpoint definitions alone.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Responders to Acute Treatment", "Non-Responders to Acute Treatment")
n      <- c(10, 9)
events <- c(10, 9)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Responders to Acute Treatment  10  10  100.0  69.2  100.0  72.2  100.0  71.5  99.8
ARM  Non-Responders to Acute Treatment  9  9  100.0  66.4  100.0  70.1  100.0  69.2  99.7

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists five outcome measures and five analyses. It identifies a primary outcome and several secondary outcomes, including a subcortical Boyko measure and fractional anisotropy measures for the left and right orbitofrontal cortex among subjects who received continuation treatment. However, a plan for handling multiple outcomes or analyses is not reported in the registry record. It is not stated whether any adjustment for multiple testing was planned or used, or how the primary and secondary findings were to be interpreted together.

As a general statistical point, examining multiple outcomes can increase the chance of seeing an apparently notable result by chance if each is assessed separately without an appropriate interpretive plan. Adjustments are not the only possible response; a prespecified hierarchy or clear distinction between confirmatory and exploratory analyses can also help readers understand the evidence. The registry record does not report such a hierarchy or classification beyond identifying the primary endpoint.

Missing-data methods are also not reported in the registry record. It does not state how many participants had usable MRI or clinical assessments, how missing observations were handled, or whether any participants were excluded from particular analyses. Enrollment of 19 should not be treated as the denominator for every outcome without those details.

The analysis populations are not described. The registry record does not state whether analyses included all enrolled participants, only those meeting stated criteria, or another defined group. It also does not report an interim-analysis plan or a statistical analysis plan. These absences prevent a full assessment of analysis selection, missingness, and the relationship between planned and reported analyses.

12. 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 0% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: lithium toxicity (5.3%).

The registry record lists a naturalistic treatment group with 19 participants at risk for serious adverse events and 1 participant affected. For other adverse events, it lists 19 at risk and 0 affected. Deaths are not reported in the registry record. Lithium toxicity is listed among the serious adverse event terms, with 5.3% recorded. The record does not provide further detail about timing, severity, clinical course, or attribution for that term.

These figures describe the safety entries in the registry record; they do not establish whether an event was caused by a treatment. The record does not report a randomized comparison that could support a comparative safety estimate. It also does not explain how adverse events were collected, coded, or monitored, or whether all participants had the same duration of safety follow-up.

As a general point, a small number of recorded events can provide limited precision about the frequency of uncommon harms. A zero recorded count does not demonstrate that an event cannot occur, and counts should be interpreted in light of the population at risk, observation period, and event definitions. The registry record does not report a safety analysis plan or additional detail needed to assess those factors.

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

The record describes a completed study examining brain imaging findings in people with bipolar disorder, with a primary endpoint focused on the number of participants whose Boyko deep white matter hyperintensity value was above 2 after 3 weeks of acute treatment for mania. It identifies the treatment context as naturalistic and refers to valproic acid as the primary intervention. Those details define the research question, but they do not establish what the observed endpoint result was.

The primary result count, its analysis denominator, and a measure of uncertainty are not reported in the registry record. The registry therefore does not allow readers to determine the observed frequency of the imaging finding or how precisely it was estimated. It also does not support a quantitative comparison between the listed groups.

The record’s note that the study was not a randomized comparison and that the arms are not like-for-like controls is important. Group definitions based on imaging ratings and a week 3 YMRS criterion do not show that differences between groups were caused by treatment. A causal claim would require a design and analysis capable of addressing differences between groups, and those details are not reported.

The endpoint is an imaging classification, not a direct measure of symptom relief, daily functioning, or long-term clinical outcome. The record’s purpose includes examining structural brain changes, but it does not report the clinical meaning of any imaging result. No conclusion about treatment benefit or lack of benefit follows from the endpoint definition or the safety counts alone.

In general, study results should be read in relation to the measurement method, analysis population, uncertainty, and design. The registry record supplies a useful outline of the question and outcome definitions, while leaving the primary result and key analysis details unreported. Its content therefore supports a cautious description of what was planned or assessed, not a numerical conclusion about the finding.

14. Why This Trial Matters Statistically

This study illustrates why a clear endpoint definition is only one part of statistical interpretation. The primary outcome is precisely framed as a participant count based on a Boyko deep white matter threshold and a 3-week timeframe. But without the observed count, denominator, and uncertainty, readers cannot assess the frequency or precision of that outcome. A well-defined measure does not substitute for a reported result.

It also highlights the importance of distinguishing a treatment description from a treatment comparison. The record describes naturalistic treatment starting with valproic acid, but explicitly notes that the arms are not a like-for-like control comparison and that allocation is not stated. This distinction is central to biostatistics: a contrast between groups can be descriptive without estimating what would happen under alternative assigned treatments.

The record’s enrollment of 19 places the analysis in a setting where detailed reporting of who contributed data is especially important. The registry record does not report outcome-specific denominators, missing-data handling, or analysis populations. Without that information, readers cannot tell whether the assessed participants represent the enrolled group or how missing observations might affect interpretation.

Finally, the five listed outcomes and analyses underscore the need for transparent multiplicity and analysis reporting. A registry description can identify outcomes without showing how they were prioritized statistically or interpreted together. This record is a useful example of why readers need both the clinical question and the full statistical context: outcome definitions, denominators, methods, uncertainty, and design constraints all shape what a result can support.

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 main limitation for interpreting this specific record is that the primary endpoint result is not reported in the registry record. The count meeting the Boyko threshold, the denominator assessed, and uncertainty around the estimate are absent. Enrollment is 19, but that figure does not establish how many participants contributed to each outcome or analysis.

The registry record states that this was not a randomized comparison, that allocation was not stated, and that the arms are not a like-for-like control comparison. The arm descriptions use imaging thresholds and a week 3 YMRS criterion, but the record does not explain the rationale for the group definitions or how their characteristics were handled in analysis. Differences between groups therefore cannot be interpreted as treatment effects from the record alone.

Further limitations are that the statistical analysis plan, analysis populations, missing-data methods, and multiplicity approach are not reported in the registry record. The record lists five outcomes and five analyses, but does not describe whether all were prespecified or how results across them were interpreted. It also does not report details of imaging assessment, assessor blinding, or consistency of ratings.

The intervention details are limited. The record refers to naturalistic treatment starting with valproic acid and describes valproic acid as the primary intervention, but does not report dose, treatment changes, adherence, or concomitant care. Safety information is also limited to the listed counts and term; deaths are not reported in the registry record. These gaps constrain interpretation of both the endpoint and safety entries.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Bipolar Disorder in Late Life — NCT00787930.
• Registry record last updated 2014-07-31; status last verified 2014-07.
• Results first posted 2014-07-31.
• 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

NCT00787930 is listed as a completed study titled “Bipolar Disorder in Late Life,” sponsored by Duke University, with enrollment of 19. The study start is recorded as 2005-10 and primary completion as 2010-09. Its stated purpose is to examine structural brain changes in people with bipolar disorder or a history of the condition.

The primary endpoint is the number of participants with a Boyko deep white matter hyperintensity value above 2 after 3 weeks of acute treatment for mania. The treatment description refers to a naturalistic protocol starting with valproic acid. The record lists five outcome measures and five analyses, but does not report the primary result, analysis denominator, confidence interval, or statistical analysis plan.

The registry record notes that allocation was not stated and the arms are not a like-for-like control comparison. Safety entries list 1 serious adverse event among 19 at risk and 0 other adverse events among 19 at risk; deaths are not reported. Lithium toxicity is listed as a serious adverse event term with 5.3% recorded.