Sensoril(Ashwaganhda)for Bipolar Disorder: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sensoril(Ashwaganhda)for Bipolar Disorder (NCT00761761), a randomized phase 3 study. The primary outcome is Change From Baseline in Digit-Span Score at 8 Weeks.

2. Clinical Question

3. Trial Design

The registry identifies NCT00761761 as a Phase 3, randomized, parallel-group study with quadruple masking. It is titled as a study of Sensoril, a standardized Ashwagandha extract, for cognitive enhancement in people with bipolar disorder. The recorded status is completed. The University of Pittsburgh is listed as sponsor, and the registry dates the study start to 2008-10 and primary completion to 2011-03.

The registry summary describes an 8 week, placebo-controlled add-on treatment trial. Participants were to continue existing mood stabilizer treatment while receiving either Sensoril or placebo. The listed conditions are Bipolar I Disorder, Bipolar II Disorder, and Bipolar Disorder NOS. The summary says that 60 subjects with DSM IV-TR bipolar disorder were to be recruited. The endpoint-specific outcome record lists 24.0 participants for Sensoril and 29.0 for placebo. The registry record does not report why the outcome counts differ from the enrollment figure, or whether the listed counts reflect a particular analysis population.

The Sensoril arm description gives a starting dose of 250 mg/day, increasing to 500 mg/day by the second week. The placebo arm is recorded with the same stated dose schedule. The record identifies four outcome measures and three posted secondary outcomes focused on residual depressive symptoms, symptoms of mania, and anxiety symptoms. The primary posted outcome is change from baseline in digit-span score at 8 weeks. Although the record specifies randomization, parallel groups, placebo control, and quadruple masking, further details of sequence generation, allocation concealment, and who was masked are not reported in the registry record.

The summary says that cognitive, psychopathology, and laboratory measures, as well as safety assessments, were to be used. The registry record does not provide a detailed protocol, the full schedule of assessments, or the trial's original statistical plan. The design description therefore establishes the broad comparison and intended treatment period, but does not fill in those unreported operational details.

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.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

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

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Digit-Span Score at 8 Weeks, measured over 8 week treatment.

The primary outcome is change from baseline in digit-span score at the end of the 8 week treatment period. Digit span is described as a test of short-term or working memory. Its raw scores range from a minimum of 2 to a maximum of 8, and the registry states that a longer span and higher score indicate better cognitive performance. This makes the direction of the outcome interpretable: a higher change score represents a greater increase in digit-span performance, while a lower change score represents less improvement or a decrease, depending on the starting score.

The registry describes the reported endpoint as a least-squares mean, with standard error as the measure of dispersion. The posted arm values are 0.73 for Sensoril and 0.17 for placebo. These are group-level summaries, not individual participant results. The reported mean difference is 0.56 in favor of the Sensoril arm when expressed as Sensoril minus placebo. This endpoint is relevant to the stated cognitive question because working memory is among the cognitive abilities named in the study summary. It is a specific test result, however, and should not be treated as a measure of every aspect of cognition or of overall functioning.

The registry says a comprehensive cognitive battery was assessed but presents digit-span results for the primary outcome. Further details about how the least-squares means were estimated, including covariates or handling of missing measurements, are not reported in the registry record.

7. Results Posted in the Registry

• Sensoril: mean 0.73 versus 0.17 on control; difference 0.56 (95% CI 0.05 to 1.07), standardized effect d = 0.59; z-test p = 0.032.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: digit-span change is recorded on a numerical scale, and the registry reports group means rather than event counts. For a continuous outcome, a difference between group means describes how far apart the group-level average changes are, in the units of the outcome. Here the reported mean difference is 0.56, with the comparison expressed as Sensoril minus placebo. The direction is meaningful because the endpoint description says higher digit-span scores indicate better performance.

What does a least-squares mean represent?

A least-squares mean is a model-based group average. Depending on the statistical model, it may account for specified covariates or other features of the analysis. The registry labels the primary result as a least-squares mean, but does not report the model specification, covariates, or estimation details. It is therefore not possible to determine from the registry record what adjustments, if any, distinguish these values from ordinary arithmetic averages. The posted arm summaries are 0.73 for Sensoril and 0.17 for placebo.

What is the standard error?

The registry gives standard errors of 0.19 for the Sensoril arm and 0.18 for the placebo arm. A standard error describes the estimated sampling variability of a statistic, such as a mean. In general, it reflects how precisely the group mean is estimated, rather than the spread of individual participant scores. It should not be confused with a standard deviation, which describes variation among observations. The record labels the dispersion measure as standard error; it does not provide the participant-level score distribution.

How should the mean difference and p-value be read?

The comparison reports a mean difference of 0.56, with a confidence interval from 0.047 to 1.073 and an unadjusted p-value of 0.032. A p-value measures how compatible the observed result is with a specified null hypothesis under the assumptions of the analysis. It is not the probability that the null hypothesis is true, nor does it quantify the size or clinical importance of an effect. Because the value is unadjusted, it does not account for any multiplicity unless that adjustment is included elsewhere; such an adjustment is not reported in the registry record.

The confidence interval describes a range of values that are compatible with the data and the interval's statistical assumptions. In general, a narrower interval indicates greater precision, while a wider interval indicates more uncertainty about the effect size. The interval lies above zero, but that observation should be read alongside the unadjusted status of the test, the missing analysis details, and the absence of a reported clinically important threshold.

Why can over-dispersion matter?

Over-dispersion occurs when observed variability is greater than a statistical model expects. In general, if a model understates variability, standard errors may be too small and uncertainty may appear lower than it is. This issue is especially relevant to count outcomes, where event variation can exceed a simple model's assumptions. The primary outcome in this record is continuous and is summarized with standard errors; the registry does not report a dispersion assessment or enough model detail to evaluate this issue for the endpoint analysis. It should not be assumed that over-dispersion was present or absent.

What does the standardized effect describe?

The comparison record also gives a standardized effect-size value, but a standardized effect is not expressed in the original digit-span units. In general, standardization divides a group difference by a measure of variability, making the result less dependent on the original measurement scale. The record indicates that the pooled standard deviation used in this calculation is approximate. The registry does not report the derivation in detail. Since the digit-span scale and mean difference are directly interpretable, the primary reading should remain focused on the difference in scale units, with standardized quantities treated as secondary context rather than as evidence of clinical importance.

The registry lists 24.0 and 29.0 as the endpoint group counts. It does not report the analysis-population definition, the handling of missing measurements, or whether the least-squares mean analysis used all randomized participants. Those details matter because analysis choices can affect both estimates and their uncertainty. No additional assumptions about those choices can be inferred from the result labels alone.

10. Confidence Intervals and Precision

The mean difference in digit-span change is 0.56, with a confidence interval from 0.047 to 1.073. The interval communicates uncertainty about the size of the difference: values across that range are compatible with the result under the interval's assumptions. It is more informative than the point estimate alone because it shows that the true mean difference is not known exactly. The interval is entirely above zero, consistent with the positive direction of the posted estimate, but it does not establish clinical importance.

The interval extends across values that may represent meaningfully different magnitudes of change. The registry record does not identify a minimal clinically important difference or another benchmark for deciding whether any particular value in the interval matters to participants. It also does not report whether the interval is adjusted for multiple outcomes or analyses. Therefore, the interval should be interpreted as precision for the posted comparison, not as a definitive statement about the total evidence from the study.

Standard errors of 0.19 and 0.18 are reported for the Sensoril and placebo arm means. These indicate uncertainty in the estimated means, rather than the range of scores across participants. The registry does not provide standard deviations for the arm-level digit-span results or participant-level data, so the amount and shape of individual variation cannot be assessed directly from these summaries.

Precision also depends on the analysis population and model. The endpoint summaries list 24.0 and 29.0 participants, while recorded enrollment is 60. The registry record does not explain this difference or report missing-data handling and model details. The numerical interval is thus available, but important context for evaluating its robustness is not.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on fractions of the observed effect, not the trial's original design calculations. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect. The trial's own sample-size justification and original design assumptions are not reported in the registry record.

The figures are participants per arm, not total participants or total events. For each assumed effect, the 80% power column gives the number of participants per arm associated with an 80% chance of detecting that effect under the planning assumptions if the effect truly exists. The 90% power column gives the corresponding planning quantity for a 90% chance. These are design calculations, not descriptions of the study's observed results.

The rows consider planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. The listed participants per arm are 46 and 61 for planning at 100% of the observed effect; 81 and 109 for 75%; and 182 and 244 for 50%, respectively, for 80% and 90% power. Smaller effects need more participants to detect under the same assumptions, which is why the planning figures increase as the target effect becomes a smaller fraction of the observed effect.

This is not observed power. The registry record does not report the original power calculation, the assumptions used by the investigators, or whether any planned sample-size reassessment occurred.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.560; pooled standard deviation σ = 0.952.
2. n = 2 × (1.96 + 0.842)² × 0.952² / 0.560² = 45.4, rounded up to 46 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.56                      # observed difference in means
sigma <- 0.952149271539982                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(761761)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 46 (80% power), 61 (90% power)
 75% of observed effect: n per arm = 81 (80% power), 109 (90% power)
 50% of observed effect: n per arm = 182 (80% power), 244 (90% power)
power.t.test: 46.4 
Simulated power at n = 46 per arm: 0.790 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures and identifies digit-span change at 8 weeks as the primary outcome. It also lists three secondary outcomes concerning residual depressive symptoms, symptoms of mania, and anxiety symptoms. The primary comparison has an unadjusted p-value of 0.032. The registry record does not report a multiplicity procedure, such as an adjustment to account for testing multiple outcomes or comparisons, or a hierarchical testing strategy. The posted p-value should therefore be understood as unadjusted, and the record does not establish how the investigators treated the full set of outcomes statistically.

The endpoint analysis counts are 24.0 for Sensoril and 29.0 for placebo, whereas recorded enrollment is 60. The registry record does not explain the relationship between these figures. It does not report how many randomized participants were included in each analysis population, why any participants were not represented in the endpoint summaries, or whether participants were analyzed according to their randomized assignment.

Missing-data methods are also not reported in the registry record. There is no posted description of imputation, likelihood-based approaches, sensitivity analyses, or assumptions about why data may be missing. Without that information, the potential effect of incomplete outcome data on the estimate cannot be assessed from these summaries.

The record labels the arm values as least-squares means, but does not provide the model specification, covariates, or a full statistical analysis plan. It does not report whether analyses were adjusted for baseline score, whether interim analyses were planned or conducted, or whether the primary analysis was prespecified in the protocol. These points are unknown from the registry record and should not be inferred from the posted result. They limit the ability to evaluate selective reporting, analytic flexibility, and the robustness of the nominal statistical evidence.

14. 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 5% reporting threshold.

Most frequent non-serious adverse event terms

The safety record lists serious adverse events for 0 of 30 participants at risk in the Sensoril group and 0 of 30 participants at risk in the placebo group. The registry record does not provide serious adverse-event terms, consistent with the absence of listed serious events in these entries. Deaths are recorded as null for both groups, which means death results are not reported in the registry record; null should not be interpreted as a confirmed count of zero.

Other adverse events are recorded for 5 of 30 participants at risk in the Sensoril group and 1 of 30 participants at risk in the placebo group. The registry record does not provide the event terms, severity, timing, or whether an individual could contribute more than one event. These are participant counts marked as affected, not necessarily counts of all event episodes.

These summaries are descriptive and do not establish that the arms differ in safety. The registry record does not provide detail on ascertainment, follow-up completeness, or statistical comparison of adverse events. The figures should be read alongside the limited scope of the posted safety information, rather than as a complete safety profile.

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

The posted result describes a positive mean difference in change in digit-span score for Sensoril compared with placebo: 0.56, with a confidence interval from 0.047 to 1.073. The endpoint description identifies higher scores with better performance, so the estimated direction favors Sensoril on digit span. The unadjusted p-value is 0.032. These statistics concern one recorded cognitive measure at the end of an 8 week treatment period.

The result does not establish that Sensoril improves cognition in every sense, or that the estimated change is clinically important. Digit span is a specific test of short-term or working memory, and the registry does not provide a benchmark for a meaningful change. Nor does the posted result establish that the difference would persist beyond the treatment period: longer-term outcome results are not reported in the registry record.

The result also does not settle the interpretation of all outcomes from the study. The record lists other cognitive and secondary outcomes, but does not provide an adjustment strategy for multiple testing. The primary p-value is unadjusted. Analysis methods, missing-data handling, and the relationship between enrollment and endpoint counts are not described sufficiently to assess how these issues might affect the estimate.

Finally, a registry summary is not a full account of the evidence needed to judge generalizability or overall benefit and harm. The comparison was an add-on study in people with specified bipolar disorder diagnoses, with an 8 week treatment period and a particular reported cognitive endpoint. The safety information is limited, and deaths are not reported in the registry record. The statistics support a cautious account of the observed endpoint difference and its uncertainty, not medical advice or a treatment recommendation.

16. Why This Trial Matters Statistically

This trial is statistically informative because it reports a continuous cognitive outcome as both arm-level summaries and a between-arm mean difference with a confidence interval. That combination helps separate the direction and estimated size of the result from its precision. The digit-span description also clarifies the direction of interpretation: higher scores are better on this measure.

The record illustrates why statistical significance and practical meaning are different questions. The unadjusted p-value is 0.032, and the interval for the mean difference lies above zero, but the registry does not give a clinically important threshold. A positive estimate can therefore be described without claiming that its size matters to participants or represents broad cognitive improvement.

It also highlights the importance of analysis transparency. The endpoint is labeled a least-squares mean, but the model, covariates, missing-data approach, and analysis population are not described. The record lists multiple outcomes without reporting a multiplicity plan. These details affect how confidently a reader can evaluate the precision and evidential weight of the result.

The endpoint counts differ from the enrollment figure, and the registry does not explain that difference. This is a practical reminder that enrollment alone does not reveal how many participants contributed to a particular outcome analysis. Clear reporting of analysis populations and missing-data handling is essential to interpreting group comparisons.

Finally, the planning illustrations show how sample-size requirements depend on the effect a future study aims to detect. Smaller target effects require more participants under the stated assumptions. Those calculations are useful for planning context but do not represent observed power or the original design rationale for this completed study.

17. 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 leaves several trial-specific uncertainties. Although enrollment is recorded as 60, the primary outcome lists 24.0 participants for Sensoril and 29.0 for placebo. Reasons for the differing counts, participant flow, withdrawals, and the analysis population are not reported in the registry record. Missing-data methods and sensitivity analyses are also not reported.

The primary endpoint is described as a least-squares mean, but the registry does not give the statistical model or identify any covariates. It is therefore unclear from the record how the estimates were adjusted, if at all. The comparison p-value is unadjusted, while the record lists multiple outcome measures and secondary outcomes. A multiplicity plan is not reported.

The primary result focuses on digit span, a specific measure of working memory. Although the study summary describes a comprehensive cognitive battery, results for the other cognitive measures are not included in the primary outcome entry. The registry record does not report a threshold for clinically important change or longer-term cognitive outcomes. The clinical relevance and durability of the observed difference therefore cannot be assessed from this entry.

The study summary describes add-on treatment alongside existing mood stabilizers, but the registry record does not provide details about those background treatments or their consistency across groups. It also does not report detailed masking procedures, allocation concealment, protocol deviations, or interim analyses. Safety reporting is limited: serious and other adverse-event summaries lack event terms and timing, and deaths are not reported in the registry record. These gaps restrict assessment of trial conduct, result robustness, generalizability, and safety interpretation.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Sensoril(Ashwaganhda)for Bipolar Disorder — NCT00761761.
• Registry record last updated 2016-04-14; status last verified 2016-03.
• Results first posted 2016-04-14.
• 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.

21. Record Summary

NCT00761761 is recorded as a completed Phase 3, randomized, parallel-group, placebo-controlled study sponsored by the University of Pittsburgh. It tested Sensoril as add-on treatment for people with Bipolar I Disorder, Bipolar II Disorder, or Bipolar Disorder NOS. Enrollment is listed as 60, with an 8 week treatment period and quadruple masking.

The primary posted outcome is change from baseline in digit-span score. The mean difference for Sensoril versus placebo is 0.56, with a confidence interval from 0.047 to 1.073 and an unadjusted p-value of 0.032. Higher digit-span scores are described as better performance. The registry record does not report the statistical analysis plan, missing-data approach, multiplicity procedure, or explanation for the difference between enrollment and endpoint counts.