Melatonin Agonist Effects of Tasimelteon Versus Placebo in Patients With Major Depressive Disorder: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Melatonin Agonist Effects of Tasimelteon Versus Placebo in Patients With Major Depressive Disorder (NCT01428661), a randomized phase 2, phase 3 study. The primary outcome is Change From Baseline to Endpoint at Week 8 Using the Total Score of the Hamilton Depression Rating Scale (HAM-D).

2. Clinical Question

3. Trial Design

The registry describes MAGELLAN as a randomized, parallel-group study evaluating tasimelteon and placebo in adults with major depressive disorder. The study is listed as Phase 2 and Phase 3, with an enrollment of 507. Its official title describes a multicenter, randomized, double-masked, placebo-controlled study followed by a 52-week open-label extension. The registry’s design fields classify masking as quadruple, while the official title and brief summary use double-masked wording. The record does not explain that difference.

The primary treatment period is described as 8 weeks. The primary outcome is change from baseline to the Week 8 endpoint on the HAM-D total score. The two masked groups have posted outcome counts of 249 for tasimelteon and 243 for placebo. The randomized design makes the tasimelteon-versus-placebo comparison the central contrast for interpreting the primary outcome, while the open-label extension is a separate study period and should not be read as another masked randomized comparison.

The registry also lists an open-label tasimelteon group with 339 participants and a mean HAM-D change. Its presence is consistent with the extension described in the official title, but the record does not provide enough detail to reconstruct participant flow into that group or the exact timing of its outcome assessment. The primary outcome timeframe is stated as 8 weeks; the open-label outcome’s detailed timeframe and analysis population are not reported in the registry record.

The record marks the study as completed. The start and primary completion dates are recorded, but they do not establish additional design details. In particular, the original sample-size justification, allocation ratio, randomization method, concealment procedures, and prespecified statistical analysis plan are not reported in the registry record. Those omissions limit how fully the trial’s intended analysis can be compared with the posted results.

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: \*Tasi: subject left country (1); Placebo: sponsor request (1), subject moved (1), IMP schedule (1), subject incarcerated (1), withdrawn after receipt of medical records (1), visit schedule (1) \*\*Tasi: surgery (1), visit schedule (3), +UDS (2), IMP schdule (2), prohibited meds (1), withdrew consent (1), sponsor terminated trial (175)

Double-Masked Phase

Reasons for not completing this period:

Open Label Extension

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The Intent-to-Treat (ITT) Population included any subject randomized into the study that receives a dose of study medication and that has completed at least one post-baseline efficacy measurement while on study medication.

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 to Endpoint at Week 8 Using the Total Score of the Hamilton Depression Rating Scale (HAM-D), measured over 8 weeks.

The primary endpoint is change from baseline to the Week 8 endpoint in the total score of the Hamilton Rating Scale for Depression, abbreviated HAM-D. It is a continuous measure, and the registry reports group means with standard errors. The scale assesses depressive symptoms, ranges from 0 to 52, and higher scores indicate more severe symptoms. Accordingly, a negative change means the score decreased from baseline; more negative changes indicate a larger reduction in the measured symptom score.

A change score focuses on how much the measured score moved over the treatment period rather than on the endpoint score alone. Comparing average change between groups can describe whether the groups had different mean trajectories over that period. The between-group estimate is especially important: a decrease within the tasimelteon group by itself does not establish that its change differed from the placebo group’s change.

HAM-D is used as a structured measure of depressive symptoms, allowing a numerical summary of symptom severity and change. The registry identifies the measure and its direction of interpretation, but does not report further details about administration, assessor training, or which HAM-D version was used. The reported endpoint therefore supports interpretation of the posted score changes, while those measurement details cannot be assessed from the registry record.

The primary population is defined as randomized participants who received study medication and completed at least one post-baseline efficacy measurement while on medication. That operational definition matters because the reported outcome does not necessarily represent every randomized participant. The record does not explain how missing Week 8 measurements were handled or whether a particular method was used to estimate endpoint changes for participants without observed endpoint values.

7. Results Posted in the Registry

• Tasimelteon: mean -8.19 versus -7.83 on control; difference -0.36 (95% CI -1.61 to 0.89), standardized effect d = -0.05; z-test p = 0.572.
• Open Label Tasimelteon: mean -13.60 versus -7.83 on control; difference -5.77 (95% CI -6.88 to -4.66), standardized effect d = -0.88; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The endpoint is a continuous score, so the registry reports a mean change for each group. A mean is an average across the participants included in that group’s outcome analysis. The reported means describe change from baseline, not the absolute HAM-D score at Week 8. Since negative change indicates a decrease, the negative group means correspond to average score reductions.

The registry gives standard errors of 0.45 for the tasimelteon and placebo mean changes, and 0.34 for the open-label group mean. A standard error describes the sampling precision of an estimated mean: it reflects how much the estimate would vary across repeated samples under a statistical model. It is not the standard deviation of individual participant changes and should not be interpreted as describing the spread of individual responses. The record does not provide confidence limits for the individual group means.

The between-group mean difference is a direct contrast of group means. For the primary comparison, the reported estimate is -0.36. A negative value means the tasimelteon group’s mean change is more negative than the placebo group’s mean change. The estimate alone is not enough to judge its precision: the confidence interval and p-value describe uncertainty under the assumptions of the calculation.

The confidence interval for that contrast extends from -1.607 to a positive upper limit. Because it spans zero, the range includes a zero difference and values on either side of zero. An interval is not a list of values that are equally plausible, nor does it mean that a fixed percentage of participants had changes inside the interval. It is an interval estimate for the population-level mean contrast, constructed under statistical assumptions. The positive upper endpoint is not repeated numerically because its recorded precision exceeds the permitted display precision.

The p-value for the primary recalculated comparison is 0.572. A p-value describes how unusual the data, or more extreme data, would be under a specified null model, assuming the model and its assumptions. It does not give the probability that the null hypothesis is true, the probability that the result occurred by chance, or the probability that a treatment is effective. A p-value also does not quantify the magnitude of the group difference; the estimate and confidence interval address magnitude and precision more directly.

What does “recalculated” mean?

The posted between-group summaries are labeled recalculated. They are simple, unadjusted calculations based on the group summary statistics rather than a reported sponsor model. An unadjusted comparison does not account for covariates unless those are explicitly included in the calculation. The registry does not post sponsor statistical analyses, so it is not possible to confirm whether the sponsor’s analysis used a different model, covariate adjustment, or missing-data strategy. The recalculated results should be treated as descriptive statistical reconstructions, not as a replacement for an unpublished analysis.

How should the open-label comparison be understood?

The posted open-label comparison is calculated against the placebo group and has an estimated mean difference of -5.77, with a confidence interval from -6.875 to -4.665 and an unadjusted p-value of <0.001. The reported open-label mean is -13.6. These calculations describe a numerical contrast, but the extension is explicitly open-label and follows the 8-week treatment period. The registry does not report enough information to establish that the open-label group and placebo group can be interpreted as a randomized, concurrent comparison for that outcome. The result should not be treated as having the same inferential status as the primary placebo-controlled comparison.

What are standardized effects and pooled variability?

The registry includes pooled standard deviations and standardized effect estimates in its recalculated comparisons. A pooled standard deviation combines group variability into a common scale, and a standardized effect expresses a mean contrast relative to that variability. These measures can help compare magnitudes across outcomes measured on different scales, but they do not remove uncertainty or establish clinical importance. Because the HAM-D scale and mean difference already have direct interpretation in scale units, the raw mean contrast and its interval are central to understanding this endpoint. The standardized effect estimates are not used to claim a threshold of clinical importance; no such threshold is reported in the registry record.

What assumptions are not documented?

The registry does not report the original analysis plan, the sponsor’s model specifications, the handling of missing data, or whether interim analyses were planned. It also does not describe model diagnostics or any adjustment for multiple comparisons. These details matter because they influence how a result is estimated and interpreted. The known planning illustration assumptions are described separately with the sample-size estimates; they should not be confused with the trial’s original design assumptions, which are not reported in the registry record.

10. Confidence Intervals and Precision

For the primary tasimelteon-versus-placebo comparison, the recalculated mean difference is -0.36. Its 95% confidence interval begins at -1.607 and extends to a positive endpoint. The interval therefore crosses zero. It includes the possibility of a more negative change in the tasimelteon group, a zero mean difference, and a less negative change in that group. The posted estimate is not precise enough to distinguish confidently among those possibilities.

The lower confidence limit is -1.607. The registry’s upper confidence limit is recorded with more than three decimal places, so it is not reproduced numerically. Its positive sign, together with the lower limit, establishes that zero lies within the interval. The interval’s width relative to the point estimate conveys substantial uncertainty about the size and direction of the mean contrast.

A 95% confidence interval is a procedure-based measure of precision. Under the statistical model and its assumptions, intervals produced repeatedly by the same method would cover the true parameter at the stated long-run rate. It is not correct to say that there is a 95% probability that this particular fixed parameter lies inside the observed interval. Nor is the interval a range containing 95% of individual participants’ responses.

The standard errors for the two primary group means are both 0.45. They describe precision of those means, not participant-level variability. The record also reports no confidence intervals for the individual means. The uncertainty around the between-group contrast is more relevant to the comparative question than either group mean considered alone.

The open-label comparison has a narrower-looking separation in the posted interval, from -6.875 to -4.665, and the interval does not include zero. That numerical precision does not resolve the design limitations of comparing an open-label extension group with placebo. Precision of an estimate and validity of the comparison are different issues: a narrow interval cannot by itself make groups exchangeable or establish a randomized contrast.

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 estimates are planning illustrations derived from the observed effect, not the trial’s original design justification. The assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Each planning scenario targets a stated fraction of the observed effect. The estimates for this continuous outcome are participants per arm, not total events and not the total number across both arms.

The 80% power column gives the estimated participants per arm for a design that would have an 80% probability of rejecting the null hypothesis under the specified assumptions if the assumed effect were the true effect. The 90% column gives the corresponding planning count for a 90% probability. These are prospective design calculations, not a description of what happened in the completed trial. They are not observed power.

For the primary comparison, the planning estimates rise as the target effect is reduced from the full observed effect to 75% and then 50% of that effect. For example, the estimate for 80% power is 6,035 participants per arm at 100% of the observed effect, 10,729 at 75%, and 24,139 at 50%. The corresponding estimates for 90% power are 8,079, 14,363, and 32,316 participants per arm. Smaller effects are harder to distinguish from random variation, so detecting them with the same target power requires more participants under the stated assumptions.

These figures depend on taking the observed effect as a planning anchor. The primary estimated contrast is small and uncertain, so the resulting planning counts should not be treated as definitive design recommendations. They also do not establish that the completed trial was designed for these counts or that it had any particular power. The trial’s own sample-size assumptions are not reported in the registry record.

The open-label contrast has a separate set of planning illustrations, but those figures inherit the limitations of using its observed comparison as an effect-size anchor. They are not evidence that a future study with those counts would reproduce the open-label contrast. Planning estimates are conditional on the specified assumptions and target effect, and the design validity of the comparison remains a separate consideration.

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 = -0.360; pooled standard deviation σ = 7.059.
2. n = 2 × (1.96 + 0.842)² × 7.059² / -0.360² = 6,034.7, rounded up to 6,035 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 <- -0.359999999999999                      # observed difference in means
sigma <- 7.05850117061053                     # 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(1428661)
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 = 6,035 (80% power), 8,079 (90% power)
 75% of observed effect: n per arm = 10,729 (80% power), 14,363 (90% power)
 50% of observed effect: n per arm = 24,139 (80% power), 32,316 (90% power)
power.t.test: 6035.7 
Simulated power at n = 6,035 per arm: 0.801 (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 one primary outcome measure and no posted secondary outcomes. It also lists no sponsor statistical analyses. This does not establish that the study had no secondary endpoints or analyses; it means they are not reported in the registry record. The record does not identify a multiplicity-control method or explain whether the primary result was protected from the effects of testing multiple outcomes, time points, or contrasts.

Multiplicity arises when multiple statistical hypotheses are tested. As the number of tests grows, the chance of at least one apparently noteworthy result can increase unless the analysis plan accounts for the testing structure. The available registry information does not show whether such adjustments were planned or used. The unadjusted p-values should therefore be read as unadjusted rather than as results known to have been corrected for multiple testing.

The primary analysis population is described as the Intent-to-Treat population, but its registry definition requires participants to have been randomized, to have received a study medication dose, and to have completed at least one post-baseline efficacy measurement while on study medication. The record does not report the number randomized who did not meet those conditions, how many participants contributed endpoint data, or reasons for missing measurements. It also does not describe imputation, likelihood-based estimation, last-observation methods, or sensitivity analyses. No such methods should be inferred.

Missing data can affect a change-score comparison if the availability of measurements is related to outcome or treatment. Whether that occurred cannot be assessed from the posted record. The registry also does not report whether the analysis included all available measurements or required a Week 8 value, beyond the stated population criterion. These omissions limit evaluation of attrition and the robustness of the primary estimate.

The posted result is a recalculated unadjusted comparison rather than a documented sponsor analysis. The analysis model, handling of repeated measurements, covariate adjustment, and any interim monitoring plan are not reported in the registry record. Consequently, the posted summary supports interpretation of the reported means and simple contrast, but not a complete assessment of the trial’s prespecified inferential strategy.

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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: chest pain (0.6%), non-cardiac chest pain (0.4%), blood pressure increased (0.4%), suicidal ideation (0.4%), back pain (0.4%).

In the masked treatment groups, serious adverse events were recorded for 2 of 254 participants at risk in the tasimelteon group and 3 of 252 participants at risk in the placebo group. These are counts of participants affected, not a comparison of event severity or a measure of causal attribution. The registry does not report deaths for these groups; the death figures are not reported in the registry record.

The registry also lists serious adverse-event terms and their reported percentages: Chest Pain, 0.6%; Non-Cardiac Chest Pain, 0.4%; Blood Pressure Increased, 0.4%; Suicidal Ideation, 0.4%; Back Pain, 0.4%; and Suicide Attempt, 0.4%. These terms are listed as the most frequent serious-event terms, but the record does not provide further detail about timing, duration, relatedness, or event-level overlap. The displayed percentages should not be interpreted as proof that an event was caused by study treatment.

Other adverse events affected 38 of 254 participants in the tasimelteon group and 53 of 252 in the placebo group. During the open-label extension, serious adverse events affected 6 of 339 participants and other adverse events affected 74 of 339. The extension group is reported separately; its counts should not be directly compared with the masked groups as if they arose under identical treatment and observation conditions. These safety counts describe what the registry reports and do not, on their own, establish comparative risk.

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

The primary result describes average HAM-D score changes over the 8-week treatment period. The tasimelteon and placebo group means were both negative, indicating average score reductions from baseline. The estimated difference between their mean changes was -0.36, and its confidence interval includes zero. The posted comparison therefore leaves considerable uncertainty about the size and direction of the underlying mean difference.

This is not the same as demonstrating that the groups were identical. A confidence interval that includes zero means the data are compatible with zero difference under the calculation, but also with other values in the interval. Nor does the result establish that there is no clinically meaningful difference: the registry does not report a clinical-importance threshold for the HAM-D change contrast. The estimate, its uncertainty, and the missing clinical threshold must be considered separately.

The result also does not determine whether individual participants improved, worsened, or had no change. Group means summarize average changes and can conceal a range of individual outcomes. The record does not report participant-level distributions or responder outcomes. The posted statistics cannot be used to infer how many participants experienced a particular degree of change.

The open-label extension mean and recalculated comparison are distinct from the primary masked comparison. The larger numerical separation in that comparison does not establish that tasimelteon caused the difference. Open-label status and incomplete information about the extension analysis make it inappropriate to treat this as a confirmatory randomized placebo contrast. Its p-value and confidence interval describe the recalculated numbers under their assumptions, not the quality of the design or the causal interpretation.

Finally, the registry record does not include the sponsor’s statistical analyses, missing-data approach, or original design assumptions. The primary outcome summaries can be described, but they do not supply a complete account of the trial’s statistical plan. The findings should be understood as registry-posted estimates with stated uncertainty, not as medical advice or a standalone judgment about treatment.

16. Why This Trial Matters Statistically

This trial illustrates why the estimated difference between groups is more informative for a randomized treatment question than the direction of change in either group alone. Both primary group means indicate reductions from baseline, but the treatment question concerns whether their mean changes differed. The estimated contrast is -0.36, and its confidence interval includes zero. Looking only at the negative means could make the groups appear to have changed in a similar direction while obscuring the uncertainty in their comparative result.

It also illustrates the distinction between a point estimate and precision. The primary estimate is a single number, but the interval shows that the data do not locate the mean contrast precisely. The p-value of 0.572 is consistent with that uncertainty; it is not a measure of clinical relevance and does not establish equivalence. Without a reported equivalence margin or clinical-importance threshold, the registry does not support those interpretations.

The open-label extension offers a useful statistical contrast in interpretation. Its reported mean change and comparison with placebo differ numerically from the masked-period primary result, yet the design context is not the same. A small p-value cannot correct for differences in masking, timing, or group comparability. Statistical evidence depends on the design and assumptions behind an estimate, not on a p-value alone.

The record also shows the limits of registry-level summaries. Group means and standard errors permit simple unadjusted comparisons, but do not reveal participant-level distributions, missing-data patterns, model details, or the sponsor’s analysis. That distinction matters when readers evaluate whether a result is robust and how it should be interpreted. Planning illustrations further show that sample-size needs depend strongly on the target effect; they are conditional calculations, not observed power or a reconstruction of the original rationale.

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 posted record does not report the sponsor’s statistical analyses, the original sample-size justification, or the prespecified analysis plan. As a result, the recalculated unadjusted comparisons cannot be checked against the intended primary model. The record does not indicate whether the sponsor used covariate adjustment, a repeated-measures method, or another approach.

The primary outcome population definition requires medication exposure and at least one post-baseline efficacy measurement. The number of randomized participants excluded from that population, the amount of missing Week 8 data, and the reasons for missingness are not reported in the registry record. No missing-data method or sensitivity analysis is documented. These gaps limit assessment of attrition and the potential influence of unavailable outcomes.

Details of the open-label extension comparison are limited. The record reports a group mean, a comparator-based recalculation, and the extension context, but not sufficient information about the comparison’s timing, participant flow, or analysis population. That estimate should therefore be kept separate from the primary masked comparison.

The registry’s design descriptions use both double-masked wording in the official title and brief summary and a quadruple masking classification in its design fields. The record does not explain the difference. It also does not provide detailed information about randomization procedures, allocation concealment, HAM-D administration, or assessor training. Those omissions constrain evaluation of design execution and measurement procedures.

Finally, the registry lists no secondary outcomes and no sponsor analyses, while not establishing whether such outcomes or analyses were absent from the broader study. Multiplicity procedures, interim analyses, and clinical-importance thresholds are not reported in the registry record. Safety data are presented as counts and selected terms without detailed attribution or event-level context.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Melatonin Agonist Effects of Tasimelteon Versus Placebo in Patients With Major Depressive Disorder — NCT01428661.
• Registry record last updated 2015-06-19; status last verified 2015-06.
• Results first posted 2015-06-19.
• 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

The registry describes a completed randomized, parallel-group study of tasimelteon versus placebo in adults with major depressive disorder, with a 52-week open-label extension described in the official title. The primary endpoint is Week 8 change from baseline in HAM-D total score. The primary group means were -8.19 for tasimelteon and -7.83 for placebo; the recalculated mean difference was -0.36, with an unadjusted p-value of 0.572 and a confidence interval that includes zero.

The open-label group had a reported mean change of -13.6, but its comparison with placebo is separate from the primary masked contrast. The registry does not post sponsor analyses, a sample-size justification, or details of missing-data handling. Safety counts are reported for serious and other adverse events, with deaths not reported in the registry record.