Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care for Depression: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care for Depression (NCT01407094), a randomized phase 2 study. The primary outcome is Hamilton Rating Scale for Depression.

2. Clinical Question

The trial was designed to examine markers and biosignatures related to antidepressant response, with the stated aim of informing a Depression Treatment Response Index. The posted result addressed is narrower: a continuous symptom-severity score at Week 8 for sertraline and placebo. It does not report whether any marker predicted or moderated treatment response, nor does the primary comparison establish the usefulness of an index for matching individual patients to treatments.

3. Trial Design

The registry identifies this as a Phase 2, randomized, quadruple-masked, sequential-model study, sponsored by the University of Texas Southwestern Medical Center. Enrollment is listed as 296, and the study status is Completed. The recorded start date is 2011-07-29, and primary completion is listed as 2016-04. The registry names Depression as the condition.

Sertraline, placebo, and BupropionXL appear in the intervention list. The arm descriptions identify sertraline as SSRI monotherapy and placebo as placebo control; BupropionXL is listed as the bupropion arm. The posted primary-outcome comparison is sertraline versus placebo. The registry does not post a Week 8 primary-outcome mean or standard deviation for BupropionXL, and the corresponding listed outcome sample size is 0. The record therefore does not support a quantitative interpretation of bupropion against either of the other groups from this endpoint.

Randomized allocation means treatment assignment was intended to be assigned by chance. Quadruple masking indicates that the registry labels the study as masked at four roles, but the identities of those roles are not reported in the registry record. The sequential model is recorded, but the operational details of that model are not reported in the registry record. These design labels alone do not specify how allocation was implemented, how masking was maintained, or how the sequence of study procedures was organized.

The primary endpoint is the mean Hamilton Rating Scale for Depression score at Week 8. The registry describes an efficacy population of participants who completed eight weeks: 115 in the sertraline group and 123 in placebo. It separately posts outcome group sizes of 146 and 150, respectively. The record does not explain the relationship between these outcome denominators and the completer counts. The planned original sample-size justification and the protocol’s own analysis assumptions 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: Final enrollment Stage 1: 296 (146 SERT, 150 PBO). Individuals who responded to their Stage 1 treatment remained on that treatment in Stage 2. Individuals who did not respond to their Stage 1 treatment were switched to a different treatment in Stage 2.

Stage 1

Stage 2

Analysis population for the primary outcome, as defined in the registry: Note subjects in the efficacy trial were those that completed 8 weeks of treatment, 115 in the sertraline group, and 123 in the placebo group. SERT had 34 wash-outs, PBO had 27.

5. Baseline Characteristics

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

Population: Stage 1 is made up of Treatment A (SERT) and Treatment B (PBO). Non-responders to Treatment A were switched in Stage 2 to BUP, and non-responders in Treatment B were switched to SERT. Between randomization and baseline, 3 participants didn't meet inclusion criteria, hence the discrepancy between official enrollment and baseline population numbers.

6. Primary Endpoint

The registry lists the primary outcome as Hamilton Rating Scale for Depression, measured over Week 8.

The primary endpoint is the Hamilton Rating Scale for Depression score at Week 8. The registry describes the measure as a rating of depressive severity, also identifying it as HAM-D17 or HDRS. Scores range from 0-52, with lower scores indicating less depressive symptomatology. This is a continuous outcome, and the posted group summaries are means and standard deviations.

A group mean describes the average score among the participants included in that group’s outcome summary. Comparing means gives a direct summary of how the groups’ average recorded symptom severity differed at the specified time. The mean difference is reported as sertraline minus placebo: a negative value means the posted sertraline mean is lower than the posted placebo mean. This direction describes the score contrast only; it does not by itself establish the reliability, clinical importance, or cause of the difference.

Week 8 is the endpoint timeframe stated in the registry. A single time-specific score can summarize the status of participants at that point, but it does not describe the full course of symptoms or changes at other visits. Although the study’s stated purpose includes investigating markers and treatment response, the posted primary endpoint is a symptom scale score, not a biosignature, marker, or treatment-response index. Marker-specific results are not reported in the registry record.

The registry notes that the efficacy-trial subjects were those who completed eight weeks of treatment, while the posted outcome summaries list different group sizes. Because the record does not explain this difference, the endpoint summaries should be interpreted in light of that unresolved population detail rather than assumed to represent one clearly specified analysis population.

7. Results Posted in the Registry

• Sertraline: mean 11.06 versus 12.52 on control; difference -1.46 (95% CI -3.10 to 0.18), standardized effect d = -0.20; z-test p = 0.081.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a continuous score, and the registry reports a mean and standard deviation for each of the two groups with posted results. The comparison is expressed as a difference in means. This scale of analysis retains the original score units and addresses how far apart the group averages are. The sign convention is apparent from the posted means: the reported difference is sertraline minus placebo, so a negative estimate corresponds to a lower sertraline mean.

What does a mean difference describe?

A mean difference summarizes the contrast between two group averages. It is not the average change for each participant unless the analysis specifically uses participant-level changes, and no such change-score analysis is reported in the registry record. Here, the reported endpoint is a Week 8 score. The outcome description states that lower scores indicate less depressive symptomatology, so the direction of the difference can be described in relation to that scoring rule. The difference remains a summary of group means; it does not say that every participant in one group scored lower than every participant in the other.

The standard deviation describes how spread out individual scores are around a group mean. The posted standard deviations are 6.71 for sertraline and 7.68 for placebo. These are not standard errors and are not confidence intervals. A standard deviation characterizes variation among participant scores, whereas a standard error characterizes uncertainty in an estimated quantity, such as a mean or mean difference. The registry does not post participant-level data, so the distribution of scores cannot be evaluated from these summaries alone.

What does the confidence interval add?

A confidence interval gives a range of values that are compatible with the observed data under the statistical procedure and its assumptions. In general, a confidence interval is a way to express estimation uncertainty, not a range containing a specified proportion of individual participants and not a probability statement about the fixed true effect. The reported interval for the mean difference is -3.102 to 0.182. Since it includes zero, the result does not rule out a zero mean difference at the interval’s confidence level. The confidence level itself is not reported in the registry record.

The posted unadjusted p-value is 0.081. In general, a p-value measures how unusual data at least as extreme as those observed would be under a specified null hypothesis and model. It is not the probability that the null hypothesis is true, nor does it measure the importance of a difference. The record labels this p-value unadjusted. Whether any adjustment was planned or applied for multiple comparisons is not reported in the registry record. No other primary-outcome analyses are posted in this record.

How should the group sizes be understood?

The outcome summaries list 146 participants for sertraline and 150 for placebo. The endpoint description also states that the efficacy-trial subjects were those completing eight weeks, with 115 in sertraline and 123 in placebo, and lists wash-outs of 34 and 27. These counts describe distinct pieces of registry information, but the record does not explain how they align. It is therefore not possible to identify from this record alone whether the posted means and comparison use the completer population, another population, or a different handling of missing endpoint observations. The analysis population definition and missing-data methods are not reported in the registry record.

The arm-level standard deviations are not labelled as approximate in the comparison details. The posted mean-difference estimate and confidence limits are available as registry results; the calculation method, model specification, covariate adjustment, and assumptions used for the sponsor’s analysis are not reported in the registry record. The unadjusted recalculation information does not replace the sponsor’s reported analysis. There is no discrepancy flagged between the posted sponsor result and the simple recalculation.

What is over-dispersion, in general?

Over-dispersion means that observed variability is greater than a statistical model assumes. When a model understates variability, standard errors can be too small and intervals can be too narrow, making evidence appear more precise than it is. That issue can arise in some outcome models, particularly when the outcome’s variability is not adequately represented by the assumed distribution. The primary measure is a continuous score reported with standard deviations. The registry does not report an over-dispersion assessment, and the posted summary statistics alone do not establish whether it was relevant to the sponsor’s analysis.

What is not specified about the analysis?

The registry record does not report the original statistical analysis plan, the original sample-size justification, the prespecified significance threshold, whether interim analyses were conducted, or how any such analyses would have affected inference. It also does not report the exact test or model used for the primary comparison, whether baseline scores or other covariates were included, or the handling of protocol deviations. The trial is labelled randomized, quadruple-masked, and sequential, but those design labels do not supply these missing analytical details.

The primary endpoint is one posted continuous outcome, and the registry reports zero posted secondary outcome measures and zero posted analyses. This absence of additional posted results limits assessment of whether other planned endpoints or subgroup and marker analyses were conducted. It should not be interpreted as evidence that no such analyses took place; their status is not reported in the registry record. In general, statistical interpretation depends not only on a point estimate but also on its uncertainty, the analysis population, the model, and whether the analysis corresponds to a prespecified question. Several of those details are unavailable in this record.

10. Confidence Intervals and Precision

The registry reports a mean difference of -1.46 for sertraline relative to placebo, with a confidence interval from -3.102 to 0.182. The interval is the main indication of precision for the estimated group contrast. It spans zero, so the data do not narrowly identify a difference in one direction. Values on the negative side correspond to a lower average score for sertraline under the stated subtraction order; the interval also reaches slightly into positive values.

In general, a confidence interval expresses uncertainty in an estimate under the method used to construct it. It does not show where most individual participant scores fall, and it should not be read as a probability that the true value lies in this particular realized interval. Its width indicates that the reported point estimate is not highly precise. The registry does not report the confidence level or the detailed procedure used to calculate the interval, so those features cannot be assessed from the record.

The outcome standard deviations are 6.71 for sertraline and 7.68 for placebo. They describe score variation within the groups, not the uncertainty in the mean difference. The outcome summary lists 146 and 150 participants, while the endpoint description identifies 115 and 123 eight-week completers. Because the record does not explain the relationship between these denominators and the reported analysis, the precision cannot be fully interpreted in relation to a clearly defined analysis population.

The unadjusted p-value is 0.081. It complements the interval but does not replace it: the interval shows a range of compatible mean differences, while the p-value summarizes evidence relative to a null value under the test assumptions. Neither quantity addresses whether the score difference is important to patients, and the registry record does not report a threshold for clinical importance. The most defensible description is that the estimate is negative but uncertain, with an interval that includes zero.

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 planning display is an illustration based on the observed mean difference, not the trial’s original design calculation. The figures are participants per arm, not total participants or total events. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The target effect is set to the stated fraction of the observed effect. The original trial’s own sample-size assumptions and justification are not reported in the registry record.

The 80% power column describes the number of participants per arm in the planning calculation for a design that would have an 80% chance of detecting the specified effect under the listed assumptions, if that effect and model assumptions were correct. The 90% power column gives the corresponding planning count for a 90% chance. These are planning concepts, not guarantees about a future study’s result; actual power depends on the true effect and whether the assumptions hold.

The rows use 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. For 100%, the illustration lists 384 participants per arm for 80% power and 514 per arm for 90% power. For 75%, it lists 683 per arm for 80% power and 914 per arm for 90% power. For 50%, it lists 1,535 per arm for 80% power and 2,055 per arm for 90% power. Smaller effects need more participants to detect them under the same assumptions, which is why the required counts increase as the target fraction decreases.

This is not observed power. The figures are hypothetical planning illustrations using the observed estimate as their starting point; they do not tell how much power the completed trial had or whether the treatment effect is real. Because the observed estimate itself is uncertain, planning around a fraction of it should not be mistaken for a precise estimate of the effect that a future study would encounter.

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 = -1.460; pooled standard deviation σ = 7.218.
2. n = 2 × (1.96 + 0.842)² × 7.218² / -1.460² = 383.7, rounded up to 384 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 <- -1.46                      # observed difference in means
sigma <- 7.21790859056616                     # 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(1407094)
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 = 384 (80% power), 514 (90% power)
 75% of observed effect: n per arm = 683 (80% power), 914 (90% power)
 50% of observed effect: n per arm = 1,535 (80% power), 2,055 (90% power)
power.t.test: 384.6 
Simulated power at n = 384 per arm: 0.796 (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 zero posted secondary outcome measures, with zero analyses posted. The primary outcome is the Week 8 Hamilton Rating Scale for Depression mean. The comparison reports an unadjusted p-value of 0.081. The record does not report whether the trial had other prespecified outcomes, whether additional analyses were conducted, or whether any multiplicity adjustment was planned or applied. The absence of posted secondary results does not establish that no secondary analyses occurred.

Multiplicity arises when several hypotheses are tested. In general, as the number of tests increases, the chance of obtaining at least one apparently notable result by chance can increase unless the analysis plan accounts for this. Whether that concern applies, and how it was handled, cannot be determined from the registry record. The p-value is explicitly unadjusted, so it should not be described as adjusted for a family of comparisons.

The endpoint description identifies the efficacy-trial subjects as participants who completed eight weeks: 115 in the sertraline group and 123 in the placebo group. It also reports wash-outs of 34 and 27. The arm-level outcome summaries instead list 146 and 150 participants. The record does not explain the relationship among these numbers. It does not specify whether the primary analysis used all randomized participants, participants with observed Week 8 scores, completers, or another analysis population.

Missing-data handling is not reported in the registry record. In general, missing outcome data can affect interpretation if the availability of scores is related to treatment assignment or to participant outcomes. Different methods can make different assumptions about unobserved scores. Without a stated method or a clear analysis population, it is not possible to assess how missing Week 8 observations were handled or how sensitive the result may be to those assumptions.

The registry also does not report whether the analysis adjusted for baseline score or other participant characteristics, how protocol deviations were handled, or whether any subgroups or biomarker interactions were tested. Although the study’s stated aim concerns moderators and mediators, no marker-specific analysis results are posted in the record. These gaps limit evaluation of the full analysis strategy and of the potential for selective reporting.

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: seizure (1.9%), pregnancy (1.0%), er visit (0.5%), suicidality (0.5%), panic attack (0.5%).

The serious adverse event listings give the number at risk and the number affected for each group. For sertraline, the record lists 283 at risk and 10 affected; for placebo, 196 at risk and 7 affected; for BupropionXL, 53 at risk and 1 affected. These are the recorded counts and denominators. They are not the Week 8 endpoint sample sizes, and the registry does not explain the relationship between these safety populations and the primary-outcome populations.

The record also lists serious adverse event terms and percentages: Seizure, 1.9%; Pregnancy, 1.0%; ER Visit, 0.5%; Suicidality, 0.5%; Panic attack, 0.5%; and Hospitalization, 0.4%. The listed terms are not assigned to particular treatment groups in the record, so group-specific comparisons should not be inferred. The percentages are reproduced as reported and should not be combined or treated as an overall event rate.

Deaths are not reported in the registry record. The counts describe reported serious adverse events, not a complete account of every safety outcome or the severity and timing of each event. The record does not provide enough detail to compare safety across groups or to establish that a treatment caused an event. These listings are descriptive registry entries and should be interpreted with their population and reporting limitations in mind.

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

The posted primary result describes average Week 8 symptom scores in the sertraline and placebo groups. The sertraline mean was 11.06 and the placebo mean was 12.52; the reported mean difference was -1.46. Since lower Hamilton Rating Scale for Depression scores indicate less depressive symptomatology, the estimate’s direction is toward a lower group mean for sertraline. The confidence interval, from -3.102 to 0.182, includes zero, so the result remains uncertain about the direction and size of the underlying mean difference.

This finding is a group-level comparison, not a prediction of how any individual participant would respond. Group averages can conceal substantial variation among participants, and the posted standard deviations show that scores varied within both groups. The record does not report an individual-response analysis, a clinically important difference threshold, or a marker-based prediction result. It therefore cannot establish how the observed mean contrast should be interpreted for a particular person.

The study’s stated purpose was to examine clinical and biological markers of antidepressant response and to inform development of a Depression Treatment Response Index. That purpose is not itself evidence that the index was completed, validated, or useful. The registry record reports no marker-specific results, no measure of predictive performance, and no outcome showing that using an index improves treatment selection. Those conclusions are not supported by the posted primary endpoint alone.

The available primary outcome result is limited to the sertraline-placebo comparison at Week 8. BupropionXL is listed as an intervention, but its primary-outcome sample size is 0 and its mean and standard deviation are absent. The record therefore does not support a corresponding estimate for that group. It also does not establish longer-term outcomes or changes at time points other than Week 8.

Finally, the endpoint description’s completer counts differ from the outcome-summary counts, and the registry does not explain how the populations align. Missing-data procedures and the original statistical analysis plan are not reported in the registry record. These gaps mean that even a careful reading of the estimate must remain bounded by uncertainty about the analyzed population and methods. The result describes what the registry posts; it should not be generalized beyond that evidence.

16. Why This Trial Matters Statistically

This trial is statistically notable because its stated aim extends beyond comparing average symptom scores. The study was designed to examine clinical and biological markers as possible moderators and mediators of antidepressant response, with the goal of informing a Depression Treatment Response Index. That aim raises questions about prediction, treatment-effect heterogeneity, and how multiple sources of information might be combined. However, the registry record posts no marker-specific analyses or index-performance results, so the statistical ambition should be distinguished from the evidence available for interpretation.

The primary result illustrates why estimation is more informative than focusing only on a single test result. The mean difference is -1.46, while the confidence interval extends from -3.102 to 0.182 and includes zero. The point estimate gives a direction and scale in the original outcome units; the interval communicates uncertainty around that estimate. The unadjusted p-value of 0.081 does not, by itself, determine whether the result is important or precise.

Population definitions matter as much as the model output. The posted group summary counts are 146 and 150, whereas the endpoint description identifies 115 and 123 eight-week completers and also reports wash-outs. Since the record does not reconcile these figures, readers cannot confidently connect the displayed precision to a single clearly described analysis population. This is a practical reminder that an outcome estimate should be read together with its denominator and missing-data explanation.

The planning illustration also demonstrates the dependence of study size on the target effect: smaller target effects require more participants per arm under the stated assumptions. Such calculations are useful for planning, but the observed effect is uncertain and the illustration is not evidence of the completed trial’s power. Overall, the record gives a limited but interpretable symptom-score contrast, while leaving the central biomarker questions unanswered in the posted results.

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 important details unresolved. The primary endpoint description says that efficacy subjects completed eight weeks, with 115 in the sertraline group and 123 in placebo, and lists wash-outs of 34 and 27. Yet the posted outcome summaries list 146 and 150 participants. The record does not reconcile these counts or state which population generated the posted means and comparison. This limits assessment of the endpoint denominator and how participant exclusions or missing scores may have affected the estimate.

Although BupropionXL is listed as an intervention, its primary outcome sample size is 0 and no mean or standard deviation is posted. The reported primary comparison is therefore limited to sertraline and placebo. The study’s stated interest in clinical and biological markers is not accompanied by posted marker-specific findings, moderator or mediator analyses, or validation results for the proposed Depression Treatment Response Index.

The original sample-size justification, statistical analysis plan, exact model, confidence level, missing-data procedure, and any interim analysis details are not reported in the registry record. The record also does not explain whether baseline adjustment or other covariate adjustment was used. The p-value is labelled unadjusted, but the record does not state how multiplicity across any other planned comparisons was handled.

The outcome is reported at Week 8, so the posted result does not describe longer-term scores or the trajectory of scores over time. The record does not give a threshold for clinical importance, and the group mean contrast cannot establish individual response. Safety listings also have a different set of group denominators from the primary outcome summaries; the record does not explain how these populations relate. These limitations do not negate the posted result, but they restrict the conclusions that can be drawn from it.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care for Depression — NCT01407094.
• Registry record last updated 2018-12-26; status last verified 2018-12.
• Results first posted 2018-12-26.
• 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.

21. Record Summary

The registry describes a completed Phase 2, randomized, quadruple-masked, sequential-model study of depression, sponsored by the University of Texas Southwestern Medical Center. Enrollment is listed as 296. The recorded start date is 2011-07-29, and primary completion is listed as 2016-04. Interventions include sertraline, placebo, and BupropionXL.

The posted primary outcome is the Week 8 Hamilton Rating Scale for Depression mean. The sertraline mean was 11.06 (standard deviation 6.71) and the placebo mean was 12.52 (standard deviation 7.68). The reported sertraline-minus-placebo mean difference is -1.46, with a confidence interval from -3.102 to 0.182 and an unadjusted p-value of 0.081. The interval includes zero, indicating uncertainty about the mean difference.

The endpoint description gives completer counts of 115 and 123, while the outcome summaries list 146 and 150; the record does not explain the difference. No primary outcome result is posted for BupropionXL, and marker-specific findings are not reported in the registry record.