A Study of Adjunctive Treatment to Antidepressant Therapy for Adults With Major Depressive Disorder: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Adjunctive Treatment to Antidepressant Therapy for Adults With Major Depressive Disorder (NCT00840034), a randomized phase 2 study. The primary outcome is Change From Baseline to 8 Weeks in Montgomery-Asberg Depression Rating Scale (MADRS).

2. Clinical Question

The trial was designed to study efficacy, safety, and tolerability in this specific partial-responder population. Its primary endpoint was change on the Montgomery-Asberg Depression Rating Scale (MADRS), a measure of depressive mood symptom severity. The comparison is between adjusted average changes, rather than a comparison of each participant’s individual change or a simple comparison of final scores.

The clinical question is therefore about the relative average change over the stated follow-up period, under the trial’s assigned treatment conditions. The registry record does not report results for every outcome listed as planned, so the available results do not answer every question about symptoms, functioning, fatigue, anxiety, or pain that the study considered. Nor does the record establish whether any observed average difference is important to individual participants.

3. Trial Design

The registry identifies this as a randomized, quadruple-masked, parallel-group phase 2 trial. Participants were assigned to LY2216684 or placebo. The study concerned adjunctive treatment for adults with major depressive disorder who had partially responded to an adequate course of SSRI treatment. The registry record does not provide fuller arm descriptions, so details beyond the named interventions and adjunctive purpose are not reported.

Enrollment was 227. The registry lists the study as completed, with a start date of 2009-02 and primary completion in 2010-01. Eli Lilly and Company is listed as the sponsor. These record details describe the study’s registration and conduct period; they do not by themselves establish how many participants contributed data to each particular outcome. For the primary MADRS results, the reported analysis group sizes are 62 for LY2216684 and 68 for placebo.

Random allocation is intended, in general, to make treatment groups comparable on average at the point of assignment. Masking is intended, in general, to limit the influence of treatment knowledge on participants’ experiences and on study assessments. The registry describes the masking as quadruple but does not identify which roles were masked. A parallel-group design compares treatment groups over the same study period rather than switching participants between the study drugs.

The primary assessment timeframe was baseline to 8 weeks. Secondary outcomes included other symptom, functioning, fatigue, and pain measures, with several also assessed over that timeframe. The registry record does not report the original sample-size justification, details of any interim analyses, or a full statistical analysis plan. Those omissions limit what can be said about how the planned design related to the reported analyses.

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: In the Acute Treatment Phase participants are randomized to either LY2216684 or placebo treatment groups. Participants who complete the Acute Treatment Phase or discontinue early after 4 or more weeks in the Acute Treatment phase are eligible to participate in the 2 week Taper Phase, but not all elected to participate.

Acute Treatment Phase

Reasons for not completing this period:

Taper Phase

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-treat population analysis is determined by the treatment groups to which participants are randomly assigned regardless of treatment received, and includes all participants who do not meet response criteria during Weeks 1-2, have baseline and at least 1 post-baseline MADRS Total Score.

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: All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline to 8 Weeks in Montgomery-Asberg Depression Rating Scale (MADRS), measured over Baseline, 8 weeks.

The primary endpoint is change from baseline to 8 weeks in the Montgomery-Asberg Depression Rating Scale (MADRS). The registry describes MADRS as a 10-item checklist measuring severity of depressive mood symptoms. Each item is rated on a scale of 0-6, and the total score ranges from 0 (low severity of depressive symptoms) to 60 (high severity of depressive symptoms). A change below zero therefore represents a reduction in the score from baseline.

MADRS is used as a structured measure of symptom severity, allowing the trial to compare average change between the assigned treatment groups. Looking at change rather than the endpoint score alone accounts for each participant’s starting measurement in the outcome definition; the model also adjusts for baseline score. The registry reports least-squares means from a mixed model repeating measures (MMRM), adjusted for investigator, treatment-by-visit, baseline score, and baseline-by-visit.

The reported least-squares means are model-adjusted group averages, not raw arithmetic averages. The primary analysis population was restricted to participants who did not meet response criteria during Weeks 1-2 and who had a baseline and at least one post-baseline MADRS Total Score. Thus, the endpoint’s interpretation depends both on the scale and on the registry’s stated analysis-population definition. The registry record does not report a threshold for a clinically important MADRS change, so the numerical change alone cannot establish how meaningful it would be to participants.

7. Results Posted in the Registry

• LY2216684: mean -8.13 versus -5.76 on control; difference -2.37 (95% CI -5.83 to 1.09), standardized effect d = -0.23; z-test p = 0.180.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change From Baseline to 8 Weeks in 16-Item Quick Inventory of Depressive Symptomatology-Self Rated (QIDS-SR16) Total Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • LY2216684: mean -4.71 versus -2.97 on control; difference -1.74 (95% CI -3.48 to -0.00), standardized effect d = -0.34; z-test p = 0.050.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for LY2216684 than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline to 8 Weeks in Clinical Global Impression of Severity (CGI-S)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • LY2216684: mean -0.74 versus -0.71 on control; difference -0.03 (95% CI -0.40 to 0.34), standardized effect d = -0.03; z-test p = 0.875.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between LY2216684 and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline to 8 Weeks in Massachusetts General Hospital Cognitive and Physical Functioning Questionnaire (CPFQ)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • LY2216684: mean -5.69 versus -2.62 on control; difference -3.07 (95% CI -5.38 to -0.76), standardized effect d = -0.46; z-test p = 0.009.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for LY2216684 than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry lists several secondary questions covering self-rated depressive symptoms, anxiety and depression, clinician-rated severity, cognitive and physical functioning, fatigue, and pain. Only some of those outcomes have numerical results in the record. The available results show that the direction and degree of separation between LY2216684 and placebo are not uniform across measures. That mixed pattern is useful context, but it does not replace the primary MADRS comparison.

Some posted secondary estimates point toward greater average reductions in the LY2216684 group, while the clinician-rated severity comparison is close between groups. This is broadly compatible with the primary point estimate’s direction in some measures, but the primary confidence interval includes no difference. Secondary results therefore cannot convert the uncertain primary result into a definitive finding. Measures can differ in what they capture, and an apparent pattern across outcomes is not itself a formal combined analysis.

The record gives unadjusted comparisons for the secondary results and does not report a procedure to control for the number of outcomes examined. In general, when many outcomes are tested, the chance of seeing an apparently notable result by chance increases unless the analysis accounts for multiple testing. An unadjusted secondary p-value should consequently be treated as exploratory evidence rather than a stand-alone confirmation.

The registry record does not report numerical results for every listed secondary outcome, so a comprehensive assessment of consistency across all planned measures is not possible. Nor are detailed secondary analysis populations or missing-data procedures reported. The available results can describe what was posted, but they cannot establish a coordinated effect across symptom, functioning, fatigue, and pain domains.

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

• Change From Baseline to 8 Weeks in 16-Item Quick Inventory of Depressive Symptomatology-Self Rated (QIDS-SR16) Total Score
• Change From Baseline to 8 Weeks in 16-Item Quick Inventory of Depressive Symptomatology-Self Rated (QIDS-SR16) Individual Items
• Change From Baseline to 8 Weeks in Hospital Anxiety and Depression Scale (HADS)
• Change From Baseline to 8 Weeks in Clinical Global Impression of Severity (CGI-S)
• Change From Baseline to 8 Weeks in Massachusetts General Hospital Cognitive and Physical Functioning Questionnaire (CPFQ)
• Change From Baseline to 8 Weeks in Fatigue Associated With Depression Participant-Reported Outcome (FAsD PRO)
• Change From Baseline to 8 Weeks in Visual Analog Scale for Fatigue (VAS-F)
• Change From Baseline to 8 Weeks in Visual Analog Scale for Pain (VAS-P)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: it is a score whose possible values fall along a numerical scale. The registry reports least-squares means and standard errors, rather than a count of participants who crossed a response threshold. In general, a group mean summarizes the average outcome, while a standard error describes the uncertainty in estimating that mean. Neither quantity describes the full range of individual participant experiences.

What does a least-squares mean represent?

The reported group values are least-squares (LS) means. In general, these are model-based averages adjusted for selected features of the analysis rather than unadjusted averages computed directly from the observed values. For the MADRS endpoint, the registry states that the MMRM was adjusted for investigator, treatment-by-visit, baseline score, and baseline-by-visit. This means the primary group estimates reflect the model and its specified adjustment terms. The registry record does not provide further model details, such as the covariance structure or the exact handling of all missing observations.

The reported MADRS LS mean changes were -8.13 for LY2216684 and -5.76 for placebo. The model-based difference was -2.37. Because lower MADRS scores represent less severe symptoms, a more negative change indicates a larger average decrease from baseline. The direction of that estimate is descriptive; whether it is sufficiently precise or important is a separate question.

Why use a mixed model repeating measures?

In general, a mixed model repeating measures can analyze repeated outcome measurements while accounting for their relationship within a participant. It can estimate treatment-group patterns over visits and incorporate baseline and other specified terms. The registry identifies this approach for the MADRS least-squares means and states that the adjustment included treatment-by-visit. That term permits the modeled treatment contrast to vary by visit rather than assuming that the treatment pattern is identical at every measurement occasion.

A model’s ability to use repeated measurements does not remove the need to understand missing data. In general, model-based estimates rely on assumptions about the observed data and the process that led to unobserved measurements. The registry record does not report enough detail to establish which missing-data assumptions were used or whether sensitivity analyses tested alternatives. The described primary analysis population required baseline and at least one post-baseline MADRS score, but this eligibility rule alone does not explain how missing observations within that population were handled.

What does the group comparison say?

The group comparison is a difference between the adjusted average changes. A negative difference means the estimate is more negative for LY2216684 than for placebo. The comparison is -2.37, with a confidence interval from -5.835 to 1.095. Since the interval includes zero, the data do not pin down the direction of the underlying group difference with high precision. The unadjusted p-value is 0.18. These quantities address different aspects of the evidence: the estimate gives the observed direction and scale, the interval describes statistical uncertainty, and the p-value assesses compatibility with a null comparison under a specified statistical framework.

The reported comparison is an unadjusted recalculation from posted summary statistics, not a substitute for any sponsor analysis. The discrepancy indicator is false, and no conflicting result is identified in the registry record. The record does not report a separate sponsor inferential analysis for this comparison, so the posted estimate and the stated primary model should be distinguished from unreported details of a full analysis plan.

How should standard errors and confidence intervals be read?

The primary group standard errors are 1.251 for LY2216684 and 1.249 for placebo. In general, a standard error measures uncertainty about an estimated quantity, not variation among individual participants. A confidence interval combines the estimate with its uncertainty to give a range of values compatible with the data under the method used. It is not a range containing a specified proportion of individual scores, and it is not a guarantee about the true value.

The interval around the difference runs from -5.835 to 1.095. Because that range crosses zero, the posted result does not clearly distinguish a negative difference from no difference or a positive difference. The interval is therefore central to interpreting the result: the point estimate alone would give an incomplete impression of the evidence.

What is not specified?

The registry record does not report the original sample-size justification, the detailed statistical analysis plan, interim analyses, or a multiplicity adjustment. It does not identify a missing-data sensitivity analysis for the primary endpoint. The method used to generate the reported comparison’s interval is not described beyond the available summary statistics and the primary model description. These omissions should be treated as limits on interpretation, not filled in by assuming a conventional protocol.

General statistical concepts can clarify the posted results, but they cannot supply unreported trial procedures. For instance, randomization generally supports a fair comparison at assignment, while analysis restrictions and missing outcome data can affect which participants contribute to a particular estimate. The registry’s defined primary population and the reported group sizes should therefore remain part of any reading of the result.

11. Confidence Intervals and Precision

The primary comparison is an estimated mean difference of -2.37 MADRS units, with a confidence interval from -5.835 to 1.095. The point estimate indicates a larger average reduction in the LY2216684 group, but the interval includes zero. It also extends into positive values, so the posted results do not rule out a difference in the opposite direction. The estimate should be read together with this range rather than as a precise value known to be the true treatment contrast.

In general, a confidence interval describes uncertainty around an estimate under the analysis method. It does not show the distribution of individual participant changes, and it does not give the probability that the true difference lies within the particular interval. Here, the interval spans values consistent with a range of possible comparisons, including no difference. This limits certainty about both the direction and magnitude of the underlying contrast.

The primary group standard errors are 1.251 and 1.249. They describe uncertainty in the estimated LS means; they are not standard deviations of participant outcomes. The registry record reports no confidence intervals for the individual group means in the arm entries, while the comparison interval is available. The interval’s width is best considered in the context of the MADRS scale and the absence of a reported clinical-importance threshold.

The unadjusted p-value of 0.18 is consistent with the interval’s inclusion of zero, but it should not replace the interval-based reading. A non-small p-value does not establish equivalence, and a point estimate favoring one group does not establish a dependable difference. The registry record does not provide an equivalence margin or a non-inferiority margin.

12. Planning a Similar Trial: Sample Size

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

The planning illustrations report participants per arm for continuous outcomes. They do not report total event counts; the registry’s sample-size assumptions distinguish the participant-per-arm format used for other outcomes from total-event calculations used for time-to-event outcomes. The listed outcomes discussed in these results are continuous measures.

The planning 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 at the stated fraction of the observed effect, with illustrations for 100%, 75%, and 50% of that effect. The original trial’s own sample-size assumptions and justification are not reported in the registry record; these planning assumptions are not evidence of what the investigators used when designing the trial.

The 80% power column means the number of participants per arm estimated to give an 80% chance of detecting the specified effect under those planning assumptions, if that effect is present. The 90% column expresses the corresponding planning target for a 90% chance. These are design calculations, not guarantees that a future study will produce a particular result. When the effect to detect is smaller, more participants are needed to distinguish it from random variation under the same assumptions.

These values are planning illustrations based on an observed effect, not observed power. They should not be read as a retrospective measure of whether the completed trial had enough power, nor as a reconstruction of the trial’s original design rationale. An effect estimated in a trial can be uncertain, so using it as the basis for planning does not make it a reliable prediction of the effect in another study.

13. How the Sample Size Is Derived

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

Formula

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 = -2.370; pooled standard deviation σ = 10.088.
2. n = 2 × (1.96 + 0.842)² × 10.088² / -2.370² = 284.4, rounded up to 285 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 <- -2.37                      # observed difference in means
sigma <- 10.0879717741291                     # 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(840034)
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 = 285 (80% power), 381 (90% power)
 75% of observed effect: n per arm = 506 (80% power), 677 (90% power)
 50% of observed effect: n per arm = 1,138 (80% power), 1,523 (90% power)
power.t.test: 285.4 
Simulated power at n = 285 per arm: 0.799 (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.

14. Multiplicity, Missing Data and Analysis Populations

The registry defines the primary analysis population as participants assigned to the treatment groups who did not meet response criteria during Weeks 1-2 and who had a baseline and at least one post-baseline MADRS Total Score. This definition is narrower than a simple statement that every randomized participant was analyzed. Although the registry calls the population intent-to-treat, it also specifies these response and measurement requirements. The reported primary group sizes are 62 for LY2216684 and 68 for placebo.

That population definition matters because estimates describe the participants meeting the stated criteria, rather than automatically describing everyone enrolled. The record does not explain in detail how participants with missing later measurements were handled within the analysis population. It also does not report a sensitivity analysis for missing data. In general, results can depend on assumptions about missing outcomes, especially when the probability of a missing measurement is related to participants’ health or treatment experience.

The registry lists multiple secondary outcomes and provides results for some of them. It does not report a multiplicity strategy or an adjustment of secondary comparisons for the number of outcomes assessed. When many hypotheses are examined, unadjusted p-values can overstate how much evidence a particular result offers. This makes secondary comparisons less conclusive, especially when they are considered individually and without a prespecified hierarchy.

The primary model is described as an MMRM adjusted for investigator, treatment-by-visit, baseline score, and baseline-by-visit. The registry record does not report the complete statistical analysis plan, including any additional model specifications or planned sensitivity analyses. Interim analyses and the original sample-size justification are also not reported in the registry record. These missing details prevent a full assessment of whether all analyses followed a prespecified plan or how choices about analysis were made.

These limitations do not invalidate the posted results, but they narrow what can be concluded from them. The primary estimate should be interpreted according to its stated population and model, and secondary findings should be viewed in light of the unreported multiplicity and missing-data procedures.

15. 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

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: haemoglobin decreased (1.1%), angina pectoris (0.9%), paraesthesia (0.9%), colon cancer (0.9%), suicidal ideation (0.9%).

The registry reports serious adverse events separately for the treatment and taper phases. During the treatment phase, 3 of 111 participants in the LY2216684 group and 2 of 116 in the placebo group were recorded as having a serious adverse event. During the taper phase, the corresponding figures were 1 of 90 for LY2216684 and 0 of 82 for placebo. These are the registry’s reported counts and denominators for the named groups and phases; the taper figures should not be combined with treatment-phase figures.

The registry lists serious adverse-event terms that include Haemoglobin decreased (1.1%), Angina pectoris (0.9%), Paraesthesia (0.9%), Colon cancer (0.9%), and Suicidal ideation (0.9%). These terms are not enough to determine whether an event was caused by a study drug, and the record does not provide further context for each listed term. Death counts are not reported in the registry record.

In general, adverse-event counts describe events recorded in the study, not proof of causation. The registry record does not provide the detail needed to compare the clinical circumstances, timing, or severity beyond the serious-event classification and listed terms. Safety figures should be considered alongside the trial’s duration and analysis population; they do not by themselves establish the overall safety profile of either study drug.

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

The primary result describes an adjusted average change in MADRS score over 8 weeks among the registry-defined analysis population. The LY2216684 estimate was more negative than the placebo estimate, indicating a numerically larger average reduction in the measured symptom score. The comparison’s confidence interval includes zero, however, so the observed difference is not estimated with enough precision to distinguish clearly between a difference favoring LY2216684, no difference, or a difference favoring placebo.

This result does not prove that the study drugs have identical effects. A confidence interval that includes zero is not an equivalence finding, and the registry does not report an equivalence margin. It also does not prove that the numerical advantage in the point estimate is a dependable treatment effect. The p-value of 0.18 and the interval both indicate uncertainty in the primary comparison.

The result does not establish how an individual participant would respond. Group averages can conceal variation among participants, and the record does not report individual response distributions for the primary outcome. It also does not give a threshold for a clinically important change on MADRS, so the size of the estimated difference cannot be classified from the record alone as meaningful or unimportant to participants.

The interpretation is limited to adults with major depressive disorder identified as partial responders to SSRI treatment, the stated analysis-population requirements, and the 8-week endpoint. It should not be extended automatically to other populations, other treatment settings, or longer follow-up. Secondary results add context but do not resolve the uncertainty in the primary result, particularly because the record does not report multiplicity adjustments or complete numerical results for every listed secondary outcome.

The most direct conclusion is that the posted primary estimate favors a greater average score reduction with LY2216684, but its uncertainty is substantial enough that the comparison remains inconclusive. This is a description of the registry results, not medical advice or a recommendation about treatment.

17. Why This Trial Matters Statistically

This trial illustrates why a treatment comparison should be read through its estimate and its uncertainty, rather than through the direction of the point estimate alone. The primary LS mean change was more negative for LY2216684, but the confidence interval for the difference extends across zero. The data therefore leave meaningful uncertainty about the direction and size of the group contrast.

It also demonstrates the importance of distinguishing model-adjusted estimates from raw averages. The primary outcome was analyzed with an MMRM and adjusted for investigator, treatment-by-visit, baseline score, and baseline-by-visit. Those choices define the reported estimate. A statistical reader needs both the numerical result and the description of the model that produced it to understand what is being compared.

The analysis-population definition is another important feature. The registry labels the population intent-to-treat but specifies exclusion of participants who met response criteria during Weeks 1-2 and requires baseline and at least one post-baseline MADRS measurement. That operational definition affects which participants contribute to the primary outcome and should be considered alongside the enrollment figure.

Finally, several secondary domains were assessed, but their results are not uniformly aligned and the registry record does not report a multiplicity procedure. This is a useful example of why multiple supportive-looking comparisons should not be treated as independent confirmation without considering the broader analysis plan. The record’s limits on missing-data methods and original design assumptions also show how incomplete statistical reporting can constrain interpretation even when key outcome estimates are available.

18. 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 does not report the original sample-size justification or design assumptions. The planning illustrations based on observed effects should not be mistaken for the trial’s own rationale. Without the original justification, it is not possible to assess from the record how the enrollment target related to the anticipated primary effect.

The primary analysis population has specific response and measurement requirements despite being described as intent-to-treat. It excludes participants who met response criteria during Weeks 1-2 and requires a baseline plus at least one post-baseline MADRS score. This limits a simple interpretation of the estimate as applying to every randomized participant. The record does not detail the handling of further missing observations or report sensitivity analyses for them.

Some secondary results are posted, but the registry record does not provide numerical results for every listed secondary outcome. It also does not report a multiplicity adjustment. This makes it difficult to assess the full consistency of the outcome pattern and reduces the weight that should be placed on individual unadjusted secondary findings.

The primary comparison has a confidence interval that includes zero, and the record does not report a clinical-importance threshold for the MADRS difference. Thus, the result is uncertain statistically and cannot be classified against a registry-reported threshold for clinical importance. The follow-up reported for the primary endpoint is 8 weeks, so longer-term outcomes are not established by these data.

The registry describes quadruple masking without identifying the roles masked. It does not provide fuller arm descriptions, detailed model specifications, a complete analysis plan, or information about interim analyses. These are reporting limitations; they should not be filled in by assumptions about the protocol.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study of Adjunctive Treatment to Antidepressant Therapy for Adults With Major Depressive Disorder — NCT00840034.
• Registry record last updated 2018-04-24; status last verified 2018-03.
• Results first posted 2018-04-24.
• 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.

22. Record Summary

This completed, randomized, quadruple-masked, parallel-group phase 2 trial studied LY2216684 as adjunctive treatment for adults with major depressive disorder who partially responded to SSRI treatment. The registry lists enrollment of 227 and identifies Eli Lilly and Company as the sponsor.

The primary endpoint was change from baseline to 8 weeks in MADRS score, analyzed using adjusted least-squares means from an MMRM. The estimated change was -8.13 with LY2216684 and -5.76 with placebo; the difference was -2.37, with a confidence interval from -5.835 to 1.095 and an unadjusted p-value of 0.18. The point estimate favors a greater reduction with LY2216684, but the interval includes no difference.

Some secondary results point in the same direction, but the available pattern is mixed. The registry record does not report a multiplicity procedure, complete missing-data details, or the trial’s original sample-size justification. Serious adverse-event counts are reported by treatment and taper phase; death counts are not reported in the registry record.