Paroxetine-referenced Study Evaluating Three Doses of DVS SR in Outpatients With MDD: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Paroxetine-referenced Study Evaluating Three Doses of DVS SR in Outpatients With MDD (NCT00445679), a randomized phase 3 study. The primary outcome is Percentage of Responders With a 50% or Greater Decrease From Baseline on the Hamilton Rating Scale for Depression, 17-item (HAM-D17).

2. Clinical Question

3. Trial Design

The registry describes a completed Phase 3, randomized, parallel-group study of DVS SR and paroxetine in adult outpatients with major depressive disorder. The official title calls the study multicenter, randomized, double-blind, and paroxetine-referenced. The registry's masking classification is triple. The enrollment recorded for the study is 807.

There were four treatment groups: DVS SR 50mg/day, DVS SR 100mg/day, DVS SR 200mg/day, and paroxetine 20mg/day. The listed primary outcome timeframe is 8 weeks. The primary result entries give 200 participants for DVS SR 50, 200 for DVS SR 100, 197 for DVS SR 200, and 192 for paroxetine 20. These outcome denominators are not necessarily interchangeable with enrollment or safety denominators.

The registry lists a start date of 2007-07 and a primary completion date of 2009-02. It identifies the sponsor as Wyeth is now a wholly owned subsidiary of Pfizer. The stated study purpose was to assess the safety, tolerability, and efficacy of DVS SR in participants with major depressive disorder.

Six secondary outcomes are listed, including clinician global-impression measures, MADRS, pain intensity, HAM-D6, and anxiety scores. The registry record does not report the randomization procedure, allocation concealment, treatment adherence, or the original statistical plan. The masking label and the title's double-blind description are both registry entries; further operational information about who was masked is 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: After a 4 to 21 day screening period, eligible subjects were treated for up to 8 weeks.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The intent-to-treat (ITT) population included all subjects randomly assigned to treatment who had a baseline HAM-D17 evaluation, took at least 1 dose of double-blind test article, and had at least 1 postbaseline HAM-D17 evaluation.

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 Percentage of Responders With a 50% or Greater Decrease From Baseline on the Hamilton Rating Scale for Depression, 17-item (HAM-D17), measured over 8 weeks.

The primary endpoint is the percentage of responders at 8 weeks, where response means a 50% or greater decrease from baseline on the HAM-D17. This is a binary outcome: each participant either meets the response definition or does not. The reported group percentages therefore describe the share classified as responders, rather than average change in symptom score.

The HAM-D17 is a standardized, clinician-administered rating scale assessing 17 items associated with major depression. Items use either a 3-point scale scored from 0 to 2 or a 5-point scale scored from 0 to 4; the maximum total score is 50. The endpoint uses change from each participant's baseline score, so the response definition is relative to that participant's starting measurement.

General statistical context: a responder endpoint can make a clinical threshold easier to interpret than a mean score change, but it also reduces the underlying score information to two categories. Results can depend on the chosen threshold and on how baseline and follow-up measurements are obtained. The registry record does not report a rationale for choosing this response threshold or additional details about endpoint assessment procedures.

7. Results Posted in the Registry

• DVS SR 100: 67.0% (134/200) versus 63.0% (126/200) on control; risk difference 4.0 percentage points (95% CI -5.3 to 13.3), risk ratio 1.06 (0.92–1.23); unadjusted two-proportion z-test p = 0.402.
• DVS SR 200: 62.9% (124/197) versus 63.0% (126/200) on control; risk difference -0.1 percentage points (95% CI -9.6 to 9.4), risk ratio 1.00 (0.86–1.16); unadjusted two-proportion z-test p = 0.991.
• Paroxetine 20: 66.1% (127/192) versus 63.0% (126/200) on control; risk difference 3.1 percentage points (95% CI -6.3 to 12.6), risk ratio 1.05 (0.91–1.22); unadjusted two-proportion z-test p = 0.515.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary, because a participant is classified as a responder or nonresponder using the specified HAM-D17 threshold. The registry reports the percentage of responders for each group. For the recalculated comparisons, the measure is a risk difference: the response proportion in one group minus the response proportion in its reference group. A positive estimate indicates a higher observed response proportion in the first group as compared with the reference; a negative estimate indicates a lower observed proportion. This is a general explanation of the measure, not a claim about clinical importance.

What does a risk difference convey?

A risk difference keeps the comparison on the absolute scale of proportions. It can be read as the separation between the observed group response proportions, while a relative measure such as a risk ratio expresses one proportion relative to another. The registry recalculations include risk differences, and the associated posted summary counts permit unadjusted comparisons. The recalculated risk differences and confidence limits are not repeated because the registry also reports its own analyses on a separate basis.

General statistical context: absolute and relative measures answer different questions. Neither is automatically more informative in every setting. Absolute differences can be easier to connect to the proportion of participants experiencing an outcome, while relative measures can look larger or smaller depending on the baseline risk. A complete interpretation considers the estimate, its uncertainty, the outcome definition, and the clinical context.

How should the p-values be read?

The recalculated unadjusted p-values for DVS SR 100, DVS SR 200, and paroxetine, each compared with DVS SR 50, are 0.402, 0.991, and 0.515. A p-value describes how compatible the observed data are with a specified null model, under the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. These values do not provide strong evidence against no difference under conventional interpretations, but they do not prove that the groups have identical response rates.

The p-values are labeled unadjusted. That matters because the registry includes multiple arms and multiple outcomes, while no multiplicity strategy is reported. If many hypotheses are tested, the chance of at least one apparently unusual result can increase. A prespecified multiplicity approach can address that issue, but its use in this trial is not reported in the registry record. The listed p-values should therefore be read as unadjusted recalculations, not as evidence that a study-wide multiplicity procedure was applied.

What is the role of confidence intervals?

A confidence interval gives a range of values reasonably compatible with the data and the method used to calculate it. It communicates precision as well as direction: a narrower interval represents greater precision, while a wider interval leaves more uncertainty about the size and direction of the effect. It is not a guarantee that the true effect lies in the interval, nor is it a probability statement about a particular treatment effect.

For each recalculated comparison, the confidence interval includes no difference. The intervals therefore do not distinguish the estimated contrast from no difference at the usual two-sided interpretation. They also extend into values on either side of no difference, so they do not establish that the groups are exactly alike. Judging whether the intervals exclude an important effect requires a meaningful margin, which is not reported for these recalculations.

Why distinguish the recalculations from the registry analyses?

The recalculations are simple, unadjusted comparisons from the posted response summaries. They do not reproduce any adjustment, model, stratification, or missing-data procedure that might have been used in the trial's own analysis. The registry lists sponsor analyses as non-inferiority or equivalence-type results, but the method field is not populated and no p-values are reported for those entries. Those analyses are therefore best described using their posted estimates and confidence intervals without inferring unreported procedures.

General statistical context: adjusted analyses can account for design features or prognostic factors, while unadjusted comparisons use the observed group summaries directly. The two approaches may differ. A prespecified trial analysis ordinarily has a different evidential role from a later reconstruction from summary statistics, but the registry record does not provide enough detail to assess the exact relationship between its sponsor analysis and the recalculation.

10. Confidence Intervals and Precision

The recalculated comparisons are imprecise enough that their confidence intervals include no difference and allow effects in either direction. This is consistent with the unadjusted p-values: 0.402 for DVS SR 100 versus DVS SR 50, 0.991 for DVS SR 200 versus DVS SR 50, and 0.515 for paroxetine versus DVS SR 50. The p-values and intervals address different aspects of the same uncertainty; neither alone shows that the groups are equivalent.

The registry's own reported comparisons with paroxetine give 95% confidence intervals of -12.62 to 6.37 for DVS SR 50, -8.50 to 10.25 for DVS SR 100, and -11.68 to 7.33 for DVS SR 200. Each includes zero. These intervals also span values on both sides of zero, indicating that the reported analyses leave uncertainty about the direction of the contrast. The registry record does not specify the clinical equivalence or non-inferiority margin needed to decide whether any of these ranges is sufficiently narrow for that purpose.

Precision depends on the amount of information and the variability in the data, as well as the analysis method. The outcome denominators listed for the primary results are 200, 200, 197, and 192 across the four groups. No confidence intervals for the individual response percentages are listed in the primary outcome entries. The sponsor's analysis methods, handling of missing endpoint measurements, and exact analysis population for those reported contrasts are not detailed enough in the registry record to reconcile the two sets of intervals fully.

General statistical context: failure to rule out no difference is not the same as showing equivalence. An equivalence claim typically requires a prespecified range of differences considered acceptably small, and the confidence interval must be interpreted against that range. The relevant margin is not reported in the registry record, so the reported results should not be recast as proof of equivalence.

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 planning figures are participants per arm, not total event counts. The outcome is a binary responder measure, and the planning entries use equal allocation between two arms. The stated planning assumptions are a two-sided significance level of 0.05 and a normal approximation; the effect to detect is the listed fraction of the observed effect. The trial's own original sample-size assumptions are not reported in the registry record.

The 80% power column gives the illustrative number of participants per arm for a design that would have an 80% chance of detecting the specified effect if that effect were real and the planning assumptions held. The 90% power column gives the corresponding illustrative number per arm for a 90% chance. These are prospective planning quantities, not a measure of what the completed study achieved. This is not observed power.

Rows planning for 100%, 75%, and 50% of the observed effect illustrate how the required size changes with the target effect. Smaller effects require more participants to detect under otherwise unchanged assumptions. For example, for the DVS SR 100 comparison, planning for 100% of the observed effect gives 2,231 participants per arm for 80% power and 2,986 participants per arm for 90% power; planning for 50% of that effect gives 9,041 and 12,103 participants per arm, respectively.

The planning figures are illustrations derived from the observed effect, not a reconstruction of the trial's original design decision. In particular, the estimated effect can be uncertain, and planning directly from an observed effect can produce very large or unstable estimates when that effect is close to no difference. The record's planning assumptions do not establish what target effect, power, or sample-size justification investigators used when designing the trial. The curve represents the same general relationship: under the stated assumptions, aiming to detect a smaller effect calls for a larger planned sample.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.630; treatment proportion p₁ = 0.670; average p̄ = 0.650.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6745 = 1.3221.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6739 = 0.5672.
4. n = (1.3221 + 0.5672)² / (0.0400)² = 2,230.8, rounded up to 2,231 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
p0 <- 0.63                          # control proportion
p1 <- 0.67                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(445679)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 2,231 (80% power), 2,986 (90% power)
 75% of observed effect: n per arm = 3,993 (80% power), 5,345 (90% power)
 50% of observed effect: n per arm = 9,041 (80% power), 12,103 (90% power)
power.prop.test: 2230.8 
Simulated power at n = 2,231 per arm: 0.798 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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. Analyses Reported by the Sponsor

The registry lists three sponsor analyses comparing a DVS SR dose with paroxetine. For DVS SR 50, the reported difference is -3.13 with a 95% confidence interval from -12.62 to 6.37. For DVS SR 100, the difference is 0.88 with a 95% confidence interval from -8.50 to 10.25. For DVS SR 200, the difference is -2.17 with a 95% confidence interval from -11.68 to 7.33. All three intervals include zero. The analyses are classified in the registry as non-inferiority or equivalence-type analyses, but their method and p-value are not reported.

The recalculated results use unadjusted risk differences from posted group summaries and compare DVS SR 100, DVS SR 200, and paroxetine with DVS SR 50. They are not the same contrasts or necessarily the same analysis as the sponsor's comparisons of each DVS SR dose with paroxetine. The registry record does not provide enough method detail to reconcile the estimates directly.

The discrepancy indicator is false, but the recalculations still should not replace the registry's own reported analyses. The sponsor results are the trial's reported comparisons; the recalculations are a separate, unadjusted reading of the posted summaries. Neither set of results establishes equivalence without the relevant prespecified margin and analysis details.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and six secondary outcomes. The secondary outcomes include CGI-I and CGI-S scores, MADRS total-score change, VAS-PI pain-intensity score change, HAM-D6 score change, and Covi Anxiety Scale score change. The record also indicates 7 outcome measures and 9 analyses. It does not report how the analyses were grouped or how hypotheses were prioritized beyond identifying the primary outcome.

A multiplicity plan describes how inference is handled when several outcomes, treatment comparisons, time points, or subgroup tests are examined. General statistical context: without an appropriate plan, interpreting each test separately can increase the overall chance of a false-positive finding. The multiplicity strategy for this trial, including any adjustment or hierarchy among outcomes, is not reported in the registry record. The recalculated p-values are explicitly unadjusted and should not be treated as multiplicity-controlled.

The primary outcome population is described as the intent-to-treat population, but its stated eligibility requires a baseline HAM-D17 evaluation, at least 1 dose of double-blind test article, and at least 1 postbaseline HAM-D17 evaluation. This definition is more specific than simply including everyone randomized. The registry record does not report how many randomized participants were excluded from this endpoint population, how missing follow-up scores were handled, or whether any imputation or sensitivity analyses were used.

The method fields for the sponsor's reported comparisons are not populated. Details about model specification, covariate adjustment, assumptions, and the analysis population for those entries are not reported in the registry record. The record also does not report an interim-analysis plan or procedures for handling protocol deviations. These omissions limit the ability to evaluate whether the posted endpoint results follow a prespecified analysis strategy or how robust they are to missing data and alternative analytic choices.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: suicidal ideation (1.0%), suicide attempt (1.0%), suicidal behavior (1.0%), drug abuse (0.5%), major depressive disorder worsening (0.5%).

The registry's safety entries list serious adverse events for each treatment group as counts among those at risk: 4 among 203 for DVS SR 50, 1 among 203 for DVS SR 100, 5 among 205 for DVS SR 200, and 2 among 196 for paroxetine 20. These are group-specific counts with their listed at-risk denominators. They should not be added across groups or interpreted as a comparison adjusted for exposure or follow-up time.

The registry also lists top serious adverse-event terms and percentages: suicidal ideation, suicide attempt, and suicidal behavior are each listed as 1.0%; drug abuse, major depressive disorder worsening, and cholelithiasis are each listed as 0.5%. The record does not provide further term-level details in these entries, such as group-specific counts or timing. The death counts are not reported in the registry record.

General statistical context: serious adverse-event counts are descriptive safety information. Comparing them requires attention to group sizes, follow-up, event definitions, and whether participants can contribute more than one event. The registry entries alone do not establish that a difference between groups is caused by treatment.

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

The posted primary results describe similar-looking response percentages across the four groups: 63.0% for DVS SR 50, 67.0% for DVS SR 100, 63.0% for DVS SR 200, and 66.0% for paroxetine 20. The recalculated unadjusted comparisons do not provide clear evidence of a difference from the DVS SR 50 group, and their confidence intervals include no difference. The registry's reported comparisons with paroxetine also have confidence intervals that include zero.

These findings do not show that the treatments are exactly equal. A non-significant test cannot prove no effect, and an interval that crosses zero can include effects in both directions. Nor can the posted results establish formal non-inferiority or equivalence without the margin and analysis details needed to evaluate those questions. The registry categorizes the sponsor's analyses as non-inferiority or equivalence-type analyses, but the details needed for an independent assessment are not reported.

The outcome is response on a clinician-administered depression scale at 8 weeks, not every aspect of symptoms, functioning, safety, or longer-term outcomes. The registry lists secondary measures, but their detailed results are not included in the registry record used for this interpretation. Results from adult outpatients in this study also do not, by themselves, describe outcomes in populations or settings not represented in the registry record.

General statistical context: statistical uncertainty and clinical importance are different considerations. A range of effects can be statistically compatible with the observed data, while only some of those effects may matter clinically. The registry record does not report a clinically important difference threshold for the recalculated comparisons. The appropriate conclusion is therefore limited to the posted response proportions, their uncertainty, and the stated analysis limitations; it is not a treatment recommendation.

17. Why This Trial Matters Statistically

This study offers a useful example of why the endpoint and comparison must be read together. The primary result is a responder percentage based on a defined change from baseline, while the recalculated comparisons express contrasts in those proportions. A group percentage alone is descriptive; a comparison estimate and its interval address the uncertainty around differences between groups.

It also illustrates the distinction between a trial's reported analysis and a reconstruction from summary results. The registry includes sponsor analyses against paroxetine and allows unadjusted recalculations for comparisons against DVS SR 50. Those analyses answer different comparison questions and are not interchangeable. Clear reporting of methods, populations, missing-data handling, and margins is important when interpreting claims framed as non-inferiority or equivalence.

The planning illustrations demonstrate how sample-size requirements depend on the effect a future study aims to detect. A smaller target effect requires a larger planned sample under the stated assumptions. Because the observed contrasts are near no difference for some comparisons, planning directly from them can yield very large illustrations; these values do not reveal the original design rationale.

Finally, the record highlights important limits of registry reporting. It identifies the outcome timeframe, response rule, primary population, treatment groups, and selected safety counts, but does not report key details about the statistical plan or missing data. For biostatisticians and clinical researchers, separating what is directly reported from what can be inferred is central to a transparent reading of trial results.

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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry record gives limited detail about the analysis methods. The sponsor-analysis method fields are not populated, and the record does not specify the model, covariate adjustments, or full analysis populations for those comparisons. The original sample-size justification and design assumptions are also not reported, so the planning illustrations cannot be treated as a description of the study's initial design.

The primary endpoint population requires baseline and postbaseline HAM-D17 assessments and at least 1 dose of study treatment. The number of randomized participants without qualifying endpoint data, reasons for missing assessments, and missing-data procedures are not reported in the registry record. These details matter for understanding how the analyzed population relates to random assignment and how sensitive the estimates may be to missing observations.

The listed primary response percentages do not include confidence intervals in the outcome entries. Recalculated uncertainty is based on posted summaries and is unadjusted. The registry's sponsor comparisons use a different stated comparison framework, but no non-inferiority or equivalence margin is reported. The two sets of results consequently cannot be fully reconciled from the registry information alone.

The primary outcome is measured at 8 weeks, and the registry record does not report longer-term results in the registry record used. Although secondary outcomes are named, their numerical results are not included. The safety entries provide serious-event counts and selected serious terms, but not detailed event timing, exposure-adjusted comparisons, or death counts. The registry's masking classification is triple while its official title says double-blind; details that would explain the terminology are not reported in the registry record.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Paroxetine-referenced Study Evaluating Three Doses of DVS SR in Outpatients With MDD — NCT00445679.
• Registry record last updated 2013-11-04; status last verified 2013-10.
• Results first posted 2013-11-04.
• 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

NCT00445679 is a completed Phase 3 randomized, parallel-group study in adult outpatients with major depressive disorder. It compared DVS SR at 50mg/day, 100mg/day, and 200mg/day with paroxetine 20mg/day. The registry lists enrollment of 807, a start date of 2007-07, and primary completion in 2009-02.

The primary endpoint was the percentage with a 50% or greater decrease from baseline on HAM-D17 at 8 weeks. Posted response percentages were 63.0%, 67.0%, 63.0%, and 66.0% for DVS SR 50, DVS SR 100, DVS SR 200, and paroxetine 20, respectively. Recalculated unadjusted comparisons do not establish a difference, and the registry's reported comparison intervals include zero. Missing-data procedures, detailed sponsor methods, and the non-inferiority or equivalence margin are not reported in the registry record.