Postpartum Depression: Transdermal Estradiol Versus Sertraline: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Postpartum Depression: Transdermal Estradiol Versus Sertraline (NCT00744328), a randomized phase 4 study. The primary outcome is To Test the Efficacy of Estradiol for the Treatment of Postpartum Depression - Percent Change in SIGH-ADS29.

2. Clinical Question

3. Trial Design

The registry identifies this as a Phase 4, randomized, parallel-group study with quadruple masking. The listed treatment groups were transdermal estradiol, sertraline, and placebo. Enrollment was 85. The study began in 2008-08, and its primary completion date is 2013-09. Its current registry status is terminated; the reason for termination is not reported in the registry record.

The treatment assignments were designed to look alike in both delivery forms. Women in the estradiol group wore a weekly estradiol patch and took inactive capsules. Women in the sertraline group wore an inactive patch and took sertraline capsules. The placebo group wore an inactive patch and took inactive capsules. The sertraline dose range was 25 to 200mg/day. The estradiol patch dose range was 50 to 200 mcg/day. Details about how doses were selected or adjusted are not reported in the registry record.

The primary outcome summaries include Estradiol, with 26 participants; Sertraline, with 30; and Placebo, with 29. These are the group sizes attached to this outcome, not a description of the number who completed the study. The registry record does not report additional details about the randomization sequence, allocation concealment, masking procedures, or the analysis populations. It lists infant serum concentrations of estradiol in the three treatment arms as a secondary outcome, but secondary outcome results are not reported in the posted results.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analyses presented are Last Observation Carried Forward. For women who completed the 8 week trial the percent change was measured from baseline to week 8. For non-completers, the percent change was measured from baseline to last observation.

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 To Test the Efficacy of Estradiol for the Treatment of Postpartum Depression - Percent Change in SIGH-ADS29, measured over Week 8.

The primary endpoint is percent change in SIGH-ADS29 score at Week 8. SIGH-ADS29 is the Structured Interview Guide for the Hamilton Depression Rating Scale - Atypical Depression Symptoms Version. It incorporates the 17 and 21-item Hamilton Rating Scales for Depression and 8 atypical symptoms of depression. Scores range from 0 to 90; higher scores correspond to higher levels of depressive symptomatology.

Percent change expresses the score change relative to baseline, rather than the score at Week 8 alone. A negative percent change in the posted summaries therefore represents a decrease from baseline in the measured score. The registry reports the mean and standard deviation for each treatment group. A group mean describes the average change, while the standard deviation describes how spread out participant changes were around that average. This endpoint is useful for comparing average symptom-score change across the randomized groups, while the spread and uncertainty around comparisons matter for judging how precisely those averages distinguish the groups.

The analysis population is described as last observation carried forward. For women who completed the 8 week trial, percent change was measured from baseline to Week 8. For non-completers, the change was measured from baseline to the last observation. The registry does not report additional details about the timing or extent of missing measurements.

7. Results Posted in the Registry

• Estradiol: mean -38.00 versus -48.00 on control; difference 10.00 (95% CI -6.89 to 26.89), standardized effect d = 0.32; z-test p = 0.246.
• Sertraline: mean -49.00 versus -48.00 on control; difference -1.00 (95% CI -14.86 to 12.86), standardized effect d = -0.04; z-test p = 0.888.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a continuous measure: percent change in SIGH-ADS29 score. The registry reports a mean and standard deviation for each arm. A mean is the arithmetic average of the observations included in the analysis. A standard deviation describes dispersion around that average; it is not the uncertainty of the mean difference itself. In general, larger dispersion makes an arm average less precise when the group size is held constant.

What does the mean difference represent?

The posted comparisons are unadjusted recalculations from the arm-level means and standard deviations. Each mean difference is expressed as the first named treatment group minus Placebo. For Estradiol, the comparison is 10.0; for Sertraline, it is -1.0. These estimates retain the endpoint's scale, percentage change in SIGH-ADS29 score. They summarize differences in group averages, not differences for every participant and not an adjusted estimate controlling for other variables.

The comparison calculations use pooled standard deviations of 31.7 for Estradiol versus Placebo and 27.05 for Sertraline versus Placebo. A pooled standard deviation combines information about within-group spread to represent variability for a two-group comparison. The corresponding standardized estimates, d = 0.316 and d = -0.037, express the mean differences relative to pooled spread. Standardization can make effects from different scales easier to compare in general, but it does not remove uncertainty, establish clinical importance, or adjust for confounding. The registry does not report a clinically important threshold for this endpoint.

How should the intervals and p-values be read?

The intervals quantify uncertainty around the recalculated mean differences. In general, an interval gives a range of estimates compatible with the data and the assumptions of the calculation. Both posted intervals include zero: Estradiol versus Placebo ranges from -6.895 to 26.89, while Sertraline versus Placebo ranges from -14.86 to 12.86. The confidence level for these intervals is not reported in the registry record, so it should not be inferred from the values alone.

The unadjusted p-values are 0.246 for Estradiol versus Placebo and 0.888 for Sertraline versus Placebo. A p-value measures how unusual the observed data, or more extreme data, would be under a specified null model; it is not the probability that a treatment works, nor the probability that the null hypothesis is true. These recalculations are unadjusted. They do not account for testing two treatment comparisons, the secondary outcome, or any other analyses that may have been conducted. The registry lists 0 analyses, and a formal analysis plan is not reported.

What does last observation carried forward mean?

For participants completing the 8 week trial, the endpoint uses change from baseline to Week 8. For non-completers, the registry says the change uses baseline and the last observation. This approach is described as last observation carried forward. In general, carrying a participant's last observed measurement forward treats it as a substitute for the missing later measurement. That can produce a result different from an analysis that models measurements over time or uses a different missing-data assumption. It can also understate or overstate uncertainty if the last observation is not representative of the missing outcome. The number of non-completers and the reasons for missing observations are not reported in the registry record.

Which assumptions remain unknown?

The registry record does not report the confidence level of the comparison intervals, whether the sponsor conducted an adjusted analysis, the model or test used for any sponsor analysis, or the analysis population beyond the endpoint's last-observation-carried-forward description. Details of randomization implementation and any prespecified multiplicity strategy are also not reported. These omissions limit assessment of whether the recalculated comparisons reproduce the trial's intended primary analysis. They should be understood as unadjusted summaries rather than a replacement for a full statistical analysis plan.

For continuous data, methods may also depend on distributional shape and whether variability differs between groups. General statistical methods can account for unequal variances or non-normal data, but the registry record does not describe such diagnostics or methods. Over-dispersion is most directly relevant to count outcomes when observed variability exceeds a model's assumptions; it is not a feature to infer from this continuous endpoint's summaries. No additional model-based analysis can be justified from the reported information alone.

10. Confidence Intervals and Precision

Precision is reflected in the width of an interval around an estimate: a narrower interval indicates less uncertainty under the calculation's assumptions, while a wider interval leaves more possible values compatible with the data. The recalculated Estradiol-versus-Placebo mean difference is 10.0, with an interval from -6.895 to 26.89. The recalculated Sertraline-versus-Placebo mean difference is -1.0, with an interval from -14.86 to 12.86. Both ranges include zero and span values on both sides of it.

The estimates should therefore not be read as precise evidence of a particular direction or size of difference. In particular, an interval that includes zero does not prove that the groups are identical. It indicates that the posted summaries do not exclude zero under the calculation used, while also allowing differences in either direction within the interval. The interval confidence level is not reported in the registry record, and the comparisons are unadjusted recalculations from posted summaries.

The arm standard deviations—33.7 for Estradiol, 24.1 for Sertraline, and 29.8 for Placebo—show considerable spread in percent change scores. The endpoint group sizes were 26, 30, and 29, respectively. In general, greater variability and smaller group sizes reduce the precision of estimated group means and differences. The record does not report a prespecified precision target or the trial's original sample-size justification, so the observed intervals should be interpreted as the uncertainty associated with the posted summaries, not as evidence that the study met a planned precision objective.

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 for the continuous-outcome comparisons, not total event counts. They are illustrations based on the observed effect, not the trial's own design assumptions. The calculation assumptions listed in the registry data are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. The original trial's sample-size assumptions and justification are not reported in the registry record.

For each comparison, the 80% power column describes the number of participants per arm estimated to give an 80% chance of detecting the selected effect under those planning assumptions, if that effect and assumptions hold. The 90% power column makes the same kind of planning statement for a 90% chance. These are design calculations, not guarantees about what a future study would find. They are not observed power.

The rows use 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. A smaller target effect is harder to distinguish from random variation, so the planning calculation requires more participants per arm. This pattern is especially important when the observed comparison is close to zero: calculations based on a very small observed effect can lead to very large projected group sizes, and those values should not be mistaken for a reliable estimate of the effect that a future study would observe.

The curve represents how the planned sample requirement changes as the target effect changes under the stated assumptions. It is a planning aid, not a display of enrollment or completed participants. The trial's observed outcome group sizes are distinct from these illustrative per-arm requirements.

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 = 10.000; pooled standard deviation σ = 31.699.
2. n = 2 × (1.96 + 0.842)² × 31.699² / 10.000² = 157.7, rounded up to 158 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 <- 10                      # observed difference in means
sigma <- 31.699464313074                     # 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(744328)
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 = 158 (80% power), 212 (90% power)
 75% of observed effect: n per arm = 281 (80% power), 376 (90% power)
 50% of observed effect: n per arm = 631 (80% power), 845 (90% power)
power.t.test: 158.7 
Simulated power at n = 158 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.

13. Multiplicity, Missing Data and Analysis Populations

The registry record lists 0 analyses, although it includes primary outcome arm summaries and recalculated unadjusted comparisons. It also lists one secondary outcome: infant serum concentrations of estradiol in the three treatment arms. Results for that secondary outcome are not reported in the registry record. There are two outcome measures listed, but the record does not describe a prespecified strategy for controlling multiplicity across outcomes or treatment comparisons.

The recalculated p-values of 0.246 and 0.888 are unadjusted. A p-value from one comparison does not account for the number of other comparisons or outcomes considered. If several hypotheses are tested, the chance of at least one apparently unusual result can increase in general. Whether any adjustment was planned or performed is not reported in the registry record, so the displayed p-values should not be interpreted as multiplicity-adjusted evidence.

Missing-data handling is partly described: completers use baseline-to-Week-8 change, while non-completers use baseline-to-last-observation change under last observation carried forward. The number of missing endpoint values, reasons for missingness, timing of last observations, and sensitivity analyses are not reported in the registry record. The record also does not state whether missingness differed across treatment groups or whether alternative assumptions were examined.

The analysis population is not described beyond the endpoint's last-observation-carried-forward statement. The registry does not report whether all randomized participants were included, whether participants were excluded from the endpoint analysis, or how adherence and protocol deviations were handled. It also does not report a full statistical analysis plan, an adjusted model, or a prespecified approach to interim analyses. These unknowns make it difficult to determine whether the recalculated summaries align with the intended confirmatory analysis.

14. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports serious adverse event counts of 0 among 26 at risk in Estradiol, 0 among 30 at risk in Sertraline, and 0 among 29 at risk in Placebo. The listed serious adverse event terms are empty. These are reported counts for the registry's safety categories; they do not establish that no events occurred outside the categories captured or reported in the record.

Other adverse events are reported as 0 among 26 at risk in Estradiol, 4 among 30 at risk in Sertraline, and 1 among 29 at risk in Placebo. The record does not provide event descriptions for these counts in the posted summary. Death counts are not reported in the registry record. The safety figures should be read as descriptive counts, not as adjusted comparisons or estimates of relative risk. The registry information does not specify a statistical safety analysis, event attribution method, or further details needed to compare safety profiles.

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

The posted results describe average percent change in SIGH-ADS29 scores at Week 8 under a last-observation-carried-forward approach. Estradiol's mean was -38.0, Sertraline's was -49.0, and Placebo's was -48.0. The recalculated unadjusted comparisons with Placebo have intervals that include zero, and the available summaries therefore leave uncertainty about the direction and size of between-group differences.

These findings do not establish that estradiol, sertraline, or placebo is more effective. A mean describes a group, not each participant, and the endpoint's score change is not by itself a full account of the clinical course. The record does not report a threshold for a clinically meaningful difference, the confidence level for the comparison intervals, or adjusted analyses. The unadjusted comparisons are not a substitute for a prespecified trial analysis.

The registry record describes the study as terminated but does not state why. It reports outcome group sizes and summary statistics, but not details about completion, missingness beyond the carried-forward approach, analysis populations, or sensitivity analyses. The secondary outcome about infant serum concentrations of estradiol is listed without results. Safety counts are also limited to the categories and counts posted, with death counts not reported.

In general, a randomized comparison can support causal interpretation when the trial is conducted and analyzed according to sound prespecified methods. The available registry information does not establish all of those details. The appropriate statistical reading is therefore descriptive and cautious: the arm summaries and recalculated intervals show the estimates and their uncertainty, but they do not settle whether the treatments differ or what the findings mean for an individual.

16. Why This Trial Matters Statistically

This trial illustrates why group means and uncertainty intervals need to be read together. The Estradiol-versus-Placebo estimate is 10.0, but its interval ranges from -6.895 to 26.89. The Sertraline-versus-Placebo estimate is -1.0, with an interval from -14.86 to 12.86. Each interval includes zero and leaves room for effects in either direction. Focusing only on the point estimates would conceal that uncertainty.

It also illustrates the distinction between posted outcome summaries and a trial's complete analysis. The comparisons are unadjusted recalculations from arm summaries; the registry record lists 0 analyses and does not report a full statistical analysis plan. A randomized design is important, but interpretation also depends on how outcomes, missing observations, multiplicity, and analysis populations were handled. Those details are not all available in this record.

The endpoint uses last observation carried forward for non-completers. This makes the stated missing-data rule visible, while also highlighting a general limitation: substituting an earlier observation for an unobserved later outcome requires assumptions about how the outcome would have changed. Without information about missingness or sensitivity analyses, readers cannot assess how dependent the estimates are on that choice.

Finally, the sample-size illustrations emphasize that planning depends strongly on the effect a future study aims to detect. Smaller target effects require more participants per arm under the stated assumptions. Those calculations are not observed power and do not explain the original trial's enrollment target, which is not reported in the registry record.

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 reports a terminated status but does not report the reason for termination. It does not describe the study's original sample-size justification, recruitment progress, or whether termination affected the planned analyses. Although enrollment is 85, the record does not provide a participant flow account or explain how many participants completed each stage.

The primary endpoint is summarized using last observation carried forward for non-completers. The number of non-completers, reasons for missing observations, and sensitivity analyses are not reported. The record also does not describe whether the analysis included all randomized participants or how protocol deviations and treatment adherence were handled. These details matter when assessing the robustness of the endpoint estimates.

The comparisons are unadjusted recalculations from the posted arm means and standard deviations. The registry lists 0 analyses and does not report the sponsor's statistical model, the confidence level for the intervals, any multiplicity strategy, or prespecified subgroup analyses. No clinically meaningful threshold is reported for interpreting the magnitude of the score changes. The secondary outcome concerning infant serum concentrations of estradiol is listed without results, and safety counts lack further event descriptions in the posted summary.

The record describes a randomized, quadruple-masked, parallel design, but does not give operational details of sequence generation, allocation concealment, masking assessment, or dose management. These limitations are specific to what is documented for this trial and constrain interpretation of the posted summaries; they do not by themselves show that a particular aspect of conduct was inadequate.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Postpartum Depression: Transdermal Estradiol Versus Sertraline — NCT00744328.
• Registry record last updated 2019-09-10; status last verified 2018-08.
• Results first posted 2014-10-23.
• 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

NCT00744328 is a randomized, quadruple-masked, parallel-group Phase 4 trial of transdermal estradiol, sertraline, and placebo for postpartum depression. Enrollment was 85. The study began in 2008-08, had a primary completion date of 2013-09, and is listed as terminated. The reason for termination is not reported in the registry record.

The primary endpoint was percent change in SIGH-ADS29 score at Week 8, summarized as a mean with standard deviation using last observation carried forward for non-completers. Posted group means were -38.0 for Estradiol, -49.0 for Sertraline, and -48.0 for Placebo. Unadjusted recalculated comparisons with Placebo have intervals that include zero. The record lists an infant serum estradiol secondary outcome without posted results and reports limited adverse event counts. The original sample-size assumptions, analysis plan, and missing-data details are not reported in the registry record.