Drug Therapy to Treat Minor Depression: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Drug Therapy to Treat Minor Depression (NCT00048815), a randomized na study. The primary outcome is Efficacy Assessed Using the Inventory of Depressive Symptomatology - Clinician Rated (IDS-C).

2. Clinical Question

3. Trial Design

This was a randomized, double-blind, parallel-group trial with enrollment of 73. The registry lists citalopram, St. John's Wort, and placebo as the study interventions. The outcome groups include 24 participants for citalopram, 26 for St. John's Wort, and 23 for placebo. These group counts sum to the overall enrollment, but the registry does not explain whether all enrolled participants contributed to the primary outcome analysis or how missing observations were handled.

The arm descriptions state that citalopram was taken orally at 20mg/day for 12 weeks and St. John's Wort was taken orally at 810 mg/day, in three tablets of 270mg each, for 12 weeks. The placebo arm received double-dummy, look-alike placebo for 12 weeks. The brief summary calls this a 6-month study, while the primary endpoint is measured from baseline to Week 12. Details about visits or treatment after Week 12 are not reported in the registry record.

The listed phase is NA. The sponsor is Massachusetts General Hospital, the status is completed, and the registry lists the start as 2003-02 and primary completion as 2007-04. These record details establish the broad study context but do not describe the trial's conduct in depth. The sequence-generation method, allocation concealment, masking procedures, protocol deviations, and analysis population are not reported in the registry record. The trial's original sample-size justification, planned statistical analysis, and any interim analyses are also 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.

Overall Study

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 Efficacy Assessed Using the Inventory of Depressive Symptomatology - Clinician Rated (IDS-C), measured over Change from Baseline to Week 12.

The primary endpoint was efficacy assessed using change in the Inventory of Depressive Symptomatology - Clinician Rated (IDS-C) from baseline to Week 12. The registry describes the IDS-C as a clinician-rated measure with a total score range from 0 to 84, where 0 means not depressed at all and 84 means the most depressed. It states that change is calculated by subtracting the Week 12 score from the baseline score.

With that stated subtraction convention, a positive change represents a lower Week 12 score than baseline, while a negative change represents a higher Week 12 score than baseline. The posted group estimates are negative: -11.47 for citalopram, -9.35 for St. John's Wort, and -10.49 for placebo. Details about the fitted model and the exact handling of the change measure are not reported in the registry record, so the results should be described using the registry's endpoint definition rather than assigning a clinical interpretation beyond it.

A clinician-rated symptom measure can capture change in an outcome relevant to the study question. Blinding is relevant because ratings may be influenced by knowledge of treatment assignment; the record describes the study as double-blind. The procedures used to preserve blinding and the reliability of ratings are not reported in the registry record.

7. Results Posted in the Registry

• Citalopram: mean -11.47 versus -10.49 on control; difference -0.98 (95% CI -4.76 to 2.80), standardized effect d = -0.15; z-test p = 0.612.
• St. John's Wort: mean -9.35 versus -10.49 on control; difference 1.14 (95% CI -2.47 to 4.75), standardized effect d = 0.18; z-test p = 0.536.

8. Statistical Interpretation

9. Statistical Methods Explained

The outcome is continuous: it is a score-based measure rather than an event count. For each group, the registry posts a least-squares mean and its standard error. A mean describes the central level of a continuous outcome. A least-squares mean is generally a model-based group estimate, but the model specification, covariates, and estimation details are not reported in the registry record. The standard error describes uncertainty in the estimated mean; it is not the spread of individual participant scores. The record does not post a standard deviation for the outcome groups.

The primary endpoint is defined as change from baseline to Week 12 on the IDS-C, with change calculated as baseline minus Week 12. The posted group estimates are -11.47 for citalopram, -9.35 for St. John's Wort, and -10.49 for placebo, each in units on a scale. The scale ranges from 0 to 84. The registry does not post the baseline and Week 12 group means separately, so the change estimates cannot be broken down into their starting and ending score components from the record alone.

The comparison measure is a mean difference. For each active treatment, the recalculated contrast subtracts the placebo mean from the active-treatment mean. A negative contrast therefore indicates a lower posted mean change for the active group than placebo, while a positive contrast indicates a higher posted mean change for the active group. This is an arithmetic description of the direction of the contrast, not a judgment about clinical benefit. Because the endpoint's change convention affects how signs relate to score movement, interpretation should keep both the contrast definition and the baseline-minus-Week-12 convention in view.

A confidence interval gives a range of values compatible with the data and the statistical assumptions used to calculate it. The citalopram comparison is -0.98, with a 95% confidence interval from -4.764 to 2.804. The St. John's Wort comparison is 1.14, with a 95% confidence interval from -2.469 to 4.749. Each range crosses zero, the value corresponding to no mean difference. The ranges therefore do not isolate a difference in one direction. Their breadth also signals uncertainty about the size and direction of the contrast.

The listed p-values, 0.612 for citalopram and 0.536 for St. John's Wort, are unadjusted recalculations from the posted summaries. A p-value describes the compatibility of the data with a specified null model; it is not the probability that the null hypothesis is true, nor the probability that the result arose by chance. These calculations do not substitute for a prespecified trial analysis. The registry reports zero posted analyses, and it does not describe the model used for its least-squares means or any sponsor-run comparison procedure. If an analysis was conducted outside the record, its details are not reported in the registry record.

The recalculated comparisons use pooled standard deviations estimated from the group standard errors and sample sizes. The record marks these as approximate. This approach depends on assumptions about the relationship between the reported standard errors and the variability of participant-level outcomes, and about combining variability across groups. Because the underlying group-level standard deviations and analysis details are absent, the approximation cannot be independently checked from the registry record. Such recalculations are useful for an approximate unadjusted comparison, but they should not be mistaken for a verified reproduction of the trial's original model.

General statistical point: over-dispersion matters when variability in the observed data exceeds what an assumed model expects. If that extra variability is not accommodated, uncertainty may be understated and tests may appear more decisive than warranted. This is a general concern in statistical modeling, not a documented feature of this trial. Whether any model needed an over-dispersion adjustment is not reported in the registry record.

General statistical point: randomization is intended to balance measured and unmeasured factors on average across treatment groups, while blinding can reduce the influence of treatment knowledge on care or outcome assessment. Neither feature guarantees that every group is comparable in a small sample or that missing outcomes cannot bias an estimate. The record identifies randomized and double-blind design features but does not report methods for checking baseline balance, preserving allocation concealment, assessing masking, or addressing missing data.

The registry's sample-size planning illustrations use the observed effect as a reference and vary the effect to be detected. They assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They are planning illustrations rather than a reconstruction of the trial's original design. The original sample-size assumptions and justification are not reported in the registry record.

10. Confidence Intervals and Precision

Precision concerns how narrowly the data estimate a treatment contrast. The recalculated citalopram-versus-placebo difference is -0.98 units, with a 95% confidence interval from -4.764 to 2.804. The recalculated St. John's Wort-versus-placebo difference is 1.14 units, with a 95% confidence interval from -2.469 to 4.749. Each interval includes zero and spans differences on both sides of zero.

A confidence interval is not a guarantee that the true value lies in the displayed range, and it is not a range of individual participant outcomes. In general, a 95% confidence interval is generated by a procedure that would cover the true parameter in 95% of repeated studies under its assumptions. Its width reflects both the amount of information and the variability represented by the analysis. Here, the posted intervals allow substantial uncertainty about the direction and magnitude of each mean difference.

The confidence intervals are recalculated from the posted summaries, not reported as the result of a trial analysis in the registry record. The recalculations use approximate pooled variability inferred from standard errors. Since the original model, covariates, and handling of missing observations are not reported in the registry record, these intervals may not capture all sources of uncertainty relevant to a full trial analysis. They should be read as approximate unadjusted intervals, not as a substitute for the trial's intended inferential results.

The p-values do not resolve the uncertainty in the intervals. A non-small p-value does not demonstrate equivalence, and an interval that includes zero can still contain differences that may matter under a separate clinical or scientific judgment. The registry does not report a threshold for a clinically important difference, so the practical importance of values within these intervals cannot be established from this record alone.

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. They are illustrations based on the observed contrast, with rows planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. For each row, the 80% power column gives the estimated participants per arm under the stated assumptions for an 80% chance of detecting an effect of that size if it is present. The 90% power column gives the corresponding planning estimate for a 90% chance.

The listed 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 the stated fraction of the observed effect. These are planning assumptions for the illustration; the trial's own original design assumptions and sample-size justification are not reported in the registry record. This is not observed power.

The illustrations are generated separately for each active-treatment comparison with placebo. For citalopram, planning for 100% of the observed effect gives 716 participants per arm for 80% power and 959 participants per arm for 90% power; planning for 75% gives 1,273 and 1,704; planning for 50% gives 2,864 and 3,834. For St. John's Wort, the corresponding figures are 497 and 666 for 100%, 884 and 1,183 for 75%, and 1,988 and 2,661 for 50%.

The direction of the relationship is important: a smaller target effect requires more participants to distinguish it from random variation at the same stated power and significance level. These values are not forecasts of the size of a future study, and they do not show what the completed trial would have found with a different enrollment. They are conditional illustrations whose outputs depend on the observed effect and the stated statistical assumptions.

12. How the Sample Size Is Derived

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

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.980; pooled standard deviation σ = 6.618.
2. n = 2 × (1.96 + 0.842)² × 6.618² / -0.980² = 715.8, rounded up to 716 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.98                      # observed difference in means
sigma <- 6.61763737631826                     # 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(48815)
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 = 716 (80% power), 959 (90% power)
 75% of observed effect: n per arm = 1,273 (80% power), 1,704 (90% power)
 50% of observed effect: n per arm = 2,864 (80% power), 3,834 (90% power)
power.t.test: 716.8 
Simulated power at n = 716 per arm: 0.803 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures but posts no secondary results. It identifies the IDS-C change from baseline to Week 12 as the primary efficacy measure. The registry does not report how the additional outcome measure was defined, when it was assessed, or whether it was included in a planned analysis.

Multiplicity arises when several outcomes, time points, treatment comparisons, or analytical choices are tested. Repeated testing can increase the chance of at least one apparently notable result even when no true effect exists. Statistical plans may address this by defining a primary comparison, controlling an error rate, or distinguishing confirmatory from exploratory analyses. The registry record does not report a multiplicity adjustment or explain whether one was planned. The two recalculated active-treatment comparisons are unadjusted, so their p-values should not be interpreted as multiplicity-controlled results.

The record does not report a statistical analysis plan, the original protocol's analysis details, or a prespecified hierarchy for outcomes and comparisons. It also does not report whether the posted least-squares means were adjusted for baseline values or other covariates, which population was included in each analysis, or how protocol deviations were handled. The trial's sample-size justification and any interim analyses are not reported in the registry record.

Missing-data handling is also not reported in the registry record. Although the outcome group counts are 24, 26, and 23, those counts alone do not establish whether all randomized participants were analyzed, whether any endpoint values were missing, or how missing values were treated. Methods such as complete-case analysis, imputation, or model-based use of available observations can rely on different assumptions, and the appropriate interpretation depends on what was actually done.

These gaps do not show that a particular procedure was absent from the trial; they mean its use cannot be assessed from the registry record. The comparison results should therefore be read as limited to the posted endpoint summaries and unadjusted recalculations, without assuming an unreported multiplicity or missing-data strategy.

14. Safety Results

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

Other (non-serious) adverse events

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

Most frequent non-serious adverse event terms

The serious adverse event record lists 0 affected participants in the citalopram group, 0 in the St. John's Wort group, and 0 in the placebo group. The corresponding numbers at risk are 24, 26, and 23. These figures describe the registry's posted serious adverse event counts; they do not establish that no serious events occurred outside the reported observation or reporting period.

The death counts are not reported in the registry record. The serious adverse event terms list is empty, so no specific serious event types are identified in the record. Other adverse events are listed for 23 participants in the citalopram group, 22 in the St. John's Wort group, and 18 in the placebo group, out of 24, 26, and 23 at risk, respectively. These are counts, not rates calculated.

Safety comparisons depend on how events were defined, collected, and followed, and on the duration and completeness of observation. The registry record does not give detailed event definitions, exposure-adjusted analyses, or information about event severity beyond the serious-event classification. The counts should be read as reporting information from the record, not as a complete assessment of comparative safety.

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

The registry's posted results describe Week 12 change in clinician-rated depressive symptom scores for participants assigned to citalopram, St. John's Wort, or placebo. The group estimates are -11.47, -9.35, and -10.49 units, respectively. The recalculated unadjusted mean differences versus placebo are -0.98 for citalopram and 1.14 for St. John's Wort. Both 95% confidence intervals include zero and extend to values on either side of it.

These results do not establish that the groups had identical outcomes. The intervals leave uncertainty about the direction and size of the differences, and the registry does not provide a clinically important difference threshold for judging whether values within those ranges would matter. The p-values of 0.612 and 0.536 likewise do not prove the absence of an effect. They are unadjusted recalculations, not posted sponsor analyses.

The registry's endpoint description defines change as baseline minus Week 12. Under this convention, a negative change corresponds to a higher Week 12 score than baseline. The posted group estimates are negative, but the record does not include the detailed analysis model or baseline and Week 12 component means needed to examine the result further. Accordingly, the estimates should be described according to the reported scale and convention, without asserting a treatment benefit or lack of benefit.

The findings concern this study's enrolled participants, treatments, and Week 12 endpoint. They do not on their own answer questions about other populations, different treatment durations, longer-term outcomes, or outcomes not posted in the registry. The record also omits key information about missing data, analysis populations, multiplicity, and the original statistical plan. Those omissions constrain how confidently the posted summaries can be connected to a broader conclusion.

In general, a randomized design supports comparisons between assigned groups, but interpretation still depends on outcome measurement, follow-up, analysis choices, and reporting completeness. The safest reading is that the available summaries produce imprecise unadjusted contrasts that are compatible with a range of possibilities, while the registry record does not supply enough detail to reconstruct a full confirmatory analysis.

16. Why This Trial Matters Statistically

This trial is statistically instructive because its primary outcome is continuous and its posted group summaries are model-based least-squares means with standard errors. Those summaries differ from raw participant-level data: without the model specification or individual observations, independent reconstruction is limited. A mean difference offers a direct scale-based contrast, while its confidence interval makes the uncertainty around that contrast visible.

The two active-treatment comparisons have confidence intervals that cross zero and include values in either direction. This illustrates why a point estimate alone is not enough. The citalopram contrast is -0.98, and the St. John's Wort contrast is 1.14, but each estimate is uncertain. Reporting the full interval helps avoid treating the point estimate as a precise description of the underlying effect.

The sample-size illustrations also show the dependence of planning on the target effect. Their required participants per arm increase when the planning target is reduced from 100% to 75% or 50% of the observed effect. This is a general design principle: smaller effects are harder to distinguish from variability, so more information is needed under otherwise similar assumptions. The illustrations are not the original trial's planning rationale, which is not reported in the registry record.

Finally, the record highlights the difference between posted outcome summaries and a complete statistical report. A full interpretation would benefit from the planned model, analysis population, missing-data approach, multiplicity strategy, and rationale for sample size. Their absence does not invalidate the posted numbers, but it limits the ability to judge how closely the recalculated comparisons reflect the trial's intended inferential analysis.

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 contains limited detail about trial conduct and analysis. It identifies randomization, double masking, and a parallel-group design, but does not report the allocation-generation method, allocation concealment, or masking procedures. It gives arm descriptions and treatment durations, but does not provide a detailed schedule of follow-up or the amount of treatment exposure actually received.

The primary outcome is defined as change from baseline to Week 12, but the record does not post separate baseline and Week 12 group scores, participant-level data, or standard deviations. It posts least-squares means and standard errors without describing the statistical model or covariates. The comparisons are unadjusted recalculations using approximate pooled variability, and cannot be verified against a full analysis output.

Important analysis details are not reported in the registry record: the original statistical plan, sample-size justification, analysis population, handling of missing observations, protocol deviations, multiplicity approach, and any interim analyses. The registry lists two outcome measures but posts no secondary results. These gaps make it difficult to assess selective reporting or determine whether the posted primary outcome analysis was the prespecified analysis.

The sample is relatively small in the context of the planning illustrations, but no new inference about power should be drawn from that observation. The registry's original enrollment target and rationale are not reported, and the planning calculations are illustrations based on observed effects rather than the trial's design. Finally, safety information is limited: serious event counts are posted as zero in each group, deaths are not reported in the registry record, and detailed event definitions and follow-up procedures are not included.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Drug Therapy to Treat Minor Depression — NCT00048815.
• Registry record last updated 2018-05-11; status last verified 2018-05.
• Results first posted 2018-05-11.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

The completed randomized, double-blind, parallel-group study compared citalopram, St. John's Wort, and placebo in minor depression. Overall enrollment was 73. The primary endpoint was change from baseline to Week 12 on the IDS-C, with posted least-squares mean changes of -11.47, -9.35, and -10.49 units for citalopram, St. John's Wort, and placebo.

Unadjusted recalculated mean differences versus placebo were -0.98 for citalopram and 1.14 for St. John's Wort; both 95% confidence intervals include zero. No secondary results or trial analyses are posted. The registry does not report the original statistical plan, sample-size rationale, analysis population, or missing-data procedures.