Psilocybin vs Escitalopram for Major Depressive Disorder: Comparative Mechanisms: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Psilocybin vs Escitalopram for Major Depressive Disorder: Comparative Mechanisms (NCT03429075), a randomized phase 2 study. The primary outcome is Percentage Change of the BOLD Signal.

2. Clinical Question

The registry lists changes in depression scales and a response outcome as secondary measures. Those outcomes provide clinical symptom context alongside the primary imaging measure, but they do not change what the primary endpoint itself measures. The registry record does not report a prespecified interpretation linking a particular BOLD change to a clinical benefit.

3. Trial Design

The registry describes a completed, randomized, double-blind, parallel-group phase 2 trial sponsored by Imperial College London. Enrollment is listed as 59. The recorded study start date is 2019-01-07, and the primary completion date is 2020-04-17. The record does not give additional detail about the randomization procedure, concealment, or how double blinding was achieved.

The intervention names are Psilocybin + Placebo and Psilocybin + Escitalopram, while the arm descriptions say that patients receive psilocybin or escitalopram. The primary outcome summaries use the arm labels Psilocybin and Escitalopram. The registry’s primary endpoint is percentage change in BOLD signal, measured with functional magnetic resonance imaging in response to emotional faces, from baseline to 6 weeks after the first psilocybin dose.

The primary outcome summaries list 25 participants in the Psilocybin arm and 21 in the Escitalopram arm. The primary endpoint description separately says that data from an imaging group used to predict responses to the trial drugs have not been analyzed. The record does not explain the relationship between that statement and the posted primary outcome summaries, so the two should be read as distinct registry statements rather than reconciled by assumption.

Secondary outcomes include changes in HAMD-17, BDI-IA, and MADRS, along with QIDS-16 response and remission outcomes. Numerical summaries are posted for HAMD-17, MADRS, and QIDS-16 response. Numerical results for BDI-IA and QIDS-16 remission are not reported in the registry record. The registry lists one analysis and seven outcome measures.

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

Analysis population for the primary outcome, as defined in the registry: The data obtained from an imaging group in the trial, in which functional MRI was used to predict responses to the trial drugs, have not been analyzed.

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 Change of the BOLD Signal, measured over Baseline measure vs 6 weeks post 1st psilocybin dosing.

The primary endpoint is the mean percentage change in the BOLD signal, a measure derived from functional MRI. Participants were shown emotional faces, and the pre-treatment and post-treatment BOLD values were used to estimate a percentage change for each participant; the individual values were then used to estimate a group mean. The timeframe is baseline compared with 6 weeks after the first psilocybin dose.

BOLD is an imaging measure of brain response, rather than a depression symptom score or a direct measure of daily functioning. In general, an imaging endpoint can be used to study brain responses associated with a treatment or task. It may help address questions about mechanisms, but a difference in this measure alone does not establish that participants experienced a corresponding difference in symptoms or clinical benefit.

The posted means are -0.08 for Psilocybin and 0.43 for Escitalopram, with standard deviations of 0.45 and 0.62, respectively. The reported mean difference is 0.51 for Escitalopram relative to Psilocybin. The registry record does not report a clinical threshold for interpreting the size or direction of a BOLD change, nor does it report an analysis linking individual BOLD changes to clinical outcomes. The endpoint should therefore be interpreted as an imaging comparison at the stated follow-up time.

7. Results Posted in the Registry

• Escitalopram: mean 0.43 versus -0.08 on control; difference 0.51 (95% CI 0.19 to 0.83), standardized effect d = 0.96; z-test p = 0.002.

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 in Hamilton Depression Scale (HAMD-17)

• 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:
  • Escitalopram: mean -5.10 versus -10.50 on control; difference 5.40 (95% CI 2.63 to 8.17), standardized effect d = 0.99; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Escitalopram than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change in MADRS

• 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:
  • Escitalopram: mean -7.20 versus 14.40 on control; difference -21.60 (95% CI -26.31 to -16.89), standardized effect d = -2.34; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Escitalopram than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Number of Patients Who "Responded": Quick Inventory of Depressive Symptomatology (QIDS-16) Response at 6 Weeks

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Escitalopram: 48.3% (14/29) versus 70.0% (21/30) on control; risk difference -21.7 percentage points (95% CI -46.2 to 2.8), risk ratio 0.69 (0.44–1.07); unadjusted two-proportion z-test p = 0.089.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Escitalopram and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover different aspects of depressive symptoms: changes in HAMD-17 and MADRS scores, and a QIDS-16 response outcome. Their estimates do not all point in the same numerical direction, and the scales do not share a common unit. For that reason, the outcomes should be considered separately rather than combined informally into one overall measure. The registry also lists BDI-IA change and QIDS-16 remission, but numerical results for those outcomes are not reported in the registry record.

These outcomes offer clinical symptom context for the primary fMRI comparison, but they are not interchangeable with the BOLD endpoint. An imaging contrast can address a question about brain response, while symptom scales and response definitions concern reported depression measures. A pattern across secondary outcomes might help contextualize the primary result, but it cannot establish a mechanism or explain why a group difference occurred without a reported analysis connecting the measures.

The posted secondary results include estimates with unadjusted p-values. When multiple outcomes are examined, unadjusted results carry less weight as confirmatory evidence because the chance of at least one apparently notable result can rise across tests. The registry record does not report a multiplicity strategy or a complete plan identifying which secondary outcomes were confirmatory. Accordingly, secondary findings are most useful as supportive or exploratory evidence, interpreted with their confidence intervals, scale definitions, and analysis limitations in view.

The posted summaries do not provide numerical results for every listed secondary outcome. That absence should not be treated as evidence that an outcome showed no difference. It means that a full assessment of the secondary outcome family cannot be made from the registry record. The primary imaging result and the secondary clinical measures answer related but distinct questions, and neither should be used to overstate what the other establishes.

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

• Change in Hamilton Depression Scale (HAMD-17)
• Change in Beck Depression Inventory (BDI-IA)
• Change in MADRS
• Number of Patients Who "Responded": Quick Inventory of Depressive Symptomatology (QIDS-16) Response at 6 Weeks
• Number of Patients Who "Remitted": QIDS-16 Remission Rate

9. Statistical Interpretation

10. Statistical Methods Explained

Continuous outcomes are reported as group means. A mean describes the average observed value in an arm, while a comparison of means describes the difference between the arm averages. In these comparisons, the registry summaries use the Psilocybin arm as the first reference and report the contrast for Escitalopram relative to Psilocybin. The direction of a positive or negative difference therefore depends on both the order of the arms and the endpoint’s scale.

What does the spread measure?

For the primary BOLD outcome, the reported spread is a standard deviation. In general, a standard deviation describes how dispersed individual observations are around the group mean. It is not the same as the uncertainty in the estimated mean. For HAMD-17 and MADRS, the posted spread is a standard error. In general, a standard error describes the sampling uncertainty of an estimated mean, rather than the spread of individual participant values. Keeping these measures distinct matters when judging how variable the data are and how precise a group average may be.

The mean difference is accompanied by a confidence interval. In general, a confidence interval gives a range of values compatible with the data and the stated estimation procedure. It communicates uncertainty about the estimated contrast; it does not mean that a particular proportion of participants had results inside that interval. If an interval crosses zero for a difference, values on either side of zero may be compatible with the data. If it stays on one side, the interval indicates the direction of the estimated contrast under that method, without proving clinical importance.

What does the p-value say?

A p-value describes how surprising data at least as incompatible with a null hypothesis would be if that null hypothesis and the analysis assumptions held. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The posted p-values are unadjusted. The registry record does not report a multiplicity correction or enough detail to determine whether the posted tests were prespecified in a full statistical analysis plan.

The primary comparison reports an unadjusted p-value of 0.002. The HAMD-17 and MADRS comparisons report p-values of <0.001. The QIDS-16 response comparison reports 0.089. These values should be read alongside the effect estimates and confidence intervals, not as standalone rankings of which outcome matters most. Differences in outcome type, measurement scale, and uncertainty affect interpretation.

How should the binary outcome be understood?

QIDS-16 response is reported as a number of participants meeting the response outcome. A risk difference compares the proportions in the two arms, with the reported contrast expressed for Escitalopram relative to Psilocybin. The registry also gives relative measures, but the risk difference provides an absolute-scale contrast. In general, the confidence interval for a risk difference helps show whether the data are compatible with no difference as well as with differences in either direction. The reported interval for this outcome includes zero.

For this binary result, it is important not to confuse the count with the group size or with the estimated risk difference. The counts are 21 and 14, and the analyzed group sizes are 30 and 29. The reported risk difference is -0.217. The p-value is unadjusted, and the registry does not report a model-based adjusted comparison for this outcome.

Why does over-dispersion matter?

In general, over-dispersion occurs when observed variability is greater than a statistical model expects. If a model assumes less variation than is present, standard errors may be too small and confidence intervals too narrow. The registry record does not identify an over-dispersion issue or describe a model requiring an over-dispersion adjustment for the reported endpoints. The concept is relevant generally, but there is no basis in the record to say that it affected these estimates.

The planning illustrations use the observed effect as a starting point and show hypothetical participant requirements for different assumed fractions of that effect. These are not evidence about the trial’s original design. The record’s planning assumptions specify a two-sided significance level of 0.05, equal allocation, and a normal approximation. The trial’s own original sample-size assumptions are not reported in the registry record.

For the primary continuous outcome, the posted summaries list the group means, standard deviations, and a mean difference with an interval. The registry also reports a standardized effect estimate of 0.955. Such an estimate expresses the contrast in relation to pooled variability; it does not replace the original scale or establish clinical relevance. No covariate-adjusted analysis, subgroup result, or interim analysis is described in the registry record. Those analyses should not be inferred from the summary statistics.

11. Confidence Intervals and Precision

Precision concerns how much uncertainty surrounds an estimate. The primary mean difference is 0.51, and its reported confidence interval extends from 0.192 to 0.829. The interval lies above zero, which indicates that the estimated contrast is positive under the reported comparison and interval method. Its width also shows that the exact size of the population contrast is not determined precisely by the point estimate alone.

For the HAMD-17 mean difference, the estimate is 5.4 and the interval runs from 2.628 to 8.172. For MADRS, the reported mean difference is -21.6, with an interval from -26.31 to -16.89. These intervals do not cross zero. However, each estimate remains tied to its measure, its direction of scoring, and the registry’s reported analysis. The record does not describe a clinically important threshold for these scales.

For QIDS-16 response, the risk difference is -0.217 and the reported interval runs from -0.462 to 0.028. Because the interval includes zero, the data are compatible with no difference as well as with differences in either direction under this estimation procedure. The unadjusted p-value is 0.089. The interval is useful because it displays this uncertainty directly; a p-value alone would not communicate the range of plausible contrasts.

Confidence intervals do not account automatically for every limitation. They do not correct for multiple testing, missing observations, selection of outcomes, or a mismatch between the analyzed participants and the intended population. The registry record does not report the missing-data methods or analysis populations in enough detail to assess those issues. The displayed intervals should therefore be interpreted as uncertainty under the reported comparisons, not as a complete measure of uncertainty about all possible sources of bias.

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 figures are hypothetical requirements, not a description of enrollment or a reconstruction of the trial’s original design. For continuous outcomes and the binary response outcome, the figures are participants per arm. The registry’s general planning note says time-to-event illustrations, if present, give total events; these outcomes are not time-to-event measures.

The listed assumptions 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 design assumptions and sample-size justification are not reported in the registry record. The planning values are not observed power.

The 80% power column represents the number of participants per arm that the planning calculation estimates would be needed to have an 80% chance of detecting the assumed effect under those assumptions. The 90% power column uses a higher target chance of detection and therefore generally requires more participants per arm. Neither column says how likely the observed trial result was to be true, nor does either column establish the probability that a treatment effect exists.

The rows that plan for 100%, 75%, and 50% of the observed effect illustrate how the assumed effect size changes the requirement. Smaller effects need more participants to detect at the same target power, because they are harder to distinguish from random variation. These calculations use the observed effect as an input, which can be an unstable basis for planning, especially in a small study. They should be treated as illustrations for interpreting the planning display, not as evidence about the trial’s actual power or adequacy.

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 = 0.510; pooled standard deviation σ = 0.534.
2. n = 2 × (1.96 + 0.842)² × 0.534² / 0.510² = 17.2, rounded up to 18 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 <- 0.51                      # observed difference in means
sigma <- 0.534024173780381                     # 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(3429075)
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 = 18 (80% power), 24 (90% power)
 75% of observed effect: n per arm = 31 (80% power), 41 (90% power)
 50% of observed effect: n per arm = 69 (80% power), 93 (90% power)
power.t.test: 18.2 
Simulated power at n = 18 per arm: 0.795 (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 record lists multiple outcome measures, including the primary BOLD endpoint and secondary depression-scale and response outcomes. Testing several outcomes can increase the chance of finding at least one nominally small p-value even when there is no true effect for any of them. In general, multiplicity methods can control that risk by adjusting the testing procedure or by defining a hierarchy of outcomes. The registry record does not report such a method, so the posted p-values should be read as unadjusted rather than as family-wise confirmatory evidence.

The registry lists one analysis, but that count does not describe the full statistical analysis plan. Details about model specification, covariate adjustment, handling of repeated measurements, planned subgroup analyses, or interim analyses are not reported in the registry record. No interim analysis should be inferred from the reported results.

Missing-data handling is also not described. The primary outcome summaries list 25 participants in the Psilocybin arm and 21 in the Escitalopram arm, while the secondary HAMD-17, MADRS, and QIDS-16 response summaries list 30 and 29. The record does not explain these differences in analysis group sizes, the reasons for any unavailable measurements, or whether missing observations were imputed. The primary outcome description also states that data from an imaging group have not been analyzed, without explaining its relationship to the posted primary summaries.

The analysis population is not clearly specified beyond the arm-specific numbers associated with each outcome. The registry record does not say whether the comparisons follow an intention-to-treat approach, include only participants with available outcome measurements, or use another population definition. These omissions limit assessment of possible bias from missing data or exclusions. They do not establish that the results are biased, but they prevent a complete evaluation of how the reported estimates relate to all randomized participants.

15. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry’s serious adverse-event records list 0 serious adverse events among 30 participants at risk in the Psilocybin group and 0 among 29 participants at risk in the Escitalopram group. Deaths are listed as 0 in both groups. No top serious-event terms are listed.

The record separately lists 26 and 24 for other adverse events in the Psilocybin and Escitalopram groups, respectively, with 30 and 29 participants at risk. These figures should not be confused with the serious adverse-event counts. The registry excerpt does not provide event descriptions, severity grades, exposure-adjusted rates, or further detail about how these other events were counted.

A zero count in the recorded serious-event category means no such events were listed for these groups in the registry record; it does not establish that serious events are impossible or that the treatments have no safety risks. The figures also do not, by themselves, allow a complete comparison of safety, especially without further information about event definitions, follow-up, and ascertainment.

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

The primary result describes a difference in the group mean percentage change in BOLD signal between the Psilocybin and Escitalopram arms at the specified imaging follow-up. The estimate is positive for Escitalopram relative to Psilocybin, with a confidence interval above zero and an unadjusted p-value of 0.002. This is evidence of a difference in the posted imaging summaries under the reported comparison, not a direct demonstration of a difference in depression symptoms or overall clinical benefit.

The secondary findings address symptom-scale changes and a response definition. They provide additional outcome-specific observations, but they are unadjusted, and the registry record does not report a multiplicity plan. The outcome definitions, scale directions, and analysis populations also need to be considered. The registry does not report a statistical analysis linking the fMRI result to the symptom outcomes, so a causal or mechanistic explanation cannot be drawn from their coexistence in the record.

The results do not establish that either treatment is superior in all relevant respects, that the estimates apply to people outside the trial, or that the observed differences would persist in a larger study. A phase 2 comparison and its posted summaries can inform further research questions, but the record does not report the full protocol, analysis plan, or original sample-size rationale needed for a complete appraisal.

The distinction between statistical evidence and clinical importance is central. An interval that excludes zero does not by itself show that an effect is large enough to matter to patients, and a result whose interval includes zero does not prove that there is no effect. The registry supplies no clinical threshold for the BOLD measure and no threshold-based interpretation for the posted contrasts. Conclusions should therefore remain specific to the measures, time points, and analyses recorded.

17. Why This Trial Matters Statistically

This trial illustrates why endpoint definition matters in biostatistics. Its primary outcome is an fMRI-derived BOLD measure, while its secondary measures include depression scales and a response count. Those outcomes can be related in a research program, but they represent different constructs and cannot be treated as substitutes for one another. A statistically clear imaging contrast still needs a separate argument before it can be interpreted as clinically important.

The record also illustrates the value of reading point estimates with their uncertainty. A mean difference, confidence interval, and p-value each answer different questions: the estimated contrast, its precision under the analysis, and compatibility with a null hypothesis under stated assumptions. The QIDS-16 response interval includes zero, whereas the primary interval does not. Neither distinction alone settles the clinical meaning of the outcomes.

Finally, the record makes visible several issues that affect interpretation beyond the arithmetic of a comparison: secondary testing without a reported multiplicity strategy, outcome-specific analysis group sizes, absent detail on missing-data handling, and a statement about unanalysed imaging-group data alongside posted primary summaries. These are reasons to keep claims narrow and transparent. They also show why registry records are best read as evidence with scope and limitations, rather than as a complete substitute for a protocol and full analysis report.

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.

Several trial-specific details needed to assess the findings are not reported in the registry record. The record does not describe the randomization procedure, allocation concealment, how double blinding was maintained, or the treatment procedures in enough detail to assess their implementation. The named interventions and arm descriptions should not be used to fill in those missing details.

The primary endpoint description says that data from an imaging group used to predict responses to the trial drugs have not been analyzed, while numerical primary outcome summaries are also posted. The record does not clarify the relationship between these statements. That ambiguity limits interpretation of the primary outcome’s analysis population and should not be resolved by assumption.

Group sizes differ across the primary and secondary outcome summaries, and the record does not explain why. Missing-data methods, participant flow, analysis-population definitions, and reasons for unavailable measurements are not reported. The registry also does not provide a complete statistical analysis plan, adjustment for multiple testing, original sample-size assumptions, or a clinical threshold for interpreting the BOLD measure.

Numerical results are not reported for all listed secondary outcomes. The posted p-values are unadjusted, and no interim analyses or covariate-adjusted results are described. These limitations do not negate the posted estimates, but they restrict how confidently the results can be generalized or combined into a broad conclusion about efficacy, mechanisms, or comparative safety.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Psilocybin vs Escitalopram for Major Depressive Disorder: Comparative Mechanisms — NCT03429075.
• Registry record last updated 2024-10-24; status last verified 2024-08.
• Results first posted 2024-10-24.
• 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

NCT03429075 is a completed randomized, double-blind, parallel-group phase 2 trial sponsored by Imperial College London. It enrolled 59 participants and studied major depressive disorder. The primary outcome was mean percentage change in BOLD signal, based on fMRI responses to emotional faces at baseline and 6 weeks after the first psilocybin dose.

The primary comparison reports a mean difference of 0.51 for Escitalopram relative to Psilocybin, with an unadjusted p-value of 0.002. Posted secondary results include HAMD-17 change, MADRS change, and QIDS-16 response; numerical results for BDI-IA change and QIDS-16 remission are not reported in the registry record. The registry does not report a multiplicity method, missing-data approach, or the trial’s original sample-size assumptions.