Poststroke Depression in Hemorrhagic Stroke: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Poststroke Depression in Hemorrhagic Stroke (NCT03826875), a randomized phase 2 study. The primary outcome is Depression.

2. Clinical Question

3. Trial Design

The registry record describes a phase 2, randomized, parallel-group trial comparing fluoxetine with placebo in survivors of aneurysmal subarachnoid hemorrhage. Its listed enrollment is 8. Participants assigned to the treatment group were initially prescribed fluoxetine 20mg/day for one year; those assigned to placebo were initially prescribed placebo 20mg/day for one year. The registry record lists both interventions as drugs.

The masking information is not fully consistent within the registry record: the trial-design description says triple masking, while the brief summary calls the trial double-blinded. The record does not clarify which people were masked or explain the difference between these descriptions. The allocation is listed as randomized and the model as parallel.

The record lists a start date of 2019-03-01, primary completion on 2025-05-07, and status as completed. The primary outcome is depression at one year. Other listed areas include anxiety, fatigue, healthcare utilization, social support, sleep disturbance, and function. Function appears twice in the listed secondary outcomes. The record lists 9 outcome measures, 7 posted secondary outcomes, and no posted analyses. These details describe what is registered and posted; they do not establish the trial’s recruitment experience, adherence, or the reasons participants may not appear in an outcome group.

The registry record does not report a sample-size justification or enough detail to establish how the stated enrollment relates to a planned statistical analysis. A phase designation and randomized design alone do not indicate whether a study was powered to detect a particular effect.

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: Patients withdrew

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 Depression, measured over 1 year.

The primary endpoint is depression measured at one year with the Hamilton Rating Scale for Depression (HAM-D). The registry describes the HAM-D as a 21-item questionnaire addressing different symptoms or aspects of depression, including mood, guilt feelings, and suicidal ideation. The first 17 questions are used for scoring: 8 items are scored 0-4 and 9 are scored 0-2, with a stated range of 0-50. The registry record gives severity bands of 0-7 for normal, 8-13 for mild, 14-18 for moderate, 19-22 for severe, and >=23 for very severe depression.

A continuous score retains gradations across the scale, rather than reducing every participant to a category. The posted endpoint is summarized as a mean with a standard deviation. In general, the mean describes the group’s average score, while the standard deviation describes how much individual scores vary around that average. Neither statistic alone describes each participant’s experience or establishes a difference between treatment groups.

The HAM-D is the registry’s stated instrument for assessing the primary depression outcome. Its use makes the endpoint a score on a defined measure, but the registry record does not provide details about who administered the questionnaire, how assessments were conducted, or how missing responses were handled.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean and standard deviation describe?

For a continuous endpoint such as a scale score, the mean is the arithmetic average of the recorded scores in a group. The standard deviation describes the spread of those scores around the mean. A small standard deviation indicates that the observed scores are relatively close to the group average; a larger one indicates more variation among the observed scores. These are descriptive summaries, not measures of the uncertainty in a comparison between randomized groups.

For this trial, the placebo group’s posted mean is 12.33 and its standard deviation is 3.21, based on 3 participants. The Drug Cohort’s posted mean is 6.0 and standard deviation is 5.66, based on 2 participants. Each statistic describes only the participants included in that group’s summary. In general, estimates of variability based on very few observations can themselves be unstable, and an individual score may substantially affect the mean and standard deviation.

What would a between-group analysis add?

A between-group analysis would estimate a contrast between the randomized groups and quantify uncertainty around that estimate. A confidence interval, in general, gives a range of values compatible with the observed data under the assumptions of the analysis. It is not a guarantee that the true effect lies in that range, and its interpretation depends on the statistical model and study conduct. A p-value, in general, evaluates how compatible the observed data are with a specified null hypothesis under the analysis assumptions; it is not the probability that the null hypothesis is true.

No between-group estimate, confidence interval, or p-value is reported for the primary continuous endpoint. The record indicates that values were missing or not estimable for a two-arm comparison. Thus, readers should not infer a formal test from the separate group means or assume that an unreported contrast was statistically significant or non-significant.

How does randomization matter?

Random allocation is intended to make treatment assignment independent of participants’ characteristics, on average, so that a suitably conducted comparison can support a causal interpretation. Randomization does not ensure that small groups will have similar baseline characteristics, and it cannot compensate for missing outcomes or selective analysis. The registry record lists randomized allocation, but does not report the sequence-generation method, allocation concealment, baseline characteristics, or how assignment was implemented.

The trial is listed as parallel, meaning the record describes participants as assigned to one of the groups rather than as receiving each intervention in sequence. The record also contains differing descriptions of masking: triple in the design information and double-blinded in the brief summary. It does not identify the masked roles or reconcile the descriptions. Masking can reduce the influence of treatment knowledge on care, assessment, or reporting, but the record does not provide enough detail to determine how masking operated in practice.

What do the planning assumptions mean?

In general, a sample-size calculation estimates how many participants may be needed to detect a specified effect under chosen assumptions. The registry record characterizes its planning approach as an illustration based on the observed effect, with two-sided significance level 0.05, equal allocation, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect. Such planning figures, when available, are illustrations rather than observed power and do not establish that the trial was designed around those assumptions.

The record does not report the trial’s own sample-size justification. The planning assumptions also do not replace a reported analysis of the outcome data. In particular, a calculation based on an observed effect cannot establish that the effect is reproducible, and a planning illustration cannot show that the completed trial had adequate precision. No planning result is stated in the registry record analyzed here, so no participant target for a particular power level can be inferred.

Why does over-dispersion matter for count outcomes?

Over-dispersion is a general issue for count data: it occurs when observed variability is greater than a statistical model assumes. It can make standard errors too small and uncertainty look narrower if the model does not account for the extra variation. The primary endpoint in this trial is continuous, not a count endpoint, so count-model over-dispersion is not the central issue for interpreting the posted HAM-D summaries. The record does not report a count-model analysis for the primary outcome.

More broadly, every statistical summary depends on the observations included and the assumptions used. Here, the small outcome-group sizes, the absence of a reported comparative analysis, and missing details about the analysis population limit what can be concluded from the descriptive means and standard deviations. General statistical principles can explain what these summaries mean, but they cannot fill in methods that are not reported in the registry record.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Placebo", "Drug Cohort")
n   <- c(3, 2)
m   <- c(12.33, 6)
sp  <- c(3.21, 5.66)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Placebo  3  12.330  3.210  4.356  20.304
CARM  Drug Cohort  2  6.000  5.660  -44.853  56.853

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists secondary outcomes relating to anxiety, fatigue, healthcare utilization, social support, sleep disturbance, and function; function is listed twice. It also lists 9 outcome measures and 7 posted secondary outcomes. Multiple outcomes create a general multiplicity concern: when many hypotheses are tested, the chance of at least one apparently noteworthy result can increase. Methods for addressing multiplicity may include defining a primary comparison in advance, adjusting inference, or interpreting secondary findings as exploratory. Whether any such approach was planned or used is not reported in the registry record. No secondary analyses are posted.

Missing-data methods are not reported in the registry record. The primary-outcome population is labeled “Patients withdrew,” but the registry record does not explain whether this refers to the analysis population, participant disposition, or something else. It does not report reasons for missing outcomes, the timing of withdrawals, imputation, sensitivity analyses, or assumptions about participants without a one-year score. The available information therefore does not support conclusions about whether missingness could have influenced the summaries.

The posted primary endpoint summaries include 3 placebo participants and 2 Drug Cohort participants, while listed enrollment is 8. The record does not explain the difference. It also does not report whether the analysis followed intention-to-treat, per-protocol, or another population definition. Those populations can answer different questions, so the analysis set should not be inferred from the group labels or from enrollment alone.

12. 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 5% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: meningitis (25.0%).

The safety record lists 4 participants at risk in each group. It reports 1 participant with a serious adverse event in the placebo group and 0 in the Drug Cohort. Other adverse events are listed for 0 participants in each group, and deaths are listed as 0 in both groups. These counts describe the registry’s safety entries; they do not by themselves establish that an event was caused by an intervention or that no unlisted event occurred.

Meningitis is listed among the top serious terms with 25.0%. The registry record does not clarify the denominator or additional context for that percentage in the term listing. It should not be treated as a comparative event rate or as evidence of a treatment relationship.

The safety groups each list 4 participants at risk, while the outcome summaries include 3 placebo participants and 2 Drug Cohort participants and listed enrollment is 8. The record does not explain how these group counts relate to the outcome populations. With small group counts, the safety entries are descriptive and provide limited information about the frequency or range of possible harms.

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

The posted result is a set of one-year depression-score summaries, not a reported treatment-effect estimate. The placebo mean is 12.33, based on 3 participants, and the Drug Cohort mean is 6.0, based on 2 participants. The corresponding standard deviations are 3.21 and 5.66. These numbers describe observed scores in the posted groups; they do not quantify the uncertainty in a comparison or show how scores changed from baseline.

The registry record does not report a two-arm analysis, confidence interval, or p-value for this endpoint. The numerical ordering of the means alone cannot establish whether the groups differed beyond what could arise from sampling variation. Nor does the lack of a reported analysis establish that there was no difference. The record’s absence of a comparative result is an important limitation on interpretation.

The result also does not establish what happened to all enrolled participants. The reason for the outcome sample sizes, the meaning of the “Patients withdrew” population label, and the handling of missing outcomes are not reported in the registry record. No individual scores or severity-category counts are posted, so the group means cannot show how many participants had scores in a particular HAM-D category.

More generally, a randomized design can support a meaningful comparison when outcomes and analyses are available and the study is conducted and reported appropriately. The registry record identifies a randomized comparison, but the posted primary result is too limited to support a precise comparative conclusion. These observations should not be taken as medical advice or as proof that either intervention does or does not have an effect.

14. Why This Trial Matters Statistically

This trial illustrates why a randomized label does not, on its own, make every posted result a comparative treatment estimate. The record describes random allocation and a placebo group, but its posted primary outcome consists of group means and standard deviations without a reported two-arm analysis. For statistical interpretation, the distinction between descriptive summaries and an estimated contrast is essential.

The small outcome groups make that distinction especially important. The reported mean and standard deviation are based on 3 placebo participants and 2 Drug Cohort participants. A reader can describe the values, but the record does not provide the uncertainty needed to judge how precisely they estimate a group contrast. The difference in group means should not be converted into a claim about treatment effect without a suitable analysis and adequate context.

The record also highlights how endpoint definitions and analysis populations shape interpretation. It names HAM-D at one year, but does not report baseline-to-follow-up change, missing-data methods, or a clear analysis-population definition. The population label “Patients withdrew” is not explained. These omissions leave open questions about which participants the summaries represent and how the posted values relate to the planned trial question.

Finally, the differing masking descriptions and the gap between listed enrollment and outcome sample sizes show why readers need clear reporting alongside design labels. The registry record is useful for identifying the intended comparison, measurement, and posted summaries, while also leaving key aspects of analysis and conduct unreported. Statistically, the central lesson is to match the strength of interpretation to the information actually available.

15. 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 primary outcome is reported for 3 placebo participants and 2 Drug Cohort participants, despite listed enrollment of 8. The registry record does not explain why the outcome group counts differ from enrollment, identify the analysis population, or clarify the primary-population label “Patients withdrew.” The results may therefore represent a restricted group, but the record does not establish how participants were selected for these summaries.

Baseline HAM-D results, change scores, individual observations, and severity-category counts are not reported in the registry record. The posted group means cannot show whether the groups differed at baseline, how participants’ scores changed, or how many participants fell within each severity band. The record also does not describe the administration of the HAM-D or the approach to missing questionnaire items and follow-up outcomes.

No comparative analysis, confidence interval, or p-value is posted for the continuous primary endpoint. The record does not report covariate adjustment, a statistical analysis plan, multiplicity procedures, or the trial’s own sample-size justification. It also does not provide reasons for missing outcomes or sensitivity analyses that could assess the effect of missing data.

The design information lists triple masking, while the brief summary describes a double-blinded trial. The record does not reconcile these descriptions or specify who was masked. The safety entries are also limited: each group lists 4 participants at risk, but their relationship to the outcome groups is not explained. These trial-specific reporting gaps limit assessment of precision, possible bias, and the consistency between intended design and posted results.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Poststroke Depression in Hemorrhagic Stroke — NCT03826875.
• Registry record last updated 2025-08-06; status last verified 2025-07.
• Results first posted 2025-08-06.
• 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.

19. Record Summary

The registry record describes a completed, phase 2 randomized parallel trial of fluoxetine versus placebo for preventative depression treatment in survivors of aneurysmal subarachnoid hemorrhage. Both groups were initially prescribed their assigned intervention at 20mg/day for one year. Depression at one year was assessed with the HAM-D.

The posted placebo summary is a mean of 12.33 with standard deviation 3.21 in 3 participants; the Drug Cohort summary is a mean of 6.0 with standard deviation 5.66 in 2 participants. No two-arm analysis is reported for this continuous outcome. Listed enrollment is 8, and the reason for the difference from the outcome sample sizes is not reported in the registry record.

The record lists 7 secondary outcomes, 9 outcome measures, and no posted analyses. Missing-data handling, the analysis population, and a sample-size justification are not reported. The posted summaries should therefore be read as descriptive rather than as a precise estimate of a treatment-group effect.