Expectation of Unpleasant Events in Anxiety Disorders: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Expectation of Unpleasant Events in Anxiety Disorders (NCT00055224), a randomized na study. The primary outcome is Change in Percent of Correct Button Presses During Threat Compared to the Safe Condition (AAST).

2. Clinical Question

The broader study examined responses to unpleasant learning experiences in volunteers with varying anxiety levels and in people with anxiety disorders. Its registry description includes several substudies and tasks, while this endpoint pertains specifically to substudy 7 and the active avoidance signal task. The endpoint result should therefore be read as a result for this task and analysis population, not as a summary of every study activity or every participant enrolled.

3. Trial Design

The registry record describes a study of threat-related responses in healthy participants and people with anxiety disorders. Its stated purpose was to examine changes in the body and brain during unpleasant learning experiences and to investigate expectations of unpleasant events. The study included multiple substudies, including tasks involving predictable and unpredictable threat, working memory, face Stroop, and active avoidance and sustained attention. The primary endpoint is specifically from substudy 7, the active avoidance signal task.

The record lists an overall enrollment of 921 and an endpoint analysis population of 22.0 healthy participants and 23.0 anxiety patients. These figures refer to different scopes: overall enrollment is for the registered study, whereas the endpoint counts apply to the analysis described for this task. The endpoint population description notes that two healthy controls were excluded from analyses. The registry does not report further details about how the endpoint population was selected or how participants entered this substudy.

The intervention is listed as threat of shock, and the endpoint compares task performance during threat with performance during a safe condition. In the safe condition, no shock could be administered. The reported group comparison is between anxiety patients and healthy participants; it is not a randomized comparison of those diagnoses. The record labels allocation as randomized, masking as none, and the model as single group. It does not explain how those design labels relate to the participant-group comparison for this endpoint, so they should not be taken to imply that diagnosis or anxiety status was assigned.

The record lists the study status as completed and gives the start date as 2003-03-24 and primary completion date as 2020-03-11. It identifies the National Institute of Mental Health as sponsor. A protocol-specific account of the original design rationale, recruitment plan, or sample-size justification is not reported in the registry record.

4. Participant Flow and Accounting

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

Before assignment: All sub-studies were conducted at different time points so subjects had the option to participate in multiple sub-studies.

Overall Study Participants

Substudy 1

Substudy 2

Substudy 5

Analysis population for the primary outcome, as defined in the registry: Analysis included only participants in sub-study 7 who completed the active avoidance signal task (AAST). Two healthy controls were excluded from analyses.

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 Change in Percent of Correct Button Presses During Threat Compared to the Safe Condition (AAST), measured over 2000 ms during trial.

The primary endpoint is the change in the percentage of correct responses during threat compared with the safe condition on the active avoidance signal task. Participants performed a go/no-go task: they were asked to respond to frequent go stimuli and withhold a response to infrequent no-go stimuli. Accuracy was assessed separately by condition and trial type. A correct go response was a response during the trial, and a correct no-go response was withholding a response during the trial. The registry describes the measurement window as 2000 ms during trial.

This measure is used to describe task performance under threat relative to safety. A higher percentage of correct responses during threat compared with safety was defined in the registry description as improvement in accuracy. The endpoint therefore represents a within-person condition contrast, while the posted comparison asks whether that contrast differs between anxiety patients and healthy participants.

The endpoint is a continuous measure reported as a mean with standard error. The healthy-participant mean is -0.016, with standard error 0.03; the anxiety-patient mean is 0.008, with standard error 0.04. The registry description does not clarify in the outcome title whether the change is encoded as a proportion or percentage-point scale, so the estimates are best retained in the reported form rather than converted to another scale. This task-specific performance measure does not by itself describe all responses to threat, all symptoms, or the results of the study's other tasks.

7. Results Posted in the Registry

• Substudy 7/ Active Avoidance Signal Task (AAST) - Anxiety Patients: mean 0.01 versus -0.02 on control; difference 0.02 (95% CI -0.07 to 0.12), standardized effect d = 0.14; z-test p = 0.631.

8. Statistical Interpretation

9. Statistical Methods Explained

The endpoint is a continuous measure, summarized as a mean and standard error. A mean describes the average value in a group. A standard error describes the statistical uncertainty in an estimated mean; it is not the same as the spread of individual participant values. The registry record gives standard errors of 0.03 for healthy participants and 0.04 for anxiety patients. The endpoint-specific group counts are 22.0 and 23.0, respectively.

What does the group difference represent?

The primary endpoint first compares task accuracy during threat with accuracy during safety. The group-level means then summarize that condition contrast separately for healthy participants and anxiety patients. The unadjusted recalculation of the difference between those means is 0.024. This is a contrast between groups in the reported endpoint, not a comparison of the raw accuracy scores in only one task condition. It is also not an estimate of the effect of assigning a participant to an anxiety diagnosis.

The comparison uses the posted group summaries and is unadjusted. An unadjusted analysis considers the stated group contrast without accounting for other variables or additional modeling choices. The registry record does not report covariate adjustment, a model specification, or an analysis plan for the endpoint. Because the record lists no posted statistical analyses, these inferential calculations should be understood as recalculations from the summary information rather than as a reported sponsor analysis. The discrepancy indicator is false, so there is no stated discrepancy between a sponsor estimate and this recalculation to resolve.

How should the confidence interval be understood?

A confidence interval gives a range of estimates compatible with the data under the calculation's assumptions and method. It communicates precision as well as the direction and size of an estimate. For the group difference, the interval is -0.074 to 0.122. It spans negative and positive values, including zero. That breadth means the estimate is not precise enough to isolate a single direction or a narrow range of plausible group differences. The interval is not a guarantee that the true value lies within its endpoints, nor is it a measure of the probability that a particular hypothesis is true.

The p-value of 0.631 is also an unadjusted recalculation. A p-value summarizes how unusual data at least as discrepant as those observed would be under a specified null model and its assumptions. It is not a measure of clinical importance, and it should not be read as the probability that there is no difference. The record does not give a sponsor-reported p-value or test method for this endpoint.

Why can a mean-based comparison have uncertainty?

Even when the endpoint is a mean, an estimate varies from sample to sample. Standard errors and confidence intervals represent this sampling uncertainty, subject to assumptions about the data and calculation. General statistical principle: small analysis groups often yield less precise estimates than larger groups when other features are similar. The actual precision also depends on variability, the analysis method, and the distribution of the measure; those details are not fully described in the registry record.

The registry also reports a planning illustration based on the observed effect. Such illustrations ask how many participants per arm might be needed under stated assumptions to detect an effect of a selected size with a stated power. They are forward-looking planning calculations, not a reconstruction of the study's original design and not evidence about what power the completed study actually had. The original sample-size justification and design assumptions are not reported in the registry record.

What does over-dispersion mean in general?

General statistical principle: over-dispersion occurs when observed variation exceeds what a model assumes. When a model understates variation, its standard errors can be too small and its intervals too narrow, making evidence appear more precise than it is. This endpoint is reported as a continuous measure with means and standard errors; the registry record does not report an over-dispersion assessment or a model for over-dispersed data. The concept is relevant as a general modeling consideration, not as a finding about this endpoint.

Finally, the posted endpoint population is task-specific. Results from this subset should not be treated as if they represent every participant in the overall enrollment or every task in the research program. The registry record does not report enough detail to assess all assumptions behind the analysis, including data distributions, handling of incomplete task observations, or the specific inferential procedure used by the sponsor.

10. Confidence Intervals and Precision

The recalculated group difference is 0.024, with a confidence interval from -0.074 to 0.122. The interval is broad relative to the point estimate and crosses zero. Thus, the estimate does not pin down a single direction: values indicating a negative difference, no difference, or a positive difference all fall within the interval. This is a statement about uncertainty in the estimate, not a conclusion that all values in the interval are equally likely or clinically equivalent.

The group means are -0.016 for healthy participants and 0.008 for anxiety patients. Their reported standard errors are 0.03 and 0.04. Standard errors describe uncertainty around estimated means, not the range of individual responses. The confidence interval for the difference is the more direct summary of uncertainty in the group contrast. No confidence limits are posted for the individual group means in the registry record.

The analysis population is limited to 22.0 healthy participants and 23.0 anxiety patients. With a task-specific analysis of this size, the estimate may be sensitive to participant-level variation, but the registry record does not provide individual data or enough distributional information to assess that sensitivity. The interval should therefore be considered alongside the endpoint definition and the fact that the analysis was limited to participants included in this substudy.

The interval and p-value discussed are from an unadjusted recalculation based on posted summary statistics; the registry record does not report a sponsor analysis for this endpoint. No clinical or practical threshold for interpreting the interval is reported. Consequently, the data support describing the estimated contrast and its uncertainty, but not a definitive claim that the groups have identical responses or that one group has a reliably greater response.

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 sample-size illustration gives participants per arm, not total event counts. It uses two-sided significance level 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to the stated fraction of the observed effect. The planning rows consider 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. These are planning scenarios, not the study's original design assumptions; the original sample-size justification is not reported in the registry record.

The 80% power column gives the number of participants per arm for 80% power under the stated planning assumptions and effect scenario. The 90% power column gives participants per arm for 90% power under those same assumptions. Power is the probability, under a particular assumed effect and statistical model, of detecting an effect according to the chosen significance rule. A higher target power generally calls for a larger planned sample when other assumptions are held constant.

The listed planning requirements are 777 and 1,040 participants per arm for 80% and 90% power when planning for 100% of the observed effect; 1,381 and 1,849 when planning for 75% of the observed effect; and 3,107 and 4,159 when planning for 50% of the observed effect. The figures increase as the effect to be detected becomes smaller, because smaller differences are harder to distinguish from sampling variation.

These calculations are planning illustrations based on the observed effect, not observed power for the completed study. They do not show that the original study was designed around these assumptions or that its analysis had a particular probability of detecting an effect.

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 = 0.024; pooled standard deviation σ = 0.169.
2. n = 2 × (1.96 + 0.842)² × 0.169² / 0.024² = 776.6, rounded up to 777 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 <- 0.024                      # observed difference in means
sigma <- 0.168812542243178                     # 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(55224)
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 = 777 (80% power), 1,040 (90% power)
 75% of observed effect: n per arm = 1,381 (80% power), 1,849 (90% power)
 50% of observed effect: n per arm = 3,107 (80% power), 4,159 (90% power)
power.t.test: 777.6 
Simulated power at n = 777 per arm: 0.801 (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 describes multiple substudies and tasks, and lists 24 outcome measures. It does not report a multiplicity adjustment or explain how the primary endpoint was selected in relation to the other outcomes. When many outcomes or comparisons are examined, the chance of at least one apparently unusual result can increase; statistical approaches to multiplicity can address that issue, but the record does not identify whether any such approach was used. The unadjusted recalculated p-value should therefore not be interpreted as a multiplicity-adjusted result.

The record lists zero posted statistical analyses and zero posted secondary results for this summary. It does not report a statistical analysis plan, a prespecified hierarchy of outcomes, or whether all planned analyses were completed. This absence limits assessment of whether the comparison was specified in advance, how the analysis was selected, and whether other analyses of the same endpoint were conducted.

The endpoint population description says the analysis included participants who completed the active avoidance signal task and that two healthy controls were excluded from analyses. The record does not describe the reasons for those exclusions, the number of participants who began but did not complete the task, or the handling of incomplete observations. It also does not report an imputation method or sensitivity analyses for missing data. The reported endpoint counts should be read as the analysis group sizes, not as proof that there were no missing task observations.

The registry labels allocation as randomized, but the outcome comparison is between anxiety patients and healthy participants, whose diagnostic group is not described as assigned. The record does not explain the randomization procedure, its role in this endpoint, or whether randomization applied to task conditions. It also does not report whether the analysis followed an intention-to-treat principle or used another analysis population. These design and analysis details are not reported in the registry record, so the recalculated group contrast cannot be assumed to reflect a particular protocol-defined estimand.

14. 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.

The registry lists serious adverse events, other adverse events, and deaths for several task and pilot groups. In every listed group, the reported count of serious adverse events is 0, the count of other adverse events is 0, and the number of deaths is 0. The groups have different numbers of participants at risk, including groups for the predictable and unpredictable response task, working memory, face Stroop, active avoidance and sustained attention, and pilot tasks.

These figures are registry counts for the listed groups. They should not be interpreted as evidence that no event of any kind occurred outside the defined reporting categories or observation periods. The record does not provide detailed event narratives, follow-up duration for safety reporting, or additional subgroup-specific context. The endpoint analysis population is also distinct from some of the safety groups, so safety counts should not be treated as if they were collected only from the 22.0 healthy participants and 23.0 anxiety patients in the endpoint comparison.

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

The posted endpoint summaries show a mean change of -0.016 among healthy participants and 0.008 among anxiety patients for correct responses during threat compared with safety. The unadjusted recalculated difference is 0.024, and its confidence interval ranges from -0.074 to 0.122. The interval includes zero and extends in both directions. The estimate is therefore uncertain about the direction and size of the group contrast.

These results do not establish that the groups are the same. A wide interval that includes zero can reflect limited precision as well as a small underlying difference. Nor do the results establish that one group responds more accurately under threat. The registry record does not report a clinically meaningful threshold for the endpoint, and the task measure alone does not establish the importance of a difference in broader behavioral or clinical terms.

The comparison is specific to an active avoidance signal task and to participants included in substudy 7. It does not represent all 921 participants enrolled across the study, all anxiety disorders, or all the other tasks described in the registry. It is not a randomized comparison of anxiety patients with healthy participants, even though the record lists randomized allocation as a design label. No causal interpretation of group status follows from the posted means.

The calculations are unadjusted recalculations from the posted summaries because the registry record does not post a sponsor statistical analysis for this endpoint. Details such as covariate adjustment, missing-data handling, and multiplicity procedures are not reported. The most defensible reading is descriptive: the group means differ by the stated estimate in this analysis, but uncertainty is substantial, and the record does not provide enough analysis detail to support stronger conclusions about the underlying response or its clinical significance.

16. Why This Trial Matters Statistically

This result illustrates why an endpoint's scope matters. The overall study enrolled 921 participants and examined several forms of threat-related response, but the reported endpoint concerns a much narrower analysis: completion of the active avoidance signal task in substudy 7. Separating the task-specific analysis from the broader research program prevents an endpoint estimate from being mistaken for a finding about every participant or every task.

It also illustrates the value of reporting uncertainty alongside a point estimate. The group difference of 0.024 is accompanied by a confidence interval from -0.074 to 0.122. Reading only the point estimate would hide how broadly the data are compatible with different group contrasts. The interval makes clear that the estimate is not precise enough to settle the direction of the difference.

The record also highlights a distinction between registry outcome summaries and a complete statistical report. It provides group means, standard errors, and analysis counts, but no posted statistical analyses, no reported analysis plan, and no details about missing data or multiplicity. Recalculations can help describe the posted summaries, but they cannot fill in protocol decisions that are not documented.

Finally, planning calculations based on an observed effect are not retrospective evidence of a trial's power. Their assumptions should be kept explicit, and their role should remain planning. Together, these points make this endpoint useful for teaching careful interpretation: identify the estimand, match the population to the outcome, report uncertainty, and distinguish the registry entries from unreported methods.

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 endpoint is drawn from one substudy within a larger study that included several tasks. Its analysis population is limited to participants who completed the active avoidance signal task, and the registry description notes that two healthy controls were excluded. The reasons for exclusion and the broader flow of participants through the task are not reported. The endpoint counts are much narrower in scope than the overall enrollment of 921.

The registry record does not report the original sample-size justification, endpoint-specific design assumptions, or a statistical analysis plan. It lists randomized allocation but does not explain how randomization relates to a comparison between anxiety patients and healthy participants. Masking is listed as none, but the record does not discuss whether outcome assessment could be influenced by that design. These missing details limit assessment of design and analysis choices.

The reported endpoint is a task-performance measure, with the stated measurement period of 2000 ms during trial. The registry does not report individual-level data, detailed distributions, or confidence limits for each group mean. It also does not clarify the scale interpretation beyond the outcome unit and description. The group comparison is therefore best retained in its reported form rather than translated into a different scale or clinical interpretation.

The registry record does not report missing-data methods, multiplicity control, adjusted analyses, or sensitivity analyses. The unadjusted recalculation cannot address these omissions. In addition, no clinical importance threshold is reported for the endpoint. These limitations do not invalidate the descriptive summaries, but they constrain how confidently the difference can be generalized or interpreted beyond this task-specific analysis.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Expectation of Unpleasant Events in Anxiety Disorders — NCT00055224.
• Registry record last updated 2024-01-09; status last verified 2022-08.
• Results first posted 2024-01-09.
• 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 registry record describes a completed study of predictability and aversive expectancies in anxiety and depressive disorders, sponsored by the National Institute of Mental Health. It lists overall enrollment of 921, randomized allocation, no masking, and a single-group model. The study included healthy participants and anxiety patients across multiple substudies.

The primary endpoint considered in this statistical analysis measures change in correct responses during threat compared with safety on the active avoidance signal task. Its analysis included 22.0 healthy participants and 23.0 anxiety patients. The posted means are -0.016 and 0.008, respectively; the unadjusted recalculated difference is 0.024, with a confidence interval from -0.074 to 0.122 and p-value 0.631.

The registry record does not post a sponsor statistical analysis for this endpoint or report the original sample-size justification, missing-data procedures, multiplicity approach, or detailed analysis plan. Safety listings report zero serious adverse events, zero other adverse events, and zero deaths in each listed group.