Clinical Investigation of Wysa: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Clinical Investigation of Wysa (NCT05533190), a randomized na study. The primary outcome is Depression Severity.

2. Clinical Question

The study concerns access to digital support during the referral and waiting period, rather than a comparison of Wysa with a completed course of IAPT treatment. The registry describes Wysa as an early-intervention and support tool that patients could explore while waiting for assessment and treatment. Its primary outcome is a patient-reported depression score; anxiety severity, uptake, dropout, app use, engagement, acceptability, crisis identification and general health state are among the listed secondary outcomes.

The posted primary comparison is a mean difference of 0.87 PHQ-9 units, with the Wysa group mean higher than the waitlist mean. The confidence interval extends from -2.132 to 3.872, so the data are compatible with differences in either direction as well as with a difference near zero. The estimate is not precise enough to establish a clear separation between groups. Interpretation also depends on how the outcomes were collected and analysed; detailed information about analysis populations and missing-data handling is not reported in the registry record.

3. Trial Design

The registry describes a completed randomized, open-label, parallel-group trial of Wysa compared with a waitlist control. It lists the study phase as not applicable. The sponsor is the University of Plymouth. The registry dates the study start to 2022-12-13 and primary completion to 2023-07-23.

The intervention was the Wysa AI chatbot mental health app. The description characterizes it as a guided self-help and triaging tool delivered through an app or widget. It uses Natural Language Processing to understand users’ written inputs but not to generate responses. The waitlist comparator was selected to represent patients awaiting standard IAPT assessment and treatment. Because masking was none, participants and study personnel were not masked to group assignment, according to the registry design fields. The precise procedures used to maintain allocation concealment are not reported in the registry record.

The registry records enrollment of 76. For the primary outcome at 3 months post-randomisation, it reports a PHQ-9 mean and standard deviation for 30 participants in the Wysa arm and 16 in the waitlist arm. These outcome counts should not be treated as an explanation of who was included, excluded, or missing; the registry record does not report the analysis population or the reasons for the difference between enrollment and those outcome counts.

Depression severity was the primary outcome. The registry also lists secondary measures covering anxiety, uptake, dropout, crisis identification, app use, engagement, acceptability and general health state. The posted record contains results for anxiety severity, uptake rates and dropout rates. It lists 12 outcome measures overall, while 11 secondary outcomes are posted. Detailed protocol information, including the original sample-size justification and the planned statistical analysis, 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.

Overall Study

Reasons for not completing this period:

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 Severity, measured over 3 months post-randomisation.

The primary endpoint is depression severity measured by the PHQ-9 at 3 months post-randomisation. The registry describes the outcome as a mean score in units on a scale, with a standard deviation. Total scores range from 0 to 27, and higher scores indicate more severe depression.

A continuous score preserves information about the degree of symptoms rather than reducing each participant’s result to a category. Comparing group means can therefore describe the average difference in reported depression severity at the stated follow-up. The standard deviation describes how spread out individual scores are within each group; it is not the uncertainty around the group mean or the treatment difference.

The PHQ-9 is used as the study’s standard measure of depression symptoms. It gives the primary comparison a defined outcome and timeframe, but a group mean alone does not describe every participant’s experience or establish why any difference occurred. The registry record does not report a threshold for a clinically important change or provide a responder analysis for this primary outcome.

At the stated follow-up, the posted mean was 9.28 in the Wysa arm and 8.41 in the waitlist arm. The mean difference is expressed as Wysa minus waitlist, so the positive estimate places the Wysa group mean higher on this scale. The confidence interval is important to interpretation: it includes values on both sides of no difference and is broad enough to leave substantial uncertainty about the direction and size of the group difference.

7. Results Posted in the Registry

• Wysa AI Chatbot Mental Health App: mean 9.28 versus 8.41 on control; difference 0.87 (95% CI -2.13 to 3.87), standardized effect d = 0.15; z-test p = 0.570.

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.

Anxiety Severity

• 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 deviations.
• Result:
  • Wysa AI Chatbot Mental Health App: mean 8.98 versus 8.19 on control; difference 0.79 (95% CI -1.91 to 3.49), standardized effect d = 0.15; z-test p = 0.567.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Wysa AI Chatbot Mental Health App and the control. The interval gives the range of effects that remain compatible with the data.

Uptake Rates

• 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:
  • Wysa AI Chatbot Mental Health App: 56.6% (30/53) versus 69.6% (16/23) on control; risk difference -13.0 percentage points (95% CI -36.0 to 10.1), risk ratio 0.81 (0.57–1.16); unadjusted two-proportion z-test p = 0.288.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Wysa AI Chatbot Mental Health App and the control. The interval gives the range of effects that remain compatible with the data.

Dropout Rates

• 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:
  • Wysa AI Chatbot Mental Health App: 43.4% (23/53) versus 30.4% (7/23) on control; risk difference 13.0 percentage points (95% CI -10.1 to 36.0), risk ratio 1.43 (0.72–2.84); unadjusted two-proportion z-test p = 0.288.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Wysa AI Chatbot Mental Health App and the control. The interval gives the range of effects that remain compatible with the data.

The registry posts secondary results for anxiety severity, uptake rates and dropout rates. Their overall statistical pattern is one of uncertainty: the continuous anxiety comparison and the binary uptake and dropout comparisons have confidence intervals that include no difference, and their p-values are unadjusted. These findings do not provide a clear, precise contrast between Wysa and the waitlist group.

The anxiety outcome is a symptom measure related to the primary depression measure, but it is a separate endpoint and cannot be treated as a replication of the PHQ-9 result. Uptake and dropout describe participation or study engagement rather than symptom severity. They may help characterize how participants interacted with the study or intervention, but they answer different questions from the primary endpoint.

These outcomes are relevant context for the primary comparison, but they do not change its estimate or resolve its uncertainty. A coherent reading is that the posted secondary comparisons, like the primary result, do not establish a clear group difference on the outcomes reported. The registry does not provide results for every secondary outcome it lists, so no overall claim about app use, engagement, acceptability, crisis identification or general health state can be drawn from the posted secondary comparisons alone.

Unadjusted results carry less confirmatory weight when multiple outcomes are considered. General statistical explanation: testing many outcomes increases the chance that at least one result may appear notable by chance, even when no corresponding effect exists. Methods such as a prespecified hierarchy or multiplicity adjustment can address this issue, but no such procedure is reported in the registry record. The secondary comparisons should therefore be treated as descriptive or exploratory unless further analysis documentation establishes otherwise, rather than as independent confirmation of the primary result.

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

• Anxiety Severity
• Crisis Identification
• Uptake Rates
• Dropout Rates
• Use of the App
• Engagement
• Patient Perceptions of Acceptability
• General Health State

9. Statistical Interpretation

10. Statistical Methods Explained

What does the mean difference represent?

The primary outcome is continuous, so the comparison is expressed as a difference between group means. The estimate is calculated in the order Wysa minus waitlist. A positive value therefore means the average PHQ-9 score was higher in the Wysa arm, while a negative value would mean it was lower. This direction is about the scale score; it does not by itself establish a causal effect or the importance of a difference.

The posted means are 9.28 for Wysa and 8.41 for waitlist, and the mean difference is 0.87. The standard deviations, 6.56 and 3.82 respectively, describe the variability of scores among participants in each group. A standard deviation is not a confidence interval and should not be used as though it directly gives the precision of the group comparison.

How should a confidence interval be read?

General statistical explanation: a confidence interval represents the range of estimates reasonably compatible with the observed data and the method’s assumptions. It communicates uncertainty about the estimated difference, rather than the spread of individual scores. The primary interval, from -2.132 to 3.872, crosses zero, the value corresponding to no mean difference. It therefore does not rule out either direction of difference. Its width also shows why the point estimate alone is an incomplete description.

A confidence interval is not a guarantee that the true effect lies inside its endpoints, and it is not a range containing most participants’ scores. Its interpretation depends on the study design, measurement, analysis and assumptions. The registry record does not report enough detail to assess all of those assumptions or the handling of missing outcomes.

What does the unadjusted p-value say?

The primary unadjusted p-value is 0.57. General statistical explanation: a p-value describes how unusual data at least as inconsistent with a specified null hypothesis would be if that null hypothesis and the analysis assumptions were true. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. A p-value that does not cross a chosen threshold is not proof of no difference. The registry record does not report a prespecified analysis threshold for the primary endpoint.

The word “unadjusted” matters because a p-value may not account for testing multiple outcomes or comparisons. The registry lists multiple outcome measures, and it does not report a multiplicity procedure. Accordingly, these p-values should be read as unadjusted comparisons, not as a complete confirmatory analysis accounting for all testing.

Why do sample sizes and variability matter?

General statistical explanation: estimates become less precise when there is substantial variation in scores or limited information for estimating the group contrast. The observed primary outcome counts are 30 for Wysa and 16 for waitlist. The group standard deviations are not identical, and scores show variability within both arms. These features help contextualize the interval, but they do not identify why the outcome counts differ from the enrollment recorded for the trial.

For a continuous endpoint, the mean captures an average, while the standard deviation describes spread around that average. Neither alone conveys the full distribution of scores. The registry record does not provide individual-level data, baseline outcome summaries, or a reported adjusted model, so it is not possible from the record to evaluate the balance or effect of baseline score differences on the posted comparison.

What is the role of the planning illustrations?

The planning figures use the observed effect as a starting point and show illustrative participant requirements under the stated assumptions. They are not the number enrolled in this trial and do not indicate the trial’s original design rationale. Smaller target effects require more participants to distinguish them from no difference with the same power assumptions. These illustrations are planning aids, not evidence about the probability that the trial result is true and not observed power.

General statistical explanation: power is the probability that a study would reject a specified null hypothesis under a particular assumed effect and analysis, if those assumptions hold. A planning calculation needs an assumed effect size, significance level, allocation, outcome variability and analytic method. The illustration’s assumptions are reported, but the trial’s own original design assumptions and sample-size justification are not reported in the registry record.

What can and cannot be inferred from the posted analysis?

The posted arm summaries and comparisons describe outcomes at the specified follow-up. Randomization is intended to support a fair comparison by allocating participants to groups, but the registry record’s design label alone does not confirm how allocation was generated or concealed. The open-label design means masking was none, and the record does not describe any steps taken to reduce the influence of awareness of assignment on reporting or outcome collection.

The registry record does not report the planned analysis population, baseline-adjustment strategy, treatment of missing measurements, or any interim analyses. Those details should not be inferred from the presence of group means, confidence intervals or p-values. The reported comparison is therefore best read as the posted group-level estimate with its stated uncertainty, rather than as a fully documented account of the analysis plan.

11. Confidence Intervals and Precision

The primary mean difference is 0.87 PHQ-9 units, with a 95% confidence interval from -2.132 to 3.872. The point estimate is a single best estimate from the posted comparison, while the interval shows substantial uncertainty about the size and direction of the difference. Because the interval includes zero, the result is compatible with no mean difference as well as with differences favoring either group in terms of lower symptom scores.

The interval’s lower endpoint is negative and its upper endpoint positive. This does not mean that every value in that range is equally likely, nor does it establish that all values are clinically plausible. The record does not state a threshold for clinical importance, so the interval cannot be divided into clinically meaningful and unimportant effects using the available information.

At the primary follow-up, the registry reports outcomes for 30 participants in the Wysa arm and 16 in the waitlist arm. The standard deviations were 6.56 and 3.82, respectively. These measure score variability within groups, whereas the confidence interval describes precision of the estimated contrast. Confusing these quantities can lead to an incorrect sense of how certain the group difference is.

The anxiety comparison is also imprecise: the mean difference is 0.79, and its confidence interval ranges from -1.912 to 3.492. The registry also posts intervals for uptake and dropout, each spanning no difference. These related findings do not make the primary interval narrower or remove its uncertainty. Nor do the posted estimates establish equivalence, since no equivalence margin is reported in the registry record.

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 illustrative requirements expressed as participants per arm for 80% power and participants per arm for 90% power. They are not total sample sizes, and the stated sample-size assumptions specify equal allocation between two arms. For the continuous and binary outcomes posted, the figures refer to participants per arm, not total events.

The calculation 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 set to the stated fraction of the observed effect. The figures are planning illustrations computed from the observed effect, not the trial’s own design. The original sample-size justification and assumptions for this trial are not reported in the registry record.

In plain terms, the 80% power column gives the illustrative number of participants per arm for a study designed to have an 80% chance of detecting the assumed effect under the planning assumptions. The 90% power column uses a higher target chance of detection, and therefore has larger participant requirements. These are conditional design calculations; they do not tell whether the trial’s result is correct or how likely it is that a real effect exists.

The rows represent planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. Smaller effects are harder to distinguish from no difference, so they require more participants under the same assumptions. For the primary endpoint, the figures rise from 692 participants per arm for 80% power when planning for 100% of the observed effect, to 2,766 participants per arm when planning for 50% of that effect. The corresponding 90% power figures are 926 and 3,703.

This is not observed power. It is a forward-looking illustration of sample sizes under stated assumptions, not a retrospective measure of what the completed study achieved.

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 = 0.870; pooled standard deviation σ = 5.774.
2. n = 2 × (1.96 + 0.842)² × 5.774² / 0.870² = 691.4, rounded up to 692 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 <- 0.869999999999999                      # observed difference in means
sigma <- 5.77388399291468                     # 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(5533190)
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 = 692 (80% power), 926 (90% power)
 75% of observed effect: n per arm = 1,230 (80% power), 1,646 (90% power)
 50% of observed effect: n per arm = 2,766 (80% power), 3,703 (90% power)
power.t.test: 692.4 
Simulated power at n = 692 per arm: 0.800 (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 registry lists 12 outcome measures overall and posts 11 secondary outcomes, but it reports results for only some of those secondary measures. No multiplicity adjustment, testing hierarchy, or designated confirmatory status for the secondary outcomes is reported in the registry record. When several outcomes are examined, unadjusted p-values do not account for the number of opportunities for a chance finding. The secondary p-values should therefore be interpreted cautiously and should not be treated as if each were the sole prespecified test.

The reported-analyses field contains no analyses. The registry record does not report whether the primary comparison was adjusted for baseline PHQ-9 score, whether any other covariates were included, or whether an analysis plan specified the model before results were examined. It also does not report interim analyses or stopping rules. These details cannot be inferred from the posted means and confidence intervals.

The primary outcome counts are 30 for Wysa and 16 for waitlist, whereas enrollment is reported as 76. The registry record does not explain the relationship between enrollment and the outcome counts. It does not report the analysis population, the number with missing primary outcomes, reasons for missingness, or any imputation or sensitivity-analysis method. Consequently, the extent to which missing data could affect the comparison cannot be assessed from the record.

The record also does not report whether the analysis followed an intention-to-treat approach, how participants were classified after allocation, or whether protocol deviations occurred. These omissions limit assessment of whether the posted outcome summaries preserve the benefits of randomization and how robust the comparisons are to different analysis choices. They do not establish that the analysis was inappropriate; they indicate that the relevant methods are not documented in the registry record.

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

For serious adverse events, the registry reports zero affected participants in the Wysa arm and zero in the waitlist arm. It also reports zero deaths in both arms. No serious adverse event terms are listed. These are the recorded serious-event figures for the groups and should be read as such, not as proof that no serious event could occur or that the interventions have been established as safe.

For other, non-serious adverse events, the registry reports 5 affected participants in the Wysa arm and zero in the waitlist arm. The record does not list the event descriptions or provide further detail about severity, timing, relatedness to the app, or how events were collected. The difference in recorded counts alone cannot establish that the app caused an event.

Safety interpretation is limited by the scale and duration of the reported study and by the available detail in the registry. The record does not report enough information to evaluate how completely events were monitored or whether participants in each arm had comparable opportunities to report them. These figures describe posted adverse-event reporting; they are not a substitute for a fuller safety assessment.

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

The posted primary result is a mean difference of 0.87 PHQ-9 units at 3 months post-randomisation, with a 95% confidence interval from -2.132 to 3.872. The Wysa group had the higher mean score, but the interval includes differences in either direction and no difference. The finding is therefore uncertain and does not establish a clear separation in average depression severity between the groups.

This result does not prove that Wysa works, does not work, or is equivalent to a waitlist. A non-significant unadjusted p-value is not evidence of equivalence, and the registry does not report an equivalence margin or a clinically important difference threshold. The interval also does not establish that every value it contains would be meaningful to patients. The data describe an estimate with uncertainty, rather than a definitive answer about individual benefit.

The study addresses the use of a digital support tool while people are waiting in the IAPT referral pathway. It does not, on the information reported, compare Wysa with completed standard IAPT assessment and treatment. The waitlist group is described as the current standard of care for people awaiting those services. The findings should be understood within that particular setting and comparison.

Likewise, the posted secondary outcomes do not overturn or confirm the primary finding. They concern different aspects of symptoms and participation, and their results are unadjusted. The registry record does not report results for all listed secondary outcomes or enough analysis detail to support broader conclusions about app engagement, acceptability, crisis identification or general health.

The results are best understood as a limited comparison at the reported follow-up, with uncertainty about both the magnitude and direction of group differences. They do not establish what would happen in other care settings, over other follow-up periods, or among populations not represented by this study. Such questions are not answered by the posted record.

17. Why This Trial Matters Statistically

This trial addresses a practical statistical question: whether a digital support tool offered during a waiting period is associated with a measurable difference in patient-reported symptoms compared with a waitlist control. The randomized parallel-group design provides a framework for comparing groups, while the open-label status and the registry’s limited analysis detail remain relevant when interpreting the estimates.

The primary outcome is a continuous symptom score, and the mean difference gives a direct estimate of the average group contrast. Equally important is its confidence interval. Reporting the interval alongside the estimate helps distinguish a small point estimate from a precise finding. In this trial, the interval spans both directions, so the estimate does not settle whether average depression severity differed between groups.

The record also illustrates why absence of a statistically clear result should not be confused with proof of no effect. The outcome counts, score variability and broad interval all matter to the precision of the estimate. Planning illustrations based on the observed effect further show how much larger future studies could need to be if they aim to detect smaller differences, under the stated assumptions. Those calculations do not reconstruct the original design or establish the adequacy of the completed trial’s sample size.

Finally, outcomes beyond symptom scores matter for understanding implementation, but multiple unadjusted comparisons require care. Uptake and dropout are not interchangeable with clinical outcomes, and incomplete reporting of other listed outcomes limits the overall picture. The registry record is useful for seeing both what was measured and how much uncertainty remains, while its unreported analysis and missing-data details prevent a complete statistical appraisal.

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.

The registry record leaves several trial-specific questions unanswered. It does not report the original sample-size justification, the planned primary analysis, allocation-concealment procedures, analysis populations, or missing-data methods. In particular, it does not explain why the primary outcome counts are 30 in the Wysa group and 16 in the waitlist group when enrollment is recorded as 76. The possible effect of missing outcome data on the estimate cannot therefore be assessed.

The trial was open-label, with masking listed as none. For self-reported symptom measures, awareness of group assignment may be relevant to how participants experience or report outcomes. The registry record does not describe procedures to reduce such influences, so their role cannot be evaluated from the posted information.

The reported primary endpoint is at 3 months post-randomisation. Results at other time points are not reported in the record. The primary outcome is a PHQ-9 mean, and no clinically important difference threshold or responder definition is reported. As a result, the statistical estimate cannot be translated into a registry-defined judgment about meaningful improvement.

The registry lists a range of secondary outcomes but posts only some secondary results. It does not report a multiplicity strategy, and the comparisons are unadjusted. This limits confirmatory interpretation of secondary findings and leaves uncertainty about whether any outcomes were intended as exploratory or confirmatory.

The waitlist comparison concerns the period before standard IAPT assessment and treatment, rather than a comparison with completed treatment. Results should not be generalized beyond that setting on the basis of this record. Further details about recruitment, participant characteristics, intervention exposure, adherence and reasons for dropout are not reported in the registry record.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Clinical Investigation of Wysa — NCT05533190.
• Registry record last updated 2025-02-03; status last verified 2025-01.
• Results first posted 2025-02-03.
• 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

NCT05533190, titled “Clinical Investigation of Wysa,” is a completed randomized, open-label, parallel-group study sponsored by the University of Plymouth. The registry reports enrollment of 76 and compares the Wysa AI chatbot mental health app with a waitlist control for patients in the IAPT referral pathway.

The primary outcome is PHQ-9 depression severity at 3 months post-randomisation. The posted mean difference is 0.87 units, Wysa minus waitlist, with a 95% confidence interval from -2.132 to 3.872 and an unadjusted p-value of 0.57. The interval spans no difference and both directions, leaving substantial uncertainty.

Secondary results are posted for anxiety severity, uptake and dropout. The registry reports zero serious adverse events and zero deaths in both groups, and 5 participants with other adverse events in the Wysa group and zero in the waitlist group. Details about analysis populations, missing data, multiplicity and the original sample-size rationale are not reported in the registry record.