Treatment-Resistant Depression Registry: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment-Resistant Depression Registry (NCT00320372), a registered phase not stated study. The primary outcome is Montgomery Asberg Depression Rating Scale (MADRS)% Responders (>/= 50% Improvement From Baseline).

2. Clinical Question

This is an observational comparison rather than a randomized treatment contrast. The registry record does not state how treatment allocation occurred. Patients receiving VNS and those not receiving it may differ in ways associated with both treatment choice and later depression scores, so a difference in their outcomes would not by itself establish that VNS caused the difference.

3. Trial Design

This was a long-term, prospective, observational, multicenter patient outcome registry for people with treatment-resistant depression who were in a major depressive episode. The registry record lists an enrollment of 795 and a status of completed. Its start is recorded as 2006-01 and primary completion as 2015-03. Cyberonics, Inc. is listed as the sponsor.

The two arm descriptions identify patients treated with VNS Therapy and patients not receiving VNS Therapy. The arm labels are “500 VNS Patients” and “300 Non-VNS Patients.” The registry does not state that patients were randomly allocated, and its allocation field is not stated. The comparison should therefore be understood as observational, not as a like-for-like randomized control comparison.

For the primary outcome, the record names an intent-to-treat population comprising a VNS Therapy Population of n=489, with D-23=330 and D-21=159, plus a TAU population of n=276. The record also cautions that the number of patients contributing to an outcome may be lower than the intent-to-treat population when assessment data are missing. The reasons for any differences between arm labels, enrollment, and the stated outcome populations are not reported in the registry record.

Thirteen outcome measures and 26 analyses are listed. The primary outcome is MADRS response, assessed from 3-month through 60-month post-baseline visits. Secondary outcomes include remission, time until recurrence among patients who achieved remission, and analyses concerning predictors of suicidality based on MADRS item 10 and other listed characteristics. The registry record does not report a detailed statistical analysis plan, the timing of any interim analyses, or how treatment groups were formed in practice.

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: A total of 878 subjects were screened in the TRD Registry Study. Thirty-seven were not eligible and 46 were eligible but not treated for various reasons. This left a total of 795 patients in the Safety Population (SP).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-To-Treat (ITT) Population: VNS Therapy Population (n=489 (D-23=330 + D-21=159)) + (n= 276) TAU population

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.

Population: A total of 878 subjects were screened. Thirty-seven were not eligible and 46 were eligible but not treated for various reasons. This left a total of 795 patients in the Safety Population (SP). The total number of patients in each group may be lower than SP if there was missing D-21 assessment data, as D-21 baseline data was not identical to D-23.

6. Primary Endpoint

The registry lists the primary outcome as Montgomery Asberg Depression Rating Scale (MADRS)% Responders (>/= 50% Improvement From Baseline), measured over 3-Month Through 60-Month (Post Baseline).

The primary endpoint is the percentage of participants who achieved at least a 50% reduction from baseline in total Montgomery Asberg Depression Rating Scale (MADRS) score at a post-baseline visit. Assessments are specified from 3 months through 60 months after baseline. At each visit, a participant meeting the reduction threshold is classified as a responder; a participant not meeting it is classified as a non-responder.

The MADRS is described in the registry as a ten-item questionnaire used to measure the severity of depressive episodes in people with mood disorders. Each item ranges from 0 to 6, and the total score ranges from 0 to 60. Higher total scores indicate more severe depression, while lower scores indicate less symptom severity. The response definition uses change relative to each participant’s baseline score, rather than a single absolute follow-up score.

A percentage makes the outcome interpretable as the share of assessed participants meeting the registry’s response criterion. It also converts a continuous score change into a binary classification, which can simplify comparison across visits and groups. That simplification discards information about the size of score changes and about participants close to the threshold. The registry record does not report the response percentages, their uncertainty intervals, or the comparative estimates, so the endpoint’s observed pattern and precision cannot be described from the record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry’s primary outcome is a binary response indicator evaluated at post-baseline visits. The statistical meaning of this endpoint differs from an analysis of average MADRS score: it counts participants who cross a specified improvement threshold, rather than preserving each participant’s full score change. The record defines response as at least a 50% reduction from baseline in total MADRS score. It identifies the outcome unit as percentage of participants and states that the response rate is summarized at each visit.

What does a response percentage represent?

At a given visit, a response percentage is the number of assessed participants meeting the response definition relative to the participants included in that assessment. This gives a direct description of how common the defined response was among those assessed. The registry record does not report the actual percentages or the visit-specific denominators, so the observed response rates cannot be reconstructed or interpreted.

Turning a score into a responder/non-responder classification has a trade-off. It can make a threshold-based outcome easier to communicate, but participants with similar score changes can fall on opposite sides of the cutoff. It also treats all responders alike even if their score reductions differ substantially. These are general properties of dichotomizing a continuous measure, not findings about participants in this registry.

Why does the observational design affect comparison?

In a randomized study, random allocation is intended to balance measured and unmeasured baseline characteristics across treatment groups on average. This registry is described as observational, and allocation is not stated. The VNS and non-VNS groups therefore cannot be assumed to be exchangeable. Differences in patient characteristics, prior treatment, clinical course, or other factors could be related both to whether a patient received VNS and to later MADRS response.

General statistical explanation: confounding occurs when a factor is associated with both the exposure or treatment and the outcome, distorting an unadjusted comparison. Adjustment methods can address measured factors when their data and appropriate modeling assumptions are available, but cannot automatically resolve unmeasured confounding. The registry record does not report the adjustment strategy, covariates, model forms, or estimates. It is therefore not possible to determine whether reported group comparisons, if any, account for such differences.

How should uncertainty be assessed?

A point estimate such as a response percentage describes an observed proportion, but does not by itself show how precisely that proportion is estimated. General statistical explanation: a confidence interval gives a range of values compatible with the data and the calculation method under stated assumptions. A narrower interval indicates greater precision than a wider interval, while neither interval width nor a p-value resolves bias from observational group differences. No confidence intervals, standard errors, or p-values for the primary outcome are reported in the registry record.

The statistical plan is not reported in the registry record. In particular, it does not state whether comparisons were adjusted, which tests or models were used, how repeated assessments were handled, or whether interim analyses were conducted. These omissions limit assessment of the assumptions behind any analysis, beyond the endpoint definition itself.

What does over-dispersion mean in general?

General statistical explanation: for count outcomes, over-dispersion means that observed variability exceeds what a simple count model assumes. A model that fails to allow for that extra variability can produce uncertainty estimates that are too small. The primary endpoint is a binary percentage, not a count-rate outcome as described in the registry record, and no over-dispersion assessment or model details are reported. This general concept should not be mistaken for a finding about the registry.

How should planning calculations be read?

The record describes any sample-size calculations as planning illustrations based on an observed effect, not as the registry’s own design justification. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect to detect expressed as a fraction of the observed effect. For time-to-event calculations, the total number of events is the planning quantity; for other calculations, the quantity is participants per arm. Such illustrations address hypothetical future planning under their assumptions. They are not observed power and do not show that the registry was designed to detect a particular effect. The registry’s sample-size justification is not reported in the registry record, and the outcome estimates needed to assess the illustrations are not included in the record.

Because the record lists repeated post-baseline assessments, interpretation would also depend on how those visits were analyzed. General statistical explanation: repeated measurements from the same participant are correlated, so treating each visit as an independent observation can misstate uncertainty. The record does not report whether analyses modeled this correlation, analyzed each visit separately, or used another approach. The number of analyses listed alone does not identify the methods used.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 outcome measures and 26 analyses, including a primary response endpoint and secondary outcomes concerning remission, recurrence, and suicidality predictors. Examining multiple outcomes, visits, or analyses creates multiplicity: the more statistical comparisons made, the greater the chance that some results may appear notable by chance. General statistical explanation: prespecified testing hierarchies or adjustments can help control the interpretation of multiple tests, but the approach depends on the analysis plan. A multiplicity strategy is not reported in the registry record. The number of listed outcomes and analyses does not reveal whether adjustment was used.

Missing assessment data are explicitly relevant to the primary outcome. The record states that the number of patients in each group may be lower than the intent-to-treat population when assessment data are missing. It does not report visit-specific numbers assessed, the extent or pattern of missingness, reasons for missing assessments, or methods used to address missing outcomes. Without these details, it is not possible to determine how much the analyzed populations differed from the stated populations or whether missingness could affect response estimates.

The outcome population is identified as a VNS Therapy Population of n=489, including D-23=330 and D-21=159, plus a TAU population of n=276. The record does not explain how those populations relate to the arm labels or the enrollment count, so those differences should not be assigned an explanation. An analysis population name also does not establish how missing data were handled in each individual visit-level analysis.

Interim analyses and stopping rules are not reported in the registry record. Nor is a detailed statistical analysis plan available in the record. These gaps limit the ability to evaluate whether the listed outcomes and analyses were prespecified, how multiplicity was handled, and whether analysis decisions were made before results were examined.

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

The registry record describes a prospective observational effort to follow patients with treatment-resistant depression receiving VNS and patients not receiving VNS. Its primary endpoint is the proportion meeting a defined MADRS response threshold at post-baseline visits. These details establish what the registry intended to measure and which groups it intended to observe.

The record does not report the primary response percentages, group contrasts, or uncertainty estimates. It therefore does not show how often participants met the response definition, whether the outcome differed between groups, or how precise any estimated difference was. It also does not report the secondary outcome results. No conclusion about the direction or size of clinical outcomes can be drawn from the endpoint description alone.

Even with complete outcome estimates, this observational comparison would require caution. Treatment allocation is not stated, and the groups are not described as a randomized, like-for-like comparison. A difference in outcomes could reflect differences between patients or other aspects of care, as well as any association with VNS use. The registry record does not report enough about adjustment or other analytical methods to distinguish these possibilities.

General statistical explanation: a statistically detectable difference, if one were reported, would not by itself establish that a treatment caused the difference; design, bias, uncertainty, and the size and relevance of the estimate all matter. Likewise, absence of a reported statistical result in the registry record is not evidence that groups had identical outcomes. The record simply does not include the results needed to assess a difference.

The safety entry is also insufficient for a population-level safety interpretation: its listed at-risk counts are zero, its event counts are zero, and the death value is null. Those registry entries do not establish the frequency of events among enrolled participants. Overall, the record supports a description of the study question, endpoint definition, and stated populations, but not an estimate of treatment-associated outcomes.

12. Why This Trial Matters Statistically

This registry is statistically relevant because it addresses a clinical question through prospective observation rather than random assignment. Registries can follow patients in clinical care over time and examine outcomes across extended follow-up. In this case, the primary endpoint was defined at repeated post-baseline visits from 3 months through 60 months, and the record also lists outcomes related to remission, recurrence, and suicidality predictors.

The design makes careful interpretation especially important. Patients receiving VNS and patients not receiving it may not have comparable characteristics, and the registry record does not state how treatment allocation occurred. That means the distinction between describing an observed association and estimating a causal effect is central to any analysis of the groups. Adjustment can help with measured differences when suitable information and methods are available, but the record does not report its analysis strategy.

The endpoint also illustrates a common statistical choice: translating a continuous symptom score into a binary threshold. This makes the responder definition easy to state, while losing detail about the magnitude of each participant’s score change. Repeated assessments and possible missing data add further considerations, because the number contributing to a visit may be lower than the stated intent-to-treat population.

The record’s value for statistical interpretation is limited by the absence of outcome estimates, uncertainty measures, and analysis details. Its enrollment and outcome definitions describe the scope and question of the registry, but cannot replace results. The case therefore highlights why study design, denominators, missing-data handling, multiplicity, and confidence intervals are needed alongside an endpoint definition when assessing a clinical comparison.

13. 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.

A key limitation is that the registry record does not report the primary outcome results. There are no response percentages, group comparisons, confidence intervals, standard errors, or p-values to describe. Results for the listed secondary outcomes are also not reported in the registry record. Accordingly, the magnitude, direction, and precision of observed outcomes cannot be assessed.

The observational design limits causal interpretation. Allocation is not stated, and the registry explicitly distinguishes VNS and non-VNS patients rather than describing a randomized comparison. The record does not report baseline characteristics, group comparability, adjustment variables, or the statistical models used. Potential confounding and the adequacy of any adjustment therefore cannot be evaluated.

For the primary endpoint, the record warns that missing assessment data can make group counts lower than the intent-to-treat population. It does not report visit-specific denominators, missingness patterns, reasons for missing data, or sensitivity analyses. This prevents assessment of how missing assessments may affect the response proportions.

The registry lists 13 outcome measures and 26 analyses, but does not report a multiplicity approach or a detailed statistical analysis plan. Interim analyses are not reported in the registry record. The relationship between the arm labels, enrollment, and stated outcome populations is also not explained, so no reconciliation can be made from the record.

Finally, the safety entry gives zero at-risk and event counts, with deaths recorded as null. It does not provide a usable account of safety outcomes among the registry population. These limitations are specific to the information and design described in this record; they do not establish what the unreported results would have shown.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: Treatment-Resistant Depression Registry — NCT00320372.
• Registry record last updated 2015-12-23; status last verified 2015-12.
• Results first posted 2015-12-23.
• 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.

17. Record Summary

NCT00320372 is a completed, prospective, observational, multicenter registry of patients with treatment-resistant depression who were in a major depressive episode. It lists an enrollment of 795 and compares patients treated with VNS Therapy with patients not receiving VNS. The primary endpoint is the percentage meeting the definition of at least a 50% reduction from baseline in total MADRS score at visits from 3 months through 60 months after baseline.

The record identifies a VNS Therapy Population of n=489, including D-23=330 and D-21=159, and a TAU population of n=276. It lists 13 outcome measures and 26 analyses, and notes that assessment-related missing data may reduce group counts. The outcome estimates, detailed statistical methods, multiplicity strategy, and missing-data analyses are not reported in the registry record. Thus, the record describes the study question and endpoint, but does not permit interpretation of the observed response or a comparative effect.