This page separates information reported in the ClinicalTrials.gov record from statistical interpretation and methodological explanation. The registry record reports a primary binary endpoint and posted outcome measures, but the registry-reported statistical-analyses section contains no formal comparative analysis results.
1. Trial at a Glance
ZeNix was a randomized, parallel-group, triple-masked, phase 3 treatment trial enrolling 181 participants with drug-resistant tuberculosis. The registered primary endpoint was a binary outcome assessing bacteriologic failure, relapse, or clinical failure through follow-up until 26 weeks after the end of treatment.
| Feature | ZeNix |
|---|---|
| Trial name | ZeNix |
| NCT identifier | NCT03086486 |
| Phase | Phase 3 |
| Therapeutic area | Infectious Disease |
| Population | Participants with drug-resistant tuberculosis, including pulmonary tuberculosis, multidrug-resistant tuberculosis, pre-XDR-TB, and XDR-TB |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 181 |
| Number of arms | 4 |
| Primary endpoint type | Binary |
| Registered primary endpoints | 1 |
| Outcome measures posted | 4 |
| Trial status | COMPLETED |
| Lead sponsor | Global Alliance for TB Drug Development |
2. Clinical Question
The ZeNix trial addresses a treatment question in participants with drug-resistant tuberculosis: how do different doses and durations of linezolid, given with pretomanid and bedaquiline, relate to the occurrence of an unfavourable clinical or bacteriologic outcome during the registered follow-up period?
Population
Participants with drug-resistant tuberculosis, including pulmonary tuberculosis, multidrug-resistant tuberculosis, pre-XDR-TB, and XDR-TB.
Intervention framework
Linezolid plus pretomanid and bedaquiline, with different linezolid doses and treatment durations represented across the four study arms.
Comparator framework
The four-arm randomized design permits comparison across the registered linezolid dose-and-duration regimens.
Primary question
What is the incidence of an unfavourable status defined by bacteriologic failure, relapse, or clinical failure through 26 weeks after the end of treatment?
3. Trial Design
The ClinicalTrials.gov record identifies ZeNix as a randomized, parallel-group, triple-masked phase 3 treatment trial with 181 participants and four arms. The intervention list contains pretomanid, linezolid, bedaquiline, and placebo linezolid.
1200mg L × 26 Weeks + Pa + B
- Linezolid: 1200 mg
- Linezolid duration: 26 weeks
- Pretomanid and bedaquiline included
- Serious AEs: 3/45
1200 mg L × 9 Weeks + Pa + B
- Linezolid: 1200 mg
- Linezolid duration: 9 weeks
- Pretomanid and bedaquiline included
- Serious AEs: 4/46
600 mg L × 26 Weeks + Pa + B
- Linezolid: 600 mg
- Linezolid duration: 26 weeks
- Pretomanid and bedaquiline included
- Serious AEs: 1/45
600 mg L × 9 Weeks + Pa + B
- Linezolid: 600 mg
- Linezolid duration: 9 weeks
- Pretomanid and bedaquiline included
- Serious AEs: 3/45
4. Randomization and Masking
ZeNix used randomized allocation in a parallel-group design. The registry identifies the masking level as triple. Randomization and masking address different sources of bias: randomization creates the basis for comparing treatment groups under the trial design, while masking can reduce the influence of knowledge of treatment assignment on participant behavior, clinical management, assessment, or other trial processes.
Participants were assigned within a randomized four-arm parallel design.
The registry classifies the study as triple-masked.
The treatment groups operate as parallel study arms rather than sequential treatment periods.
The registered primary purpose is treatment.
Why randomization matters statistically
Randomization is the foundation for a causal comparison between study arms. In expectation, randomization balances measured and unmeasured baseline factors across treatment assignments. With 181 participants distributed among four arms, however, the realized groups need not be identical on every characteristic. That is why baseline balance should be described rather than assumed.
Why masking matters statistically
For an endpoint involving clinical failure and microbiologic classification, masking can be important because knowledge of treatment assignment may otherwise influence assessment or behavior. Triple masking is therefore a design feature relevant to the credibility of the endpoint comparison, although the ClinicalTrials.gov record does not specify which three parties were masked.
5. Primary Endpoint
| Endpoint | Registry definition and time frame | Type |
|---|---|---|
| Incidence of Bacteriologic Failure or Relapse or Clinical Failure (Unfavorables) | Through follow up until 26 weeks after the end of treatment. The registry definition specifies unfavourable status based on participants not classified as having achieved or maintained culture negative status when last seen; participants previously classified as having culture negative status who, following the end of treatment, have two positive cultures without an intervening negative culture; and participants who had a positive culture not followed by at least two negative cultures when last seen 4. Participants dying from any cause during treatment, except from violent or accidental cause, not including suicide 5. Participants definitely or possibly dying from TB related cause during the follow-up phase 6. Participants requiring an extension of their treatment beyond that permitted by the protocol a restart or a change of treatment for any reason except reinfection or pregnancy 7. Participants who have had surgery and the resected tissue is cultured and is positive for MTB 8. Participants lost to follow up or withdrawn from the study before the end of treatment | Binary |
What a binary endpoint means here
A binary endpoint classifies each participant according to whether the prespecified unfavourable outcome occurred during the defined assessment period. Unlike a continuous endpoint, the primary analysis does not fundamentally ask how large a numerical change occurred within each participant. Instead, it asks how the proportion with the prespecified outcome differs across randomized treatment groups.
A treatment comparison can then be expressed through an absolute risk difference, a relative risk, an odds ratio, or a model-based estimate. The ClinicalTrials.gov record does not identify which formal comparative estimator or hypothesis test was used for the posted outcome.
6. Results Availability
ClinicalTrials.gov reports that results have been posted and that 4 outcome measures are posted. However, the ClinicalTrials.gov record contains no formal statistical analyses for the primary endpoint.
Accordingly, the page does not report a treatment-effect estimate, confidence interval, or p-value for the primary endpoint. Those values are not present in the ClinicalTrials.gov record.
7. Planned Analysis
The registered primary endpoint is binary, so the natural statistical framework is a comparison of the incidence of an unfavourable status across the randomized treatment groups. A complete formal analysis would begin by specifying the analysis population, defining the denominator for each arm, and then comparing the observed event proportions using a prespecified method.
Risk difference
The absolute difference between event proportions directly describes the change in the probability of an unfavourable outcome between two groups.
Relative risk
The ratio of event probabilities describes the event rate in one group relative to another. It is particularly intuitive when the outcome is not rare.
Odds ratio
The ratio of event odds can be estimated using a 2×2 contingency-table framework or logistic regression. It should not be interpreted as a risk ratio when the event is common.
Exact or asymptotic testing
Depending on the prespecified analysis plan and cell counts, a categorical comparison could use a chi-square-type test or an exact method. The ClinicalTrials.gov record does not specify which was used.
What a complete results record would ideally contain
For this endpoint, a statistically interpretable results presentation would identify the number of participants analyzed in each arm, the number and proportion classified as unfavourable, the treatment contrast used, its confidence interval, and the hypothesis-testing result if one was prespecified. Without those quantities, the posted outcome status alone does not establish the magnitude or statistical uncertainty of a treatment comparison.
8. Serious Adverse Events
The ClinicalTrials.gov record provides serious adverse-event counts by arm. These are presented as affected participants divided by participants at risk.
| Arm | Serious adverse events | Affected / at risk |
|---|---|---|
| 1200mg L × 26 Weeks + Pa + B | 3 | 3/45 |
| 1200 mg L × 9 Weeks + Pa + B | 4 | 4/46 |
| 600 mg L × 26 Weeks + Pa + B | 1 | 1/45 |
| 600 mg L × 9 Weeks + Pa + B | 3 | 3/45 |
| Total | 11 | 11/181 |
The total reported serious-adverse-event count is 11 affected participants among 181 participants at risk. The arm-specific denominators in the registry-reported safety summary are 45, 46, 45, and 45, which sum to 181.
The affected/at-risk values describe the observed frequency of serious adverse events in each arm. They do not by themselves establish that differences between arms are statistically significant, nor do they identify whether an observed difference is attributable to linezolid dose, linezolid duration, or another feature of the regimen. A formal comparative safety analysis would require the prespecified analysis population, event definitions, exposure information, and statistical method.
9. Statistical Methodology
Binary endpoint analysis
The primary endpoint is explicitly classified as Binary. This means each participant is ultimately classified into an endpoint category for the defined follow-up period. The fundamental quantity is therefore an event proportion rather than a mean change or a time-to-event hazard.
where x is the number of participants with an unfavourable outcome and n is the number of participants included in the relevant analysis.
Risk difference
For two treatment groups, the risk difference is
A negative value would indicate a lower observed event probability in group 1 relative to group 0, while a positive value would indicate a higher observed event probability. In a multi-arm study such as ZeNix, the choice of reference group and the set of planned pairwise comparisons should be prespecified.
Relative risk
The relative risk is
An RR of 1 corresponds to equal observed event probabilities. Values below 1 correspond to a lower event probability in the numerator group, while values above 1 correspond to a higher event probability.
Odds ratio
For a binary endpoint, the odds ratio compares the odds of the outcome between groups:
The odds ratio is mathematically different from the relative risk. When the outcome is relatively common, the numerical values can differ substantially, so an odds ratio should not automatically be described as a risk ratio.
Confidence intervals
A confidence interval quantifies statistical uncertainty around an estimated treatment contrast. For a risk difference, the interval is expressed on the absolute-risk scale; for a relative risk or odds ratio, it is expressed on a multiplicative scale. The registry-reported ZeNix data do not report a confidence interval for the primary endpoint.
Hypothesis testing
A p-value, if reported from a prespecified hypothesis test, would quantify how compatible the observed data are with the null hypothesis under that test's assumptions. It would not measure the size of the treatment effect. Effect estimates and confidence intervals are therefore needed to understand the magnitude and precision of the comparison.
10. Statistical Methods Explained
Why is the primary endpoint binary?
The registered endpoint classifies participants according to whether they meet criteria for an unfavourable status involving bacteriologic failure, relapse, or clinical failure. A binary endpoint is useful when the clinically relevant question is whether an outcome occurred by a defined follow-up point rather than exactly when it occurred.
Why does the 26-week time frame matter?
The endpoint is defined through follow-up until 26 weeks after the end of treatment. This fixes the observation window for the primary outcome. Without a defined time frame, participants with different amounts of follow-up could be classified under different opportunities for observing an unfavourable outcome.
Why does randomization matter for a binary endpoint?
Randomization makes treatment assignment the organizing variable for the primary comparison. If the randomization process is properly implemented, systematic differences in participant characteristics are less likely to explain observed between-arm differences than they would be in an uncontrolled comparison.
Why does triple masking matter?
Masking can reduce the influence of treatment knowledge on participant behavior, clinical management, and outcome assessment. This is particularly relevant when the endpoint includes a clinical-failure component, although the ClinicalTrials.gov record does not specify the identities of the three masked parties.
Why can a four-arm trial require special attention to multiplicity?
With four treatment arms, there can be several possible pairwise comparisons. If many hypotheses are tested without an appropriate multiplicity strategy, the probability of at least one false-positive result can increase. The registry-reported ZeNix data do not report a multiplicity adjustment or endpoint hierarchy, so no such procedure is attributed to the trial here.
Why should safety and efficacy be analyzed separately?
The primary endpoint measures an efficacy outcome, whereas serious adverse events describe safety. They answer different questions and generally require different denominators, analysis populations, and interpretations. Combining them into a single numerical judgment would obscure those distinctions.
11. Analysis Population and Missing Data Considerations
The ClinicalTrials.gov record identifies enrollment and the registered primary endpoint but do not specify an analysis population such as intention-to-treat, per-protocol, modified intention-to-treat, or safety population for the primary endpoint.
| Question | Information in the ClinicalTrials.gov record |
|---|---|
| Enrollment | 181 |
| Primary endpoint | Binary unfavourable status |
| Formal primary analysis population | Not specified in the ClinicalTrials.gov record |
| Missing-data/imputation method | Not specified in the ClinicalTrials.gov record |
| Formal statistical comparison | Not present in the posted results |
| Multiplicity strategy | Not specified in the ClinicalTrials.gov record |
These omissions matter because binary outcomes can be sensitive to how participants with incomplete follow-up are classified. A complete statistical analysis should specify how participants who do not have enough information to determine endpoint status are handled and whether the analysis is based on all randomized participants, an evaluable population, or another prespecified population.
12. Multiplicity in a Four-Arm Design
A four-arm randomized trial naturally creates a multiple-comparison question. If the scientific objective involves comparing several regimens, the statistical plan needs to distinguish the primary comparison or comparisons from secondary comparisons and determine how the type I error rate is controlled.
Pairwise comparisons
Four arms create multiple possible pairwise treatment comparisons. The number of potential comparisons grows faster than the number of arms themselves.
Primary endpoint
ZeNix registered one primary endpoint, but the ClinicalTrials.gov record does not specify which arm comparisons constituted the formal confirmatory hypotheses.
Multiplicity adjustment
The ClinicalTrials.gov record does not report a Bonferroni, Holm, Hochberg, gatekeeping, hierarchical, or other multiplicity procedure.
Interpretation
Individual nominal p-values cannot be reconstructed or interpreted as confirmatory evidence without knowing the prespecified comparison structure and error-control strategy.
13. Non-Inferiority, Crossover, Factorial Design, and Bayesian Methods
The registry-reported ZeNix data do not identify a non-inferiority margin, crossover component, factorial design, interim-analysis procedure, or Bayesian statistical method.
| Design topic | Supported by the ClinicalTrials.gov record? | Interpretation |
|---|---|---|
| Non-inferiority margin | No | No non-inferiority margin is reported in the ClinicalTrials.gov record. |
| Crossover | No | The registered design is parallel; no crossover procedure is reported. |
| Factorial design | No | The design is identified as parallel rather than factorial. |
| Interim analysis | No | No interim-analysis procedure is reported. |
| Bayesian methods | No | No Bayesian analysis is identified in the ClinicalTrials.gov record. |
| Multiplicity procedure | No | No formal multiplicity strategy is identified in the ClinicalTrials.gov record. |
These distinctions are important because the correct interpretation of a clinical-trial result depends not only on the endpoint but also on the inferential framework established before the data are analyzed.
14. What the Posted Safety Data Do — and Do Not — Mean
The reported serious-adverse-event values range from 1/45 to 4/46 across the four arms, with 11/181 affected participants overall. These are descriptive frequencies in the ClinicalTrials.gov record.
The serious-adverse-event counts do not establish that one regimen is statistically safer than another. No confidence intervals, p-values, rate ratios, or other formal comparative safety analyses are included in the posted results.
The four arms vary by linezolid dose and duration, but the ClinicalTrials.gov record does not provide a factorial statistical model that separates dose effects from duration effects. A descriptive difference between arms therefore cannot be decomposed into independent dose and duration effects from the ClinicalTrials.gov record alone.
15. Trial Timeline
Trial start
The registry profile gives 2017-11-21 as the study start date.
Randomized phase 3 study period
The trial used randomized allocation, a parallel design, triple masking, four arms, and a treatment purpose.
Primary completion
The registry profile gives 2021-02-15 as the primary completion date.
Completed with results posted
The ClinicalTrials.gov record classifies the trial as COMPLETED and indicate that results have been posted, with four outcome measures reported.
16. Trial Status and Data Structure
The distinction between results posted and formal statistical analyses reported is especially important for this record. The registry profile says that results are posted, but the registry-reported statistical-analysis array is empty.
| Data element | Supplied value | Statistical implication |
|---|---|---|
| Results posted | Yes | Outcome information is available in the registry record. |
| Outcome measures posted | 4 | Four outcome measures are identified as posted. |
| Statistical analyses | None reported | No formal effect estimate, CI, or p-value can be attributed to the registry analysis from the ClinicalTrials.gov record. |
| Analyses omitted for length | 0 | The absence of the posted statistical analyses is not due to the ClinicalTrials.gov recordset being shortened. |
This distinction prevents a common reporting error: treating the presence of posted outcome data as evidence that a particular statistical test, effect estimate, or confidence interval has been reported. The statistical method used for a binary outcome can be explained, but it should not be retroactively assigned to the trial without supporting documentation.
17. Limitations
- No formal statistical analyses reported: no formal statistical analyses were posted, so no primary treatment-effect estimate, confidence interval, or p-value is reported here.
- Incomplete endpoint definition in registry-reported excerpt: the final portion of the registry definition ends after "when last", so the missing wording is not reconstructed.
- Analysis population not specified: the ClinicalTrials.gov record does not identify whether the primary endpoint used an intention-to-treat, per-protocol, modified intention-to-treat, or another population.
- Missing-data handling not specified: the ClinicalTrials.gov record does not identify how incomplete endpoint information was handled.
- Multiplicity not specified: a four-arm design can generate multiple comparisons, but the ClinicalTrials.gov record does not identify a formal error-control strategy.
- No non-inferiority margin reported: the ClinicalTrials.gov record does not identify a non-inferiority framework or margin.
- No interim-analysis method reported: the ClinicalTrials.gov record does not identify an interim monitoring plan.
- Safety interpretation is descriptive: serious-adverse-event counts are available, but formal between-arm safety analyses are not reported.
- Dose and duration effects cannot be separated from the registry-reported summaries: the four regimens vary by both linezolid dose and duration, without a reported factorial analysis in the ClinicalTrials.gov record.
- No Bayesian analysis reported: the ClinicalTrials.gov record does not identify Bayesian methods.
18. Why This Trial Matters Statistically
ZeNix is a useful teaching case because it combines a randomized four-arm design with a binary clinical and bacteriologic endpoint. It illustrates how trial design, endpoint definition, analysis population, and multiplicity decisions all influence what can legitimately be concluded from a set of outcome counts.
| Statistical concept | How it appears in ZeNix |
|---|---|
| Randomization | The trial uses randomized allocation across four parallel arms. |
| Masking | The registry classifies the study as triple-masked. |
| Binary endpoint | The primary endpoint is classified as binary. |
| Composite clinical outcome | The primary endpoint incorporates bacteriologic failure, relapse, or clinical failure into an unfavourable-status classification. |
| Fixed follow-up window | The primary endpoint is assessed through 26 weeks after the end of treatment. |
| Multi-arm comparison | Four treatment arms create a potentially complex comparison structure. |
| Multiplicity | Multiple possible comparisons make prespecification and type I error control important. |
| Safety analysis | Serious adverse events are reported separately by arm. |
| Missing data | The ClinicalTrials.gov record does not specify the method used to handle incomplete endpoint information. |
| Statistical transparency | Results are posted, but no formal statistical analyses are included in the ClinicalTrials.gov record. |
Why the endpoint definition is especially important
For a binary endpoint, the classification rule determines what enters the numerator. In ZeNix, the registry defines an unfavourable status through several bacteriologic and clinical pathways. That means statistical interpretation cannot be separated from the endpoint algorithm itself: changing the classification rule can change the event count even when the underlying patient data are unchanged.
Why the four-arm structure matters
A four-arm trial can provide more information than a simple two-group comparison, but it also creates a larger inferential space. The statistical analysis must identify which comparisons are confirmatory, which are supportive, and which are exploratory. Without that hierarchy, an isolated significant result among several comparisons can be difficult to interpret.
Why the absence of a reported effect estimate matters
Knowing that an outcome was posted does not tell the reader whether the observed difference between treatment groups was large or small, precise or imprecise, or statistically compatible with the null hypothesis. Those questions require the underlying event counts and a specified comparison method.
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Statistical Calculators
21. Sources
- ClinicalTrials.gov: NCT03086486 — ZeNix. Trial profile and ClinicalTrials.gov record.
- Linked publication: PubMed record PMID 42429813.
- Linked publication: PubMed record PMID 36053506.
Continue through Clinical Biostats
Explore the statistical concepts behind randomized trials, binary endpoints, effect measures, and multi-arm study design.
22. Record Summary
ZeNix is a randomized, parallel-group, triple-masked phase 3 trial with 181 participants and four treatment arms evaluating different linezolid doses and durations in combination with pretomanid and bedaquiline. Its registered primary endpoint is a binary measure of bacteriologic failure, relapse, or clinical failure through 26 weeks after the end of treatment. The ClinicalTrials.gov record reports posted results and four outcome measures, together with serious-adverse-event counts by arm, but they do not contain formal statistical-analysis entries for the primary endpoint.
For that reason, the central statistical lesson is methodological rather than a reconstructed treatment-effect claim. A valid analysis of the primary endpoint would require the endpoint counts by arm, the prespecified analysis population, the comparison structure, the effect measure, confidence interval method, hypothesis-testing procedure, and any multiplicity or missing-data rules. The ClinicalTrials.gov record establishes the design and endpoint framework, but they do not provide those formal comparative results.