This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Using Biomarkers to Predict TB Treatment Duration (NCT02821832), a randomized phase 2 study. The primary outcome is Comparison of the Rate of Treatment Success at 18 Months (After Treatment Initiation) Between Arms B and C.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT02821832 |
| Phase | Phase 2 |
| Condition | Pulmonary Tuberculosis |
| Allocation | Randomized |
| Design model | Sequential |
| Masking | None (open label) |
| Enrollment | 946 |
| Arms in the primary outcome | n/a |
| Lead sponsor | National Institute of Allergy and Infectious Diseases (NIAID) |
| Results posted | Yes |
2. Clinical Question
Population
People with pulmonary tuberculosis treatable with standard tuberculosis drugs were eligible according to the registry summary. The stated age range was 18–75 years. Participants underwent screening that included clinical assessment, laboratory testing, sputum collection, chest x-ray, and PET/CT scans during the study. The trial also distinguished participants according to whether their predicted risk of relapse was expected to be high or low. The eligibility criteria and biomarker rules in full are not reported in the registry record.
Intervention
Arm C received shortened tuberculosis treatment. The registry describes this arm as having an expected low risk of relapse and receiving shortened treatment. The objective was to assess a four-month treatment duration in people thought likely to be cured at that point. The specific biomarkers or decision thresholds used to identify those participants are not reported in the registry record.
Comparator
Arm B received standard-duration treatment and was also described as having an expected low risk of relapse. The registry summary says participants thought likely to be cured after four months were randomly assigned to four or six months of treatment. Arm A was described as having an expected high risk of relapse and receiving standard-of-care treatment; it was not part of the primary comparison.
Primary question
Among participants in Arms B and C, was the treatment-success rate at 18 months after treatment initiation consistent with the shortened-duration arm being noninferior to the standard-duration arm? The registry defines the noninferiority criterion through the lower bound of a one-sided 95% confidence interval for the difference in success rates: that bound must be greater than -7%. The primary outcome analysis itself is not reported in the registry record.
3. Trial Design
This was a phase 2, randomized, open-label trial classified in the registry as having a sequential model. Its stated enrollment was 946, and its status is completed. The recorded start date was 2017-06-21, and the primary completion date was 2021-10-09. The sponsor was the National Institute of Allergy and Infectious Diseases (NIAID).
The trial concerned pulmonary tuberculosis and the possibility of using PET/CT lung imaging together with an estimate of tuberculosis burden to identify people who might be treated for a shorter duration. The registry summary describes participants beginning tuberculosis drugs and returning for visits across an 18-month study. Visits included clinical assessments, sputum samples, and blood tests; PET/CT scans and chest x-rays were also planned. The listed drug regimen includes isoniazid, rifampicin, pyrazinamide, and ethambutol.
Participants thought likely to be cured after four months were randomly assigned to standard-duration treatment or shortened treatment. The registry describes Arm B as low expected relapse risk with standard treatment, Arm C as low expected relapse risk with shortened treatment, and Arm A as high expected relapse risk with standard-of-care treatment. The primary comparison was between Arms B and C. The precise randomization procedure, allocation ratio, meaning of the sequential-model classification, and operational biomarker algorithm are not reported in the registry record. Masking is recorded as none.
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: Enrolled participants may have been excluded from participation for the following reasons: identification of baseline drug-resistant TB, withdrawn consent, sputum culture negative on Lowenstein-Jensen medium at baseline, poorly adherent to treatment, developed extrapulmonary TB, or had a significant incidental finding on baseline PET/CT requiring study withdrawal.
Overall Study
| Stage | Arm A | Arm B | Arm C |
|---|---|---|---|
| STARTED | 251 | 154 | 141 |
| COMPLETED | 186 | 106 | 109 |
| NOT COMPLETED | 65 | 48 | 32 |
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.
| Characteristic | Unit | Arm A | Arm B | Arm C | Total |
|---|---|---|---|---|---|
| Participants analysed | 251 | 154 | 141 | 546 | |
| Age, Continuous | years | 37.5 (18.2 to 75.9) | 35.5 (18.0 to 68.5) | 33.4 (18.4 to 73.8) | 35.8 (18.0 to 75.9) |
| Sex: Female, Male: Female | Participants | 58 | 46 | 46 | 150 |
| Sex: Female, Male: Male | Participants | 193 | 108 | 95 | 396 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Asian | Participants | 72 | 55 | 53 | 180 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Black or African American | Participants | 179 | 99 | 88 | 366 |
| Race (NIH/OMB): White | Participants | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 | 0 |
| Region of Enrollment: South Africa | participants | 179 | 99 | 88 | 366 |
| Region of Enrollment: China | participants | 72 | 55 | 53 | 180 |
| Weight | kilograms | 55.3 (SD 9.5) | 55.0 (SD 9.7) | 57.1 (SD 9.9) | 55.7 (SD 9.7) |
| BMI | kg/m^2 | 19.5 (SD 2.9) | 19.7 (SD 2.8) | 20.2 (SD 3.2) | 19.7 (SD 3.0) |
6. Primary Endpoint
The registry lists the primary outcome as Comparison of the Rate of Treatment Success at 18 Months (After Treatment Initiation) Between Arms B and C, measured over 18 months.
| Element | Registry entry |
|---|---|
| Result type | Count Of Participants |
| Measure of spread | not applicable |
| Unit | Participants |
| Endpoint classification | Binary Endpoint |
The primary endpoint is the comparison of treatment-success rates between Arms B and C at 18 months after treatment initiation. It is a binary outcome summarized as a count of participants, with treatment success assessed in the context of a trial comparing shortened treatment with standard-duration treatment. The registry description does not give the operational criteria for success, such as the specific clinical or microbiological events used to classify a participant. Those criteria are not reported in the registry record.
The endpoint is used for a noninferiority question. Rather than asking whether the two treatment durations produce exactly the same success rate, the stated approach asks whether the data exclude a disadvantage larger than the prespecified margin for the shortened-duration arm. The registry describes the contrast through the lower bound of a one-sided 95% confidence interval for the difference in success rates. Its stated criterion is that this lower bound be greater than -7%.
This framework is relevant because the trial’s objective was to assess whether treatment could be shortened without crossing the stated margin. The endpoint description identifies the intended time point and decision rule, but does not report the observed success counts, estimated difference, confidence interval, or analysis population. Consequently, the registry record does not establish whether the criterion was met.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record describes a randomized comparison of treatment durations in a selected group expected to have a low risk of relapse. Its primary question is a noninferiority question: whether the shortened-treatment arm’s treatment-success rate at 18 months is sufficiently close to the standard-duration arm’s rate under the stated margin. The registry specifies that noninferiority would be supported if the lower bound of a one-sided 95% confidence interval for the difference in success rates were greater than -7%.
However, the primary outcome analysis is not reported in the registry record. There are no posted primary success counts, estimated difference, confidence interval, or primary analysis result. Therefore, the record does not show whether the lower-bound criterion was met, nor how precisely the treatment-success rates were estimated. The study’s completed status and enrollment of 946 do not substitute for an outcome analysis.
The registry describes a total of three arms, but the primary comparison is limited to Arms B and C. Arm A is characterized as high expected relapse risk with standard-of-care treatment, while Arms B and C are both characterized as low expected relapse risk and differ in treatment duration. The primary comparison therefore concerns the two low-risk arms rather than all trial participants. How the risk classification was determined and how participants were allocated among all arms are not reported in the registry record.
The confidence-interval criterion is a decision rule, not a result. General statistical interpretation: a confidence interval represents uncertainty around an estimated effect under the analysis assumptions. In a noninferiority assessment, the lower bound matters because it indicates how large a disadvantage remains compatible with the data. Without the actual estimate and interval, neither the size nor the uncertainty of the observed comparison can be evaluated.
The posted safety counts describe serious adverse events and deaths by arm, but they do not answer the primary noninferiority question. The registry also lists radiologic, immunologic, and microbiologic measures as secondary outcomes, but does not report analyses for them. Overall, the record establishes the intended design and criterion, not the trial’s comparative efficacy findings.
9. Statistical Methods Explained
The primary outcome is a binary participant-level measure: a participant is classified according to whether treatment success was achieved at 18 months after treatment initiation. The registry describes the outcome as a count of participants, but does not give the detailed success definition or the number analyzed. A binary endpoint can be summarized as a success rate in each arm, then contrasted between arms. The registry does not report those rates or the estimated contrast.
Understanding the noninferiority criterion
General statistical explanation: a noninferiority trial asks whether an intervention is not unacceptably worse than a comparator according to a margin chosen in advance. This differs from seeking proof that the two approaches are identical. The margin defines the largest disadvantage that would still be considered compatible with the trial’s stated objective. In this trial, the registry gives a margin of -7% and specifies that the lower bound of a one-sided 95% confidence interval for the difference in success rates must be greater than that margin to support noninferiority.
The direction and exact calculation of the contrast matter when interpreting such a rule. The registry says the analysis is a difference in success rates between Arms B and C, but does not state the subtraction order in the description. The noninferiority criterion is reported explicitly, but the record does not provide the analysis estimate or interval needed to apply it. No conclusion about meeting the criterion can be drawn from the design description alone.
What a confidence interval conveys
General statistical explanation: a confidence interval combines an estimated effect with its uncertainty. A narrow interval indicates greater precision under the model and data used; a wide interval indicates less precision. For a noninferiority analysis, the position of the lower bound relative to the margin is central. An interval that does not meet the stated boundary would not support the registry’s criterion, even if the point estimate appeared favorable. Since the interval is not reported, its precision and relationship to the margin are unknown.
Randomization, masking, and selection
Randomization is recorded, and the summary says participants thought likely to be cured at four months were assigned to either standard-duration or shortened treatment. In general, random assignment helps make comparison groups similar on average at baseline, while masking can reduce some forms of influence on measurement or behavior. Masking is recorded as none, so the study was open-label. The registry does not detail the randomization procedure, allocation ratio, concealment, or whether outcome assessment was masked. The primary comparison is between the low expected relapse-risk arms, not the high-risk standard-of-care arm.
The registry classifies the design model as sequential. That label alone does not establish whether there were interim looks at outcomes, a sequential stopping rule, or any particular adaptive procedure. Such operational details are not reported in the registry record. They should not be inferred from the model label.
Planning assumptions and analysis results
General statistical explanation: sample-size planning uses assumptions about the effect, variability, allocation, and chosen error rates to estimate the amount of information needed. Planning illustrations are not observed power and do not describe how many participants actually contributed to an outcome analysis. The registry’s planning assumptions describe a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and effects expressed as a stated fraction of the observed effect. These are described as planning illustrations, not as the trial’s own design. The record does not report the trial’s sample-size justification or the numerical outputs of such illustrations.
The registry records enrollment of 946, but does not report the number randomized to each arm for the primary comparison, the number included in the outcome analysis, or the handling of participants with missing outcome data. No primary analysis estimate or test is reported. Thus, the design and planning information cannot be used to reconstruct the trial result.
Why analysis details matter
General statistical explanation: interpretation of a noninferiority result depends on the analysis population, endpoint definition, missing-data handling, and assumptions used to estimate uncertainty. Missing outcomes can affect arm comparisons, and different analysis populations can produce different estimates. The registry does not report these details for the primary endpoint. The posted record therefore allows the planned question and decision threshold to be described, but not an independent assessment of the primary analysis.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists the primary comparison between Arms B and C and identifies radiologic, immunologic, and microbiologic measures as secondary outcomes. It records two outcome measures overall, but does not provide a full statistical analysis plan, a testing hierarchy, or procedures for controlling multiplicity. It is therefore not possible to determine from the registry record how the investigators intended to manage the chance of misleading findings when assessing multiple outcomes or comparisons.
Primary outcome analysis results are not reported in the registry record. The record also does not specify the primary analysis population, whether the analysis was intention-to-treat, per-protocol, or another population, or how participants with missing outcome data were handled. The outcome definition and rules for classifying participants with incomplete follow-up are not reported. These details are particularly relevant to a noninferiority question because exclusions and missing data can affect the estimated difference and its uncertainty.
Interim analyses, stopping guidelines, and any monitoring plan are not reported in the registry record. The design is classified as sequential, but the meaning of that classification and any associated analysis procedures are not described. No specific interim testing approach or stopping rule can be inferred from the label alone.
General statistical explanation: multiplicity concerns arise when several hypotheses or outcomes are assessed, because repeated testing can increase the chance of at least one apparently notable result. Missing-data methods also require assumptions about why values are absent and how that absence relates to outcomes. Without the analysis plan and outcome reporting, the registry record does not allow evaluation of these issues for this trial.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Arm A | 20 | 251 | 8.0% |
| Arm B | 5 | 154 | 3.2% |
| Arm C | 3 | 141 | 2.1% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Arm A | 4 of 251 |
| Arm B | 3 of 154 |
| Arm C | 0 of 141 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Arm A | 0 of 251 |
| Arm B | 0 of 154 |
| Arm C | 0 of 141 |
Most frequent serious adverse event terms
| Event term | Organ system | Arm A | Arm B | Arm C |
|---|---|---|---|---|
| Stab wound | Injury, poisoning and procedural complications | 3 of 251 | 1 of 154 | 0 of 141 |
| Abortion spontaneous | Pregnancy, puerperium and perinatal conditions | 0 of 251 | 0 of 154 | 1 of 141 |
| Alcoholic pancreatitis | Gastrointestinal disorders | 1 of 251 | 0 of 154 | 0 of 141 |
| Animal bite | Injury, poisoning and procedural complications | 1 of 251 | 0 of 154 | 0 of 141 |
| Appendicitis | Gastrointestinal disorders | 1 of 251 | 0 of 154 | 0 of 141 |
| Arteriosclerosis | Vascular disorders | 0 of 251 | 1 of 154 | 0 of 141 |
| Bartholin's cyst | Reproductive system and breast disorders | 1 of 251 | 0 of 154 | 0 of 141 |
| Breast cancer | Neoplasms benign, malignant and unspecified (incl cysts and polyps) | 0 of 251 | 0 of 154 | 1 of 141 |
| Bronchospasm | Respiratory, thoracic and mediastinal disorders | 1 of 251 | 0 of 154 | 0 of 141 |
| Chronic Obstructive Pulmonary Disease | Respiratory, thoracic and mediastinal disorders | 1 of 251 | 0 of 154 | 0 of 141 |
Serious events with the highest arm-level rate: stab wound (1.2%), abortion spontaneous (0.7%), breast cancer (0.7%), pneumonia (0.7%), pyrexia (0.7%).
The registry reports serious adverse events for three arms. Arm A had 20 participants with serious adverse events among 251 at risk; Arm B had 5 among 154 at risk; and Arm C had 3 among 141 at risk. The deaths recorded were 4 in Arm A, 3 in Arm B, and 0 in Arm C. The registry reports 0 other adverse events in each of the three arms. These are descriptive counts and should be read in the context of the arm-specific numbers at risk.
The listed serious terms include Stab wound (1.2%), Abortion spontaneous (0.7%), Breast cancer (0.7%), Pneumonia (0.7%), Pyrexia (0.7%), and Arteriosclerosis (0.6%). The registry record does not provide further context for these term percentages, such as the specific denominator used for each percentage. The counts do not by themselves establish that an event was caused by treatment, and they are not a substitute for a comparative safety analysis.
The registry does not report a statistical analysis of adverse events, adjusted comparisons, or a detailed account of event timing and relatedness. The arm counts should therefore be treated as posted descriptive safety information, not as evidence of a treatment effect or a complete assessment of safety.
12. What the Result Does, and Does Not, Mean
The registry record describes a trial designed to test whether a shortened tuberculosis-treatment duration could meet a stated noninferiority criterion compared with standard-duration treatment among participants classified as having low expected relapse risk. It specifies an 18-month treatment-success outcome and a lower-bound confidence-interval threshold of greater than -7% for the difference in success rates.
The record does not report the primary outcome results. No success counts, estimated difference, confidence interval, or analysis result are available in the registry record. It is therefore not possible to say whether the noninferiority criterion was met, how large the observed difference was, or how precisely the two arms were compared. The enrollment number alone cannot answer those questions.
General statistical explanation: a noninferiority conclusion depends on both the estimated effect and its uncertainty in relation to the prespecified margin. A favorable point estimate alone would not establish noninferiority if the interval left substantial uncertainty about disadvantage. Conversely, the design statement and threshold do not imply that the trial achieved its objective.
The registry’s safety entries provide serious adverse event and death counts by arm, but do not establish causation or supply a complete comparative safety analysis. The listed secondary measures likewise do not have reported results in this record. The statistical meaning of the record is therefore limited to the study question, design features, endpoint description, decision rule, and posted descriptive safety counts. It does not provide the comparative findings required to judge the primary question.
13. Why This Trial Matters Statistically
Statistically, this trial is notable for making treatment duration conditional on an assessment of expected relapse risk and for posing a noninferiority question in the low-risk comparison. That structure distinguishes the primary comparison from a simple comparison of all participants receiving different durations: the primary contrast is between Arm B, described as low expected relapse risk with standard treatment, and Arm C, described as low expected relapse risk with shortened treatment.
The stated margin and confidence-interval rule make the intended decision framework explicit. In general, noninferiority designs require attention to the direction of the contrast, the chosen margin, the analysis population, and the precision of the estimate. In this record, the decision threshold is described, but the primary estimate and interval are not reported. This illustrates why a protocol objective cannot be treated as a finding.
The trial also connects biomarker-based classification with the population targeted for the duration comparison. The registry summary mentions PET/CT imaging and estimating tuberculosis burden, but does not give the biomarker algorithm or thresholds. Without those details and the primary analysis, the record does not show how reproducibly the lower-risk group could be identified or how the intended classification related to the outcome.
The design includes an open-label classification and three arms, while the primary question focuses on two of them. These features make clear why the analysis population and outcome definitions matter. The absence of those details and of primary results limits statistical interpretation, even though the registry records the trial as completed.
14. Limitations
- The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
- Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.
The primary limitation for interpreting this trial is that the registry record does not report the primary endpoint analysis. It provides neither the treatment-success counts nor the estimated difference and confidence interval needed to assess the stated noninferiority criterion. The outcome’s operational definition is also not reported, so the classification of success cannot be evaluated from the record.
The registry summary describes biomarker-informed risk classification and PET/CT imaging, but does not specify the biomarkers, prediction model, cutoffs, or validation procedures used to decide who was likely to be cured after four months. The meaning of the recorded sequential-model classification is also not detailed. These omissions prevent a precise account of how participants entered the primary comparison.
Masking is recorded as none. The registry does not report whether outcome assessors were masked, how randomization was implemented, or the allocation ratio. The analysis population, missing-data rules, interim procedures, and multiplicity strategy are not reported in the registry record. These gaps limit assessment of bias, precision, and robustness.
The enrollment of 946 is not accompanied by arm-specific primary outcome analysis counts. The safety data are descriptive and lack a reported comparative analysis, event timing, and treatment-relatedness details. Finally, the registry lists secondary measures without their results. These limitations concern what can be concluded from the registry record; they do not establish that unreported procedures or analyses were absent from the full study.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Using Biomarkers to Predict TB Treatment Duration — NCT02821832.
- Registry record last updated 2024-04-17; status last verified 2024-04.
- Results first posted 2022-12-20.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT02821832, titled “Using Biomarkers to Predict TB Treatment Duration,” is a completed phase 2 randomized, open-label trial sponsored by the National Institute of Allergy and Infectious Diseases (NIAID). Its recorded enrollment is 946. The study concerns pulmonary tuberculosis and compares standard-duration with shortened treatment in participants described as having low expected relapse risk. The primary comparison is between Arms B and C, with treatment success assessed at 18 months after treatment initiation.
The registry’s stated noninferiority criterion is that the lower bound of a one-sided 95% confidence interval for the difference in success rates be greater than -7%. The primary outcome analysis is not reported in the registry record. Serious adverse event counts and deaths are posted by arm, but the record does not provide a comparative safety analysis. Details about the success definition, biomarker algorithm, analysis populations, missing-data procedures, and multiplicity strategy are also not reported.