This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
RIFASHORT was a randomized phase 3 trial evaluating the toxicity and efficacy of 1200mg and 1800mg rifampicin for pulmonary tuberculosis. The registry reports an enrollment of 672 participants and three arms.
| Feature | RIFASHORT |
|---|---|
| Phase | Phase 3 |
| Condition | Pulmonary Tuberculosis |
| Allocation | Randomized |
| Design model | Single group |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 672 |
| Number of arms | 3 |
| Lead sponsor | St George's, University of London |
| Status | Completed |
| ClinicalTrials.gov | NCT02581527 |
2. Clinical Question
The clinical question concerns whether higher rifampicin doses of 1200mg and 1800mg can be evaluated for both toxicity and efficacy in pulmonary tuberculosis. The registry title identifies these two rifampicin doses as the central intervention comparison, while the listed interventions include rifampicin, isoniazid, ethambutol, and pyrazinamide.
Population
Participants with pulmonary tuberculosis.
Intervention focus
Rifampicin doses of 1200mg and 1800mg.
Background therapy
The registry lists rifampicin, isoniazid, ethambutol, and pyrazinamide as study interventions.
Primary questions
How frequently do grade 3 or 4 adverse events occur, and what is the combined rate of treatment failure and subsequent relapse in the specified efficacy population?
3. Trial Design
The registry therefore describes a randomized, unmasked phase 3 treatment study with three arms. It does not provide the arm-level sample sizes or detailed dosing schedules for each arm in the information reported here. Those details are important for reconstructing arm-specific statistical comparisons and are not assumed from the number of arms alone.
4. Trial Population and Interventions
The registry identifies pulmonary tuberculosis as the condition under study and lists four drug interventions: rifampicin, isoniazid, ethambutol, and pyrazinamide. The study title specifically identifies 1200mg and 1800mg rifampicin as the dose levels being evaluated.
| Study feature | Registry description |
|---|---|
| Condition | Pulmonary Tuberculosis |
| Rifampicin doses identified in title | 1200mg and 1800mg |
| Other listed drugs | Isoniazid, ethambutol, pyrazinamide |
| Total enrollment | 672 |
| Number of arms | 3 |
The available registry information does not specify the number randomized to each arm, the exact composition of every arm, or the dose schedule for the non-rifampicin components. Those distinctions matter statistically because the appropriate treatment contrast depends on the actual randomized comparison.
5. Primary Endpoints
| Endpoint | Registry definition | Time frame | Statistical type |
|---|---|---|---|
| Grade 3 or 4 adverse events | The occurrence of grade 3 or 4 adverse events at any time during chemotherapy. | 18 months | Binary safety outcome over follow-up |
| Combined failure and relapse | The combined rate of failure at the end of treatment and relapse during the subsequent 12 months in smear positive patients in the modified intent to treat population. | 18 months | Binary/combined efficacy outcome |
The two primary endpoints represent different statistical questions. The first is a safety endpoint based on whether a participant experiences a grade 3 or 4 adverse event at any time during chemotherapy. The second combines treatment failure at the end of treatment with relapse during the subsequent 12 months, and is specifically defined for smear positive patients in the modified intent-to-treat population.
6. Planned Analysis
No formal statistical analyses were posted to ClinicalTrials.gov. The registry nevertheless identifies two primary outcomes that can be understood in terms of their appropriate statistical structure.
Grade 3 or 4 adverse events
The first endpoint is an occurrence endpoint: each participant can be classified according to whether at least one grade 3 or 4 adverse event occurred at any time during chemotherapy. A conventional primary comparison for such an endpoint would use the proportion of participants experiencing the event in each randomized treatment group.
Treatment groups can then be compared using a risk difference, risk ratio, odds ratio, or an appropriate exact or asymptotic hypothesis test. The most informative presentation would include both the event proportions and an uncertainty interval for the between-group effect.
Because the endpoint asks whether an adverse event occurred at any time during chemotherapy, a participant with multiple qualifying events would generally remain a single participant with an event for this particular occurrence analysis. Counts of repeated events would represent a different estimand and would require a different statistical model.
Combined treatment failure and relapse
The second primary endpoint is also naturally expressed as a proportion: the combined rate of failure at the end of treatment plus relapse during the subsequent 12 months. The registry restricts this endpoint to smear positive patients in the modified intent-to-treat population.
For a randomized comparison, the principal effect could be summarized with an absolute risk difference and a relative measure such as a risk ratio. A confidence interval would quantify uncertainty around the estimated treatment contrast.
The definition combines two clinically different points in the treatment pathway: failure at the end of treatment and relapse during the following 12 months. The combined endpoint therefore asks whether either component occurred, rather than estimating two completely independent outcomes.
7. Why the Endpoint Definitions Matter
Statistical interpretation begins with the estimand: the precise quantity the trial seeks to compare. RIFASHORT's two primary endpoints have different event structures, populations, and follow-up definitions.
Safety estimand
Whether a participant experiences a grade 3 or 4 adverse event at any time during chemotherapy over the 18-month timeframe.
Efficacy estimand
The combined rate of treatment failure at the end of treatment and relapse during the subsequent 12 months.
Population matters
The efficacy endpoint is explicitly restricted to smear positive patients in the modified intent-to-treat population.
Time matters
The registry specifies an 18-month timeframe for both primary endpoint listings, while the efficacy definition itself identifies relapse during the subsequent 12 months.
These distinctions prevent a common statistical error: treating every trial endpoint as if it were simply a generic percentage. The denominator, event definition, treatment period, follow-up period, and analysis population determine what a reported rate actually means.
8. Statistical Methodology
Comparison of binary outcomes
Both primary endpoints can be represented as participant-level occurrence outcomes. For example, for grade 3 or 4 adverse events, each participant can be classified as having experienced at least one qualifying event or not. The same framework can be used for the combined failure-or-relapse endpoint.
The risk difference expresses the absolute difference in the proportion experiencing the endpoint between two groups. Unlike a relative measure, it directly describes the change in event probability on the original percentage scale.
Risk ratio
A risk ratio compares the event probability between groups. A value below 1 indicates a lower observed event probability in the numerator group relative to the denominator group, while a value above 1 indicates a higher probability.
For example, an RR of 0.80 would correspond to an event probability 20% lower on a relative scale, provided the same endpoint and populations are being compared.
Odds ratio
An odds ratio is another possible measure for a binary endpoint. It compares the odds of an event rather than the probabilities themselves. Odds ratios can differ substantially from risk ratios when the outcome is common, so the measure should always be labeled clearly.
Confidence intervals
A confidence interval describes statistical uncertainty around an estimated effect under the specified statistical framework. It should be interpreted alongside the point estimate rather than as a range of individual patient outcomes.
Modified intention-to-treat analysis
The registry explicitly identifies a modified intent-to-treat population for the primary efficacy outcome and further restricts the endpoint to smear positive patients. This differs from simply analyzing every enrolled participant. The exact eligibility rules defining the modified population are not reported here.
9. Statistical Methods Explained
Why is the adverse-event endpoint treated as an occurrence outcome?
The registry asks about the occurrence of grade 3 or 4 adverse events at any time during chemotherapy. That wording naturally defines a participant-level event indicator: whether the participant experienced at least one qualifying event. If repeated adverse events were instead analyzed as separate observations, the analysis would address a different question and would need to account for within-participant correlation.
Why would a risk difference be useful for this trial?
A risk difference puts the comparison on an absolute scale. If one treatment group had an event probability of 30% and another had 40%, the risk difference would be 10 percentage points. This is often easier to interpret clinically than a relative measure alone because it describes the actual difference in event probability.
Why might a risk ratio also be reported?
A risk ratio expresses the treatment contrast relative to the comparator group's event probability. It can be useful when the scientific question concerns proportional differences in event frequency. Risk differences and risk ratios answer different questions, so reporting both can provide a fuller description when appropriate.
Why does the efficacy endpoint specify a modified intent-to-treat population?
The analysis population determines who contributes to the denominator and therefore directly affects the estimated event rate. Here, the registry specifically defines the primary efficacy outcome in smear positive patients in the modified intent-to-treat population. Applying the endpoint to a different population would change the estimand.
Why combine treatment failure and relapse?
The registry defines the primary efficacy measure as a combined rate. Statistically, this creates a composite binary outcome in which a participant is counted if either treatment failure at the end of treatment or relapse during the subsequent 12 months occurs. The benefit of combining components is that the analysis can address a broader treatment-success concept with one primary outcome, but interpretation should still consider the individual components separately when their clinical meanings differ.
Why is the 18-month timeframe important?
An event rate is inseparable from its observation window. The same treatment could produce different event rates over different follow-up periods. The registry specifies 18 months for the primary endpoint timeframes, while the efficacy definition specifically describes relapse during the subsequent 12 months after treatment.
10. Safety Endpoint: Statistical Interpretation
The first primary endpoint is the occurrence of grade 3 or 4 adverse events at any time during chemotherapy. This is fundamentally different from comparing the number of adverse-event episodes.
The endpoint measures the proportion of participants who experience at least one grade 3 or 4 adverse event during the specified treatment period. It does not by itself describe the number of events per participant, their duration, their timing, or their individual clinical severity beyond the grade threshold specified by the endpoint.
A risk difference would describe the absolute difference in the probability of experiencing the qualifying adverse-event endpoint between randomized groups. A risk ratio would describe the same contrast on a relative scale.
An estimated treatment-group difference is subject to sampling variability. A confidence interval would show how precisely the study estimated that difference under the selected statistical method. A narrow interval generally indicates greater precision than a wide interval, but precision should not be confused with clinical importance.
11. Efficacy Endpoint: Statistical Interpretation
The primary efficacy outcome combines failure at the end of treatment with relapse during the subsequent 12 months in smear positive patients in the modified intent-to-treat population.
The endpoint asks whether either of two specified unfavorable outcomes occurs: failure at the end of treatment or relapse during the subsequent 12 months. It therefore summarizes treatment failure across the defined follow-up period rather than measuring a continuous outcome such as time to relapse.
A risk ratio below 1 would indicate a lower observed probability of the combined endpoint in the numerator treatment group relative to the comparator group. It would not mean that every patient experienced the same proportional reduction in risk.
Because the registry specifies smear positive patients in the modified intent-to-treat population, the denominator is narrower than the total enrollment of 672. The efficacy rate should therefore never be interpreted as a rate among all enrolled participants unless the analysis population is shown to be identical.
12. Multiplicity and Three-Arm Design
The registry reports three arms and two primary endpoints. A multi-arm trial creates more than one possible treatment contrast, while multiple primary endpoints create more than one formal efficacy or safety question. The statistical interpretation therefore depends on which comparisons were prespecified as confirmatory.
| Design feature | Statistical implication |
|---|---|
| Three arms | More than one treatment comparison may be possible. |
| Two primary endpoints | Safety and efficacy are both designated as primary outcomes. |
| Randomized allocation | Randomization supports comparisons between assigned treatment groups. |
| No masking | Treatment assignment was not masked, which can matter particularly for subjective outcomes and treatment management. |
When multiple hypotheses are tested, the familywise probability of at least one false-positive conclusion can increase if no adjustment or hierarchical strategy is specified. The registry information available here does not report an alpha-allocation strategy, multiplicity adjustment, or endpoint hierarchy, so those design details are not assigned to the trial.
13. Analysis Populations and Denominators
The distinction between enrollment and analysis population is particularly important for RIFASHORT because the primary efficacy endpoint is explicitly restricted to a modified intent-to-treat population of smear positive patients.
| Population concept | What is reported | Why it matters |
|---|---|---|
| Total enrollment | 672 | Describes the overall number enrolled in the study. |
| Primary efficacy population | Smear positive patients in the modified intent-to-treat population | Defines who contributes to the primary efficacy endpoint. |
| Safety endpoint population | The endpoint is defined by occurrence during chemotherapy | The exact safety analysis population is not specified in the available registry information. |
Using the enrollment number as the denominator for the efficacy endpoint without verifying the modified intent-to-treat definition could produce a misleading rate. In clinical-trial statistics, the population is part of the endpoint definition rather than a secondary bookkeeping detail.
14. Missing Data and Follow-up
The registry information identifies the endpoint timeframes but does not report a specific missing-data or imputation strategy. For an occurrence endpoint observed over follow-up, incomplete outcome ascertainment can affect the estimated event proportion if participants do not have sufficient follow-up to establish whether the endpoint occurred.
Safety follow-up
The safety endpoint concerns events occurring at any time during chemotherapy, with an 18-month timeframe listed for the primary endpoint.
Efficacy follow-up
The efficacy definition incorporates treatment failure and relapse during the subsequent 12 months, within the listed 18-month timeframe.
Incomplete observation
Participants without complete ascertainment can create uncertainty about whether an event would have occurred later in the specified window.
Imputation
No specific imputation method is reported in the available registry information.
For a binary endpoint, the appropriate handling of missing outcome status depends on why observations are missing and on the prespecified analysis plan. Simple assumptions such as treating every missing participant as a success or every missing participant as a failure can produce biased results when missingness is informative.
15. Time-to-Event Analysis vs Fixed-Time Rates
The primary endpoints listed for RIFASHORT are expressed as occurrence or combined rates over a defined period rather than as explicitly stated time-to-event endpoints. This distinction affects the statistical method.
| Endpoint structure | Typical statistical representation | Primary interpretation |
|---|---|---|
| Event occurrence over a defined window | Risk / proportion | Probability of experiencing the endpoint during the specified period |
| Time until an event | Kaplan-Meier / Cox model | Distribution and relative rate of event occurrence over time |
| Repeated events | Repeated-event or count model | Frequency or burden of multiple events |
This distinction matters for the grade 3 or 4 adverse-event endpoint. The phrase "occurrence ... at any time" supports a participant-level occurrence analysis. It does not, by itself, establish that the number or timing of repeated adverse events should be analyzed with a recurrent-event model.
16. Why This Trial Matters Statistically
RIFASHORT provides a useful teaching example because the study combines randomized treatment allocation, multiple arms, two primary endpoints, a safety outcome, a composite efficacy outcome, and a specifically defined modified intent-to-treat population.
| Concept | How it appears in RIFASHORT |
|---|---|
| Randomization | The registry identifies allocation as randomized. |
| Multi-arm design | The study has three arms. |
| Safety analysis | Grade 3 or 4 adverse-event occurrence is a primary endpoint. |
| Composite outcome | The primary efficacy endpoint combines treatment failure and subsequent relapse. |
| Analysis population | The efficacy endpoint is restricted to smear positive patients in the modified intent-to-treat population. |
| Fixed follow-up | The registry gives an 18-month timeframe for both primary endpoint listings. |
| Unmasked treatment | The masking designation is none. |
| Multiplicity | Three arms and two primary endpoints create a potentially important hypothesis-testing structure. |
| Endpoint definition | The precise event definition determines the appropriate statistical estimand. |
The trial is especially useful for illustrating why a statistical analysis cannot be separated from endpoint definitions. A percentage without its population, event definition, and follow-up period is incomplete statistical information.
17. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record does not provide formal statistical analysis results for the primary endpoints.
- No numerical outcome estimates: the available registry information does not provide event counts, treatment-group rates, effect estimates, confidence intervals, or p-values for the primary endpoints.
- Arm-level detail: the registry information does not provide the sample size or complete intervention schedule for each of the three arms.
- Comparison structure: the available information does not specify which pairwise or multi-arm comparisons were prespecified as primary.
- Analysis-plan detail: a specific multiplicity strategy, alpha allocation, missing-data method, and statistical test are not reported in the available registry information.
- Composite endpoint: the efficacy outcome combines treatment failure and relapse, which may have different clinical interpretations even though they are summarized together.
- Modified intent-to-treat population: the registry identifies this population but does not provide the detailed criteria needed to reconstruct its exact denominator.
- Unmasked design: the trial has no masking, which can be relevant to treatment behavior, assessment, and reporting, depending on the endpoint.
18. Trial Timeline
Study start
The registry lists February 1, 2017 as the study start date.
Study period
The completed phase 3 study evaluated toxicity and efficacy in pulmonary tuberculosis with an enrollment of 672.
Primary completion
The registry lists January 1, 2022 as the primary completion date.
Registry status
The current registry record identifies the study as completed.
19. Interpreting a Future Risk Difference
For the primary safety endpoint, a risk difference would be particularly straightforward to interpret because the endpoint is a participant-level occurrence measure.
Risk difference
The value represents the absolute difference in the probability of experiencing at least one grade 3 or 4 adverse event between two specified randomized groups.
An RD of +0.05 would mean that the event probability was 5 percentage points higher in the first group than in the second group. An RD of −0.05 would mean it was 5 percentage points lower.
A risk difference would not describe the number of adverse-event episodes, the severity distribution below or above the grade threshold, or the timing of events. It would describe only the specified occurrence endpoint.
A confidence interval around the risk difference would quantify statistical uncertainty around the estimated between-group difference. Its width would depend on factors including sample size and the observed event frequencies.
20. Interpreting a Future Risk Ratio
The same primary endpoint could also be summarized with a risk ratio. This would put the comparison on a relative rather than absolute scale.
Risk ratio
A value of 1 represents equal event probabilities between the two specified groups; values below or above 1 describe relative differences in opposite directions.
The risk ratio should not be confused with an odds ratio. When events are relatively common, the odds ratio can be numerically farther from 1 than the corresponding risk ratio. Clear labeling of the effect measure is therefore essential.
21. Composite Endpoint: Failure and Relapse
The primary efficacy outcome is explicitly a combined rate of failure at the end of treatment and relapse during the subsequent 12 months. This is a composite endpoint with two components that occur at different stages of treatment and follow-up.
| Component | Timing in the registry definition | Statistical role |
|---|---|---|
| Failure | At the end of treatment | First component of the combined primary efficacy endpoint |
| Relapse | During the subsequent 12 months | Second component of the combined primary efficacy endpoint |
| Combined outcome | 18-month endpoint timeframe | Primary efficacy measure |
Composite endpoints can increase the number of observed events, which can improve statistical efficiency, but they require careful interpretation. A treatment effect on the composite does not necessarily imply the same magnitude of effect on each component. Component-specific results are therefore useful when available.
22. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The trial's primary endpoints are binary or composite occurrence measures that can be expressed using treatment-group risks and between-group effect measures such as risk differences or risk ratios.
Clinical interpretation
The safety endpoint concerns grade 3 or 4 adverse events, while the efficacy endpoint concerns treatment failure and subsequent relapse. These outcomes represent different dimensions of treatment effect and should not be collapsed into a single summary judgment.
The distinction is important because statistical significance and clinical importance are separate concepts. A small absolute difference can be statistically detectable in a large study, while a larger-looking estimate can remain uncertain in a smaller subgroup or analysis population.
23. What Is and Is Not Reported by the Registry
| Information | Registry status |
|---|---|
| Study phase | Reported: Phase 3 |
| Enrollment | Reported: 672 |
| Number of arms | Reported: 3 |
| Randomized allocation | Reported |
| Primary endpoints | Reported |
| Endpoint timeframes | Reported: 18 months |
| Primary efficacy population | Reported as smear positive patients in the modified intent-to-treat population |
| Primary endpoint event counts | Not reported in the available registry information |
| Effect estimates | Not reported in the available registry information |
| Confidence intervals | Not reported in the available registry information |
| p-values | Not reported in the available registry information |
| Posted statistical analyses | None |
This distinction is central to reading a completed clinical trial record. A study can have a completed status and clearly defined primary endpoints without the registry containing numerical statistical results for those endpoints.
24. Why This Trial Matters for Clinical Trial Statistics
RIFASHORT illustrates several recurring statistical issues in randomized clinical research. The most important is the relationship between the clinical question and the statistical estimand.
The safety endpoint asks whether a participant experiences a severe adverse event during chemotherapy. The efficacy endpoint asks whether a participant experiences treatment failure or later relapse within the defined follow-up. Those are not interchangeable outcomes, and they require different interpretive frameworks even though both can ultimately be summarized as rates.
The trial also illustrates why randomized allocation, analysis population, endpoint definition, and follow-up period must be considered together. A reported percentage without those four pieces of context can be difficult to interpret correctly.
25. Sources
- ClinicalTrials.gov: RIFASHORT, NCT02581527.
- PubMed: PubMed record associated with the RIFASHORT study.
Continue through the Clinical Biostats clinical-trial library
Clinical trial analysis is most useful when study design, endpoint definitions, analysis populations, effect measures, and uncertainty are interpreted together.
26. Record Summary
RIFASHORT is a completed randomized phase 3 trial with 672 participants and three arms, evaluating the toxicity and efficacy of 1200mg and 1800mg rifampicin for pulmonary tuberculosis. Its two primary endpoints address different questions: the occurrence of grade 3 or 4 adverse events during chemotherapy, and the combined rate of treatment failure at the end of treatment plus relapse during the subsequent 12 months in smear positive patients in the modified intent-to-treat population.
Statistically, the study is a useful example of binary safety analysis, composite efficacy endpoints, modified intent-to-treat populations, multi-arm trial structure, and the importance of defining the denominator and observation period before interpreting an event rate. The registry identifies the primary outcomes and their 18-month timeframes, but no formal statistical analyses are posted in the ClinicalTrials.gov record.