This page separates reported trial results from statistical interpretation. Numerical results are restricted to the ClinicalTrials.gov record and the posted statistical analysis. The registry provides the official trial record.
1. Trial at a Glance
TREXA was a randomized, parallel-group, quadruple-masked phase 3 treatment trial in asthma. The registry reports 288 participants, four arms, and a primary endpoint concerning asthma exacerbations requiring systemic corticosteroid therapy during the 44-week treatment period.
| Feature | TREXA |
|---|---|
| Trial name | TREXA |
| NCT identifier | NCT00394329 |
| Therapeutic area | Pulmonology |
| Condition | Asthma |
| Phase | Phase 3 |
| Status | Completed |
| Start | 2006-11 |
| Primary completion | 2011-12 |
| Lead sponsor | Milton S. Hershey Medical Center |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 288 |
| Number of arms | 4 |
2. Clinical Question
The registered primary question concerns whether treatment strategies involving beclomethasone dipropionate affect the occurrence of an asthma exacerbation requiring systemic corticosteroid therapy during the 44-week treatment period.
Population
Participants with asthma enrolled in the phase 3 TREXA treatment trial. The ClinicalTrials.gov record identifies the condition as asthma but do not provide additional baseline demographic or disease-severity detail for this page.
Interventions
The registry lists beclomethasone dipropionate twice among the drug interventions and also lists albuterol sulfate as a drug intervention. The ClinicalTrials.gov record does not provide sufficient arm-level dosing or schedule details to expand those descriptions.
Comparator
The posted primary analysis compares Arm A, Daily ICS + Rescue ICS, with Arm D, Placebo.
Primary question
Does the Arm A strategy change the time to an asthma exacerbation requiring systemic corticosteroid therapy compared with placebo during the 44-week treatment period?
3. Trial Design
Daily ICS + Rescue ICS
The posted primary statistical analysis compares this arm with Arm D. The ClinicalTrials.gov record does not provide a numeric arm-level enrollment for Arm A in the primary efficacy analysis.
Placebo
The placebo arm is the comparator in the posted primary Cox proportional-hazards analysis.
The registry lists four total arms. The registry-reported statistical analysis specifically evaluates Arm A versus Arm D; the available data do not provide formal statistical analyses for the other arm comparisons.
4. Endpoints
| Endpoint | Registry definition / time frame | Analysis information |
|---|---|---|
| Participants Experiencing an Asthma Exacerbation That Requires Systemic Corticosteroid Therapy | Measured during the 44-week treatment period | Cox proportional-hazards model; hazard ratio; all randomized participants included in the time-to-event analysis |
5. Statistical Methodology
Cox proportional-hazards model
The posted primary analysis used regression with a Cox proportional-hazards model. This is a standard survival-analysis model for comparing the timing of an event between groups while accounting for the fact that participants can have different observed follow-up times.
In a two-group setting, the hazard ratio can be represented as exp(β). A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the treatment group relative to the comparator, subject to the model assumptions.
Hazard ratio as the effect measure
The registry reports the effect measure as a Cox proportional hazard, normalized here as a hazard ratio. The primary estimate was 0.56, with a two-sided 95% confidence interval of 0.32 to 0.96.
Analysis population
The posted analysis states that all randomized participants were included in the time-to-event analysis. This is important because analyzing participants according to randomized assignment preserves the treatment comparison established by randomization.
Superiority hypothesis
The analysis is classified as a superiority hypothesis. Thus, the inferential question is whether the randomized treatment comparison provides evidence of a difference in the time-to-event endpoint, rather than whether one strategy is merely no worse than another within a prespecified non-inferiority margin.
Kaplan-Meier framework
The registry-reported statistical analysis identifies the endpoint as time-to-event, making Kaplan-Meier estimation the natural descriptive companion to the Cox model. Kaplan-Meier estimation describes the event-free probability over time while accommodating right-censoring. The ClinicalTrials.gov record does not provide Kaplan-Meier estimates or a numerical survival curve for this page.
Here, di represents events at an event time and ni represents participants at risk immediately beforehand. Censoring allows participants to contribute information up to the point at which their event status is no longer observed.
6. Results
Primary Endpoint: Asthma Exacerbation Requiring Systemic Corticosteroid Therapy
The primary endpoint was measured during the 44-week treatment period. The posted analysis compared Arm A: Daily ICS + Rescue ICS with Arm D: Placebo using a Cox proportional-hazards model.
Hazard ratio for asthma exacerbation
95% CI: 0.32–0.96 · P = 0.066
Two-sided 95% confidence interval · Superiority hypothesis
| Primary endpoint | Arm A | Arm D | Effect estimate |
|---|---|---|---|
| Participants experiencing an asthma exacerbation that requires systemic corticosteroid therapy | Daily ICS + Rescue ICS | Placebo | HR 0.56 (95% CI 0.32–0.96); P = 0.066 |
A hazard ratio of 0.56 means that, under the fitted Cox model, the estimated instantaneous rate of experiencing the primary event in Arm A was approximately 44% lower than in the placebo arm. This is a relative time-to-event measure, not an absolute reduction in the probability of an exacerbation.
The estimate does not mean that 44% of participants avoided an exacerbation, that each participant had exactly a 44% reduction in risk, or that the absolute probability of an exacerbation was reduced by 44 percentage points. A hazard ratio describes a modeled relative event rate over time.
The two-sided 95% confidence interval of 0.32–0.96 describes uncertainty around the estimated hazard ratio under the statistical model and sampling framework. It does not describe the range of effects that individual participants experienced.
The P = 0.066 value addresses the statistical evidence against the null hypothesis under the specified inferential framework; it does not measure the size or clinical importance of the treatment effect. The effect size is described by the hazard ratio and its confidence interval, while the p-value describes evidence relative to a null hypothesis.
Because the analysis uses a Cox proportional-hazards model, interpretation of a single hazard ratio also depends on the proportional-hazards assumption. The ClinicalTrials.gov record does not report a formal assessment of that assumption. The analysis population was all randomized participants, which is consistent with preserving the randomized comparison for the time-to-event analysis.
7. Secondary Endpoint Results
The ClinicalTrials.gov record states that 10 outcome measures were posted, but they provide a formal statistical analysis for only 1 endpoint: the primary endpoint of asthma exacerbation requiring systemic corticosteroid therapy.
Because the ClinicalTrials.gov record does not contain secondary-endpoint estimates, confidence intervals, or p-values, no secondary efficacy result is reproduced here. The available statistical-analyses record contains no additional analyses omitted for length.
| Results information | Registry data reported |
|---|---|
| Outcome measures posted | 10 |
| Statistical analyses posted | 1 |
| Primary-endpoint analyses | 1 |
| Primary analyses with estimate + CI | 1 |
| Analyses omitted for length | 0 |
8. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized arm as affected participants over participants at risk. These figures are presented exactly as reported in the registry rather than converted into percentages.
| Arm | Serious adverse events | Affected / at risk |
|---|---|---|
| A: Daily ICS + Rescue ICS | Serious adverse events | 4/71 |
| B: Daily ICS | Serious adverse events | 2/72 |
| C: Rescue ICS | Serious adverse events | 6/71 |
| D: Placebo | Serious adverse events | 17/74 |
The serious-adverse-event data show the number affected and the corresponding number at risk for each of the four arms. These are descriptive safety results. The ClinicalTrials.gov record does not provide a formal statistical comparison of serious adverse events between arms, so no inferential conclusion about differences between the groups is added here.
Safety and efficacy also answer different statistical questions. The primary efficacy analysis is a time-to-event model based on randomized participants, whereas serious adverse events are presented here as arm-level affected/at-risk counts. The two types of evidence should therefore not be combined into a single numerical treatment-effect measure.
9. Statistical Methods Explained
Why was a Cox proportional-hazards model used?
The primary endpoint concerns the occurrence of an asthma exacerbation during a defined treatment period, and the posted analysis classifies it as a time-to-event endpoint. A Cox model is designed to compare the event rate over time while accommodating censoring and unequal event times. It produces a hazard ratio as a relative measure of the treatment effect.
What does a hazard ratio of 0.56 mean?
A hazard ratio of 0.56 indicates a lower estimated instantaneous event rate for Arm A relative to placebo under the fitted model. Equivalently, 0.56 corresponds to an estimated relative reduction of approximately 44% in the hazard. It does not mean that the probability of an exacerbation was exactly 44% lower for every participant.
Why is the confidence interval important?
The 95% confidence interval of 0.32–0.96 shows that the estimate has uncertainty. The point estimate alone, 0.56, can give a false impression of precision if its interval is ignored. The interval describes uncertainty in the estimated treatment effect; it is not a prediction interval for individual participants.
Why doesn't the p-value measure effect size?
The P = 0.066 value is an inferential measure under the specified null-hypothesis framework. It is affected by both the magnitude of the observed effect and the amount of information in the analysis. A p-value therefore cannot replace the hazard ratio or its confidence interval when describing the size and precision of the treatment effect.
What does randomization contribute to the analysis?
Randomization establishes the treatment assignment mechanism before outcomes are observed. When all randomized participants are included in the time-to-event analysis according to their randomized groups, the primary comparison retains the principal causal advantage created by random allocation.
What does quadruple masking contribute?
Quadruple masking reduces opportunities for knowledge of treatment assignment to influence trial conduct or outcome-related processes. The ClinicalTrials.gov record identifies the trial as quadruple-masked but do not specify the four masked roles, so the analysis does not assign particular roles without evidence.
Why might Kaplan-Meier estimation accompany the Cox model?
Kaplan-Meier estimation provides a descriptive picture of the event-free experience over time, while the Cox model provides a relative hazard estimate. For a time-to-event endpoint such as the TREXA primary endpoint, the two approaches answer related but distinct questions: Kaplan-Meier describes the time-dependent event experience, while the Cox model summarizes the relative hazard between treatment groups.
10. Understanding the Primary Analysis More Deeply
Relative effect
The HR of 0.56 is a relative measure comparing the modeled event hazard in Arm A with that in placebo.
Precision
The 95% CI of 0.32–0.96 communicates uncertainty around the hazard-ratio estimate.
Inference
The two-sided P = 0.066 describes evidence against the relevant null hypothesis; it is not an effect-size measure.
Population
The posted time-to-event analysis included all randomized participants.
One of the most important statistical distinctions in this trial is the difference between a time-to-event endpoint and a simple binary endpoint. If analysis only asked whether each participant ever experienced an exacerbation during follow-up, the result could be represented as a binary outcome. The posted analysis instead identifies the endpoint as time-to-event and uses a Cox model, meaning the timing of the event and censoring information contribute to the analysis.
This distinction matters because two participants could both experience an exacerbation during the treatment period but contribute different information if their events occur at different times. Similarly, participants without an observed event at the end of available follow-up are not necessarily treated as though they had zero lifetime risk; their observations can be censored within the survival-analysis framework.
Hazard ratio: describes the estimated relative effect. Confidence interval: describes uncertainty around that estimate. P-value: describes evidence relative to a null hypothesis under the specified statistical framework.
11. What the Primary Hazard Ratio Does — and Does Not — Mean
The HR of 0.56 indicates a lower estimated hazard of the primary event in the Daily ICS + Rescue ICS arm compared with placebo. The direct derived interpretation is an approximately 44% lower estimated hazard under the fitted Cox model.
The HR does not say that 44% of participants avoided an exacerbation, that the absolute event probability decreased by 44 percentage points, or that every participant experienced the same proportional change in risk.
The 95% CI of 0.32–0.96 describes uncertainty around the estimated hazard ratio. A confidence interval should be interpreted as an interval produced by the statistical procedure under its assumptions, not as a statement that there is a 95% probability that the fixed treatment effect lies inside this particular interval.
The two-sided P = 0.066 should not be interpreted as a 6.6% probability that the observed treatment effect occurred by chance, nor as a 93.4% probability that the treatment is effective. The p-value is tied to a null-hypothesis testing framework and depends on the observed data and analysis assumptions.
12. Analysis Population and Censoring
The posted statistical analysis explicitly states that all randomized participants were included in the time-to-event analysis. This is an important feature of the primary efficacy analysis because treatment assignment, rather than subsequent treatment experience, defines the randomized comparison.
Time-to-event analysis also requires a rule for participants whose event is not observed during the relevant observation period. Such observations can be censored rather than treated as if the event could never occur. The ClinicalTrials.gov record does not provide the detailed censoring rules or missing-data procedures, so no additional assumptions about how censoring was implemented are made here.
13. Proportional-Hazards Assumption
The Cox proportional-hazards model assumes that the relative hazard between the compared groups is reasonably represented by a proportional relationship over time. Conceptually, if the hazard ratio is 0.56, the model treats the relative hazard as being summarized by that value across the analyzed time scale rather than requiring a separate treatment effect at every time point.
The proportional-hazards assumption means that the ratio of the hazards is treated as approximately constant over time. If hazards are not proportional, a single HR can become less descriptive of how treatment effects evolve during follow-up.
The ClinicalTrials.gov record does not report a formal test or diagnostic for proportional hazards. Consequently, the HR of 0.56 should be understood as the effect measure produced by the posted Cox analysis, without adding an unsupported claim that the proportional-hazards assumption was formally demonstrated.
14. Multiplicity, Interim Analysis, and Other Design Features
The ClinicalTrials.gov record does not report a non-inferiority margin, crossover design, factorial design, multiplicity procedure, interim-analysis plan, alpha-spending method, or Bayesian analysis. These topics therefore are not assigned specific methods or interpretations for TREXA.
| Design topic | Information reported |
|---|---|
| Non-inferiority margin | Not reported in the ClinicalTrials.gov record; the primary hypothesis is classified as superiority. |
| Crossover | Not reported in the ClinicalTrials.gov record. |
| Factorial design | Not reported; the registry identifies a parallel design. |
| Multiplicity adjustment | Not reported in the ClinicalTrials.gov record. |
| Interim analysis | Not reported in the ClinicalTrials.gov record. |
| Missing-data / imputation method | Not reported in the ClinicalTrials.gov record. |
| Bayesian methods | Not reported; the posted method is a Cox proportional-hazards regression. |
| Stratification factors | Not reported in the ClinicalTrials.gov record. |
The absence of a reported design feature in the ClinicalTrials.gov record should not be interpreted as proof that the feature was never considered in the underlying protocol. It means only that the available trial data for this page do not support a more specific statement.
15. Trial Timeline
Trial start
The TREXA trial began in November 2006 according to the ClinicalTrials.gov record.
Parallel-group treatment study
The registry identifies a phase 3 randomized, parallel-group trial with quadruple masking and four arms.
Primary endpoint observation window
The primary endpoint was measured during the 44-week treatment period.
Primary completion
The ClinicalTrials.gov record gives December 2011 as the primary completion date.
16. Why This Trial Matters Statistically
TREXA is a useful statistical teaching case because the primary analysis illustrates how a clinically defined binary-looking event can be analyzed using its time-to-event structure. The endpoint asks whether participants experience an asthma exacerbation requiring systemic corticosteroid therapy, but the posted analysis uses the timing of that event and a Cox proportional-hazards model rather than reducing the analysis to a simple event/no-event comparison.
| Statistical concept | How it appears in TREXA |
|---|---|
| Randomization | The trial used randomized allocation. |
| Parallel-group design | The registry identifies the design model as parallel. |
| Quadruple masking | The registry identifies the trial as quadruple-masked. |
| Time-to-event endpoint | The posted primary statistical analysis identifies the primary endpoint as time-to-event. |
| Cox proportional-hazards model | The primary analysis used Cox regression. |
| Hazard ratio | The primary effect measure was a hazard ratio of 0.56. |
| Confidence interval | The primary estimate had a two-sided 95% CI of 0.32–0.96. |
| P-value | The primary analysis reported P = 0.066. |
| Analysis population | All randomized participants were included in the time-to-event analysis. |
| Superiority testing | The posted hypothesis type was superiority. |
| Safety by randomized arm | Serious adverse events were reported as affected/at-risk counts for all four arms. |
17. A Statistical Reading of the TREXA Result
The primary result can be read in layers rather than as a single number. First, the hazard ratio of 0.56 gives the estimated relative treatment effect from the Cox model. Second, the 95% CI of 0.32–0.96 provides information about the precision of that estimate. Third, the two-sided P = 0.066 supplies the reported hypothesis-testing result.
These quantities should not be collapsed into one statement. For example, saying simply that the result "was significant" or "was not significant" would discard the information contained in the effect estimate and its confidence interval. Conversely, focusing only on the HR of 0.56 would omit the uncertainty represented by the interval and the inferential result.
The analysis is also specific to the comparison that was posted: Daily ICS + Rescue ICS versus placebo. The presence of four randomized arms does not justify extending the HR of 0.56 to the other two arms. Each treatment comparison requires its own analysis if an inferential comparison is to be made.
18. Limitations
- Limited reported efficacy detail: The ClinicalTrials.gov record contains one formal statistical analysis, so the page cannot provide a broader set of efficacy estimates without introducing unsupported results.
- Four-arm design but one posted analysis: The trial has four arms, while the registry-reported formal primary analysis compares only Daily ICS + Rescue ICS with placebo.
- Endpoint classification discrepancy: The registry profile labels the registered primary endpoint as binary, while the posted analysis identifies it as time-to-event. Interpretation here follows the posted Cox analysis.
- No reported baseline table: The ClinicalTrials.gov record does not include baseline demographic or clinical characteristics, so no balance assessment is attempted.
- No reported Kaplan-Meier estimates: Although the endpoint is analyzed as time-to-event, the ClinicalTrials.gov record does not provide numerical Kaplan-Meier estimates or a curve.
- No formal proportional-hazards diagnostic: The ClinicalTrials.gov record does not report an assessment of the Cox proportional-hazards assumption.
- No detailed censoring rules: The ClinicalTrials.gov record does not provide the complete censoring or missing-data methodology.
- No reported multiplicity framework: The ClinicalTrials.gov record does not describe adjustment for multiple endpoints, multiple treatment comparisons, or repeated looks.
- No reported interim-analysis framework: The ClinicalTrials.gov record does not identify an interim-analysis schedule or alpha-spending approach.
- No secondary statistical estimates reported: Although 10 outcome measures were posted, only one statistical analysis is included in the ClinicalTrials.gov record.
- Safety inference is descriptive: Serious adverse-event counts are reported by arm, but no formal between-arm statistical analysis is reported.
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The posted Cox analysis estimated a hazard ratio of 0.56 for Daily ICS + Rescue ICS versus placebo, with a two-sided 95% CI of 0.32–0.96 and P = 0.066.
Clinical interpretation
The result concerns the timing of asthma exacerbations requiring systemic corticosteroid therapy during the 44-week treatment period. The ClinicalTrials.gov record does not provide enough absolute event-time information to characterize the magnitude of benefit using absolute risks or median event times.
The statistical result therefore supports a precise description of the estimated relative event hazard, but the ClinicalTrials.gov record does not support a broader claim about absolute exacerbation risk, median time to exacerbation, or treatment effects for the other two arms.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Calculators
22. Sources
- ClinicalTrials.gov: TREXA — NCT00394329.
- Linked publication: PubMed record for PMID 21324520.
- Linked publication: PubMed record for PMID 36161875.
Continue with Clinical Biostats
Explore statistical tutorials and analysis tools related to randomized trials, time-to-event endpoints, hazard ratios, confidence intervals, and hypothesis testing.
23. Record Summary
TREXA is a phase 3 randomized, parallel-group, quadruple-masked asthma trial with 288 participants and four arms. Its registered primary endpoint was participants experiencing an asthma exacerbation requiring systemic corticosteroid therapy, measured during the 44-week treatment period. The posted primary analysis treated this endpoint as time-to-event and used a Cox proportional-hazards model in all randomized participants.
The primary statistical comparison was Daily ICS + Rescue ICS versus placebo. The reported hazard ratio was 0.56, with a two-sided 95% confidence interval of 0.32–0.96 and P = 0.066. The most informative statistical reading is to consider these three quantities together: the hazard ratio describes the estimated relative event hazard, the confidence interval describes uncertainty around that estimate, and the p-value describes evidence relative to the specified null hypothesis.
The ClinicalTrials.gov record also report serious adverse events for all four arms: 4/71 in Daily ICS + Rescue ICS, 2/72 in Daily ICS, 6/71 in Rescue ICS, and 17/74 in placebo. These safety data are descriptive in the ClinicalTrials.gov record because no formal statistical comparison is provided.