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
ORBIT-3 was a randomized, parallel-group, quadruple-masked phase 3 study evaluating ciprofloxacin dispersion for inhalation against placebo in patients with non-cystic-fibrosis bronchiectasis. The registry identifies time to first exacerbation over one year as the primary endpoint.
| Feature | ORBIT-3 |
|---|---|
| Study title | Phase 3 Study With Ciprofloxacin Dispersion for Inhalation in Non-CF Bronchiectasis (ORBIT-3) |
| Phase | Phase 3 |
| Condition | Non Cystic Fibrosis Bronchiectasis |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 278 |
| Number of arms | 2 |
| Interventions | Ciprofloxacin dispersion for inhalation; placebo |
| Lead sponsor | Aradigm Corporation |
| Sponsor type | Industry |
| Study status | Completed |
2. Clinical Question
The central statistical question is whether treatment assignment to ciprofloxacin dispersion for inhalation changes the time until the first exacerbation compared with placebo during the registry-defined one-year observation period.
Population
Participants with non-cystic-fibrosis bronchiectasis enrolled in the phase 3 ORBIT-3 study.
Intervention
Ciprofloxacin dispersion for inhalation.
Comparator
Placebo.
Primary question
How does randomized treatment assignment relate to the time to first exacerbation over one year?
3. Trial Design
Study timeline
Study start
The registry lists March 31, 2014 as the study start date.
Primary completion
The registry lists August 17, 2016 as the primary completion date.
Registry status
The study is listed as completed.
4. Treatment Arms
Ciprofloxacin dispersion for inhalation
- Intervention type: drug
- Study role: active intervention
- Trial design: randomized, parallel
Placebo
- Intervention type: drug
- Study role: comparator
- Trial design: randomized, parallel
5. Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Time to first exacerbation | Time to first exacerbation | One Year |
The endpoint is a time-to-event outcome. Unlike a simple binary endpoint, it incorporates the timing of the first exacerbation. A participant who experiences the first exacerbation earlier contributes a different event time from a participant who experiences it later.
The one-year time frame defines the period over which the registry specifies this primary endpoint. The statistical analysis therefore needs to preserve information about when the first event occurs rather than reducing the endpoint immediately to a yes/no indicator.
6. Why Time to First Exacerbation Is a Time-to-Event Endpoint
A time-to-first-exacerbation endpoint has two pieces of information: whether the event occurred during observation and, if it occurred, when it occurred. This distinction is important because participants can have different follow-up times.
Event time
For a participant who experiences an exacerbation, the analysis records the time from the relevant study starting point to the first qualifying exacerbation.
Censoring
If a participant has not experienced the event by the end of available follow-up, the observation can contribute information up to the time at which follow-up ends or the participant is otherwise censored.
Why timing matters
Two participants can both remain event-free at a particular assessment while having different amounts of observed follow-up. Time-to-event methods retain that distinction.
Primary comparison
The randomized treatment groups can be compared across the distribution of time until the first exacerbation rather than only by the proportion with an event.
7. Planned Analysis
The registry identifies time to first exacerbation over one year as the primary endpoint. No posted statistical analyses are present in the ClinicalTrials.gov information summarized for this record, so the registry does not provide an outcome estimate, confidence interval, or p-value for this endpoint.
Primary statistical question
The appropriate analysis preserves the ordering of event times and accounts for participants whose first exacerbation is not observed during their available follow-up.
Typical analysis framework
For a randomized trial with a time-to-first-event endpoint, a conventional primary framework would use Kaplan-Meier estimation to describe the time-to-event distribution in each treatment group and a time-to-event comparison such as a log-rank test to compare those distributions. A Cox proportional-hazards model is commonly used when a hazard ratio is the chosen measure of relative treatment effect.
Here, S(t) represents the probability of remaining free of the first event beyond time t. For ORBIT-3, the event of interest is the first exacerbation.
The ClinicalTrials.gov record does not report a posted statistical analysis specifying which of these methods was actually used for the primary endpoint. The distinction matters: The statistical method should be documented from the protocol or statistical analysis plan before a specific inferential result is attributed to the trial.
8. Statistical Methodology
Kaplan-Meier estimation
Kaplan-Meier estimation is designed for time-to-event data with right censoring. It estimates the probability of remaining event-free as follow-up progresses, updating the estimate when events occur while retaining information from participants who are censored.
At each observed event time ti, di is the number of events and ni is the number at risk immediately beforehand.
For ORBIT-3, the Kaplan-Meier quantity would be interpreted as the estimated probability of remaining without a first exacerbation through each point in the one-year observation period.
Log-rank comparison
A log-rank test compares the observed and expected numbers of events between treatment groups across the follow-up period. It is particularly suited to randomized trials in which the endpoint is the time until a first event.
The test does not measure the magnitude of a treatment effect. A small p-value, if one were reported, would indicate evidence against a specified null hypothesis of comparable event-time distributions; it would not itself tell the reader how large the treatment difference is.
Cox proportional-hazards model
A Cox model provides a way to estimate a relative hazard associated with treatment assignment while accounting for the timing of events and censoring. If a hazard ratio were reported, a value below 1 would indicate a lower estimated instantaneous event rate in the intervention group relative to the comparator under the fitted model.
For a treatment indicator X, the treatment hazard ratio is represented by exp(β). The model separates the baseline hazard from the relative effect associated with the treatment covariate.
Hazard ratio versus probability of an exacerbation
A hazard ratio is not the same as a risk ratio. It describes a relative rate of experiencing the event at a given point in time under a fitted time-to-event model. A probability such as the proportion experiencing an exacerbation by one year answers a different question and depends on the full event-time distribution.
Analysis population
Because the study is randomized, the principal efficacy comparison would ordinarily be anchored to treatment assignment rather than selectively excluding participants after randomization. The exact analysis population and censoring rules for ORBIT-3 are not reported in the registry information summarized here.
9. Statistical Methods Explained
Why is time to first exacerbation different from a binary endpoint?
A binary analysis would ask whether an exacerbation occurred during a specified period. A time-to-event analysis additionally asks when the first exacerbation occurred. That extra timing information can make the analysis more informative when participants have different follow-up times.
Why use Kaplan-Meier estimation?
Kaplan-Meier estimation accommodates right-censored observations. A participant who has not experienced the first exacerbation by the end of observed follow-up does not simply disappear from the analysis; their event-free information contributes up to the censoring time.
What does a hazard ratio represent?
A hazard ratio compares estimated instantaneous event rates between treatment groups within a fitted survival model. For a time-to-first-exacerbation endpoint, a hazard ratio below 1 would represent a lower estimated instantaneous rate of first exacerbation in the intervention group under that model.
Why is a hazard ratio not the same as a one-year risk ratio?
The hazard ratio summarizes a relative event rate over time, whereas a one-year risk ratio compares two cumulative probabilities at a specific time point. The two measures can differ substantially because they summarize different aspects of the event-time distribution.
Why does censoring matter?
Participants may reach the end of their available follow-up without an observed first exacerbation. Time-to-event methods can use their event-free follow-up while treating the observation as censored. The validity of the resulting inference depends in part on whether the censoring mechanism is compatible with the assumptions of the analysis.
Why does the proportional-hazards assumption matter?
A Cox hazard ratio is easiest to interpret as a common relative hazard when the treatment hazards are reasonably represented by proportional hazards over time. If the relative hazards change substantially, a single hazard ratio can compress a more complicated time-varying treatment effect into one summary number.
10. Understanding the Primary Endpoint Statistically
Time to first exacerbation measures how long participants remain free of their first qualifying exacerbation. It therefore contains more information than simply recording whether an event occurred at some point.
A treatment effect on this endpoint would indicate a difference between randomized groups in the distribution of time until the first exacerbation. The exact numerical magnitude of that effect is not reported in the registry information summarized here.
A difference in time to first exacerbation would not automatically describe the number of subsequent exacerbations, the severity of those exacerbations, or every other clinical outcome. Those questions require their own prespecified endpoints and analyses.
The registry specifies a one-year time frame. A treatment effect estimated over one year should not automatically be generalized to a different follow-up period without considering how the event process and censoring change over time.
11. Censoring and Follow-Up
Time-to-event analysis requires explicit rules for determining when a participant's event-free observation ends. Participants who experience the first exacerbation contribute an observed event time. Participants who do not experience the event during their available observation can contribute censored follow-up.
| Observation pattern | Statistical role |
|---|---|
| First exacerbation observed | Contributes an event at the observed time. |
| No first exacerbation during available follow-up | Can contribute censored event-free follow-up. |
| Different follow-up durations | Time-to-event methods can retain the information available before censoring. |
The registry information summarized here does not report the detailed censoring rules, whether particular intercurrent events altered endpoint definitions, or how missing follow-up was handled. Those details can materially affect a time-to-event analysis and ordinarily belong in the statistical analysis plan.
12. Non-Inferiority, Equivalence, and Superiority
The ClinicalTrials.gov information summarized here does not identify a non-inferiority or equivalence objective. The registered primary endpoint is time to first exacerbation, with a one-year time frame.
This distinction is important because non-inferiority trials require a prespecified margin and a specific interpretation of the confidence interval relative to that margin. Those design elements cannot be assumed simply because a trial has a randomized comparator.
Non-inferiority
Requires a prespecified margin defining how much loss of efficacy would still be considered acceptable for the trial's objective.
Superiority
Tests whether the treatment groups differ in the specified direction, subject to the prespecified hypothesis-testing framework.
No non-inferiority margin is reported in the registry information summarized here, so no margin-based interpretation is appropriate for ORBIT-3 from this record alone.
13. Multiplicity and Multiple Endpoints
The registry identifies one registered primary endpoint: time to first exacerbation. The information summarized here does not report additional registered primary endpoints, a multiplicity strategy, or an alpha-allocation procedure.
| Feature | Registry information | Statistical implication |
|---|---|---|
| Registered primary endpoint | Time to first exacerbation | Defines the principal efficacy outcome recorded in the registry. |
| Primary time frame | One Year | Defines the registered observation period for the endpoint. |
| Multiplicity strategy | Not reported | No specific multiplicity adjustment should be attributed to the trial from this record. |
| Alpha-spending procedure | Not reported | No interim-error-control method should be assumed from the registry information. |
14. Interim Analysis
The ClinicalTrials.gov information summarized here does not report an interim analysis plan, interim efficacy boundary, alpha-spending procedure, or stopping rule.
For a time-to-event endpoint, an interim analysis can be based on accumulated events or another prespecified information measure. If such analyses are performed, repeated examination of the data can affect the overall type I error unless the monitoring strategy is incorporated into the trial design.
15. Missing Data and Imputation
The registry information summarized here does not report a missing-data or imputation strategy for the primary endpoint.
For a time-to-first-exacerbation endpoint, missing follow-up is generally addressed through the censoring framework rather than by simply filling in an unobserved event time. The validity of that approach depends on assumptions about why observations become censored and whether censoring is related to the underlying event process.
Censoring is not ordinary missingness
A censored participant contributes information about remaining event-free up to the censoring time. Treating that observation as if no information were available would discard usable follow-up.
Imputation is not automatically appropriate
Imputing an exact first-exacerbation time would require assumptions about an event that was not observed. The appropriate approach depends on the prespecified endpoint and analysis plan.
16. Stratification and Covariate Adjustment
The registry information summarized here identifies randomized allocation but does not report stratification factors for ORBIT-3.
In a randomized trial, stratification can be incorporated into the analysis when it was part of the randomization procedure. A stratified time-to-event analysis compares event experience while accounting for prespecified strata, whereas an unstratified analysis treats the treatment groups as a single population.
17. Bayesian Methods
No Bayesian statistical method is reported in the ClinicalTrials.gov information summarized here.
A Bayesian time-to-event analysis would combine a prior distribution with the observed event and censoring information to produce a posterior distribution for quantities such as a treatment effect. That framework differs from conventional frequentist confidence intervals and p-values and should not be attributed to ORBIT-3 without documentation that it was part of the trial's statistical plan.
18. What a Formal Result Would Need to Show
A complete statistical presentation of the registered primary endpoint would normally provide more than a single conclusion. A reader evaluating the analysis would ideally see the event-time estimate, the treatment-effect measure, its uncertainty, and the hypothesis-testing framework used to obtain it.
| Result component | Why it matters |
|---|---|
| Number of participants analyzed | Defines the population contributing to the primary analysis. |
| Number of first exacerbations | Indicates the amount of event information available. |
| Kaplan-Meier estimates | Describe the event-free distribution over time. |
| Median time to first exacerbation, if estimable | Provides an interpretable summary of the event-time distribution. |
| Hazard ratio, if prespecified | Provides a relative treatment-effect measure under a Cox model. |
| Confidence interval | Describes statistical uncertainty around the estimated effect. |
| P-value, if part of the prespecified test | Quantifies evidence against the corresponding null hypothesis under the stated testing framework. |
None of these numerical outcome estimates is reported in the ClinicalTrials.gov information summarized here. The registry therefore establishes the primary endpoint and study design but does not provide a posted statistical result for that endpoint.
19. How to Read a Future Hazard Ratio
If a Cox model for time to first exacerbation produced a hazard ratio of 0.70, the model would indicate an estimated instantaneous exacerbation rate 30% lower in the intervention group than in the comparator group, because 1 − 0.70 = 0.30.
This would not mean that exactly 30% fewer participants experienced an exacerbation, nor that each participant had a 30% reduction in personal risk. The hazard ratio summarizes the fitted time-to-event comparison.
The confidence interval would show the statistical uncertainty surrounding the hazard-ratio estimate. A narrow interval indicates greater statistical precision than a wide interval, all else equal. The interval is not a range containing the individual treatment effects experienced by patients.
A p-value addresses evidence against a specified null hypothesis under the analysis model. It does not measure the size, clinical importance, or precision of the treatment effect. Those questions require the effect estimate and its confidence interval.
20. Statistical Assumptions That Matter
Time-to-event analyses are powerful because they retain information about when events occur, but their interpretation depends on assumptions and design choices.
| Issue | Why it matters for ORBIT-3 |
|---|---|
| Independent randomization | The randomized allocation is the foundation for comparing treatment groups without systematic treatment-selection bias. |
| Censoring | Participants without an observed first exacerbation may contribute censored follow-up; assumptions about censoring affect inference. |
| Proportional hazards | A single Cox hazard ratio has a straightforward interpretation when the relative hazards are reasonably stable over time. |
| Endpoint definition | The analysis must use the prespecified definition of an exacerbation and the corresponding event date. |
| Analysis population | Different analysis populations can answer different questions and should be defined before outcome evaluation. |
| Multiplicity | If multiple confirmatory hypotheses or repeated looks were planned, the testing framework needs to control the relevant error rate. |
21. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record does not provide a posted numerical analysis of the registered primary endpoint in the information summarized here.
- No outcome estimates: the registry information summarized here does not report an event count, hazard ratio, confidence interval, p-value, median time to first exacerbation, or time-specific event probability.
- Limited arm-level detail: total enrollment and the existence of two arms are reported, but arm-specific enrollment counts are not reported in the ClinicalTrials.gov record.
- Limited analysis-plan detail: the registry information summarized here does not specify the exact statistical model, censoring rules, analysis population, stratification factors, missing-data approach, or multiplicity strategy.
- No subgroup results: the record information summarized here does not report subgroup estimates for the primary endpoint.
- No safety results: the record information summarized here does not provide serious adverse-event counts by treatment arm.
- No non-inferiority margin: no non-inferiority or equivalence margin is reported in the registry information summarized here.
- No interim-analysis details: no interim statistical boundary or alpha-spending procedure is reported in the registry information summarized here.
These limitations affect how far the statistical interpretation can go. The study design and registered endpoint establish the framework for analysis, but numerical treatment-effect conclusions require the corresponding posted results or detailed statistical reporting.
22. Why This Trial Matters Statistically
ORBIT-3 is a useful teaching example because its registered primary endpoint illustrates an important shift from simple event counting to time-to-event analysis. The statistical question is not merely whether an exacerbation occurred, but how the distribution of time until the first exacerbation differs between randomized groups over a defined observation period.
| Concept | How it appears in ORBIT-3 |
|---|---|
| Randomization | The study uses randomized allocation to compare two treatment arms. |
| Parallel-group design | The registry identifies a parallel design with two arms. |
| Masking | The registry classifies the study as quadruple-masked. |
| Time-to-event endpoint | The primary endpoint is time to first exacerbation. |
| Censoring | Time-to-event analysis can preserve information from participants whose event is not observed during available follow-up. |
| Kaplan-Meier estimation | Provides a natural framework for describing event-free probability over time. |
| Log-rank testing | Provides a conventional way to compare event-time distributions between randomized groups. |
| Cox regression | Provides a conventional model-based framework for estimating a hazard ratio. |
| Confidence intervals | Would quantify uncertainty around a treatment-effect estimate when results are reported. |
| Analysis assumptions | Censoring and proportional-hazards assumptions can affect interpretation of time-to-event results. |
23. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered primary endpoint is a time-to-event outcome. Its analysis should compare the distribution of time to first exacerbation between randomized treatment groups while accounting for censoring and the prespecified inferential framework.
Clinical interpretation
The endpoint is clinically oriented toward the timing of the first exacerbation during the one-year time frame. A numerical treatment effect would need to be interpreted alongside the event definition, follow-up, uncertainty, and the broader clinical context.
24. Results Status
The registry establishes that the study was a completed phase 3 randomized parallel-group trial with 278 enrolled participants and two intervention arms. It identifies ciprofloxacin dispersion for inhalation and placebo as the interventions and specifies time to first exacerbation over one year as the primary endpoint.
Because the registry does not report the corresponding numerical outcome analysis, there is no reported hazard ratio, confidence interval, p-value, event count, or median time to first exacerbation to interpret from this record.
25. Sources
- ClinicalTrials.gov: ORBIT-3, NCT01515007.
Continue through Clinical Biostats
Explore additional clinical trial analyses, statistical tutorials, and practical statistical methods.
26. Record Summary
ORBIT-3 was a completed phase 3 randomized, parallel-group, quadruple-masked treatment study with 278 enrolled participants and two arms evaluating ciprofloxacin dispersion for inhalation versus placebo in non-cystic-fibrosis bronchiectasis. The registry identifies time to first exacerbation over one year as the primary endpoint.
Statistically, the defining feature of the trial is its time-to-event primary outcome. An appropriate analysis preserves the timing of the first exacerbation, accounts for censored observations, and can use Kaplan-Meier estimation, a log-rank comparison, and a Cox proportional-hazards model when those methods are specified by the study's statistical plan. The interpretation of any resulting hazard ratio must remain distinct from cumulative event probabilities and must consider censoring and model assumptions.